diff options
Diffstat (limited to 'lib/sdp/ao_emitter.d')
| -rw-r--r-- | lib/sdp/ao_emitter.d | 1322 |
1 files changed, 1322 insertions, 0 deletions
diff --git a/lib/sdp/ao_emitter.d b/lib/sdp/ao_emitter.d new file mode 100644 index 0000000..dd7103d --- /dev/null +++ b/lib/sdp/ao_emitter.d @@ -0,0 +1,1322 @@ +/* +#+OPTIONS: ^:nil _:nil#+OPTIONS: ^:nil _:nil +* sisu_emitter.d +*/ +mixin template Emitters() { + mixin InternalMarkup; + class CLI { + string[string] extract_actions(string cmdlnins, string[string] actions) + in { } + body { + switch (cmdlnins) { + case "--no-assert": + actions["assert"] = "no"; + break; + default: + break; + } + return actions; + } + } + class OCNemitter : AssertOCN { + int ocn, ocn_; + int ocn_emitter(int ocn_status_flag) + in { assert(ocn_status_flag <= 2); } + body { + if (ocn_status_flag == 0) { + ocn=++ocn_; + } else { + ocn=0; + } + return ocn; + } + invariant() { + } + } + class ObjAttributes { + string[string] obj_txt; + string para_and_blocks(string obj_txt_in) + in { } + body { + auto rgx = new Rgx(); + obj_txt["munge"]=obj_txt_in; + if (match(obj_txt_in, rgx.para_bullet)) { + obj_txt["attrib"] =" \"bullet\": \"true\"," + ~ " \"indent_first\": 0," + ~ " \"indent_rest\": 0,"; + } else if (auto m = match(obj_txt_in, rgx.para_bullet_indent)) { + obj_txt["attrib"] =" \"bullet\": \"true\"," + ~ " \"indent_first\": " ~ to!string(m.captures[1]) ~ "," + ~ " \"indent_rest\": " ~ to!string(m.captures[1]) ~ ","; + } else if (auto m = match(obj_txt_in, rgx.para_indent_hang)) { + obj_txt["attrib"] =" \"bullet\": \"false\"," + ~ " \"indent_first\": " ~ to!string(m.captures[1]) ~ "," + ~ " \"indent_rest\": " ~ to!string(m.captures[2]) ~ ","; + } else if (auto m = match(obj_txt_in, rgx.para_indent)) { + obj_txt["attrib"] =" \"bullet\": \"false\"," + ~ " \"indent_first\": " ~ to!string(m.captures[1]) ~ "," + ~ " \"indent_rest\": " ~ to!string(m.captures[1]) ~ ","; + } else { + obj_txt["attrib"] =" \"bullet\": \"false\"," + ~ " \"indent_first\": 0," + ~ " \"indent_rest\": 0,"; + } + return obj_txt["attrib"]; + } + string para(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"para\"," + ~ " \"is\": \"para\""; + return obj_txt["attrib"]; + } + invariant() { + } + string heading(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"para\"," + ~ " \"is\": \"heading\""; + return obj_txt["attrib"]; + } + invariant() { + } + string header_make(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"head\"," + ~ " \"of\": \"header\"," + ~ " \"is\": \"header_make\""; + return obj_txt["attrib"]; + } + invariant() { + } + string header_metadata(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"head\"," + ~ " \"of\": \"header\"," + ~ " \"is\": \"header_metadata\""; + return obj_txt["attrib"]; + } + invariant() { + } + string code(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"code\""; + return obj_txt["attrib"]; + } + invariant() { + } + string group(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"group\""; + return obj_txt["attrib"]; + } + invariant() { + } + string block(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"block\""; + return obj_txt["attrib"]; + } + invariant() { + } + string verse(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"verse\""; + return obj_txt["attrib"]; + } + invariant() { + } + string quote(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"quote\""; + return obj_txt["attrib"]; + } + invariant() { + } + string