From a87a80c4528ce1bfd0090b425f4194eddf33699c Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Thu, 23 Jun 2016 00:55:33 -0400 Subject: step5.1 headers sdlang (like native headers) converted to json for common internal format --- src/sdp/ao_abstract_doc_source.d | 10 +- src/sdp/ao_defaults.d | 265 +++++--- src/sdp/ao_header_extract.d | 1307 ++++++++++++++++++++++++++++++++++++++ src/sdp/ao_read_source_files.d | 57 +- src/sdp/ao_rgx.d | 46 +- 5 files changed, 1512 insertions(+), 173 deletions(-) create mode 100644 src/sdp/ao_header_extract.d (limited to 'src/sdp') diff --git a/src/sdp/ao_abstract_doc_source.d b/src/sdp/ao_abstract_doc_source.d index c814c15..aca4944 100644 --- a/src/sdp/ao_abstract_doc_source.d +++ b/src/sdp/ao_abstract_doc_source.d @@ -338,11 +338,11 @@ template SiSUdocAbstraction() { && ((type["para"] == State.off) && (type["heading"] == State.off))) { /+ heading or para but neither flag nor line exists +/ - // if ((to!string(dochead_make["make"]["headings"]).length > 2) + // if ((to!string(dochead_make_json["make"]["headings"]).length > 2) // && (type["make_headings"] == State.off)) { // /+ heading found +/ // auto dochead_make_headings = - // to!string(dochead_make["make"]["headings"]); + // to!string(dochead_make_json["make"]["headings"]); // heading_found(line, dochead_make_headings, heading_match_str, heading_match_rgx, type); // } if ((type["make_headings"] == State.on) @@ -1691,10 +1691,8 @@ template SiSUdocAbstraction() { assertions_doc_structure(an_object, lv); // includes most of the logic for collapsed levels switch (an_object["lev"]) { case "A": - // an_object["obj"]=replaceFirst(an_object["obj"], rgx.head_value_title, to!string(dochead_metadata["title"]["main"])); - // an_object["obj"]=replaceFirst(an_object["obj"], rgx.head_value_author, to!string(dochead_metadata["creator"]["author"])); - // // an_object["obj"]=replaceFirst(an_object["obj"], rgx.head_value_title, to!string(parseJSON(dochead_metadata["title"]["main"]))); - // // an_object["obj"]=replaceFirst(an_object["obj"], rgx.head_value_author, to!string(parseJSON(dochead_metadata["creator"]["author"]))); + // an_object["obj"]=replaceFirst(an_object["obj"], rgx.variable_doc_title, to!string(dochead_meta_json["title"]["main"])); + // an_object["obj"]=replaceFirst(an_object["obj"], rgx.variable_doc_author, to!string(dochead_meta_json["creator"]["author"])); collapsed_lev["h0"] = 1; an_object["lev_collapsed_number"] = to!string(collapsed_lev["h0"]); diff --git a/src/sdp/ao_defaults.d b/src/sdp/ao_defaults.d index 087067c..b6eeae1 100644 --- a/src/sdp/ao_defaults.d +++ b/src/sdp/ao_defaults.d @@ -5,148 +5,125 @@ template SiSUheaderSkel() { auto header_make_jsonstr = `{ "make": { + "bold" : "", + "breaks" : "", "cover_image" : "", - "home_button_image" : "", - "home_button_text" : "", + "css" : "", + "emphasis" : "", "footer" : "", "headings" : "", + "home_button_image" : "", + "home_button_text" : "", + "italics" : "", "num_top" : "", - "breaks" : "", "substitute" : "", - "bold" : "", - "italics" : "", - "emphasis" : "", - "texpdf_font" : "", - "css" : "" + "texpdf_font" : "" } }`; auto header_meta_jsonstr = `{ + "classify": { + "dewey" : "", + "keywords" : "", + "loc" : "", + "subject" : "", + "topic_register" : "" + }, "creator": { "author" : "", - "translator" : "", - "illustrator" : "" - }, - "title": { - "main" : "", - "sub" : "", - "full" : "", - "language" : "", - "edition" : "", - "note" : "" - }, - "rights": { - "copyright" : "", - "illustrations" : "", - "license" : "", - "cover" : "" + "author_email" : "", + "illustrator" : "", + "translator" : "" }, "date": { - "published" : "", + "added_to_site" : "", + "available" : "", "created" : "", "issued" : "", - "available" : "", - "valid" : "", "modified" : "", - "added_to_site" : "" - }, - "original": { - "title" : "", - "language" : "", - "source" : "" - }, - "classify": { - "topic_register" : "", - "subject" : "", - "keywords" : "", - "loc" : "", - "dewey" : "" + "published" : "", + "valid" : "" }, "identifier": { + "isbn" : "", "oclc" : "", - "pg" : "", - "isbn" : "" + "pg" : "" + }, + "links": { + "link" : "" }, "notes": { "abstract" : "", "description" : "" }, + "original": { + "language" : "", + "source" : "", + "title" : "" + }, "publisher": { "name" : "" }, - "links": { + "rights": { + "copyright" : "", + "cover" : "", + "illustrations" : "", + "license" : "" + }, + "title": { + "edition" : "", + "full" : "", + "language" : "", + "main" : "", + "note" : "", + "sub" : "" } }`; // links + auto pointer_head_sub_classify = + [ + "dewey", + "keywords", // + "loc", + "subject", + "topic_register" + ]; auto pointer_head_main = [ + "classify", "creator", - "title", - "rights", "date", - "original", - "classify", "identifier", + "links", + "make", /+ make +/ + "original", "notes", - "make", - "links" + "rights", + "title" ]; auto pointer_head_sub_creator = [ "author", - "translator", + "author_email", + "cover", "illustrator", - "cover" - ]; - auto pointer_head_sub_title = - [ - "main", - "sub", - "full", - "language", - "edition", - "note" - ]; - auto pointer_head_sub_rights = - [ - "copyright", - "illustrations", - "license" + "translator" ]; auto pointer_head_sub_date = [ - "published", + "added_to_site", + "available", "created", "issued", - "valid", "modified", - "added_to_site" - ]; - auto pointer_head_sub_original = - [ - "title", - "language", - "source" - ]; - auto pointer_head_sub_classify = - [ - "topic_register", - "subject", - "keywords", - "loc", - "dewey" + "published", + "valid" ]; auto pointer_head_sub_identifier = [ + "isbn", "oclc", - "pg", - "isbn" - ]; - auto pointer_head_sub_notes = - [ - "abstract", - "description" + "pg" ]; - auto pointer_head_sub_publisher = - [ "name" ]; + /+ make +/ auto pointer_head_sub_make = [ "cover_image", @@ -162,9 +139,112 @@ template SiSUheaderSkel() { "texpdf_font", "css" ]; + auto pointer_head_sub_notes = + [ + "abstract", + "description" + ]; + auto pointer_head_sub_original = + [ + "language", + "source", + "title" + ]; + auto pointer_head_sub_publisher = + [ "name" ]; + auto pointer_head_sub_rights = + [ + "copyright", + "cover", + "illustrations", + "license" + ]; + auto