From e95c49b76f4ac7bf72c383ee43a0567dfcbf1603 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Mon, 10 Sep 2018 18:15:02 -0400 Subject: 0.1.0 renamed doc-reform, doc_reform (& rad) - from sdp --- src/doc_reform/meta/conf_make_meta_json.d | 607 +++ src/doc_reform/meta/conf_make_meta_structs.d | 328 ++ src/doc_reform/meta/conf_make_meta_toml.d | 77 + src/doc_reform/meta/defaults.d | 241 + src/doc_reform/meta/doc_debugs.d | 652 +++ src/doc_reform/meta/metadoc.d | 177 + src/doc_reform/meta/metadoc_from_src.d | 6322 ++++++++++++++++++++++++++ src/doc_reform/meta/metadoc_summary.d | 101 + src/doc_reform/meta/object_setter.d | 126 + src/doc_reform/meta/package.d | 17 + src/doc_reform/meta/rgx.d | 290 ++ 11 files changed, 8938 insertions(+) create mode 100644 src/doc_reform/meta/conf_make_meta_json.d create mode 100644 src/doc_reform/meta/conf_make_meta_structs.d create mode 100644 src/doc_reform/meta/conf_make_meta_toml.d create mode 100644 src/doc_reform/meta/defaults.d create mode 100644 src/doc_reform/meta/doc_debugs.d create mode 100644 src/doc_reform/meta/metadoc.d create mode 100644 src/doc_reform/meta/metadoc_from_src.d create mode 100644 src/doc_reform/meta/metadoc_summary.d create mode 100644 src/doc_reform/meta/object_setter.d create mode 100644 src/doc_reform/meta/package.d create mode 100644 src/doc_reform/meta/rgx.d (limited to 'src/doc_reform/meta') diff --git a/src/doc_reform/meta/conf_make_meta_json.d b/src/doc_reform/meta/conf_make_meta_json.d new file mode 100644 index 0000000..d6adfa7 --- /dev/null +++ b/src/doc_reform/meta/conf_make_meta_json.d @@ -0,0 +1,607 @@ +/++ + json headers
+ extract json header return json ++/ +module doc_reform.meta.conf_make_meta_json; +static template contentJSONtoSiSUstruct() { + import + std.exception, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.utf, + std.conv : to; + import + doc_reform.meta.conf_make_meta_structs, + doc_reform.meta.conf_make_meta_json, + doc_reform.meta.rgx; + ConfCompositePlus _struct_composite; + auto contentJSONtoSiSUstruct(C, J)(C _struct_composite, J _json, string _identifier) { + mixin SiSUrgxInit; + static auto _rgx = Rgx(); + debug (json) { + writeln(">> --------------------------- >>"); + foreach (tag0; _json.object.byKeyValue) { + if (tag0.value.stringof == "string") { + writeln(tag0.key, ": ", tag0.value); + } else { + // writeln(tag0.key, ":"); + foreach (tag1; tag0.value.object.byKeyValue) { + writeln(tag0.key, ":", tag1.key, ": ", tag1.value); + } + } + } + writeln("<< --------------------------- <<"); + } + confCompositeMakeBuild _mk; + /+ make ------------------------------------------------------------------- +/ + if ("make" in _json.object) { + if ("bold" in _json.object["make"] + && (_json.object["make"]["bold"].type().to!string == "STRING") + ) { + _struct_composite.make_str.bold = _json.object["make"]["bold"].str; + } + if ("breaks" in _json.object["make"] + && (_json.object["make"]["breaks"].type().to!string == "STRING") + ) { + _struct_composite.make_str.breaks = _json.object["make"]["breaks"].str; + } + if ("cover_image" in _json.object["make"] + && (_json.object["make"]["cover_image"].type().to!string == "STRING") + ) { + _struct_composite.make_str.cover_image = _json.object["make"]["cover_image"].str; + } + if ("css" in _json.object["make"] + && (_json.object["make"]["css"].type().to!string == "STRING") + ) { + _struct_composite.make_str.css = _json.object["make"]["css"].str; + } + if ("emphasis" in _json.object["make"] + && (_json.object["make"]["emphasis"].type().to!string == "STRING") + ) { + _struct_composite.make_str.emphasis = _json.object["make"]["emphasis"].str; + } + if ("footer" in _json.object["make"] + && (_json.object["make"]["footer"].type().to!string == "STRING") + ) { + char[][] __match_footer_array + = (cast(char[]) _json.object["make"]["footer"].str) + .split(_rgx.make_heading_delimiter); + _struct_composite.make_str.footer = __match_footer_array.to!(string[]); + } else if ("footer" in _json.object["make"] + && _json.object["make"]["footer"].type().to!string == "ARRAY") { + string[] _match_footer_array; + foreach (_match_heading; _json.object["make"]["footer"].arrayNoRef) { + _match_footer_array ~= _match_heading.str; + } + _struct_composite.make_str.footer = _match_footer_array; + } + if ("headings" in _json.object["make"] + && (_json.object["make"]["headings"].type().to!string == "STRING") + ) { + char[][] __match_headings_array + = (cast(char[]) _json.object["make"]["headings"].str) + .split(_rgx.make_heading_delimiter); + _struct_composite.make_str.headings = __match_headings_array.to!(string[]); + } else if ("headings" in _json.object["make"] + && _json.object["make"]["headings"].type().to!string == "ARRAY") { + string[] _match_headings_array; + foreach (_match_heading; _json.object["make"]["headings"].arrayNoRef) { + _match_headings_array ~= _match_heading.str; + } + _struct_composite.make_str.headings = _match_headings_array; + } + if ("home_button_image" in _json.object["make"] + && (_json.object["make"]["home_button_image"].type().to!string == "STRING") + ) { + char[][] __match_home_button_image_array + = (cast(char[]) _json.object["make"]["home_button_image"].str) + .split(_rgx.make_heading_delimiter); + _struct_composite.make_str.home_button_image = __match_home_button_image_array.to!(string[]); + } else if ("home_button_image" in _json.object["make"] + && _json.object["make"]["home_button_image"].type().to!string == "ARRAY") { + string[] _match_home_button_image_array; + foreach (_match_heading; _json.object["make"]["home_button_image"].arrayNoRef) { + _match_home_button_image_array ~= _match_heading.str; + } + _struct_composite.make_str.home_button_image = _match_home_button_image_array; + } + if ("home_button_text" in _json.object["make"] + && (_json.object["make"]["home_button_text"].type().to!string == "STRING") + ) { + _struct_composite.make_str.home_button_text = _json.object["make"]["home_button_text"].str; + } else if ("home_button_text" in _json.object["make"] + && _json.object["make"]["home_button_text"].type().to!string == "ARRAY" + ) { + string[] _match_home_button_text_array; + foreach (_match_heading; _json.object["make"]["home_button_text"].arrayNoRef) { + _match_home_button_text_array ~= _match_heading.str; + } + string _match_home_button_text_str = (_match_home_button_text_array).join("; "); + _struct_composite.make_str.home_button_text = _match_home_button_text_str; + } + if ("italics" in _json.object["make"] + && (_json.object["make"]["italics"].type().to!string == "STRING") + ) { + _struct_composite.make_str.italics = _json.object["make"]["italics"].str; + } + if ("auto_num_top_at_level" in _json.object["make"] // str == A - D, 1 - 4 + && (_json.object["make"]["auto_num_top_at_level"].type().to!string == "STRING") + ) { + _struct_composite.make_str.auto_num_top_at_level = _json.object["make"]["auto_num_top_at_level"].str; + switch (_json.object["make"]["auto_num_top_at_level"].str) { + case "A": + break; + case "B": _struct_composite.make_str.auto_num_top_lv = 1; + break; + case "C": _struct_composite.make_str.auto_num_top_lv = 2; + break; + case "D": _struct_composite.make_str.auto_num_top_lv = 3; + break; + case "1": _struct_composite.make_str.auto_num_top_lv = 4; + break; + case "2": _struct_composite.make_str.auto_num_top_lv = 5; + break; + case "3": _struct_composite.make_str.auto_num_top_lv = 6; + break; + case "4": _struct_composite.make_str.auto_num_top_lv = 7; + break; + default: + break; + } + } + if ("auto_num_depth" in _json.object["make"] + && (_json.object["make"]["auto_num_depth"].type().to!string == "INTEGER") + ) { + _struct_composite.make_str.auto_num_depth = _json.object["make"]["auto_num_depth"].integer.to!int; + } else if ("auto_num_depth" in _json.object["make"] + && (_json.object["make"]["auto_num_depth"].type().to!string == "STRING") + ) { + _struct_composite.make_str.auto_num_depth = _json.object["make"]["auto_num_depth"].str.to!int; + } + if ("substitute" in _json.object["make"]) { + string[][] _sub; + if ( + (_json.object["make"]["substitute"].type().to!string == "ARRAY") + && (_json.object["make"]["substitute"][0].type().to!string == "ARRAY") + ) { + foreach (substitute_pair; _json.object["make"]["substitute"].arrayNoRef) { + if ((substitute_pair.type().to!string) == "ARRAY") { + if (!empty(substitute_pair[0].str) && !empty(substitute_pair[1].str)) { + _sub ~= [ substitute_pair[0].str, substitute_pair[1].str]; + } + } + } + } else if ( + (_json.object["make"]["substitute"].type().to!string == "ARRAY") + && (_json.object["make"]["substitute"][0].type().to!string == "STRING") + ) { + if (!empty(_json.object["make"]["substitute"][0].str) && !empty(_json.object["make"]["substitute"][1].str)) { + _sub = [[_json.object["make"]["substitute"][0].str, _json.object["make"]["substitute"][1].str]]; + } + } + // writeln(_sub); + _struct_composite.make_str.substitute = _sub; + } + if ("texpdf_font" in _json.object["make"] + && (_json.object["make"]["texpdf_font"].type().to!string == "STRING") + ) { + _struct_composite.make_str.texpdf_font = _json.object["make"]["texpdf_font"].str; + } + _struct_composite.make.bold = _mk.bold(_struct_composite.make_str.bold); + _struct_composite.make.breaks = _mk.breaks(_struct_composite.make_str.breaks); + _struct_composite.make.cover_image = _mk.cover_image(_struct_composite.make_str.cover_image); + _struct_composite.make.css = _mk.css(_struct_composite.make_str.css); + _struct_composite.make.emphasis = _mk.emphasis(_struct_composite.make_str.emphasis); + _struct_composite.make.footer = _mk.footer(_struct_composite.make_str.footer); + _struct_composite.make.headings = _mk.headings(_struct_composite.make_str.headings); + _struct_composite.make.home_button_image = _mk.home_button_image(_struct_composite.make_str.home_button_image); + _struct_composite.make.home_button_text = _mk.home_button_text(_struct_composite.make_str.home_button_text); + _struct_composite.make.italics = _mk.italics(_struct_composite.make_str.italics); + _struct_composite.make.auto_num_top_at_level = _mk.auto_num_top_at_level(_struct_composite.make_str.auto_num_top_at_level); + _struct_composite.make.auto_num_top_lv = _mk.auto_num_top_lv(_struct_composite.make_str.auto_num_top_lv); + _struct_composite.make.auto_num_depth = _mk.auto_num_depth(_struct_composite.make_str.auto_num_depth); + _struct_composite.make.substitute = _mk.substitute(_struct_composite.make_str.substitute); + _struct_composite.make.texpdf_font = _mk.texpdf_font(_struct_composite.make_str.texpdf_font); + } + /+ conf ------------------------------------------------------------------- +/ + if ("webserv" in _json.object) { + if ("url_root" in _json.object["webserv"] + && (_json.object["webserv"]["url_root"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_url_root = _json.object["webserv"]["url_root"].str; + } + if ("path" in _json.object["webserv"] + && (_json.object["webserv"]["path"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_path = _json.object["webserv"]["path"].str; + } + if ("images" in _json.object["webserv"] + && (_json.object["webserv"]["images"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_images = _json.object["webserv"]["images"].str; + } + if ("cgi" in _json.object["webserv"] + && (_json.object["webserv"]["cgi"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi = _json.object["webserv"]["cgi"].str; + } + if ("cgi_host" in _json.object["webserv"] + && (_json.object["webserv"]["cgi_host"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi_host = _json.object["webserv"]["cgi_host"].str; + } + if ("cgi_host_path" in _json.object["webserv"] + && (_json.object["webserv"]["cgi_host_path"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi_host_path = _json.object["webserv"]["cgi_host_path"].str; + } + if ("cgi_port" in _json.object["webserv"] + && (_json.object["webserv"]["cgi_port"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi_port = _json.object["webserv"]["cgi_port"].str; + } + if ("cgi_user" in _json.object["webserv"] + && (_json.object["webserv"]["cgi_user"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi_user = _json.object["webserv"]["cgi_user"].str; + } + if ("cgi_file_links" in _json.object["webserv"] + && (_json.object["webserv"]["cgi_file_links"].type().to!string == "STRING") + ) { + _struct_composite.conf.webserv_cgi_file_links = _json.object["webserv"]["cgi_file_links"].str; + } + } + if ("processing" in _json.object) { + if ("path" in _json.object["processing"] + && (_json.object["processing"]["path"].type().to!string == "STRING") + ) { + _struct_composite.conf.processing_path = _json.object["processing"]["path"].str; + } + if ("dir" in _json.object["processing"] + && (_json.object["processing"]["dir"].type().to!string == "STRING") + ) { + _struct_composite.conf.processing_dir = _json.object["processing"]["dir"].str; + } + if ("concord_max" in _json.object["processing"] + && (_json.object["processing"]["concord_max"].type().to!string == "STRING") + ) { + _struct_composite.conf.processing_concord_max = _json.object["processing"]["concord_max"].str; + } + } + if ("flag" in _json.object) { + if ("act0" in _json.object["flag"] + && (_json.object["flag"]["act0"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act0 = _json.object["flag"]["act0"].str; + } + if ("act1" in _json.object["flag"] + && (_json.object["flag"]["act1"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act1 = _json.object["flag"]["act1"].str; + } + if ("act2" in _json.object["flag"] + && (_json.object["flag"]["act2"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act2 = _json.object["flag"]["act2"].str; + } + if ("act3" in _json.object["flag"] + && (_json.object["flag"]["act3"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act3 = _json.object["flag"]["act3"].str; + } + if ("act4" in _json.object["flag"] + && (_json.object["flag"]["act4"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act4 = _json.object["flag"]["act4"].str; + } + if ("act5" in _json.object["flag"] + && (_json.object["flag"]["act5"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act5 = _json.object["flag"]["act5"].str; + } + if ("act6" in _json.object["flag"] + && (_json.object["flag"]["act6"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act6 = _json.object["flag"]["act6"].str; + } + if ("act7" in _json.object["flag"] + && (_json.object["flag"]["act7"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act7 = _json.object["flag"]["act7"].str; + } + if ("act8" in _json.object["flag"] + && (_json.object["flag"]["act8"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act8 = _json.object["flag"]["act8"].str; + } + if ("act9" in _json.object["flag"] + && (_json.object["flag"]["act9"].type().to!string == "STRING") + ) { + _struct_composite.conf.flag_act9 = _json.object["flag"]["act9"].str; + } + } + if ("default" in _json.object) { + if ("papersize" in _json.object["default"] + && (_json.object["default"]["papersize"].type().to!string == "STRING") + ) { + _struct_composite.conf.default_papersize = _json.object["default"]["papersize"].str; + } + if ("text_wrap" in _json.object["default"] + && (_json.object["default"]["text_wrap"].type().to!string == "STRING") + ) { + _struct_composite.conf.default_text_wrap = _json.object["default"]["text_wrap"].str; + } + if ("emphasis" in _json.object["default"] + && (_json.object["default"]["emphasis"].type().to!string == "STRING") + ) { + _struct_composite.conf.default_emphasis = _json.object["default"]["emphasis"].str; + } + if ("language" in _json.object["default"] + && (_json.object["default"]["language"].type().to!string == "STRING") + ) { + _struct_composite.conf.default_language = _json.object["default"]["language"].str; + } + if ("digest" in _json.object["default"] + && (_json.object["default"]["digest"].type().to!string == "STRING") + ) { + _struct_composite.conf.default_digest = _json.object["default"]["digest"].str; + } + } + if ("search" in _json.object) { + if ("flag" in _json.object["search"] + && (_json.object["search"]["flag"].type().to!string == "STRING") + ) { + _struct_composite.conf.search_flag = _json.object["search"]["flag"].str; + } + if ("action" in _json.object["search"] + && (_json.object["search"]["action"].type().to!string == "STRING") + ) { + _struct_composite.conf.search_action = _json.object["search"]["action"].str; + } + if ("db" in _json.object["search"] + && (_json.object["search"]["db"].type().to!string == "STRING") + ) { + _struct_composite.conf.search_db = _json.object["search"]["db"].str; + } + if ("title" in _json.object["search"] + && (_json.object["search"]["title"].type().to!string == "STRING") + ) { + _struct_composite.conf.search_title = _json.object["search"]["title"].str; + } + } + /+ meta ------------------------------------------------------------------- +/ + if ("classify" in _json.object) { + if ("dewey" in _json.object["classify"] + && (_json.object["classify"]["dewey"].type().to!string == "STRING") + ) { + _struct_composite.meta.classify_dewey = _json.object["classify"]["dewey"].str; + } + if ("keywords" in _json.object["classify"] + && (_json.object["classify"]["keywords"].type().to!string == "STRING") + ) { + _struct_composite.meta.classify_keywords = _json.object["classify"]["keywords"].str; + } + if ("loc" in _json.object["classify"] + && (_json.object["classify"]["loc"].type().to!string == "STRING") + ) { + _struct_composite.meta.classify_loc = _json.object["classify"]["loc"].str; + } + if ("subject" in _json.object["classify"] + && (_json.object["classify"]["subject"].type().to!string == "STRING") + ) { + _struct_composite.meta.classify_subject = _json.object["classify"]["subject"].str; + } + if ("topic_register" in _json.object["classify"] + && (_json.object["classify"]["topic_register"].type().to!string == "STRING") + ) { + _struct_composite.meta.classify_topic_register = _json.object["classify"]["topic_register"].str; + } + } + if ("date" in _json.object) { + if ("added_to_site" in _json.object["date"] + && (_json.object["date"]["added_to_site"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_added_to_site = _json.object["date"]["added_to_site"].str; + } + if ("available" in _json.object["date"] + && (_json.object["date"]["available"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_available = _json.object["date"]["available"].str; + } + if ("created" in _json.object["date"] + && (_json.object["date"]["created"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_created = _json.object["date"]["created"].str; + } + if ("issued" in _json.object["date"] + && (_json.object["date"]["issued"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_issued = _json.object["date"]["issued"].str; + } + if ("modified" in _json.object["date"] + && (_json.object["date"]["modified"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_modified = _json.object["date"]["modified"].str; + } + if ("published" in _json.object["date"] + && (_json.object["date"]["published"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_published = _json.object["date"]["published"].str; + } + if ("valid" in _json.object["date"] + && (_json.object["date"]["valid"].type().to!string == "STRING") + ) { + _struct_composite.meta.date_valid = _json.object["date"]["valid"].str; + } + } + if ("links" in _json.object) {} + if ("notes" in _json.object) { + if ("abstract" in _json.object["notes"] + && (_json.object["notes"]["abstract"].type().to!string == "STRING") + ) { + _struct_composite.meta.notes_abstract = _json.object["notes"]["abstract"].str; + } + if ("description" in _json.object["notes"] + && (_json.object["notes"]["description"].type().to!string == "STRING") + ) { + _struct_composite.meta.notes_description = _json.object["notes"]["description"].str; + } + } + if ("original" in _json.object) { + if ("language" in _json.object["original"] + && (_json.object["original"]["language"].type().to!string == "STRING") + ) { + _struct_composite.meta.original_language = _json.object["original"]["language"].str; + } + if ("language_char" in _json.object["original"] + && (_json.object["original"]["language_char"].type().to!string == "STRING") + ) { + _struct_composite.meta.original_language_char = _json.object["original"]["language_char"].str; + } + if ("source" in _json.object["original"] + && (_json.object["original"]["source"].type().to!string == "STRING") + ) { + _struct_composite.meta.original_source = _json.object["original"]["source"].str; + } + if ("title" in _json.object["original"] + && (_json.object["original"]["title"].type().to!string == "STRING") + ) { + _struct_composite.meta.original_title = _json.object["original"]["title"].str; + } + } + if ("publisher" in _json.object) {} + if ("rights" in _json.object) { + if ("copyright" in _json.object["rights"] + && (_json.object["rights"]["copyright"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright = _json.object["rights"]["copyright"].str; + } + if ("copyright_text" in _json.object["rights"] + && (_json.object["rights"]["copyright_text"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_text = _json.object["rights"]["copyright_text"].str; + } + if ("copyright_audio" in _json.object["rights"] + && (_json.object["rights"]["copyright_audio"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_audio = _json.object["rights"]["copyright_audio"].str; + } + if ("copyright_cover" in _json.object["rights"] + && (_json.object["rights"]["copyright_cover"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_cover = _json.object["rights"]["copyright_cover"].str; + } + if ("copyright_illustrations" in _json.object["rights"] + && (_json.object["rights"]["copyright_illustrations"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_illustrations = _json.object["rights"]["copyright_illustrations"].str; + } + if ("copyright_photographs" in _json.object["rights"] + && (_json.object["rights"]["copyright_photographs"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_photographs = _json.object["rights"]["copyright_photographs"].str; + } + if ("copyright_translation" in _json.object["rights"] + && (_json.object["rights"]["copyright_translation"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_translation = _json.object["rights"]["copyright_translation"].str; + } + if ("copyright_video" in _json.object["rights"] + && (_json.object["rights"]["copyright_video"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_copyright_video = _json.object["rights"]["copyright_video"].str; + } + if ("license" in _json.object["rights"] + && (_json.object["rights"]["license"].type().to!string == "STRING") + ) { + _struct_composite.meta.rights_license = _json.object["rights"]["license"].str; + } + } + if (_struct_composite.meta.creator_author.empty) { + if ("creator" in _json.object) { + if ("author" in _json.object["creator"] + && (_json.object["creator"]["author"].type().to!string == "STRING") + ) { + _struct_composite.meta.creator_author = _json.object["creator"]["author"].str; + } + if ("email" in _json.object["creator"] + && (_json.object["creator"]["email"].type().to!string == "STRING") + ) { + _struct_composite.meta.creator_author_email = _json.object["creator"]["email"].str; + } + if ("illustrator" in _json.object["creator"] + && (_json.object["creator"]["illustrator"].type().to!string == "STRING") + ) { + _struct_composite.meta.creator_illustrator = _json.object["creator"]["illustrator"].str; + } + if ("translator" in _json.object["creator"] + && (_json.object["creator"]["translator"].type().to!string == "STRING") + ) { + _struct_composite.meta.creator_translator = _json.object["creator"]["translator"].str; + } + } + string[] authors_arr; + auto authors_raw_arr = _struct_composite.meta.creator_author.split(_rgx.arr_delimiter); + foreach (author_raw; authors_raw_arr) { + authors_arr ~= author_raw.replace(_rgx.raw_author_munge, "$2 $1"); + } + _struct_composite.meta.creator_author = join(authors_arr, ", ").chomp.chomp; + } + if (_struct_composite.meta.title_main.empty) { + if ("title" in _json.object) { + if ((_json.object["title"].type().to!string) == "STRING") { + _struct_composite.meta.title_main = _json.object["title"].str; + } else { + if ("edition" in _json.object["title"] + && (_json.object["title"]["edition"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_edition = _json.object["title"]["edition"].str; + } + if ("full" in _json.object["title"] + && (_json.object["title"]["full"].type().to!string == "STRING") + ) {} + if ("language" in _json.object["title"] + && (_json.object["title"]["language"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_language = _json.object["title"]["language"].str; + } + if ("main" in _json.object["title"] + && (_json.object["title"]["main"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_main = _json.object["title"]["main"].str; + } + if ("note" in _json.object["title"] + && (_json.object["title"]["note"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_note = _json.object["title"]["note"].str; + } + if ("sub" in _json.object["title"] + && (_json.object["title"]["sub"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_sub = _json.object["title"]["sub"].str; + } + if ("subtitle" in _json.object["title"] + && (_json.object["title"]["subtitle"].type().to!string == "STRING") + ) { + _struct_composite.meta.title_subtitle = _json.object["title"]["subtitle"].str; + } + } + } + if ((!(_struct_composite.meta.title_subtitle.empty)) + && (_struct_composite.meta.title_sub.empty)) { + _struct_composite.meta.title_sub = _struct_composite.meta.title_subtitle; + } + _struct_composite.meta.title_full = (_struct_composite.meta.title_sub.empty) + ? _struct_composite.meta.title_main + : format( + "%s - %s", + _struct_composite.meta.title_main, + _struct_composite.meta.title_sub, + ); + } + return _struct_composite; + } +} diff --git a/src/doc_reform/meta/conf_make_meta_structs.d b/src/doc_reform/meta/conf_make_meta_structs.d new file mode 100644 index 0000000..cf279a5 --- /dev/null +++ b/src/doc_reform/meta/conf_make_meta_structs.d @@ -0,0 +1,328 @@ +module doc_reform.meta.conf_make_meta_structs; +import + std.exception, + std.json, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.utf, + std.conv : to; +import + doc_reform.meta.defaults, + doc_reform.meta.rgx; +mixin SiSUrgxInit; +static auto _rgx = Rgx(); +mixin InternalMarkup; +auto _mkup = InlineMarkup(); +auto url_markup(string line) { + auto line_ = (line) + .replaceAll( + _rgx.smid_inline_link_markup_regular, + ("$1" + ~ _mkup.lnk_o ~ "$2" ~ _mkup.lnk_c + ~ _mkup.url_o ~ "$3" ~ _mkup.url_c + ~ "$4") // ("$1{ $2 }$3$4") + ) + .replaceAll( + _rgx.smid_inline_link_naked_url, + ("$1" + ~ _mkup.lnk_o ~ "$2" ~ _mkup.lnk_c + ~ _mkup.url_o ~ "$2" ~ _mkup.url_c + ~ "$3") // ("$1{ $2 }$2$3") + ) + .replaceAll( + _rgx.arr_delimiter, + _mkup.br_line + ); + return line_; +} +struct ConfCompositeMakeStr { + string bold; + string breaks; + string cover_image; + string css; + string emphasis; + string[] footer; + string[] headings; + string[] home_button_image; + string home_button_text = "{Doc Reform}http://www.doc-reform.org;" + ~ " {www.doc-reform.org}http://www.doc-reform.org;" + ~ " {sources / git}https://git.doc-reform.org/software/doc-reform"; + string italics; + string auto_num_top_at_level; + int auto_num_top_lv = 9; + int auto_num_depth = 2; + string[][] substitute; + string texpdf_font; +} +struct confCompositeMakeBuild { + auto bold(string _mk) { + string[] _out; + if (_mk) { + _out = [ (cast(string) (`(` ~ _mk.dup ~ `)`)), "*{$1}*", "$1"]; + } + return _out; + } + auto breaks(string _mk) { + return _mk; + } + auto cover_image(string _mk) { + return _mk; + } + auto css(string _mk) { + return _mk; + } + auto emphasis(string _mk) { + string[] _out; + if (_mk) { + _out = [ (cast(string) (`(` ~ _mk.dup ~ `)`)), "!{$1}!", "$1" ]; + } + return _out; + } + auto footer(string[] _mk) { + string line_; + string[] _mk2; + foreach (line; _mk) { + _mk2 ~= url_markup(line); + } + return _mk2; + } + auto headings(string[] _mk) { + return _mk; + } + auto home_button_image(string[] _mk) { + return _mk; + } + auto home_button_text(string _mk) { + return url_markup(_mk); + } + auto italics(string _mk) { + string[] _out; + if (_mk) { + _out = [ (cast(string) (`(` ~ _mk.dup ~ `)`)), "/{$1}/", "$1" ]; + } + return _out; + } + auto auto_num_top_at_level(string _mk) { + return _mk; + } + auto auto_num_top_lv(int _mk) { + return _mk; + } + auto auto_num_depth(int _mk) { + return _mk; + } + auto substitute(string[][] _mk) { + return _mk; + } + auto texpdf_font(string _mk) { + return _mk; + } +} +struct ConfCompositeMakeInit { + string[] bold; + string breaks; + string cover_image; + string css; + string[] emphasis; + string[] footer; + string[] headings; + string[] home_button_image; + string home_button_text = "{Doc Reform}http://www.doc-reform.org;" + ~ " {www.doc-reform.org}http://www.doc-reform.org;" + ~ " {sources / git}https://git.doc-reform.org/software/doc-reform"; + string[] italics; + string auto_num_top_at_level; + int auto_num_top_lv = 9; + int auto_num_depth = 2; + string[][] substitute; + string texpdf_font; +} +struct ConfCompositeSiteLocal { + string webserv_url_root; + string webserv_path; + string webserv_images; + string webserv_cgi; + string webserv_cgi_host; + string webserv_cgi_host_path; + string webserv_cgi_port; + string webserv_cgi_user; + string webserv_cgi_file_links; + string processing_path; + string processing_dir; + string processing_concord_max; + string flag_act0; + string flag_act1; + string flag_act2; + string flag_act3; + string flag_act4; + string flag_act5; + string flag_act6; + string flag_act7; + string flag_act8; + string flag_act9; + string default_papersize; + string default_text_wrap; + string default_emphasis; + string default_language; + string default_digest; + string permission_share_source; + string search_flag; + string search_action; + string search_db; + string search_title; +} +struct MetaComposite { + string classify_dewey; + string classify_keywords; + string classify_loc; + string classify_subject; + string classify_topic_register; + string creator_author; + string creator_author_email; + string creator_illustrator; + string creator_translator; + string date_added_to_site; + string date_available; + string date_created; + string date_issued; + string date_modified; + string date_published; + string date_valid; + string identifier_isbn; + string identifier_oclc; + string identifier_pg; + string language_document; + string language_document_char; + string links; + string notes_abstract; + string notes_description; + string original_language; + string original_language_char; + string original_publisher; + string original_source; + string original_title; + string publisher; + string rights_copyright; + string rights_copyright_audio; + string rights_copyright_cover; + string rights_copyright_illustrations; + string rights_copyright_photographs; + string rights_copyright_text; + string rights_copyright_translation; + string rights_copyright_video; + string rights_license; + string title_edition; + string title_full; + string title_language; + string title_main; + string title_note; + string title_short; + string title_sub; + string title_subtitle; +} +struct ConfComposite { + MetaComposite meta; + ConfCompositeMakeInit make; + ConfCompositeSiteLocal conf; +} +struct ConfCompositePlus { + MetaComposite meta; + ConfCompositeMakeInit make; + ConfCompositeMakeStr make_str; + ConfCompositeSiteLocal conf; +} +static auto ptr_head_main + = [ + "classify", + "creator", + "date", + "identifier", + "links", + "make", + "original", + "notes", + "rights", + "title" + ]; +static auto ptr_head_sub_classify + = [ + "dewey", + "keywords", + "loc", + "subject", + "topic_register" + ]; +static auto ptr_head_sub_creator + = [ + "author", + "author_email", + "cover", + "illustrator", + "translator" + ]; +static auto ptr_head_sub_date + = [ + "added_to_site", + "available", + "created", + "issued", + "modified", + "published", + "valid" + ]; +static auto ptr_head_sub_identifier + = [ + "isbn", + "oclc", + "pg" + ]; +/+ make +/ +static auto ptr_head_sub_make + = [ + "cover_image", + "home_button_image", + "home_button_text", + "footer", "headings", + "auto_num_top_at_level", "auto_num_top_lv", "auto_num_depth", + "breaks", + "substitute", + "bold", + "italics", + "emphasis", + "texpdf_font", + "css" + ]; +static auto ptr_head_sub_notes + = [ + "abstract", + "description" + ]; +static auto ptr_head_sub_original + = [ + "language", + "source", + "title" + ]; +static auto ptr_head_sub_publisher + = [ "name" ]; +static auto ptr_head_sub_rights + = [ + "copyright", + "cover", + "illustrations", + "license" + ]; +static auto ptr_head_sub_title + = [ + "edition", + "full", + "language", + "main", + "note", + "sub" + ]; +auto config_jsonstr = `{ +}`; diff --git a/src/doc_reform/meta/conf_make_meta_toml.d b/src/doc_reform/meta/conf_make_meta_toml.d new file mode 100644 index 0000000..d308ae4 --- /dev/null +++ b/src/doc_reform/meta/conf_make_meta_toml.d @@ -0,0 +1,77 @@ +/++ + extract native/orig header return associative array