table(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"content\"," + ~ " \"of\": \"block\"," + ~ " \"is\": \"table\""; + return obj_txt["attrib"]; + } + invariant() { + } + string comment(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["attrib"] = " \"use\": \"comment\"," + ~ " \"of\": \"comment\"," + ~ " \"is\": \"comment\""; + return obj_txt["attrib"]; + } + invariant() { + } + } + class ObjInlineMarkupMunge { + string[string] obj_txt; + int n_foot, n_foot_reg, n_foot_sp_asterisk, n_foot_sp_plus; + string obj_txt_out, tail, note; + private auto initialize_note_numbers() { + n_foot = 0; + n_foot_reg = 0; + n_foot_sp_asterisk = 0; + n_foot_sp_plus = 0; + } + private auto object_notes_(string obj_txt_in) + in { } + body { + auto rgx = new Rgx(); + auto mkup = new InternalMarkup(); + obj_txt_out = ""; + tail = ""; + obj_txt_in = replaceAll( + obj_txt_in, + rgx.inline_notes_curly_sp_asterisk, + (mkup.en_a_o ~ "*" ~ " $1" ~ mkup.en_a_c) + ); + obj_txt_in = + replaceAll( + obj_txt_in, + rgx.inline_notes_curly_sp_plus, + (mkup.en_a_o ~ "+" ~ " $1" ~ mkup.en_a_c) + ); + obj_txt_in = + replaceAll( + obj_txt_in, + rgx.inline_notes_curly, + (mkup.en_a_o ~ " $1" ~ mkup.en_a_c) + ); + if (match(obj_txt_in, rgx.inline_notes_al_gen)) { + foreach(m; matchAll(obj_txt_in, rgx.inline_text_and_note_al)) { + if (match(obj_txt_in, rgx.inline_al_delimiter_open_asterisk)) { + n_foot_sp_asterisk++; + n_foot=n_foot_sp_asterisk; + } else if (match(obj_txt_in, rgx.inline_al_delimiter_open_plus)) { + n_foot_sp_plus++; + n_foot=n_foot_sp_plus; + } else { + n_foot_reg++; + n_foot=n_foot_reg; + } + obj_txt_out ~= replaceFirst( + m.hit, + rgx.inline_al_delimiter_open_regular, + (mkup.en_a_o ~ to!string(n_foot)) + ); + tail = m.post; + } + } else { + obj_txt_out = obj_txt_in; + } + 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); + } + } + return obj_txt_out; + } + string para(string obj_txt_in) + in { } + body { + auto rgx = new Rgx(); + obj_txt["munge"]=obj_txt_in; + obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.para_attribs, ""); + obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.ocn_off_all, ""); + obj_txt["munge"]=object_notes_(obj_txt["munge"]); + debug(munge) { + writeln(__LINE__); + writeln(obj_txt_in); + writeln(__LINE__); + writeln(to!string(obj_txt["munge"])); + } + return obj_txt["munge"]; + } + string heading(string obj_txt_in) + in { } + body { + auto rgx = new Rgx(); + obj_txt["munge"]=obj_txt_in; + obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.heading, ""); + obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.ocn_off_all, ""); + obj_txt["munge"]=object_notes_(obj_txt["munge"]); + debug(munge) { + writeln(__LINE__); + writeln(obj_txt_in); + writeln(__LINE__); + writeln(to!string(obj_txt["munge"])); + } + return obj_txt["munge"]; + } + invariant() { + } + string header_make(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string header_metadata(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string code(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string group(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["munge"]=object_notes_(obj_txt["munge"]); + return obj_txt["munge"]; + } + invariant() { + } + string block(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["munge"]=object_notes_(obj_txt["munge"]); + return obj_txt["munge"]; + } + invariant() { + } + string verse(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + obj_txt["munge"]=object_notes_(obj_txt["munge"]); + return obj_txt["munge"]; + } + invariant() { + } + string quote(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string table(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + string comment(string obj_txt_in) + in { } + body { + obj_txt["munge"]=obj_txt_in; + return obj_txt["munge"]; + } + invariant() { + } + } + class ObjInlineMarkup : AssertObjInlineMarkup { + auto munge = new ObjInlineMarkupMunge(); + string[string] obj_txt; + string obj_inline_markup(string obj_is_, string obj_raw) + in { } + body { + obj_txt["munge"]=obj_raw.dup; + obj_txt["munge"]=(match(obj_is_, ctRegex!(`verse|code`))) + ? obj_txt["munge"] + : strip(obj_txt["munge"]); + switch (obj_is_) { + case "header_make": + obj_txt["munge"]=munge.header_make(obj_txt["munge"]); + break; + case "header_metadata": + obj_txt["munge"]=munge.header_metadata(obj_txt["munge"]); + break; + case "heading": + obj_txt["munge"]=munge.heading(obj_txt["munge"]); + break; + case "para": + obj_txt["munge"]=munge.para(obj_txt["munge"]); + break; + case "code": + obj_txt["munge"]=munge.code(obj_txt["munge"]); + break; + case "group": + obj_txt["munge"]=munge.group(obj_txt["munge"]); + break; + case "block": + obj_txt["munge"]=munge.block(obj_txt["munge"]); + break; + case "verse": + obj_txt["munge"]=munge.verse(obj_txt["munge"]); + break; + case "quote": + obj_txt["munge"]=munge.quote(obj_txt["munge"]); + break; + case "table": + obj_txt["munge"]=munge.table(obj_txt["munge"]); + break; + case "comment": + obj_txt["munge"]=munge.comment(obj_txt["munge"]); + break; + case "doc_end_reset": + munge.initialize_note_numbers(); + break; + default: + break; + } + return obj_txt["munge"]; + } + invariant() { + } + } + class ObjAttrib : AssertObjAttrib { +// auto sink = appender!(char[])(); + auto attrib = new ObjAttributes(); + string[string] obj_attrib; + string obj_attributes(string obj_is_, string obj_raw, string node) + in { } + body { + scope(exit) { + destroy(obj_raw); + destroy(node); + } + JSONValue node_j = parseJSON(node); + obj_attrib.remove("json"); + obj_attrib["json"] ="{"; + switch (obj_is_) { + case "header_make": + obj_attrib["json"] ~= attrib.header_make(obj_raw); + break; + case "header_metadata": + obj_attrib["json"] ~= attrib.header_metadata(obj_raw); + break; + case "heading": + obj_attrib["json"] ~= attrib.heading(obj_raw); // + break; + case "para": + obj_attrib["json"] ~= attrib.para_and_blocks(obj_raw) + ~ attrib.para(obj_raw); + break; + case "code": + obj_attrib["json"] ~= attrib.code(obj_raw); + break; + case "group": + obj_attrib["json"] ~= attrib.para_and_blocks(obj_raw) + ~ attrib.group(obj_raw); + break; + case "block": + obj_attrib["json"] ~= attrib.para_and_blocks(obj_raw) + ~ attrib.block(obj_raw); + break; + case "verse": + obj_attrib["json"] ~= attrib.verse(obj_raw); + break; + case "quote": + obj_attrib["json"] ~= attrib.quote(obj_raw); + break; + case "table": + obj_attrib["json"] ~= attrib.table(obj_raw); + break; + case "comment": + obj_attrib["json"] ~= attrib.comment(obj_raw); + break; + default: + obj_attrib["json"] ~= attrib.para(obj_raw); + break; + } + obj_attrib["json"] ~=" }"; + JSONValue oa_j = parseJSON(obj_attrib["json"]); + assert( + (oa_j.type == JSON_TYPE.OBJECT) && + (node_j.type == JSON_TYPE.OBJECT) + ); + if (obj_is_ == "heading") { + oa_j.object["ocn"] = node_j["ocn"]; + oa_j.object["lvn"] = node_j["lvn"]; + oa_j.object["lcn"] = node_j["lcn"]; + oa_j.object["heading_pointer"] = + node_j["heading_pointer"]; // check + oa_j.object["doc_object_pointer"] = + node_j["doc_object_pointer"]; // check + } + oa_j.object["parent_ocn"] = node_j["parent_ocn"]; + oa_j.object["parent_lvn"] = node_j["parent_lvn"]; + obj_attrib["json"] = oa_j.toString(); + debug(structattrib) { + if (oa_j["is"].str() == "heading") { + writeln(obj_attrib["json"]); + writeln( + "is: ", oa_j["is"].str(), + "; ocn: ", oa_j["ocn"].integer() + ); + } + } + return obj_attrib["json"]; + } + invariant() { + } + } + class HeaderDocMetadataMakeJson { + auto rgx = new Rgx(); + string hm, hs; + auto header_metadata_and_make_jsonstr( + string header, + JSONValue[string] dochead_metadata, + JSONValue[string] dochead_make + ) + in { } + body { + scope(exit) { + destroy(header); + destroy(dochead_metadata); + destroy(dochead_make); + } + if (auto t = match(header, rgx.head_main)) { + char[][] obj_spl = split( + cast(char[]) header, + rgx.line_delimiter_ws_strip + ); + auto hm = to!string(t.captures[1]); + if (match(hm, rgx.main_headers)) { + foreach (line; obj_spl) { + if (auto m = match(line, rgx.head_main)) { + if (!empty(m.captures[2])) { + if (hm == "creator") { + dochead_metadata[hm]["author"].str = + to!string(m.captures[2]); + } else if (hm == "title") { + dochead_metadata[hm]["main"].str = + to!string(m.captures[2]); + } else if (hm == "publisher") { + dochead_metadata[hm]["name"].str = + to!string(m.captures[2]); + } + } + } else if (auto s = match(line, rgx.head_sub)) { + if (!empty(s.captures[2])) { + auto hs = to!string(s.captures[1]); + if ((hm == "make" ) + && (dochead_make[hm].type() == JSON_TYPE.OBJECT)) { + switch (hm) { + case "make": + if (match(hs, rgx.subhead_make)) { + if (dochead_make[hm][hs].type() == JSON_TYPE.STRING) { + dochead_make[hm][hs].str = to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + default: + break; + } + } else if (dochead_metadata[hm].type() == JSON_TYPE.OBJECT) { + switch (hm) { + case "creator": + if (match(hs, rgx.subhead_creator)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "title": + if (match(hs, rgx.subhead_title)) { + if ((hs == "subtitle") + && (dochead_metadata[hm]["sub"].type() == JSON_TYPE.STRING)) { + dochead_metadata[hm]["sub"].str = + to!string(s.captures[2]); + } else if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "rights": + if (match(hs, rgx.subhead_rights)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "date": + if (match(hs, rgx.subhead_date)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "original": + if (match(hs, rgx.subhead_original)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "classify": + if (match(hs, rgx.subhead_classify)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "identifier": + if (match(hs, rgx.subhead_identifier)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "notes": + if (match(hs, rgx.subhead_notes)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + to!string(s.captures[2]); + } + } else { + writeln("not a valid header type:", hm, ":", hs); + destroy(hm); + destroy(hs); + } + break; + case "publisher": + if (match(hs, rgx.subhead_publisher)) { + if (dochead_metadata[hm][hs].type() == JSON_TYPE.STRING) { + dochead_metadata[hm][hs].str = + 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_metadata, dochead_make); + static assert(!isTypeTuple!