From 22cea68385b57a1593f5751e49cfdd41d8067997 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Thu, 13 Jul 2017 07:44:03 -0400 Subject: 0.18.0 rename meta from ao (considered adr) --- src/sdp/meta/composite_make.d | 109 + src/sdp/meta/conf_make_meta.d | 49 + src/sdp/meta/conf_make_meta_native.d | 330 ++ src/sdp/meta/conf_make_meta_sdlang.d | 155 + src/sdp/meta/defaults.d | 489 +++ src/sdp/meta/doc_debugs.d | 659 ++++ src/sdp/meta/metadoc.d | 141 + src/sdp/meta/metadoc_from_src.d | 5590 ++++++++++++++++++++++++++++++++++ src/sdp/meta/metadoc_summary.d | 81 + src/sdp/meta/object_setter.d | 65 + src/sdp/meta/package.d | 17 + src/sdp/meta/read_config_files.d | 80 + src/sdp/meta/read_source_files.d | 294 ++ src/sdp/meta/rgx.d | 256 ++ 14 files changed, 8315 insertions(+) create mode 100644 src/sdp/meta/composite_make.d create mode 100644 src/sdp/meta/conf_make_meta.d create mode 100644 src/sdp/meta/conf_make_meta_native.d create mode 100644 src/sdp/meta/conf_make_meta_sdlang.d create mode 100644 src/sdp/meta/defaults.d create mode 100644 src/sdp/meta/doc_debugs.d create mode 100644 src/sdp/meta/metadoc.d create mode 100644 src/sdp/meta/metadoc_from_src.d create mode 100644 src/sdp/meta/metadoc_summary.d create mode 100644 src/sdp/meta/object_setter.d create mode 100644 src/sdp/meta/package.d create mode 100644 src/sdp/meta/read_config_files.d create mode 100644 src/sdp/meta/read_source_files.d create mode 100644 src/sdp/meta/rgx.d (limited to 'src/sdp/meta') diff --git a/src/sdp/meta/composite_make.d b/src/sdp/meta/composite_make.d new file mode 100644 index 0000000..dc28206 --- /dev/null +++ b/src/sdp/meta/composite_make.d @@ -0,0 +1,109 @@ +/++ + output hub
+ check & generate output types requested ++/ +module sdp.meta.composite_make; +template compositeMkCnf() { + import sdp.meta; + import std.array; + mixin SiSUrgxInit; + string[] _substitutions; + string[string][] _sub; + string _bold; + string _italics; + string _emphasis; + auto compositeMkCnf(Mks...)(Mks makes) { + foreach (make; makes) { + auto rgx = Rgx(); + if (auto z = "make" in make) { + if (auto x = "substitute" in *z) { + foreach (m; (*x).matchAll(rgx.make_simple_substitutions_d)) { + _sub ~= ["match": (m["match"]), "replace": (m["replace"])]; + _substitutions ~= format("`%s`,\"%s\"", + m["match"], + m["replace"], + ); + } + foreach (m; (*x).matchAll(rgx.make_simple_substitutions_rb)) { + _sub ~= ["match": (m["match"]), "replace": (m["replace"])]; + _substitutions ~= format("`%s`,\"%s\"", + m["match"], + m["replace"], + ); + } + } + if (auto x = "bold" in *z) { + _bold = (*x).to!string; + } + if (auto x = "italics" in *z) { + _italics = (*x).to!string; + } + if (auto x = "emphasis" in *z) { + _emphasis = (*x).to!string; + } + } + } + struct _composite_make { + auto substitutions() { + auto _k = _sub; + return _k; + } + auto substitute() { + auto _k = _substitutions; + return _k; + } + auto italics() { + auto _k = _italics; + return _k; + } + auto bold() { + auto _k = _bold; + return _k; + } + auto emphasis() { + auto _k = _emphasis; + return _k; + } + } + return _composite_make(); + } +} +/++ + output hub
+ check & generate output types requested ++/ +template compositeMkCnfAA() { + import sdp.meta; + import std.array; + mixin SiSUrgxInit; + string[] _substitutions; + string[string][] _sub; + auto rgx = Rgx(); + auto compositeMkCnfAA(Mks...)(Mks makes) { + /+ + - skip.first_make which is the "composite make output" + - pass config files as associative arrays, + - skip.last_make which is getopt, treat separately + +/ + auto _composite_make = makes[0]; + auto _opts = makes[$-1]; + foreach (make; makes[1..$-1]) { + if (auto z = "make" in make) { // document head: make (part of individual document) + if (auto x = "substitute" in *z) { + _composite_make["make"]["substitute"] = *x; + } + if (auto x = "italics" in *z) { + _composite_make["make"]["italics"] = *x; + } + if (auto x = "bold" in *z) { + _composite_make["make"]["bold"] = *x; + } + if (auto x = "emphasis" in *z) { + _composite_make["make"]["emphasis"] = *x; + } + } + } + /+ logic for adding _opts make instructions to _composite make +/ + return _composite_make; + } +} diff --git a/src/sdp/meta/conf_make_meta.d b/src/sdp/meta/conf_make_meta.d new file mode 100644 index 0000000..5beeae2 --- /dev/null +++ b/src/sdp/meta/conf_make_meta.d @@ -0,0 +1,49 @@ +/++ + 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 sdp.meta.conf_make_meta; +template docHeaderMakeAndMetaTupExtractAndConvertToAA() { + import + std.exception, + std.regex, + std.stdio, + std.traits, + std.typecons, + std.utf, + std.conv : to; + import sdlang; + import + sdp.meta.conf_make_meta_native, + sdp.meta.conf_make_meta_sdlang, + sdp.meta.rgx; + mixin SiSUrgxInit; + mixin SiSUheaderExtractNative; + mixin SiSUextractSDLang; + auto rgx = Rgx(); + auto docHeaderMakeAndMetaTupExtractAndConvertToAA(DocMake, Src)( + DocMake conf_doc_make_aa, + Src header_src, + ) { + debug(asserts){ + static assert(is(typeof(header_src) == char[])); + static assert(is(typeof(conf_doc_make_aa) == string[string][string])); + } + auto head_native = HeaderDocMetadataAndMakeNativeToAA(); + auto header_sdlang_tag = (!(header_src.match(rgx.native_header_meta_title))) + ? extractSDL().docHeaderSDLtagGet(header_src) // sdlang.ast.Tag + : null; + auto header_make_and_meta_tuple = (header_src.match(rgx.native_header_meta_title)) + ? (head_native.headerNativeToAA(header_src)) + : (extractSDL().docHeaderSDLtoAA(conf_doc_make_aa, header_sdlang_tag)); + static assert(!isTypeTuple!(header_make_and_meta_tuple)); + static assert(header_make_and_meta_tuple.length==2); + return header_make_and_meta_tuple; + } +} diff --git a/src/sdp/meta/conf_make_meta_native.d b/src/sdp/meta/conf_make_meta_native.d new file mode 100644 index 0000000..fc7912f --- /dev/null +++ b/src/sdp/meta/conf_make_meta_native.d @@ -0,0 +1,330 @@ +/++ + native headers using
@title:
:subtitle:
type tags
+ extract native/orig header return associative array ++/ +module sdp.meta.conf_make_meta_native; +template SiSUheaderExtractNative() { + import + std.exception, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.utf, + std.conv : to; + import + sdp.meta.defaults, + sdp.meta.rgx; + struct HeaderDocMetadataAndMakeNativeToAA { + mixin SiSUregisters; + mixin SiSUrgxInitFlags; + mixin SiSUrgxInit; + auto rgx = Rgx(); + enum State { off, on } + string hm, hs; + auto header_metadata_and_make_aa(H,Me,Ma)( + H header, + Me dochead_meta, + Ma dochead_make + ) + in { + debug(asserts){ + static assert(is(typeof(header) == string)); + static assert(is(typeof(dochead_meta) == string[string][string])); + static assert(is(typeof(dochead_make) == string[string][string])); + } + } + body { + scope(exit) { + destroy(header); + destroy(dochead_meta); + destroy(dochead_make); + } + if (auto t = header.match(rgx.native_header_main)) { + char[][] header_obj_spl = + (cast(char[]) header).split(rgx.line_delimiter_ws_strip); + auto hm = to!string(t.captures[1]); + if (hm.match(rgx.main_headers)) { + foreach (line; header_obj_spl) { + if (auto m = line.match(rgx.native_header_main)) { + if (!empty(m.captures[2])) { + if (hm == "creator") { + dochead_meta[hm]["author"] = + to!string(m.captures[2]); + } else if (hm == "title") { + dochead_meta[hm]["main"] = + to!string(m.captures[2]); + } else if (hm == "publisher") { + dochead_meta[hm]["name"] = + to!string(m.captures[2]); + } + } + } else if (auto s = match(line, rgx.native_header_sub)) { + if (!empty(s.captures[2])) { + auto hs = to!string(s.captures[1]); + if ((hm == "make" ) + && (dochead_make[hm])) { + switch (hm) { + case "make": + if (hs.match(rgx.native_subhead_make)) { + if (dochead_make[hm][hs]) { + dochead_make[hm][hs] = to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + default: + break; + } + } else if (dochead_meta[hm]) { + switch (hm) { + case "creator": + if (hs.match(rgx.native_subhead_creator)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "title": + if (hs.match(rgx.native_subhead_title)) { + if ((hs == "subtitle") + && (dochead_meta[hm]["sub"])) { + dochead_meta[hm]["sub"] = + to!string(s.captures[2]); + } else if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "rights": + if (hs.match(rgx.native_subhead_rights)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "date": + if (hs.match(rgx.native_subhead_date)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "original": + if (hs.match(rgx.native_subhead_original)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "classify": + if (hs.match(rgx.native_subhead_classify)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "identifier": + if (hs.match(rgx.native_subhead_identifier)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "notes": + if (hs.match(rgx.native_subhead_notes)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "publisher": + if (hs.match(rgx.native_subhead_publisher)) { + if (dochead_meta[hm][hs]) { + dochead_meta[hm][hs] = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "links": + destroy(hm); + destroy(hs); + break; + default: + break; + } + } + } + } + } + } else { + writeln("not a valid header type:", hm); + } + } + auto t = tuple(dochead_meta, dochead_make); + static assert(t.length==2); + return t; + } + private auto native_header_extract(L,Lo,O,T)( + L line, + return ref Lo line_occur, + return ref O an_object, + return ref T type + ) { + 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(type) == int[string])); + } + if (line.matchFirst(rgx.native_header_make)) { /+ matched header_make +/ + debug(header1) { /+ writeln(line); +/ } + type["header"] = State.on; + type["header_make"] = State.on; + type["header_meta"] = State.off; + ++line_occur["header_make"]; + an_object["body_nugget"] ~= line ~= "\n"; + } else if (line.matchFirst(rgx.native_header)) { /+ matched header_metadata +/ + /+ (generic header match and not previously caught by header_make) +/ + debug(header1) { /+ writeln(line); +/ } + type["header"] = State.on; + type["header_make"] = State.off; + type["header_meta"] = State.on; + ++line_occur["header_meta"]; + an_object["body_nugget"] ~= line ~= "\n"; + } else if (type["header_make"] == State.on + && (line_occur["header_make"] > State.off)) { /+ header_make flag set +/ + if (line.matchFirst(rgx.native_header_sub)) { /+ sub-header +/ + debug(header1) { /+ writeln(line); +/ } + ++line_occur["header_make"]; + an_object["body_nugget"] ~= line ~= "\n"; + } + } else if (type["header_meta"] == State.on + && (line_occur["header_meta"] > State.off)) { /+ header_metadata flag set +/ + if (line.matchFirst(rgx.native_header_sub)) { /+ sub-header +/ + debug(header1) { /+ writeln(line); +/ } + ++line_occur["header_meta"]; + an_object["body_nugget"] ~= line ~= "\n"; + } + } + return an_object; + } + auto header_reset_states_common(Lo,O,T)( + return ref Lo line_occur, + return ref O an_object, + return ref T type + ) { + debug(asserts){ + static assert(is(typeof(line_occur) == int[string])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + line_occur["header_make"] = State.off; + line_occur["header_meta"] = State.off; + type["header"] = State.off; + an_object.remove("body_nugget"); + an_object.remove("is"); + an_object.remove("attrib"); + } + private auto headerNativeToAA(Hn)(Hn src_header) { + debug(asserts){ + static assert(is(typeof(src_header) == char[])); + } + auto type = flags_type_init; + type = [ + "header" : State.off, + "header_make" : State.off, + "header_meta" : State.off, + ]; + string[string] an_object; + int[string] line_occur; + auto dochead_make = conf_aa_empty; + auto dochead_meta = meta_aa_empty; + auto set_header = HeaderDocMetadataAndMakeNativeToAA(); + char[][] source_header_arr = + (cast(char[]) src_header).split(rgx.newline_eol_delimiter); + foreach(header_line; source_header_arr) { + if (auto m = header_line.matchFirst(rgx.comment)) { + /+ matched comment +/ + debug(comment) { + } + header_reset_states_common(line_occur, an_object, type); + } else if ((header_line.matchFirst(rgx.native_header)) + || (type["header_make"] == State.on + && (line_occur["header_make"] > State.off)) + || (type["header_meta"] == State.on + && (line_occur["header_meta"] > State.off))) { + if (header_line.length == 0) { + /+ header_make instructions (current line empty) +/ + auto dochead_metadata_and_make = + set_header.header_metadata_and_make_aa(strip(an_object["body_nugget"]), dochead_meta, dochead_make); + static assert(!isTypeTuple!(dochead_metadata_and_make)); + dochead_meta = dochead_metadata_and_make[0]; + dochead_make = dochead_metadata_and_make[1]; + header_reset_states_common(line_occur, an_object, type); + type["header_make"] = State.off; + type["header_meta"] = State.off; + debug(headersdlang) { + writeln(dochead_metadata_and_make); + } + } else { + an_object = native_header_extract(header_line, line_occur, an_object, type); + } + } + } + auto t = tuple( + dochead_make, + dochead_meta, + ); + return t; + } + } +} diff --git a/src/sdp/meta/conf_make_meta_sdlang.d b/src/sdp/meta/conf_make_meta_sdlang.d new file mode 100644 index 0000000..f4af643 --- /dev/null +++ b/src/sdp/meta/conf_make_meta_sdlang.d @@ -0,0 +1,155 @@ +/++ + sdlang headers
+ extract sdlang header return sdlang ++/ +module sdp.meta.conf_make_meta_sdlang; +template SiSUextractSDLang() { + import + std.exception, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.utf, + std.conv : to; + import + sdp.meta.defaults, + sdp.meta.rgx; + struct extractSDL { + mixin SiSUregisters; + mixin SiSUrgxInit; + auto rgx = Rgx(); + private auto docHeaderSDLtagGet(Hs)(Hs src_header) { + debug(asserts){ + static assert(is(typeof(src_header) == char[])); + } + char[][] source_header_arr = + (cast(char[]) src_header).split(rgx.newline_eol_delimiter); + char[] _src_header; + foreach(header_line; source_header_arr) { + if (!match(header_line, rgx.comments)) { + _src_header ~= header_line ~ "\n"; + } + } + scope(failure) { + stderr.writefln( + "%s\n%s\n%s:%s failed here:\n _src_header: %s", + __MODULE__, __FUNCTION__, + __FILE__, __LINE__, + _src_header, + ); + } + Tag sdl_root_header; + try { + sdl_root_header = parseSource(_src_header.to!string); + } + catch(ParseException e) { + stderr.writeln("SDLang problem with this document header:"); + stderr.writeln(_src_header); + // Error messages of the form: + // myFile.sdl(5:28): Error: Invalid integer suffix. + stderr.writeln(e.msg); + } + debug(sdlang) { + writeln("header SDL:"); + writeln(__LINE__, ": ", sdl_root_header.toSDLDocument()); + writeln(__LINE__, ": ", sdl_root_header.maybe.namespaces); + writeln("header make sdlang: ", sdl_root_header.toSDLDocument()); + writeln(__LINE__, ": ", sdl_root_header.getTagValues("title")); + writeln(__LINE__, ": ", sdl_root_header.getTagValues("creator")); + Tag creator = sdl_root_header.getTag("creator"); + if (creator !is null) { + if ("author" in creator.maybe.tags) { + writeln(__LINE__, ": ", creator.getTagValues("author")); + } else if ("author" in creator.maybe.attributes) { + writeln(__LINE__, ": ", creator.maybe.attributes["author"][0].value); + } + } + } + return sdl_root_header; // sdlang.ast.Tag + } + private auto sdlangToAA(C,Tag)(C conf, Tag conf_sdlang) { + debug(asserts){ + static assert(is(typeof(conf) == string[string][string])); + } + foreach (maintag, subtags; conf) { + /+ writeln(__LINE__, ": ", maintag, ":- ", subtags); +/ + foreach (subtag, content; subtags) { + if (maintag in conf_sdlang.maybe.tags) { + Tag _maintag = conf_sdlang.getTag(maintag); + if ((subtag in _maintag.maybe.tags) + && (_maintag.getTagValues(subtag).length > 0)) { + debug(headersdlang) { + writeln(__LINE__, ": ", maintag, ":", subtag, ": ", _maintag.getTagValues(subtag)[0]); + } + if (_maintag.getTagValues(subtag).length == 1) { + conf[maintag][subtag] = + (_maintag.getTagValues(subtag)[0]).to!string; + } else if (_maintag.getTagValues(subtag).length > 1) { + foreach (st; _maintag.getTagValues(subtag)) { + conf[maintag][subtag] ~= + st.to!string ~ ";"; + } + } + } else if ((subtag in _maintag.maybe.attributes) + && (_maintag.maybe.attributes[subtag][0].value.length > 0)) { + debug(headersdlang) { + writeln(__LINE__, ": ", maintag, ":", subtag, ": ", conf_sdlang.tags[maintag][0].attributes[subtag][0].value); + } + conf[maintag][subtag] = + (conf_sdlang.tags[maintag][0].attributes[subtag][0].value).to!string; + } + } + } + } + return conf; + } + private auto docHeaderSDLtoAA(Ma, Tag)(Ma dochead_make, Tag header_sdlang) { + debug(asserts){ + static assert(is(typeof(dochead_make) == string[string][string])); + } + dochead_make = sdlangToAA(dochead_make, header_sdlang); + auto dochead_meta = sdlangToAA(meta_aa_empty, header_sdlang); + if (dochead_meta["title"]["main"].empty) { + { + Tag _maintag = header_sdlang.getTag("title"); + if (_maintag !is null) { + if ("main" in _maintag.maybe.tags) { + dochead_meta["title"]["main"] = + to!string(_maintag.getTagValues("main")); + } else if ("main" !in _maintag.maybe.attributes) { + writeln(__LINE__, ": ", _maintag.values[0]); + dochead_meta["title"]["main"] = + (_maintag.values[0]).to!string; // test that this exists + } + } + } + } + if (!