+ + the header is passed as text (lopped off top of a sisu markup file until the + required first heading ^A~), determine whether is a native header or sdlang one + with a regex check if whether it contains the "native header" required tag/field + @title: then process accordingly as a "native header" or "sdlang header" + converting the metadata and make instructions to a common json format used by + program internally. Moved to associative array. ++/ +module doc_reform.meta.conf_make_meta_toml; +static template configParseTOMLreturnJSON() { + import + toml, + toml.json; + auto configParseTOMLreturnJSON(T)( + T _text + ){ + TOMLDocument _doc; + _doc = parseTOML(cast(string)(_text.content)); + auto _doc_json = toJSON(_doc); + return _doc_json; + } +} +static template configParseTOMLreturnSiSUstruct() { + import + toml, + toml.json; + import + doc_reform.meta.conf_make_meta_structs, + doc_reform.meta.conf_make_meta_json; + mixin contentJSONtoSiSUstruct; + auto configParseTOMLreturnSiSUstruct(CCm, T)( + CCm _make_and_meta_struct, + T _document_struct + ){ + TOMLDocument _doc = parseTOML(cast(string)(_document_struct.content)); + auto _doc_json = toJSON(_doc); + _make_and_meta_struct = contentJSONtoSiSUstruct!()(_make_and_meta_struct, _doc_json, _document_struct.filename); // struct from json + return _make_and_meta_struct; + } +} +static template docHeaderMakeAndMetaTupTomlExtractAndConvertToStruct() { + import + std.exception, + std.regex, + std.stdio, + std.traits, + std.typecons, + std.utf, + std.conv : to; + import + toml, + toml.json; + import + doc_reform.meta.conf_make_meta_structs, + doc_reform.meta.conf_make_meta_json, + doc_reform.meta.rgx; + mixin SiSUrgxInit; + mixin contentJSONtoSiSUstruct; + static auto rgx = Rgx(); + auto docHeaderMakeAndMetaTupTomlExtractAndConvertToStruct(CCm, Src)( + CCm _make_and_meta_struct, + Src header_src, + ) { + TOMLDocument _doc; + if (header_src.match(rgx.toml_header_meta_title)) { + debug (json) { + writeln(">>> document header is toml, convert to JSON"); + } + _doc = parseTOML(cast(string)(header_src)); + } + auto _doc_json = toJSON(_doc); + auto _header_and_make_and_meta_struct = contentJSONtoSiSUstruct!()(_make_and_meta_struct, _doc_json, "header"); + return _header_and_make_and_meta_struct; + } +} diff --git a/src/doc_reform/meta/defaults.d b/src/doc_reform/meta/defaults.d new file mode 100644 index 0000000..2f8c068 --- /dev/null +++ b/src/doc_reform/meta/defaults.d @@ -0,0 +1,241 @@ +/++ + default settings ++/ +module doc_reform.meta.defaults; +template SiSUrgxInitFlags() { + /+ regex flags +/ + static int[string] flags_type_init() { + int[string] flags_type_init = [ + "make_headings" : 0, + "header_make" : 0, + "header_meta" : 0, + "heading" : 0, + "biblio_section" : 0, + "glossary_section" : 0, + "blurb_section" : 0, + "para" : 0, + "blocks" : 0, // 0..2 generic + "code" : 0, // 0..2 + "poem" : 0, // 0..2 + "table" : 0, // 0..2 + "group" : 0, // 0..2 + "block" : 0, // 0..2 + "quote" : 0, // 0..2 + "verse_new" : 0, + "curly_code" : 0, + "curly_poem" : 0, + "curly_group" : 0, + "curly_block" : 0, + "curly_quote" : 0, + "curly_table" : 0, + "curly_table_special_markup" : 0, + "tic_code" : 0, + "tic_poem" : 0, + "tic_group" : 0, + "tic_block" : 0, + "tic_quote" : 0, + "tic_table" : 0, + "ocn_status" : 0, // 0 object_number; 1 no object_number; 2 no object_number & dummy headings + "ocn_status_off_for_multiple_objects" : 0, // 0 object_number; 1 no object_number; 2 no object_number & dummy headings + "book_index" : 0, + ]; + return flags_type_init; + } +} +template SiSUnode() { + static string[string] node_metadata_heading_str() { + auto _node = [ + "is" : "", + "ocn" : "", + "marked_up_lev" : "", + "segment_anchor_tag" : "", + "attrib" : "", + ]; + return _node; + } + static int[string] node_metadata_heading_int() { + auto _node = [ + "ocn" : 0, // decide whether to use or keep? + "ptr_doc_object" : 0, + "ptr_html_segnames" : 0, + "ptr_heading" : 0, + "heading_lev_markup" : 9, + "heading_lev_collapsed" : 9, + "parent_ocn" : 0, + "parent_lev_markup" : 9, + ]; + return _node; + } + static string[string] node_metadata_para_str() { + auto _node = [ + "is" : "", + "ocn" : "", + "attrib" : "", + ]; + return _node; + } + static int[string] node_metadata_para_int() { + auto _node = [ + "ocn" : 0, + "indent_base" : 0, + "indent_hang" : 0, + "bullet" : 0, // bool (0|1) + ]; + return _node; + } +} +template SiSUbiblio() { + // required: deemed_author (author || editor); year; fulltitle; + struct BibJsnStr { + static auto biblio_entry_tags_jsonstr() { + string x = `{ + "is" : "", + "sortby_deemed_author_year_title" : "", + "deemed_author" : "", + "author_raw" : "", + "author" : "", + "author_arr" : [ "" ], + "editor_raw" : "", + "editor" : "", + "editor_arr" : [ "" ], + "title" : "", + "subtitle" : "", + "fulltitle" : "", + "language" : "", + "trans" : "", + "src" : "", + "journal" : "", + "in" : "", + "volume" : "", + "edition" : "", + "year" : "", + "place" : "", + "publisher" : "", + "url" : "", + "pages" : "", + "note" : "", + "short_name" : "", + "id" : "" + }`; // is: book, article, magazine, newspaper, blog, other + return x; + } + } +} +template InternalMarkup() { + import std.array; + static struct InlineMarkup { + auto en_a_o = "【"; auto en_a_c = "】"; + auto en_b_o = "〖"; auto en_b_c = "〗"; + auto lnk_o = "┥"; auto lnk_c = "┝"; + auto url_o = "┤"; auto url_c = "├"; + auto mark_internal_site_lnk = "¤"; + auto nbsp = "░"; + auto br_line = "┘"; + auto br_nl = "┙"; + auto br_paragraph = "┚"; + auto br_obj = "break_obj"; + auto br_page_line = "┼"; + auto br_page = "┿"; + auto br_page_new = "╂"; + auto tc_s = "┊"; + auto tc_o = "┏"; + auto tc_c = "┚"; + auto tc_p = "┆"; + auto mono = "■"; + auto img = "☼"; + static string indent_by_spaces_provided(int indent, string _indent_spaces ="░░") { + _indent_spaces = replicate(_indent_spaces, indent); + return _indent_spaces; + } + static string repeat_character_by_number_provided(C,N)(C _character ="-", N number=10) { + _character = replicate(_character, number); + return _character; + } + } +} +template SiSUlanguageCodes() { + /+ language codes +/ + struct Lang { + static string[string][string] codes() { + auto _lang_codes = [ + "am": [ "c": "am", "n": "Amharic", "t": "Amharic", "xlp": "amharic" ], + "bg": [ "c": "bg", "n": "Bulgarian", "t": "Български (Bəlgarski)", "xlp": "bulgarian" ], + "bn": [ "c": "bn", "n": "Bengali", "t": "Bengali", "xlp": "bengali" ], + "br": [ "c": "br", "n": "Breton", "t": "Breton", "xlp": "breton" ], + "ca": [ "c": "ca", "n": "Catalan", "t": "catalan", "xlp": "catalan" ], + "cs": [ "c": "cs", "n": "Czech", "t": "česky", "xlp": "czech" ], + "cy": [ "c": "cy", "n": "Welsh", "t": "Welsh", "xlp": "welsh" ], + "da": [ "c": "da", "n": "Danish", "t": "dansk", "xlp": "danish" ], + "de": [ "c": "de", "n": "German", "t": "Deutsch", "xlp": "german" ], + "el": [ "c": "el", "n": "Greek", "t": "Ελληνικά (Ellinika)", "xlp": "greek" ], + "en": [ "c": "en", "n": "English", "t": "English", "xlp": "english" ], + "eo": [ "c": "eo", "n": "Esperanto", "t": "Esperanto", "xlp": "esperanto" ], + "es": [ "c": "es", "n": "Spanish", "t": "español", "xlp": "spanish" ], + "et": [ "c": "et", "n": "Estonian", "t": "Estonian", "xlp": "estonian" ], + "eu": [ "c": "eu", "n": "Basque", "t": "basque", "xlp": "basque" ], + "fi": [ "c": "fi", "n": "Finnish", "t": "suomi", "xlp": "finnish" ], + "fr": [ "c": "fr", "n": "French", "t": "français", "xlp": "french" ], + "ga": [ "c": "ga", "n": "Irish", "t": "Irish", "xlp": "irish" ], + "gl": [ "c": "gl", "n": "Galician", "t": "Galician", "xlp": "galician" ], + "he": [ "c": "he", "n": "Hebrew", "t": "Hebrew", "xlp": "hebrew" ], + "hi": [ "c": "hi", "n": "Hindi", "t": "Hindi", "xlp": "hindi" ], + "hr": [ "c": "hr", "n": "Croatian", "t": "Croatian", "xlp": "croatian" ], + "hy": [ "c": "hy", "n": "Armenian", "t": "Armenian", "xlp": "armenian" ], + "ia": [ "c": "ia", "n": "Interlingua", "t": "Interlingua", "xlp": "interlingua" ], + "is": [ "c": "is", "n": "Icelandic", "t": "Icelandic", "xlp": "icelandic" ], + "it": [ "c": "it", "n": "Italian", "t": "Italiano", "xlp": "italian" ], + "ja": [ "c": "ja", "n": "Japanese", "t": "日本語 (Nihongo)", "xlp": "japanese" ], + "ko": [ "c": "ko", "n": "Korean", "t": "Korean", "xlp": "korean" ], + "la": [ "c": "la", "n": "Latin", "t": "Latin", "xlp": "latin" ], + "lo": [ "c": "lo", "n": "Lao", "t": "Lao", "xlp": "lao" ], + "lt": [ "c": "lt", "n": "Lithuanian", "t": "Lithuanian", "xlp": "lithuanian" ], + "lv": [ "c": "lv", "n": "Latvian", "t": "Latvian", "xlp": "latvian" ], + "ml": [ "c": "ml", "n": "Malayalam", "t": "Malayalam", "xlp": "malayalam" ], + "mr": [ "c": "mr", "n": "Marathi", "t": "Marathi", "xlp": "marathi" ], + "nl": [ "c": "nl", "n": "Dutch", "t": "Nederlands", "xlp": "dutch" ], + "no": [ "c": "no", "n": "Norwegian", "t": "norsk", "xlp": "norsk" ], + "nn": [ "c": "nn", "n": "Norwegian Nynorsk", "t": "nynorsk", "xlp": "nynorsk" ], + "oc": [ "c": "oc", "n": "Occitan", "t": "Occitan", "xlp": "occitan" ], + "pl": [ "c": "pl", "n": "Polish", "t": "polski", "xlp": "polish" ], + "pt": [ "c": "pt", "n": "Portuguese", "t": "Português", "xlp": "portuges" ], + "pt_BR": [ "c": "pt_BR", "n": "Portuguese Brazil", "t": "Brazilian Português", "xlp": "brazilian" ], + "ro": [ "c": "ro", "n": "Romanian", "t": "română", "xlp": "romanian" ], + "ru": [ "c": "ru", "n": "Russian", "t": "Русский (Russkij)", "xlp": "russian" ], + "sa": [ "c": "sa", "n": "Sanskrit", "t": "Sanskrit", "xlp": "sanskrit" ], + "se": [ "c": "se", "n": "Sami", "t": "Samin", "xlp": "samin" ], + "sk": [ "c": "sk", "n": "Slovak", "t": "slovensky", "xlp": "slovak" ], + "sl": [ "c": "sl", "n": "Slovenian", "t": "Slovenian", "xlp": "slovenian" ], + "sq": [ "c": "sq", "n": "Albanian", "t": "Albanian", "xlp": "albanian" ], + "sr": [ "c": "sr", "n": "Serbian", "t": "Serbian", "xlp": "serbian" ], + "sv": [ "c": "sv", "n": "Swedish", "t": "svenska", "xlp": "swedish" ], + "ta": [ "c": "ta", "n": "Tamil", "t": "Tamil", "xlp": "tamil" ], + "te": [ "c": "te", "n": "Telugu", "t": "Telugu", "xlp": "telugu" ], + "th": [ "c": "th", "n": "Thai", "t": "Thai", "xlp": "thai" ], + "tk": [ "c": "tk", "n": "Turkmen", "t": "Turkmen", "xlp": "turkmen" ], + "tr": [ "c": "tr", "n": "Turkish", "t": "Türkçe", "xlp": "turkish" ], + "uk": [ "c": "uk", "n": "Ukranian", "t": "українська (ukrajins\"ka)", "xlp": "ukrainian" ], + "ur": [ "c": "ur", "n": "Urdu", "t": "Urdu", "xlp": "urdu" ], + "us": [ "c": "en", "n": "English (American)","t": "English", "xlp": "english" ], + "vi": [ "c": "vi", "n": "Vietnamese", "t": "Vietnamese", "xlp": "vietnamese" ], + "zh": [ "c": "zh", "n": "Chinese", "t": "中文", "xlp": "chinese" ], + "en": [ "c": "en", "n": "English", "t": "English", "xlp": "english" ], + "xx": [ "c": "xx", "n": "Default", "t": "English", "xlp": "english" ], + ]; + return _lang_codes; + } + static string[] code_arr_ptr() { + auto _lang_codes = ["am", "bg", "bn", "br", "ca", "cs", "cy", "da", "de", "el", "en", "eo", "es", "et", "eu", "fi", "fr", "ga", "gl", "he", "hi", "hr", "hy", "ia", "is", "it", "ja", "ko", "la", "lo", "lt", "lv", "ml", "mr", "nl", "no", "nn", "oc", "pl", "pt", "pt_BR", "ro", "ru", "sa", "se", "sk", "sl", "sq", "sr", "sv", "ta", "te", "th", "tk", "tr", "uk", "ur", "us", "vi", "zh", "en", "xx",]; + return _lang_codes; + } + static string[] code_arr() { + auto _lang_codes = ["am", "bg", "bn", "br", "ca", "cs", "cy", "da", "de", "el", "en", "eo", "es", "et", "eu", "fi", "fr", "ga", "gl", "he", "hi", "hr", "hy", "ia", "is", "it", "ja", "ko", "la", "lo", "lt", "lv", "ml", "mr", "nl", "no", "nn", "oc", "pl", "pt", "pt_BR", "ro", "ru", "sa", "se", "sk", "sl", "sq", "sr", "sv", "ta", "te", "th", "tk", "tr", "uk", "ur", "vi", "zh"]; + return _lang_codes; + } + static auto codes_() { + return "(" ~ join(code_arr,"|") ~ ")"; + } + static auto codes_regex() { + return regex(codes_); + } + } +} diff --git a/src/doc_reform/meta/doc_debugs.d b/src/doc_reform/meta/doc_debugs.d new file mode 100644 index 0000000..04dedb2 --- /dev/null +++ b/src/doc_reform/meta/doc_debugs.d @@ -0,0 +1,652 @@ +/++ + output debugs ++/ +module doc_reform.meta.doc_debugs; +template SiSUdebugs() { + import + doc_reform.meta.defaults, + doc_reform.meta.rgx; + import + std.algorithm, + std.array, + std.container, + std.exception, + std.json, + std.stdio, + std.file, + std.path, + std.range, + std.regex, + std.string, + std.traits, + std.typecons, + std.utf, + std.conv : to; + auto SiSUdebugs(S,T)( + const S contents, + T doc_matters, + ) { + mixin SiSUrgxInit; + mixin InternalMarkup; + static auto rgx = Rgx(); + auto markup = InlineMarkup(); + string key; + debug(parent) { + writefln( + "%s:%s", + __FILE__, + __LINE__, + ); + foreach (key; doc_matters.xml.keys_seq.seg) { + foreach (obj; contents[key]) { + if (obj.metainfo.is_of_part != "empty") { + if (obj.metainfo.is_a == "heading") { + writefln( + "%s node: %s heading: %s %s", + obj.object_number, + obj.node, + obj.heading_lev_markup, + obj.text, + ); + } + } + } + } + } + debug(dumpdoc) { + writefln( + "> %s:%s", + __FILE__, + __LINE__, + ); + } + debug(section_head) { + key="head"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(section_toc) { + key="toc_seg"; + out_toc(contents, key); + } + debug(section_toc_seg) { + key="toc_seg"; + out_toc(contents, key); + } + debug(section_toc_scroll) { + key="toc_scroll"; + out_toc(contents, key); + } + debug(segnames) { + writeln(__LINE__); + out_segnames(contents, doc_matters); + } + debug(section_body) { + key="body"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(toc_nav_dom) { + enum DomTags { none, open, close, close_and_open, open_still, } + foreach (sect; doc_matters.xml.keys_seq.seg) { + foreach (obj; contents[sect]) { + if (obj.metainfo.is_a == "heading") { + foreach_reverse (k; 0 .. 7) { + switch (obj.dom_structure_markedup_tags_status[k]) { + case DomTags.close : + writeln(markup.indent_by_spaces_provided(k), ""); + break; + case DomTags.close_and_open : + writeln(markup.indent_by_spaces_provided(k), ""); + writeln(markup.indent_by_spaces_provided(k), + "<", k, ">", obj.text, + " file: ", obj.segment_anchor_tag, ".xhtml#", obj.ocn); + break; + case DomTags.open : + writeln(markup.indent_by_spaces_provided(k), + "<", k, ">", obj.text, + " file: ", obj.segment_anchor_tag, ".xhtml#", obj.ocn); + break; + default : + break; + } + } + } + } + } + writeln("--------------------"); + foreach (sect; doc_matters.xml.keys_seq.seg) { + foreach (obj; contents[sect]) { + if (obj.metainfo.is_a == "heading") { + foreach_reverse (k; 0 .. 7) { + switch (obj.dom_structure_collapsed_tags_status[k]) { + case DomTags.close : + writeln(markup.indent_by_spaces_provided(k), ""); + break; + case DomTags.close_and_open : + writeln(markup.indent_by_spaces_provided(k), ""); + writeln(markup.indent_by_spaces_provided(k), + "<", k, ">", obj.text, + " file: ", obj.segment_anchor_tag, ".xhtml#", obj.ocn); + break; + case DomTags.open : + writeln(markup.indent_by_spaces_provided(k), + "<", k, ">", obj.text, + " file: ", obj.segment_anchor_tag, ".xhtml#", obj.ocn); + break; + default : + break; + } + } + } + } + } + } + debug(decendants) { + foreach (sect; doc_matters.xml.keys_seq.scroll) { + foreach (obj; contents[sect]) { + if (obj.metainfo.is_a == "heading") { + writeln(obj.metainfo.ocn, " .. ", obj.metainfo.last_decendant_ocn); + } + } + } + } + debug(section_endnotes) { + key="endnotes"; + out_endnotes(contents, key); + } + debug(section_endnotes_seg) { + key="endnotes"; + out_endnotes(contents, key); + } + debug(section_glossary) { + key="glossary"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(section_bibliography) { + key="bibliography"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(section_bookindex) { + key="bookindex_seg"; + out_bookindex(contents, key); + } + debug(section_bookindex_seg) { + key="bookindex_seg"; + out_bookindex(contents, key); + } + debug(section_bookindex_scroll) { + key="bookindex_scroll"; + out_bookindex(contents, key); + } + debug(blurb_section) { + key="blurb"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(objects) { + writefln( + "> %s:%s", + __FILE__, + __LINE__, + ); + foreach (obj; contents[key]) { + if (obj.metainfo.is_of_part != "empty") { + writefln( + "* [%s][%s] %s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + debug(headermakejson) { + writefln( + "%s\n%s\n%s", + "document header, metadata & make instructions:", + doc_matters.conf_make_meta.meta, + ptr_head_main, + ); + foreach (main_header; ptr_head_main) { + switch (main_header) { + case "make": + foreach (sub_header; ptr_head_sub_make) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + default: + break; + } + } + } + debug(headermetadatajson) { + writefln( + "%s\n%s\n%s", + "document header, metadata & make instructions:", + doc_matters.conf_make_meta.meta, + ptr_head_main, + ); + foreach (main_header; ptr_head_main) { + switch (main_header) { + case "creator": + foreach (sub_header; ptr_head_sub_creator) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full, + ); + } + } + break; + case "title": + foreach (sub_header; ptr_head_sub_title) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "rights": + foreach (sub_header; ptr_head_sub_rights) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "date": + foreach (sub_header; ptr_head_sub_date) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "original": + foreach (sub_header; ptr_head_sub_original) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "classify": + foreach (sub_header; ptr_head_sub_classify) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "identifier": + foreach (sub_header; ptr_head_sub_identifier) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "notes": + foreach (sub_header; ptr_head_sub_notes) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + case "publisher": + foreach (sub_header; ptr_head_sub_publisher) { + if (doc_matters.conf_make_meta.meta.title_full.to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.conf_make_meta.meta.title_full + ); + } + } + break; + default: + break; + } + } + } + debug(anchor) { + writefln( + "> %s:%s", + __FILE__, + __LINE__, + ); + foreach (k; doc_matters.xml.keys_seq.seg) { + foreach (obj; contents[k]) { + if (obj.metainfo.is_a == "heading") { + writefln( + "%s~ [%s] %s %s", + obj.marked_up_level, + obj.object_number, + obj.anchor_tags, + obj.text + ); + } + } + } + } + debug(heading) { // heading + foreach (k; doc_matters.xml.keys_seq.seg) { + foreach (o; contents[k]) { + if (o.metainfo.is_a == "heading") { + writefln( + "%s* %s\n (markup level: %s; collapsed level: %s)", + replicate(" ", o.heading_lev_markup), + strip(o.text), + o.heading_lev_markup, + o.heading_lev_collapsed, + ); + } + } + } + } + debug(headings) { + writefln( + "> %s:%s", + __FILE__, + __LINE__, + ); + foreach (k; doc_matters.xml.keys_seq.seg) { + foreach (obj; contents[k]) { + if (obj.metainfo.is_a == "heading") { + writefln( + "%s~ [%s] %s", + obj.marked_up_level, + obj.object_number, + obj.text + ); + } + } + } + } + debug(summary) { + string[string] check = [ + "last_object_number" : "NA [debug \"checkdoc\" not run]", + ]; + } + debug(checkdoc) { + if ((doc_matters.opt.action.debug_do)) { + debug(checkdoc) { + if (auto mfn=match(doc_matters.src.filename, rgx.src_fn)) { + if (doc_matters.opt.action.assertions) { + switch (mfn.captures[2]) { + // live manual: + case "live-manual.ssm": + assert(check["last_object_number"] == + "1019","last object_number should be: 1019 (check test, document is frequently updated)"); // ok + break; + // sisu_markup: + case "sisu_markup.sst": + assert(check["last_object_number"] == + "297","last object_number expected to be: 297 rather than " ~ check["last_object_number"]); // ok + // assert(check["last_object_number"] == "297","last object_number expected to be: 297 rather than " ~ check["last_object_number"]); + // notes for first divergance study sisu headings 247 250 + // sisu has issue with code that contains heading 1~ which results in no object_number! ?? + // sisu currently has incorrect last body object_number of 294! + // bug in sisu? attend + break; + // sisu-markup-samples: + case "accelerando.charles_stross.sst": + assert(check["last_object_number"] == + "2861","last object_number expected to be: 2861 rather than " ~ check["last_object_number"]); // ok + break; + case "alices_adventures_in_wonderland.lewis_carroll.sst": + assert(check["last_object_number"] == + "805","last object_number expected to be: 805 rather than " ~ check["last_object_number"]); // 808 + break; + case "autonomy_markup0.sst": + assert(check["last_object_number"] == + "77","last object_number expected to be: 77 rather than " ~ check["last_object_number"]); // ok endnotes + // assert(check["last_object_number"] == "78","last object_number expected to be: 78 rather than " ~ check["last_object_number"]); + break; + case "content.cory_doctorow.sst": + assert(check["last_object_number"] == + "953","last object_number expected to be: 953 rather than " ~ check["last_object_number"]); // 1007 way off, check object_number off switches + // assert(check["last_object_number"] == "953","last object_number expected to be: 953 rather than " ~ check["last_object_number"]); + break; + case "democratizing_innovation.eric_von_hippel.sst": + // fixed ERROR! range violation, broken check! endnotes, bookindex, biblio + // error in bookindex ... (ch1; ch6; ch8 ) + assert(check["last_object_number"] == + "905","last object_number expected to be: 905 rather than " ~ check["last_object_number"]); // 911 + break; + case "down_and_out_in_the_magic_kingdom.cory_doctorow.sst": + assert(check["last_object_number"] == + "1417","last object_number expected to be: 1417 rather than " ~ check["last_object_number"]); // 1455 check object_number off switches + break; + case "for_the_win.cory_doctorow.sst": + assert(check["last_object_number"] == + "3510","last object_number expected to be: 3510 rather than " ~ check["last_object_number"]); // 3569 check object_number off switches + break; + case "free_as_in_freedom_2.richard_stallman_and_the_free_software_revolution.sam_williams.richard_stallman.sst": + assert(check["last_object_number"] == + "1082","last object_number expected to be: 1082 rather than " ~ check["last_object_number"]); // check 1079 too few + break; + case "free_culture.lawrence_lessig.sst": + assert(check["last_object_number"] == + "1330","last object_number expected to be: 1330 rather than " ~ check["last_object_number"]); // 1312 + // fixed ERROR! range violation, broken check! + // error in bookindex ... sections piracy (ch1) & property (ch10 market concentration) fixed + break; + case "free_for_all.peter_wayner.sst": // endnotes, bookindex, biblio + assert(check["last_object_number"] == + "1559","last object_number expected to be: 1559 rather than " ~ check["last_object_number"]); // 1560, check object_number off switches, has endnotes so 2 too many + // assert(check["last_object_number"] == "1559","last object_number expected to be: 1559 rather than " ~ check["last_object_number"]); + break; + case "gpl2.fsf.sst": + assert(check["last_object_number"] == + "65","last object_number expected to be: 65 rather than " ~ check["last_object_number"]); // ok endnotes? check + // assert(check["last_object_number"] == "66","last object_number expected to be: 66 rather than " ~ check["last_object_number"]); + break; + case "gpl3.fsf.sst": + assert(check["last_object_number"] == + "123","last object_number expected to be: 123 rather than " ~ check["last_object_number"]); // ok + break; + case "gullivers_travels.jonathan_swift.sst": + assert(check["last_object_number"] == + "668","last object_number expected to be: 668 rather than " ~ check["last_object_number"]); // 674 + break; + case "little_brother.cory_doctorow.sst": + assert(check["last_object_number"] == + "3130","last object_number expected to be: 3130 rather than " ~ check["last_object_number"]); // 3204, check object_number off switches + break; + case "the_cathedral_and_the_bazaar.eric_s_raymond.sst": + assert(check["last_object_number"] == + "258","last object_number expected to be: 258 rather than " ~ check["last_object_number"]); // ok + break; + case "the_public_domain.james_boyle.sst": + assert(check["last_object_number"] == + "970","last object_number expected to be: 970 rather than " ~ check["last_object_number"]); // 978 + break; + case "the_wealth_of_networks.yochai_benkler.sst": // endnotes, bookindex + assert(check["last_object_number"] == + "829","last object_number expected to be: 829 rather than " ~ check["last_object_number"]); // ok + // assert(check["last_object_number"] == "832","last object_number expected to be: 832 rather than " ~ check["last_object_number"]); + // has endnotes and bookindex, issue with sisu.rb + break; + case "through_the_looking_glass.lewis_carroll.sst": + assert(check["last_object_number"] == + "949","last object_number expected to be: 949 rather than " ~ check["last_object_number"]); // 955 + break; + case "two_bits.christopher_kelty.sst": // endnotes, bookindex, biblio + assert(check["last_object_number"] == + "1190","last object_number expected to be: 1190 rather than " ~ check["last_object_number"]); // 1191 + // assert(check["last_object_number"] == "1193","last object_number expected to be: 1193 rather than " ~ check["last_object_number"]); // 1191 ok? + // has endnotes and bookindex, issue with sisu.rb + break; + // fixed ERROR! range violation! + // error in bookindex ... (ch3 the movement) + case "un_contracts_international_sale_of_goods_convention_1980.sst": + assert(check["last_object_number"] == + "377","last object_number expected to be: 377 rather than " ~ check["last_object_number"]); // ok + break; + case "viral_spiral.david_bollier.sst": // endnotes, bookindex + assert(check["last_object_number"] == + "1078","last object_number expected to be: 1078 rather than " ~ check["last_object_number"]); // 1100 + // fixed ERROR! range violation! + // error in bookindex ... (ch7 ... building the cc machine, an extra semi colon) + break; + default: + writeln(doc_matters.src.filename); + break; + } + } + } + } + debug(checkdoc) { + void out_segnames(S,T)( + const S contents, + T doc_matters, + ) { + foreach (key; doc_matters.xml.keys_seq.seg) { + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + if (obj.heading_lev_markup == 4) { + writeln(obj.ptr_html_segnames, ". (", doc_matters.xml.segnames[obj.ptr_html_segnames], ") -> ", obj.text); + } + } + } + } + } + } + debug(checkdoc) { + void out_toc(S)( + const S contents, + string key, + ) { + if (contents[key].length > 1) { + string indent_spaces; + foreach (obj; contents[key]) { + indent_spaces=markup.indent_by_spaces_provided(obj.indent_hang); + writefln( + "%s%s", + indent_spaces, + obj.text + ); + } + } + } + } + debug(checkdoc) { + void out_endnotes(S)( + const S contents, + string key, + ) { + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s]\n%s", + obj.metainfo.is_a, + obj.text + ); + } + } + } + } + debug(checkdoc) { + void out_bookindex(S)( + const S contents, + string key, + ) { + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.object_number, + obj.metainfo.is_a, + obj.text + ); + } + } + } + } + } + } + } +} diff --git a/src/doc_reform/meta/metadoc.d b/src/doc_reform/meta/metadoc.d new file mode 100644 index 0000000..7c0e90d --- /dev/null +++ b/src/doc_reform/meta/metadoc.d @@ -0,0 +1,177 @@ +module doc_reform.meta.metadoc; +template SiSUabstraction() { + import + std.getopt, + std.file, + std.path, + std.process; + import + doc_reform.meta, + doc_reform.meta.metadoc_summary, + doc_reform.meta.metadoc_from_src, + doc_reform.meta.conf_make_meta_structs, + doc_reform.meta.conf_make_meta_toml, + doc_reform.meta.conf_make_meta_json, + doc_reform.meta.defaults, + doc_reform.meta.doc_debugs, + doc_reform.meta.rgx, + doc_reform.source.paths_source, + doc_reform.source.read_config_files, + doc_reform.source.read_source_files, + doc_reform.output.hub; + mixin SiSUrgxInit; + mixin contentJSONtoSiSUstruct; + mixin SiSUbiblio; + mixin SiSUrgxInitFlags; + mixin outputHub; + enum headBody { header, body_content, insert_file_list, image_list } + enum makeMeta { make, meta } + enum docAbst { doc_abstraction, section_keys, segnames, segnames_0_4, images } + static auto rgx = Rgx(); + auto SiSUabstraction(E,O,M)( + E _env, + O _opt_action, + M _manifest + ){ + auto _config_document_struct = readConfigDoc!()(_manifest, _env); // document config file + auto _config_local_site_struct = readConfigSite!()(_manifest, _env); // local site config + ConfCompositePlus _make_and_meta_struct; + _make_and_meta_struct = configParseTOMLreturnSiSUstruct!()(_make_and_meta_struct, _config_document_struct); + _make_and_meta_struct = configParseTOMLreturnSiSUstruct!()(_make_and_meta_struct, _config_local_site_struct); + /+ ↓ read file (filename with path) +/ + /+ ↓ file tuple of header and content +/ + debug(steps) { + writeln("step1 commence → (get document header & body & insert file list & if needed image list)" + ); + } + auto _header_body_insertfilelist_imagelist + = SiSUrawMarkupContent!()(_opt_action, _manifest.src.path_and_fn); + static assert(!isTypeTuple!(_header_body_insertfilelist_imagelist)); + static assert(_header_body_insertfilelist_imagelist.length==4); + debug(steps) { + writeln("- step1 complete"); + } + debug(header_and_body) { + writeln(header); + writeln(_header_body_insertfilelist_imagelist.length); + writeln(_header_body_insertfilelist_imagelist.length[headBody.body_content][0]); + } + /+ ↓ split header into make and meta +/ + debug(steps) { + writeln("step2 commence → (read document header - toml, return struct)"); + } + _make_and_meta_struct = + docHeaderMakeAndMetaTupTomlExtractAndConvertToStruct!()( + _make_and_meta_struct, + _header_body_insertfilelist_imagelist[headBody.header] + ); + debug(steps) { + writeln("- step2 complete"); + } + /+ ↓ document abstraction: process document, return abstraction as tuple +/ + debug(steps) { + writeln("step3 commence → (document abstraction (da); da keys; segnames; doc_matters)"); + } + auto da = SiSUdocAbstraction!()( + _header_body_insertfilelist_imagelist[headBody.body_content], + _make_and_meta_struct, + _opt_action, + _manifest, + true, + ); + static assert(!isTypeTuple!(da)); + static assert(da.length==5); + auto doc_abstraction = da[docAbst.doc_abstraction]; /+ head ~ toc ~ body ~ endnotes_seg ~ glossary ~ bibliography ~ bookindex ~ blurb; +/ + auto _document_section_keys_sequenced = da[docAbst.section_keys]; + string[] _doc_html_segnames = da[docAbst.segnames]; + string[] _doc_epub_segnames_0_4 = da[docAbst.segnames_0_4]; + auto _images = da[docAbst.images]; + debug(steps) { + writeln("- step3 complete"); + } + debug(steps) { + writeln("step4 commence → (doc_matters)"); + } + struct DocumentMatters { + auto conf_make_meta() { // TODO meld with all make instructions + return _make_and_meta_struct; + } + auto env() { + struct Env_ { + auto pwd() { + return _manifest.env.pwd; + } + auto home() { + return _manifest.env.home; + } + } + return Env_(); + } + auto opt() { + struct Opt_ { + auto action() { + /+ getopt options, commandline instructions, raw + - processing instructions --epub --html etc. + - command line config instructions --output-path + +/ + return _opt_action; + } + } + return Opt_(); + } + auto src() { + return _manifest.src; + } + auto src_path_info() { + return SiSUpathsSRC!