(t)); + return t; + } + } + class HeaderMetadataMakeHash { + auto rgx = new Rgx(); + string header_main; + string[string] head; + string[string] header_topic_hash(string header) + in { } + body { + if (auto t = match(header, rgx.head_main)) { + char[][] obj_spl = split( + cast(char[]) header, + rgx.line_delimiter_ws_strip + ); + auto header_main = to!string(t.captures[1]); + head[header_main] = "{"; + foreach (line; obj_spl) { + if (auto m = match(line, rgx.head_main)) { + if (!empty(m.captures[2])) { + head[header_main] ~= + "\"" ~ header_main ~ + "\": \"" ~ + to!string(m.captures[2]) ~ + "\","; + } + } else if (auto s = match(line, rgx.head_sub)) { + head[header_main] ~= "\"" ~ s.captures[1] ~ "\":"; + if (!empty(s.captures[2])) { + head[header_main] ~= "\"" ~ s.captures[2] ~ "\","; + } + } + } + head[header_main] = replaceFirst( + head[header_main], + rgx.tailing_comma, + "" + ); + head[header_main] ~= "}"; + debug(headerjson) { + JSONValue j = parseJSON(head[header_main]); + assert( + (j.type == JSON_TYPE.OBJECT) + ); + } + } + return head; + } + invariant() { + } + } + class BookIndexNuggetHash : AssertBookIndexNuggetHash { + string main_term, sub_term, sub_term_bits; + uint ocn_offset, ocn_endpoint; + string[] ocns; + string[][string][string] bi; + string[][string][string] hash_nugget; + string[] bi_main_terms_split_arr; + string[][string][string] bookindex_nugget_hash(string bookindex, int ocn) + in { } + body { + auto rgx = new Rgx(); + if (!bookindex.empty) { + auto bi_main_terms_split_arr = + split(bookindex, rgx.bi_main_terms_split); + foreach (bi_main_terms_content; bi_main_terms_split_arr) { + auto bi_main_term_and_rest = + split(bi_main_terms_content, rgx.bi_main_term_plus_rest_split); + if (auto m = match( + bi_main_term_and_rest[0], + rgx.bi_term_and_ocns_match) + ) { + main_term = strip(m.captures[1]); + ocn_offset = to!uint(m.captures[2]); + ocn_endpoint=(ocn + ocn_offset); + ocns ~= (to!string(ocn) ~ "-" ~ to!string(ocn_endpoint)); + } else { + main_term = strip(bi_main_term_and_rest[0]); + ocns ~= to!string(ocn); + } + bi[main_term]["_a"] ~= ocns; + ocns=null; + if (bi_main_term_and_rest.length > 1) { + auto bi_sub_terms_split_arr = + split( + bi_main_term_and_rest[1], + rgx.bi_sub_terms_plus_ocn_offset_split + ); + foreach (sub_terms_bits; bi_sub_terms_split_arr) { + if (auto m = match(sub_terms_bits, rgx.bi_term_and_ocns_match)) { + sub_term = strip(m.captures[1]); + ocn_offset = to!uint(m.captures[2]); + ocn_endpoint=(ocn + ocn_offset); + ocns ~= (to!string(ocn) ~ " - " ~ to!string(ocn_endpoint)); + } else { + sub_term = strip(sub_terms_bits); + ocns ~= to!string(ocn); + } + if (!empty(sub_term)) { + bi[main_term][sub_term] ~= ocns; + } + ocns=null; + } + } + } + } + hash_nugget = bi; + return hash_nugget; + } + invariant() { + } + } + class BookIndexReport { + int mkn, skn; + auto bookindex_report_sorted( + string[][string][string] bookindex_unordered_hashes + ) { + auto mainkeys=bookindex_unordered_hashes.byKey.array. + sort!("toLower(a) < toLower(b)", SwapStrategy.stable).release; + foreach (mainkey; mainkeys) { + auto subkeys=bookindex_unordered_hashes[mainkey].byKey.array. + sort!