pointer_head_sub_title = + [ + "edition", + "full", + "language", + "main", + "note", + "sub" + ]; auto config_jsonstr = `{ }`; } +/+ +/+ + unify internal representation of header info for native & sdlang document headers + represent either using struct, hashes or possibly json + sdp internal representation should be identical for native & sdlang variants ++/ +header. + ├── make // make instructions + │   ├── bold + │   ├── breaks + │   ├── cover_image + │   ├── css + │   ├── emphasis + │   ├── footer + │   ├── headings + │   ├── home_button_image + │   ├── home_button_text + │   ├── italics + │   ├── num_top + │   ├── substitute + │   └── texpdf_font + └── meta // metadata +    ├── author // move author to creator:author +    ├── classify +    │   ├── dewey +    │   ├── keyword +    │   ├── loc +    │   ├── subject +    │   └── topic_register +    ├── creator +    │   ├── author + │ │ ├── [ [first_name: x0, last_name: y0], [first_name: x1, last_name: y1] ] + │ │ └── [ full_name0, full_name1 ] +    │   ├── author_email +    │   ├── illustrator +    │   └── translator +    ├── date +    │   ├── added_to_site +    │   ├── available +    │   ├── created +    │   ├── issued +    │   ├── modified +    │   ├── published +    │   └── valid +    ├── identifier +    │   ├── isbn +    │   ├── oclc +    │   └── pg +    ├── links +    ├── notes +    │   ├── abstract +    │   └── description +    ├── original +    │   ├── language +    │   ├── source +    │   └── title +    ├── publisher +    │   └── name +    ├── rights +    │   ├── copyright +    │   ├── cover +    │   ├── illustrations +    │   └── license +    └── title // move title: to title:main +    ├── edition +    ├── [ full (main + sub) ] +    ├── language +    ├── main +    ├── note +    ├── sub +    └── subtitle // move title:subtitle to title:sub + +61 leaves ++/ /+ regex flags +/ template SiSUrgxInitFlags() { int[string] flags_type_init() { @@ -203,6 +283,7 @@ template SiSUrgxInitFlags() { } } template SiSUbiblio() { + // required: deemed_author (author || editor); year; fulltitle; auto biblio_entry_tags_jsonstr = `{ "is" : "", "sortby_deemed_author_year_title" : "", diff --git a/src/sdp/ao_header_extract.d b/src/sdp/ao_header_extract.d new file mode 100644 index 0000000..c02bc7c --- /dev/null +++ b/src/sdp/ao_header_extract.d @@ -0,0 +1,1307 @@ +/+ + extract sdl header return sdl ++/ +template SiSUheaderExtractSDLang() { + private import + std.regex; + private import + ao_rgx; + struct HeaderExtractSDL { + mixin RgxInit; + auto rgx = Rgx(); + final private auto headerMakeSDLang(in string src_header) { + 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); + } + catch(SDLangParseException 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(sdl_root_header.toSDLDocument()); + } + return sdl_root_header; + } + private auto headerSDLangToJSON(in char[] src_header) { + char[][] source_header_arr = + split(cast(char[]) src_header, rgx.line_delimiter); + char[] header_clean; + // TODO + foreach(header_line; source_header_arr) { + if (!match(header_line, rgx.comments)) { + header_clean ~= header_line ~ "\n"; + // writeln(header_line); + } + } + /+ get sdlang tags +/ + auto header_sdlang=headerMakeSDLang(to!string(header_clean)); + debug(sdlang) { + writeln("--------------"); + stdout.rawWrite( header_sdlang.toSDLDocument() ); + writeln("--------------"); + Value test = header_sdlang.tags["title"][0].values[0]; + assert(test == typeid(string)); + writeln(header_sdlang.maybe.tags["title"]); + writeln(header_sdlang.maybe.tags["title"][0].maybe.attributes["subtitle"]); + } + /+ TODO separate modules? sdlang then json + return header_sdlang; + // then do json module? + +/ + /+ TODO json next +/ + /+ TODO + auto dochead_make = parseJSON(header_make_jsonstr).object; + auto dochead_meta = parseJSON(header_meta_jsonstr).object; + auto hm = to!string(t.captures[1]); + auto hs = to!string(s.captures[1]); + if (dochead_make[hm][hs].type() == JSON_TYPE.STRING) { + dochead_make[hm][hs].str = to!string(s.captures[2]); + } + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[hm][hs].str = to!string(s.captures[2]); + } + +/ + /+ dochead +/ + string hm; + string hs; + /+ make +/ + auto dochead_make = parseJSON(header_make_jsonstr).object; + if (!(header_sdlang.maybe.tags["make"].empty)) { + hm = "make"; + hs = "bold"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "breaks"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "cover_image"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "css"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "emphasis"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "footer"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "headings"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "home_button_image"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "home_button_text"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "italics"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "num_top"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "substitute"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "texpdf_font"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_make[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + /+ meta +/ + auto dochead_meta = parseJSON(header_meta_jsonstr).object; + hm = "title"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + /+ TODO Title REQUIRED +/ + hs = "main"; + if (!(header_sdlang.tags[hm].empty) + && (header_sdlang.tags[hm][0].values[0].length > 1)) { + writeln(header_sdlang.tags[hm][0].values[0]); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].values[0]); + // to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } else if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } else { + writeln("Required header metadata Title, missing"); + } + hs = "sub"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } else if (!(header_sdlang.tags[hm][0].maybe.attributes["subtitle"].empty) + && (header_sdlang.tags[hm][0].attributes["subtitle"][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes["subtitle"][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes["subtitle"][0].value); + } + // full title composite (main + sub) + // hs = "full"; + // dochead_meta[hm][hs] = dochead_meta[hm]["main"] ~ dochead_meta[hm]["sub"]; + hs = "language"; + if (!