()(_manifest.env.pwd, _manifest.src.file_with_absolute_path); // would like (to have and use) relative path + } + auto srcs() { + struct SRC_ { + auto file_insert_list() { + return _header_body_insertfilelist_imagelist[headBody.insert_file_list]; + } + auto image_list() { + return _images; + } + } + return SRC_(); + } + auto pod() { + return _manifest.pod; + } + auto sqlite() { + struct SQLite_ { + string filename() { + return _opt_action.sqlite_filename; + } + } + return SQLite_(); + } + auto xml() { + struct XML_ { + auto keys_seq() { + /+ contains .seg & .scroll sequences +/ + return _document_section_keys_sequenced; + } + string[] segnames() { + return _doc_html_segnames; + } + string[] segnames_lv_0_to_4() { + return _doc_epub_segnames_0_4; + } + } + return XML_(); + } + auto output_path() { + return _manifest.output.path; + } + } + auto doc_matters = DocumentMatters(); + debug(steps) { + writeln("- step4 complete"); + } + auto t = tuple(doc_abstraction, doc_matters); + static assert(t.length==2); + return t; + } +} diff --git a/src/doc_reform/meta/metadoc_from_src.d b/src/doc_reform/meta/metadoc_from_src.d new file mode 100644 index 0000000..c4a12bb --- /dev/null +++ b/src/doc_reform/meta/metadoc_from_src.d @@ -0,0 +1,6322 @@ +/++ + document abstraction: + abstraction of sisu markup for downstream processing + metadoc_from_src.d ++/ +module doc_reform.meta.metadoc_from_src; +template SiSUdocAbstraction() { + /+ ↓ abstraction imports +/ + import doc_reform.meta; + import + std.algorithm, + std.container, + std.file, + std.json, + std.path; + import + doc_reform.meta.defaults, + doc_reform.meta.object_setter, + doc_reform.meta.rgx; + /+ ↓ abstraction mixins +/ + mixin ObjectSetter; + mixin InternalMarkup; + mixin SiSUrgxInit; + /+ ↓ abstraction struct init +/ + /+ initialize +/ + ObjGenericComposite[][string] the_table_of_contents_section; + ObjGenericComposite[] the_document_head_section, the_document_body_section, the_bibliography_section, the_glossary_section, the_blurb_section; + ObjGenericComposite[] the_dom_tail_section; + string[string] an_object, processing; + string an_object_key; + string[] anchor_tags; + string anchor_tag_; + string segment_anchor_tag_that_object_belongs_to; + string segment_anchor_tag_that_object_belongs_to_uri; + /+ enum +/ + enum State { off, on } + enum TriState { off, on, closing } + enum DocStructMarkupHeading { + h_sect_A, + h_sect_B, + h_sect_C, + h_sect_D, + h_text_1, + h_text_2, + h_text_3, + h_text_4, + h_text_5, // extra level, drop + content_non_header + } // header section A-D; header text 1-4 + enum OCNstatus { on, off, bkidx, closing, reset, } + enum OCNtype { ocn, non, bkidx, } + /+ biblio variables +/ + string biblio_tag_name, biblio_tag_entry, st; + string[] biblio_arr_json; + string biblio_entry_str_json; + JSONValue[] bib_arr_json; + int bib_entry; + /+ counters +/ + int cntr, previous_count, previous_length; + bool reset_note_numbers=true; + int[string] line_occur; + int html_segnames_ptr=0; + int html_segnames_ptr_cntr=0; + int verse_line, heading_ptr; + /+ paragraph attributes +/ + int[string] indent; + bool bullet = true; + string content_non_header = "8"; + static auto obj_im = ObjInlineMarkup(); + static auto obj_att = ObjAttributes(); + /+ ocn +/ + struct OCNset { + int digit; + int on; + int off; + int bkidx; + int type; + } + OCNset obj_cite_digits; + int obj_cite_digit_, obj_cite_digit_off, obj_cite_digit_bkidx, obj_cite_digit_type; + auto object_citation_number = OCNemitter(); + int[] dom_structure_markedup_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_structure_markedup_tags_status_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_structure_collapsed_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_structure_collapsed_tags_status_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + enum DomTags { none, open, close, close_and_open, open_still, } + pure auto obj_heading_ancestors(O)( + O obj, + string[] lv_ancestors_txt, + ) { + switch (obj.metainfo.heading_lev_markup) { + case 0: + lv_ancestors_txt[0] = obj.text.to!string; + foreach(k; 1..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 1: + lv_ancestors_txt[1] = obj.text.to!string; + foreach(k; 2..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 2: + lv_ancestors_txt[2] = obj.text.to!string; + foreach(k; 3..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 3: + lv_ancestors_txt[3] = obj.text.to!string; + foreach(k; 4..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 4: + lv_ancestors_txt[4] = obj.text.to!string; + foreach(k; 5..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 5: + lv_ancestors_txt[5] = obj.text.to!string; + foreach(k; 6..8) { + lv_ancestors_txt[k] = ""; + } + goto default; + case 6: + lv_ancestors_txt[6] = obj.text.to!string; + lv_ancestors_txt[7] = ""; + goto default; + case 7: + lv_ancestors_txt[7] = obj.text.to!string; + goto default; + default: + obj.tags.heading_ancestors_text = lv_ancestors_txt.dup; + } + return obj; + } + pure auto obj_dom_structure_set_markup_tags(O)( + O obj, + int[] dom, + int lev + ) { + foreach (i; 0 .. 8) { + if (i < lev) { + if (dom[i] == DomTags.open + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.open_still; + } else if (dom[i] == DomTags.close) { + dom[i] = DomTags.none; + } + } else if (i == lev) { + if (lev == 0 + && dom[i] == DomTags.open_still + ) { + dom[i] = DomTags.close; + } else if (dom[i] == DomTags.open + || dom[i] == DomTags.open_still + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.close_and_open; + } else { + dom[i] = DomTags.open; + } + } else if (i > lev) { + if (dom[i] == DomTags.close) { + dom[i] = DomTags.none; + } else if (dom[i] == DomTags.open + || dom[i] == DomTags.open_still + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.close; + } + } + } + debug(dom_magic_numbers) { + writeln("marked up: ", lev, ": ", dom); + } + obj.metainfo.dom_structure_markedup_tags_status = dom.dup; + return obj; + } + pure auto obj_dom_set_collapsed_tags(O)( + O obj, + int[] dom, + int lev + ) { + foreach (i; 0 .. 8) { + if (i < lev) { + if (dom[i] == DomTags.open + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.open_still; + } else if (dom[i] == DomTags.close) { + dom[i] = DomTags.none; + } + } else if (i == lev) { + if (lev == 0 + && dom[i] == DomTags.open_still + ) { + dom[i] = DomTags.close; + } else if (dom[i] == DomTags.open + || dom[i] == DomTags.open_still + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.close_and_open; + } else { + dom[i] = DomTags.open; + } + } else if (i > lev) { + if (dom[i] == DomTags.close) { + dom[i] = DomTags.none; + } else if (dom[i] == DomTags.open + || dom[i] == DomTags.open_still + || dom[i] == DomTags.close_and_open + ) { + dom[i] = DomTags.close; + } + } + } + debug(dom_magic_numbers) { + writeln("collapsed: ", lev, ": ", dom); + } + obj.metainfo.dom_structure_collapsed_tags_status = dom.dup; + return obj; + } + static auto ocn_emit(int ocn_status_flag) { + return object_citation_number.ocn_emitter(ocn_status_flag); + } + /+ book index variables +/ + string book_idx_tmp; + string[][string][string] bookindex_unordered_hashes; + /+ node +/ + ObjGenericComposite comp_obj_heading, comp_obj_location, comp_obj_block, comp_obj_code, comp_obj_poem_ocn, comp_obj_comment; + auto node_construct = NodeStructureMetadata(); + enum sObj { content, anchor_tags, notes_reg, notes_star, links, image_no_dimensions } + /+ ↓ abstract marked up document +/ + auto SiSUdocAbstraction(Src,CMM,Opt,Mfst)( + Src markup_sourcefile_content, + CMM conf_make_meta, + Opt opt_action, + Mfst manifest_matter, + bool _new_doc + ) { + static auto rgx = Rgx(); + debug(asserts) { + static assert(is(typeof(markup_sourcefile_content) == char[][])); + } + /+ ↓ abstraction init +/ + scope(success) { + } + scope(failure) { + } + scope(exit) { + destroy(the_document_head_section); + destroy(the_table_of_contents_section); + destroy(the_document_body_section); + destroy(the_bibliography_section); + destroy(an_object); + destroy(processing); + destroy(biblio_arr_json); + previous_length=0; + reset_note_numbers=true; + } + mixin SiSUrgxInitFlags; + mixin SiSUnode; + auto node_para_int_ = node_metadata_para_int; + auto node_para_str_ = node_metadata_para_str; + ObjGenericComposite comp_obj_heading_, comp_obj_para, comp_obj_toc; + line_occur = [ + "heading" : 0, + "para" : 0, + "glossary" : 0, + "blurb" : 0, + ]; + auto obj_type_status = flags_type_init; + string[string] object_number_poem = [ + "start" : "", + "end" : "" + ]; + string[] lv_ancestors_txt = [ "", "", "", "", "", "", "", "", ]; + int[string] lv = [ + "lv" : State.off, + "h0" : State.off, + "h1" : State.off, + "h2" : State.off, + "h3" : State.off, + "h4" : State.off, + "h5" : State.off, + "h6" : State.off, + "h7" : State.off, + "lev_int_collapsed" : 0, + ]; + int[string] collapsed_lev = [ + "h0" : State.off, + "h1" : State.off, + "h2" : State.off, + "h3" : State.off, + "h4" : State.off, + "h5" : State.off, + "h6" : State.off, + "h7" : State.off + ]; + string[string] heading_match_str = [ + "h_A": "^(none)", + "h_B": "^(none)", + "h_C": "^(none)", + "h_D": "^(none)", + "h_1": "^(none)", + "h_2": "^(none)", + "h_3": "^(none)", + "h_4": "^(none)" + ]; + auto heading_match_rgx = [ + "h_A": regex(r"^(none)"), + "h_B": regex(r"^(none)"), + "h_C": regex(r"^(none)"), + "h_D": regex(r"^(none)"), + "h_1": regex(r"^(none)"), + "h_2": regex(r"^(none)"), + "h_3": regex(r"^(none)"), + "h_4": regex(r"^(none)") + ]; + string _anchor_tag; + string toc_txt_; + an_object["glossary_nugget"] = ""; + an_object["blurb_nugget"] = ""; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "frontmatter"; + comp_obj_heading_.metainfo.is_of_section = "toc"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Table of Contents"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "toc"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.ptr.html_segnames = html_segnames_ptr; + comp_obj_heading_.tags.anchor_tags = ["toc"]; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = [ 1, 1, 1, 0, 0, 0, 0, 0]; + auto toc_head = comp_obj_heading_; + html_segnames_ptr_cntr++; + the_table_of_contents_section = [ + "seg": [toc_head], + "scroll": [toc_head], + ]; + static auto mkup = InlineMarkup(); + static auto munge = ObjInlineMarkupMunge(); + auto note_section = NotesSection(); + auto bookindex_extract_hash = BookIndexNuggetHash(); + string[][string] lev4_subtoc; + string[] html_segnames=["toc"]; + int cnt1 = 1; int cnt2 = 1; int cnt3 = 1; + /+ abstraction init ↑ +/ + enum Substitute { match, markup, } + debug (substitutions) { + writeln(__LINE__, ":", __FILE__, ": DEBUG substitutions:"); + if (!(conf_make_meta.make.headings.empty)) { + writeln(conf_make_meta.make.headings); + } + if (conf_make_meta.make.substitute) { + foreach(substitution_pair; conf_make_meta.make.substitute) { + writeln("regex to match: ", substitution_pair[Substitute.match]); + writeln("substitution to make: ", substitution_pair[Substitute.markup]); + } + } + if (conf_make_meta.make.bold) { + writeln("regex to match: ", conf_make_meta.make.bold[Substitute.match]); + writeln("substitution to make: ", conf_make_meta.make.bold[Substitute.markup]); + } + if (conf_make_meta.make.emphasis) { + writeln("regex to match: ", conf_make_meta.make.emphasis[Substitute.match]); + writeln("substitution to make: ", conf_make_meta.make.emphasis[Substitute.markup]); + } + if (conf_make_meta.make.italics) { + writeln("regex to match: ", conf_make_meta.make.italics[Substitute.match]); + writeln("substitution to make: ", conf_make_meta.make.italics[Substitute.markup]); + } + } + /+ ↓ ↻ loop markup document/text line by line +/ + srcDocLoop: + foreach (line; markup_sourcefile_content) { + // "line" variable can be empty but should never be null + /+ ↓ markup document/text line by line +/ + /+ scope +/ + scope(exit) { + } + scope(failure) { + stderr.writefln( + "%s\n%s\n%s:%s failed here:\n line: %s", + __MODULE__, __FUNCTION__, + __FILE__, __LINE__, + line, + ); + } + line = (line).replaceAll(rgx.true_dollar, "$$$$"); + /+ dollar represented as $$ needed to stop submatching on $ + (substitutions using ${identifiers} must take into account (i.e. happen earlier)) + +/ + debug(source) { + writeln(line); + } + debug(srclines) { + if (!line.empty) { + writefln( + "* %s", + line + ); + } + } + if (!line.empty) { + obj_type_status = _check_ocn_status_(line, obj_type_status); + } + if (obj_type_status["code"] == TriState.on) { + /+ block object: code +/ + _code_block_(line, an_object, obj_type_status); + continue; + } else if (!matchFirst(line, rgx.skip_from_regular_parse)) { + /+ object other than "code block" object + (includes regular text paragraph, headings & blocks other than code) +/ + /+ heading, glossary, blurb, poem, group, block, quote, table +/ + if (line.matchFirst(rgx.heading_biblio) + || (obj_type_status["biblio_section"] == State.on + && (!(line.matchFirst(rgx.heading_blurb_glossary))) + && (!(line.matchFirst(rgx.heading))) + && (!(line.matchFirst(rgx.comment))))) { + /+ within section (block object): biblio +/ + obj_type_status["glossary_section"] = State.off; + obj_type_status["biblio_section"] = State.on; + obj_type_status["blurb_section"] = State.off; + if (opt_action.backmatter && opt_action.section_biblio) { + _biblio_block_(line, obj_type_status, bib_entry, biblio_entry_str_json, biblio_arr_json); + debug(bibliobuild) { + writeln("- ", biblio_entry_str_json); + writeln("-> ", biblio_arr_json.length); + } + } + continue; + } else if (line.matchFirst(rgx.heading_glossary) + || (obj_type_status["glossary_section"] == State.on + && (!(line.matchFirst(rgx.heading_biblio_blurb))) + && (!(line.matchFirst(rgx.heading))) + && (!(line.matchFirst(rgx.comment))))) { + /+ within section (block object): glossary +/ + debug(glossary) { + writeln(__LINE__); + writeln(line); + } + obj_type_status["glossary_section"] = State.on; + obj_type_status["biblio_section"] = State.off; + obj_type_status["blurb_section"] = State.off; + if (opt_action.backmatter && opt_action.section_glossary) { + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + obj_type_status["para"] = State.on; + line_occur["para"] = State.off; + an_object_key="glossary_nugget"; // + if (line.matchFirst(rgx.heading_glossary)) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "glossary"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Glossary"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "_part_glossary"; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = [ 1, 1, 0, 0, 0, 0, 0, 0]; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = [ 1, 1, 0, 0, 0, 0, 0, 0]; + the_glossary_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "glossary"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Glossary"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "glossary"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 2; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = [ 1, 1, 1, 0, 0, 0, 0, 0]; + comp_obj_heading_.tags.anchor_tags = ["glossary"]; + the_glossary_section ~= comp_obj_heading_; + } else { + _para_match_(line, an_object, an_object_key, indent, bullet, obj_type_status, line_occur); + comp_obj_para = comp_obj_para.init; + comp_obj_para.metainfo.is_of_part = "backmatter"; + comp_obj_para.metainfo.is_of_section = "glossary"; + comp_obj_para.metainfo.is_of_type = "para"; + comp_obj_para.metainfo.is_a = "glossary"; + comp_obj_para.text = line.to!string.strip; + comp_obj_para.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_para.attrib.indent_hang = indent["hang_position"]; + comp_obj_para.attrib.indent_base = indent["base_position"]; + comp_obj_para.attrib.bullet = bullet; + the_glossary_section ~= comp_obj_para; + } + obj_type_status["ocn_status"] = OCNstatus.on; + } + continue; + } else if (line.matchFirst(rgx.heading_blurb) + || (obj_type_status["blurb_section"] == State.on + && (!(line.matchFirst(rgx.heading_biblio_glossary))) + && (!(line.matchFirst(rgx.heading))) + && (!(line.matchFirst(rgx.comment))))) { + /+ within section (block object): blurb +/ + debug(blurb) { + writeln(__LINE__); + writeln(line); + } + obj_type_status["glossary_section"] = State.off; + obj_type_status["biblio_section"] = State.off; + obj_type_status["blurb_section"] = State.on; + if (opt_action.backmatter && opt_action.section_blurb) { + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + if (auto m = line.matchFirst(rgx.para_indent)) { + debug(paraindent) { + writeln(line); + } + indent["hang_position"] = (m.captures[1]).to!int; + indent["base_position"] = 0; + } else if (line.matchFirst(rgx.para_bullet)) { + debug(parabullet) { + writeln(line); + } + bullet = true; + } else if (auto m = line.matchFirst(rgx.para_indent_hang)) { + debug(paraindenthang) { + writeln(line); + } + indent=[ + "hang_position" : (m.captures[1]).to!int, + "base_position" : (m.captures[2]).to!int, + ]; + } else if (auto m = line.matchFirst(rgx.para_bullet_indent)) { + debug(parabulletindent) { + writeln(line); + } + indent=[ + "hang_position" : (m.captures[1]).to!int, + "base_position" : 0, + ]; + bullet = true; + } + obj_type_status["para"] = State.on; + line_occur["para"] = State.off; + an_object_key="blurb_nugget"; + if (line.matchFirst(rgx.heading_blurb)) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "blurb"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Blurb"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "_part_blurb"; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = [ 1, 1, 0, 0, 0, 0, 0, 0]; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = [ 1, 1, 0, 0, 0, 0, 0, 0]; + the_blurb_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "blurb"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Blurb"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "blurb"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 2; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.tags.anchor_tags = ["blurb"]; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = [ 1, 1, 1, 0, 0, 0, 0, 0]; + the_blurb_section ~= comp_obj_heading_; + } else if (line.matchFirst(rgx.heading) + && (opt_action.backmatter && opt_action.section_blurb)) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "blurb"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = line.to!string; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "blurb"; + comp_obj_heading_.metainfo.heading_lev_markup = an_object["lev_markup_number"].to!int; // make int, remove need to conv + comp_obj_heading_.metainfo.heading_lev_collapsed = an_object["lev_collapsed_number"].to!int; // make int, remove need to conv + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_blurb_section ~= comp_obj_heading_; + } else { + _para_match_(line, an_object, an_object_key, indent, bullet, obj_type_status, line_occur); + comp_obj_para = comp_obj_para.init; + comp_obj_para.metainfo.is_of_part = "backmatter"; + comp_obj_para.metainfo.is_of_section = "blurb"; + comp_obj_para.metainfo.is_of_type = "para"; + comp_obj_para.metainfo.is_a = "blurb"; + comp_obj_para.text = munge.url_links(line.to!string.strip).replaceFirst(rgx.para_attribs, ""); + comp_obj_para.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_para.attrib.indent_hang = indent["hang_position"]; + comp_obj_para.attrib.indent_base = indent["base_position"]; + comp_obj_para.has.inline_links = true; + comp_obj_para.attrib.bullet = bullet; + the_blurb_section ~= comp_obj_para; + } + obj_type_status["ocn_status"] = OCNstatus.on; + } + continue; + } else if (obj_type_status["quote"] == TriState.on) { + /+ within block object: quote +/ + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + line = _doc_header_and_make_substitutions_fontface_(line, conf_make_meta); + _quote_block_(line, an_object, obj_type_status); + continue; + /+ within block object: group +/ + } else if (obj_type_status["group"] == TriState.on) { + /+ within block object: group +/ + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + line = _doc_header_and_make_substitutions_fontface_(line, conf_make_meta); + line = (line) + .replaceAll(rgx.para_delimiter, mkup.br_paragraph ~ "$1"); + _group_block_(line, an_object, obj_type_status); + continue; + } else if (obj_type_status["block"] == TriState.on) { + /+ within block object: block +/ + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + line = _doc_header_and_make_substitutions_fontface_(line, conf_make_meta); + if (auto m = line.match(rgx.spaces_line_start)) { + line = (line) + .replaceAll(rgx.spaces_line_start, (m.captures[1]).translate([ ' ' : mkup.nbsp ])); + } + if (auto m = line.match(rgx.spaces_multiple)) { + line = (line) + .replaceAll(rgx.spaces_multiple, (m.captures[1]).translate([ ' ' : mkup.nbsp ])); + } + _block_block_(line, an_object, obj_type_status); + continue; + } else if (obj_type_status["poem"] == TriState.on) { + /+ within block object: poem +/ + _poem_block_(line, an_object, obj_type_status, cntr, object_number_poem, conf_make_meta); + continue; + } else if (obj_type_status["table"] == TriState.on) { + /+ within block object: table +/ + _table_block_(line, an_object, obj_type_status, conf_make_meta); + continue; + } else { + /+ not within a block group +/ + assert( + (obj_type_status["blocks"] == TriState.off) + || (obj_type_status["blocks"] == TriState.closing), + "block status: none or closed" + ); + assertions_flag_types_block_status_none_or_closed(obj_type_status); + if (line.matchFirst(rgx.block_open)) { + if (line.matchFirst(rgx.block_poem_open)) { + /+ poem to verse exceptions! +/ + object_reset(an_object); + processing.remove("verse"); + object_number_poem["start"] = obj_cite_digits.on.to!string; + } + _start_block_(line, obj_type_status, object_number_poem); + continue; + } else if (!line.empty) { + /+ line not empty +/ + /+ non blocks (headings, paragraphs) & closed blocks +/ + assert( + !line.empty, + "line tested, line not empty surely:\n \"" ~ line ~ "\"" + ); + assert( + (obj_type_status["blocks"] == TriState.off) + || (obj_type_status["blocks"] == TriState.closing), + "code block status: none or closed" + ); + if (obj_type_status["blocks"] == TriState.closing) { + debug(check) { // block + writeln(__LINE__); + writeln(line); + } + assert( + line.matchFirst(rgx.book_index) + || line.matchFirst(rgx.book_index_open) + || obj_type_status["book_index"] == State.on, + "\nblocks closed, unless followed by book index, non-matching line:\n \"" + ~ line ~ "\"" + ); + } + if (line.matchFirst(rgx.book_index) + || line.matchFirst(rgx.book_index_open) + || obj_type_status["book_index"] == State.on ) { + /+ book_index +/ + _book_index_(line, book_idx_tmp, an_object, obj_type_status, opt_action); + } else { + /+ not book_index +/ + an_object_key="body_nugget"; + if (auto m = matchFirst(line, rgx.comment)) { + /+ matched comment +/ + debug(comment) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; + comp_obj_comment = comp_obj_comment.init; + comp_obj_comment.metainfo.is_of_part = "comment"; // breaks flow + comp_obj_comment.metainfo.is_of_section = "comment"; // breaks flow + comp_obj_comment.metainfo.is_of_type = "comment"; + comp_obj_comment.metainfo.is_a = "comment"; + comp_obj_comment.text = an_object[an_object_key].strip; + the_document_body_section ~= comp_obj_comment; + _common_reset_(line_occur, an_object, obj_type_status); + processing.remove("verse"); + ++cntr; + } else if (((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((obj_type_status["para"] == State.off) + && (obj_type_status["heading"] == State.off))) { + /+ heading or para but neither flag nor line exists +/ + if ((conf_make_meta.make.headings.length > 2) + && (obj_type_status["make_headings"] == State.off)) { + /+ heading found +/ + _heading_found_(line, conf_make_meta.make.headings, heading_match_str, heading_match_rgx, obj_type_status); + } + if ((obj_type_status["make_headings"] == State.on) + && ((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((obj_type_status["para"] == State.off) + && (obj_type_status["heading"] == State.off))) { + /+ heading make set +/ + line = _heading_make_set_(line, line_occur, heading_match_rgx, obj_type_status); + } + /+ TODO node info: all headings identified at this point, + - extract node info here?? + - how long can it wait? + - should be incorporated in composite objects + - should happen before endnote links set (they need to be moved down?) + +/ + if (line.matchFirst(rgx.heading)) { + /+ heading match +/ + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + _heading_matched_(line, line_occur, an_object, an_object_key, lv, collapsed_lev, obj_type_status, conf_make_meta); + } else if (line_occur["para"] == State.off) { + /+ para match +/ + an_object_key="body_nugget"; + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + line = _doc_header_and_make_substitutions_fontface_(line, conf_make_meta); + _para_match_(line, an_object, an_object_key, indent, bullet, obj_type_status, line_occur); + } + } else if (line_occur["heading"] > State.off) { + /+ heading +/ + debug(heading) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; + ++line_occur["heading"]; + } else if (line_occur["para"] > State.off) { + /+ paragraph +/ + debug(para) { + writeln(an_object_key, "-> ", line); + } + line = _doc_header_and_make_substitutions_(line, conf_make_meta); + line = _doc_header_and_make_substitutions_fontface_(line, conf_make_meta); + an_object[an_object_key] ~= " " ~ line; + ++line_occur["para"]; + } + } + } else if (obj_type_status["blocks"] == TriState.closing) { + /+ line empty, with blocks flag +/ + _block_flag_line_empty_( + bookindex_extract_hash, + line, + an_object, + the_document_body_section, + bookindex_unordered_hashes, + obj_cite_digits, + comp_obj_heading, + cntr, + obj_type_status, + object_number_poem, + conf_make_meta, + ); + } else { + /+ line.empty, post contents, empty variables: +/ + assert( + line.empty, + "\nline should be empty:\n \"" + ~ line ~ "\"" + ); + assert( + (obj_type_status["blocks"] == State.off), + "code block status: none" + ); + if ((obj_type_status["heading"] == State.on) + && (line_occur["heading"] > State.off)) { + /+ heading object (current line empty) +/ + obj_cite_digits = (an_object["lev_markup_number"].to!int == 0) + ? ocn_emit(OCNstatus.reset) + : ocn_emit(OCNstatus.on); + an_object["is"] = "heading"; + an_object_key="body_nugget"; + auto substantive_object_and_anchor_tags_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, _new_doc); + an_object["substantive"] = substantive_object_and_anchor_tags_tuple[sObj.content]; + anchor_tags = substantive_object_and_anchor_tags_tuple[sObj.anchor_tags]; + if (_new_doc) { + cnt1 = 1; + cnt2 = 1; + cnt3 = 1; + _new_doc = false; + } + if ( + an_object["lev_markup_number"].to!int == 4 + && anchor_tags.length > 0 + ) { + segment_anchor_tag_that_object_belongs_to = anchor_tags[0]; + segment_anchor_tag_that_object_belongs_to_uri = anchor_tags[0] ~ ".fnSuffix"; + anchor_tag_ = anchor_tags[0]; + } else if (an_object["lev_markup_number"].to!int > 4) { + segment_anchor_tag_that_object_belongs_to = anchor_tag_; + segment_anchor_tag_that_object_belongs_to_uri = anchor_tag_ ~ ".fnSuffix#" ~ obj_cite_digits.on.to!string; + } else if (an_object["lev_markup_number"].to!int < 4) { + string segn; + switch (an_object["lev_markup_number"].to!int) { + case 0: + segn = "_the_title"; + goto default; + case 1: + segn = "_part_" ~ cnt1.to!string; + ++cnt1; + goto default; + case 2: + segn = "_part_" ~ cnt1.to!string ~ "_" ~ cnt2.to!string; + ++cnt2; + goto default; + case 3: + segn = "_part_" ~ cnt1.to!string ~ "_" ~ cnt2.to!string ~ "_" ~ cnt3.to!string; + ++cnt3; + goto default; + default: + segment_anchor_tag_that_object_belongs_to = segn; + segment_anchor_tag_that_object_belongs_to_uri = segn ~ ".fnSuffix"; + break; + } + } + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash(an_object["bookindex_nugget"], obj_cite_digits, segment_anchor_tag_that_object_belongs_to); + /+ (incrementally build toc) table of contents here! +/ + _anchor_tag=to!string(obj_cite_digits.on); + the_table_of_contents_section = obj_im.table_of_contents_gather_headings( + an_object, + conf_make_meta, + segment_anchor_tag_that_object_belongs_to, + _anchor_tag, + lev4_subtoc, + the_table_of_contents_section, + ); + if (an_object["lev_markup_number"] == "4") { + html_segnames ~= segment_anchor_tag_that_object_belongs_to; + html_segnames_ptr = html_segnames_ptr_cntr; + html_segnames_ptr_cntr++; + } + auto comp_obj_heading + = node_construct.node_emitter_heading( + an_object["substantive"], // string + an_object["lev_markup_number"], // string + an_object["lev_collapsed_number"], // string + segment_anchor_tag_that_object_belongs_to, // string + obj_cite_digits, // OCNset + cntr, // int + heading_ptr, // int + lv_ancestors_txt, // string[] + an_object["is"], // string + html_segnames_ptr, // int + substantive_object_and_anchor_tags_tuple[sObj.notes_reg], + substantive_object_and_anchor_tags_tuple[sObj.notes_star], + substantive_object_and_anchor_tags_tuple[sObj.links], + ); + ++heading_ptr; + debug(segments) { + writeln(an_object["lev_markup_number"]); + writeln(segment_anchor_tag_that_object_belongs_to); + } + the_document_body_section ~= comp_obj_heading; + debug(objectrelated1) { // check + writeln(line); + } + _common_reset_(line_occur, an_object, obj_type_status); + an_object.remove("lev"); + an_object.remove("lev_markup_number"); + processing.remove("verse"); + ++cntr; + } else if ((obj_type_status["para"] == State.on) + && (line_occur["para"] > State.off)) { + /+ paragraph object (current line empty) +/ + obj_cite_digits = ocn_emit(obj_type_status["ocn_status"]); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash(an_object["bookindex_nugget"], obj_cite_digits, segment_anchor_tag_that_object_belongs_to); + an_object["is"] = "para"; + auto comp_obj_heading + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"], + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_para = comp_obj_para.init; + comp_obj_para.metainfo.is_of_part = "body"; + comp_obj_para.metainfo.is_of_section = "body"; + comp_obj_para.metainfo.is_of_type = "para"; + comp_obj_para.metainfo.is_a = "para"; + comp_obj_para.text = an_object["substantive"].to!string.strip; + comp_obj_para.metainfo.ocn = obj_cite_digits.digit; + comp_obj_para.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_para.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_para.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_para.attrib.indent_hang = indent["hang_position"]; + comp_obj_para.attrib.indent_base = indent["base_position"]; + comp_obj_para.attrib.bullet = bullet; + comp_obj_para.tags.anchor_tags = anchor_tags; + comp_obj_para.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_para.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_para.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + comp_obj_para.has.contains_image_without_dimensions = substantive_obj_misc_tuple[sObj.image_no_dimensions]; + the_document_body_section ~= comp_obj_para; + _common_reset_(line_occur, an_object, obj_type_status); + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + processing.remove("verse"); + ++cntr; + } else { + // could be useful to test line variable should be empty and never null + } + } // close else for line empty + } // close else for not the above + } // close after non code, other blocks or regular text + /+ unless (the_document_body_section.length == 0) ? +/ + if (the_document_body_section.length > 0) { + if (((the_document_body_section[$-1].metainfo.is_a == "para") + || (the_document_body_section[$-1].metainfo.is_a == "heading") + || (the_document_body_section[$-1].metainfo.is_a == "quote") + || (the_document_body_section[$-1].metainfo.is_a == "group") + || (the_document_body_section[$-1].metainfo.is_a == "block") + || (the_document_body_section[$-1].metainfo.is_a == "verse")) + && (the_document_body_section.length > previous_length)) { + if ((the_document_body_section[$-1].metainfo.is_a == "heading") + && (the_document_body_section[$-1].metainfo.heading_lev_markup < 5)) { + obj_type_status["biblio_section"] = State.off; + obj_type_status["glossary_section"] = State.off; + obj_type_status["blurb_section"] = State.off; + } + if (the_document_body_section[$-1].metainfo.is_a == "verse") { + /+ scan for endnotes for whole poem (each verse in poem) +/ + foreach (i; previous_length .. the_document_body_section.length) { + if (the_document_body_section[i].metainfo.is_a == "verse") { + if ((the_document_body_section[i].text).match( + rgx.inline_notes_delimiter_al_regular_number_note + )) { + note_section.gather_notes_for_endnote_section( + the_document_body_section, + segment_anchor_tag_that_object_belongs_to, + (i).to!int, + ); + } + } + } + } else { + /+ scan object for endnotes +/ + previous_length = the_document_body_section.length.to!int; + if ((the_document_body_section[$-1].text).match( + rgx.inline_notes_delimiter_al_regular_number_note + )) { + previous_count=(the_document_body_section.length -1).to!int; + note_section.gather_notes_for_endnote_section( + the_document_body_section, + segment_anchor_tag_that_object_belongs_to, + (the_document_body_section.length-1).to!int, + ); + } + } + previous_length = the_document_body_section.length.to!int; + } + } + } /+ ← srcDocLoop closed: loop markup document/text line by line +/ + /+ ↓ post loop markup document/text +/ + auto en_tuple + = note_section.endnote_objects(obj_cite_digits, opt_action); + static assert(!isTypeTuple!