("toLower(a) < toLower(b)", SwapStrategy.stable).release; + foreach (subkey; subkeys) { + debug(bookindex) { + writeln( + mainkey, ": ", + subkey, ": ", + to!string(bookindex_unordered_hashes[mainkey][subkey]) + ); + } + skn++; + } + mkn++; + } + } + } + class BookIndexReportIndent { + int mkn, skn; + auto bookindex_report_indented( + string[][string][string] bookindex_unordered_hashes + ) { + auto mainkeys= + bookindex_unordered_hashes.byKey.array.sort().release; + foreach (mainkey; mainkeys) { + debug(bookindex) { + writeln(mainkey); + } + auto subkeys= + bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + debug(bookindex) { + writeln(" ", subkey); + writeln(" ", to!string( + bookindex_unordered_hashes[mainkey][subkey] + )); + } + skn++; + } + mkn++; + } + } + } + class BookIndexReportSection { + mixin ObjectSetters; + int mkn, skn; + auto rgx = new Rgx(); + auto bookindex_write_section( + string[][string][string] bookindex_unordered_hashes + ) { + 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 = replaceAll(ref_, 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 = replaceAll(ref_, rgx.book_index_go, "$1"); + write(" {", ref_, "}#", go, ", "); + } + writeln(" \\\\"); + skn++; + } + mkn++; + } + } + auto bookindex_build_section( + string[][string][string] bookindex_unordered_hashes, + int ocn + ) { + string type; + int type_heading; + string lev, lvn, lcn; + string attrib; + string indent_first; + string indent_second; + auto set_oa = new ObjectAbstractSet(); + auto mainkeys = + bookindex_unordered_hashes.byKey.array.sort().release; + string bi_tmp; + string[string][1024] bookindex_arbitrary_max_length_set; + writeln(mainkeys.length); + type_heading=1; + bi_tmp = "Book Index"; + attrib=""; + lev="B"; + lvn="1"; + lcn="1"; + bookindex_arbitrary_max_length_set[mkn] = + set_oa.contents_heading( + type_heading, + bi_tmp, + attrib, + ocn, + lev, + lvn, + lcn + ); + ocn++; + mkn++; + type_heading=1; + bi_tmp = "Index"; + attrib=""; + lev="1"; + lvn="4"; + lcn="2"; + bookindex_arbitrary_max_length_set[mkn] = + set_oa.contents_heading( + type_heading, + bi_tmp, + attrib, + ocn, + lev, + lvn, + lcn + ); + ocn++; + mkn++; + foreach (mainkey; mainkeys) { + bi_tmp = "!{" ~ mainkey ~ "}! "; + foreach (ref_; bookindex_unordered_hashes[mainkey]["_a"]) { + auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + bi_tmp ~= " {" ~ ref_ ~ "}#" ~ go ~ ", "; + } + bi_tmp ~= " \\\\\n "; + bookindex_unordered_hashes[mainkey].remove("_a"); + auto subkeys = + bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + bi_tmp ~= subkey ~ ", "; + foreach (ref_; bookindex_unordered_hashes[mainkey][subkey]) { + auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + bi_tmp ~= " {" ~ ref_ ~ "}#" ~ go ~ ", "; + } + bi_tmp ~= " \\\\\n "; + skn++; + } + bi_tmp = replaceFirst(bi_tmp, rgx.trailing_linebreak, ""); + type="para"; + attrib=""; + indent_first = "0"; + indent_second = "1"; + attrib=""; + bookindex_arbitrary_max_length_set[mkn] = + set_oa.contents_para( + type, + bi_tmp, + attrib, + ocn, + indent_first, + indent_second, + false + ); + ocn++; + mkn++; + } + auto bookindex = + bookindex_arbitrary_max_length_set[0..mkn].dup; + auto t = tuple(bookindex, ocn); + return t; + } + auto bookindex_build_section_( + string[][string][string] bookindex_unordered_hashes + ) { + auto mainkeys = + bookindex_unordered_hashes.byKey.array.sort().release; + string bi_tmp; + string[1024] bookindex_arbitrary_max_length_set; + writeln(mainkeys.length); + foreach (mainkey; mainkeys) { + bi_tmp = "_0_1 !