(header_sdlang.tags[hm][0].maybe.attributes["lang"].empty) + && (header_sdlang.tags[hm][0].attributes["lang"][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes["lang"][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes["lang"][0].value); + } else if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "edition"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "note"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "creator"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + /+ Creator / Author REQUIRED +/ + /+ TODO + - decide on representation for & deal with multiple authors; + - author(s) full name; + - author(s) surname & other parts + +/ + hs = "author"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "author_email"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "illustrator"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "translator"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "classify"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "dewey"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "keywords"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "loc"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "subject"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "topic_register"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "date"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "added_to_site"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "available"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "created"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "issued"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "modified"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "published"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "valid"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "identifier"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "isbn"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "oclc"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "pg"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "links"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + /+ TODO + stuff to fix + +/ + // hs = "link"; + // if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + // && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + // writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + // dochead_meta[hm][hs].str = + // to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + // } + } + hm = "notes"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "abstract"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "description"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "original"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "language"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "source"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "title"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "publisher"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "name"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + hm = "rights"; + if (!(header_sdlang.maybe.tags[hm].empty)) { + hs = "copyright"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "cover"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "illustrations"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + hs = "license"; + if (!(header_sdlang.tags[hm][0].maybe.attributes[hs].empty) + && (header_sdlang.tags[hm][0].attributes[hs][0].value.length > 1)) { + writeln(header_sdlang.tags[hm][0].attributes[hs][0].value); + dochead_meta[hm][hs].str = + to!string(header_sdlang.tags[hm][0].attributes[hs][0].value); + } + } + // writeln(dochead_make); + // writeln(dochead_meta); + auto t = tuple(dochead_make, dochead_meta); + static assert(!isTypeTuple!(t)); + return t; + } + private auto headerSDLangToHash(in char[] src_header) { + char[][] source_header_arr = + split(cast(char[]) src_header, rgx.line_delimiter); + char[] header_clean; + string[string] header_make; + string[string][string] header_metadata; + foreach(header_line; source_header_arr) { + if (!match(header_line, rgx.comments)) { + header_clean ~= header_line ~ "\n"; + // writeln(header_line); + } + } + /+ get sdlang tags +/ + auto header_sdlang=headerMakeSDLang(to!string(header_clean)); + Value test = header_sdlang.tags["title"][0].values[0]; + // assert(test == typeid(string)); + // writeln(header_sdlang.maybe.tags["title"]); + // writeln(header_sdlang.maybe.tags["title"][0].attributes["subtitle"]); + if (!(header_sdlang.maybe.tags["make"].empty)) { + if (!(header_sdlang.tags["make"][0].maybe.attributes["bold"].empty) + && (header_sdlang.tags["make"][0].attributes["bold"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["bold"][0].value); + header_make["bold"] = + to!string(header_sdlang.tags["make"][0].attributes["bold"][0].value); + } else { + header_make["bold"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["breaks"].empty) + && (header_sdlang.tags["make"][0].attributes["breaks"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["breaks"][0].value); + header_make["breaks"] = + to!string(header_sdlang.tags["make"][0].attributes["breaks"][0].value); + } else { + header_make["breaks"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["cover_image"].empty) + && (header_sdlang.tags["make"][0].attributes["cover_image"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["cover_image"][0].value); + header_make["cover_image"] = + to!string(header_sdlang.tags["make"][0].attributes["cover_image"][0].value); + } else { + header_make["cover_image"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["css"].empty) + && (header_sdlang.tags["make"][0].attributes["css"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["css"][0].value); + header_make["css"] = + to!string(header_sdlang.tags["make"][0].attributes["css"][0].value); + } else { + header_make["css"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["emphasis"].empty) + && (header_sdlang.tags["make"][0].attributes["emphasis"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["emphasis"][0].value); + header_make["emphasis"] = + to!string(header_sdlang.tags["make"][0].attributes["emphasis"][0].value); + } else { + header_make["emphasis"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["footer"].empty) + && (header_sdlang.tags["make"][0].attributes["footer"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["footer"][0].value); + header_make["footer"] = + to!string(header_sdlang.tags["make"][0].attributes["footer"][0].value); + } else { + header_make["footer"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["headings"].empty) + && (header_sdlang.tags["make"][0].attributes["headings"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["headings"][0].value); + header_make["headings"] = + to!string(header_sdlang.tags["make"][0].attributes["headings"][0].value); + } else { + header_make["headings"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["home_button_image"].empty) + && (header_sdlang.tags["make"][0].attributes["home_button_image"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["home_button_image"][0].value); + header_make["home_button_image"] = + to!string(header_sdlang.tags["make"][0].attributes["home_button_image"][0].value); + } else { + header_make["home_button_image"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["home_button_text"].empty) + && (header_sdlang.tags["make"][0].attributes["home_button_text"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["home_button_text"][0].value); + header_make["home_button_text"] = + to!string(header_sdlang.tags["make"][0].attributes["home_button_text"][0].value); + } else { + header_make["home_button_text"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["italics"].empty) + && (header_sdlang.tags["make"][0].attributes["italics"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["italics"][0].value); + header_make["italics"] = + to!string(header_sdlang.tags["make"][0].attributes["italics"][0].value); + } else { + header_make["italics"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["num_top"].empty) + && (header_sdlang.tags["make"][0].attributes["num_top"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["num_top"][0].value); + header_make["num_top"] = + to!string(header_sdlang.tags["make"][0].attributes["num_top"][0].value); + } else { + header_make["num_top"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["substitute"].empty) + && (header_sdlang.tags["make"][0].attributes["substitute"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["substitute"][0].value); + header_make["substitute"] = + to!string(header_sdlang.tags["make"][0].attributes["substitute"][0].value); + } else { + header_make["substitute"] = ""; + } + if (!(header_sdlang.tags["make"][0].maybe.attributes["texpdf_font"].empty) + && (header_sdlang.tags["make"][0].attributes["texpdf_font"][0].value.length > 1)) { + writeln(header_sdlang.tags["make"][0].attributes["texpdf_font"][0].value); + header_make["texpdf_font"] = + to!string(header_sdlang.tags["make"][0].attributes["texpdf_font"][0].value); + } else { + header_make["texpdf_font"] = ""; + } + } + if (!(header_sdlang.maybe.tags["title"].empty)) { + /+ Title REQUIRED +/ + if (!(header_sdlang.tags["title"].empty) + && (header_sdlang.tags["title"][0].values[0].length > 1)) { + writeln(header_sdlang.tags["title"][0].values[0]); + header_metadata["title"]["main"] = + to!string(header_sdlang.tags["title"][0].values[0]); + } else if (!(header_sdlang.tags["title"][0].maybe.attributes["main"].empty) + && (header_sdlang.tags["title"][0].attributes["main"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["main"][0].value); + header_metadata["title"]["main"] = + to!string(header_sdlang.tags["title"][0].attributes["main"][0].value); + } else { + writeln("Required header metadata Title, missing"); + header_metadata["title"]["main"] = ""; + } + if (!(header_sdlang.tags["title"][0].maybe.attributes["sub"].empty) + && (header_sdlang.tags["title"][0].attributes["sub"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["sub"][0].value); + header_metadata["title"]["sub"] = + to!string(header_sdlang.tags["title"][0].attributes["sub"][0].value); + } else if (!(header_sdlang.tags["title"][0].maybe.attributes["subtitle"].empty) + && (header_sdlang.tags["title"][0].attributes["subtitle"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["subtitle"][0].value); + header_metadata["title"]["sub"] = + to!string(header_sdlang.tags["title"][0].attributes["subtitle"][0].value); + } else { + header_metadata["title"]["sub"] = ""; + } + // full title composite (main + sub) + header_metadata["title"]["full"] = header_metadata["title"]["main"] ~ header_metadata["title"]["sub"]; + if (!(header_sdlang.tags["title"][0].maybe.attributes["lang"].empty) + && (header_sdlang.tags["title"][0].attributes["lang"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["lang"][0].value); + header_metadata["title"]["language"] = + to!string(header_sdlang.tags["title"][0].attributes["lang"][0].value); + } else if (!(header_sdlang.tags["title"][0].maybe.attributes["language"].empty) + && (header_sdlang.tags["title"][0].attributes["language"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["language"][0].value); + header_metadata["title"]["language"] = + to!string(header_sdlang.tags["title"][0].attributes["language"][0].value); + } else { + header_metadata["title"]["language"] = ""; + } + if (!(header_sdlang.tags["title"][0].maybe.attributes["edition"].empty) + && (header_sdlang.tags["title"][0].attributes["edition"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["edition"][0].value); + header_metadata["title"]["edition"] = + to!string(header_sdlang.tags["title"][0].attributes["edition"][0].value); + } else { + header_metadata["title"]["edition"] = ""; + } + if (!(header_sdlang.tags["title"][0].maybe.attributes["note"].empty) + && (header_sdlang.tags["title"][0].attributes["note"][0].value.length > 1)) { + writeln(header_sdlang.tags["title"][0].attributes["note"][0].value); + header_metadata["title"]["note"] = + to!string(header_sdlang.tags["title"][0].attributes["note"][0].value); + } else { + header_metadata["title"]["note"] = ""; + } + } + if (!