(en_tuple)); + auto the_endnotes_section = en_tuple[0]; + obj_cite_digits = en_tuple[1]; + debug(endnotes) { + writefln( + "%s %s", + __LINE__, + the_endnotes_section.length + ); + foreach (o; the_endnotes_section) { + writeln(o); + } + } + if (an_object["glossary_nugget"].length == 0) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "empty"; + comp_obj_heading_.metainfo.is_of_section = "empty"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Glossary section"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_glossary_section ~= comp_obj_heading_; + } + debug(glossary) { + foreach (gloss; the_glossary_section) { + writeln(gloss.text); + } + } + auto biblio_unsorted_incomplete = biblio_arr_json.dup; + auto biblio = Bibliography(); + auto biblio_ordered + = biblio._bibliography_(biblio_unsorted_incomplete, bib_arr_json); + if (biblio_ordered.length > 0) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "bibliography"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Bibliography"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "_part_bibliography"; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_bibliography_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "bibliography"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Bibliography"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "bibliography"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 2; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.tags.anchor_tags = ["bibliography"]; + the_bibliography_section ~= comp_obj_heading_; + } else { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "empty"; + comp_obj_heading_.metainfo.is_of_section = "empty"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Bibliography"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_bibliography_section ~= comp_obj_heading_; + } + string out_; + foreach (entry; biblio_ordered) { + out_ = format( + "%s \"%s\"%s%s%s%s%s%s%s%s%s.", + ((entry["author"].str.empty) ? entry["editor"].str : entry["author"].str), + entry["fulltitle"].str, + ((entry["journal"].str.empty) ? "" : ", /{" ~ entry["journal"].str ~ "}/"), + ((entry["volume"].str.empty) ? "" : ", " ~ entry["volume"].str), + ((entry["in"].str.empty) ? "" : ", " ~ entry["in"].str), + ((!(entry["author"].str.empty) && (!(entry["editor"].str.empty))) ? entry["editor"].str : ""), + ", " ~ entry["year"].str, + ((entry["pages"].str.empty) ? "" : ", " ~ entry["pages"].str), + ((entry["publisher"].str.empty) ? "" : ", " ~ entry["publisher"].str), + ((entry["place"].str.empty) ? "" : ", " ~ entry["place"].str), + ((entry["url"].str.empty) ? "" : ", [" ~ entry["url"].str ~ "]"), + ); + comp_obj_para = comp_obj_para.init; + comp_obj_para.metainfo.is_of_part = "backmatter"; + comp_obj_para.metainfo.is_of_section = "bibliography"; + comp_obj_para.metainfo.is_of_type = "para"; + comp_obj_para.metainfo.is_a = "bibliography"; + comp_obj_para.text = out_.to!string.strip; + comp_obj_para.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_para.attrib.indent_hang = 0; + comp_obj_para.attrib.indent_base = 1; + comp_obj_para.attrib.bullet = bullet; + comp_obj_para.tags.anchor_tags = anchor_tags; + the_bibliography_section ~= comp_obj_para; + } + debug(bibliosection) { + foreach (o; the_bibliography_section) { + writeln(o.text); + } + } + auto bi = BookIndexReportSection(); + auto bi_tuple + = bi.bookindex_build_abstraction_section( + bookindex_unordered_hashes, + obj_cite_digits, + opt_action, + ); + destroy(bookindex_unordered_hashes); + static assert(!isTypeTuple!(bi_tuple)); + auto the_bookindex_section = bi_tuple[0]; + obj_cite_digits = bi_tuple[1]; + debug(bookindex) { + foreach (bi_entry; the_bookindex_section["seg"]) { + writeln(bi_entry); + } + } + if (an_object["blurb_nugget"].length == 0) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "empty"; + comp_obj_heading_.metainfo.is_of_section = "empty"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Blurb section"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = ""; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_blurb_section ~= comp_obj_heading_; + } + debug(blurb) { + foreach (blurb; the_blurb_section) { + writeln(blurb.text); + } + } + indent=[ + "hang_position" : 1, + "base_position" : 1, + ]; + comp_obj_toc = comp_obj_toc.init; + comp_obj_toc.metainfo.is_of_part = "frontmatter"; + comp_obj_toc.metainfo.is_of_section = "toc"; + comp_obj_toc.metainfo.is_of_type = "para"; + comp_obj_toc.metainfo.is_a = "toc"; + comp_obj_toc.metainfo.ocn = 0; + comp_obj_toc.metainfo.object_number_off = ""; + comp_obj_toc.metainfo.object_number_type = 0; + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.attrib.bullet = false; + if (the_endnotes_section.length > 1) { + toc_txt_ = format( + "{ %s }%s%s%s", + "Endnotes", + mkup.mark_internal_site_lnk, + "endnotes", // segment_anchor_tag_that_object_belongs_to + ".fnSuffix", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + } + if (the_glossary_section.length > 1) { + toc_txt_ = format( + "{ %s }%s%s%s", + "Glossary", + mkup.mark_internal_site_lnk, + "glossary", // segment_anchor_tag_that_object_belongs_to + ".fnSuffix", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + toc_txt_ = format( + "{ %s }#%s", + "Glossary", + "glossary", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + if (the_bibliography_section.length > 1){ + toc_txt_ = format( + "{ %s }%s%s%s", + "Bibliography", + mkup.mark_internal_site_lnk, + "bibliography", // segment_anchor_tag_that_object_belongs_to + ".fnSuffix", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + + toc_txt_ = format( + "{ %s }#%s", + "Bibliography", + "bibliography", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + if (the_bookindex_section["seg"].length > 1) { + toc_txt_ = format( + "{ %s }%s%s%s", + "Book Index", + mkup.mark_internal_site_lnk, + "bookindex", // segment_anchor_tag_that_object_belongs_to + ".fnSuffix", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + } + if (the_bookindex_section["scroll"].length > 1) { + toc_txt_ = format( + "{ %s }#%s", + "Book Index", + "bookindex", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + if (the_blurb_section.length > 1) { + toc_txt_ = format( + "{ %s }%s%s%s", + "Blurb", + mkup.mark_internal_site_lnk, + "blurb", // segment_anchor_tag_that_object_belongs_to + ".fnSuffix", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + toc_txt_ = format( + "{ %s }#%s", + "Blurb", + "blurb", + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.has.inline_links = true; + comp_obj_toc.text = toc_txt_.to!string.strip; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + debug(toc) { + writefln( + "%s %s", + __LINE__, + the_table_of_contents_section["seg"].length + ); + foreach (toc_linked_heading; the_table_of_contents_section["seg"]) { + writeln(mkup.indent_by_spaces_provided(toc_linked_heading.attrib.indent_hang), toc_linked_heading.text); + } + } + debug(tocscroll) { + writefln( + "%s %s", + __LINE__, + the_table_of_contents_section["seg"].length + ); + foreach (toc_linked_heading; the_table_of_contents_section["scroll"]) { + writeln(mkup.indent_by_spaces_provided(toc_linked_heading.attrib.indent_hang), toc_linked_heading.text); + } + } + the_document_head_section ~= the_document_body_section[0]; + the_document_body_section=the_document_body_section[1..$]; + int[] _get_ancestors_markup(O)(O obj, ref int[] _ancestors_markup) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.metainfo.heading_lev_markup == 1) { + _ancestors_markup = [ + _ancestors_markup[0], + 0,0,0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 2) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + 0,0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 3) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + 0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 4) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + _ancestors_markup[3], + 0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 5) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + _ancestors_markup[3], + _ancestors_markup[4], + 0,0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 6) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + _ancestors_markup[3], + _ancestors_markup[4], + _ancestors_markup[5], + 0,0 + ]; + } + if (obj.metainfo.heading_lev_markup == 7) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + _ancestors_markup[3], + _ancestors_markup[4], + _ancestors_markup[5], + _ancestors_markup[6], + 0 + ]; + } + if (obj.metainfo.heading_lev_markup == 8) { + _ancestors_markup = [ + _ancestors_markup[0], + _ancestors_markup[1], + _ancestors_markup[2], + _ancestors_markup[3], + _ancestors_markup[4], + _ancestors_markup[5], + _ancestors_markup[6], + _ancestors_markup[7] + ]; + } + _ancestors_markup[obj.metainfo.heading_lev_markup] = obj.metainfo.ocn; + } + debug(ancestor_markup) { + writeln("marked up: ", _ancestors_markup); + } + return _ancestors_markup; + } + int[] _get_ancestors_collapsed(O)(O obj, ref int[] _ancestors_collapsed) { + if (obj.metainfo.is_a == "heading") { + if (obj.metainfo.heading_lev_collapsed == 1) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + 0,0,0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 2) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + 0,0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 3) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + 0,0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 4) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + _ancestors_collapsed[3], + 0,0,0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 5) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + _ancestors_collapsed[3], + _ancestors_collapsed[4], + 0,0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 6) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + _ancestors_collapsed[3], + _ancestors_collapsed[4], + _ancestors_collapsed[5], + 0,0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 7) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + _ancestors_collapsed[3], + _ancestors_collapsed[4], + _ancestors_collapsed[5], + _ancestors_collapsed[6], + 0 + ]; + } + if (obj.metainfo.heading_lev_collapsed == 8) { + _ancestors_collapsed = [ + _ancestors_collapsed[0], + _ancestors_collapsed[1], + _ancestors_collapsed[2], + _ancestors_collapsed[3], + _ancestors_collapsed[4], + _ancestors_collapsed[5], + _ancestors_collapsed[6], + _ancestors_collapsed[7] + ]; + } + _ancestors_collapsed[obj.metainfo.heading_lev_collapsed] = obj.metainfo.ocn; + } + debug(ancestor_collapsed) { + writeln("collapsed: ", _ancestors_collapsed); + } + return _ancestors_collapsed; + } + /+ multiple 1~ levels, loop through document body +/ + if (the_document_body_section.length > 1) { + int[] _ancestors_markup = [0,0,0,0,0,0,0,0]; + int[][] _ancestors_markup_; + _ancestors_markup = [1,0,0,0,0,0,0,0]; + _ancestors_markup_ ~= _ancestors_markup; + int[] _ancestors_collapsed = [0,0,0,0,0,0,0,0]; + int[][] _ancestors_collapsed_; + _ancestors_collapsed = [1,0,0,0,0,0,0,0]; + _ancestors_collapsed_ ~= _ancestors_collapsed; + foreach (ref obj; the_document_body_section) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.markedup_ancestors = _get_ancestors_markup(obj, _ancestors_markup); + obj.metainfo.collapsed_ancestors = _get_ancestors_collapsed(obj, _ancestors_collapsed); + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + } + debug(ancestors) { + writeln("ancestors markup o_n: ", obj.metainfo.markedup_ancestors); + writeln("ancestors collapsed o_n: ", obj.metainfo.markedup_ancestors); + } + } + if (the_endnotes_section.length > 1) { + html_segnames ~= "endnotes"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref obj; the_endnotes_section) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + if (the_glossary_section.length > 1) { + html_segnames ~= "glossary"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref obj; the_glossary_section) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + if (the_bibliography_section.length > 1) { + html_segnames ~= "bibliography"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref obj; the_bibliography_section) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + if (the_bookindex_section["scroll"].length > 1) { + html_segnames ~= "bookindex"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref obj; the_bookindex_section["scroll"]) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + foreach (ref obj; the_bookindex_section["seg"]) { + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + if (the_blurb_section.length > 1) { + html_segnames ~= "blurb"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref obj; the_blurb_section) { + if (obj.metainfo.is_a == "heading") { + obj.metainfo.parent_ocn = obj.metainfo.markedup_ancestors[obj.metainfo.parent_lev_markup]; + } + if (obj.metainfo.heading_lev_markup == 4) { + obj.ptr.html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + auto get_decendants(S)(S document_sections) { + int[string] _heading_ocn_decendants; + string[] _ocn_open_key = ["","","","","","","",""]; + auto _doc_sect_length = document_sections.length - 1; + int _last_ocn; + foreach (_lg, ref obj; document_sections) { + if (obj.metainfo.is_a == "heading") { + foreach (_dts_lv, dom_tag_status; obj.metainfo.dom_structure_markedup_tags_status) { + switch (dom_tag_status) { + case DomTags.none: break; + case DomTags.open: + _ocn_open_key[_dts_lv] = (obj.metainfo.ocn).to!string; + _heading_ocn_decendants[_ocn_open_key[_dts_lv]] = obj.metainfo.ocn; + break; + case DomTags.close: + if (_ocn_open_key[_dts_lv].empty) { + _ocn_open_key[_dts_lv] = "0"; + } + _heading_ocn_decendants[_ocn_open_key[_dts_lv]] = obj.metainfo.ocn - 1; + _ocn_open_key[_dts_lv] = (0).to!string; + break; + case DomTags.close_and_open: + if (_ocn_open_key[_dts_lv].empty) { + _ocn_open_key[_dts_lv] = "0"; + } + _heading_ocn_decendants[_ocn_open_key[_dts_lv]] = obj.metainfo.ocn - 1; + _ocn_open_key[_dts_lv] = (obj.metainfo.ocn).to!string; + _heading_ocn_decendants[_ocn_open_key[_dts_lv]] = obj.metainfo.ocn; + break; + case DomTags.open_still: break; + default: break; + } + } + } + if (obj.metainfo.ocn > 0) { + _last_ocn = obj.metainfo.ocn; + } + if (_lg == _doc_sect_length) { + _heading_ocn_decendants["1"] = _last_ocn; // close existing o_n key + } + } + Tuple!(int, int)[] pairs; + foreach (pair; _heading_ocn_decendants.byPair) { + pairs ~= tuple(pair[0].to!int, pair[1]); + } + return pairs.sort; + } + string[] _images; + auto extract_images(S)(S content_block) { + string[] images_; + if (auto m = content_block.matchAll(rgx.image)) { + images_ ~= m.captures[1]; + } + return images_; + } + string[] segnames_0_4; + auto _image_dimensions(M,O)(M manifest_matter, O obj) { + if (obj.has.contains_image_without_dimensions) { + import std.math; + import imageformats; + int w, h, chans; + real _w, _h; + int max_width = 640; + foreach (m; obj.text.matchAll(rgx.inline_image_without_dimensions)) { + debug(images) { + writeln(manifest_matter.src.image_dir_path ~ "/" ~ m.captures["img"]); + } + read_image_info(manifest_matter.src.image_dir_path ~ "/" ~ m.captures["img"], w, h, chans); + // calculate, decide max width and proportionally reduce to keep w & h within it + debug(images) { + writeln("width: ", w, ", height: ", h); + } + if (w > max_width) { + _w = max_width; + _h = round((max_width / w.to!real) * h.to!real); + } else { + _w = w; + _h = h; + } + obj.text = obj.text.replaceFirst( + rgx.inline_image_without_dimensions, + ("$1☼$3,w" ~ _w.to!string ~ "h" ~ _h.to!string ~ " $6") + ); + } + debug(images) { + writeln("image without dimensions: ", obj.text); + } + } + return obj; + } + foreach (ref obj; the_document_head_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + } + if (obj.metainfo.heading_lev_markup == 0) { + /+ TODO second hit (of two) with same assertion failure, check, fix and reinstate + assert( obj.metainfo.ocn == 1, + "Title OCN should be 1 not: " ~ obj.metainfo.ocn.to!string); // bug introduced 0.18.1 + +/ + obj.metainfo.ocn = 1; + obj.metainfo.object_number_type = OCNtype.ocn; + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + if (the_table_of_contents_section["scroll"].length > 1) { + /+ scroll +/ + dom_structure_markedup_tags_status_buffer = dom_structure_markedup_tags_status.dup; + dom_structure_collapsed_tags_status_buffer = dom_structure_collapsed_tags_status.dup; + foreach (ref obj; the_table_of_contents_section["scroll"]) { + if (obj.metainfo.is_a == "heading") { + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + /+ seg +/ + dom_structure_markedup_tags_status = dom_structure_markedup_tags_status_buffer.dup; + dom_structure_collapsed_tags_status = dom_structure_collapsed_tags_status_buffer.dup; + foreach (ref obj; the_table_of_contents_section["seg"]) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ multiple 1~ levels, loop through document body +/ + if (the_document_body_section.length > 1) { + foreach (ref obj; the_document_body_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.lev4_subtoc = lev4_subtoc[obj.tags.segment_anchor_tag]; + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "para") { + _images ~= extract_images(obj.text); + obj = _image_dimensions(manifest_matter, obj); + } + } + } + auto images=uniq(_images.sort()); + /+ optional only one 1~ level +/ + if (the_endnotes_section.length > 1) { + dom_structure_markedup_tags_status_buffer = dom_structure_markedup_tags_status.dup; + dom_structure_collapsed_tags_status_buffer = dom_structure_collapsed_tags_status.dup; + dom_structure_markedup_tags_status = dom_structure_markedup_tags_status_buffer.dup; + dom_structure_collapsed_tags_status = dom_structure_collapsed_tags_status_buffer.dup; + foreach (ref obj; the_endnotes_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.digit; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ optional only one 1~ level +/ + if (the_glossary_section.length > 1) { + foreach (ref obj; the_glossary_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.digit; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "glossary") { + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.digit; + } + } + } + /+ optional only one 1~ level +/ + if (the_bibliography_section.length > 1) { + foreach (ref obj; the_bibliography_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.digit; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "bibliography") { + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.on; + } + } + } + /+ optional only one 1~ level +/ + int ocn_ = obj_cite_digits.on; + int ocn_bkidx_ = 0; + int ocn_bidx_; + if (the_bookindex_section["scroll"].length > 1) { + /+ scroll +/ + dom_structure_markedup_tags_status_buffer = dom_structure_markedup_tags_status.dup; + dom_structure_collapsed_tags_status_buffer = dom_structure_collapsed_tags_status.dup; + foreach (ref obj; the_bookindex_section["scroll"]) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + } + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.on; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "bookindex") { + obj_cite_digits = ocn_emit(OCNstatus.bkidx); + obj.metainfo.ocn = obj_cite_digits.digit; + obj.metainfo.o_n_book_index = obj_cite_digits.bkidx; + obj.metainfo.object_number_type = OCNtype.bkidx; + } + } + /+ seg +/ + dom_structure_markedup_tags_status = dom_structure_markedup_tags_status_buffer.dup; + dom_structure_collapsed_tags_status = dom_structure_collapsed_tags_status_buffer.dup; + foreach (ref obj; the_bookindex_section["seg"]) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + obj.metainfo.ocn = ++ocn_; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "bookindex") { + ocn_bidx_ = ++ocn_bkidx_; + obj.metainfo.o_n_book_index = ocn_bidx_; // FIX need to distinguish from regular ocn + obj.metainfo.object_number_type = OCNtype.bkidx; + } + } + /+ TODO assert failure, reinstate + assert(obj_cite_digit_bkidx == ocn_bidx_ + obj_cite_digit_bkidx ~ " == ocn_" ~ ocn_ ~ "?"); + +/ + } + /+ optional only one 1~ level +/ + if (the_blurb_section.length > 1) { + foreach (ref obj; the_blurb_section) { + if (obj.metainfo.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + obj_cite_digits = ocn_emit(OCNstatus.on); + obj.metainfo.ocn = obj_cite_digits.on; + if (obj.metainfo.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.tags.segment_anchor_tag; + if (obj.metainfo.heading_lev_markup == 4) { + obj.tags.segname_prev = html_segnames[obj.ptr.html_segnames - 1]; + if (html_segnames.length > obj.ptr.html_segnames + 1) { + obj.tags.segname_next = html_segnames[obj.ptr.html_segnames + 1]; + } + assert(obj.tags.segment_anchor_tag == html_segnames[obj.ptr.html_segnames], + obj.tags.segment_anchor_tag ~ "!=" ~ html_segnames[obj.ptr.html_segnames]); + } + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub) + || (opt_action.sqlite_discrete) + || (opt_action.sqlite_update)) { + obj = obj_dom_structure_set_markup_tags(obj, dom_structure_markedup_tags_status, obj.metainfo.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_structure_collapsed_tags_status, obj.metainfo.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.metainfo.is_a == "blurb") { + obj_cite_digits = ocn_emit(OCNstatus.off); + obj.metainfo.object_number_off = obj_cite_digits.off.to!string; + obj.metainfo.object_number_type = OCNtype.non; + } + } + } + if (the_document_body_section.length > 1) { + auto pairs = get_decendants( + the_document_head_section ~ + the_document_body_section ~ + the_endnotes_section ~ + the_glossary_section ~ + the_bibliography_section ~ + the_bookindex_section["scroll"] ~ + the_blurb_section + ); + debug(decendants_tuple) { + pairs = pairs.sort(); + foreach (pair; pairs) { // (pair; pairs.sort()) + writeln(pair[0], "..", pair[1]); + } + } + foreach (ref obj; the_document_head_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + if (the_document_body_section.length > 1) { + foreach (ref obj; the_document_body_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + if (the_endnotes_section.length > 1) { + foreach (ref obj; the_endnotes_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + if (the_glossary_section.length > 1) { + foreach (ref obj; the_glossary_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + if (the_bibliography_section.length > 1) { + foreach (ref obj; the_bibliography_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + if (the_bookindex_section["scroll"].length > 1) { + foreach (ref obj; the_bookindex_section["scroll"]) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + if (the_blurb_section.length > 1) { + foreach (ref obj; the_blurb_section) { + if (obj.metainfo.is_a == "heading") { + foreach (pair; pairs) { + if (obj.metainfo.ocn == pair[0]) { + obj.metainfo.last_decendant_ocn = pair[1]; + } + } + } + } + } + } + /+ TODO + - note create/insert heading object sole purpose eof close all open tags + sort out: + - obj.metainfo.dom_structure_markedup_tags_status = dom_structure_markedup_tags_status; + - obj.metainfo.dom_structure_collapsed_tags_status = dom_structure_collapsed_tags_status; + +/ + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "empty"; + comp_obj_heading_.metainfo.is_of_section = "empty"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = ""; + comp_obj_heading_.metainfo.heading_lev_markup = 9; + comp_obj_heading_.metainfo.heading_lev_collapsed = 9; + comp_obj_heading_.metainfo.parent_ocn = 0; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.metainfo.dom_structure_markedup_tags_status = dom_structure_markedup_tags_status.dup; + comp_obj_heading_.metainfo.dom_structure_collapsed_tags_status = dom_structure_collapsed_tags_status.dup; + comp_obj_heading_ = obj_dom_structure_set_markup_tags(comp_obj_heading_, dom_structure_markedup_tags_status, 0); + comp_obj_heading_ = obj_dom_set_collapsed_tags(comp_obj_heading_, dom_structure_collapsed_tags_status, 0); + comp_obj_heading_ = obj_heading_ancestors(comp_obj_heading_, lv_ancestors_txt); + // the_dom_tail_section ~= comp_obj_heading_; // remove tail for now, decide on later + auto document_the = [ + "head": the_document_head_section, + "toc_seg": the_table_of_contents_section["seg"], + "toc_scroll": the_table_of_contents_section["scroll"], + /+ substantive/body: +/ + "body": the_document_body_section, + /+ backmatter: +/ + "endnotes": the_endnotes_section, + "glossary": the_glossary_section, + "bibliography": the_bibliography_section, + "bookindex_scroll": the_bookindex_section["scroll"], + "bookindex_seg": the_bookindex_section["seg"], + "blurb": the_blurb_section, + /+ dom tail only +/ + "tail": the_dom_tail_section, + ]; + string[][string] document_section_keys_sequenced = [ + "seg": ["head", "toc_seg", "body",], + "scroll": ["head", "toc_scroll", "body",], + "sql": ["head", "body",] + ]; + if (document_the["endnotes"].length > 1) { + document_section_keys_sequenced["seg"] ~= "endnotes"; + document_section_keys_sequenced["scroll"] ~= "endnotes"; + } + if (document_the["glossary"].length > 1) { + document_section_keys_sequenced["seg"] ~= "glossary"; + document_section_keys_sequenced["scroll"] ~= "glossary"; + document_section_keys_sequenced["sql"] ~= "glossary"; + } + if (document_the["bibliography"].length > 1) { + document_section_keys_sequenced["seg"] ~= "bibliography"; + document_section_keys_sequenced["scroll"] ~= "bibliography"; + document_section_keys_sequenced["sql"] ~= "bibliography"; + } + if (document_the["bookindex_seg"].length > 1) { + document_section_keys_sequenced["seg"] ~= "bookindex_seg"; + document_section_keys_sequenced["sql"] ~= "bookindex_seg"; + } + if (document_the["bookindex_scroll"].length > 1) { + document_section_keys_sequenced["scroll"] ~= "bookindex_scroll"; + } + if (document_the["blurb"].length > 1) { + document_section_keys_sequenced["seg"] ~= "blurb"; + document_section_keys_sequenced["scroll"] ~= "blurb"; + document_section_keys_sequenced["sql"] ~= "blurb"; + } + if ((opt_action.html) + || (opt_action.html_scroll) + || (opt_action.html_seg) + || (opt_action.epub)) { + document_section_keys_sequenced["seg"] ~= "tail"; + document_section_keys_sequenced["scroll"] ~= "tail"; + } + auto sequenced_document_keys = docSectKeysSeq!()(document_section_keys_sequenced); + auto segnames = html_segnames.dup; + destroy(the_document_head_section); + destroy(the_table_of_contents_section); + destroy(the_document_body_section); + destroy(the_endnotes_section); + destroy(the_glossary_section); + destroy(the_bibliography_section); + destroy(the_bookindex_section); + destroy(the_blurb_section); + destroy(html_segnames); + destroy(bookindex_unordered_hashes); + destroy(an_object); + obj_cite_digits = ocn_emit(OCNstatus.reset); + biblio_arr_json = []; + obj_cite_digit_ = 0; + html_segnames_ptr = 0; + html_segnames_ptr_cntr = 0; + content_non_header = "8"; + dom_structure_markedup_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_structure_markedup_tags_status_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_structure_collapsed_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_structure_collapsed_tags_status_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + debug(abstraction) { + foreach (part; sequenced_document_keys.scroll) { + foreach (ref obj; document_the[part]) { + debug(node_misc) { + if (obj.metainfo.is_a == "heading") { + writeln("heading ancestors text?: ", obj.tags.heading_ancestors_text); + } + } + debug(node_headings) { + if (obj.metainfo.is_a == "heading") { + writeln("---"); + writeln(obj.text); + writeln(" node: ", obj.metainfo.node); + writeln(" ocn: ", obj.metainfo.ocn); + writeln(" markedup heading lev: ", obj.metainfo.heading_lev_markup); + writeln(" collapsed heading lev ", obj.metainfo.heading_lev_collapsed); + writeln(" parent ocn: ", obj.metainfo.parent_ocn); + writeln(" parent lev: ", obj.metainfo.parent_lev_markup); + writeln(" markedup ancestors o_n: ", obj.metainfo.markedup_ancestors); + writeln(" dom markedup: ", obj.metainfo.dom_structure_markedup_tags_status); + writeln(" collapsed ancestors o_n: ", obj.metainfo.collapsed_ancestors); + writeln(" dom collapsed: ", obj.metainfo.dom_structure_collapsed_tags_status); + } + } + debug(node_all) { + writeln("---"); + if (obj.metainfo.is_a == "heading") { + writeln(obj.text); + } + writeln("node: ", obj.metainfo.node); + writeln("ocn: ", obj.metainfo.ocn); + writeln("markedup heading lev: ", obj.metainfo.heading_lev_markup); + writeln("collapsed heading lev ", obj.metainfo.heading_lev_collapsed); + writeln("markedup ancestors o_n: ", obj.metainfo.markedup_ancestors); + writeln("collapsed ancestors o_n: ", obj.metainfo.collapsed_ancestors); + writeln("dom markedup: ", obj.metainfo.dom_structure_markedup_tags_status); + writeln("dom collapsed: ", obj.metainfo.dom_structure_collapsed_tags_status); + writeln("parent ocn: ", obj.metainfo.parent_ocn); + writeln("parent lev: ", obj.metainfo.parent_lev_markup); + writeln("Node: ", obj.metainfo.ocn); + } + } + } + } + auto t = tuple( + document_the, + sequenced_document_keys, + segnames, + segnames_0_4, + images, + ); + return t; + /+ post loop markup document/text ↑ +/ + } /+ ← closed: abstract doc source +/ + /+ ↓ abstraction functions +/ + static auto object_reset(O)(O an_object) { + debug(asserts) { + static assert(is(typeof(an_object) == string[string])); + } + an_object.remove("body_nugget"); + an_object.remove("substantive"); + an_object.remove("is"); + an_object.remove("attrib"); + an_object.remove("bookindex_nugget"); + return an_object; + } + auto _common_reset_(L,O,T)( + return ref L line_occur, + return ref O an_object, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line_occur) == int[string])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + line_occur["heading"] = State.off; + line_occur["para"] = State.off; + obj_type_status["heading"] = State.off; + obj_type_status["para"] = State.off; + an_object = object_reset(an_object); + } + static auto _check_ocn_status_(L,T)( + L line, + T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if ((!line.empty) && (obj_type_status["ocn_status_off_for_multiple_objects"] == TriState.off)) { + /+ not multi-line object, check whether object_number is on or turned off +/ + if (line.matchFirst(rgx.object_number_block_marks)) { + /+ switch off object_number +/ + if (line.matchFirst(rgx.object_number_off_block)) { + obj_type_status["ocn_status_off_for_multiple_objects"] = TriState.on; + debug(ocnoff) { + writeln(line); + } + } + if (line.matchFirst(rgx.object_number_off_block_dh)) { + obj_type_status["ocn_status_off_for_multiple_objects"] = TriState.closing; + debug(ocnoff) { + writeln(line); + } + } + } else { + if (obj_type_status["ocn_status_off_for_multiple_objects"] == TriState.off) { + if (line.matchFirst(rgx.object_number_off)) { + obj_type_status["ocn_status"] = OCNstatus.off; + } else if (line.matchFirst(rgx.object_number_off_dh)) { + obj_type_status["ocn_status"] = OCNstatus.closing; + } else { + obj_type_status["ocn_status"] = OCNstatus.on; + } + } else { + obj_type_status["ocn_status"] + = obj_type_status["ocn_status_off_for_multiple_objects"]; + } + } + } else if ((!line.empty) && (obj_type_status["ocn_status_off_for_multiple_objects"] > TriState.off)) { + if (line.matchFirst(rgx.object_number_off_block_close)) { + obj_type_status["ocn_status_off_for_multiple_objects"] = TriState.off; + obj_type_status["ocn_status"] = OCNstatus.on; + debug(ocnoff) { + writeln(line); + } + } + } + return obj_type_status; + } + auto _doc_header_and_make_substitutions_(L,CMM)( + L line, + CMM conf_make_meta, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + } + enum Substitute { match, markup, } + if (conf_make_meta.make.substitute) { + foreach(substitution_pair; conf_make_meta.make.substitute) { + line = line.replaceAll( + regex("\b" ~ substitution_pair[Substitute.match]), + substitution_pair[Substitute.markup] + ); + } + } + return line; + } + auto _doc_header_and_make_substitutions_fontface_(L,CMM)( + L line, + CMM conf_make_meta, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + } + enum Substitute { match, markup, } + if ( conf_make_meta.make.bold) { + line = line.replaceAll( + regex("\b" ~ conf_make_meta.make.bold[Substitute.match]), + conf_make_meta.make.bold[Substitute.markup] + ); + } + if (conf_make_meta.make.emphasis) { + line = line.replaceAll( + regex("\b" ~ conf_make_meta.make.emphasis[Substitute.match]), + conf_make_meta.make.emphasis[Substitute.markup] + ); + } + if (conf_make_meta.make.italics) { + line = line.replaceAll( + regex("\b" ~ conf_make_meta.make.italics[Substitute.match]), + conf_make_meta.make.italics[Substitute.markup] + ); + } + return line; + } + void _start_block_(L,T,N)( + L line, + return ref T obj_type_status, + return ref N object_number_poem + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(obj_type_status) == int[string])); + static assert(is(typeof(object_number_poem) == string[string])); + } + static auto rgx = Rgx(); + string code_block_syntax = ""; + bool code_block_numbered = false; + if (auto m = line.matchFirst(rgx.block_curly_code_open)) { + /+ curly code open +/ + code_block_syntax = (m.captures[1]) ? m.captures[1].to!string : ""; + code_block_numbered = (m.captures[2] == "#") ? true : false; + debug(code) { // code (curly) open + writefln( + "* [code curly] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["code"] = TriState.on; + obj_type_status["curly_code"] = TriState.on; + } else if (line.matchFirst(rgx.block_curly_poem_open)) { + /+ curly poem open +/ + debug(poem) { // poem (curly) open + writefln( + "* [poem curly] %s", + line + ); + } + object_number_poem["start"] = obj_cite_digits.on.to!string; + obj_type_status["blocks"] = TriState.on; + obj_type_status["verse_new"] = State.on; + obj_type_status["poem"] = TriState.on; + obj_type_status["curly_poem"] = TriState.on; + } else if (line.matchFirst(rgx.block_curly_group_open)) { + /+ curly group open +/ + debug(group) { // group (curly) open + writefln( + "* [group curly] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["group"] = TriState.on; + obj_type_status["curly_group"] = TriState.on; + } else if (line.matchFirst(rgx.block_curly_block_open)) { + /+ curly block open +/ + debug(block) { // block (curly) open + writefln( + "* [block curly] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["block"] = TriState.on; + obj_type_status["curly_block"] = TriState.on; + } else if (line.matchFirst(rgx.block_curly_quote_open)) { + /+ curly quote open +/ + debug(quote) { // quote (curly) open + writefln( + "* [quote curly] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["quote"] = TriState.on; + obj_type_status["curly_quote"] = TriState.on; + } else if (auto m = line.matchFirst(rgx.block_curly_table_open)) { + /+ curly table open +/ + debug(table) { // table (curly) open + writefln( + "* [table curly] %s", + line + ); + } + an_object["table_head"] = m.captures[1].to!string; + an_object["block_type"] = "curly"; + obj_type_status["blocks"] = TriState.on; + obj_type_status["table"] = TriState.on; + obj_type_status["curly_table"] = TriState.on; + } else if (auto m = line.matchFirst(rgx.block_curly_table_special_markup)) { + /+ table: special table block markup syntax! +/ + an_object["table_head"] = m.captures[1].to!string; + an_object["block_type"] = "special"; + obj_type_status["blocks"] = TriState.on; + obj_type_status["table"] = TriState.on; + obj_type_status["curly_table_special_markup"] = TriState.on; + } else if (auto m = line.matchFirst(rgx.block_tic_code_open)) { + /+ tic code open +/ + code_block_syntax = (m.captures[1]) ? m.captures[1].to!string : ""; + code_block_numbered = (m.captures[2] == "#") ? true : false; + debug(code) { // code (tic) open + writefln( + "* [code tic] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["code"] = TriState.on; + obj_type_status["tic_code"] = TriState.on; + } else if (line.matchFirst(rgx.block_tic_poem_open)) { + /+ tic poem open +/ + debug(poem) { // poem (tic) open + writefln( + "* [poem tic] %s", + line + ); + } + object_number_poem["start"] = obj_cite_digits.on.to!string; + obj_type_status["blocks"] = TriState.on; + obj_type_status["verse_new"] = State.on; + obj_type_status["poem"] = TriState.on; + obj_type_status["tic_poem"] = TriState.on; + } else if (line.matchFirst(rgx.block_tic_group_open)) { + /+ tic group open +/ + debug(group) { // group (tic) open + writefln( + "* [group tic] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["group"] = TriState.on; + obj_type_status["tic_group"] = TriState.on; + } else if (line.matchFirst(rgx.block_tic_block_open)) { + /+ tic block open +/ + debug(block) { // block (tic) open + writefln( + "* [block tic] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["block"] = TriState.on; + obj_type_status["tic_block"] = TriState.on; + } else if (line.matchFirst(rgx.block_tic_quote_open)) { + /+ tic quote open +/ + debug(quote) { // quote (tic) open + writefln( + "* [quote tic] %s", + line + ); + } + obj_type_status["blocks"] = TriState.on; + obj_type_status["quote"] = TriState.on; + obj_type_status["tic_quote"] = TriState.on; + } else if (auto m = line.matchFirst(rgx.block_tic_table_open)) { + /+ tic table open +/ + debug(table) { // table (tic) open + writefln( + "* [table tic] %s", + line + ); + } + an_object["table_head"] = m.captures[1].to!string; + an_object["block_type"] = "tic"; + obj_type_status["blocks"] = TriState.on; + obj_type_status["table"] = TriState.on; + obj_type_status["tic_table"] = TriState.on; + } + } + void _quote_block_(L,O,T)( + L line, + return ref O an_object, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_quote"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_quote_close)) { + debug(quote) { // quote (curly) close + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["quote"] = TriState.closing; + obj_type_status["curly_quote"] = TriState.off; + } else { + debug(quote) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build quote array (or string) + } + } else if (obj_type_status["tic_quote"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(quote) { // quote (tic) close + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["quote"] = TriState.closing; + obj_type_status["tic_quote"] = TriState.off; + } else { + debug(quote) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build quote array (or string) + } + } + } + void _group_block_(L,O,T)( + L line, + return ref O an_object, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_group"] == State.on) { + if (line.matchFirst(rgx.block_curly_group_close)) { + debug(group) { + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["group"] = TriState.closing; + obj_type_status["curly_group"] = TriState.off; + } else { + debug(group) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build group array (or string) + } + } else if (obj_type_status["tic_group"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(group) { + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["group"] = TriState.closing; + obj_type_status["tic_group"] = TriState.off; + } else { + debug(group) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build group array (or string) + } + } + } + void _block_block_(L,O,T)( + L line, + return ref O an_object, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_block"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_block_close)) { + debug(block) { // block (curly) close + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["block"] = TriState.closing; + obj_type_status["curly_block"] = TriState.off; + } else { + debug(block) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build block array (or string) + } + } else if (obj_type_status["tic_block"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(block) { + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key].stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["block"] = TriState.closing; + obj_type_status["tic_block"] = TriState.off; + } else { + debug(block) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build block array (or string) + } + } + } + void _poem_block_(L,O,T,C,N,CMM)( + L line, + return ref O an_object, + return ref T obj_type_status, + return ref C cntr, + N object_number_poem, + CMM conf_make_meta, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + static assert(is(typeof(cntr) == int)); + static assert(is(typeof(object_number_poem) == string[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_poem"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_poem_close)) { + if (an_object_key in an_object + || processing.length > 0) { + an_object[an_object_key] = ""; + debug(poem) { + writefln( + "* [poem curly] %s", + line + ); + } + if (processing.length > 0) { + an_object[an_object_key] = processing["verse"]; + } + debug(poem) { + writeln(__LINE__); + writefln( + "* %s %s", + obj_cite_digits.on, + line + ); + } + if (an_object.length > 0) { + debug(poem) { // poem (curly) close + writeln( + obj_cite_digits.on, + an_object[an_object_key] + ); + } + an_object["is"] = "verse"; + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "verse"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + } + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + object_number_poem["end"] = obj_cite_digits.on.to!string; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["poem"] = TriState.closing; + obj_type_status["curly_poem"] = TriState.off; + } else { + processing["verse"] ~= line ~= "\n"; + if (obj_type_status["verse_new"] == State.on) { + obj_cite_digits = ocn_emit(OCNstatus.on); + obj_type_status["verse_new"] = State.off; + } else if (line.matchFirst(rgx.newline_eol_delimiter_only)) { + processing["verse"] = processing["verse"].stripRight; + verse_line = TriState.off; + obj_type_status["verse_new"] = State.on; + } + if (obj_type_status["verse_new"] == State.on) { + verse_line=1; + an_object[an_object_key] = processing["verse"]; + debug(poem) { // poem verse + writefln( + "* %s curly\n%s", + obj_cite_digits.on, + an_object[an_object_key] + ); + } + processing.remove("verse"); + an_object["is"] = "verse"; + auto comp_obj_location = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "verse"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + } + } else if (obj_type_status["tic_poem"] == TriState.on) { + if (auto m = line.matchFirst(rgx.block_tic_close)) { // tic_poem_close + an_object[an_object_key]="verse"; + debug(poem) { + writefln( + "* [poem tic] %s", + line + ); + } + if (processing.length > 0) { + an_object[an_object_key] = processing["verse"]; + } + if (an_object.length > 0) { + debug(poem) { // poem (tic) close + writeln(__LINE__); + writeln(obj_cite_digits.on, line); + } + processing.remove("verse"); + an_object["is"] = "verse"; + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "verse"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + object_number_poem["end"] = obj_cite_digits.on.to!string; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + obj_type_status["blocks"] = TriState.closing; + obj_type_status["poem"] = TriState.closing; + obj_type_status["tic_poem"] = TriState.off; + } else { + processing["verse"] ~= line ~= "\n"; + if (obj_type_status["verse_new"] == State.on) { + obj_cite_digits = ocn_emit(OCNstatus.on); + obj_type_status["verse_new"] = State.off; + } else if (line.matchFirst(rgx.newline_eol_delimiter_only)) { + processing["verse"] = processing["verse"].stripRight; + obj_type_status["verse_new"] = State.on; + verse_line = TriState.off; + } + if (obj_type_status["verse_new"] == State.on) { + verse_line=1; + an_object[an_object_key] = processing["verse"]; + debug(poem) { // poem (tic) close + writefln( + "* %s tic\n%s", + obj_cite_digits.on, + an_object[an_object_key] + ); + } + processing.remove("verse"); + an_object["is"] = "verse"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "verse"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + } + } + } + void _code_block_(L,O,T)( + L line, + return ref O an_object, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_code"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_code_close)) { + debug(code) { // code (curly) close + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key] + .replaceFirst(rgx.newline_eol_delimiter_only, "") + .stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["code"] = TriState.closing; + obj_type_status["curly_code"] = TriState.off; + } else { + debug(code) { // code (curly) line + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // code (curly) line + } + } else if (obj_type_status["tic_code"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(code) { // code (tic) close + writeln(line); + } + an_object[an_object_key] = an_object[an_object_key] + .replaceFirst(rgx.newline_eol_delimiter_only, "") + .stripRight; + obj_type_status["blocks"] = TriState.closing; + obj_type_status["code"] = TriState.closing; + obj_type_status["tic_code"] = TriState.off; + } else { + debug(code) { // code (tic) line + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // code (tic) line + } + } + } + void _table_block_(L,O,T,CMM)( + L line, + return ref O an_object, + return ref T obj_type_status, + return ref CMM conf_make_meta, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (obj_type_status["curly_table"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_table_close)) { + debug(table) { // table (curly) close + writeln(line); + } + obj_type_status["blocks"] = TriState.closing; + obj_type_status["table"] = TriState.closing; + obj_type_status["curly_table"] = TriState.off; + } else { + debug(table) { // table + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build table array (or string) + } + } else if (obj_type_status["curly_table_special_markup"] == TriState.on) { + if (line.empty) { + obj_type_status["blocks"] = TriState.off; + obj_type_status["table"] = TriState.off; + obj_type_status["curly_table_special_markup"] = TriState.off; + _table_closed_make_special_notation_table_( + line, + an_object, + the_document_body_section, + obj_cite_digits, + comp_obj_heading, + cntr, + obj_type_status, + conf_make_meta, + ); + } else { + debug(table) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; + } + } else if (obj_type_status["tic_table"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(table) { // table (tic) close + writeln(line); + } + obj_type_status["blocks"] = TriState.closing; + obj_type_status["table"] = TriState.closing; + obj_type_status["tic_table"] = TriState.off; + } else { + debug(table) { // table + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build table array (or string) + } + } + } + final string biblio_tag_map(A)(A abr) { + debug(asserts) { + static assert(is(typeof(abr) == string)); + } + auto btm = [ + "au" : "author_raw", + "ed" : "editor_raw", + "ti" : "fulltitle", + "lng" : "language", + "jo" : "journal", + "vol" : "volume", + "edn" : "edition", + "yr" : "year", + "pl" : "place", + "pb" : "publisher", + "pub" : "publisher", + "pg" : "pages", + "pgs" : "pages", + "sn" : "short_name" + ]; + return btm[abr]; + } + void _biblio_block_( + char[] line, + return ref int[string] obj_type_status, + return ref int bib_entry, + return ref string biblio_entry_str_json, + return ref string[] biblio_arr_json + ) { + mixin SiSUbiblio; + auto jsn = BibJsnStr(); + static auto rgx = Rgx(); + if (line.matchFirst(rgx.heading_biblio)) { + obj_type_status["biblio_section"] = TriState.on; + obj_type_status["blurb_section"] = State.off; + obj_type_status["glossary_section"] = State.off; + } + if (line.empty) { + debug { + debug(biblioblock) { + writeln("---"); + } + debug(biblioblockinclude) { + writeln(biblio_entry_str_json.length); + } + } + if ((bib_entry == State.off) + && (biblio_entry_str_json.empty)) { + bib_entry = State.on; + biblio_entry_str_json = jsn.biblio_entry_tags_jsonstr; + } else if (!(biblio_entry_str_json.empty)) { + bib_entry = State.off; + if (!(biblio_entry_str_json == jsn.biblio_entry_tags_jsonstr)) { + auto biblio_entry = parseJSON(biblio_entry_str_json); + if (biblio_entry["fulltitle"].str.empty) { + writeln("check problem entry (Title missing): ", biblio_entry_str_json); + } else if ((biblio_entry["author_raw"].str.empty) && (biblio_entry["editor_raw"].str.empty)) { + writeln("check problem entry (No author and no editor): ", biblio_entry_str_json); + } else { + biblio_arr_json ~= biblio_entry_str_json; + } + biblio_entry_str_json = jsn.biblio_entry_tags_jsonstr; + } + } else { // CHECK ERROR + writeln("?? 2. ERROR ", biblio_entry_str_json, "??"); + biblio_entry_str_json = ""; + } + } else if (line.matchFirst(rgx.biblio_tags)) { + debug(biblioblock) { + writeln(line); + } + auto bt = line.match(rgx.biblio_tags); + bib_entry = State.off; + st = bt.captures[1].to!string; + auto header_tag_value=(bt.captures[2]).to!string; + JSONValue j = parseJSON(biblio_entry_str_json); + biblio_tag_name = (st.match(rgx.biblio_abbreviations)) + ? (biblio_tag_map(st)) + : st; + j.object[biblio_tag_name] = header_tag_value; + debug(bibliounsortedcheckduplicates) { + writeln(biblio_tag_name, ": ", header_tag_value); + writeln("--"); + } + switch (biblio_tag_name) { + case "author_raw": // author_arr author (fn sn) + j["author_arr"] + = header_tag_value.split(rgx.arr_delimiter); + string tmp; + biblioAuthorLoop: + foreach (au; j["author_arr"].array) { + if (auto x = au.str.match(rgx.name_delimiter)) { + tmp ~= x.captures[2] ~ " " ~ x.captures[1] ~ ", "; + } else { + tmp ~= au.str; + } + } + tmp = (tmp).replace(rgx.trailing_comma, ""); + j["author"].str = tmp; + goto default; + case "editor_raw": // editor_arr editor (fn sn) + j["editor_arr"] + = header_tag_value.split(rgx.arr_delimiter); + string tmp; + biblioEditorLoop: + foreach (ed; j["editor_arr"].array) { + if (auto x = ed.str.match(rgx.name_delimiter)) { + tmp ~= x.captures[2] ~ " " ~ x.captures[1] ~ ", "; + } else { + tmp ~= ed.str; + } + } + tmp = (tmp).replace(rgx.trailing_comma, ""); + j["editor"].str = tmp; + goto default; + case "fulltitle": // title & subtitle + goto default; + default: + break; + } + auto s = j.toString(); + debug(biblio1) { + writefln( + "* %s: %s\n%s", + biblio_tag_name, + biblio_tag_entry, + j[biblio_tag_name] + ); + } + if (line.match(rgx.comment)) { + writeln("ERROR", line, "COMMENT"); + writeln("ERROR", s, "%%"); + } + if (!(match(line, rgx.comment))) { + debug(biblioblockinclude) { + writeln(line); + } + biblio_entry_str_json = s; + } else { + biblio_entry_str_json = ""; + } + header_tag_value=""; + } + } + void _table_closed_make_special_notation_table_(N,CMM)( + char[] line, + return ref string[string] an_object, + return ref ObjGenericComposite[] the_document_body_section, + return ref N obj_cite_digits, + return ref ObjGenericComposite _comp_obj_heading, + return ref int cntr, + return ref int[string] obj_type_status, + CMM conf_make_meta + ) { + comp_obj_block = comp_obj_block.init; + obj_cite_digits = ocn_emit(OCNstatus.on); + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + "table" + ); + an_object["is"] = "table"; + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, "body_nugget", conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block = table_instructions(comp_obj_block, an_object["table_head"]); + comp_obj_block = table_substantive_munge_special(comp_obj_block, an_object["substantive"]); + the_document_body_section ~= comp_obj_block; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + void _block_flag_line_empty_(B,N,CMM)( + B bookindex_extract_hash, + char[] line, + return ref string[string] an_object, + return ref ObjGenericComposite[] the_document_body_section, + return ref string[][string][string] bookindex_unordered_hashes, + return ref N obj_cite_digits, + return ref ObjGenericComposite _comp_obj_heading, + return ref int cntr, + return ref int[string] obj_type_status, + string[string] object_number_poem, + CMM conf_make_meta, + ) { + assert( + line.empty, + "\nline should be empty:\n \"" + ~ line ~ "\"" + ); + assert( + (obj_type_status["blocks"] == TriState.closing), + "code block status: closed" + ); + assertions_flag_types_block_status_none_or_closed(obj_type_status); + if (obj_type_status["quote"] == TriState.closing) { + obj_cite_digits = ocn_emit(OCNstatus.on); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "quote"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "quote"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digit_type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + obj_type_status["blocks"] = TriState.off; + obj_type_status["quote"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (obj_type_status["group"] == TriState.closing) { + obj_cite_digits = ocn_emit(OCNstatus.on); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "group"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "group"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + obj_type_status["blocks"] = TriState.off; + obj_type_status["group"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (obj_type_status["block"] == TriState.closing) { + obj_cite_digits = ocn_emit(OCNstatus.on); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "block"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.is_of_part = "body"; + comp_obj_block.metainfo.is_of_section = "body"; + comp_obj_block.metainfo.is_of_type = "block"; + comp_obj_block.metainfo.is_a = "block"; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digit_type; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + obj_type_status["blocks"] = TriState.off; + obj_type_status["block"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (obj_type_status["poem"] == TriState.closing) { + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "verse"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + comp_obj_poem_ocn = comp_obj_poem_ocn.init; + comp_obj_poem_ocn.metainfo.is_of_part = "body"; + comp_obj_poem_ocn.metainfo.is_of_section = "body"; + comp_obj_poem_ocn.metainfo.is_of_type = "block"; + comp_obj_poem_ocn.metainfo.is_a = "poem"; + comp_obj_poem_ocn.metainfo.ocn = obj_cite_digits.on; + comp_obj_poem_ocn.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; // + comp_obj_poem_ocn.metainfo.o_n_book_index = obj_cite_digits.bkidx; // + comp_obj_poem_ocn.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_poem_ocn.text = ""; + the_document_body_section ~= comp_obj_poem_ocn; + obj_type_status["blocks"] = TriState.off; + obj_type_status["poem"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + } else if (obj_type_status["code"] == TriState.closing) { + obj_cite_digits = ocn_emit(OCNstatus.on); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "code"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + anchor_tags = substantive_obj_misc_tuple[sObj.anchor_tags]; + comp_obj_code = comp_obj_code.init; + comp_obj_code.metainfo.is_of_part = "body"; + comp_obj_code.metainfo.is_of_section = "body"; + comp_obj_code.metainfo.is_of_type = "block"; + comp_obj_code.metainfo.is_a = "code"; + comp_obj_code.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_code.text = an_object["substantive"]; + comp_obj_code.has.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_code.has.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_code.has.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_code; + obj_type_status["blocks"] = TriState.off; + obj_type_status["code"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (obj_type_status["table"] == TriState.closing) { + comp_obj_block = comp_obj_block.init; + obj_cite_digits = ocn_emit(OCNstatus.on); + an_object["bookindex_nugget"] + = ("bookindex_nugget" in an_object) ? an_object["bookindex_nugget"] : ""; + bookindex_unordered_hashes + = bookindex_extract_hash.bookindex_nugget_hash( + an_object["bookindex_nugget"], + obj_cite_digits, + segment_anchor_tag_that_object_belongs_to + ); + an_object["is"] = "table"; + auto comp_obj_location + = node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_digits, + cntr, + heading_ptr-1, + an_object["is"] + ); + auto substantive_obj_misc_tuple + = obj_im.obj_inline_markup_and_anchor_tags_and_misc(an_object, an_object_key, conf_make_meta, false); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.metainfo.ocn = obj_cite_digits.on; + comp_obj_block.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + comp_obj_block.metainfo.o_n_book_index = obj_cite_digits.bkidx; + comp_obj_block.metainfo.object_number_type = obj_cite_digits.type; + comp_obj_block = table_instructions(comp_obj_block, an_object["table_head"]); + comp_obj_block = table_substantive_munge(comp_obj_block, an_object["substantive"]); + the_document_body_section ~= comp_obj_block; + obj_type_status["blocks"] = TriState.off; + obj_type_status["table"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + } + auto _book_index_(L,I,O,T,B)( + L line, + return ref I book_idx_tmp, + return ref O an_object, + return ref T obj_type_status, + B opt_action, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(book_idx_tmp) == string)); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (auto m = line.match(rgx.book_index)) { + /+ match book_index +/ + debug(bookindexmatch) { + writefln( + "* [bookindex] %s\n", + m.captures[1].to!string, + ); + } + an_object["bookindex_nugget"] = m.captures[1].to!string; + } else if (auto m = line.match(rgx.book_index_open)) { + /+ match open book_index +/ + obj_type_status["book_index"] = State.on; + if (opt_action.backmatter && opt_action.section_bookindex) { + book_idx_tmp = m.captures[1].to!string; + debug(bookindexmatch) { + writefln( + "* [bookindex] %s\n", + book_idx_tmp, + ); + } + } + } else if (obj_type_status["book_index"] == State.on ) { + /+ book_index flag set +/ + if (auto m = line.match(rgx.book_index_close)) { + obj_type_status["book_index"] = State.off; + if (opt_action.backmatter + && opt_action.section_bookindex) { + an_object["bookindex_nugget"] = book_idx_tmp ~ m.captures[1].to!string; + debug(bookindexmatch) { + writefln( + "* [bookindex] %s\n", + book_idx_tmp, + ); + } + } + book_idx_tmp = ""; + } else { + if (opt_action.backmatter + && opt_action.section_bookindex) { + book_idx_tmp ~= line; + } + } + } + } + auto _heading_found_(L,X,H,R,T)( + L line, + X _make_unmarked_headings, + return ref H heading_match_str, + return ref R heading_match_rgx, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(_make_unmarked_headings) == string[])); + static assert(is(typeof(heading_match_str) == string[string])); + static assert(is(typeof(heading_match_rgx) == Regex!(char)[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if ((_make_unmarked_headings.length > 2) + && (obj_type_status["make_headings"] == State.off)) { + /+ headings found +/ + debug(headingsfound) { + writeln(_make_unmarked_headings); + } + debug(headingsfound) { + writeln(_make_unmarked_headings.length); + writeln(_make_unmarked_headings); + } + switch (_make_unmarked_headings.length) { + case 7 : + if (!empty(_make_unmarked_headings[6])) { + heading_match_str["h_4"] + = "^(" ~ _make_unmarked_headings[6].to!string ~ ")"; + heading_match_rgx["h_4"] + = regex(heading_match_str["h_4"]); + } + goto case; + case 6 : + if (!empty(_make_unmarked_headings[5])) { + heading_match_str["h_3"] + = "^(" ~ _make_unmarked_headings[5].to!string ~ ")"; + heading_match_rgx["h_3"] + = regex(heading_match_str["h_3"]); + } + goto case; + case 5 : + if (!empty(_make_unmarked_headings[4])) { + heading_match_str["h_2"] + = "^(" ~ _make_unmarked_headings[4].to!string ~ ")"; + heading_match_rgx["h_2"] + = regex(heading_match_str["h_2"]); + } + goto case; + case 4 : + if (!empty(_make_unmarked_headings[3])) { + heading_match_str["h_1"] + = "^(" ~ _make_unmarked_headings[3].to!string ~ ")"; + heading_match_rgx["h_1"] + = regex(heading_match_str["h_1"]); + } + goto case; + case 3 : + if (!empty(_make_unmarked_headings[2])) { + heading_match_str["h_D"] + = "^(" ~ _make_unmarked_headings[2].to!string ~ ")"; + heading_match_rgx["h_D"] + = regex(heading_match_str["h_D"]); + } + goto case; + case 2 : + if (!empty(_make_unmarked_headings[1])) { + heading_match_str["h_C"] + = "^(" ~ _make_unmarked_headings[1].to!string ~ ")"; + heading_match_rgx["h_C"] + = regex(heading_match_str["h_C"]); + } + goto case; + case 1 : + if (!empty(_make_unmarked_headings[0])) { + heading_match_str["h_B"] + = "^(" ~ _make_unmarked_headings[0].to!string ~ ")"; + heading_match_rgx["h_B"] + = regex(heading_match_str["h_B"]); + } + break; + default: + break; + } + obj_type_status["make_headings"] = State.on; + } + } + auto _heading_make_set_(L,C,R,T)( + L line, + C line_occur, + return ref R heading_match_rgx, + return ref T obj_type_status + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(line_occur) == int[string])); + static assert(is(typeof(heading_match_rgx) == Regex!(char)[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + if ((obj_type_status["make_headings"] == State.on) + && ((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((obj_type_status["para"] == State.off) + && (obj_type_status["heading"] == State.off))) { + /+ heading make set +/ + if (line.matchFirst(heading_match_rgx["h_B"])) { + line = "B~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_C"])) { + line = "C~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_D"])) { + line = "D~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_1"])) { + line = "1~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_2"])) { + line = "2~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_3"])) { + line = "3~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + if (line.matchFirst(heading_match_rgx["h_4"])) { + line = "4~ " ~ line; + debug(headingsfound) { + writeln(line); + } + } + } + return line; + } + auto _heading_matched_(L,C,O,K,Lv,Lc,T,CMM)( + L line, + return ref C line_occur, + return ref O an_object, + return ref K an_object_key, + return ref Lv lv, + return ref Lc collapsed_lev, + return ref T obj_type_status, + return ref CMM conf_make_meta, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(line_occur) == int[string])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(an_object_key) == string)); + static assert(is(typeof(lv) == int[string])); + static assert(is(typeof(collapsed_lev) == int[string])); + static assert(is(typeof(obj_type_status) == int[string])); + } + static auto rgx = Rgx(); + if (auto m = line.match(rgx.heading)) { + /+ heading match +/ + ++line_occur["heading"]; + obj_type_status["heading"] = State.on; + obj_type_status["para"] = State.off; + if (line.match(rgx.heading_seg_and_above)) { + obj_type_status["biblio_section"] = State.off; + obj_type_status["glossary_section"] = State.off; + obj_type_status["blurb_section"] = State.off; + } + an_object[an_object_key] ~= line ~= "\n"; + an_object["lev"] ~= m.captures[1]; + assertions_doc_structure(an_object, lv); // includes most of the logic for collapsed levels + switch (an_object["lev"]) { + case "A": // Title set + an_object[an_object_key]=(an_object[an_object_key]) + .replaceFirst(rgx.variable_doc_title, + (conf_make_meta.meta.title_full ~ ",")) + .replaceFirst(rgx.variable_doc_author, + conf_make_meta.meta.creator_author); + collapsed_lev["h0"] = 0; + an_object["lev_collapsed_number"] + = collapsed_lev["h0"].to!string; + lv["lv"] = DocStructMarkupHeading.h_sect_A; + ++lv["h0"]; + lv["h1"] = State.off; + lv["h2"] = State.off; + lv["h3"] = State.off; + lv["h4"] = State.off; + lv["h5"] = State.off; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "B": + collapsed_lev["h1"] = collapsed_lev["h0"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h1"].to!string; + lv["lv"] = DocStructMarkupHeading.h_sect_B; + ++lv["h1"]; + lv["h2"] = State.off; + lv["h3"] = State.off; + lv["h4"] = State.off; + lv["h5"] = State.off; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "C": + collapsed_lev["h2"] = collapsed_lev["h1"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h2"].to!string; + lv["lv"] = DocStructMarkupHeading.h_sect_C; + ++lv["h2"]; + lv["h3"] = State.off; + lv["h4"] = State.off; + lv["h5"] = State.off; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "D": + collapsed_lev["h3"] = collapsed_lev["h2"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h3"].to!string; + lv["lv"] = DocStructMarkupHeading.h_sect_D; + ++lv["h3"]; + lv["h4"] = State.off; + lv["h5"] = State.off; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "1": + if (lv["h3"] > State.off) { + collapsed_lev["h4"] = collapsed_lev["h3"] + 1; + } else if (lv["h2"] > State.off) { + collapsed_lev["h4"] = collapsed_lev["h2"] + 1; + } else if (lv["h1"] > State.off) { + collapsed_lev["h4"] = collapsed_lev["h1"] + 1; + } else if (lv["h0"] > State.off) { + collapsed_lev["h4"] = collapsed_lev["h0"] + 1; + } + an_object["lev_collapsed_number"] + = collapsed_lev["h4"].to!string; + lv["lv"] = DocStructMarkupHeading.h_text_1; + ++lv["h4"]; + lv["h5"] = State.off; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "2": + if (lv["h5"] > State.off) { + an_object["lev_collapsed_number"] + = collapsed_lev["h5"].to!string; + } else if (lv["h4"] > State.off) { + collapsed_lev["h5"] = collapsed_lev["h4"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h5"].to!string; + } + lv["lv"] = DocStructMarkupHeading.h_text_2; + ++lv["h5"]; + lv["h6"] = State.off; + lv["h7"] = State.off; + goto default; + case "3": + if (lv["h6"] > State.off) { + an_object["lev_collapsed_number"] + = collapsed_lev["h6"].to!string; + } else if (lv["h5"] > State.off) { + collapsed_lev["h6"] = collapsed_lev["h5"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h6"].to!string; + } + lv["lv"] = DocStructMarkupHeading.h_text_3; + ++lv["h6"]; + lv["h7"] = State.off; + goto default; + case "4": + if (lv["h7"] > State.off) { + an_object["lev_collapsed_number"] + = collapsed_lev["h7"].to!string; + } else if (lv["h6"] > State.off) { + collapsed_lev["h7"] = collapsed_lev["h6"] + 1; + an_object["lev_collapsed_number"] + = collapsed_lev["h7"].to!string; + } + lv["lv"] = DocStructMarkupHeading.h_text_4; + ++lv["h7"]; + goto default; + default: + an_object["lev_markup_number"] = lv["lv"].to!string; + } + debug(heading) { + writeln(line.strip); + } + } + } + void _para_match_(L,O,K,I,B,T,C)( + L line, + return ref O an_object, + return ref K an_object_key, + return ref I indent, + return ref B bullet, + return ref T obj_type_status, + return ref C line_occur, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(an_object_key) == string)); + static assert(is(typeof(indent) == int[string])); + static assert(is(typeof(bullet) == bool)); + static assert(is(typeof(obj_type_status) == int[string])); + static assert(is(typeof(line_occur) == int[string])); + } + static auto rgx = Rgx(); + if (line_occur["para"] == State.off) { + line = font_faces_line(line); + /+ para matches +/ + obj_type_status["para"] = State.on; + an_object[an_object_key] ~= line; + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + if (auto m = line.matchFirst(rgx.para_indent)) { + debug(paraindent) { + writeln(line); + } + indent["hang_position"] = (m.captures[1]).to!int; + indent["base_position"] = 0; + } else if (line.matchFirst(rgx.para_bullet)) { + debug(parabullet) { + writeln(line); + } + bullet = true; + } else if (auto m = line.matchFirst(rgx.para_indent_hang)) { + debug(paraindenthang) { + writeln(line); + } + indent=[ + "hang_position" : (m.captures[1]).to!int, + "base_position" : (m.captures[2]).to!int, + ]; + } else if (auto m = line.matchFirst(rgx.para_bullet_indent)) { + debug(parabulletindent) { + writeln(line); + } + indent=[ + "hang_position" : (m.captures[1]).to!int, + "base_position" : 0, + ]; + bullet = true; + } + ++line_occur["para"]; + } + } + auto font_faces_line(T)( + T textline, + ) { + static auto rgx = Rgx(); + if (textline.match(rgx.inline_faces_line)) { + textline = (textline) + .replaceFirst(rgx.inline_emphasis_line, ("*{$1}*$2")) + .replaceFirst(rgx.inline_bold_line, ("!