(dochead_meta["title"]["subtitle"].empty) + && (dochead_meta["title"]["sub"].empty)) { + dochead_meta["title"]["sub"] ~= dochead_meta["title"]["subtitle"]; + } + dochead_meta["title"].remove("subtitle"); + if (dochead_meta["title"]["sub"].empty) { + dochead_meta["title"]["full"] ~= dochead_meta["title"]["main"]; + } else { + dochead_meta["title"]["full"] ~= format( + "%s - %s", + dochead_meta["title"]["main"], + dochead_meta["title"]["sub"], + ); + } + dochead_meta["creator"]["author_raw"] = dochead_meta["creator"]["author"]; + string[] authors_arr; + auto authors_raw_arr = dochead_meta["creator"]["author"].split(rgx.arr_delimiter); + foreach (author_raw; authors_raw_arr) { + authors_arr ~= author_raw.replace(rgx.raw_author_munge, "$2 $1"); + } + dochead_meta["creator"]["author"] = join(authors_arr, ", ").chomp.chomp; + auto t = tuple(dochead_make, dochead_meta); + static assert(t.length==2); + return t; + } + } +} diff --git a/src/sdp/meta/defaults.d b/src/sdp/meta/defaults.d new file mode 100644 index 0000000..fb95e3a --- /dev/null +++ b/src/sdp/meta/defaults.d @@ -0,0 +1,489 @@ +/++ + default settings ++/ +module sdp.meta.defaults; +template SiSUregisters() { + import + std.algorithm, + std.array, + std.container, + std.exception, + std.file, + std.getopt, + std.json, + std.path, + std.process, + std.range, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.uni, + std.utf, + std.conv : to; + string[string][string] conf_aa_empty() { + auto conf_ = [ + "webserv": [ + "url_root" : "", + "path" : "~/sdp_www" , + "images" : "" , + "cgi" : "/usr/local/lib/sdp-cgi" + ], + "webserv_cgi": [ + "host" : "localhost", + "base_path" : "", + "port" : "8081", + "user" : "", + "file_links" : "www.sisudoc.org" + ], + "processing": [ + "path" : "~", + "dir" : "_sisu_processing", + "concord_max" : "400000" + ], + "flag": [ + "act0" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --xhtml --xml-sax --xml-dom --sqlite --manifest --verbose", + "act1" : "--digest --text --html --manifest", + "act2" : "--digest --text --html --epub --pdf --manifest", + "act3" : "--digest --qrcode --text --html --epub --concordance --pdf --manifest", + "act4" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --manifest", + "act5" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --sqlite --manifest", + "act6" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --xhtml --xml-sax --xml-dom --sqlite --manifest", + "act7" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --xhtml --xml-sax --xml-dom --sqlite --source --sisupod --manifest", + "act8" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --xhtml --xml-sax --xml-dom --pg --update --manifest", + "act9" : "--digest --qrcode --text --html --epub --concordance --pdf --odf --docbook --xhtml --xml-sax --xml-dom --pg --update --source --sisupod --manifest" + ], + "default": [ + "papersize" : "a4,letter", + "text_wrap" : "78", + "emphasis" : "bold", + "language" : "en", + "digest" : "sha256" + ], + "permission": [ + "share_source" : "" + ], + "program_select": [ + "editor" : "vim", + "epub_viewer" : "ebook-viewer", + "html_viewer" : "xombrero", + "odf_viewer" : "lowriter", + "pdf_viewer" : "evince", + "xml_viewer" : "xml-viewer" + ], + "search": [ + "flag" : "", + "action" : "", + "db" : "", + "title" : "" + ], + "make": [ + "bold" : "", + "breaks" : "", + "cover_image" : "", + "css" : "", + "emphasis" : "", + "footer" : "", + "headings" : "", + "home_button_image" : "", + "home_button_text" : "", + "italics" : "", + "num_top" : "", + "num_depth" : "", + "substitute" : "", + "texpdf_font" : "" + ], + ]; + return conf_; + } + string[string][string] meta_aa_empty() { + auto meta_ = [ + "classify": [ + "dewey" : "", + "keywords" : "", + "loc" : "", + "subject" : "", + "topic_register" : "" + ], + "creator": [ + "author" : "", + "author_email" : "", + "illustrator" : "", + "translator" : "" + ], + "date": [ + "added_to_site" : "", + "available" : "", + "created" : "", + "issued" : "", + "modified" : "", + "published" : "", + "valid" : "" + ], + "identifier": [ + "isbn" : "", + "oclc" : "", + "pg" : "" + ], + "links": [ + "link" : "" + ], + "notes": [ + "abstract" : "", + "description" : "" + ], + "original": [ + "language" : "", + "source" : "", + "title" : "" + ], + "publisher": [ + "name" : "" + ], + "rights": [ + "copyright" : "", + "cover" : "", + "illustrations" : "", + "license" : "" + ], + "title": [ + "edition" : "", + "full" : "", + "language" : "", + "main" : "", + "note" : "", + "sub" : "", + "subtitle" : "" + ] + ]; + return meta_; + } + auto ptr_head_main = + [ + "classify", + "creator", + "date", + "identifier", + "links", + "make", + "original", + "notes", + "rights", + "title" + ]; + auto ptr_head_sub_classify = + [ + "dewey", + "keywords", + "loc", + "subject", + "topic_register" + ]; + auto ptr_head_sub_creator = + [ + "author", + "author_email", + "cover", + "illustrator", + "translator" + ]; + auto ptr_head_sub_date = + [ + "added_to_site", + "available", + "created", + "issued", + "modified", + "published", + "valid" + ]; + auto ptr_head_sub_identifier = + [ + "isbn", + "oclc", + "pg" + ]; + /+ make +/ + auto ptr_head_sub_make = + [ + "cover_image", + "home_button_image", + "home_button_text", + "footer", "headings", + "num_top", "num_depth", + "breaks", + "substitute", + "bold", + "italics", + "emphasis", + "texpdf_font", + "css" + ]; + auto ptr_head_sub_notes = + [ + "abstract", + "description" + ]; + auto ptr_head_sub_original = + [ + "language", + "source", + "title" + ]; + auto ptr_head_sub_publisher = + [ "name" ]; + auto ptr_head_sub_rights = + [ + "copyright", + "cover", + "illustrations", + "license" + ]; + auto ptr_head_sub_title = + [ + "edition", + "full", + "language", + "main", + "note", + "sub" + ]; + auto config_jsonstr = `{ + }`; +} +template SiSUrgxInitFlags() { + /+ regex flags +/ + 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 obj_cite_number; 1 no obj_cite_number; 2 no obj_cite_number & dummy headings + "ocn_status_multi_obj" : 0, // 0 obj_cite_number; 1 no obj_cite_number; 2 no obj_cite_number & dummy headings + "book_index" : 0, + ]; + return flags_type_init; + } +} +template SiSUnode() { + string[string] node_metadata_heading_str() { + auto _node = [ + "is" : "", + "ocn" : "", + "marked_up_lev" : "", + "segment_anchor_tag" : "", + "attrib" : "", + ]; + return _node; + } + 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; + } + string[string] node_metadata_para_str() { + auto _node = [ + "is" : "", + "ocn" : "", + "attrib" : "", + ]; + return _node; + } + 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 { + 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() { + 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 = "■"; + string indent_by_spaces_provided(int indent, string _indent_spaces ="░░") { + _indent_spaces = replicate(_indent_spaces, indent); + return _indent_spaces; + } + 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 { + 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; + } + 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; + } + 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; + } + auto codes_() { + return "(" ~ join(code_arr,"|") ~ ")"; + } + auto codes_regex() { + return regex(codes_); + } + } +} diff --git a/src/sdp/meta/doc_debugs.d b/src/sdp/meta/doc_debugs.d new file mode 100644 index 0000000..6d94b05 --- /dev/null +++ b/src/sdp/meta/doc_debugs.d @@ -0,0 +1,659 @@ +/++ + output debugs ++/ +module sdp.meta.doc_debugs; +template SiSUdebugs() { + import + sdp.meta.defaults, + sdp.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)( + auto return ref const S contents, + auto return ref T doc_matters, + ) { + mixin SiSUrgxInit; + mixin InternalMarkup; + auto rgx = Rgx(); + auto markup = InlineMarkup(); + string key; + debug(parent) { + writefln( + "%s:%s", + __FILE__, + __LINE__, + ); + foreach (key; doc_matters.keys_seq.seg) { + foreach (obj; contents[key]) { + if (obj.use != "empty") { + if (obj.is_a == "heading") { + writefln( + "%s node: %s heading: %s %s", + obj.obj_cite_number, + obj.node, + obj.heading_lev_markup, + obj.text, + ); + } + } + } + } + } + debug(dumpdoc) { + writefln( + "%s\n%s:%s", + "-------------------------------", + __FILE__, + __LINE__, + ); + foreach (obj; contents[key]) { + if (obj.use != "empty") { + writefln( + "[%s][%s]\n%s", + obj.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + debug(section_head) { + key="head"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.obj_cite_number, + obj.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.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + debug(toc_nav_dom) { + enum DomTags { none, open, close, close_and_open, open_still, } + foreach (sect; doc_matters.keys_seq.seg) { + foreach (obj; contents[sect]) { + if (obj.is_a == "heading") { + foreach_reverse (k; 0 .. 7) { + switch (obj.dom_markedup[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.keys_seq.seg) { + foreach (obj; contents[sect]) { + if (obj.is_a == "heading") { + foreach_reverse (k; 0 .. 7) { + switch (obj.dom_collapsed[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(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.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + debug(section_bibliography) { + key="bibliography"; + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.obj_cite_number, + obj.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.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + debug(objects) { + writefln( + "%s\n%s:%s", + "-------------------------------", + __FILE__, + __LINE__, + ); + foreach (obj; contents[key]) { + if (obj.use != "empty") { + writefln( + "* [%s][%s] %s", + obj.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + debug(headermakejson) { + writefln( + "%s\n%s\n%s", + "document header, metadata & make instructions:", + doc_matters.dochead_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.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + default: + break; + } + } + } + debug(headermetadatajson) { + writefln( + "%s\n%s\n%s", + "document header, metadata & make instructions:", + doc_matters.dochead_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.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "title": + foreach (sub_header; ptr_head_sub_title) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "rights": + foreach (sub_header; ptr_head_sub_rights) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "date": + foreach (sub_header; ptr_head_sub_date) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "original": + foreach (sub_header; ptr_head_sub_original) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "classify": + foreach (sub_header; ptr_head_sub_classify) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "identifier": + foreach (sub_header; ptr_head_sub_identifier) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "notes": + foreach (sub_header; ptr_head_sub_notes) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + case "publisher": + foreach (sub_header; ptr_head_sub_publisher) { + if (doc_matters.dochead_meta[main_header][sub_header].to!string.length > 2) { + writefln( + "%s:%s: %s", + main_header, + sub_header, + doc_matters.dochead_meta[main_header][sub_header] + ); + } + } + break; + default: + break; + } + } + } + debug(anchor) { + writefln( + "%s\n%s:%s", + "-------------------------------", + __FILE__, + __LINE__, + ); + foreach (k; doc_matters.keys_seq.seg) { + foreach (obj; contents[k]) { + if (obj.is_a == "heading") { + writefln( + "%s~ [%s] %s %s", + obj.marked_up_level, + obj.obj_cite_number, + obj.anchor_tags, + // "[", obj["is"], "] ", + obj.text + ); + } + } + } + } + debug(heading) { // heading + foreach (k; doc_matters.keys_seq.seg) { + foreach (o; contents[k]) { + if (o.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\n%s:%s", + "-------------------------------", + __FILE__, + __LINE__, + ); + foreach (k; doc_matters.keys_seq.seg) { + foreach (obj; contents[k]) { + if (obj.is_a == "heading") { + writefln( + "%s~ [%s] %s", + obj.marked_up_level, + obj.obj_cite_number, + // "[", obj["is"], "] ", + obj.text + ); + } + } + } + } + debug(summary) { + string[string] check = [ + "last_obj_cite_number" : "NA [debug \"checkdoc\" not run]", + ]; + } + debug(checkdoc) { + if ((doc_matters.opt_action["debug"])) { + debug(checkdoc) { + if (auto mfn=match(doc_matters.source_filename, rgx.src_fn)) { + if (doc_matters.opt_action["assertions"]) { + switch (mfn.captures[2]) { + // live manual: + case "live-manual.ssm": + assert(check["last_obj_cite_number"] == + "1019","last obj_cite_number should be: 1019 (check test, document is frequently updated)"); // ok + break; + // sisu_markup: + case "sisu_markup.sst": + assert(check["last_obj_cite_number"] == + "297","last obj_cite_number expected to be: 297 rather than " ~ check["last_obj_cite_number"]); // ok + // assert(check["last_obj_cite_number"] == "297","last obj_cite_number expected to be: 297 rather than " ~ check["last_obj_cite_number"]); + // notes for first divergance study sisu headings 247 250 + // sisu has issue with code that contains heading 1~ which results in no obj_cite_number! ?? + // sisu currently has incorrect last body obj_cite_number of 294! + // bug in sisu? attend + break; + // sisu-markup-samples: + case "accelerando.charles_stross.sst": + assert(check["last_obj_cite_number"] == + "2861","last obj_cite_number expected to be: 2861 rather than " ~ check["last_obj_cite_number"]); // ok + break; + case "alices_adventures_in_wonderland.lewis_carroll.sst": + assert(check["last_obj_cite_number"] == + "805","last obj_cite_number expected to be: 805 rather than " ~ check["last_obj_cite_number"]); // 808 + break; + case "autonomy_markup0.sst": + assert(check["last_obj_cite_number"] == + "77","last obj_cite_number expected to be: 77 rather than " ~ check["last_obj_cite_number"]); // ok endnotes + // assert(check["last_obj_cite_number"] == "78","last obj_cite_number expected to be: 78 rather than " ~ check["last_obj_cite_number"]); + break; + case "content.cory_doctorow.sst": + assert(check["last_obj_cite_number"] == + "953","last obj_cite_number expected to be: 953 rather than " ~ check["last_obj_cite_number"]); // 1007 way off, check obj_cite_number off switches + // assert(check["last_obj_cite_number"] == "953","last obj_cite_number expected to be: 953 rather than " ~ check["last_obj_cite_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_obj_cite_number"] == + "905","last obj_cite_number expected to be: 905 rather than " ~ check["last_obj_cite_number"]); // 911 + break; + case "down_and_out_in_the_magic_kingdom.cory_doctorow.sst": + assert(check["last_obj_cite_number"] == + "1417","last obj_cite_number expected to be: 1417 rather than " ~ check["last_obj_cite_number"]); // 1455 check obj_cite_number off switches + break; + case "for_the_win.cory_doctorow.sst": + assert(check["last_obj_cite_number"] == + "3510","last obj_cite_number expected to be: 3510 rather than " ~ check["last_obj_cite_number"]); // 3569 check obj_cite_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_obj_cite_number"] == + "1082","last obj_cite_number expected to be: 1082 rather than " ~ check["last_obj_cite_number"]); // check 1079 too few + break; + case "free_culture.lawrence_lessig.sst": + assert(check["last_obj_cite_number"] == + "1330","last obj_cite_number expected to be: 1330 rather than " ~ check["last_obj_cite_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_obj_cite_number"] == + "1559","last obj_cite_number expected to be: 1559 rather than " ~ check["last_obj_cite_number"]); // 1560, check obj_cite_number off switches, has endnotes so 2 too many + // assert(check["last_obj_cite_number"] == "1559","last obj_cite_number expected to be: 1559 rather than " ~ check["last_obj_cite_number"]); + break; + case "gpl2.fsf.sst": + assert(check["last_obj_cite_number"] == + "65","last obj_cite_number expected to be: 65 rather than " ~ check["last_obj_cite_number"]); // ok endnotes? check + // assert(check["last_obj_cite_number"] == "66","last obj_cite_number expected to be: 66 rather than " ~ check["last_obj_cite_number"]); + break; + case "gpl3.fsf.sst": + assert(check["last_obj_cite_number"] == + "123","last obj_cite_number expected to be: 123 rather than " ~ check["last_obj_cite_number"]); // ok + break; + case "gullivers_travels.jonathan_swift.sst": + assert(check["last_obj_cite_number"] == + "668","last obj_cite_number expected to be: 668 rather than " ~ check["last_obj_cite_number"]); // 674 + break; + case "little_brother.cory_doctorow.sst": + assert(check["last_obj_cite_number"] == + "3130","last obj_cite_number expected to be: 3130 rather than " ~ check["last_obj_cite_number"]); // 3204, check obj_cite_number off switches + break; + case "the_cathedral_and_the_bazaar.eric_s_raymond.sst": + assert(check["last_obj_cite_number"] == + "258","last obj_cite_number expected to be: 258 rather than " ~ check["last_obj_cite_number"]); // ok + break; + case "the_public_domain.james_boyle.sst": + assert(check["last_obj_cite_number"] == + "970","last obj_cite_number expected to be: 970 rather than " ~ check["last_obj_cite_number"]); // 978 + break; + case "the_wealth_of_networks.yochai_benkler.sst": // endnotes, bookindex + assert(check["last_obj_cite_number"] == + "829","last obj_cite_number expected to be: 829 rather than " ~ check["last_obj_cite_number"]); // ok + // assert(check["last_obj_cite_number"] == "832","last obj_cite_number expected to be: 832 rather than " ~ check["last_obj_cite_number"]); + // has endnotes and bookindex, issue with sisu.rb + break; + case "through_the_looking_glass.lewis_carroll.sst": + assert(check["last_obj_cite_number"] == + "949","last obj_cite_number expected to be: 949 rather than " ~ check["last_obj_cite_number"]); // 955 + break; + case "two_bits.christopher_kelty.sst": // endnotes, bookindex, biblio + assert(check["last_obj_cite_number"] == + "1190","last obj_cite_number expected to be: 1190 rather than " ~ check["last_obj_cite_number"]); // 1191 + // assert(check["last_obj_cite_number"] == "1193","last obj_cite_number expected to be: 1193 rather than " ~ check["last_obj_cite_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_obj_cite_number"] == + "377","last obj_cite_number expected to be: 377 rather than " ~ check["last_obj_cite_number"]); // ok + break; + case "viral_spiral.david_bollier.sst": // endnotes, bookindex + assert(check["last_obj_cite_number"] == + "1078","last obj_cite_number expected to be: 1078 rather than " ~ check["last_obj_cite_number"]); // 1100 + // fixed ERROR! range violation! + // error in bookindex ... (ch7 ... building the cc machine, an extra semi colon) + break; + default: + writeln(doc_matters.source_filename); + break; + } + } + } + } + debug(checkdoc) { + void out_segnames(S,T)( + auto return ref const S contents, + auto return ref T doc_matters, + ) { + foreach (key; doc_matters.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.segnames[obj.ptr_html_segnames], ") -> ", obj.text); + } + } + } + } + } + } + debug(checkdoc) { + void out_toc(S)( + auto return ref 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)( + auto return ref const S contents, + string key, + ) { + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s]\n%s", + obj.is_a, + obj.text + ); + } + } + } + } + debug(checkdoc) { + void out_bookindex(S)( + auto return ref const S contents, + string key, + ) { + if (contents[key].length > 1) { + foreach (obj; contents[key]) { + writefln( + "[%s][%s]\n%s", + obj.obj_cite_number, + obj.is_a, + obj.text + ); + } + } + } + } + } + } + } +} diff --git a/src/sdp/meta/metadoc.d b/src/sdp/meta/metadoc.d new file mode 100644 index 0000000..4174b67 --- /dev/null +++ b/src/sdp/meta/metadoc.d @@ -0,0 +1,141 @@ +module sdp.meta.metadoc; +template SiSUabstraction() { + /+ sdp: sisu document parser, see http://sisudoc.org +/ + import sdp.meta; + import + std.getopt, + std.process; + import + sdp.meta.metadoc_summary, + sdp.meta.metadoc_from_src, + sdp.meta.composite_make, + sdp.meta.conf_make_meta, + // sdp.meta.conf_make_meta_native, + sdp.meta.conf_make_meta_sdlang, + sdp.meta.defaults, + sdp.meta.doc_debugs, + sdp.meta.read_config_files, + sdp.meta.read_source_files, + sdp.meta.rgx, + sdp.output.hub, + sdp.output.paths_source; + mixin SiSUrgxInit; + mixin SiSUregisters; + mixin SiSUextractSDLang; + mixin SiSUnode; + mixin SiSUbiblio; + mixin SiSUrgxInitFlags; + mixin outputHub; + enum headBody { header, body_content, insert_filelist } + enum makeMeta { make, meta } + enum docAbst { doc_abstraction, section_keys, segnames, segnames_0_4, images } + auto rgx = Rgx(); + auto SiSUabstraction(Fn,O,E)(Fn fn_src, O opts, E env){ + auto sdl_root_config_share = configRead!()("config_local", env); + auto sdl_root_config_local = configRead!()("config_share", env); + auto conf_composite_static_aa = extractSDL().sdlangToAA(conf_aa_empty, sdl_root_config_share); + conf_composite_static_aa = extractSDL().sdlangToAA(conf_composite_static_aa, sdl_root_config_local); + /+ ↓ read file (filename with path) +/ + /+ ↓ file tuple of header and content +/ + auto _header_body_inserts = + SiSUrawMarkupContent!()(fn_src); + static assert(!isTypeTuple!(_header_body_inserts)); + static assert(_header_body_inserts.length==3); + debug(header_and_body) { + writeln(header); + writeln(_header_body_inserts.length); + writeln(_header_body_inserts.length[headBody.body_content][0]); + } + /+ ↓ split header into make and meta +/ + auto _make_and_meta_tup = + docHeaderMakeAndMetaTupExtractAndConvertToAA!()(conf_composite_static_aa, _header_body_inserts[headBody.header]); + static assert(!isTypeTuple!(_make_and_meta_tup)); + static assert(_make_and_meta_tup.length==2); + auto _make_config = compositeMkCnf!()( + conf_composite_static_aa, + _make_and_meta_tup[makeMeta.make], + // opts, + ); + auto _make_and_conf_composite_static_plus_docheader_aa = compositeMkCnfAA!()( + conf_aa_empty, + conf_composite_static_aa, + _make_and_meta_tup[makeMeta.make], + opts, + ); + /+ ↓ document abstraction: process document, return abstraction as tuple +/ + auto da = SiSUdocAbstraction!()( + _header_body_inserts[headBody.body_content], + _make_and_meta_tup[makeMeta.make], + _make_and_meta_tup[makeMeta.meta], + opts + ); + 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]; + struct DocumentMatters { + auto keys_seq() { + /+ contains .seg & .scroll sequences +/ + auto _k = _document_section_keys_sequenced; + return _k; + } + string[] segnames() { + string[] _k = _doc_html_segnames; + return _k; + } + string[] segnames_lv_0_to_4() { + string[] _k = _doc_epub_segnames_0_4; + return _k; + } + auto dochead_meta() { + string[string][string] _k = _make_and_meta_tup[makeMeta.meta]; + return _k; + } + auto dochead_make() { + string[string][string] _k = _make_and_meta_tup[makeMeta.make]; + return _k; + } + auto source_filename() { + string _k = fn_src; + return _k; + } + auto src_path_info() { + string _pwd = env["pwd"]; + auto _k = SiSUpathsSRC!()(_pwd, fn_src); + return _k; + } + auto opt_action() { + bool[string] _k = opts; + return _k; + } + auto environment() { + auto _k = env; + return _k; + } + auto language() { + string _k; + if (auto m = fn_src.match(rgx.language_code_and_filename)) { + _k = m.captures[1]; + } else { + _k = "en"; + } + return _k; + } + auto file_insert_list() { + string[] _k = _header_body_inserts[headBody.insert_filelist]; + return _k; + } + auto image_list() { + auto _k = _images; + return _k; + } + } + auto doc_matters = DocumentMatters(); + auto t = tuple(doc_abstraction, doc_matters); + static assert(t.length==2); + return t; + } +} diff --git a/src/sdp/meta/metadoc_from_src.d b/src/sdp/meta/metadoc_from_src.d new file mode 100644 index 0000000..ab7f883 --- /dev/null +++ b/src/sdp/meta/metadoc_from_src.d @@ -0,0 +1,5590 @@ +/++ + document abstraction: + abstraction of sisu markup for downstream processing + meta_metadoc_from_src.d ++/ +module sdp.meta.metadoc_from_src; +template SiSUdocAbstraction() { + /+ ↓ abstraction imports +/ + import sdp.meta; + import + std.algorithm, + std.container, + std.file, + std.json, + std.path; + import + sdp.meta.defaults, + sdp.meta.object_setter, + sdp.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 + /+ 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"; + auto obj_im = ObjInlineMarkup(); + auto obj_att = ObjAttributes(); + /+ ocn +/ + int obj_cite_number, obj_cite_number_; + auto object_citation_number = OCNemitter(); + int[] dom_markedup = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_markedup_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_collapsed = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_collapsed_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + enum DomTags { none, open, close, close_and_open, open_still, } + auto obj_heading_ancestors(O)( + O obj, + string[] lv_ancestors_txt, + ) { + switch (obj.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.heading_ancestors_text = lv_ancestors_txt.dup; + } + return obj; + } + auto obj_dom_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.dom_markedup = dom.dup; + return obj; + } + 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.dom_collapsed = dom.dup; + return obj; + } + int 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 } + /+ ↓ abstract marked up document +/ + auto SiSUdocAbstraction(Src,Make,Meta,Opt)( + Src markup_sourcefile_content, + Make dochead_make_aa, + Meta dochead_meta_aa, + Opt opt_action, + ) { + auto rgx = Rgx(); + debug(asserts) { + static assert(is(typeof(markup_sourcefile_content) == char[][])); + static assert(is(typeof(dochead_make_aa) == string[string][string])); + static assert(is(typeof(dochead_meta_aa) == string[string][string])); + static assert(is(typeof(opt_action) == bool[string])); + } + /+ ↓ 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 type = flags_type_init; + string[string] obj_cite_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_.use = "frontmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Table of Contents"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "toc"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.ptr_html_segnames = html_segnames_ptr; + comp_obj_heading_.anchor_tags = ["toc"]; + comp_obj_heading_.dom_markedup = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.dom_collapsed = [ 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], + ]; + auto mkup = InlineMarkup(); + 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 ↑ +/ + /+ ↓ loop markup document/text line by line +/ + srcDocLoop: + foreach (line; markup_sourcefile_content) { + /+ ↓ 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) { // source lines + writeln(line); + } + debug(srclines) { + if (!line.empty) { // source lines, not empty + writefln( + "* %s", + line + ); + } + } + if (!line.empty) { + _check_ocn_status_(line, type); + } + if (type["code"] == TriState.on) { + /+ block object: code +/ + _code_block_(line, an_object, type); + 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) + || (type["biblio_section"] == State.on + && (!(line.matchFirst(rgx.heading_blurb_glossary))) + && (!(line.matchFirst(rgx.heading))) + && (!(line.matchFirst(rgx.comment))))) { + /+ within section (block object): biblio +/ + type["glossary_section"] = State.off; + type["biblio_section"] = State.on; + type["blurb_section"] = State.off; + if (opt_action["backmatter"] && opt_action["section_biblio"]) { + _biblio_block_(line, type, 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) + || (type["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); + } + type["glossary_section"] = State.on; + type["biblio_section"] = State.off; + type["blurb_section"] = State.off; + if (opt_action["backmatter"] && opt_action["section_glossary"]) { + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + type["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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Glossary"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "_part_glossary"; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.dom_markedup = [ 1, 1, 0, 0, 0, 0, 0, 0]; + comp_obj_heading_.dom_collapsed = [ 1, 1, 0, 0, 0, 0, 0, 0]; + the_glossary_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Glossary"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "glossary"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 2; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.dom_markedup = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.dom_collapsed = [ 1, 1, 1, 0, 0, 0, 0, 0]; + comp_obj_heading_.anchor_tags = ["glossary"]; + the_glossary_section ~= comp_obj_heading_; + } else { + _para_match_(line, an_object, an_object_key, indent, bullet, type, line_occur); + comp_obj_para = comp_obj_para.init; + comp_obj_para.use = "backmatter"; + comp_obj_para.is_of = "para"; + comp_obj_para.is_a = "glossary"; + comp_obj_para.text = line.to!string.strip; + comp_obj_para.ocn = 0; + comp_obj_para.obj_cite_number = ""; + comp_obj_para.indent_hang = indent["hang_position"]; + comp_obj_para.indent_base = indent["base_position"]; + comp_obj_para.bullet = bullet; + the_glossary_section ~= comp_obj_para; + } + type["ocn_status"] = TriState.off; + } + continue; + } else if (line.matchFirst(rgx.heading_blurb) + || (type["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); + } + type["glossary_section"] = State.off; + type["biblio_section"] = State.off; + type["blurb_section"] = State.on; + if (opt_action["backmatter"] && opt_action["section_blurb"]) { + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + type["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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Blurb"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "_part_blurb"; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.dom_markedup = [ 1, 1, 0, 0, 0, 0, 0, 0]; + comp_obj_heading_.dom_collapsed = [ 1, 1, 0, 0, 0, 0, 0, 0]; + the_blurb_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Blurb"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "blurb"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 2; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.anchor_tags = ["blurb"]; + comp_obj_heading_.dom_markedup = [ 1, 1, 0, 0, 1, 0, 0, 0]; + comp_obj_heading_.dom_collapsed = [ 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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = line.to!string; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "blurb"; + comp_obj_heading_.marked_up_level = an_object["lev"].to!string; + comp_obj_heading_.heading_lev_markup = an_object["lev_markup_number"].to!int; // make int, remove need to conv + comp_obj_heading_.heading_lev_collapsed = an_object["lev_collapsed_number"].to!int; // make int, remove need to conv + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + the_blurb_section ~= comp_obj_heading_; + } else { + _para_match_(line, an_object, an_object_key, indent, bullet, type, line_occur); + comp_obj_para = comp_obj_para.init; + comp_obj_para.use = "backmatter"; + comp_obj_para.is_of = "para"; + comp_obj_para.is_a = "blurb"; + comp_obj_para.text = line.to!string.strip; + comp_obj_para.ocn = 0; + comp_obj_para.obj_cite_number = ""; + comp_obj_para.indent_hang = indent["hang_position"]; + comp_obj_para.indent_base = indent["base_position"]; + comp_obj_para.bullet = bullet; + the_blurb_section ~= comp_obj_para; + } + type["ocn_status"] = TriState.off; + } + continue; + } else if (type["quote"] == TriState.on) { + /+ within block object: quote +/ + _quote_block_(line, an_object, type); + continue; + /+ within block object: group +/ + } else if (type["group"] == TriState.on) { + /+ within block object: group +/ + line = (line) + .replaceAll(rgx.para_delimiter, mkup.br_paragraph ~ "$1"); + _group_block_(line, an_object, type); + continue; + } else if (type["block"] == TriState.on) { + /+ within block object: block +/ + 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, type); + continue; + } else if (type["poem"] == TriState.on) { + /+ within block object: poem +/ + _poem_block_(line, an_object, type, cntr, obj_cite_number_poem, dochead_make_aa); + continue; + } else if (type["table"] == TriState.on) { + /+ within block object: table +/ + _table_block_(line, an_object, type, dochead_make_aa); + continue; + } else { + /+ not within a block group +/ + assert( + (type["blocks"] == TriState.off) + || (type["blocks"] == TriState.closing), + "block status: none or closed" + ); + assertions_flag_types_block_status_none_or_closed(type); + if (line.matchFirst(rgx.block_open)) { + if (line.matchFirst(rgx.block_poem_open)) { + /+ poem to verse exceptions! +/ + object_reset(an_object); + processing.remove("verse"); + obj_cite_number_poem["start"] = obj_cite_number.to!string; + } + _start_block_(line, type, obj_cite_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( + (type["blocks"] == TriState.off) + || (type["blocks"] == TriState.closing), + "code block status: none or closed" + ); + if (type["blocks"] == TriState.closing) { + debug(check) { // block + writeln(__LINE__); + writeln(line); + } + assert( + line.matchFirst(rgx.book_index) + || line.matchFirst(rgx.book_index_open) + || type["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) + || type["book_index"] == State.on ) { + /+ book_index +/ + _book_index_(line, book_idx_tmp, an_object, type, 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.use = "comment"; + comp_obj_comment.is_of = "comment"; + comp_obj_comment.