{" ~ mainkey ~ "}! "; + foreach (ref_; bookindex_unordered_hashes[mainkey]["_a"]) { + auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + bi_tmp ~= " {" ~ ref_ ~ "}#" ~ go ~ ", "; + } + bi_tmp ~= " \\\\\n "; + bookindex_unordered_hashes[mainkey].remove("_a"); + auto subkeys = + bookindex_unordered_hashes[mainkey].byKey.array.sort().release; + foreach (subkey; subkeys) { + bi_tmp ~= subkey ~ ", "; + foreach (ref_; bookindex_unordered_hashes[mainkey][subkey]) { + auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + bi_tmp ~= " {" ~ ref_ ~ "}#" ~ go ~ ", "; + } + bi_tmp ~= " \\\\\n "; + skn++; + } + bi_tmp = replaceFirst(bi_tmp, rgx.trailing_linebreak, ""); + bookindex_arbitrary_max_length_set[mkn] = bi_tmp; + mkn++; + } + auto bookindex = + bookindex_arbitrary_max_length_set[0..mkn].dup; + return bookindex; + } + } + class NotesSection { + mixin ObjectSetters; + string object_notes; + int previous_count; + int mkn; + auto rgx = new Rgx(); + private auto gather_notes_for_endnote_section( + string[string][131072] contents_arbitrary_max_length_set, + int counter + ) + in { + assert((contents_arbitrary_max_length_set[counter]["is"] == "para") + || (contents_arbitrary_max_length_set[counter]["is"] == "heading")); + assert(counter > previous_count); + previous_count=counter; + assert( + match(contents_arbitrary_max_length_set[counter]["obj"], + rgx.inline_notes_delimiter_al_regular_number_note) + ); + } + body { + foreach(m; + matchAll(contents_arbitrary_max_length_set[counter]["obj"], + rgx.inline_notes_delimiter_al_regular_number_note)) { + debug(endnotes_build) { + writeln( + "{^{", m.captures[1], ".}^}#noteref_", m.captures[1], " ", + m.captures[2]); // sometimes need segment name (segmented html & epub) + } + object_notes ~= + "{^{" ~ m.captures[1] ~ ".}^}#noteref_" ~ + m.captures[1] ~ " " ~ m.captures[2] ~ "』"; + } + return object_notes; + } + private auto gathered_notes() + in { + } + body { + string[] endnotes_; + if (object_notes.length > 1) { + endnotes_ = (split(object_notes, rgx.break_string))[0..$-1]; + } + return endnotes_; + } + private auto endnote_objects(int ocn) + in { + } + body { + auto set_oa = new ObjectAbstractSet(); + string[string][1024] endnotes_arbitrary_max_length_set; + auto endnotes_ = gathered_notes(); + string type; + int type_heading; + string lev, lvn, lcn; + string attrib; + string indent_first; + string indent_second; + type_heading=1; + attrib=""; + lev="B"; + lvn="1"; + lcn="1"; + endnotes_arbitrary_max_length_set[mkn] = + set_oa.contents_heading( + type_heading, + "Endnotes", + attrib, + ocn, + lev, + lvn, + lcn + ); + ocn++; + mkn++; + type_heading=1; + attrib=""; + lev="1"; + lvn="4"; + lcn="2"; + endnotes_arbitrary_max_length_set[mkn] = + set_oa.contents_heading( + type_heading, + "Endnotes", + attrib, + ocn, + lev, + lvn, + lcn + ); + ocn++; + mkn++; + foreach (endnote; endnotes_) { + type="para"; + attrib=""; + indent_first = "0"; + indent_second = "0"; + attrib=""; + endnotes_arbitrary_max_length_set[mkn] = + set_oa.contents_para( + type, + endnote, + attrib, + ocn, + indent_first, + indent_second, + false + ); + ocn++; + mkn++; + } + auto endnotes = + endnotes_arbitrary_max_length_set[0..mkn].dup; + auto t = tuple(endnotes, ocn); + return t; + } + } + class Bibliography { + public JSONValue[] bibliography(string[] biblio_unsorted_incomplete) + in { } + body { + JSONValue[] biblio_unsorted = + biblio_unsorted_complete(biblio_unsorted_incomplete); + JSONValue[] biblio_sorted = biblio_sort(biblio_unsorted); + biblio_debug(biblio_sorted); + return biblio_sorted; + } + final private JSONValue[] biblio_unsorted_complete( + string[] biblio_unordered + ) { + JSONValue[1024] bib_arr_json; + int count_biblio_entry; + count_biblio_entry=0; + foreach (bibent; biblio_unordered) { + 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[count_biblio_entry] = j; + count_biblio_entry++; + } + JSONValue[] biblio_unsorted_array_of_json_objects = + bib_arr_json[0..