(header_sdlang.maybe.tags["creator"].empty)) { + /+ Creator / Author REQUIRED +/ + /+ TODO + - decide on representation for & deal with multiple authors; + - author(s) full name; + - author(s) surname & other parts + +/ + if (!(header_sdlang.tags["creator"][0].maybe.attributes["author"].empty) + && (header_sdlang.tags["creator"][0].attributes["author"][0].value.length > 1)) { + writeln(header_sdlang.tags["creator"][0].attributes["author"][0].value); + header_metadata["creator"]["author"] = + to!string(header_sdlang.tags["creator"][0].attributes["author"][0].value); + } else { + header_metadata["creator"]["author"] = ""; + } + if (!(header_sdlang.tags["creator"][0].maybe.attributes["author_email"].empty) + && (header_sdlang.tags["creator"][0].attributes["author_email"][0].value.length > 1)) { + writeln(header_sdlang.tags["creator"][0].attributes["author_email"][0].value); + header_metadata["creator"]["author_email"] = + to!string(header_sdlang.tags["creator"][0].attributes["author_email"][0].value); + } else { + header_metadata["creator"]["author_email"] = ""; + } + if (!(header_sdlang.tags["creator"][0].maybe.attributes["illustrator"].empty) + && (header_sdlang.tags["creator"][0].attributes["illustrator"][0].value.length > 1)) { + writeln(header_sdlang.tags["creator"][0].attributes["illustrator"][0].value); + header_metadata["creator"][""] = + to!string(header_sdlang.tags["creator"][0].attributes["illustrator"][0].value); + } else { + header_metadata["creator"]["illustrator"] = ""; + } + if (!(header_sdlang.tags["creator"][0].maybe.attributes["translator"].empty) + && (header_sdlang.tags["creator"][0].attributes["translator"][0].value.length > 1)) { + writeln(header_sdlang.tags["creator"][0].attributes["translator"][0].value); + header_metadata["creator"]["translator"] = + to!string(header_sdlang.tags["creator"][0].attributes["translator"][0].value); + } else { + header_metadata["creator"][""] = ""; + } + } + if (!(header_sdlang.maybe.tags["classify"].empty)) { + if (!(header_sdlang.tags["classify"][0].maybe.attributes["dewey"].empty) + && (header_sdlang.tags["classify"][0].attributes["dewey"][0].value.length > 1)) { + writeln(header_sdlang.tags["classify"][0].attributes["dewey"][0].value); + header_metadata["classify"]["dewey"] = + to!string(header_sdlang.tags["classify"][0].attributes["dewey"][0].value); + } else { + header_metadata["classify"]["dewey"] = ""; + } + if (!(header_sdlang.tags["classify"][0].maybe.attributes["keyword"].empty) + && (header_sdlang.tags["classify"][0].attributes["keyword"][0].value.length > 1)) { + writeln(header_sdlang.tags["classify"][0].attributes["keyword"][0].value); + header_metadata["classify"]["keyword"] = + to!string(header_sdlang.tags["classify"][0].attributes["keyword"][0].value); + } else { + header_metadata["classify"]["keyword"] = ""; + } + if (!(header_sdlang.tags["classify"][0].maybe.attributes["loc"].empty) + && (header_sdlang.tags["classify"][0].attributes["loc"][0].value.length > 1)) { + writeln(header_sdlang.tags["classify"][0].attributes["loc"][0].value); + header_metadata["classify"]["loc"] = + to!string(header_sdlang.tags["classify"][0].attributes["loc"][0].value); + } else { + header_metadata["classify"]["loc"] = ""; + } + if (!(header_sdlang.tags["classify"][0].maybe.attributes["subject"].empty) + && (header_sdlang.tags["classify"][0].attributes["subject"][0].value.length > 1)) { + writeln(header_sdlang.tags["classify"][0].attributes["subject"][0].value); + header_metadata["classify"]["subject"] = + to!string(header_sdlang.tags["classify"][0].attributes["subject"][0].value); + } else { + header_metadata["classify"]["subject"] = ""; + } + if (!(header_sdlang.tags["classify"][0].maybe.attributes["topic_register"].empty) + && (header_sdlang.tags["classify"][0].attributes["topic_register"][0].value.length > 1)) { + writeln(header_sdlang.tags["classify"][0].attributes["topic_register"][0].value); + header_metadata["classify"]["topic_register"] = + to!string(header_sdlang.tags["classify"][0].attributes["topic_register"][0].value); + } else { + header_metadata["classify"]["topic_register"] = ""; + } + } + if (!(header_sdlang.maybe.tags["date"].empty)) { + if (!(header_sdlang.tags["date"][0].maybe.attributes["added_to_site"].empty) + && (header_sdlang.tags["date"][0].attributes["added_to_site"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["added_to_site"][0].value); + header_metadata["date"]["added_to_site"] = + to!string(header_sdlang.tags["date"][0].attributes["added_to_site"][0].value); + } else { + header_metadata["date"]["added_to_site"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["available"].empty) + && (header_sdlang.tags["date"][0].attributes["available"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["available"][0].value); + header_metadata["date"]["available"] = + to!string(header_sdlang.tags["date"][0].attributes["available"][0].value); + } else { + header_metadata["date"]["available"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["created"].empty) + && (header_sdlang.tags["date"][0].attributes["created"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["created"][0].value); + header_metadata["date"]["created"] = + to!string(header_sdlang.tags["date"][0].attributes["created"][0].value); + } else { + header_metadata["date"]["created"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["issued"].empty) + && (header_sdlang.tags["date"][0].attributes["issued"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["issued"][0].value); + header_metadata["date"]["issued"] = + to!string(header_sdlang.tags["date"][0].attributes["issued"][0].value); + } else { + header_metadata["date"]["issued"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["modified"].empty) + && (header_sdlang.tags["date"][0].attributes["modified"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["modified"][0].value); + header_metadata["date"]["modified"] = + to!string(header_sdlang.tags["date"][0].attributes["modified"][0].value); + } else { + header_metadata["date"]["modified"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["published"].empty) + && (header_sdlang.tags["date"][0].attributes["published"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["published"][0].value); + header_metadata["date"]["published"] = + to!string(header_sdlang.tags["date"][0].attributes["published"][0].value); + } else { + header_metadata["date"]["published"] = ""; + } + if (!