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, type); + processing.remove("verse"); + ++cntr; + } else if (((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((type["para"] == State.off) + && (type["heading"] == State.off))) { + /+ heading or para but neither flag nor line exists +/ + if ((dochead_make_aa["make"]["headings"].length > 2) + && (type["make_headings"] == State.off)) { + /+ heading found +/ + _heading_found_(line, dochead_make_aa["make"]["headings"], heading_match_str, heading_match_rgx, type); + } + if ((type["make_headings"] == State.on) + && ((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((type["para"] == State.off) + && (type["heading"] == State.off))) { + /+ heading make set +/ + line = _heading_make_set_(line, line_occur, heading_match_rgx, type); + } + /+ 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 +/ + _heading_matched_(line, line_occur, an_object, an_object_key, lv, collapsed_lev, type, dochead_meta_aa); + } else if (line_occur["para"] == State.off) { + /+ para match +/ + an_object_key="body_nugget"; + _para_match_(line, an_object, an_object_key, indent, bullet, type, 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); + } + an_object[an_object_key] ~= " " ~ line; + ++line_occur["para"]; + } + } + } else if (type["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_number, + comp_obj_heading, + cntr, + type, + obj_cite_number_poem, + dochead_make_aa + ); + } else { + /+ line.empty, post contents, empty variables: +/ + assert( + line.empty, + "\nline should be empty:\n \"" + ~ line ~ "\"" + ); + assert( + (type["blocks"] == State.off), + "code block status: none" + ); + if ((type["heading"] == State.on) + && (line_occur["heading"] > State.off)) { + /+ heading object (current line empty) +/ + obj_cite_number = (an_object["lev_markup_number"].to!int == 0) + ? (ocn_emit(3)) + : (obj_cite_number = ocn_emit(type["ocn_status"])); + 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, dochead_make_aa); + an_object["substantive"] = substantive_object_and_anchor_tags_tuple[sObj.content]; + anchor_tags = substantive_object_and_anchor_tags_tuple[sObj.anchor_tags]; + if (an_object["lev_markup_number"].to!int == 4) { + 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_number.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_number, segment_anchor_tag_that_object_belongs_to); + /+ (incrementally build toc) table of contents here! +/ + _anchor_tag=to!string(obj_cite_number); + the_table_of_contents_section = obj_im.table_of_contents_gather_headings( + an_object, + dochead_make_aa, + 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"], // string + an_object["lev_markup_number"], // string + an_object["lev_collapsed_number"], // string + segment_anchor_tag_that_object_belongs_to, // string + obj_cite_number, // int + 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, type); + an_object.remove("lev"); + an_object.remove("lev_markup_number"); + processing.remove("verse"); + ++cntr; + } else if ((type["para"] == State.on) + && (line_occur["para"] > State.off)) { + /+ paragraph object (current line empty) +/ + obj_cite_number = ocn_emit(type["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_number, 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_para.is_of = "para"; + comp_obj_para.is_a = "para"; + comp_obj_para.text = an_object["substantive"].to!string.strip; + comp_obj_para.ocn = obj_cite_number; + comp_obj_para.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_para.indent_hang = indent["hang_position"]; + comp_obj_para.indent_base = indent["base_position"]; + comp_obj_para.bullet = bullet; + comp_obj_para.anchor_tags = anchor_tags; + comp_obj_para.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_para.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_para.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_para; + _common_reset_(line_occur, an_object, type); + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + processing.remove("verse"); + ++cntr; + } else { + assert( + line == null, + "line variable should be empty, should not occur" + ); + } + } // 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].is_a == "para") + || (the_document_body_section[$-1].is_a == "heading") + || (the_document_body_section[$-1].is_a == "quote") + || (the_document_body_section[$-1].is_a == "group") + || (the_document_body_section[$-1].is_a == "block") + || (the_document_body_section[$-1].is_a == "verse")) + && (the_document_body_section.length > previous_length)) { + if ((the_document_body_section[$-1].is_a == "heading") + && (the_document_body_section[$-1].heading_lev_markup < 5)) { + type["biblio_section"] = State.off; + type["glossary_section"] = State.off; + type["blurb_section"] = State.off; + } + if (the_document_body_section[$-1].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].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, + to!int(i), + ); + } + } + } + } 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, + to!int(the_document_body_section.length-1), + ); + } + } + previous_length = the_document_body_section.length.to!int; + } + } + } /+ ← closed: loop markup document/text line by line +/ + /+ ↓ post loop markup document/text +/ + auto en_tuple = + note_section.endnote_objects(obj_cite_number, opt_action); + static assert(!isTypeTuple!(en_tuple)); + auto the_endnotes_section = en_tuple[0]; + obj_cite_number = 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_.use = "empty"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Glossary section"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + the_glossary_section ~= comp_obj_heading_; + } else { + writeln("gloss"); + } + 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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Bibliography"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "_part_bibliography"; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + the_bibliography_section ~= comp_obj_heading_; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Bibliography"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "bibliography"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 2; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.anchor_tags = ["bibliography"]; + the_bibliography_section ~= comp_obj_heading_; + } else { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "empty"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Bibliography"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.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.use = "backmatter"; + comp_obj_para.is_of = "para"; + comp_obj_para.is_a = "bibliography"; + comp_obj_para.text = out_.to!string.strip; + comp_obj_para.ocn = 0; + comp_obj_para.obj_cite_number = ""; + comp_obj_para.indent_hang = 0; + comp_obj_para.indent_base = 1; + comp_obj_para.bullet = bullet; + comp_obj_para.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_number, + opt_action, + ); + destroy(bookindex_unordered_hashes); + static assert(!isTypeTuple!(bi_tuple)); + auto the_bookindex_section = bi_tuple[0]; + obj_cite_number = 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_.use = "empty"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "(skip) there is no Blurb section"; + comp_obj_heading_.ocn = 0; + comp_obj_para.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = ""; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.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.use = "frontmatter"; + comp_obj_toc.is_of = "para"; + comp_obj_toc.is_a = "toc"; + comp_obj_toc.ocn = 0; + comp_obj_toc.obj_cite_number = ""; + comp_obj_toc.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.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.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.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + toc_txt_ = format( + "{ %s }#%s", + "Glossary", + "glossary", // _anchor_tag + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + + toc_txt_ = format( + "{ %s }#%s", + "Bibliography", + "bibliography", // _anchor_tag + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.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", // _anchor_tag + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + toc_txt_ = format( + "{ %s }#%s", + "Blurb", + "blurb", // _anchor_tag + ); + toc_txt_= munge.url_links(toc_txt_); + comp_obj_toc.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.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.indent_hang), toc_linked_heading.text); + } + } + the_document_head_section ~= the_document_body_section[0]; + the_document_body_section=the_document_body_section[1..$]; + if (the_endnotes_section.length > 1) { + html_segnames ~= "endnotes"; + html_segnames_ptr = html_segnames_ptr_cntr; + foreach (ref section; the_endnotes_section) { + if (section.heading_lev_markup == 4) { + section.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 section; the_glossary_section) { + if (section.heading_lev_markup == 4) { + section.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 section; the_bibliography_section) { + if (section.heading_lev_markup == 4) { + section.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 section; the_bookindex_section["scroll"]) { + if (section.heading_lev_markup == 4) { + section.ptr_html_segnames = html_segnames_ptr; + break; + } + } + foreach (ref section; the_bookindex_section["seg"]) { + if (section.heading_lev_markup == 4) { + section.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 section; the_blurb_section) { + if (section.heading_lev_markup == 4) { + section.ptr_html_segnames = html_segnames_ptr; + break; + } + } + html_segnames_ptr_cntr++; + } + 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; + foreach (ref obj; the_document_head_section) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + } + /+ dom structure (marked up & collapsed) +/ + if ((opt_action["html"]) + || (opt_action["html-scroll"]) + || (opt_action["html-seg"]) + || (opt_action["epub"]) + || (opt_action["sqlite"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + if (the_table_of_contents_section["scroll"].length > 1) { + /+ scroll +/ + dom_markedup_buffer = dom_markedup.dup; + dom_collapsed_buffer = dom_collapsed.dup; + foreach (ref obj; the_table_of_contents_section["scroll"]) { + if (obj.is_a == "heading") { + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + /+ seg +/ + dom_markedup = dom_markedup_buffer.dup; + dom_collapsed = dom_collapsed_buffer.dup; + foreach (ref obj; the_table_of_contents_section["seg"]) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + assert(obj.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_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.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.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.lev4_subtoc = lev4_subtoc[obj.segment_anchor_tag]; + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } else if (obj.is_a == "para") { + _images ~= extract_images(obj.text); + } + } + } + auto images=uniq(_images.sort()); + /+ optional only one 1~ level +/ + if (the_endnotes_section.length > 1) { + dom_markedup_buffer = dom_markedup.dup; + dom_collapsed_buffer = dom_collapsed.dup; + dom_markedup = dom_markedup_buffer.dup; + dom_collapsed = dom_collapsed_buffer.dup; + foreach (ref obj; the_endnotes_section) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.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.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ optional only one 1~ level +/ + if (the_bibliography_section.length > 1) { + foreach (ref obj; the_bibliography_section) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ optional only one 1~ level +/ + if (the_bookindex_section["scroll"].length > 1) { + /+ scroll +/ + dom_markedup_buffer = dom_markedup.dup; + dom_collapsed_buffer = dom_collapsed.dup; + foreach (ref obj; the_bookindex_section["scroll"]) { + if (obj.is_a == "heading") { + debug(dom) { + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + /+ seg +/ + dom_markedup = dom_markedup_buffer.dup; + dom_collapsed = dom_collapsed_buffer.dup; + foreach (ref obj; the_bookindex_section["seg"]) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ optional only one 1~ level +/ + if (the_blurb_section.length > 1) { + foreach (ref obj; the_blurb_section) { + if (obj.is_a == "heading") { + debug(dom) { + writeln(obj.text); + } + if (obj.heading_lev_markup <= 4) { + segnames_0_4 ~= obj.segment_anchor_tag; + if (obj.heading_lev_markup == 4) { + obj.segname_prev = html_segnames[obj.ptr_html_segnames - 1]; + if (html_segnames.length > obj.ptr_html_segnames + 1) { + obj.segname_next = html_segnames[obj.ptr_html_segnames + 1]; + } + assert(obj.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"]) + || (opt_action["postgresql"])) { + obj = obj_dom_set_markup_tags(obj, dom_markedup, obj.heading_lev_markup); + obj = obj_dom_set_collapsed_tags(obj, dom_collapsed, obj.heading_lev_collapsed); + } + obj = obj_heading_ancestors(obj, lv_ancestors_txt); + } + } + } + /+ TODO + - note create/insert heading object sole purpose eof close all open tags + sort out: + - obj.dom_markedup = dom_markedup; + - obj.dom_collapsed = dom_collapsed; + +/ + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "empty"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.ocn = 0; + comp_obj_para.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = ""; + comp_obj_heading_.marked_up_level = ""; + comp_obj_heading_.heading_lev_markup = 9; + comp_obj_heading_.heading_lev_collapsed = 9; + comp_obj_heading_.parent_ocn = 0; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.dom_markedup = dom_markedup.dup; + comp_obj_heading_.dom_collapsed = dom_collapsed.dup; + obj_dom_set_markup_tags(comp_obj_heading_, dom_markedup, 0); + obj_dom_set_collapsed_tags(comp_obj_heading_, dom_collapsed, 0); + obj_heading_ancestors(comp_obj_heading_, lv_ancestors_txt); + comp_obj_heading_ = obj_dom_set_markup_tags(comp_obj_heading_, dom_markedup, 0); + comp_obj_heading_ = obj_dom_set_collapsed_tags(comp_obj_heading_, dom_collapsed, 0); + comp_obj_heading_ = obj_heading_ancestors(comp_obj_heading_, lv_ancestors_txt); + the_dom_tail_section ~= comp_obj_heading_; + 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 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); + biblio_arr_json = []; + obj_cite_number=0; + obj_cite_number_=0; + html_segnames_ptr=0; + html_segnames_ptr_cntr=0; + content_non_header = "8"; + dom_markedup = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_markedup_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_collapsed = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + dom_collapsed_buffer = [ 0, 0, 0, 0, 0, 0, 0, 0, 0,]; + auto t = tuple( + document_the, + docSectKeysSeq!