(count_biblio_entry)].dup; + return biblio_unsorted_array_of_json_objects; + } + final private JSONValue[] biblio_sort(JSONValue[] biblio_unordered) { + 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; + } + auto biblio_debug(JSONValue[] biblio_sorted) { + debug(biblio) { + foreach (j; biblio_sorted) { + if (!empty(j["fulltitle"].str)) { + writeln(j["sortby_deemed_author_year_title"]); + } + } + } + } + } + class NodeStructureMetadata : AssertNodeJSON { + int lv, lv0, lv1, lv2, lv3, lv4, lv5, lv6, lv7; + uint ocn; + uint[string] p_; // p_ parent_ + string node; + string node_emitter( + string lvn, + int ocn_, + int counter_, + int pointer_, + string is_ + ) + in { + auto rgx = new Rgx(); + } + body { + assert(is_ != "heading"); // should not be necessary + assert(to!int(ocn_) >= 0); // should not be necessary + uint ocn=to!uint(ocn_); + if (lv7 > 0) { + p_["lvn"] = 7; p_["ocn"] = lv7; + } else if (lv6 > 0) { + p_["lvn"] = 6; p_["ocn"] = lv6; + } else if (lv5 > 0) { + p_["lvn"] = 5; p_["ocn"] = lv5; + } else { + p_["lvn"] = 4; p_["ocn"] = lv4; + } + node=("{ " ~ + "\"is\": \"" ~ is_ ~ "\"" ~ + ", \"heading_pointer\": " ~ to!string(pointer_) ~ + ", \"doc_object_pointer\": " ~ to!string(counter_) ~ + ", \"ocn\": " ~ to!string(ocn_) ~ + ", \"parent_ocn\": " ~ to!string(p_["ocn"]) ~ + ", \"parent_lvn\": " ~ to!string(p_["lvn"]) ~ + " }" + ); + return node; + } + invariant() { + } + string node_emitter_heading( + string lvn, + string lcn, + int ocn_, + int counter_, + int pointer_, + string is_ + ) + in { + auto rgx = new Rgx(); + } + body { + uint ocn=to!uint(ocn_); + switch (lvn) { // switch (to!string(lv)) { + case "0": + lv=0; + lv0=ocn; lv1=0; lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lvn"] = 0; p_["ocn"] = 0; + break; + case "1": + lv=1; + lv1=ocn; lv2=0; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lvn"] = 0; p_["ocn"] = lv0; + break; + case "2": + lv=2; + lv2=ocn; lv3=0; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lvn"] = 1; p_["ocn"] = lv1; + break; + case "3": + lv=3; + lv3=ocn; lv4=0; lv5=0; lv6=0; lv7=0; + p_["lvn"] = 2; p_["ocn"] = lv2; + break; + case "4": + lv=4; + lv4=ocn; lv5=0; lv6=0; lv7=0; + if (lv3 > 0) { + p_["lvn"] = 3; p_["ocn"] = lv3; + } else if (lv2 > 0) { + p_["lvn"] = 2; p_["ocn"] = lv2; + } else if (lv1 > 0) { + p_["lvn"] = 1; p_["ocn"] = lv1; + } else { + p_["lvn"] = 0; p_["ocn"] = lv0; + } + break; + case "5": + lv=5; + lv5=ocn; lv6=0; lv7=0; + p_["lvn"] = 4; p_["ocn"] = lv4; + break; + case "6": + lv=6; + lv6=ocn; lv7=0; + p_["lvn"] = 5; p_["ocn"] = lv5; + break; + case "7": + lv=7; + lv7=ocn; + p_["lvn"] = 6; p_["ocn"] = lv6; + break; + default: + break; + } + node=("{ " ~ + "\"is\": \"" ~ is_ ~ "\"" ~ + ", \"heading_pointer\": " ~ to!string(pointer_) ~ + ", \"doc_object_pointer\": " ~ to!string(counter_) ~ + ", \"ocn\": " ~ to!string(ocn_) ~ + ", \"lvn\": " ~ to!string(lvn) ~ + ", \"lcn\": " ~ to!string(lcn) ~ + ", \"parent_ocn\": " ~ to!string(p_["ocn"]) ~ + ", \"parent_lvn\": " ~ to!string(p_["lvn"]) ~ + " }" + ); + return node; + } + invariant() { + } + } +} |