(header_sdlang.tags["date"][0].maybe.attributes["valid"].empty) + && (header_sdlang.tags["date"][0].attributes["valid"][0].value.length > 1)) { + writeln(header_sdlang.tags["date"][0].attributes["valid"][0].value); + header_metadata["date"]["valid"] = + to!string(header_sdlang.tags["date"][0].attributes["valid"][0].value); + } else { + header_metadata["date"]["valid"] = ""; + } + } + if (!(header_sdlang.maybe.tags["identifier"].empty)) { + if (!(header_sdlang.tags["identifier"][0].maybe.attributes["isbn"].empty) + && (header_sdlang.tags["identifier"][0].attributes["isbn"][0].value.length > 1)) { + writeln(header_sdlang.tags["identifier"][0].attributes["isbn"][0].value); + header_metadata["identifier"]["isbn"] = + to!string(header_sdlang.tags["identifier"][0].attributes["isbn"][0].value); + } else { + header_metadata["identifier"]["isbn"] = ""; + } + if (!(header_sdlang.tags["identifier"][0].maybe.attributes["oclc"].empty) + && (header_sdlang.tags["identifier"][0].attributes["oclc"][0].value.length > 1)) { + writeln(header_sdlang.tags["identifier"][0].attributes["oclc"][0].value); + header_metadata["identifier"]["oclc"] = + to!string(header_sdlang.tags["identifier"][0].attributes["oclc"][0].value); + } else { + header_metadata["identifier"]["oclc"] = ""; + } + if (!(header_sdlang.tags["identifier"][0].maybe.attributes["pg"].empty) + && (header_sdlang.tags["identifier"][0].attributes["pg"][0].value.length > 1)) { + writeln(header_sdlang.tags["identifier"][0].attributes["pg"][0].value); + header_metadata["identifier"]["pg"] = + to!string(header_sdlang.tags["identifier"][0].attributes["pg"][0].value); + } else { + header_metadata["identifier"]["pg"] = ""; + } + } + if (!(header_sdlang.maybe.tags["links"].empty)) { + /+ TODO + stuff to fix + +/ + if (!(header_sdlang.tags["links"][0].maybe.attributes["link"].empty) + && (header_sdlang.tags["links"][0].attributes["link"][0].value.length > 1)) { + writeln(header_sdlang.tags["links"][0].attributes["link"][0].value); + // header_metadata["links"]["list"] = + // to!string(header_sdlang.tags["links"][0].attributes["link"][0].value); + header_metadata["links"]["list"] = ""; + } else { + header_metadata["links"]["list"] = ""; + } + } + if (!(header_sdlang.maybe.tags["notes"].empty)) { + if (!(header_sdlang.tags["notes"][0].maybe.attributes["abstract"].empty) + && (header_sdlang.tags["notes"][0].attributes["abstract"][0].value.length > 1)) { + writeln(header_sdlang.tags["notes"][0].attributes["abstract"][0].value); + header_metadata["notes"]["abstract"] = + to!string(header_sdlang.tags["notes"][0].attributes["abstract"][0].value); + } else { + header_metadata["notes"]["abstract"] = ""; + } + if (!(header_sdlang.tags["notes"][0].maybe.attributes["description"].empty) + && (header_sdlang.tags["notes"][0].attributes["description"][0].value.length > 1)) { + writeln(header_sdlang.tags["notes"][0].attributes["description"][0].value); + header_metadata["notes"]["description"] = + to!string(header_sdlang.tags["notes"][0].attributes["description"][0].value); + } else { + header_metadata["notes"]["description"] = ""; + } + } + if (!(header_sdlang.maybe.tags["original"].empty)) { + if (!(header_sdlang.tags["original"][0].maybe.attributes["language"].empty) + && (header_sdlang.tags["original"][0].attributes["language"][0].value.length > 1)) { + writeln(header_sdlang.tags["original"][0].attributes["language"][0].value); + header_metadata["original"]["language"] = + to!string(header_sdlang.tags["original"][0].attributes["language"][0].value); + } else { + header_metadata["original"]["language"] = ""; + } + if (!(header_sdlang.tags["original"][0].maybe.attributes["source"].empty) + && (header_sdlang.tags["original"][0].attributes["source"][0].value.length > 1)) { + writeln(header_sdlang.tags["original"][0].attributes["source"][0].value); + header_metadata["original"]["source"] = + to!string(header_sdlang.tags["original"][0].attributes["source"][0].value); + } else { + header_metadata["original"]["source"] = ""; + } + if (!(header_sdlang.tags["original"][0].maybe.attributes["title"].empty) + && (header_sdlang.tags["original"][0].attributes["title"][0].value.length > 1)) { + writeln(header_sdlang.tags["original"][0].attributes["title"][0].value); + header_metadata["original"]["title"] = + to!string(header_sdlang.tags["original"][0].attributes["title"][0].value); + } else { + header_metadata["original"]["title"] = ""; + } + } + if (!(header_sdlang.maybe.tags["publisher"].empty)) { + if (!(header_sdlang.tags["publisher"][0].maybe.attributes["name"].empty) + && (header_sdlang.tags["publisher"][0].attributes["name"][0].value.length > 1)) { + writeln(header_sdlang.tags["publisher"][0].attributes["name"][0].value); + header_metadata["publisher"]["name"] = + to!string(header_sdlang.tags["publisher"][0].attributes["name"][0].value); + } else { + header_metadata["publisher"]["name"] = ""; + } + } + if (!(header_sdlang.maybe.tags["rights"].empty)) { + if (!(header_sdlang.tags["rights"][0].maybe.attributes["copyright"].empty) + && (header_sdlang.tags["rights"][0].attributes["copyright"][0].value.length > 1)) { + writeln(header_sdlang.tags["rights"][0].attributes["copyright"][0].value); + header_metadata["rights"]["copyright"] = + to!string(header_sdlang.tags["rights"][0].attributes["copyright"][0].value); + } else { + header_metadata["rights"]["copyright"] = ""; + } + if (!(header_sdlang.tags["rights"][0].maybe.attributes["cover"].empty) + && (header_sdlang.tags["rights"][0].attributes["cover"][0].value.length > 1)) { + writeln(header_sdlang.tags["rights"][0].attributes["cover"][0].value); + header_metadata["rights"]["cover"] = + to!string(header_sdlang.tags["rights"][0].attributes["cover"][0].value); + } else { + header_metadata["rights"]["cover"] = ""; + } + if (!