()(document_section_keys_sequenced), + segnames, + segnames_0_4, + images, + ); + return t; + /+ post loop markup document/text ↑ +/ + } /+ ← closed: abstract doc source +/ + /+ ↓ abstraction functions +/ + auto object_reset(O)(ref 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"); + } + auto _common_reset_(L,O,T)( + return ref L line_occur, + return ref O an_object, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line_occur) == int[string])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + line_occur["heading"] = State.off; + line_occur["para"] = State.off; + type["heading"] = State.off; + type["para"] = State.off; + object_reset(an_object); + } + void _check_ocn_status_(L,T)( + L line, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if ((!line.empty) && (type["ocn_status_multi_obj"] == TriState.off)) { + /+ not multi-line object, check whether obj_cite_number is on or turned off +/ + if (line.matchFirst(rgx.obj_cite_number_block_marks)) { + /+ switch off obj_cite_number +/ + if (line.matchFirst(rgx.obj_cite_number_off_block)) { + type["ocn_status_multi_obj"] = TriState.on; + debug(ocnoff) { + writeln(line); + } + } + if (line.matchFirst(rgx.obj_cite_number_off_block_dh)) { + type["ocn_status_multi_obj"] = TriState.closing; + debug(ocnoff) { + writeln(line); + } + } + } else { + if (type["ocn_status_multi_obj"] == TriState.off) { + if (line.matchFirst(rgx.obj_cite_number_off)) { + type["ocn_status"] = TriState.on; + } else if (line.matchFirst(rgx.obj_cite_number_off_dh)) { + type["ocn_status"] = TriState.closing; + } else { + type["ocn_status"] = TriState.off; + } + } else { + type["ocn_status"] = + type["ocn_status_multi_obj"]; + } + } + } else if ((!line.empty) && (type["ocn_status_multi_obj"] > TriState.off)) { + if (line.matchFirst(rgx.obj_cite_number_off_block_close)) { + type["ocn_status_multi_obj"] = TriState.off; + type["ocn_status"] = TriState.off; + debug(ocnoff) { + writeln(line); + } + } + } + } + void _start_block_(L,T,N)( + L line, + return ref T type, + N obj_cite_number_poem + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(type) == int[string])); + static assert(is(typeof(obj_cite_number_poem) == string[string])); + } + 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 + ); + } + type["blocks"] = TriState.on; + type["code"] = TriState.on; + type["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 + ); + } + obj_cite_number_poem["start"] = + obj_cite_number.to!string; + type["blocks"] = TriState.on; + type["verse_new"] = State.on; + type["poem"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["group"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["block"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["quote"] = TriState.on; + type["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"; + type["blocks"] = TriState.on; + type["table"] = TriState.on; + type["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"; + type["blocks"] = TriState.on; + type["table"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["code"] = TriState.on; + type["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 + ); + } + obj_cite_number_poem["start"] = obj_cite_number.to!string; + type["blocks"] = TriState.on; + type["verse_new"] = State.on; + type["poem"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["group"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["block"] = TriState.on; + type["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 + ); + } + type["blocks"] = TriState.on; + type["quote"] = TriState.on; + type["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"; + type["blocks"] = TriState.on; + type["table"] = TriState.on; + type["tic_table"] = TriState.on; + } + } + void _quote_block_(L,O,T)( + return ref L line, + return ref O an_object, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if (type["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; + type["blocks"] = TriState.closing; + type["quote"] = TriState.closing; + type["curly_quote"] = TriState.off; + } else { + debug(quote) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build quote array (or string) + } + } else if (type["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; + type["blocks"] = TriState.closing; + type["quote"] = TriState.closing; + type["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)( + return ref L line, + return ref O an_object, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if (type["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; + type["blocks"] = TriState.closing; + type["group"] = TriState.closing; + type["curly_group"] = TriState.off; + } else { + debug(group) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build group array (or string) + } + } else if (type["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; + type["blocks"] = TriState.closing; + type["group"] = TriState.closing; + type["tic_group"] = TriState.off; + } else { + debug(group) { // group + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build group array (or string) + } + } + } + void _block_block_(L,O,T)( + return ref L line, + return ref O an_object, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if (type["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; + type["blocks"] = TriState.closing; + type["block"] = TriState.closing; + type["curly_block"] = TriState.off; + } else { + debug(block) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build block array (or string) + } + } else if (type["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; + type["blocks"] = TriState.closing; + type["block"] = TriState.closing; + type["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,Ma)( + L line, + return ref O an_object, + return ref T type, + return ref C cntr, + N obj_cite_number_poem, + Ma dochead_make_aa, + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + static assert(is(typeof(cntr) == int)); + static assert(is(typeof(obj_cite_number_poem) == string[string])); + static assert(is(typeof(dochead_make_aa) == string[string][string])); + } + auto rgx = Rgx(); + if (type["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) { // poem (curly) close + writefln( + "* [poem curly] %s", + line + ); + } + if (processing.length > 0) { + an_object[an_object_key] = processing["verse"]; + } + debug(poem) { // poem (curly) close + writeln(__LINE__); + writefln( + "* %s %s", + obj_cite_number, + line + ); + } + if (an_object.length > 0) { + debug(poem) { // poem (curly) close + writeln( + obj_cite_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "verse"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + } + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + obj_cite_number_poem["end"] = + obj_cite_number.to!string; + type["blocks"] = TriState.closing; + type["poem"] = TriState.closing; + type["curly_poem"] = TriState.off; + } else { + processing["verse"] ~= line ~= "\n"; + if (type["verse_new"] == State.on) { + obj_cite_number = + ocn_emit(type["ocn_status"]); + type["verse_new"] = State.off; + } else if (line.matchFirst(rgx.newline_eol_delimiter_only)) { + processing["verse"] = processing["verse"].stripRight; + verse_line = TriState.off; + type["verse_new"] = State.on; + } + if (type["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_number, + 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "verse"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.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 (type["tic_poem"] == TriState.on) { + if (auto m = line.matchFirst(rgx.block_tic_close)) { // tic_poem_close + an_object[an_object_key]="verse"; + debug(poem) { // poem (curly) close + 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_number, 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "verse"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + obj_cite_number_poem["end"] = obj_cite_number.to!string; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } + type["blocks"] = TriState.closing; + type["poem"] = TriState.closing; + type["tic_poem"] = TriState.off; + } else { + processing["verse"] ~= line ~= "\n"; + if (type["verse_new"] == State.on) { + obj_cite_number = + ocn_emit(type["ocn_status"]); + type["verse_new"] = State.off; + } else if (line.matchFirst(rgx.newline_eol_delimiter_only)) { + processing["verse"] = processing["verse"].stripRight; + type["verse_new"] = State.on; + verse_line = TriState.off; + } + if (type["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_number, + 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "verse"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.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)( + return ref L line, + return ref O an_object, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if (type["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; + type["blocks"] = TriState.closing; + type["code"] = TriState.closing; + type["curly_code"] = TriState.off; + } else { + debug(code) { // code (curly) line + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // code (curly) line + } + } else if (type["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; + type["blocks"] = TriState.closing; + type["code"] = TriState.closing; + type["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,Ma)( + return ref L line, + return ref O an_object, + return ref T type, + return ref Ma dochead_make_aa + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(an_object) == string[string])); + static assert(is(typeof(type) == int[string])); + } + auto rgx = Rgx(); + if (type["curly_table"] == TriState.on) { + if (line.matchFirst(rgx.block_curly_table_close)) { + debug(table) { // table (curly) close + writeln(line); + } + type["blocks"] = TriState.closing; + type["table"] = TriState.closing; + type["curly_table"] = TriState.off; + } else { + debug(table) { // table + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; // build table array (or string) + } + } else if (type["curly_table_special_markup"] == TriState.on) { + if (line.empty) { + type["blocks"] = TriState.off; + type["table"] = TriState.off; + type["curly_table_special_markup"] = TriState.off; + _table_closed_make_special_notation_table_( + line, + an_object, + the_document_body_section, + obj_cite_number, + comp_obj_heading, + cntr, + type, + dochead_make_aa + ); + } else { + debug(table) { + writeln(line); + } + an_object[an_object_key] ~= line ~= "\n"; + } + } else if (type["tic_table"] == TriState.on) { + if (line.matchFirst(rgx.block_tic_close)) { + debug(table) { // table (tic) close + writeln(line); + } + type["blocks"] = TriState.closing; + type["table"] = TriState.closing; + type["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] type, + return ref int bib_entry, + return ref string biblio_entry_str_json, + return ref string[] biblio_arr_json + ) { + mixin SiSUbiblio; + auto jsn = BibJsnStr(); + auto rgx = Rgx(); + if (line.matchFirst(rgx.heading_biblio)) { + type["biblio_section"] = TriState.on; + type["blurb_section"] = State.off; + type["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_( + char[] line, + return ref string[string] an_object, + return ref ObjGenericComposite[] the_document_body_section, + return ref int obj_cite_number, + return ref ObjGenericComposite _comp_obj_heading, + return ref int cntr, + return ref int[string] type, + string[string][string] dochead_make_aa, + ) { + comp_obj_block = comp_obj_block.init; + obj_cite_number = + ocn_emit(type["ocn_status"]); + auto comp_obj_location = + node_construct.node_location_emitter( + content_non_header, + segment_anchor_tag_that_object_belongs_to, + obj_cite_number, + 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", dochead_make_aa); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + 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)( + 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 int obj_cite_number, + return ref ObjGenericComposite _comp_obj_heading, + return ref int cntr, + return ref int[string] type, + string[string] obj_cite_number_poem, + string[string][string] dochead_make_aa, + ) { + assert( + line.empty, + "\nline should be empty:\n \"" + ~ line ~ "\"" + ); + assert( + (type["blocks"] == TriState.closing), + "code block status: closed" + ); + assertions_flag_types_block_status_none_or_closed(type); + if (type["quote"] == TriState.closing) { + obj_cite_number = + ocn_emit(type["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_number, + 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "quote"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + type["blocks"] = TriState.off; + type["quote"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (type["group"] == TriState.closing) { + obj_cite_number = + ocn_emit(type["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_number, + 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "group"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + type["blocks"] = TriState.off; + type["group"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (type["block"] == TriState.closing) { + obj_cite_number = ocn_emit(type["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_number, + 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_number, + 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, dochead_make_aa); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.use = "body"; + comp_obj_block.is_of = "block"; + comp_obj_block.is_a = "block"; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_block.text = an_object["substantive"]; + comp_obj_block.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_block.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_block.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_block; + type["blocks"] = TriState.off; + type["block"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (type["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_number, + 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_number, + cntr, + heading_ptr-1, + an_object["is"] + ); + comp_obj_poem_ocn = comp_obj_poem_ocn.init; + comp_obj_poem_ocn.use = "body"; + comp_obj_poem_ocn.is_of = "block"; + comp_obj_poem_ocn.is_a = "poem"; + comp_obj_poem_ocn.ocn = obj_cite_number; + comp_obj_poem_ocn.obj_cite_number = (obj_cite_number_poem["start"], obj_cite_number_poem["end"]); + comp_obj_poem_ocn.text = ""; + the_document_body_section ~= comp_obj_poem_ocn; + type["blocks"] = TriState.off; + type["poem"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + } else if (type["code"] == TriState.closing) { + obj_cite_number = + ocn_emit(type["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_number, + 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_number, + 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, dochead_make_aa); + 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.use = "body"; + comp_obj_code.is_of = "block"; + comp_obj_code.is_a = "code"; + comp_obj_code.ocn = obj_cite_number; + comp_obj_code.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + comp_obj_code.text = an_object["substantive"]; + comp_obj_code.inline_notes_reg = substantive_obj_misc_tuple[sObj.notes_reg]; + comp_obj_code.inline_notes_star = substantive_obj_misc_tuple[sObj.notes_star]; + comp_obj_code.inline_links = substantive_obj_misc_tuple[sObj.links]; + the_document_body_section ~= comp_obj_code; + type["blocks"] = TriState.off; + type["code"] = TriState.off; + object_reset(an_object); + processing.remove("verse"); + ++cntr; + } else if (type["table"] == TriState.closing) { + comp_obj_block = comp_obj_block.init; + obj_cite_number = + ocn_emit(type["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_number, + 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_number, + 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, dochead_make_aa); + an_object["substantive"] = substantive_obj_misc_tuple[sObj.content]; + comp_obj_block = comp_obj_block.init; + comp_obj_block.ocn = obj_cite_number; + comp_obj_block.