{$1}!$2")) + .replaceFirst(rgx.inline_underscore_line, ("_{$1}_$2")) + .replaceFirst(rgx.inline_italics_line, ("/{$1}/$2")); + } + return textline; + } + auto table_instructions(O,H)( + return ref O table_object, + return ref H table_head, + ) { + static auto rgx = Rgx(); + table_object.metainfo.is_of_part = "body"; + table_object.metainfo.is_of_section = "body"; + table_object.metainfo.is_of_type = "block"; + table_object.metainfo.is_a = "table"; + table_object.has.inline_notes_reg = false; + table_object.has.inline_notes_star = false; + table_object.has.inline_links = false; + if (auto m = table_head.matchFirst(rgx.table_head_instructions)) { + table_object.table.heading = ((m["c_heading"].length > 0) && (m["c_heading"] == "h")) ? true : false; + table_object.table.number_of_columns = ((m["c_num"].length > 0) && (m["c_num"].to!int > 0)) ? m["c_num"].to!int : 0; // double check, may be obsolete + foreach (cw; m["c_widths"].matchAll(rgx.table_col_widths)) { + auto x = cw.hit.matchFirst(rgx.table_col_widths_and_alignment); + table_object.table.column_widths ~= x["width"].to!int; + table_object.table.column_aligns ~= (x["align"].empty) ? "" : x["align"]; + } + } + return table_object; + } + auto table_array_munge(O,T)( + return ref O table_object, + return ref T table_array, + ) { + static auto rgx = Rgx(); + static auto mng = InlineMarkup(); + string _table_substantive; + ulong col_num; + ulong col_num_; + ulong col_num_chk = 0; + foreach(idx_r, row; table_array) { + debug(table_dev) { + writeln("row ", idx_r); + } + col_num_ = 0; + if (col_num == 0 + || col_num < row.length) { + col_num = row.length; + } + if (col_num_chk == 0) { + col_num_chk = col_num; + } else if (col_num == 1) { + debug(table_dev) { + writeln("table note: "); + } + } else if (col_num_chk != col_num) { + debug(table_dev) { + writeln("warning irregular number of columns: ", col_num_chk, " != ", col_num); + } + } else { + } + foreach(idx_c, col; row) { + debug(table_dev) { + write(idx_c, ", "); + } + col_num_ = idx_c; + _table_substantive ~= col ~ mng.tc_s; + if (idx_r == 0 && comp_obj_block.table.heading) { + } else if (col.match(rgx.numeric_col) && idx_r == 1) { // conditions reversed to avoid: gdc compiled program run segfault + if ((comp_obj_block.table.column_aligns.length > idx_c) + && (comp_obj_block.table.column_aligns[idx_c].matchFirst(rgx.table_col_align_match))) { + comp_obj_block.table.column_aligns[idx_c] + = comp_obj_block.table.column_aligns[idx_c]; + } else if (comp_obj_block.table.column_aligns.length > idx_c) { + comp_obj_block.table.column_aligns[idx_c] = "r"; + } else { + comp_obj_block.table.column_aligns ~= "r"; + } + } else if (idx_r == 1) { + if ((comp_obj_block.table.column_aligns.length > idx_c) + && (comp_obj_block.table.column_aligns[idx_c].matchFirst(rgx.table_col_align_match))) { + comp_obj_block.table.column_aligns[idx_c] + = comp_obj_block.table.column_aligns[idx_c]; + } else if (comp_obj_block.table.column_aligns.length > idx_c) { + comp_obj_block.table.column_aligns[idx_c] = "l"; + } else { + comp_obj_block.table.column_aligns ~= "l"; + } + } + } + debug(table_dev) { + writeln(""); + } + if (col_num_chk > 0 && (col_num != col_num_chk)) { + } else if (col_num == col_num_chk){ + } else { + col_num_chk = col_num; + } + _table_substantive = _table_substantive.replaceFirst(rgx.table_col_separator_nl, "\n"); + } + if (comp_obj_block.table.number_of_columns != col_num) { + if (comp_obj_block.table.number_of_columns == 0) { + comp_obj_block.table.number_of_columns = (col_num).to!int; + } else { + debug(table_dev) { + writeln(comp_obj_block.table.number_of_columns, " != ", col_num); + } + } + } + if (table_object.table.number_of_columns == 0 + && table_object.table.column_widths.length > 0) { + writeln(__LINE__, " ERROR"); + } + if (table_object.table.number_of_columns > 0 + && table_object.table.column_widths.length == 0) { + double col_w = (100.00 / table_object.table.number_of_columns); + foreach (i; 0..table_object.table.number_of_columns) { + table_object.table.column_widths ~= col_w; + } + } else if (table_object.table.number_of_columns + != table_object.table.column_widths.length) { + debug(table_dev) { + writeln(m.hit); // further logic required + } + if (table_object.table.number_of_columns > table_object.table.column_widths.length) { + double col_w = (100.00 - (table_object.table.column_widths).sum) + / (table_object.table.number_of_columns - table_object.table.column_widths.length); + foreach (i; 0..table_object.table.column_widths.length) { + table_object.table.column_widths ~= col_w; + } + foreach (i; 0..(table_object.table.number_of_columns - table_object.table.column_widths.length)) { + table_object.table.column_widths ~= col_w; + } + } else if (table_object.table.number_of_columns < table_object.table.column_widths.length) { + writeln(__LINE__, " warning, ERROR"); + } + } + if (table_object.table.column_widths.sum > 101 + || table_object.table.column_widths.sum < 95 ) { + writeln("sum: ", table_object.table.column_widths.sum, + ", array: ", table_object.table.column_widths, + ", cols: ", table_object.table.number_of_columns); + writeln(_table_substantive); + } + debug(table_res) { + writeln("aligns: ", comp_obj_block.table.column_aligns, "\n", + "no. of columns: ", comp_obj_block.table.number_of_columns, "\n", + "col widths: ", comp_obj_block.table.column_widths, + " sum: ", comp_obj_block.table.column_widths.sum, "\n", + _table_substantive); + } + comp_obj_block.text = _table_substantive; + return table_object; + } + auto table_array_munge_open_close(O,T)( + return ref O table_object, + return ref T table_array, + ) { + static auto rgx = Rgx(); + static auto mng = InlineMarkup(); + string _table_substantive; + foreach(row; table_array) { + foreach(col; row) { + _table_substantive ~= mng.tc_o ~ col ~ mng.tc_c; + } + _table_substantive ~= "\n"; + } + debug(table_dev) { + writeln(_table_substantive); + } + comp_obj_block.text = _table_substantive; + return table_object; + } + auto table_substantive_munge(O,T)( + return ref O table_object, + return ref T table_substantive, + ) { + static auto rgx = Rgx(); + static auto munge = ObjInlineMarkupMunge(); + string[] _table_rows = (table_substantive).split(rgx.table_row_delimiter); + string[] _table_cols; + string[][] _table; + foreach(col; _table_rows) { + _table_cols = col.split(rgx.table_col_delimiter); + _table ~= _table_cols; + } + table_object = table_array_munge(table_object, _table); + return table_object; + } + auto table_substantive_munge_special(O,T)( + return ref O table_object, + return ref T table_substantive, + ) { + static auto rgx = Rgx(); + static auto munge = ObjInlineMarkupMunge(); + string[] _table_rows = (table_substantive).split(rgx.table_row_delimiter_special); + string[] _table_cols; + string[][] _table; + foreach(col; _table_rows) { + _table_cols = col.split(rgx.table_col_delimiter_special); + _table ~= _table_cols; + } + table_object = table_array_munge(table_object, _table); + return table_object; + } + /+ abstraction functions ↑ +/ + /+ ↓ abstraction function emitters +/ + pure struct OCNemitter { + int ocn_, ocn_on, ocn_on_, ocn_off, ocn_off_, ocn_bkidx, ocn_bkidx_; + auto ocn_emitter(int ocn_status_flag) { + OCNset ocn; + assert(ocn_status_flag <= OCNstatus.reset); + ocn_on = ocn_off = ocn_bkidx = 0; + switch(ocn_status_flag) { + case OCNstatus.reset: + ocn_ = ocn_on_ = ocn_off_ = 1; + ocn_bkidx_ = 0; + break; + case OCNstatus.on: + ocn_ = ocn_on = ++ocn_on_; + break; + case OCNstatus.off: + ocn_ = ocn_off = ++ocn_off_; + break; + case OCNstatus.bkidx: + ocn_ = ocn_bkidx = ++ocn_bkidx_; + break; + case OCNstatus.closing: + break; + default: + ocn_ = 0; + } + assert(ocn_ >= 0); + ocn.digit = ocn_; + ocn.on = ocn_on; + ocn.off = ocn_off; + ocn.bkidx = ocn_bkidx; + ocn.type = ocn_status_flag; + return ocn; + } + invariant() { + } + } + /+ +/ + static struct ObjInlineMarkupMunge { + string[string] obj_txt; + int n_foot, n_foot_reg, n_foot_sp_asterisk, n_foot_sp_plus; + string asterisks_, plus_; + string obj_txt_out, tail, note; + static auto rgx = Rgx(); + static auto mkup = InlineMarkup(); + int stage_reset_note_numbers = true; + private auto initialize_note_numbers() { + n_foot = 0; + n_foot_reg = 0; + n_foot_sp_asterisk = 0; + n_foot_sp_plus = 0; + } + static auto url_links(Ot)(Ot obj_txt_in) { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + static auto mng = InlineMarkup(); + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_mono, (mng.mono ~ "{$1}" ~ mng.mono)); + /+ url matched +/ + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_notes_al_special, ""); // TODO reinstate when special footnotes are implemented + if (obj_txt_in.match(rgx.smid_inline_url_generic)) { + /+ link: naked url: http://url +/ + if (obj_txt_in.match(rgx.smid_inline_link_naked_url)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.smid_inline_link_naked_url, + ("$1" + ~ mkup.lnk_o ~ "$2" ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ "$3") // ("$1{ $2 }$2$3") + ); + } + /+ link with helper for endnote including the url: + {~^ link which includes url as footnote }http://url + maps to: + { link which includes url as footnote }http://url~{ { http://url }http://url }~ + +/ + if (obj_txt_in.match(rgx.smid_inline_link_endnote_url_helper)) { + obj_txt_in = (obj_txt_in) + .replaceAll( + rgx.smid_inline_link_endnote_url_helper_punctuated, + (mkup.lnk_o ~ "$1" ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ " }~$3") // ("{ $1 }$2~{ { $2 }$2 }~$3") + ) + .replaceAll( + rgx.smid_inline_link_endnote_url_helper, + (mkup.lnk_o ~ "$1" ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ " }~") // ("{ $1 }$2~{ { $2 }$2 }~") + ); + } + /+ link with regular markup: + { linked text or image }http://url + +/ + if (obj_txt_in.match(rgx.smid_inline_link_markup_regular)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.smid_inline_link_markup_regular, + ("$1" + ~ mkup.lnk_o ~ "$2" ~ mkup.lnk_c + ~ mkup.url_o ~ "$3" ~ mkup.url_c + ~ "$4") // ("$1{ $2 }$3$4") + ); + } + } + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_mono_box, ("#{$1}#")); + return obj_txt_in; + } + static auto images(Ot)(Ot obj_txt_in) { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + static auto mng = InlineMarkup(); + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_mono, (mng.mono ~ "{$1}" ~ mng.mono)); // figure + /+ url matched +/ + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_notes_al_special, ""); // TODO reinstate when special footnotes are implemented + if (obj_txt_in.match(rgx.smid_image_generic)) { /+ images with and without links +/ + debug(images) { + writeln("Image: ", obj_txt_in); + } + if (obj_txt_in.match(rgx.smid_image_with_dimensions)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.smid_image_with_dimensions, + ("$1" + ~ mkup.img ~ "$2,w$3h$4 " + ~ "$5") // ("$1{ $2 }$2$3") + ); + debug(images) { + writeln("IMAGE with size: ", obj_txt_in); // decide on representation + } + } else if (obj_txt_in.match(rgx.smid_image)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.smid_image, + ("$1" + ~ mkup.img ~ "$2,w0h0 " + ~ "$3") // ("$1{ $2 }$2$3") + ); + debug(images) { + writeln("IMAGE: ", obj_txt_in); // decide on representation + } + } + } + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_mono_box, ("#{$1}#")); // figure + return obj_txt_in; + } + auto footnotes_endnotes_markup_and_number_or_stars(Ot)(Ot obj_txt_in, bool reset_note_numbers) { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + /+ endnotes (regular) +/ + bool flg_notes_reg = false; + bool flg_notes_star = false; + bool flg_notes_plus = false; + obj_txt_in = (obj_txt_in).replaceAll( + rgx.inline_notes_curly, + (mkup.en_a_o ~ " $1" ~ mkup.en_a_c) + ); + if (!(stage_reset_note_numbers) && reset_note_numbers) { + stage_reset_note_numbers = true; + } + if (obj_txt_in.match(rgx.inline_notes_al_gen)) { + if (auto m = obj_txt_in.matchAll(rgx.inline_text_and_note_al_)) { + if (stage_reset_note_numbers) { + n_foot = 0; + n_foot_reg = 0; + n_foot_sp_asterisk = 0; + n_foot_sp_plus = 0; + } + stage_reset_note_numbers = false; + foreach(n; m) { + if (n.hit.to!string.match(rgx.inline_al_delimiter_open_symbol_star)) { + flg_notes_star = true; + ++n_foot_sp_asterisk; + asterisks_ = "*"; + n_foot=n_foot_sp_asterisk; + obj_txt_out ~= n.hit.to!string.replaceFirst( + rgx.inline_al_delimiter_open_symbol_star, + (mkup.en_a_o ~ replicate(asterisks_, n_foot_sp_asterisk) ~ " ") + ) ~ "\n"; + } else if (n.hit.to!string.match(rgx.inline_al_delimiter_open_symbol_plus)) { + flg_notes_plus = true; + ++n_foot_sp_plus; + plus_ = "*"; + n_foot=n_foot_sp_plus; + obj_txt_out ~= n.hit.to!string.replaceFirst( + rgx.inline_al_delimiter_open_symbol_plus, + (mkup.en_a_o ~ replicate(plus_, n_foot_sp_plus) ~ " ") + ) ~ "\n"; + } else if (n.hit.to!string.match(rgx.inline_al_delimiter_open_regular)) { + flg_notes_reg = true; + ++n_foot_reg; + n_foot=n_foot_reg; + obj_txt_out ~= n.hit.to!string.replaceFirst( + rgx.inline_al_delimiter_open_regular, + (mkup.en_a_o ~ to!string(n_foot) ~ " ") + ) ~ "\n"; + } else { + obj_txt_out ~= n.hit.to!string ~ "\n"; + } + } + } + } else { + obj_txt_out = obj_txt_in; + } + auto t = tuple( + obj_txt_out, + flg_notes_reg, + flg_notes_star, + flg_notes_plus, + ); + return t; + } + private auto object_notes_and_links_(Ot)(Ot obj_txt_in, bool reset_note_numbers=false) + in { + debug(asserts) { + assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt_out = ""; + bool urls = false; + bool images_without_dimensions = false; + tail = ""; + /+ special endnotes +/ + obj_txt_in = obj_txt_in.replaceAll( + rgx.inline_notes_curly_sp_asterisk, + (mkup.en_a_o ~ "*" ~ " $1" ~ mkup.en_a_c) + ); + obj_txt_in + = obj_txt_in.replaceAll( + rgx.inline_notes_curly_sp_plus, + (mkup.en_a_o ~ "+" ~ " $1" ~ mkup.en_a_c) + ); + /+ image matched +/ + if (obj_txt_in.match(rgx.smid_image_generic)) { + obj_txt_in = images(obj_txt_in); + if (obj_txt_in.match(rgx.smid_mod_image_without_dimensions)) { + images_without_dimensions = true; + } + } + /+ url matched +/ + if (obj_txt_in.match(rgx.smid_inline_url)) { + urls = true; + obj_txt_in = url_links(obj_txt_in); + } + auto ftn = footnotes_endnotes_markup_and_number_or_stars(obj_txt_in, reset_note_numbers); + obj_txt_out = ftn[0]; + debug(footnotes) { + writeln(obj_txt_out, tail); + } + obj_txt_out = obj_txt_out ~ tail; + debug(footnotesdone) { + foreach(m; matchAll(obj_txt_out, + (mkup.en_a_o ~ `\s*(.+?)` ~ mkup.en_a_c))) { + writeln(m.captures[1]); + writeln(m.hit); + } + } + auto t = tuple( + obj_txt_out, + ftn[1], + ftn[2], + ftn[3], + urls, + images_without_dimensions, + ); + return t; + } + auto init() + in { } + body { + auto t = object_notes_and_links_(""); + return t; + } + invariant() { + } + auto munge_heading(Ot)(Ot obj_txt_in, bool reset_note_numbers=false) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=(obj_txt_in) + .replaceFirst(rgx.heading, "") + .replaceFirst(rgx.object_number_off_all, "") + .strip; + auto t = object_notes_and_links_(obj_txt["munge"], reset_note_numbers); + debug(munge) { + writeln(__LINE__); + writeln(obj_txt_in); + writeln(__LINE__); + writeln(obj_txt["munge"].to!string); + } + return t; + } + invariant() { + } + auto munge_para(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=(obj_txt_in) + .replaceFirst(rgx.para_attribs, "") + .replaceFirst(rgx.object_number_off_all, ""); + auto t = object_notes_and_links_(obj_txt["munge"]); + debug(munge) { + writeln(__LINE__); + writeln(obj_txt_in); + writeln(__LINE__); + writeln(obj_txt["munge"].to!string); + } + return t; + } + string munge_quote(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + auto munge_group(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + auto t = object_notes_and_links_(obj_txt["munge"]); + return t; + } + invariant() { + } + auto munge_block(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=obj_txt_in; + auto t = object_notes_and_links_(obj_txt["munge"]); + return t; + } + invariant() { + } + auto munge_verse(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=obj_txt_in; + auto t = object_notes_and_links_(obj_txt["munge"]); + return t; + } + invariant() { + } + string munge_code(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt_in = (obj_txt_in).replaceAll(rgx.space, mkup.nbsp); + obj_txt["munge"] = obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string munge_table(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string munge_comment(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + } + static struct ObjInlineMarkup { + static auto rgx = Rgx(); + static auto munge = ObjInlineMarkupMunge(); + string[string] obj_txt; + static string[] anchor_tags_ = []; + static string anchor_tag = ""; + auto obj_inline_markup_and_anchor_tags_and_misc(O,K,CMM)( + O obj_, + K obj_key_, + CMM conf_make_meta, + bool _new_doc + ) + in { + debug(asserts) { + static assert(is(typeof(obj_) == string[string])); + static assert(is(typeof(obj_key_) == string)); + } + } + body { + obj_txt["munge"] = obj_[obj_key_].dup; + obj_txt["munge"] = (obj_["is"].match(ctRegex!(`verse|code`))) + ? obj_txt["munge"] + : obj_txt["munge"].strip; + if (_new_doc) { + anchor_tags_ = []; + } + auto x = munge.init; + bool[string] obj_notes_and_links; + obj_notes_and_links["notes_reg"] = false; + obj_notes_and_links["notes_star"] = false; + obj_notes_and_links["links"] = false; + obj_notes_and_links["image_no_dimensions"] = false; + switch (obj_["is"]) { + case "heading": + if (_new_doc) { + anchor_tag = ""; + } + obj_txt["munge"]=_configured_auto_heading_numbering_and_segment_anchor_tags(obj_txt["munge"], obj_, conf_make_meta, _new_doc); + obj_txt["munge"]=_make_segment_anchor_tags_if_none_provided(obj_txt["munge"], obj_["lev"], _new_doc); + if (auto m = obj_txt["munge"].match(rgx.heading_anchor_tag)) { + anchor_tag = m.captures[1]; + anchor_tags_ ~= anchor_tag; + } else if (obj_["lev"] == "1") { + writeln("heading anchor tag missing: ", obj_txt["munge"]); + } + x = munge.munge_heading(obj_txt["munge"], reset_note_numbers); + reset_note_numbers=false; + goto default; + case "para": + x = munge.munge_para(obj_txt["munge"]); + goto default; + case "group": + x = munge.munge_group(obj_txt["munge"]); + goto default; + case "block": + x = munge.munge_block(obj_txt["munge"]); + goto default; + case "verse": + x = munge.munge_verse(obj_txt["munge"]); + goto default; + case "code": + obj_txt["munge"] = munge.munge_code(obj_txt["munge"]); + break; + case "table": + obj_txt["munge"] = munge.munge_table(obj_txt["munge"]); + break; + case "quote": + obj_txt["munge"] = munge.munge_quote(obj_txt["munge"]); + break; + case "comment": + obj_txt["munge"] = munge.munge_comment(obj_txt["munge"]); + break; + case "doc_end_reset": + munge.initialize_note_numbers(); + break; + default: + /+ para, heading, group, block, verse +/ + obj_txt["munge"] = x[0]; + obj_notes_and_links["notes_reg"] = x[1]; + obj_notes_and_links["notes_star"] = x[2]; + obj_notes_and_links["notes_plus"] = x[3]; + obj_notes_and_links["links"] = x[4]; + obj_notes_and_links["image_no_dimensions"] = x[5]; + break; + } + auto t = tuple( + obj_txt["munge"], + anchor_tags_, + obj_notes_and_links["notes_reg"], + obj_notes_and_links["notes_star"], + obj_notes_and_links["links"], + obj_notes_and_links["image_no_dimensions"], + ); + anchor_tags_=[]; + return t; + } + invariant() { + } + auto _clean_heading_toc_(Toc)( + Toc heading_toc_, + ) { + debug(asserts) { + static assert(is(typeof(heading_toc_) == char[])); + } + auto m = (cast(char[]) heading_toc_).matchFirst(rgx.heading); + heading_toc_ = (m.post).replaceAll( + rgx.inline_notes_curly_gen, + ""); + return heading_toc_; + }; + auto table_of_contents_gather_headings(O,CMM,Ts,Ta,X,Toc)( + O obj_, + CMM conf_make_meta, + Ts segment_anchor_tag_that_object_belongs_to, + Ta _anchor_tag, + return ref X lev4_subtoc, + Toc the_table_of_contents_section, + ) + in { + debug(asserts) { + static assert(is(typeof(obj_) == string[string])); + static assert(is(typeof(segment_anchor_tag_that_object_belongs_to) == string)); + static assert(is(typeof(_anchor_tag) == string)); + static assert(is(typeof(lev4_subtoc) == string[][string])); + static assert(is(typeof(the_table_of_contents_section) == ObjGenericComposite[][string])); + } + } + body { + ObjGenericComposite comp_obj_toc; + mixin InternalMarkup; + static auto mkup = InlineMarkup(); + char[] heading_toc_ = (obj_["substantive"].dup.strip.to!(char[])).replaceAll(rgx.inline_notes_al, ""); + heading_toc_ = _clean_heading_toc_(heading_toc_); + auto attrib=""; + string toc_txt_, subtoc_txt_; + int[string] indent; + if (obj_["lev_markup_number"].to!int > 0) { + indent=[ + "hang_position" : obj_["lev_markup_number"].to!int, + "base_position" : obj_["lev_markup_number"].to!int, + ]; + toc_txt_ = format( + "{ %s }#%s", + heading_toc_, + _anchor_tag, + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc = comp_obj_toc.init; + comp_obj_toc.metainfo.is_of_part = "frontmatter"; + comp_obj_toc.metainfo.is_of_section = "toc"; + comp_obj_toc.metainfo.is_of_type = "para"; + comp_obj_toc.metainfo.is_a = "toc"; + comp_obj_toc.metainfo.ocn = 0; + comp_obj_toc.metainfo.object_number_off = ""; + comp_obj_toc.metainfo.object_number_type = 0; + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.attrib.bullet = false; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } else { + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + comp_obj_toc = comp_obj_toc.init; + comp_obj_toc.metainfo.is_of_part = "frontmatter"; + comp_obj_toc.metainfo.is_of_section = "toc"; + comp_obj_toc.metainfo.is_of_type = "para"; + comp_obj_toc.metainfo.is_a = "toc"; + comp_obj_toc.metainfo.ocn = 0; + comp_obj_toc.metainfo.object_number_off = ""; + comp_obj_toc.metainfo.object_number_type = 0; + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.attrib.bullet = false; + comp_obj_toc.text = "Table of Contents"; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + comp_obj_toc = comp_obj_toc.init; + comp_obj_toc.metainfo.is_of_part = "frontmatter"; + comp_obj_toc.metainfo.is_of_section = "toc"; + comp_obj_toc.metainfo.is_of_type = "para"; + comp_obj_toc.metainfo.is_a = "toc"; + comp_obj_toc.metainfo.ocn = 0; + comp_obj_toc.metainfo.object_number_off = ""; + comp_obj_toc.metainfo.object_number_type = 0; + comp_obj_toc.attrib.bullet = false; + comp_obj_toc.has.inline_links = true; + switch (obj_["lev_markup_number"].to!int) { + case 0: + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + toc_txt_ = "{ Table of Contents }" ~ mkup.mark_internal_site_lnk ~ "toc.fnSuffix"; + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + break; + case 1: .. case 3: + indent=[ + "hang_position" : obj_["lev_markup_number"].to!int, + "base_position" : obj_["lev_markup_number"].to!int, + ]; + toc_txt_ = format( + "%s", + heading_toc_, + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + break; + case 4: + toc_txt_ = format( + "{ %s }%s%s%s", + heading_toc_, + mkup.mark_internal_site_lnk, + segment_anchor_tag_that_object_belongs_to, + ".fnSuffix", + ); + lev4_subtoc[segment_anchor_tag_that_object_belongs_to] = []; + toc_txt_= munge.url_links(toc_txt_); + indent=[ + "hang_position" : obj_["lev_markup_number"].to!int, + "base_position" : obj_["lev_markup_number"].to!int, + ]; + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + break; + case 5: .. case 7: + toc_txt_ = format( + "{ %s }%s%s%s#%s", + heading_toc_, + mkup.mark_internal_site_lnk, + segment_anchor_tag_that_object_belongs_to, + ".fnSuffix", + _anchor_tag, + ); + subtoc_txt_ = format( + "{ %s }#%s", + heading_toc_, + _anchor_tag, + ); + lev4_subtoc[segment_anchor_tag_that_object_belongs_to] + ~= munge.url_links(obj_["lev_markup_number"] + ~ "~ " ~ subtoc_txt_.to!string.strip + ); + toc_txt_= munge.url_links(toc_txt_); + indent=[ + "hang_position" : obj_["lev_markup_number"].to!int, + "base_position" : obj_["lev_markup_number"].to!int, + ]; + comp_obj_toc.attrib.indent_hang = indent["hang_position"]; + comp_obj_toc.attrib.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.has.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + break; + default: + break; + } + return the_table_of_contents_section; + } + invariant() { + } + private: + static int[] heading_num = [ 0, 0, 0, 0 ]; + static string heading_number_auto_composite = ""; + static string _configured_auto_heading_numbering_and_segment_anchor_tags(M,O,CMM)( + M munge_, + O obj_, + CMM conf_make_meta, + bool _new_doc + ) { + debug(asserts) { + static assert(is(typeof(munge_) == string)); + static assert(is(typeof(obj_) == string[string])); + } + if (_new_doc) { + heading_num = [ 0, 0, 0, 0 ]; + heading_number_auto_composite = ""; + } + if (conf_make_meta.make.auto_num_top_lv) { + if (obj_["lev_markup_number"].to!int == 0) { + heading_num[0] = 0; + heading_num[1] = 0; + heading_num[2] = 0; + heading_num[3] = 0; + heading_number_auto_composite = ""; + } + /+ auto_num_depth minimum 0 + (1.) default 2 (1.1.1) max 3 (1.1.1.1) implement +/ + if ( + conf_make_meta.make.auto_num_top_lv + > obj_["lev_markup_number"].to!uint + ) { + heading_num[1] = 0; + heading_num[2] = 0; + heading_num[3] = 0; + } else if ( + conf_make_meta.make.auto_num_top_lv + == obj_["lev_markup_number"].to!uint + ) { + heading_num[0] ++; + heading_num[1] = 0; + heading_num[2] = 0; + heading_num[3] = 0; + } else if ( + conf_make_meta.make.auto_num_top_lv + == (obj_["lev_markup_number"].to!uint - 1) + ) { + heading_num[1] ++; + heading_num[2] = 0; + heading_num[3] = 0; + } else if ( + conf_make_meta.make.auto_num_top_lv + == (obj_["lev_markup_number"].to!uint - 2) + ) { + heading_num[2] ++; + heading_num[3] = 0; + } else if ( + conf_make_meta.make.auto_num_top_lv + == (obj_["lev_markup_number"].to!uint - 3) + ) { + heading_num[3] ++; + } + if (heading_num[3] > 0) { + heading_number_auto_composite + = (conf_make_meta.make.auto_num_depth.to!uint == 3) + ? ( heading_num[0].to!string ~ "." + ~ heading_num[1].to!string ~ "." + ~ heading_num[2].to!string ~ "." + ~ heading_num[3].to!string + ) + : ""; + } else if (heading_num[2] > 0) { + heading_number_auto_composite + = ((conf_make_meta.make.auto_num_depth.to!uint >= 2) + && (conf_make_meta.make.auto_num_depth.to!uint <= 3)) + ? ( heading_num[0].to!string ~ "." + ~ heading_num[1].to!string ~ "." + ~ heading_num[2].to!string + ) + : ""; + } else if (heading_num[1] > 0) { + heading_number_auto_composite + = ((conf_make_meta.make.auto_num_depth.to!uint >= 1) + && (conf_make_meta.make.auto_num_depth.to!uint <= 3)) + ? ( heading_num[0].to!string ~ "." + ~ heading_num[1].to!string + ) + : ""; + } else if (heading_num[0] > 0) { + heading_number_auto_composite + = ((conf_make_meta.make.auto_num_depth.to!uint >= 0) + && (conf_make_meta.make.auto_num_depth.to!uint <= 3)) + ? (heading_num[0].to!string) + : ""; + } else { + heading_number_auto_composite = ""; + } + debug(heading_number_auto) { + writeln(heading_number_auto_composite); + } + if ((!empty(heading_number_auto_composite)) + && (obj_["lev_markup_number"].to!uint >= conf_make_meta.make.auto_num_top_lv)) { + munge_=(munge_) + .replaceFirst(rgx.heading, + "$1~$2 " ~ heading_number_auto_composite ~ ". ") + .replaceFirst(rgx.heading_marker_missing_tag, + "$1~" ~ heading_number_auto_composite ~ " "); + } + } + return munge_; + } + static int heading_num_lev1 = 0; + static string _make_segment_anchor_tags_if_none_provided(M,Lv)( + M munge_, + Lv lev_, + bool _new_doc + ) { + debug(asserts) { + static assert(is(typeof(munge_) == string)); + static assert(is(typeof(lev_) == string)); + } + if (!