(header_sdlang.tags["rights"][0].maybe.attributes["illustrations"].empty) + && (header_sdlang.tags["rights"][0].attributes["illustrations"][0].value.length > 1)) { + writeln(header_sdlang.tags["rights"][0].attributes["illustrations"][0].value); + header_metadata["rights"]["illustrations"] = + to!string(header_sdlang.tags["rights"][0].attributes["illustrations"][0].value); + } else { + header_metadata["rights"]["illustrations"] = ""; + } + if (!(header_sdlang.tags["rights"][0].maybe.attributes["license"].empty) + && (header_sdlang.tags["rights"][0].attributes["license"][0].value.length > 1)) { + writeln(header_sdlang.tags["rights"][0].attributes["license"][0].value); + header_metadata["rights"]["license"] = + to!string(header_sdlang.tags["rights"][0].attributes["license"][0].value); + } else { + header_metadata["rights"]["license"] = ""; + } + } + writeln("--------------"); + stdout.rawWrite( header_sdlang.toSDLDocument() ); + writeln("--------------"); + /+ + namespace:name values attributes { + children + } + +/ + return header_sdlang; + } + private auto headerSDLang(in char[] src_header) { + char[][] source_header_arr = + split(cast(char[]) src_header, rgx.line_delimiter); + char[] header_clean; + foreach(header_line; source_header_arr) { + if (!match(header_line, rgx.comments)) { + header_clean ~= header_line ~ "\n"; + // writeln(header_line); + } + } + // writeln(header_clean); // consider + auto header_sdlang=headerMakeSDLang(to!string(header_clean)); + return header_sdlang; + } + + } + struct HeaderUseSDL { + mixin RgxInit; + auto rgx = Rgx(); + // Tag = sdl_header; + auto headerUseSDLang(T)(auto T sdl_root_header) { + // T sdl_root_header; + // auto headerUseSDLang(Tag sdl_root_header) { + // private auto headerUseSDLang(Tag sdl_root_header) { + // private auto headerUseSDLang(in Tag sdl_root_header) { + // auto sdl_root_header = T; + + // Value is a std.variant.Algebraic + Value test = sdl_root_header.tags["title"][0].values[0]; + // assert(test == typeid(string)); + writeln(test); + + return sdl_root_header; + } + } +} +/+ + extract native/orig header return json ++/ +template SiSUheaderExtractNative() { + private import + std.exception, + std.regex, + std.utf, + std.conv : to; + private import + ao_rgx; + struct HeaderDocMetadataAndMakeNativeToJson { + mixin SiSUrgxInitFlags; + mixin RgxInit; + auto rgx = Rgx(); + enum State { off, on } + string hm, hs; + auto header_metadata_and_make_jsonstr( + string header, + JSONValue[string] dochead_meta, + JSONValue[string] dochead_make + ) + in { } + body { + scope(exit) { + destroy(header); + destroy(dochead_meta); + destroy(dochead_make); + } + if (auto t = match(header, rgx.native_header_main)) { + char[][] header_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; header_obj_spl) { + if (auto m = match(line, rgx.native_header_main)) { + if (!empty(m.captures[2])) { + if (hm == "creator") { + dochead_meta[hm]["author"].str = + to!string(m.captures[2]); + } else if (hm == "title") { + dochead_meta[hm]["main"].str = + to!string(m.captures[2]); + } else if (hm == "publisher") { + dochead_meta[hm]["name"].str = + 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].type() == JSON_TYPE.OBJECT)) { + switch (hm) { + case "make": + if (match(hs, rgx.native_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_meta[hm].type() == JSON_TYPE.OBJECT) { + switch (hm) { + case "creator": + if (match(hs, rgx.native_subhead_creator)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_title)) { + if ((hs == "subtitle") + && (dochead_meta[hm]["sub"].type() == JSON_TYPE.STRING)) { + dochead_meta[hm]["sub"].str = + to!string(s.captures[2]); + } else if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_rights)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_date)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_original)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_classify)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_identifier)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_notes)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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.native_subhead_publisher)) { + if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + dochead_meta[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); + // if (match(hs, rgx.native_subhead_links)) { + // if (dochead_meta[hm][hs].type() == JSON_TYPE.STRING) { + // dochead_meta[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 { + writeln("not a valid header type:", hm); + } + } + auto t = tuple(dochead_meta, dochead_make); + static assert(!isTypeTuple!(t)); + return t; + } + private auto native_header_extract( + char[] line, + ref int[string] line_occur, + ref string[string] an_object, + ref int[string] type + ) { + if (matchFirst(line, 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["obj"] ~= line ~= "\n"; + } else if (matchFirst(line, 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["obj"] ~= line ~= "\n"; + } else if (type["header_make"] == State.on + && (line_occur["header_make"] > State.off)) { /+ header_make flag set +/ + if (matchFirst(line, rgx.native_header_sub)) { /+ sub-header +/ + debug(header1) { /+ writeln(line); +/ } + ++line_occur["header_make"]; + an_object["obj"] ~= line ~= "\n"; + } + } else if (type["header_meta"] == State.on + && (line_occur["header_meta"] > State.off)) { /+ header_metadata flag set +/ + if (matchFirst(line, rgx.native_header_sub)) { /+ sub-header +/ + debug(header1) { /+ writeln(line); +/ } + ++line_occur["header_meta"]; + an_object["obj"] ~= line ~= "\n"; + } + } + return an_object; + } + auto header_reset_states_common( + ref int[string] line_occur, + ref string[string] an_object, + ref int[string] type + ) { + // line_occur["header"] = State.off; + line_occur["header_make"] = State.off; + line_occur["header_meta"] = State.off; + type["header"] = State.off; + // type["header_make"] = State.off; + // type["header_meta"] = State.off; + an_object.remove("obj"); + an_object.remove("is"); + an_object.remove("attrib"); + } + private auto headerContentJSON(in char[] src_header) { + 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 = parseJSON(header_make_jsonstr).object; + auto dochead_meta = parseJSON(header_meta_jsonstr).object; + auto set_header = HeaderDocMetadataAndMakeNativeToJson(); + char[][] source_header_arr = + split(cast(char[]) src_header, rgx.line_delimiter); + foreach(header_line; source_header_arr) { + if (auto m = matchFirst(header_line, rgx.comment)) { + /+ matched comment +/ + debug(comment) { + // tell_l("blue", header_line); + } + header_reset_states_common(line_occur, an_object, type); + // type["header_make"] = State.off; + // type["header_meta"] = State.off; + } else if ((matchFirst(header_line, 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_jsonstr(strip(an_object["obj"]), 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; + writeln(dochead_metadata_and_make); + } else { + an_object = native_header_extract(header_line, line_occur, an_object, type); + } + } else { + // writeln(__LINE__); + } + } + auto t = tuple( + dochead_make, + dochead_meta, + ); + return t; + } + } +} diff --git a/src/sdp/ao_read_source_files.d b/src/sdp/ao_read_source_files.d index ef9b8b4..4d766b2 100644 --- a/src/sdp/ao_read_source_files.d +++ b/src/sdp/ao_read_source_files.d @@ -36,59 +36,6 @@ template SiSUmarkupRaw() { } } private - struct HeaderExtractSDL { - final private auto headerMakeSDLang(in string src_header) { - 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); - } - catch(SDLangParseException 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) { - // // Value is a std.variant.Algebraic - // Value output_dir_structure_by = sdl_root_header.tags["output_dir_structure_by"][0].values[0]; - // assert(output_dir_structure_by.type == typeid(string)); - // writeln(output_dir_structure_by); - - // Tag person = sdl_root_header.namespaces["myNamespace"].tags["person"][0]; - // writeln("Name: ", person.attributes["name"][0].value); - // - // int age = person.tags["age"][0].values[0].get!int(); - // writeln("Age: ", age); - - writeln("header SDL:"); - writeln(sdl_root_header.toSDLDocument()); - } - return sdl_root_header; - } - private auto headerSDLang(in char[] src_header) { - char[][] source_header_arr = - split(cast(char[]) src_header, rgx.line_delimiter); - char[] header_clean; - foreach(header_line; source_header_arr) { - if (!match(header_line, rgx.comments)) { - header_clean ~= header_line ~ "\n"; - // writeln(header_line); - } - } - // writeln(header_clean); // consider - auto header_sdlang=headerMakeSDLang(to!string(header_clean)); - return header_sdlang; - } - } struct MarkupRawUnit { private import std.file; // enum State { off, on } @@ -205,14 +152,14 @@ template SiSUmarkupRaw() { contents_insert ~= line; } else if ( (type1["header_make"] == 1) - && matchFirst(line, rgx.header_sub) + && matchFirst(line, rgx.native_header_sub) ) { type1["header_make"] = 1; type1["header_meta"] = 0; // cont_dynamic_array ~= "% " ~ line; } else if ( (type1["header_meta"] == 1) - && matchFirst(line, rgx.header_sub) + && matchFirst(line, rgx.native_header_sub) ) { type1["header_meta"] = 1; type1["header_make"] = 0; diff --git a/src/sdp/ao_rgx.d b/src/sdp/ao_rgx.d index 1245a71..4e282ee 100644 --- a/src/sdp/ao_rgx.d +++ b/src/sdp/ao_rgx.d @@ -32,18 +32,36 @@ template RgxInit() { static src_fn_find_inserts = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`); static insert_src_fn_ssi_or_sst = ctRegex!(`^<<\s*(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`); // static insert_ssi_or_sst_fn = ctRegex!(`^<<\s*[a-zA-Z0-9._-]+[.]ss[ti]`); + // static ssm_fn = ctRegex!(`^[a-zA-Z0-9._-]+[.]ssm$`); + /+ insert markup file +/ + // static insert_src_fn_ssi_or_sst = ctRegex!(`^<<\s*([a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`); /+ comments +/ static comment = ctRegex!(`^%+ `); static comments = ctRegex!(`^%+ |^%+$`); /+ header +/ - static header = ctRegex!(`^@([a-z_]+):(?:\s|$)`); - static header_make = ctRegex!(`^@(make):(?:\s|$)`); - static header_meta = ctRegex!(`^@([a-z_]+):(?:\s|$)`); - static header_sub = ctRegex!(`^[ ]+:([a-z_]+):\s`); - static head_main = ctRegex!(`^@(?P
[a-z_]+):\s*(?P.*)`, "m"); - static head_sub = ctRegex!(`^[ ]*:(?P[a-z_]+):\s+(?P.+)`, "m"); - static head_value_title = ctRegex!(`@title`); - static head_value_author = ctRegex!(`@author`); + 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_sub = ctRegex!(`^[ ]+:([a-z_]+):\s`); + static native_header_meta_title = ctRegex!(`^@title:\s`, "m"); + static variable_doc_title = ctRegex!(`@title`); + static variable_doc_author = ctRegex!(`@author`); + /+ 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)$`, "m"); + static native_subhead_make = ctRegex!(`^(?:cover_image|home_button_image|home_button_text|footer|headings|num_top|breaks|substitute|bold|italics|emphasis|texpdf_font|css)$`, "m"); /+ heading & paragraph operators +/ static heading_a = ctRegex!(`^:?[A][~] `, "m"); static heading = ctRegex!(`^:?([A-D1-4])[~]([a-z0-9_.-]*[?]?) `); @@ -160,18 +178,6 @@ template RgxInit() { static parent = ctRegex!(`([0-7]):([0-9]+)`); /+ json +/ static tailing_comma = ctRegex!(`,$`, "m"); - /+ head +/ - static main_headers = ctRegex!(`^(?:creator|title|rights|date|original|classify|identifier|notes|publisher|make|links)$`, "m"); - static subhead_creator = ctRegex!(`^(?:author|translator|illustrator)$`, "m"); - static subhead_title = ctRegex!(`^(?:main|sub(?:title)?|full|language|edition|note)$`, "m"); - static subhead_rights = ctRegex!(`^(?:copyright|illustrations|license|cover)$`, "m"); - static subhead_date = ctRegex!(`^(?:published|created|issued|available|valid|modified|added_to_site)$`, "m"); - static subhead_original = ctRegex!(`^(?:title|language|source)$`, "m"); - static subhead_classify = ctRegex!(`^(?:topic_register|subject|keywords|loc|dewey)$`, "m"); - static subhead_identifier = ctRegex!(`^(?:oclc|pg|isbn)$`, "m"); - static subhead_notes = ctRegex!(`^(?:abstract|description)$`, "m"); - static subhead_publisher = ctRegex!(`^(?:name)$`, "m"); - static subhead_make = ctRegex!(`^(?:cover_image|home_button_image|home_button_text|footer|headings|num_top|breaks|substitute|bold|italics|emphasis|texpdf_font|css)$`, "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)$`); -- cgit v1.2.3