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + 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; + type["blocks"] = TriState.off; + type["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 type, + 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(type) == int[string])); + static assert(is(typeof(opt_action) == bool[string])); + } + auto rgx = Rgx(); + if (auto m = line.match(rgx.book_index)) { + /+ match book_index +/ + debug(bookindexmatch) { // book index + 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 +/ + type["book_index"] = State.on; + if (opt_action["backmatter"] && opt_action["section_bookindex"]) { + book_idx_tmp = m.captures[1].to!string; + debug(bookindexmatch) { // book index + writefln( + "* [bookindex] %s\n", + book_idx_tmp, + ); + } + } + } else if (type["book_index"] == State.on ) { + /+ book_index flag set +/ + if (auto m = line.match(rgx.book_index_close)) { + type["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) { // book index + 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 dochead_make_identify_unmarked_headings, + return ref H heading_match_str, + return ref R heading_match_rgx, + return ref T type + ) { + debug(asserts) { + static assert(is(typeof(line) == char[])); + static assert(is(typeof(dochead_make_identify_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(type) == int[string])); + } + auto rgx = Rgx(); + if ((dochead_make_identify_unmarked_headings.length > 2) + && (type["make_headings"] == State.off)) { + /+ headings found +/ + debug(headingsfound) { + writeln(dochead_make_identify_unmarked_headings); + } + char[][] make_headings_spl = + (cast(char[]) dochead_make_identify_unmarked_headings) + .split(rgx.make_heading_delimiter); + debug(headingsfound) { + writeln(make_headings_spl.length); + writeln(make_headings_spl); + } + switch (make_headings_spl.length) { + case 7 : + if (!empty(make_headings_spl[6])) { + heading_match_str["h_4"] = + "^(" ~ make_headings_spl[6].to!string ~ ")"; + heading_match_rgx["h_4"] = + regex(heading_match_str["h_4"]); + } + goto case; + case 6 : + if (!empty(make_headings_spl[5])) { + heading_match_str["h_3"] = + "^(" ~ make_headings_spl[5].to!string ~ ")"; + heading_match_rgx["h_3"] = + regex(heading_match_str["h_3"]); + } + goto case; + case 5 : + if (!empty(make_headings_spl[4])) { + heading_match_str["h_2"] = + "^(" ~ make_headings_spl[4].to!string ~ ")"; + heading_match_rgx["h_2"] = + regex(heading_match_str["h_2"]); + } + goto case; + case 4 : + if (!empty(make_headings_spl[3])) { + heading_match_str["h_1"] = + "^(" ~ make_headings_spl[3].to!string ~ ")"; + heading_match_rgx["h_1"] = + regex(heading_match_str["h_1"]); + } + goto case; + case 3 : + if (!empty(make_headings_spl[2])) { + heading_match_str["h_D"] = + "^(" ~ make_headings_spl[2].to!string ~ ")"; + heading_match_rgx["h_D"] = + regex(heading_match_str["h_D"]); + } + goto case; + case 2 : + if (!empty(make_headings_spl[1])) { + heading_match_str["h_C"] = + "^(" ~ make_headings_spl[1].to!string ~ ")"; + heading_match_rgx["h_C"] = + regex(heading_match_str["h_C"]); + } + goto case; + case 1 : + if (!empty(make_headings_spl[0])) { + heading_match_str["h_B"] = + "^(" ~ make_headings_spl[0].to!string ~ ")"; + heading_match_rgx["h_B"] = + regex(heading_match_str["h_B"]); + } + break; + default: + break; + } + type["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 type + ) { + 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(type) == int[string])); + } + if ((type["make_headings"] == State.on) + && ((line_occur["para"] == State.off) + && (line_occur["heading"] == State.off)) + && ((type["para"] == State.off) + && (type["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,Me)( + return ref 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 type, + return ref Me dochead_meta_aa, + ) { + 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(type) == int[string])); + static assert(is(typeof(dochead_meta_aa) == string[string][string])); + } + auto rgx = Rgx(); + if (auto m = line.match(rgx.heading)) { + /+ heading match +/ + type["heading"] = State.on; + if (line.match(rgx.heading_seg_and_above)) { + type["biblio_section"] = State.off; + type["glossary_section"] = State.off; + type["blurb_section"] = State.off; + } + type["para"] = State.off; + ++line_occur["heading"]; + 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": + an_object[an_object_key]=(an_object[an_object_key]) + .replaceFirst(rgx.variable_doc_title, + (dochead_meta_aa["title"]["full"] ~ ",")) + .replaceFirst(rgx.variable_doc_author, + dochead_meta_aa["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) { // heading + writeln(line.strip); + } + } + } + void _para_match_(L,O,K,I,B,T,C)( + return ref L line, + return ref O an_object, + return ref K an_object_key, + return ref I indent, + return ref B bullet, + return ref T type, + 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(type) == int[string])); + static assert(is(typeof(line_occur) == int[string])); + } + auto rgx = Rgx(); + if (line_occur["para"] == State.off) { + line = font_faces_line(line); + /+ para matches +/ + type["para"] = State.on; + an_object[an_object_key] ~= line; // body_nugget + indent=[ + "hang_position" : 0, + "base_position" : 0, + ]; + bullet = false; + if (auto m = line.matchFirst(rgx.para_indent)) { + debug(paraindent) { // para indent + writeln(line); + } + indent["hang_position"] = (m.captures[1]).to!int; + indent["base_position"] = 0; + } else if (line.matchFirst(rgx.para_bullet)) { + debug(parabullet) { // para bullet + writeln(line); + } + bullet = true; + } else if (auto m = line.matchFirst(rgx.para_indent_hang)) { + debug(paraindenthang) { // para indent hang + 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) { // para bullet indent + writeln(line); + } + indent=[ + "hang_position" : (m.captures[1]).to!int, + "base_position" : 0, + ]; + bullet = true; + } + ++line_occur["para"]; + } + } + auto font_faces_line(T)( + return ref T textline, + ) { + 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, + ) { + auto rgx = Rgx(); + table_object.use = "body"; + table_object.is_of = "block"; + table_object.is_a = "table"; + table_object.inline_notes_reg = false; + table_object.inline_notes_star = false; + table_object.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, + ) { + auto rgx = Rgx(); + 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 (idx_r == 1 && col.match(rgx.numeric_col)) { + 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, + ) { + auto rgx = Rgx(); + 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, + ) { + auto rgx = Rgx(); + 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, + ) { + auto rgx = Rgx(); + 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 +/ + struct OCNemitter { + int obj_cite_number, obj_cite_number_; + int ocn_emitter(int ocn_status_flag) + in { assert(ocn_status_flag <= 3); } + body { + if (ocn_status_flag == 3) { + obj_cite_number = obj_cite_number_ = 1; + } else { + obj_cite_number=(ocn_status_flag == 0) + ? ++obj_cite_number_ + : 0; + } + assert(obj_cite_number >= 0); + return obj_cite_number; + } + invariant() { + } + } + /+ +/ + struct ObjInlineMarkupMunge { + string[string] obj_txt; + int n_foot, n_foot_reg, n_foot_sp_asterisk, n_foot_sp_plus; + string asterisks_; + string obj_txt_out, tail, note; + auto rgx = Rgx(); + 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; + } + string url_links(Ot)(Ot obj_txt_in) { + debug(asserts) { + static assert(is(typeof(obj_txt_in) == string)); + } + auto mng = InlineMarkup(); + obj_txt_in = obj_txt_in.replaceAll(rgx.inline_mono, (mng.mono ~ "{$1}" ~ mng.mono)); + /+ url matched +/ + if (obj_txt_in.match(rgx.inline_url_generic)) { + /+ link: naked url: http://url +/ + if (obj_txt_in.match(rgx.inline_link_naked_url)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.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.inline_link_endnote_url_helper)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.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") + ); + obj_txt_in = (obj_txt_in).replaceAll( + rgx.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.inline_link_markup_regular)) { + obj_txt_in = (obj_txt_in).replaceAll( + rgx.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; + } + 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; + 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_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, + ); + 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; + 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) + ); + /+ url matched +/ + if (obj_txt_in.match(rgx.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], + urls, + ); + 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.obj_cite_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.obj_cite_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() { + } + } + struct ObjInlineMarkup { + auto rgx = Rgx(); + auto munge = ObjInlineMarkupMunge(); + string[string] obj_txt; + auto obj_inline_markup_and_anchor_tags_and_misc(O,K,Ma)( + O obj_, + K obj_key_, + Ma dochead_make_aa + ) + in { + debug(asserts) { + static assert(is(typeof(obj_) == string[string])); + static assert(is(typeof(obj_key_) == string)); + static assert(is(typeof(dochead_make_aa) == string[string][string])); + } + } + body { + obj_txt["munge"] = obj_[obj_key_].dup; + obj_txt["munge"] = (obj_["is"].match(ctRegex!(`verse|code`))) + ? obj_txt["munge"] + : strip(obj_txt["munge"]); + static __gshared string[] 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; + switch (obj_["is"]) { + case "heading": + static __gshared string anchor_tag = ""; + obj_txt["munge"]=_configured_auto_heading_numbering_and_segment_anchor_tags(obj_txt["munge"], obj_, dochead_make_aa); + obj_txt["munge"]=_make_segment_anchor_tags_if_none_provided(obj_txt["munge"], obj_["lev"]); + 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["links"] = x[3]; + 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"], + ); + 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,Ma,Ts,Ta,X,Toc)( + O obj_, + Ma dochead_make_aa, + 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(dochead_make_aa) == string[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; + 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.use = "frontmatter"; + comp_obj_toc.is_of = "para"; + comp_obj_toc.is_a = "toc"; + comp_obj_toc.ocn = 0; + comp_obj_toc.obj_cite_number = ""; + comp_obj_toc.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.bullet = false; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.use = "frontmatter"; + comp_obj_toc.is_of = "para"; + comp_obj_toc.is_a = "toc"; + comp_obj_toc.ocn = 0; + comp_obj_toc.obj_cite_number = ""; + comp_obj_toc.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.bullet = false; + comp_obj_toc.text = "Table of Contents"; + comp_obj_toc.inline_links = true; + the_table_of_contents_section["scroll"] ~= comp_obj_toc; + } + comp_obj_toc = comp_obj_toc.init; + comp_obj_toc.use = "frontmatter"; + comp_obj_toc.is_of = "para"; + comp_obj_toc.is_a = "toc"; + comp_obj_toc.ocn = 0; + comp_obj_toc.obj_cite_number = ""; + comp_obj_toc.bullet = false; + comp_obj_toc.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.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.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.indent_hang = indent["hang_position"]; + comp_obj_toc.indent_base = indent["base_position"]; + comp_obj_toc.text = toc_txt_.to!string.strip; + comp_obj_toc.inline_links = true; + the_table_of_contents_section["seg"] ~= comp_obj_toc; + break; + default: + break; + } + return the_table_of_contents_section; + } + invariant() { + } + private: + static string _configured_auto_heading_numbering_and_segment_anchor_tags(M,O,Ma)( + M munge_, + O obj_, + Ma dochead_make_aa + ) { + debug(asserts) { + static assert(is(typeof(munge_) == string)); + static assert(is(typeof(obj_) == string[string])); + static assert(is(typeof(dochead_make_aa) == string[string][string])); + } + if (dochead_make_aa["make"]["num_top"].length > 0) { + static __gshared int heading_num_top_level=9; + static __gshared int heading_num_depth=2; + static __gshared int heading_num_0 = 0; + static __gshared int heading_num_1 = 0; + static __gshared int heading_num_2 = 0; + static __gshared int heading_num_3 = 0; + static __gshared string heading_number_auto_composite = ""; + if (heading_num_top_level==9) { + if (dochead_make_aa["make"]["num_depth"].length > 0) { + heading_num_depth = dochead_make_aa["make"]["num_depth"].to!uint; + } + switch (dochead_make_aa["make"]["num_top"]) { + case "A": + break; + case "B": + heading_num_top_level=1; + break; + case "C": + heading_num_top_level=2; + break; + case "D": + heading_num_top_level=3; + break; + case "1": + heading_num_top_level=4; + break; + case "2": + heading_num_top_level=5; + break; + case "3": + heading_num_top_level=6; + break; + case "4": + heading_num_top_level=7; + break; + default: + break; + } + } + /+ num_depth minimum 0 + (1.) default 2 (1.1.1) max 3 (1.1.1.1) implement +/ + if ( + heading_num_top_level + > obj_["lev_markup_number"].to!uint + ) { + heading_num_1 = 0; + heading_num_2 = 0; + heading_num_3 = 0; + } else if ( + heading_num_top_level + == obj_["lev_markup_number"].to!uint + ) { + heading_num_0 ++; + heading_num_1 = 0; + heading_num_2 = 0; + heading_num_3 = 0; + } else if ( + heading_num_top_level + == (obj_["lev_markup_number"].to!uint - 1) + ) { + heading_num_1 ++; + heading_num_2 = 0; + heading_num_3 = 0; + } else if ( + heading_num_top_level + == (obj_["lev_markup_number"].to!uint - 2) + ) { + heading_num_2 ++; + heading_num_3 = 0; + } else if ( + heading_num_top_level + == (obj_["lev_markup_number"].to!uint - 3) + ) { + heading_num_3 ++; + } + if (heading_num_3 > 0) { + heading_number_auto_composite = + (heading_num_depth == 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 = + ((heading_num_depth >= 2) + && (heading_num_depth <= 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 = + ((heading_num_depth >= 1) + && (heading_num_depth <= 3)) + ? ( heading_num_0.to!string ~ "." + ~ heading_num_1.to!string + ) + : ""; + } else if (heading_num_0 > 0) { + heading_number_auto_composite = + ((heading_num_depth >= 0) + && (heading_num_depth <= 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 >= heading_num_top_level)) { + 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 string _make_segment_anchor_tags_if_none_provided(M,Lv)(M munge_, Lv lev_) { + 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] ~ " "); + } 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" + static __gshared int 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(), + "; obj_cite_number: ", oa_j["obj_cite_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["obj_cite_number"] = _comp_obj_heading.ocn; + oa_j.object["lev_markup_number"] = _comp_obj_heading.heading_lev_markup; + oa_j.object["lev_collapsed_number"] = _comp_obj_heading.