(munge_.match(rgx.heading_anchor_tag))) { + if (munge_.match(rgx.heading_identify_anchor_tag)) { + if (auto m = munge_.match(rgx.heading_extract_named_anchor_tag)) { + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, + "$1~" ~ m.captures[1].toLower ~ "_" ~ m.captures[2] ~ " "); + if (auto n = munge_.match(rgx.heading_anchor_tag_plus_colon)) { + auto tag_remunge_ = n.captures[2].replaceAll(rgx.heading_marker_tag_has_colon, ".."); + munge_=(munge_).replaceFirst(rgx.heading_anchor_tag_plus_colon, n.captures[1] ~ tag_remunge_ ~ " "); + } + } else if (auto m = munge_.match(rgx.heading_extract_unnamed_anchor_tag)) { + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, + "$1~" ~ "s" ~ m.captures[1] ~ " "); + } + } else if (lev_ == "1") { // (if not successful) manufacture a unique anchor tag for lev=="1" + if (_new_doc) { + heading_num_lev1 = 0; + } + heading_num_lev1 ++; + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, + "$1~" ~ "x" ~ heading_num_lev1.to!string ~ " "); + } + } + return munge_; + } + unittest { + string txt_lev="1"; + string txt_in, txt_out; + + txt_in = "1~copyright Copyright"; + txt_out ="1~copyright Copyright"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in = "1~ 6. Writing Copyright Licenses"; + txt_out ="1~s6 6. Writing Copyright Licenses"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ 1. Reinforcing trends"; + txt_out= "1~s1 1. Reinforcing trends"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ 11 SCIENCE AS A COMMONS"; + txt_out= "1~s11 11 SCIENCE AS A COMMONS"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ Chapter 1"; + txt_out="1~chapter_1 Chapter 1"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ Chapter 1."; + txt_out="1~chapter_1 Chapter 1."; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ Chapter 1: Done"; + txt_out="1~chapter_1 Chapter 1: Done"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ Chapter 11 - The Battle Over the Institutional Ecology of the Digital Environment"; + txt_out= "1~chapter_11 Chapter 11 - The Battle Over the Institutional Ecology of the Digital Environment"; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ CHAPTER I."; + txt_out="1~x1 CHAPTER I."; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + + txt_in= "1~ CHAPTER II."; + txt_out="1~x2 CHAPTER II."; + assert(_make_segment_anchor_tags_if_none_provided(txt_in, txt_lev) == txt_out); + } + } + /+ +/ + struct ObjAttributes { + string[string] _obj_attrib; + string obj_attributes(Oi,OR,OH)( + Oi obj_is_, + OR obj_raw, + OH _comp_obj_heading, + ) + in { + debug(asserts) { + static assert(is(typeof(obj_is_) == string)); + static assert(is(typeof(obj_raw) == string)); + static assert(is(typeof(_comp_obj_heading) == ObjGenericComposite)); + } + } + body { + scope(exit) { + destroy(obj_is_); + destroy(obj_raw); + destroy(_comp_obj_heading); + } + _obj_attrib["json"] ="{"; + switch (obj_is_) { + case "heading": + _obj_attrib["json"] ~= _heading(obj_raw); + break; + case "para": + _obj_attrib["json"] ~= _para_and_blocks(obj_raw) + ~ _para(obj_raw); + break; + case "code": + _obj_attrib["json"] ~= _code(obj_raw); + break; + case "group": + _obj_attrib["json"] ~= _para_and_blocks(obj_raw) + ~ _group(obj_raw); + break; + case "block": + _obj_attrib["json"] ~= _para_and_blocks(obj_raw) + ~ _block(obj_raw); + break; + case "verse": + _obj_attrib["json"] ~= _verse(obj_raw); + break; + case "quote": + _obj_attrib["json"] ~= _quote(obj_raw); + break; + case "table": + _obj_attrib["json"] ~= _table(obj_raw); + break; + case "comment": + _obj_attrib["json"] ~= _comment(obj_raw); + break; + default: + _obj_attrib["json"] ~= _para(obj_raw); + break; + } + _obj_attrib["json"] ~=" }"; + _obj_attrib["json"]=_set_additional_values_parse_as_json(_obj_attrib["json"], obj_is_, _comp_obj_heading); + debug(structattrib) { + if (oa_j["is"].str() == "heading") { + writeln(_obj_attrib["json"]); + writeln( + "is: ", oa_j["is"].str(), + "; object_number: ", oa_j["object_number"].integer() + ); + } + } + return _obj_attrib["json"]; + } + invariant() { + } + private: + string _obj_attributes; + string _para_and_blocks(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + if (obj_txt_in.matchFirst(rgx.para_bullet)) { + _obj_attributes =" \"bullet\": \"true\"," + ~ " \"indent_hang\": 0," + ~ " \"indent_base\": 0,"; + } else if (auto m = obj_txt_in.matchFirst(rgx.para_bullet_indent)) { + _obj_attributes =" \"bullet\": \"true\"," + ~ " \"indent_hang\": " ~ m.captures[1].to!string ~ "," + ~ " \"indent_base\": " ~ m.captures[1].to!string ~ ","; + } else if (auto m = obj_txt_in.matchFirst(rgx.para_indent_hang)) { + _obj_attributes =" \"bullet\": \"false\"," + ~ " \"indent_hang\": " ~ m.captures[1].to!string ~ "," + ~ " \"indent_base\": " ~ m.captures[2].to!string ~ ","; + } else if (auto m = obj_txt_in.matchFirst(rgx.para_indent)) { + _obj_attributes =" \"bullet\": \"false\"," + ~ " \"indent_hang\": " ~ m.captures[1].to!string ~ "," + ~ " \"indent_base\": " ~ m.captures[1].to!string ~ ","; + } else { + _obj_attributes =" \"bullet\": \"false\"," + ~ " \"indent_hang\": 0," + ~ " \"indent_base\": 0,"; + } + return _obj_attributes; + } + string _heading(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"para\"," + ~ " \"is\": \"heading\""; + return _obj_attributes; + } + invariant() { + } + string _para(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"para\"," + ~ " \"is\": \"para\""; + return _obj_attributes; + } + invariant() { + } + string _quote(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"quote\""; + return _obj_attributes; + } + invariant() { + } + string _group(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"group\""; + return _obj_attributes; + } + invariant() { + } + string _block(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"block\""; + return _obj_attributes; + } + invariant() { + } + string _verse(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"verse\""; + return _obj_attributes; + } + invariant() { + } + string _code(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"code\""; + return _obj_attributes; + } + invariant() { + } + string _table(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"table\""; + return _obj_attributes; + } + invariant() { + } + string _comment(Ot)(Ot obj_txt_in) + in { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + } + body { + _obj_attributes = " \"use\": \"comment\"," + ~ " \"of\": \"comment\"," + ~ " \"is\": \"comment\""; + return _obj_attributes; + } + invariant() { + } + string _set_additional_values_parse_as_json(OA,Oi,OH)( + OA _obj_attrib, + Oi obj_is_, + OH _comp_obj_heading, + ) { + debug(asserts) { + static assert(is(typeof(_obj_attrib) == string)); + static assert(is(typeof(obj_is_) == string)); + static assert(is(typeof(_comp_obj_heading) == ObjGenericComposite)); + } + JSONValue oa_j = parseJSON(_obj_attrib); + assert( + (oa_j.type == JSON_TYPE.OBJECT) + ); + if (obj_is_ == "heading") { + oa_j.object["object_number"] = _comp_obj_heading.metainfo.ocn; + oa_j.object["lev_markup_number"] = _comp_obj_heading.metainfo.heading_lev_markup; + oa_j.object["lev_collapsed_number"] = _comp_obj_heading.metainfo.heading_lev_collapsed; + oa_j.object["heading_ptr"] + = _comp_obj_heading.ptr.heading; + oa_j.object["doc_object_ptr"] + = _comp_obj_heading.ptr.doc_object; + } + oa_j.object["parent_object_number"] = _comp_obj_heading.metainfo.parent_ocn; + oa_j.object["parent_lev_markup_number"] = _comp_obj_heading.metainfo.parent_lev_markup; + _obj_attrib = oa_j.toString(); + return _obj_attrib; + } + } + /+ +/ + struct BookIndexNuggetHash { + string main_term, sub_term, sub_term_bits; + int object_number_offset, object_number_endpoint; + string[] object_numbers; + string[][string][string] bi; + string[][string][string] hash_nugget; + string[] bi_main_terms_split_arr; + string[][string][string] bookindex_nugget_hash(BI,N,S)( + BI bookindex_section, + N obj_cite_digits, + S segment_anchor_tag, + ) + in { + debug(asserts) { + static assert(is(typeof(bookindex_section) == string)); + static assert(is(typeof(obj_cite_digits.on) == int)); + } + debug(bookindexraw) { + if (!bookindex_section.empty) { + writeln( + "* [bookindex] ", + "[", obj_cite_digits.on.to!string, ": ", segment_anchor_tag, "] ", bookindex_section + ); + } + } + } + body { + static auto rgx = Rgx(); + if (!bookindex_section.empty) { + auto bi_main_terms_split_arr + = bookindex_section.split(rgx.bi_main_terms_split); + foreach (bi_main_terms_content; bi_main_terms_split_arr) { + auto bi_main_term_and_rest + = bi_main_terms_content.split(rgx.bi_main_term_plus_rest_split); + if (auto m = bi_main_term_and_rest[0].match( + rgx.bi_term_and_object_numbers_match) + ) { + main_term = m.captures[1].strip; + object_number_offset = m.captures[2].to!int; + object_number_endpoint=(obj_cite_digits.on + object_number_offset); + object_numbers ~= (obj_cite_digits.on.to!string ~ "-" ~ to!string(object_number_endpoint) + ~ ":" ~ segment_anchor_tag); + } else { + main_term = bi_main_term_and_rest[0].strip; + object_numbers ~= obj_cite_digits.on.to!string + ~ ":" ~ segment_anchor_tag; + } + bi[main_term]["_a"] ~= object_numbers; + object_numbers=null; + if (bi_main_term_and_rest.length > 1) { + auto bi_sub_terms_split_arr + = bi_main_term_and_rest[1].split( + rgx.bi_sub_terms_plus_object_number_offset_split + ); + foreach (sub_terms_bits; bi_sub_terms_split_arr) { + if (auto m = sub_terms_bits.match(rgx.bi_term_and_object_numbers_match)) { + sub_term = m.captures[1].strip; + object_number_offset = m.captures[2].to!int; + object_number_endpoint=(obj_cite_digits.on + object_number_offset); + object_numbers ~= (obj_cite_digits.on.to!string ~ " - " ~ to!string(object_number_endpoint) + ~ ":" ~ segment_anchor_tag); + } else { + sub_term = sub_terms_bits.strip; + object_numbers ~= to!string(obj_cite_digits.on) + ~ ":" ~ segment_anchor_tag; + } + if (!empty(sub_term)) { + bi[main_term][sub_term] ~= object_numbers; + } + object_numbers=null; + } + } + } + } + hash_nugget = bi; + return hash_nugget; + } + invariant() { + } + } + struct BookIndexReportIndent { + int mkn, skn; + auto bookindex_report_indented(BI)( + BI bookindex_unordered_hashes + ) { + debug(asserts) { + static assert(is(typeof(bookindex_unordered_hashes) == string[][string][string])); + } + auto mainkeys + = bookindex_unordered_hashes.byKey.array.sort().release; + foreach (mainkey; mainkeys) { + debug(bookindex1) { + writeln(mainkey); + } + auto subkeys + = bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + debug(bookindex1) { + writeln(" ", subkey); + writeln(" ", to!string( + bookindex_unordered_hashes[mainkey][subkey] + )); + } + ++skn; + } + ++mkn; + } + } + } + struct BookIndexReportSection { + int mkn, skn; + static auto rgx = Rgx(); + static auto munge = ObjInlineMarkupMunge(); + auto bookindex_write_section(BI)( + BI bookindex_unordered_hashes + ) { + debug(asserts) { + static assert(is(typeof(bookindex_unordered_hashes) == string[][string][string])); + } + auto mainkeys=bookindex_unordered_hashes.byKey.array.sort().release; + foreach (mainkey; mainkeys) { + write("_0_1 !{", mainkey, "}! "); + foreach (ref_; bookindex_unordered_hashes[mainkey]["_a"]) { + auto go = (ref_).replaceAll(rgx.book_index_go, "$1"); + write(" {", ref_, "}#", go, ", "); + } + writeln(" \\\\"); + bookindex_unordered_hashes[mainkey].remove("_a"); + auto subkeys + = bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + write(" ", subkey, ", "); + foreach (ref_; bookindex_unordered_hashes[mainkey][subkey]) { + auto go = (ref_).replaceAll(rgx.book_index_go, "$1"); + write(" {", ref_, "}#", go, ", "); + } + writeln(" \\\\"); + ++skn; + } + ++mkn; + } + } + auto bookindex_build_abstraction_section(BI,N,B)( + BI bookindex_unordered_hashes, + N obj_cite_digits, + B opt_action, + ) { + debug(asserts) { + static assert(is(typeof(bookindex_unordered_hashes) == string[][string][string])); + static assert(is(typeof(obj_cite_digits.on) == int)); + } + mixin SiSUnode; + mixin InternalMarkup; + static auto mkup = InlineMarkup(); + string type_is; + string lev; + int heading_lev_markup, heading_lev_collapsed; + string attrib; + int[string] indent; + auto mainkeys + = bookindex_unordered_hashes.byKey.array.sort().release; + ObjGenericComposite[][string] bookindex_section; + ObjGenericComposite comp_obj_heading_, comp_obj_para; + auto node_para_int_ = node_metadata_para_int; + auto node_para_str_ = node_metadata_para_str; + if ((mainkeys.length > 0) + && (opt_action.backmatter + && opt_action.section_bookindex)) { + string bi_tmp_seg, bi_tmp_scroll; + string[] bi_tmp_tags; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "bookindex"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Book Index"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "_part_book_index"; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading.has.inline_links = true; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + ++mkn; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "bookindex"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Index"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "bookindex"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 2; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading.has.inline_links = false; + comp_obj_heading_.tags.anchor_tags = ["bookindex"]; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + ++mkn; + import std.array : appender; + auto buffer = appender!(char[])(); + string[dchar] transTable = [' ' : "_"]; + foreach (mainkey; mainkeys) { + bi_tmp_tags = [""]; + bi_tmp_scroll = "!{" ~ mainkey ~ "}! "; + buffer.clear(); + bi_tmp_tags ~= translate(mainkey, transTable); + bi_tmp_seg = "!{" ~ mainkey ~ "}! "; + auto bkidx_lnk_seg(string locs) { + string markup = ""; + if (auto m = locs.matchFirst(rgx.book_index_go_seg)) { + markup + = munge.url_links("{ " ~ m["link"] ~ " }" + ~ mkup.mark_internal_site_lnk ~ m["seg"] ~ ".fnSuffix" + ~ "#" ~ m["ocn"] ~ ", "); + } else { + writeln(__LINE__, ": ", locs); + } + return markup; + } + auto bkidx_lnk_scroll(string locs) { + string markup = ""; + if (auto m = locs.matchFirst(rgx.book_index_go)) { + markup + = munge.url_links("{ " ~ m["link"] ~ " }" + ~ mkup.mark_internal_site_lnk + ~ "#" ~ m["ocn"] ~ ", "); + } else { + writeln(__LINE__, ": ", locs); + } + return markup; + } + foreach (ref_; bookindex_unordered_hashes[mainkey]["_a"]) { + bi_tmp_scroll ~= bkidx_lnk_scroll(ref_); + bi_tmp_seg ~= bkidx_lnk_seg(ref_); + } + bi_tmp_scroll ~= " \\\\\n "; + bi_tmp_seg ~= " \\\\\n "; + bookindex_unordered_hashes[mainkey].remove("_a"); + auto subkeys + = bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + bi_tmp_scroll ~= subkey ~ ", "; + buffer.clear(); + bi_tmp_tags ~= translate(subkey, transTable); + bi_tmp_seg ~= subkey ~ ", "; + foreach (ref_; bookindex_unordered_hashes[mainkey][subkey]) { + bi_tmp_scroll ~= bkidx_lnk_scroll(ref_); + bi_tmp_seg ~= bkidx_lnk_seg(ref_); + } + bi_tmp_scroll ~= " \\\\\n "; + bi_tmp_seg ~= " \\\\\n "; + ++skn; + } + bi_tmp_scroll = (bi_tmp_scroll).replaceFirst(rgx.trailing_linebreak, ""); + bi_tmp_seg = (bi_tmp_seg).replaceFirst(rgx.trailing_linebreak, ""); + comp_obj_para = comp_obj_para.init; + comp_obj_para.metainfo.is_of_part = "backmatter"; + comp_obj_para.metainfo.is_of_section = "bookindex"; + comp_obj_para.metainfo.is_of_type = "para"; + comp_obj_para.metainfo.is_a = "bookindex"; + comp_obj_para.text = bi_tmp_scroll.to!string.strip; + comp_obj_para.metainfo.ocn = 0; + comp_obj_para.metainfo.object_number_off = ""; + comp_obj_para.metainfo.object_number_type = 0; + comp_obj_para.tags.anchor_tags = bi_tmp_tags; + comp_obj_para.attrib.indent_hang = 0; + comp_obj_para.attrib.indent_base = 1; + comp_obj_para.attrib.bullet = false; + comp_obj_para.has.inline_links = true; + bookindex_section["scroll"] ~= comp_obj_para; + comp_obj_para.text = bi_tmp_seg.to!string.strip; + bookindex_section["seg"] ~= comp_obj_para; + ++mkn; + } + } else { // no book index, (figure out what to do here) + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.text = "(skip) there is no Book Index"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + } + auto t = tuple(bookindex_section, obj_cite_digits); + return t; + } + } + /+ +/ + struct NotesSection { + string[string] object_notes; + int previous_count; + int mkn; + static auto rgx = Rgx(); + private auto gather_notes_for_endnote_section( + ObjGenericComposite[] contents_am, + string segment_anchor_tag_that_object_belongs_to, + int cntr, + ) + in { + assert((contents_am[cntr].metainfo.is_a == "para") + || (contents_am[cntr].metainfo.is_a == "heading") + || (contents_am[cntr].metainfo.is_a == "quote") + || (contents_am[cntr].metainfo.is_a == "group") + || (contents_am[cntr].metainfo.is_a == "block") + || (contents_am[cntr].metainfo.is_a == "verse")); + assert(cntr >= previous_count); + assert( + (contents_am[cntr].text).match( + rgx.inline_notes_delimiter_al_regular_number_note) + ); + } + body { + mixin InternalMarkup; + previous_count=cntr; + static auto mkup = InlineMarkup(); + static auto munge = ObjInlineMarkupMunge(); + foreach( + m; + (contents_am[cntr].text).matchAll( + rgx.inline_notes_delimiter_al_regular_number_note + ) + ) { + debug(endnotes_build) { + writeln( + "{^{", m.captures[1], ".}^}" + ~ mkup.mark_internal_site_lnk, + segment_anchor_tag_that_object_belongs_to, + ".fnSuffix#noteref_\n ", m.captures[1], " ", + m.captures[2]); // sometimes need segment name (segmented html & epub) + } + // you need anchor for segments at this point -> + object_notes["anchor"] ~= "note_" ~ m.captures[1] ~ "』"; + object_notes["notes"] ~= (segment_anchor_tag_that_object_belongs_to.empty) + ? (munge.url_links( + "{^{" ~ m.captures[1] ~ ".}^}#noteref_" + ~ m.captures[1]) ~ " " + ~ m.captures[2] ~ "』" + ) + : (munge.url_links( + "{^{" ~ m.captures[1] ~ ".}^}" + ~ mkup.mark_internal_site_lnk + ~ segment_anchor_tag_that_object_belongs_to + ~ ".fnSuffix#noteref_" + ~ m.captures[1]) ~ " " + ~ m.captures[2] ~ "』" + ); + } + return object_notes; + } + private auto gathered_notes() + in { + } + body { + string[][string] endnotes_; + if (object_notes.length > 1) { + endnotes_["notes"] = (object_notes["notes"].split(rgx.break_string))[0..$-1]; + endnotes_["anchor"] = (object_notes["anchor"].split(rgx.break_string))[0..$-1]; + } else { + endnotes_["notes"] = []; + endnotes_["anchor"] = []; + } + return endnotes_; + } + private auto endnote_objects(N,O)( + N obj_cite_digits, + O opt_action, + ) + in { + } + body { + mixin SiSUnode; + ObjGenericComposite[] the_endnotes_section; + auto endnotes_ = gathered_notes(); + string type_is; + string lev, lev_markup_number, lev_collapsed_number; + string attrib; + int[string] indent; + ObjGenericComposite comp_obj_heading_; + if ((endnotes_["notes"].length > 0) + && (opt_action.backmatter && opt_action.section_endnotes)) { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "endnotes"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Endnotes"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "_part_endnotes"; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_endnotes_section ~= comp_obj_heading_; + ++mkn; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "backmatter"; + comp_obj_heading_.metainfo.is_of_section = "endnotes"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "Endnotes"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.tags.segment_anchor_tag = "endnotes"; + comp_obj_heading_.metainfo.heading_lev_markup = 4; + comp_obj_heading_.metainfo.heading_lev_collapsed = 2; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + comp_obj_heading_.tags.anchor_tags = ["endnotes"]; + the_endnotes_section ~= comp_obj_heading_; + ++mkn; + } else { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.metainfo.is_of_part = "empty"; + comp_obj_heading_.metainfo.is_of_section = "empty"; + comp_obj_heading_.metainfo.is_of_type = "para"; + comp_obj_heading_.metainfo.is_a = "heading"; + comp_obj_heading_.text = "(skip) there are no Endnotes"; + comp_obj_heading_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_heading_.metainfo.heading_lev_markup = 1; + comp_obj_heading_.metainfo.heading_lev_collapsed = 1; + comp_obj_heading_.metainfo.parent_ocn = 1; + comp_obj_heading_.metainfo.parent_lev_markup = 0; + the_endnotes_section ~= comp_obj_heading_; + } + if (opt_action.backmatter && opt_action.section_endnotes) { + ObjGenericComposite comp_obj_endnote_; + comp_obj_endnote_ = comp_obj_endnote_.init; + comp_obj_endnote_.metainfo.is_of_part = "backmatter"; + comp_obj_endnote_.metainfo.is_of_section = "endnote"; + comp_obj_endnote_.metainfo.is_of_type = "para"; + comp_obj_endnote_.metainfo.is_a = "endnote"; + comp_obj_endnote_.metainfo.ocn = 0; + comp_obj_heading_.metainfo.object_number_off = ""; + comp_obj_heading_.metainfo.object_number_type = 0; + comp_obj_endnote_.attrib.indent_hang = 0; + comp_obj_endnote_.attrib.indent_base = 0; + comp_obj_endnote_.attrib.bullet = false; + foreach (i, endnote; endnotes_["notes"]) { + auto m = endnote.matchFirst(rgx.note_ref); + string notenumber = m.captures[1].to!string; + string anchor_tag = "note_" ~ notenumber; + comp_obj_endnote_.tags.anchor_tags = [ endnotes_["anchor"][i] ]; + comp_obj_endnote_.has.inline_links = true; + comp_obj_endnote_.text = endnote.strip; + the_endnotes_section ~= comp_obj_endnote_; + } + } + auto t = tuple(the_endnotes_section, obj_cite_digits); + return t; + } + } + /+ +/ + struct Bibliography { + public JSONValue[] _bibliography_(Bi,BJ)( + return ref Bi biblio_unsorted_incomplete, + return ref BJ bib_arr_json + ) + in { + debug(asserts) { + static assert(is(typeof(biblio_unsorted_incomplete) == string[])); + static assert(is(typeof(bib_arr_json) == JSONValue[])); + } + } + body { + JSONValue[] biblio_unsorted + = _biblio_unsorted_complete_(biblio_unsorted_incomplete, bib_arr_json); + biblio_arr_json = []; + biblio_unsorted_incomplete = []; + JSONValue[] biblio_sorted__ = biblio_sort(biblio_unsorted); + biblio_debug(biblio_sorted__); + debug(biblio0) { + writeln("---"); + writeln("unsorted incomplete: ", biblio_unsorted_incomplete.length); + writeln("json: ", bib_arr_json.length); + writeln("unsorted: ", biblio_unsorted.length); + writeln("sorted: ", biblio_sorted__.length); + int cntr; + int[7] x; + while (cntr < x.length) { + writeln(cntr, ": ", biblio_sorted__[cntr]["fulltitle"]); + cntr++; + } + } + return biblio_sorted__; + } + final private JSONValue[] _biblio_unsorted_complete_(Bi,BJ)( + Bi biblio_unordered, + return ref BJ bib_arr_json + ) { + debug(asserts) { + static assert(is(typeof(biblio_unordered) == string[])); + static assert(is(typeof(bib_arr_json) == JSONValue[])); + } + foreach (bibent; biblio_unordered) { + /+ update bib to include deemed_author, needed for: + sort_bibliography_array_by_deemed_author_year_title + either: sort on multiple fields, or; create such sort field + +/ + JSONValue j = parseJSON(bibent); + if (!empty(j["fulltitle"].str)) { + if (!empty(j["author_raw"].str)) { + j["deemed_author"]=j["author_arr"][0]; + } else if (!empty(j["editor_raw"].str)) { + j["deemed_author"]=j["editor_arr"][0]; + } + j["sortby_deemed_author_year_title"] = ( + j["deemed_author"].str ~ + "; " ~ + j["year"].str ~ + "; " ~ + j["fulltitle"].str + ); + } + bib_arr_json ~= j; + } + JSONValue[] biblio_unsorted_array_of_json_objects + = bib_arr_json.dup; + destroy(bib_arr_json); + return biblio_unsorted_array_of_json_objects; + } + final private JSONValue[] biblio_sort(BJ)(BJ biblio_unordered) { + debug(asserts) { + static assert(is(typeof(biblio_unordered) == JSONValue[])); + } + JSONValue[] biblio_sorted_; + biblio_sorted_ + = sort!((a, b){ + return ((a["sortby_deemed_author_year_title"].str) < (b["sortby_deemed_author_year_title"].str)); + })(biblio_unordered).array; + debug(bibliosorted) { + foreach (j; biblio_sorted_) { + if (!empty(j["fulltitle"].str)) { + writeln(j["sortby_deemed_author_year_title"]); + } + } + } + return biblio_sorted_; + } + void biblio_debug(BJ)(BJ biblio_sorted) { + debug(asserts) { + static assert(is(typeof(biblio_sorted) == JSONValue[])); + } + debug(biblio0) { + foreach (j; biblio_sorted) { + if (!empty(j["fulltitle"].str)) { + writeln(j["sortby_deemed_author_year_title"]); + } + } + } + } + } + /+ +/ + struct NodeStructureMetadata { + int lv, lv0, lv1, lv2, lv3, lv4, lv5, lv6, lv7; + int obj_cite_digit; + int[string] p_; // p_ parent_ + static auto rgx = Rgx(); + ObjGenericComposite node_location_emitter(Lv,Ta,N,C,P,I)( + Lv lev_markup_number, + Ta segment_anchor_tag, + N obj_cite_digits, + C cntr_, + P ptr_, + I is_ + ) + in { + debug(asserts) { + static assert(is(typeof(lev_markup_number) == string)); + static assert(is(typeof(segment_anchor_tag) == string)); + static assert(is(typeof(obj_cite_digits.on) == int)); + static assert(is(typeof(cntr_) == int)); + static assert(is(typeof(ptr_) == int)); + static assert(is(typeof(is_) == string)); + } + assert(is_ != "heading"); + assert(obj_cite_digits.on.to!int >= 0); + } + body { + assert(is_ != "heading"); // should not be necessary + assert(obj_cite_digits.on.to!int >= 0); // should not be necessary + if (lv7 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_4; + p_["object_number"] = lv7; + } else if (lv6 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_3; + p_["object_number"] = lv6; + } else if (lv5 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_2; + p_["object_number"] = lv5; + } else { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_1; + p_["object_number"] = lv4; + } + ObjGenericComposite comp_obj_location; + comp_obj_location = comp_obj_location.init; + comp_obj_location.metainfo.is_a = is_; + comp_obj_location.metainfo.ocn = obj_cite_digits.on; + comp_obj_location.tags.segment_anchor_tag = segment_anchor_tag.to!string; + comp_obj_location.metainfo.parent_ocn = p_["object_number"]; + comp_obj_location.metainfo.parent_lev_markup = p_["lev_markup_number"]; + debug(_node) { + if (lev_markup_number.match(rgx.levels_numbered_headings)) { + writeln("x ", _node.to!string); + } else { + writeln("- ", _node.to!string); + } + } + assert(comp_obj_location.metainfo.parent_lev_markup >= 4); + assert(comp_obj_location.metainfo.parent_lev_markup <= 7); + assert(comp_obj_location.metainfo.parent_ocn >= 0); + return comp_obj_location; + } + invariant() { + } + ObjGenericComposite node_emitter_heading(T,Lm,Lc,Ta,N,C,P,LA,I,PSn,fNr,fNs,fL)( + T _text, + Lm lev_markup_number, + Lc lev_collapsed_number, + Ta segment_anchor_tag, + N obj_cite_digits, + C cntr_, + P ptr_, + LA lv_ancestors_txt, + I is_, + PSn html_segnames_ptr, + fNr flag_notes_reg, + fNs flag_notes_star, + fL flag_links, + ) + in { + debug(asserts) { + static assert(is(typeof(_text) == string)); + static assert(is(typeof(lev) == string)); + static assert(is(typeof(lev_markup_number) == string)); + static assert(is(typeof(lev_collapsed_number) == string)); + static assert(is(typeof(segment_anchor_tag) == string)); + static assert(is(typeof(obj_cite_digits.on) == int)); + static assert(is(typeof(cntr_) == int)); + static assert(is(typeof(ptr_) == int)); + static assert(is(typeof(lv_ancestors_txt) == string[])); + static assert(is(typeof(is_) == string)); + static assert(is(typeof(html_segnames_ptr) == int)); + } + assert(is_ == "heading"); + assert((obj_cite_digits.on).to!int >= 0); + assert( + lev_markup_number.match(rgx.levels_numbered), + ("not a valid heading level: " ~ lev_markup_number ~ " at " ~ obj_cite_digits.on.to!string) + ); + if (lev_markup_number.match(rgx.levels_numbered)) { + if (lev_markup_number.to!int == 0) { + /+ TODO first hit (of two) with this assertion failure, check, fix & reinstate + assert(obj_cite_digits.on.to!int == 1, + "ERROR header lev markup number is: " ~ + lev_markup_number.to!string ~ + " obj_cite_digits.on.to!int should == 1 but is: " ~ + obj_cite_digits.on.to!string ~ + "\n" ~ _text); + +/ + } + } + } + body { + switch (lev_markup_number.to!int) { + case 0: + lv = DocStructMarkupHeading.h_sect_A; + lv0 = obj_cite_digit; + lv1=0; lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] = 0; + p_["object_number"] = 0; + break; + case 1: + lv = DocStructMarkupHeading.h_sect_B; + lv1 = obj_cite_digit; + lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_A; + p_["object_number"] = lv0; + break; + case 2: + lv = DocStructMarkupHeading.h_sect_C; + lv2 = obj_cite_digit; + lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_B; + p_["object_number"] = lv1; + break; + case 3: + lv = DocStructMarkupHeading.h_sect_D; + lv3=obj_cite_digit; + lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_C; + p_["object_number"] = lv2; + break; + case 4: + lv = DocStructMarkupHeading.h_text_1; + lv4 = obj_cite_digit; + lv5=0; lv6=0; lv7=0; + if (lv3 > State.off) { + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_D; + p_["object_number"] = lv3; + } else if (lv2 > State.off) { + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_C; + p_["object_number"] = lv2; + } else if (lv1 > State.off) { + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_B; + p_["object_number"] = lv1; + } else { + p_["lev_markup_number"] + = DocStructMarkupHeading.h_sect_A; + p_["object_number"] = lv0; + } + break; + case 5: + lv = DocStructMarkupHeading.h_text_2; + lv5 = obj_cite_digit; + lv6=0; lv7=0; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_text_1; + p_["object_number"] = lv4; + break; + case 6: + lv = DocStructMarkupHeading.h_text_3; + lv6 = obj_cite_digit; + lv7=0; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_text_2; + p_["object_number"] = lv5; + break; + case 7: + lv = DocStructMarkupHeading.h_text_4; + lv7 = obj_cite_digit; + p_["lev_markup_number"] + = DocStructMarkupHeading.h_text_3; + p_["object_number"] = lv6; + break; + default: + break; + } + ObjGenericComposite _comp_obj_heading_; + _comp_obj_heading_ = _comp_obj_heading_.init; + _comp_obj_heading_.metainfo.is_of_part = "body"; + _comp_obj_heading_.metainfo.is_of_section = "body"; + _comp_obj_heading_.metainfo.is_of_type = "para"; + _comp_obj_heading_.metainfo.is_a = "heading"; + _comp_obj_heading_.text = _text.to!string.strip; + _comp_obj_heading_.metainfo.ocn = obj_cite_digits.on; + _comp_obj_heading_.metainfo.object_number_off = (obj_cite_digits.off==0) ? "" : obj_cite_digits.off.to!string; + _comp_obj_heading_.metainfo.object_number_type = obj_cite_digits.type; + _comp_obj_heading_.tags.segment_anchor_tag = segment_anchor_tag.to!string; + _comp_obj_heading_.metainfo.heading_lev_markup = (!(lev_markup_number.empty) ? lev_markup_number.to!int : 0); + _comp_obj_heading_.metainfo.heading_lev_collapsed = (!(lev_collapsed_number.empty) ? lev_collapsed_number.to!int : 0); + _comp_obj_heading_.metainfo.parent_ocn = p_["object_number"]; + _comp_obj_heading_.metainfo.parent_lev_markup = p_["lev_markup_number"]; + _comp_obj_heading_.tags.heading_ancestors_text = lv_ancestors_txt; + _comp_obj_heading_.ptr.doc_object = cntr_; + _comp_obj_heading_.ptr.html_segnames = ((lev_markup_number == "4") ? html_segnames_ptr : 0); + _comp_obj_heading_.ptr.heading = ptr_; + _comp_obj_heading_.has.inline_notes_reg = flag_notes_reg; + _comp_obj_heading_.has.inline_notes_star = flag_notes_star; + _comp_obj_heading_.has.inline_links = flag_links; + debug(_node) { + if (lev_markup_number.match(rgx.levels_numbered_headings)) { + writeln("* ", _node.to!string); + } + } + debug(nodeheading) { + if (lev_markup_number.match(rgx.levels_numbered_headings)) { + writeln("* ", _node.to!string); + } + } + assert(_comp_obj_heading_.metainfo.parent_lev_markup <= 7); + assert(_comp_obj_heading_.metainfo.parent_ocn >= 0); + if (lev_markup_number.match(rgx.levels_numbered_headings)) { + assert(_comp_obj_heading_.metainfo.heading_lev_markup <= 7); + assert(_comp_obj_heading_.metainfo.ocn >= 0); + if (_comp_obj_heading_.metainfo.parent_lev_markup > 0) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup < _comp_obj_heading_.metainfo.heading_lev_markup); + if (_comp_obj_heading_.metainfo.ocn != 0) { + assert(_comp_obj_heading_.metainfo.parent_ocn < _comp_obj_heading_.metainfo.ocn); + } + } + if (_comp_obj_heading_.metainfo.heading_lev_markup == 0) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_sect_A); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_sect_B) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_sect_A); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_sect_C) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_sect_B); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_sect_D) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_sect_C); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_text_1) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup <= DocStructMarkupHeading.h_sect_D); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_text_2) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_text_1); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_text_3) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_text_2); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_text_4) { + assert(_comp_obj_heading_.metainfo.parent_lev_markup == DocStructMarkupHeading.h_text_3); + } else if (_comp_obj_heading_.metainfo.heading_lev_markup == DocStructMarkupHeading.h_text_5) { + } + } + return _comp_obj_heading_; + } + invariant() { + } + } + /+ abstraction functions emitters ↑ +/ + /+ ↓ abstraction functions assertions +/ + pure void assertions_doc_structure(O,Lv)( + O an_object, + Lv lv + ) { + debug(asserts) { + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(lv) == int[string])); + } + if (lv["h3"] > State.off) { + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h2"] > State.off); + } else if (lv["h2"] > State.off) { + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h3"] == State.off); + } else if (lv["h1"] > State.off) { + assert(lv["h0"] > State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + } else if (lv["h0"] > State.off) { + assert(lv["h1"] == State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + } else { + assert(lv["h0"] == State.off); + assert(lv["h1"] == State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + } + if (lv["h7"] > State.off) { + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + assert(lv["h6"] > State.off); + } else if (lv["h6"] > State.off) { + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + assert(lv["h7"] == State.off); + } else if (lv["h5"] > State.off) { + assert(lv["h4"] > State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } else if (lv["h4"] > State.off) { + assert(lv["h5"] == State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } else { + assert(lv["h4"] == State.off); + assert(lv["h5"] == State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } + if (lv["h0"] == State.off) { + assert(lv["h1"] == State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + assert(lv["h4"] == State.off); + assert(lv["h5"] == State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } + if (lv["h1"] == State.off) { + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + } + if (lv["h2"] == State.off) { + assert(lv["h3"] == State.off); + } + if (lv["h3"] == State.off) { + } + if (lv["h4"] == State.off) { + assert(lv["h5"] == State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } + if (lv["h5"] == State.off) { + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } + if (lv["h6"] == State.off) { + assert(lv["h7"] == State.off); + } + if (lv["h7"] == State.off) { + } + switch ((an_object["lev"]).to!string) { + case "A": + if (lv["h0"] == State.off) { + assert(lv["h1"] == State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + assert(lv["h4"] == State.off); + assert(lv["h5"] == State.off); + assert(lv["h6"] == State.off); + assert(lv["h7"] == State.off); + } else { // (lv["h0"] > State.off) + assert(lv["h0"] == State.off,"error should not enter level A a second time"); + } + break; + case "B": + if (lv["h1"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h2"] == State.off); + assert(lv["h3"] == State.off); + } else { // (lv["h1"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + } + break; + case "C": + if (lv["h2"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h3"] == State.off); + } else { // (lv["h2"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h2"] > State.off); + } + break; + case "D": + if (lv["h3"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h2"] > State.off); + } else { // (lv["h3"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h1"] > State.off); + assert(lv["h2"] > State.off); + assert(lv["h3"] > State.off); + } + break; + case "1": + if (lv["h4"] == State.off) { + assert(lv["h0"] > State.off); + } else { // (lv["h4"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + } + break; + case "2": + if (lv["h5"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + } else { // (lv["h5"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + } + break; + case "3": + if (lv["h6"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + } else { // (lv["h6"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + assert(lv["h6"] > State.off); + } + break; + case "4": + if (lv["h7"] == State.off) { + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + assert(lv["h6"] > State.off); + } else { // (lv["h7"] > State.off) + assert(lv["h0"] > State.off); + assert(lv["h4"] > State.off); + assert(lv["h5"] > State.off); + assert(lv["h6"] > State.off); + assert(lv["h7"] > State.off); + } + break; + default: + break; + } + } + pure void assertions_flag_types_block_status_none_or_closed(T)(T obj_type_status) { + debug(asserts) { + static assert(is(typeof(obj_type_status) == int[string])); + } + assert( + (obj_type_status["code"] == TriState.off) + || (obj_type_status["code"] == TriState.closing), + "code block status: off or closing"); + assert( + (obj_type_status["poem"] == TriState.off) + || (obj_type_status["poem"] == TriState.closing), + "poem status: off or closing"); + assert( + (obj_type_status["table"] == TriState.off) + || (obj_type_status["table"] == TriState.closing), + "table status: off or closing"); + assert( + (obj_type_status["group"] == TriState.off) + || (obj_type_status["group"] == TriState.closing), + "group block status: off or closing"); + assert( + (obj_type_status["block"] == TriState.off) + || (obj_type_status["block"] == TriState.closing), + "block status: off or closing"); + } + /+ abstraction functions assertions ↑ +/ +} /+ ← closed: template SiSUdocAbstraction +/ +template docSectKeysSeq() { + auto docSectKeysSeq(string[][string] document_section_keys_sequenced) { + struct doc_sect_keys_seq { + auto seg() { + return document_section_keys_sequenced["seg"]; + } + auto scroll() { + return document_section_keys_sequenced["scroll"]; + } + auto sql() { + return document_section_keys_sequenced["sql"]; + } + } + return doc_sect_keys_seq(); + } +} diff --git a/src/doc_reform/meta/metadoc_summary.d b/src/doc_reform/meta/metadoc_summary.d new file mode 100644 index 0000000..f2e0632 --- /dev/null +++ b/src/doc_reform/meta/metadoc_summary.d @@ -0,0 +1,101 @@ +module doc_reform.meta.metadoc_summary; +template SiSUabstractionSummary() { + auto SiSUabstractionSummary(S,T)( + auto ref const S doc_abstraction, + auto ref T doc_matters, + ) { + import + doc_reform.meta.defaults, + doc_reform.meta.rgx; + import + std.array, + std.exception, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.uni, + std.utf, + std.conv : to; + mixin InternalMarkup; + auto markup = InlineMarkup(); + if (doc_matters.opt.action.verbose) { + string[string] check = [ + "last_object_number" : "NA [debug \"checkdoc\" not run]", + "last_object_number_body" : "0", + "last_object_number_book_index" : "0", + ]; + foreach (k; doc_matters.xml.keys_seq.seg) { + foreach (obj; doc_abstraction[k]) { + if (obj.metainfo.is_of_part != "empty") { + if (!empty(obj.metainfo.object_number)) { + if (k == "body") { + check["last_object_number_body"] = obj.metainfo.object_number; + } + if (!