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_obj_cite_number"] = _comp_obj_heading.parent_ocn; + oa_j.object["parent_lev_markup_number"] = _comp_obj_heading.parent_lev_markup; + _obj_attrib = oa_j.toString(); + return _obj_attrib; + } + } + /+ +/ + struct BookIndexNuggetHash { + string main_term, sub_term, sub_term_bits; + int obj_cite_number_offset, obj_cite_number_endpoint; + string[] obj_cite_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_number, + S segment_anchor_tag, + ) + in { + debug(asserts) { + static assert(is(typeof(bookindex_section) == string)); + static assert(is(typeof(obj_cite_number) == int)); + } + debug(bookindexraw) { + if (!bookindex_section.empty) { + writeln( + "* [bookindex] ", + "[", obj_cite_number.to!string, ": ", segment_anchor_tag, "] ", bookindex_section + ); + } + } + } + body { + 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_obj_cite_numbers_match) + ) { + main_term = m.captures[1].strip; + obj_cite_number_offset = m.captures[2].to!int; + obj_cite_number_endpoint=(obj_cite_number + obj_cite_number_offset); + obj_cite_numbers ~= (obj_cite_number.to!string ~ "-" ~ to!string(obj_cite_number_endpoint) + ~ ":" ~ segment_anchor_tag); + } else { + main_term = bi_main_term_and_rest[0].strip; + obj_cite_numbers ~= obj_cite_number.to!string + ~ ":" ~ segment_anchor_tag; + } + bi[main_term]["_a"] ~= obj_cite_numbers; + obj_cite_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_obj_cite_number_offset_split + ); + foreach (sub_terms_bits; bi_sub_terms_split_arr) { + if (auto m = sub_terms_bits.match(rgx.bi_term_and_obj_cite_numbers_match)) { + sub_term = m.captures[1].strip; + obj_cite_number_offset = m.captures[2].to!int; + obj_cite_number_endpoint=(obj_cite_number + obj_cite_number_offset); + obj_cite_numbers ~= (obj_cite_number.to!string ~ " - " ~ to!string(obj_cite_number_endpoint) + ~ ":" ~ segment_anchor_tag); + } else { + sub_term = sub_terms_bits.strip; + obj_cite_numbers ~= to!string(obj_cite_number) + ~ ":" ~ segment_anchor_tag; + } + if (!empty(sub_term)) { + bi[main_term][sub_term] ~= obj_cite_numbers; + } + obj_cite_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; + auto rgx = Rgx(); + 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_number, + B opt_action, + ) { + debug(asserts) { + static assert(is(typeof(bookindex_unordered_hashes) == string[][string][string])); + static assert(is(typeof(obj_cite_number) == int)); + static assert(is(typeof(opt_action) == bool[string])); + } + mixin SiSUnode; + mixin InternalMarkup; + 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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Book Index"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "_part_book_index"; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading.inline_links = true; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + ++obj_cite_number; + ++mkn; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Index"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "bookindex"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 2; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading.inline_links = false; + comp_obj_heading_.anchor_tags = ["bookindex"]; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + ++obj_cite_number; + ++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.use = "backmatter"; + comp_obj_para.is_of = "para"; + comp_obj_para.is_a = "bookindex"; + comp_obj_para.text = bi_tmp_scroll.to!string.strip; + comp_obj_para.ocn = obj_cite_number; + comp_obj_para.obj_cite_number = (obj_cite_number==0) ? "" : to!string(obj_cite_number); + comp_obj_para.anchor_tags = bi_tmp_tags; + comp_obj_para.indent_hang = 0; + comp_obj_para.indent_base = 1; + comp_obj_para.bullet = false; + comp_obj_para.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; + ++obj_cite_number; + ++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_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + bookindex_section["scroll"] ~= comp_obj_heading_; + bookindex_section["seg"] ~= comp_obj_heading_; + } + auto t = tuple(bookindex_section, obj_cite_number); + return t; + } + } + /+ +/ + struct NotesSection { + string[string] object_notes; + int previous_count; + int mkn; + 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].is_a == "para") + || (contents_am[cntr].is_a == "heading") + || (contents_am[cntr].is_a == "quote") + || (contents_am[cntr].is_a == "group") + || (contents_am[cntr].is_a == "block") + || (contents_am[cntr].is_a == "verse")); + assert(cntr >= previous_count); + previous_count=cntr; + assert( + (contents_am[cntr].text).match( + rgx.inline_notes_delimiter_al_regular_number_note) + ); + } + body { + mixin InternalMarkup; + auto mkup = InlineMarkup(); + 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) + } + // TODO NEXT 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( + int obj_cite_number, + bool[string] 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_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Endnotes"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "_part_endnotes"; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + the_endnotes_section ~= comp_obj_heading_; + ++obj_cite_number; + ++mkn; + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "backmatter"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "Endnotes"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.segment_anchor_tag = "endnotes"; + comp_obj_heading_.marked_up_level = "1"; + comp_obj_heading_.heading_lev_markup = 4; + comp_obj_heading_.heading_lev_collapsed = 2; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.parent_lev_markup = 0; + comp_obj_heading_.anchor_tags = ["endnotes"]; + the_endnotes_section ~= comp_obj_heading_; + ++obj_cite_number; + ++mkn; + } else { + comp_obj_heading_ = comp_obj_heading_.init; + comp_obj_heading_.use = "empty"; + comp_obj_heading_.is_of = "para"; + comp_obj_heading_.is_a = "heading"; + comp_obj_heading_.text = "(skip) there are no Endnotes"; + comp_obj_heading_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_heading_.marked_up_level = "B"; + comp_obj_heading_.heading_lev_markup = 1; + comp_obj_heading_.heading_lev_collapsed = 1; + comp_obj_heading_.parent_ocn = 1; + comp_obj_heading_.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_.use = "backmatter"; + comp_obj_endnote_.is_of = "para"; + comp_obj_endnote_.is_a = "endnote"; + comp_obj_endnote_.ocn = 0; + comp_obj_heading_.obj_cite_number = ""; + comp_obj_endnote_.indent_hang = 0; + comp_obj_endnote_.indent_base = 0; + comp_obj_endnote_.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_.anchor_tags = [ endnotes_["anchor"][i] ]; + comp_obj_endnote_.inline_links = true; + comp_obj_endnote_.text = endnote.strip; + the_endnotes_section ~= comp_obj_endnote_; + } + } + auto t = tuple(the_endnotes_section, obj_cite_number); + 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_number; + int[string] p_; // p_ parent_ + auto rgx = Rgx(); + ObjGenericComposite node_location_emitter(Lv,Ta,N,C,P,I)( + Lv lev_markup_number, + Ta segment_anchor_tag, + N obj_cite_number_, + 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_number_) == 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_number_.to!int >= 0); + } + body { + assert(is_ != "heading"); // should not be necessary + assert(obj_cite_number_.to!int >= 0); // should not be necessary + int obj_cite_number = obj_cite_number_.to!int; + if (lv7 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_4; + p_["obj_cite_number"] = lv7; + } else if (lv6 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_3; + p_["obj_cite_number"] = lv6; + } else if (lv5 > State.off) { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_2; + p_["obj_cite_number"] = lv5; + } else { + p_["lev_markup_number"] = DocStructMarkupHeading.h_text_1; + p_["obj_cite_number"] = lv4; + } + ObjGenericComposite comp_obj_location; + comp_obj_location = comp_obj_location.init; + comp_obj_location.is_a = is_; + comp_obj_location.ocn = obj_cite_number_; + comp_obj_location.segment_anchor_tag = segment_anchor_tag.to!string; + comp_obj_location.parent_ocn = p_["obj_cite_number"]; + comp_obj_location.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.parent_lev_markup >= 4); + assert(comp_obj_location.parent_lev_markup <= 7); + assert(comp_obj_location.parent_ocn >= 0); + return comp_obj_location; + } + invariant() { + } + ObjGenericComposite node_emitter_heading(T,L,Lm,Lc,Ta,N,C,P,LA,I,PSn,fNr,fNs,fL)( + T _text, + L lev, + Lm lev_markup_number, + Lc lev_collapsed_number, + Ta segment_anchor_tag, + N obj_cite_number_, + 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_number_) == 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(to!int(obj_cite_number_) >= 0); + assert( + lev_markup_number.match(rgx.levels_numbered), + ("not a valid heading level: " ~ lev_markup_number ~ " at " ~ obj_cite_number_.to!string) + ); + if (lev_markup_number.match(rgx.levels_numbered)) { + if (lev_markup_number.to!int == 0) { + assert(obj_cite_number_.to!int == 1); + } + } + } + body { + int obj_cite_number = obj_cite_number_.to!int; + switch (lev_markup_number.to!int) { + case 0: + lv = DocStructMarkupHeading.h_sect_A; + lv0 = obj_cite_number; + lv1=0; lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] = 0; + p_["obj_cite_number"] = 0; + break; + case 1: + lv = DocStructMarkupHeading.h_sect_B; + lv1 = obj_cite_number; + lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_A; + p_["obj_cite_number"] = lv0; + break; + case 2: + lv = DocStructMarkupHeading.h_sect_C; + lv2 = obj_cite_number; + lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_B; + p_["obj_cite_number"] = lv1; + break; + case 3: + lv = DocStructMarkupHeading.h_sect_D; + lv3=obj_cite_number; + lv4=0; lv5=0; lv6=0; lv7=0; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_C; + p_["obj_cite_number"] = lv2; + break; + case 4: + lv = DocStructMarkupHeading.h_text_1; + lv4 = obj_cite_number; + lv5=0; lv6=0; lv7=0; + if (lv3 > State.off) { + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_D; + p_["obj_cite_number"] = lv3; + } else if (lv2 > State.off) { + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_C; + p_["obj_cite_number"] = lv2; + } else if (lv1 > State.off) { + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_B; + p_["obj_cite_number"] = lv1; + } else { + p_["lev_markup_number"] = + DocStructMarkupHeading.h_sect_A; + p_["obj_cite_number"] = lv0; + } + break; + case 5: + lv = DocStructMarkupHeading.h_text_2; + lv5 = obj_cite_number; + lv6=0; lv7=0; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_text_1; + p_["obj_cite_number"] = lv4; + break; + case 6: + lv = DocStructMarkupHeading.h_text_3; + lv6 = obj_cite_number; + lv7=0; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_text_2; + p_["obj_cite_number"] = lv5; + break; + case 7: + lv = DocStructMarkupHeading.h_text_4; + lv7 = obj_cite_number; + p_["lev_markup_number"] = + DocStructMarkupHeading.h_text_3; + p_["obj_cite_number"] = lv6; + break; + default: + break; + } + ObjGenericComposite _comp_obj_heading_; + _comp_obj_heading_ = _comp_obj_heading_.init; + _comp_obj_heading_.use = "body"; + _comp_obj_heading_.is_of = "para"; + _comp_obj_heading_.is_a = "heading"; + _comp_obj_heading_.text = _text.to!string.strip; + _comp_obj_heading_.ocn = obj_cite_number_; + _comp_obj_heading_.obj_cite_number = (obj_cite_number==0) ? "" : obj_cite_number.to!string; + _comp_obj_heading_.segment_anchor_tag = segment_anchor_tag.to!string; + _comp_obj_heading_.marked_up_level = lev; + _comp_obj_heading_.heading_lev_markup = (!(lev_markup_number.empty) ? lev_markup_number.to!int : 0); + _comp_obj_heading_.heading_lev_collapsed = (!(lev_collapsed_number.empty) ? lev_collapsed_number.to!int : 0); + _comp_obj_heading_.parent_ocn = p_["obj_cite_number"]; + _comp_obj_heading_.parent_lev_markup = p_["lev_markup_number"]; + _comp_obj_heading_.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_.inline_notes_reg = flag_notes_reg; + _comp_obj_heading_.inline_notes_star = flag_notes_star; + _comp_obj_heading_.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_.parent_lev_markup <= 7); + assert(_comp_obj_heading_.parent_ocn >= 0); + if (lev_markup_number.match(rgx.levels_numbered_headings)) { + assert(_comp_obj_heading_.heading_lev_markup <= 7); + assert(_comp_obj_heading_.ocn >= 0); + if (_comp_obj_heading_.parent_lev_markup > 0) { + assert(_comp_obj_heading_.parent_lev_markup < _comp_obj_heading_.heading_lev_markup); + if (_comp_obj_heading_.ocn != 0) { + assert(_comp_obj_heading_.parent_ocn < _comp_obj_heading_.ocn); + } + } + if (_comp_obj_heading_.heading_lev_markup == 0) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_sect_A); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_sect_B) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_sect_A); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_sect_C) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_sect_B); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_sect_D) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_sect_C); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_text_1) { + assert(_comp_obj_heading_.parent_lev_markup <= DocStructMarkupHeading.h_sect_D); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_text_2) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_text_1); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_text_3) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_text_2); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_text_4) { + assert(_comp_obj_heading_.parent_lev_markup == DocStructMarkupHeading.h_text_3); + } else if (_comp_obj_heading_.heading_lev_markup == DocStructMarkupHeading.h_text_5) { + } + } + return _comp_obj_heading_; + } + invariant() { + } + } + /+ abstraction functions emitters ↑ +/ + /+ ↓ abstraction functions assertions +/ + auto 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; + } + } + auto assertions_flag_types_block_status_none_or_closed(T)(T type) { + debug(asserts) { + static assert(is(typeof(type) == int[string])); + } + assert( + (type["code"] == TriState.off) + || (type["code"] == TriState.closing), + "code block status: off or closing"); + assert( + (type["poem"] == TriState.off) + || (type["poem"] == TriState.closing), + "poem status: off or closing"); + assert( + (type["table"] == TriState.off) + || (type["table"] == TriState.closing), + "table status: off or closing"); + assert( + (type["group"] == TriState.off) + || (type["group"] == TriState.closing), + "group block status: off or closing"); + assert( + (type["block"] == TriState.off) + || (type["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/sdp/meta/metadoc_summary.d b/src/sdp/meta/metadoc_summary.d new file mode 100644 index 0000000..e2b85b8 --- /dev/null +++ b/src/sdp/meta/metadoc_summary.d @@ -0,0 +1,81 @@ +module sdp.meta.metadoc_summary; +template SiSUabstractionSummary() { + auto SiSUabstractionSummary(S,T)( + auto return ref const S doc_abstraction, + auto return ref T doc_matters, + ) { + import + sdp.meta.defaults, + sdp.meta.rgx; + import + std.array, + std.exception, + std.stdio, + std.regex, + 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_obj_cite_number" : "NA [debug \"checkdoc\" not run]", + ]; + foreach (k; doc_matters.keys_seq.seg) { + foreach (obj; doc_abstraction[k]) { + if (obj.use != "empty") { + if (!empty(obj.obj_cite_number)) { + check["last_obj_cite_number"] = obj.obj_cite_number; + } + } + } + } + auto min_repeat_number = 66; + auto char_repeat_number = (doc_matters.dochead_meta["title"]["full"].length + + doc_matters.dochead_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(%s: %s)\n%s", + markup.repeat_character_by_number_provided("-", char_repeat_number), + doc_matters.dochead_meta["title"]["full"], + doc_matters.dochead_meta["creator"]["author"], + doc_matters.source_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 obj_cite_number:", + to!int(check["last_obj_cite_number"]), + "length endnotes:", + (doc_abstraction["endnotes"].length > 1) + ? (to!int(doc_abstraction["endnotes"].length)) + : 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, + "length blurb:", + (doc_abstraction["blurb"].length > 1) + ? (to!int(doc_abstraction["blurb"].length)) + : 0, + __FILE__, + __LINE__, + markup.repeat_character_by_number_provided("-", min_repeat_number), + ); + } + } +} diff --git a/src/sdp/meta/object_setter.d b/src/sdp/meta/object_setter.d new file mode 100644 index 0000000..385c3d2 --- /dev/null +++ b/src/sdp/meta/object_setter.d @@ -0,0 +1,65 @@ +/++ + object setter: + setting of sisu objects for downstream processing + meta_object_setter.d ++/ +module sdp.meta.object_setter; +template ObjectSetter() { + /+ structs +/ + struct HeadingAttrib { + string lev = "9"; + int heading_lev_markup = 9; + int heading_lev_collapsed = 9; + int[] closes_lev_collapsed = []; + int[] closes_lev_markup = []; + int array_ptr = 0; + int heading_array_ptr_segments = 0; + } + struct ObjGenericComposite { + // size_t id; + string use = ""; + string is_of = ""; + string is_a = ""; + string text = ""; + string obj_cite_number = ""; + string[] anchor_tags = []; + int indent_base = 0; + int indent_hang = 0; + bool bullet = false; + bool inline_links = false; + bool inline_notes_reg = false; + bool inline_notes_star = false; + string language = ""; // not implemented, consider + string code_block_syntax = ""; + int table_number_of_columns = 0; + double[] table_column_widths = []; + string[] table_column_aligns = []; + bool table_heading = false; + bool table_walls = false; // not implemented + int ocn = 0; + string segment_anchor_tag = ""; + string segname_prev = ""; + string segname_next = ""; + int parent_lev_markup = 0; + int parent_ocn = 0; + int[] ancestors = []; + string marked_up_level = "9"; + int heading_lev_markup = 9; + int heading_lev_collapsed = 9; + int[] dom_markedup = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] dom_collapsed = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + int[] heading_ancestors = [ 0, 0, 0, 0, 0, 0, 0, 0,]; + string[] heading_ancestors_text = [ "", "", "", "", "", "", "", "", ]; + string[] lev4_subtoc = []; + int heading_array_ptr = 0; + int ptr_doc_object = 0; + int ptr_html_segnames = 0; + int ptr_heading = 0; + int array_ptr = 0; + int heading_array_ptr_segments = 0; + string[string][string] node; + } + struct TheObjects { + ObjGenericComposite[] oca; + } +} diff --git a/src/sdp/meta/package.d b/src/sdp/meta/package.d new file mode 100644 index 0000000..fcc6b13 --- /dev/null +++ b/src/sdp/meta/package.d @@ -0,0 +1,17 @@ +module sdp.meta; +public import + sdp.meta.defaults, + sdp.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/sdp/meta/read_config_files.d b/src/sdp/meta/read_config_files.d new file mode 100644 index 0000000..2410316 --- /dev/null +++ b/src/sdp/meta/read_config_files.d @@ -0,0 +1,80 @@ +/++ + read configuration files
+ - read config files
+ meta_config_files.d ++/ +module sdp.meta.read_config_files; +template configIn() { + import + sdp.meta, + std.file, + std.path; + final string configIn(C,E)(C conf_sdl, E env) { + string dot_pwd = chainPath(to!string(env["pwd"]), ".sisu").array; + string underscore_pwd = chainPath(to!string(env["pwd"]), "_sisu").array; + string dot_home = chainPath(to!string(env["home"]), ".sisu").array; + string[] possible_config_path_locations = [ + dot_pwd, + underscore_pwd, + dot_home, + "/etc/sisu" + ]; + string config_file_str; + foreach(pth; possible_config_path_locations) { + auto conf_file = format( + "%s/%s", + pth, + conf_sdl, + ); + try { + if (exists(conf_file)) { + debug(configfile) { + writeln(conf_file); + } + config_file_str = conf_file.readText; + break; + } + } + catch (ErrnoException ex) { + } + catch (FileException ex) { + } + } + return config_file_str; + } +} +/+ + ++/ +template ConfigSDLang() { + import sdlang; + import + sdp.meta, + std.file, + std.path; + auto ConfigSDLang(string configuration, string conf_sdl_filename) { + Tag sdl_root_conf; + try { + sdl_root_conf = parseSource(configuration); + } + catch(ParseException e) { + stderr.writeln("SDLang problem with content for ", conf_sdl_filename); + stderr.writeln(e.msg); + } + return sdl_root_conf; + } +} +/+ ++/ +template configRead() { + import + sdp.meta, + std.file, + std.path; + + final auto configRead(C,E)(C conf_sdl, E env) { + auto configuration = configIn!()(conf_sdl, env); + auto sdl_root = ConfigSDLang!()(configuration, conf_sdl); + return sdl_root; + } +} diff --git a/src/sdp/meta/read_source_files.d b/src/sdp/meta/read_source_files.d new file mode 100644 index 0000000..60f0777 --- /dev/null +++ b/src/sdp/meta/read_source_files.d @@ -0,0 +1,294 @@ +/++ + module meta_read_source_files;
+ - open markup files
+ - if master file scan for addional files to import/insert ++/ +module sdp.meta.read_source_files; +template SiSUrawMarkupContent() { + import + sdp.meta.rgx; + import + sdp.meta, + std.file, + std.path; + mixin SiSUrgxInit; + auto rgx = Rgx(); + auto rawsrc = RawMarkupContent(); + auto SiSUrawMarkupContent(Fn)(Fn fn_src) { + auto _0_header_1_body_content_2_insert_filelist_tuple = + rawsrc.sourceContentSplitIntoHeaderAndBody(rawsrc.sourceContent(fn_src), fn_src); + return _0_header_1_body_content_2_insert_filelist_tuple; + } + struct RawMarkupContent { + final sourceContent(in string fn_src) { + auto raw = MarkupRawUnit(); + auto source_txt_str = + raw.markupSourceReadIn(fn_src); + return source_txt_str; + } + final auto sourceContentSplitIntoHeaderAndBody(in string source_txt_str, in string fn_src="") { + auto raw = MarkupRawUnit(); + string[] insert_file_list; + auto t = + raw.markupSourceHeaderContentRawLineTupleArray(source_txt_str); + auto header_raw = t[0]; + auto sourcefile_body_content = t[1]; + if (fn_src.match(rgx.src_fn_master)) { // filename with path needed if master file (.ssm) not otherwise + auto ins = Inserts(); + auto tu = + ins.scan_master_src_for_insert_files_and_import_content(sourcefile_body_content, fn_src); + static assert(!isTypeTuple!(tu)); + sourcefile_body_content = tu[0]; + insert_file_list = tu[1].dup; + } + t = tuple( + header_raw, + sourcefile_body_content, + insert_file_list + ); + static assert(t.length==3); + return t; + } + } + struct MarkupRawUnit { + import std.file; + final private string readInMarkupSource(in char[] fn_src) { + enforce( + exists(fn_src)!=0, + "file not found" + ); + string source_txt_str; + try { + if (exists(fn_src)) { + source_txt_str = fn_src.readText; + } + } + catch (ErrnoException ex) { + } + catch (UTFException ex) { + // Handle validation errors + } + catch (FileException ex) { + // Handle errors + } + std.utf.validate(source_txt_str); + return source_txt_str; + } + final private char[][] header0Content1(in string src_text) { + /+ split string on _first_ match of "^:?A~\s" into [header, content] array/tuple +/ + char[][] header_and_content; + auto m = (cast(char[]) src_text).matchFirst(rgx.heading_a); + header_and_content ~= m.pre; + header_and_content ~= m.hit ~ m.post; + assert(header_and_content.length == 2, + "document markup is broken, header body split == " + ~ header_and_content.length.to!string + ~ "; (header / body array split should == 2 (split is on level A~))" + ); + return header_and_content; + } + final private char[][] markupSourceLineArray(in char[] src_text) { + char[][] source_line_arr = + (cast(char[]) src_text).split(rgx.newline_eol_strip_preceding); + return source_line_arr; + } + auto markupSourceReadIn(in string fn_src) { + auto rgx = Rgx(); + enforce( + fn_src.match(rgx.src_pth), + "not a sisu markup filename" + ); + auto source_txt_str = readInMarkupSource(fn_src); + return source_txt_str; + } + auto markupSourceHeaderContentRawLineTupleArray(in string source_txt_str) { + string[] file_insert_list = []; + auto hc = header0Content1(source_txt_str); + auto header = hc[0]; + char[] source_txt = hc[1]; + auto source_line_arr = markupSourceLineArray(source_txt); + auto t = tuple( + header, + source_line_arr, + file_insert_list + ); + return t; + } + final char[][] getInsertMarkupSourceContentRawLineArray( + in char[] fn_src_insert, + Regex!(char) rgx_file + ) { + enforce( + fn_src_insert.match(rgx_file), + "not a sisu markup filename" + ); + auto source_txt_str = readInMarkupSource(fn_src_insert); + auto source_line_arr = markupSourceLineArray(source_txt_str); + return source_line_arr; + } + } + struct Inserts { + import sdp.meta.defaults; + auto scan_subdoc_source( + char[][] markup_sourcefile_insert_content, + string fn_src + ) { + mixin SiSUrgxInitFlags; + char[][] contents_insert; + auto type1 = flags_type_init; + auto fn_pth_full = fn_src.match(rgx.src_pth); + auto markup_src_file_path = fn_pth_full.captures[1]; + foreach (line; markup_sourcefile_insert_content) { + if (type1["curly_code"] == 1) { + type1["header_make"] = 0; + type1["header_meta"] = 0; + if (line.matchFirst(rgx.block_curly_code_close)) { + type1["curly_code"] = 0; + } + contents_insert ~= line; + } else if (line.matchFirst(rgx.block_curly_code_open)) { + type1["curly_code"] = 1; + type1["header_make"] = 0; + type1["header_meta"] = 0; + contents_insert ~= line; + } else if (type1["tic_code"] == 1) { + type1["header_make"] = 0; + type1["header_meta"] = 0; + if (line.matchFirst(rgx.block_tic_close)) { + type1["tic_code"] = 0; + } + contents_insert ~= line; + } else if (line.matchFirst(rgx.block_tic_code_open)) { + type1["tic_code"] = 1; + type1["header_make"] = 0; + type1["header_meta"] = 0; + contents_insert ~= line; + } else if ( + (type1["header_make"] == 1) + && line.matchFirst(rgx.native_header_sub) + ) { + type1["header_make"] = 1; + type1["header_meta"] = 0; + } else if ( + (type1["header_meta"] == 1) + && line.matchFirst(rgx.native_header_sub) + ) { + type1["header_meta"] = 1; + type1["header_make"] = 0; + } else if (auto m = line.match(rgx.insert_src_fn_ssi_or_sst)) { + type1["header_make"] = 0; + type1["header_meta"] = 0; + auto insert_fn = m.captures[2]; + auto insert_sub_pth = m.captures[1]; + auto fn_src_insert = + chainPath(markup_src_file_path, insert_sub_pth ~ insert_fn).array; + auto raw = MarkupRawUnit(); + auto markup_sourcesubfile_insert_content = + raw.getInsertMarkupSourceContentRawLineArray(fn_src_insert, rgx.src_fn_find_inserts); + debug(insert_file) { + tell_l("red", line); + tell_l("red", fn_src_insert); + tell_l("fuchsia", "ERROR"); + writeln( + " length contents insert array: ", + markup_sourcesubfile_insert_content.length + ); + } + auto ins = Inserts(); + /+ + 1. load file, + 2. read lines; + 3. scan lines, + 4. if filename insert, and insert filename + 5. repeat 1 + 6. else + 7. add line to new array; + +/ + } else { + type1["header_make"] = 0; + type1["header_meta"] = 0; + contents_insert ~= line; + } + } // end src subdoc (inserts) loop + return contents_insert; + } + auto scan_master_src_for_insert_files_and_import_content( + char[][] sourcefile_body_content, + string fn_src + ) { + mixin SiSUrgxInitFlags; + char[][] contents; + auto type = flags_type_init; + auto fn_pth_full = fn_src.match(rgx.src_pth); + auto markup_src_file_path = fn_pth_full.captures[1]; + string[] insert_file_list =[]; + foreach (line; sourcefile_body_content) { + if (type["curly_code"] == 1) { + if (line.matchFirst(rgx.block_curly_code_close)) { + type["curly_code"] = 0; + } + contents ~= line; + } else if (line.matchFirst(rgx.block_curly_code_open)) { + type["curly_code"] = 1; + contents ~= line; + } else if (type["tic_code"] == 1) { + if (line.matchFirst(rgx.block_tic_close)) { + type["tic_code"] = 0; + } + contents ~= line; + } else if (line.matchFirst(rgx.block_tic_code_open)) { + type["tic_code"] = 1; + contents ~= line; + } else if (auto m = line.match(rgx.insert_src_fn_ssi_or_sst)) { + auto insert_fn = m.captures[2]; + auto insert_sub_pth = m.captures[1]; + auto fn_src_insert = + chainPath(markup_src_file_path, insert_sub_pth ~ insert_fn).array; + insert_file_list ~= to!string(fn_src_insert); + auto raw = MarkupRawUnit(); + /+ TODO +/ + if (auto ma = line.match(rgx.src_fn_text)) { + /+ .sst when inserted, not used: headers and heading level ^:?A~ so remove +/ + writeln(__LINE__); writeln(ma); + } + auto markup_sourcefile_insert_content = + raw.getInsertMarkupSourceContentRawLineArray(fn_src_insert, rgx.src_fn_find_inserts); + debug(insert_file) { + tell_l("red", line); + tell_l("red", fn_src_insert); + writeln( + " length contents insert array: ", + markup_sourcefile_insert_content.length + ); + } + auto ins = Inserts(); + auto contents_insert = ins.scan_subdoc_source( + markup_sourcefile_insert_content, + to!string(fn_src_insert) + ); + contents ~= contents_insert; + /+ + 1. load file, + 2. read lines; + 3. scan lines, + 4. if filename insert, and insert filename + 5. repeat 1 + 6. else + 7. add line to new array; + +/ + } else { + contents ~= line; + } + } // end src doc loop + debug(insert_file) { + writeln(__LINE__); + writeln(contents.length); + } + auto t = tuple( + contents, + insert_file_list + ); + return t; + } + } +} diff --git a/src/sdp/meta/rgx.d b/src/sdp/meta/rgx.d new file mode 100644 index 0000000..854e9cc --- /dev/null +++ b/src/sdp/meta/rgx.d @@ -0,0 +1,256 @@ +/++ + regex: regular expressions used in sisu document parser ++/ +module sdp.meta.rgx; +template SiSUrgxInit() { + import sdp.meta.defaults; + 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|$)`); + static native_header_main = ctRegex!(`^@(?P
[a-z_]+):\s*(?P.*)`, "m"); + static native_header_sub = ctRegex!(`^[ ]*:(?P[a-z_]+):\s+(?P.+)`, "m"); + static native_header_meta_title = ctRegex!(`^@title:\s`, "m"); + static variable_doc_title = ctRegex!(`@title`); + static variable_doc_author = ctRegex!(`@author|@creator`); + static raw_author_munge = ctRegex!(`(\S.+?),\s+(.+)`,"i"); + /+ head +/ + static native_subhead_creator = ctRegex!(`^(?:author|translator|illustrator)$`, "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|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]+)(?:[.:;, ]|$)`); + static heading_marker_missing_tag = ctRegex!(`^:?([A-D1-4])[~] `); + static heading_title = ctRegex!(`^:?[A-D1-4][~][a-z0-9_.-]*[?]?\s+(.+?)$`); + 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 inline_url_generic = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_#]`, "mg"); + static inline_url = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg"); + static inline_link_naked_url = ctRegex!(`(?P^|[ ])(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!'"]?(?:[ ]|$))`, "mg"); + static inline_link_markup_regular = ctRegex!(`(?P^|[ ])\{\s*(?P.+?)\s*\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg"); + static inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg"); + static inline_link_endnote_url_helper = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg"); + static image = ctRegex!(`([a-zA-Z0-9._-]+?\.(?:png|gif|jpg))`, "mg"); + /+ inline markup book index +/ + static book_index = ctRegex!(`^=\{\s*(.+?)\}$`, "m"); + static book_index_open = ctRegex!(`^=\{\s*([^}]+?)$`); + static book_index_close = ctRegex!(`^(.*?)\}$`, "m"); + /+ no obj_cite_number object +/ + static obj_cite_number_off = ctRegex!(`~#$`, "m"); + static obj_cite_number_off_dh = ctRegex!(`-#$`, "m"); + static obj_cite_number_off_all = ctRegex!(`[~-]#$`, "m"); + /+ no obj_cite_number block +/ + static obj_cite_number_off_block = ctRegex!(`^--~#$`); + static obj_cite_number_off_block_dh = ctRegex!(`^---#$`); + static obj_cite_number_off_block_close = ctRegex!(`^--\+#$`); + static obj_cite_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]+)`); + /+ 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_obj_cite_number_offset_split = ctRegex!(`\s*\|\s*`); + static bi_term_and_obj_cite_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 src_pth = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[tm])$`); + static src_fn = + ctRegex!(`^([a-zA-Z0-9._-]+/)*(?P(?P[a-zA-Z0-9._-]+)[.](?Pss[tm]))$`); + 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])$`); + /+ inline markup footnotes endnotes +/ + static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); + static inline_notes_al_gen = ctRegex!(`【.+?】`, "m"); + static inline_notes_al_gen_text = ctRegex!(`【(?P.+?)】`, "m"); + static inline_notes_al_gen_ref = ctRegex!(`【(?P[*+]\s+)\s*(?P.+?)】`, "mg"); + static inline_al_delimiter_open_regular = ctRegex!(`【\s`, "m"); + static inline_al_delimiter_open_symbol_star = ctRegex!(`【[*]\s`, "m"); + static inline_al_delimiter_open_symbol_plus = ctRegex!(`【[+]\s`, "m"); + static inline_al_delimiter_close_regular = ctRegex!(`】`, "m"); + static inline_al_delimiter_open_and_close_regular = ctRegex!(`【|】`, "m"); + static inline_notes_delimiter_al_regular = ctRegex!(`【(.+?)】`, "mg"); + static inline_notes_delimiter_al_regular_number_note = ctRegex!(`【(\d+)\s+(.+?)】`, "mg"); + static inline_al_delimiter_open_asterisk = ctRegex!(`【\*`, "m"); + static inline_al_delimiter_open_plus = ctRegex!(`【\+`, "m"); + static inline_text_and_note_al = ctRegex!(`(?P.+?)【(?:[*+ ]*)(?P.+?)】`, "mg"); + static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|$))`, "mg"); + /+ inline markup footnotes endnotes +/ + static inline_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"); + /+ 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"); + /+ table delimiters +/ + static table_delimiter_col = ctRegex!("[ ]*[┊][ ]*", "mg"); + static table_delimiter_row = ctRegex!("[ ]*\n", "mg"); + } +} -- cgit v1.2.3