(obj.metainfo.object_number.empty)) { + check["last_object_number"] = obj.metainfo.object_number; + } + } + if (k == "bookindex_seg") { + if (obj.metainfo.object_number_type == 2) { + check["last_object_number_book_index"] = obj.metainfo.object_number_book_index; + } + } + } + } + } + auto min_repeat_number = 66; + auto char_repeat_number = (doc_matters.conf_make_meta.meta.title_full.length + + doc_matters.conf_make_meta.meta.creator_author.length + 4); + char_repeat_number = (char_repeat_number > min_repeat_number) + ? char_repeat_number + : min_repeat_number; + writefln( + "%s\n\"%s\", %s\n%s\n%s\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n%30-s%10-d\n(%s: %s)\n%s", + markup.repeat_character_by_number_provided("-", char_repeat_number), + doc_matters.conf_make_meta.meta.title_full, + doc_matters.conf_make_meta.meta.creator_author, + doc_matters.src.filename, + markup.repeat_character_by_number_provided("-", char_repeat_number), + "length toc arr:", + to!int(doc_abstraction["toc_seg"].length), + "length doc_abstraction arr:", + to!int(doc_abstraction["body"].length), + "last doc body ocn:", + to!int(check["last_object_number_body"]), + "last object_number:", + to!int(check["last_object_number"]), + "length endnotes:", // subtract headings + (doc_abstraction["endnotes"].length > 2) + ? (to!int(doc_abstraction["endnotes"].length - 2)) + : 0, + "length glossary:", + (doc_abstraction["glossary"].length > 1) + ? (to!int(doc_abstraction["glossary"].length)) + : 0, + "length biblio:", + (doc_abstraction["bibliography"].length > 1) + ? (to!int(doc_abstraction["bibliography"].length)) + : 0, + "length bookindex:", + (doc_abstraction["bookindex_seg"].length > 1) + ? (to!int(doc_abstraction["bookindex_seg"].length)) + : 0, + " last book idx ocn:", + to!int(check["last_object_number_book_index"]), + "length blurb:", + (doc_abstraction["blurb"].length > 1) + ? (to!int(doc_abstraction["blurb"].length)) + : 0, + "number of segments:", + (doc_matters.xml.segnames.length > 1) + ? (to!int(doc_matters.xml.segnames.length)) + : 0, + __FILE__, + __LINE__, + markup.repeat_character_by_number_provided("-", min_repeat_number), + ); + } + } +} diff --git a/src/doc_reform/meta/object_setter.d b/src/doc_reform/meta/object_setter.d new file mode 100644 index 0000000..f481ec1 --- /dev/null +++ b/src/doc_reform/meta/object_setter.d @@ -0,0 +1,126 @@ +/++ + object setter: + setting of sisu objects for downstream processing + meta_object_setter.d ++/ +module doc_reform.meta.object_setter; +template ObjectSetter() { + /+ structs +/ + struct DocObj_MetaInfo_ { + string is_of_part = ""; // frontmatter, body, backmatter + string is_of_section = ""; // toc, body, glossary, biography, book index, blurb + string is_of_type = ""; // para, block ? + string is_a = ""; // heading, para, table, code block, group, ... + alias of_part = is_of_part; + alias of_section = is_of_section; + alias is_of = is_of_type; + /+ o_n +/ + int o_n_substantive = 0; + int o_n_non_substantive = 0; + int o_n_glossary = 0; + int o_n_bibliography = 0; + int o_n_book_index = 0; + int o_n_blurb = 0; + string object_number_substantive() const @property { + return (o_n_substantive==0) ? "" : o_n_substantive.to!string; + } + string object_number_non_substantive() const @property { + return (o_n_non_substantive==0) ? "" : o_n_non_substantive.to!string; + } + string object_number_glossary() const @property { + return (o_n_glossary==0) ? "" : o_n_glossary.to!string; + } + string object_number_bibliography() const @property { + return (o_n_bibliography==0) ? "" : o_n_bibliography.to!string; + } + string object_number_book_index() const @property { + return (o_n_book_index==0) ? "" : o_n_book_index.to!string; + } + string object_number_blurb() const @property { + return (o_n_blurb==0) ? "" : o_n_blurb.to!string; + } + string object_number_off = ""; + bool visible_object_number = false; + // enum ONtype { none, substantive, non_substantive, glossary, bibliography, book_index, blurb, comment } + int object_number_type = 0; // { ocn, non, bkidx } + /+ node +/ + string[string][string] node; + int ocn = 0; + string object_number() const @property { + return (ocn==0) ? "" : ocn.to!string; + } + int o_n_type = 0; + int heading_lev_markup = 9; + int heading_lev_collapsed = 9; + string marked_up_level() const @property { + string _out; + switch (heading_lev_markup) { + case 0: _out = "A"; break; + case 1: _out = "B"; break; + case 2: _out = "C"; break; + case 3: _out = "D"; break; + case 4: _out = "1"; break; + case 5: _out = "2"; break; + case 6: _out = "3"; break; + case 7: _out = "4"; break; + default: _out = ""; break; + } + return _out; + } + int[] markedup_ancestors = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] collapsed_ancestors = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_structure_markedup_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_structure_collapsed_tags_status = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int parent_lev_markup = 0; + int parent_ocn = 0; + int last_decendant_ocn = 0; + } + struct DocObj_TxtAttrib_ { + int indent_base = 0; + int indent_hang = 0; + bool bullet = false; + string language = ""; // not implemented, consider + } + struct DocObj_Has_ { // has + bool inline_links = false; + bool inline_notes_reg = false; + bool inline_notes_star = false; + bool contains_image_without_dimensions = false; + } + struct DocObj_Table_ { + int number_of_columns = 0; + double[] column_widths = []; + string[] column_aligns = []; + bool heading = false; + bool walls = false; // not implemented + } + struct DocObj_CodeBlock_ { + string syntax = ""; + } + struct DocObj_Pointer_ { + int doc_object = 0; + int html_segnames = 0; + int heading = 0; + } + struct DocObj_Tags_ { + string[] heading_ancestors_text = [ "", "", "", "", "", "", "", "", ]; // TODO redundant? see markedup and collapsed ancestors DONE + string segment_anchor_tag = ""; + string segname_prev = ""; + string segname_next = ""; + string[] lev4_subtoc = []; + string[] anchor_tags = []; + } + struct ObjGenericComposite { + string text = ""; + DocObj_MetaInfo_ metainfo; + DocObj_TxtAttrib_ attrib; + DocObj_Tags_ tags; + DocObj_Has_ has; + DocObj_Table_ table; + DocObj_CodeBlock_ code_block; + DocObj_Pointer_ ptr; + } + struct TheObjects { + ObjGenericComposite[] oca; + } +} diff --git a/src/doc_reform/meta/package.d b/src/doc_reform/meta/package.d new file mode 100644 index 0000000..2981b6a --- /dev/null +++ b/src/doc_reform/meta/package.d @@ -0,0 +1,17 @@ +module doc_reform.meta; +public import + doc_reform.meta.defaults, + doc_reform.meta.rgx; +/+ std +/ +public import + std.array, + std.exception, + std.range, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + // std.uni, + std.utf, + std.conv : to; diff --git a/src/doc_reform/meta/rgx.d b/src/doc_reform/meta/rgx.d new file mode 100644 index 0000000..63bb4ee --- /dev/null +++ b/src/doc_reform/meta/rgx.d @@ -0,0 +1,290 @@ +/++ + regex: regular expressions used in sisu document parser ++/ +module doc_reform.meta.rgx; +static template SiSUrgxInit() { + import doc_reform.meta.defaults; + static struct Rgx { + /+ misc +/ + static true_dollar = ctRegex!(`\$`, "gm"); + static flag_action = ctRegex!(`^(--[a-z][a-z0-9-]+)$`); + static flag_action_str = ctRegex!(` (--[a-z][a-z0-9-]+)`); + static within_quotes = ctRegex!(`"(.+?)"`); + static make_heading_delimiter = ctRegex!(`[;][ ]*`); + static arr_delimiter = ctRegex!(`[ ]*[;][ ]*`); + static name_delimiter = ctRegex!(`^([^,]+)[ ]*,[ ]+(.+?)$`); + static book_index_go = ctRegex!("(?P(?P[0-9]+)(?:-[0-9]+)?)"); + static book_index_go_scroll = ctRegex!("(?P(?P[0-9]+)(?:-[0-9]+)?)"); + static book_index_go_seg = ctRegex!("(?P(?P[0-9]+)(?:-[0-9]+)?):(?P[a-z0-9_-]+)"); + static book_index_go_seg_ = ctRegex!("(?P(?P[0-9]+)(?:-[0-9]+)?)(:(?P[a-z0-9_-]+))?"); + static book_index_go_seg_anchorless = ctRegex!("(?P(?P[0-9]+)(?:-[0-9]+)?)"); + static trailing_comma = ctRegex!(",[ ]*$"); + static trailing_linebreak = ctRegex!(",[ ]{1,2}\\\\\\\\\n[ ]{4}$","m"); + static newline_eol_delimiter = ctRegex!("\n"); + static newline_eol_strip_preceding = ctRegex!("[ ]*\n"); + static newline_eol_delimiter_only = ctRegex!("^\n"); + static line_delimiter_ws_strip = ctRegex!("[ ]*\n[ ]*"); + static para_delimiter = ctRegex!("\n[ ]*\n+"); + static table_col_delimiter = ctRegex!("[ ]*\n+", "mg"); + static table_row_delimiter = ctRegex!("\n[ ]*\n+", "mg"); + static table_row_delimiter_special = ctRegex!("[ ]*\n", "mg"); + static table_col_delimiter_special = ctRegex!("[ ]*[|][ ]*", "mg"); + static levels_markup = ctRegex!(`^[A-D1-4]$`); + static levels_numbered = ctRegex!(`^[0-9]$`); + static levels_numbered_headings = ctRegex!(`^[0-7]$`); + static numeric = ctRegex!(`[ 0-9,.-]+`); + static numeric_col = ctRegex!(`^[ 0-9,.$£₤Є€€¥-]+$`); + /+ comments +/ + static comment = ctRegex!(`^%+ `); + static comments = ctRegex!(`^%+ |^%+$`); + /+ header +/ + static make_simple_substitutions_rb = ctRegex!(`(?P/(?P.+?)/,[ ]*['"](?P.+?)['"])`); + static make_simple_substitutions_d = ctRegex!(`(?P` ~ '`' ~ `(?P.+?)` ~ '`' ~ `,[ ]*['"](?P.+?)['"])`); + /+ header +/ + static main_headers = + ctRegex!(`^(?:creator|title|rights|date|original|classify|identifier|notes|publisher|make|links)$`, "m"); + static native_header = ctRegex!(`^@([a-z_]+):(?:\s|$)`); + static native_header_make = ctRegex!(`^@(make):(?:\s|$)`); + static native_header_meta = + ctRegex!(`^@(?:creator|title|rights|date|original|classify|identifier|notes|publisher|links):(?:\s|$)`); + static native_header_main = ctRegex!(`^@(?P
[a-z_]+):\s*(?P.*)`, "m"); + static native_header_sub = ctRegex!(`^[ ]*:(?P[a-z_]+):\s+(?P.+)`, "m"); + static native_header_meta_title = ctRegex!(`^@title:\s`, "m"); + static variable_doc_title = ctRegex!(`@title`); + static variable_doc_author = ctRegex!(`@author|@creator`); + static raw_author_munge = ctRegex!(`(\S.+?),\s+(.+)`,"i"); + static toml_header_meta_title = ctRegex!(`^\s*(title\s*=\s*"|\[title\])`, "m"); + /+ head +/ + static native_subhead_creator = ctRegex!(`^(?:author|translator|illustrator)$`, "m"); + static native_subhead_title = ctRegex!(`^(?:main|sub(?:title)?|full|language|edition|note)$`, "m"); + static native_subhead_rights = ctRegex!(`^(?:copyright|illustrations|license|cover)$`, "m"); + static native_subhead_date = ctRegex!(`^(?:published|created|issued|available|valid|modified|added_to_site)$`, "m"); + static native_subhead_original = ctRegex!(`^(?:title|language|source)$`, "m"); + static native_subhead_classify = ctRegex!(`^(?:topic_register|subject|keywords|loc|dewey)$`, "m"); + static native_subhead_identifier = ctRegex!(`^(?:oclc|pg|isbn)$`, "m"); + static native_subhead_notes = ctRegex!(`^(?:abstract|description)$`, "m"); + static native_subhead_publisher = ctRegex!(`^(?:name)$`, "m"); + static native_subhead_make = ctRegex!(`^(?:cover_image|home_button_image|home_button_text|footer|headings|num_top|auto_num_depth|breaks|substitute|bold|italics|emphasis|texpdf_font|css)$`, "m"); + /+ heading & paragraph operators +/ + static heading_a = ctRegex!(`^:?[A][~] `, "m"); + static heading = ctRegex!(`^:?([A-D1-4])[~]([a-z0-9_.-]*[?]?)\s+`,"i"); + static heading_seg_and_above = ctRegex!(`^:?([A-D1])[~]([a-z0-9_.-]*[?]?)\s+`,"i"); + static heading_marker = ctRegex!(`^:?([A-D1-4])[~]`); + static heading_anchor_tag = ctRegex!(`^:?[A-D1-4][~]([a-z0-9_.-]+) `,"i"); + static heading_identify_anchor_tag = ctRegex!(`^:?[A-D1-4][~]\s+(?:(?:(?:chapter|article|section|clause)\s+[0-9.]+)|(?:[0-9]+))`,"i"); + static heading_extract_named_anchor_tag = ctRegex!(`^:?[A-D1-4][~]\s+(chapter|article|section|clause)\s+((?:[0-9]+[.:])*[0-9]+)(?:[.:;, ]|$)`,"i"); + static heading_extract_unnamed_anchor_tag = ctRegex!(`^:?[A-D1-4][~]\s+((?:[0-9]+.)*[0-9]+)(?:[.:;, ]|$)`); + static heading_marker_missing_tag = ctRegex!(`^:?([A-D1-4])[~] `); + static heading_anchor_tag_plus_colon = ctRegex!(`^:?([A-D1-4][~])([a-z0-9_.:-]+) `,"i"); + static heading_marker_tag_has_colon = ctRegex!(`([:])`); + static heading_title = ctRegex!(`^:?[A-D1-4][~][a-z0-9_.-]*[?]?\s+(.+?)$`); + static heading_all = ctRegex!(`^:?([A-D1-4])[~]([a-z0-9_.-]*[?]?)\s+(.+?)$`); + static heading_backmatter = ctRegex!(`^:?1[~][!](glossary|bibliography|biblio|blurb)\s+`,"i"); + static heading_biblio = ctRegex!(`^:?(1)[~][!](biblio(?:graphy)?|references?)`); + static heading_glossary = ctRegex!(`^:?(1)[~][!](glossary)`); + static heading_blurb = ctRegex!(`^:?(1)[~][!](blurb)`); + static heading_biblio_glossary = ctRegex!(`^:?(?:(1)[~][!](?:(?:biblio(?:graphy)?|references?)|glossary)|[A-D1][~])`); + static heading_biblio_blurb = ctRegex!(`^:?(?:(1)[~][!](?:(?:biblio(?:graphy)?|references?)|blurb)|[A-D1][~])`); + static heading_blurb_glossary = ctRegex!(`^:?(?:(1)[~][!](?:blurb|glossary)|[A-D1][~])`); + static para_bullet = ctRegex!(`^_[*] `); + static para_bullet_indent = ctRegex!(`^_([1-9])[*] `); + static para_indent = ctRegex!(`^_([1-9]) `); + static para_indent_hang = ctRegex!(`^_([0-9])_([0-9]) `); + static para_attribs = ctRegex!(`^_(?:(?:[0-9])(?:_([0-9]))?|(?:[1-9])?[*]) `); + /+ blocked markup +/ + static block_open = ctRegex!("^((code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)[{].*?$)|^`{3} (code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)|^[{]table(~h)?(?P(?:[ ]+[0-9]+;)+)[}]"); + static block_poem_open = ctRegex!("^((poem[{].*?$)|`{3} poem)"); + /+ blocked markup tics +/ + static block_tic_open = ctRegex!("^`{3} (code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)"); // what of numbered code? + static block_tic_code_open = ctRegex!("^`{3} (?:code)(?:[.]([a-z][0-9a-z_]+))?(?:[ ]+([#]))?"); // extract additional info + static block_tic_poem_open = ctRegex!("^`{3} (poem)"); + static block_tic_group_open = ctRegex!("^`{3} (group)"); + static block_tic_block_open = ctRegex!("^`{3} (block)"); + static block_tic_quote_open = ctRegex!("^`{3} (quote)"); + static block_tic_table_open = ctRegex!("^`{3} table(.*)"); + static block_tic_close = ctRegex!("^(`{3})$","m"); + /+ blocked markup curly +/ + static block_curly_open = ctRegex!(`^((code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)[{].*?$)`); + static block_curly_code_open = ctRegex!(`^(?:code(?:[.]([a-z][0-9a-z_]+))?[{]([#]?)\s*$)`); // extract additional info + static block_curly_code_close = ctRegex!(`^([}]code)`); + static block_curly_poem_open = ctRegex!(`^(poem[{].*?$)`); + static block_curly_poem_close = ctRegex!(`^([}]poem)`); + static block_curly_group_open = ctRegex!(`^(group[{].*?$)`); + static block_curly_group_close = ctRegex!(`^([}]group)`); + static block_curly_block_open = ctRegex!(`^(block[{].*?$)`); + static block_curly_block_close = ctRegex!(`^([}]block)`); + static block_curly_quote_open = ctRegex!(`^(quote[{].*?$)`); + static block_curly_quote_close = ctRegex!(`^([}]quote)`); + static block_curly_table_open = ctRegex!(`^table[{](.*)`); + static block_curly_table_close = ctRegex!(`^([}]table)`); + static block_curly_table_special_markup = ctRegex!(`^[{]table((~h)?(?P(?:[ ]+[0-9]+;)+))[}]`, "mg"); + static table_head_instructions = ctRegex!(`(?Ph)?(?:[ ]+c(?P[0-9]);)?(?P(?:[ ]+[0-9]+[lr]?;)+)`); + static table_col_widths_and_alignment = ctRegex!(`(?P[0-9]+)(?P[lr]?)`); + static table_col_widths = ctRegex!(`(?P[0-9]+)`); + static table_col_align = ctRegex!(`(?P[lr]?)`); + static table_col_align_match = ctRegex!(`(?P[lr])`); + static table_col_separator = ctRegex!(`┊`); + static table_col_separator_nl = ctRegex!(`[┊]$`, "mg"); + /+ inline markup footnotes endnotes +/ + static inline_notes_curly_gen = ctRegex!(`~\{.+?\}~`, "m"); + static inline_notes_curly = ctRegex!(`~\{\s*(.+?)\}~`, "mg"); + static inline_curly_delimiter_open_and_close_regular = ctRegex!(`~\{\s*|\s*\}~`, "m"); + static inline_notes_delimiter_curly_regular = ctRegex!(`~\{[ ]*(.+?)\}~`, "m"); + static inline_notes_curly_sp = ctRegex!(`~\{[*+]+\s+(.+?)\}~`, "m"); + static inline_notes_curly_sp_asterisk = ctRegex!(`~\{[*]+\s+(.+?)\}~`, "m"); + static inline_notes_curly_sp_plus = ctRegex!(`~\{[+]+\s+(.+?)\}~`, "m"); + static inline_note_curly_delimiters = ctRegex!(`(~\{[*+]?\s*)(.+?)(\}~)`, "mg"); + static inline_notes_square = ctRegex!(`~\[\s*(.+?)\]~`, "mg"); + static inline_text_and_note_square_sp = ctRegex!(`(.+?)~\[[*+]+\s+(.+?)\]~`, "mg"); + static inline_text_and_note_square = ctRegex!(`(.+?)~\[\s*(.+?)\]~`, "mg"); + static inline_note_square_delimiters = ctRegex!(`(~\[\s*)(.+?)(\]~)`, "mg"); + static inline_curly_delimiter_open_regular = ctRegex!(`~\{\s*`, "m"); + static inline_curly_delimiter_open_symbol_star = ctRegex!(`~\{[*]\s`, "m"); + static inline_curly_delimiter_open_symbol_plus = ctRegex!(`~\{[+]\s`, "m"); + static inline_curly_delimiter_open_star_or_plus = ctRegex!(`~\{[+*]`, "m"); + static inline_curly_delimiter_close_regular = ctRegex!(`\s*\}~`, "m"); + static inline_text_and_note_curly = ctRegex!(`(?P.+?)(?:(?:[~])[{][*+ ]*)(?P.+?)(?:[}][~])`, "mg"); + static note_ref = ctRegex!(`^\S+?noteref_([0-9]+)`, "mg"); // {^{73.}^}#noteref_73 + static smid_inline_url_generic = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_#]`, "mg"); + static smid_inline_url = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg"); + static smid_inline_link_naked_url = ctRegex!(`(?P
^|[ ])(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤)\S+?)(?P[.,;:?!'"]?(?:[ ]|$))`, "mg"); // issue with #link #32
+    static smid_inline_link_markup_regular                = ctRegex!(`(?P
^|[ ]|[^\S]?)\{\s*(?P.+?)\s*\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[;:!,?.]?(?:[ )\]]|$))`, "mg"); // NEXT
+    static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg");
+    static smid_inline_link_endnote_url_helper            = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg");
+    static image                                           = ctRegex!(`([a-zA-Z0-9._-]+?\.(?:png|gif|jpg))`, "mg");
+    static smid_image_generic                              = ctRegex!(`(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*)\S+\.(?:png|gif|jpg).+?\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)[;:!,?.]?(?:[ )\]]|$)`, "mg");
+    static smid_image_with_dimensions                      = ctRegex!(`(?P
(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P\S+\.(?:png|gif|jpg))\s+(?P\d+)x(?P\d+)\s*(?P(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?:[;:!,?.]?(?:[ )\]]|$)))`, "mg");
+    static smid_image                                      = ctRegex!(`(?P
(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P\S+\.(?:png|gif|jpg))\s*(?P(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?:[;:!,?.]?(?:[ )\]]|$)))`, "mg");
+    static smid_mod_image_without_dimensions               = ctRegex!(`\{(?:~\^\s+|\s*)☼\S+\.(?:png|gif|jpg),w0h0\s+(?:.+?)\s*\}(?:image|(?:https?|git):\/\/\S+?)(?:[;:!,?.]?(?:[ )\]]|$))`, "mg");
+    /+ inline markup book index +/
+    static book_index                                     = ctRegex!(`^=\{\s*(.+?)\}$`, "m");
+    static book_index_open                                = ctRegex!(`^=\{\s*([^}]+?)$`);
+    static book_index_close                               = ctRegex!(`^(.*?)\}$`, "m");
+    /+ no object_number object +/
+    static object_number_off                            = ctRegex!(`~#$`, "m");
+    static object_number_off_dh                         = ctRegex!(`-#$`, "m");
+    static object_number_off_all                        = ctRegex!(`[~-]#$`, "m");
+    /+ no object_number block +/
+    static object_number_off_block                      = ctRegex!(`^--~#$`);
+    static object_number_off_block_dh                   = ctRegex!(`^---#$`);
+    static object_number_off_block_close                = ctRegex!(`^--\+#$`);
+    static object_number_block_marks                    = ctRegex!(`^--[+~-]#$`);
+    /+ ignore outside code blocks +/
+    static skip_from_regular_parse    = ctRegex!(`^(--[+~-]#|-[\\]{2}-|=[.\\]{2}=)$`);
+    /+ line & page breaks +/
+    static break_line_within_object                       = ctRegex!(`[\\]{2}( |$)`);
+    static break_page                                     = ctRegex!(`^-[\\]{2}-$`);
+    static break_page_new                                 = ctRegex!(`^=[\\]{2}=$`);
+    static break_page_line_across                         = ctRegex!(`^=[.]{2}=$`);
+    static break_string                                   = ctRegex!(`』`);
+    static parent                                         = ctRegex!(`([0-7]):([0-9]+)`);
+    static header_regex_content                           = ctRegex!(`([0-7]):([0-9]+)`);
+    /+ json +/
+    static tailing_comma                                  = ctRegex!(`,$`, "m");
+    /+ biblio tags +/
+    static biblio_tags                                    = ctRegex!(`^(is|au|author_raw|author|author_arr|editor_raw|ed|editor_arr|ti|title|subtitle|fulltitle|lng|language|trans|src|jo|journal|in|vol|volume|edn|edition|yr|year|pl|place|pb|pub|publisher|url|pg|pages|note|short_name|id):\s+(.+)`);
+    static biblio_abbreviations                           = ctRegex!(`^(au|ed|ti|lng|jo|vol|edn|yr|pl|pb|pub|pg|pgs|sn)$`);
+    /+ bookindex split +/
+    static bi_main_terms_split                            = ctRegex!(`\s*;\s*`);
+    static bi_main_term_plus_rest_split                   = ctRegex!(`\s*:\s*`);
+    static bi_sub_terms_plus_object_number_offset_split = ctRegex!(`\s*\|\s*`);
+    static bi_term_and_object_numbers_match             = ctRegex!(`^(.+?)\+(\d+)`);
+    /+ language codes +/
+    auto language_codes                                    =
+       ctRegex!("(am|bg|bn|br|ca|cs|cy|da|de|el|en|eo|es|et|eu|fi|fr|ga|gl|he|hi|hr|hy|ia|is|it|ja|ko|la|lo|lt|lv|ml|mr|nl|no|nn|oc|pl|pt|pt_BR|ro|ru|sa|se|sk|sl|sq|sr|sv|ta|te|th|tk|tr|uk|ur|vi|zh)");
+    auto language_code_and_filename                                    =
+       ctRegex!("(?:^|[/])(am|bg|bn|br|ca|cs|cy|da|de|el|en|eo|es|et|eu|fi|fr|ga|gl|he|hi|hr|hy|ia|is|it|ja|ko|la|lo|lt|lv|ml|mr|nl|no|nn|oc|pl|pt|pt_BR|ro|ru|sa|se|sk|sl|sq|sr|sv|ta|te|th|tk|tr|uk|ur|vi|zh)/[A-Za-z0-9._-].+?[.](?:sst|ssm)$");
+    static newline                                        = ctRegex!("\n", "mg");
+    static strip_br                                       = ctRegex!("^
\n|
\n*$");
+    static space                                          = ctRegex!(`[ ]`, "mg");
+    static spaces_line_start                              = ctRegex!(`^(?P[ ]+)`, "mg");
+    static spaces_multiple                                = ctRegex!(`(?P[ ]{2,})`, "mg");
+    static two_spaces                                     = ctRegex!(`[ ]{2}`, "mg");
+    static nbsp_char                                      = ctRegex!(`░`, "mg");
+    static nbsp_chars_line_start                          = ctRegex!(`^░+`, "mg");
+    static nbsp_and_space                                 = ctRegex!(` [ ]`, "mg");
+    static nbsp_char_and_space                            = ctRegex!(`░[ ]`, "mg");
+    static special_markup_chars                           = ctRegex!(`[【】〖〗┥┝┤├¤░┘┙┚┼┿╂┊┏┚┆■]`, "mg");
+    static src_pth_sst_or_ssm                             = ctRegex!(`^(?P[/]?(?:[a-zA-Z0-9._-]+/)*)(?P[a-zA-Z0-9._-]+[.](?Pss[tm]))$`);
+    static src_pth_pod_sst_or_ssm                         = ctRegex!(`^(?P[/]?(?:[a-zA-Z0-9._-]+/)*)media/text/[a-z]{2}/(?P[a-zA-Z0-9._-]+[.]ss[tm])$`);
+    static src_pth_contents                               = ctRegex!(`^(?P[/]?(?:[a-zA-Z0-9._-]+/)*)(?P[a-zA-Z0-9._-]+)/sisupod[.]manifest$`);
+    static src_pth_pod_root                               = ctRegex!(`^(?P(?:[/]?(?:[a-zA-Z0-9._-]+/)*)(sisupod))$`);
+    static src_pth_zip                                    = ctRegex!(`^(?P[/]?(?:[a-zA-Z0-9._-]+/)*)(?P[a-zA-Z0-9._-]+[.]zip)$`);
+    static src_pth_unzip_pod                              = ctRegex!(`^(?Pmedia/text/[a-z]{2}/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`);
+    static src_pth_types                                  =
+      ctRegex!(`^(?P[/]?[a-zA-Z0-9._-]+/)*(?P(?P[a-zA-Z0-9._-]+[.]ss[tm])|(?P[a-zA-Z0-9._-]+/sisupod[.]manifest)|(?P[a-zA-Z0-9._-]+[.]zip))$`);
+    static pod_content_location                           =
+      ctRegex!(`^(?P[a-zA-Z0-9._-]+[.]ss[tm])(?P(?:\s+[a-z]{2}(?:,|$))+)`, "mg");
+    static src_fn                                         =
+      ctRegex!(`^([/]?(?:[a-zA-Z0-9._-]+/)*)(?P(?P[a-zA-Z0-9._-]+)[.](?Pss[tm]))$`);
+    static src_fn_master                                  = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssm)$`);
+    static src_fn_text                                    = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]sst)$`);
+    static src_fn_insert                                  = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssi)$`);
+    static src_fn_find_inserts                            = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`);
+    static insert_src_fn_ssi_or_sst                       = ctRegex!(`^<<\s*(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`);
+    static src_base_parent_dir_name                       = ctRegex!(`[/](?P(?:[a-zA-Z0-9._-]+))(?:/media/text/[a-z]{2})$`); // formalizes dir structure
+    static src_base_parent_path                           = ctRegex!(`(?P(?:[/a-zA-Z0-9._-]+))(?:/media/text/[a-z]{2})$`); // formalizes dir structure
+    static src_formalised_file_path_parts                 = ctRegex!(`(?P(?:[/a-zA-Z0-9._-]+?)(?P[a-zA-Z0-9._-]+))(?:/media/text/[a-z]{2})$`); // formalizes dir structure
+    /+ line breaks +/
+    static br_line                                        = ctRegex!(`┘`, "mg");
+    static br_nl                                          = ctRegex!(`┙`, "mg");
+    static br_paragraph                                   = ctRegex!(`┚`, "mg");
+    static br_page_line                                   = ctRegex!(`┼`, "mg");
+    static br_page                                        = ctRegex!(`┿`, "mg");
+    static br_page_new                                    = ctRegex!(`╂`, "mg");
+    /+ inline markup footnotes endnotes +/
+    static inline_notes_al                                = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg");
+    static inline_notes_al_special                        = ctRegex!(`【(?:[*+]\s+)(.+?)】`, "mg"); // TODO remove match when special footnotes are implemented
+    static inline_notes_al_gen                            = ctRegex!(`【.+?】`, "m");
+    static inline_notes_al_gen_text                       = ctRegex!(`【(?P.+?)】`, "m");
+    static inline_notes_al_gen_ref                        = ctRegex!(`【(?P[*+]\s+)\s*(?P.+?)】`, "mg");
+    static inline_al_delimiter_open_regular               = ctRegex!(`【\s`, "m");
+    static inline_al_delimiter_open_symbol_star           = ctRegex!(`【[*]\s`, "m");
+    static inline_al_delimiter_open_symbol_plus           = ctRegex!(`【[+]\s`, "m");
+    static inline_al_delimiter_close_regular              = ctRegex!(`】`, "m");
+    static inline_al_delimiter_open_and_close_regular     = ctRegex!(`【|】`, "m");
+    static inline_notes_delimiter_al_regular              = ctRegex!(`【(.+?)】`, "mg");
+    static inline_notes_delimiter_al_regular_number_note  = ctRegex!(`【(\d+)\s+(.+?)】`, "mg");
+    static inline_al_delimiter_open_asterisk              = ctRegex!(`【\*`, "m");
+    static inline_al_delimiter_open_plus                  = ctRegex!(`【\+`, "m");
+    static inline_text_and_note_al                        = ctRegex!(`(?P.+?)【(?:[*+ ]*)(?P.+?)】`, "mg");
+    static inline_text_and_note_al_                       = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|$))`, "mg");
+    /+ inline markup footnotes endnotes +/
+    static inline_image                                   = ctRegex!(`(?P
┥)☼(?P(?P\S+?\.(?:jpg|gif|png)),w(?P\d+)h(?P\d+))\s*(?P.*?┝┤.+?├)`, "mg");
+    static inline_image_without_dimensions                = ctRegex!(`(?P
┥)☼(?P(?P\S+?\.(?:jpg|gif|png)),w(?P0)h(?P0))\s*(?P.*?┝┤.+?├)`, "mg");
+    static inline_link                                    = ctRegex!(`┥(?P.+?)┝┤(?P.+?)├`, "mg");
+    static inline_link_clean                              = ctRegex!(`┤(?:.+?)├|[┥┝]`, "mg");
+    static inline_a_url                                   = ctRegex!(`(┤)(\S+?)(├)`, "mg");
+    static url                                            = ctRegex!(`https?://`, "mg");
+    static inline_link_subtoc                             = ctRegex!(`^(?P[5-7])~ ┥(?P.+?)┝┤(?P.+?)├`, "mg");
+    static fn_suffix                                      = ctRegex!(`\.fnSuffix`, "mg");
+    static inline_link_fn_suffix                          = ctRegex!(`¤(.+?)(\.fnSuffix)`, "mg");
+    static inline_seg_link                                = ctRegex!(`(¤)(?:.+?)\.fnSuffix`, "mg");
+    static mark_internal_site_lnk                         = ctRegex!(`¤`, "mg");
+    static quotation_mark_sql_insert_delimiter            = ctRegex!("[']", "mg");
+    static quotation_mark_various                         = ctRegex!(q"¶['‘’“”"`´¨]¶", "mg");
+    /+ inline markup font face mod +/
+    static inline_faces                                   = ctRegex!(`(?P(?P[*!_^,+#-])\{(?P.+?)\}[*!_^,+#-])`, "mg");
+    static inline_emphasis                                = ctRegex!(`\*\{(?P.+?)\}\*`, "mg");
+    static inline_bold                                    = ctRegex!(`!\{(?P.+?)\}!`, "mg");
+    static inline_underscore                              = ctRegex!(`_\{(?P.+?)\}_`, "mg");
+    static inline_italics                                 = ctRegex!(`/\{(?P.+?)\}/`, "mg");
+    static inline_superscript                             = ctRegex!(`\^\{(?P.+?)\}\^`, "mg");
+    static inline_subscript                               = ctRegex!(`,\{(?P.+?)\},`, "mg");
+    static inline_strike                                  = ctRegex!(`-\{(?P.+?)\}-`, "mg");
+    static inline_insert                                  = ctRegex!(`\+\{(?P.+?)\}\+`, "mg");
+    static inline_mono                                    = ctRegex!(`#\{(?P.+?)\}#`, "mg");
+    static inline_mono_box                                = ctRegex!(`■\{(?P.+?)\}■`, "mg");
+    static inline_cite                                    = ctRegex!(`"\{(?P.+?)\}"`, "mg");
+    static inline_faces_line                              = ctRegex!(`^[*!/_]_ (?P.+?)((?: [\\]{2}|[~]#){0,2}$)`);
+    static inline_emphasis_line                           = ctRegex!(`^\*_ (?P.+?)((?: [\\]{2}|[~]#){0,2}$)`);
+    static inline_bold_line                               = ctRegex!(`^!_ (?P.+?)((?: [\\]{2}|[~]#){0,2}$)`);
+    static inline_italics_line                            = ctRegex!(`^/_ (?P.+?)((?: [\\]{2}|[~]#){0,2}$)`);
+    static inline_underscore_line                         = ctRegex!(`^__ (?P.+?)((?: [\\]{2}|[~]#){0,2}$)`);
+    static inline_fontface_clean                          = ctRegex!(`[*!_/^,+#■"-]\{|\}[*!_/^,+#■"-]`, "mg");
+    static no_header_rgx                                  = ctRegex!(`^=NULL$`);
+    /+ table delimiters +/
+    static table_delimiter_col                           = ctRegex!("[ ]*[┊][ ]*", "mg");
+    static table_delimiter_row                           = ctRegex!("[ ]*\n", "mg");
+  }
+}
-- 
cgit v1.2.3