From 6f27312b0e61932d820b991a15c44845ff2cee75 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Sun, 26 Feb 2017 18:27:08 -0500 Subject: 0.13.5 defaults regex reorganised; some work on (x)html output --- src/sdp/ao_abstract_doc_source.d | 157 +++++++++----------- src/sdp/ao_conf_make_meta_sdlang.d | 2 +- src/sdp/ao_rgx.d | 50 +++---- src/sdp/output_epub.d | 66 ++++++--- src/sdp/output_html.d | 289 ++++++++++++++++++++----------------- src/sdp/output_hub.d | 4 +- src/sdp/output_rgx.d | 37 +++++ src/sdp/output_xhtmls.d | 212 ++++++++++++++++++++++++++- src/sdp/source_sisupod.d | 8 +- 9 files changed, 548 insertions(+), 277 deletions(-) create mode 100644 src/sdp/output_rgx.d (limited to 'src/sdp') diff --git a/src/sdp/ao_abstract_doc_source.d b/src/sdp/ao_abstract_doc_source.d index ef64f0b..6690a9f 100644 --- a/src/sdp/ao_abstract_doc_source.d +++ b/src/sdp/ao_abstract_doc_source.d @@ -362,7 +362,7 @@ template SiSUdocAbstraction() { line, ); } - line = replaceAll(line, rgx.true_dollar, "$$$$"); + line = (line).replaceAll(rgx.true_dollar, "$$$$"); // dollar represented as $$ needed to stop submatching on $ // (substitutions using ${identifiers} must take into account (i.e. happen earlier)) debug(source) { // source lines @@ -1965,7 +1965,7 @@ template SiSUdocAbstraction() { tmp ~= au.str; } } - tmp = replace(tmp, rgx.trailing_comma, ""); + tmp = (tmp).replace(rgx.trailing_comma, ""); j["author"].str = tmp; goto default; case "editor_raw": // editor_arr editor (fn sn) @@ -1980,7 +1980,7 @@ template SiSUdocAbstraction() { tmp ~= ed.str; } } - tmp = replace(tmp, rgx.trailing_comma, ""); + tmp = (tmp).replace(rgx.trailing_comma, ""); j["editor"].str = tmp; goto default; case "fulltitle": // title & subtitle @@ -2876,10 +2876,11 @@ template SiSUdocAbstraction() { assertions_doc_structure(an_object, lv); // includes most of the logic for collapsed levels switch (an_object["lev"]) { case "A": - an_object[an_object_key]=replaceFirst(an_object[an_object_key], - rgx.variable_doc_title, (dochead_meta_aa["title"]["full"] ~ ",")); - an_object[an_object_key]=replaceFirst(an_object[an_object_key], - rgx.variable_doc_author, dochead_meta_aa["creator"]["author"]); + an_object[an_object_key]=(an_object[an_object_key]) + .replaceFirst(rgx.variable_doc_title, + (dochead_meta_aa["title"]["full"] ~ ",")) + .replaceFirst(rgx.variable_doc_author, + dochead_meta_aa["creator"]["author"]); collapsed_lev["h0"] = 0; an_object["lev_collapsed_number"] = to!string(collapsed_lev["h0"]); @@ -2994,7 +2995,7 @@ template SiSUdocAbstraction() { } } } - auto _para_match_(L,O,K,I,B,T,C)( + void _para_match_(L,O,K,I,B,T,C)( return ref L line, return ref O an_object, return ref K an_object_key, @@ -3089,17 +3090,15 @@ template SiSUdocAbstraction() { n_foot_sp_asterisk = 0; n_foot_sp_plus = 0; } - string url_links(Ot)(Ot obj_txt_in, string suffix = ".html") { + string url_links(Ot)(Ot obj_txt_in) { debug(asserts){ static assert(is(typeof(obj_txt_in) == string)); } /+ url matched +/ - if (match(obj_txt_in, rgx.inline_url)) { + if (match(obj_txt_in, rgx.inline_url_generic)) { /+ link: naked url: http://url +/ if (match(obj_txt_in, rgx.inline_link_naked_url)) { - obj_txt_in = - replaceAll( - obj_txt_in, + obj_txt_in = (obj_txt_in).replaceAll( rgx.inline_link_naked_url, ("$1" ~ mkup.lnk_o ~ "$2" ~ mkup.lnk_c @@ -3113,40 +3112,34 @@ template SiSUdocAbstraction() { { link which includes url as footnote }http://url~{ { http://url }http://url }~ +/ if (match(obj_txt_in, rgx.inline_link_endnote_url_helper)) { - obj_txt_in = - replaceAll( - obj_txt_in, - rgx.inline_link_endnote_url_helper_punctuated, - (mkup.lnk_o ~ "$1" ~ mkup.lnk_c - ~ mkup.url_o ~ "$2" ~ mkup.url_c - ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c - ~ mkup.url_o ~ "$2" ~ mkup.url_c - ~ " }~$3") // ("{ $1 }$2~{ { $2 }$2 }~$3") - ); - obj_txt_in = - replaceAll( - obj_txt_in, - rgx.inline_link_endnote_url_helper, - (mkup.lnk_o ~ "$1" ~ mkup.lnk_c - ~ mkup.url_o ~ "$2" ~ mkup.url_c - ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c - ~ mkup.url_o ~ "$2" ~ mkup.url_c - ~ " }~") // ("{ $1 }$2~{ { $2 }$2 }~") - ); + obj_txt_in = (obj_txt_in).replaceAll( + rgx.inline_link_endnote_url_helper_punctuated, + (mkup.lnk_o ~ "$1" ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ " }~$3") // ("{ $1 }$2~{ { $2 }$2 }~$3") + ); + obj_txt_in = (obj_txt_in).replaceAll( + rgx.inline_link_endnote_url_helper, + (mkup.lnk_o ~ "$1" ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ "~{ " ~ mkup.lnk_o ~ " $2 " ~ mkup.lnk_c + ~ mkup.url_o ~ "$2" ~ mkup.url_c + ~ " }~") // ("{ $1 }$2~{ { $2 }$2 }~") + ); } /+ link with regular markup: { linked text or image }http://url +/ if (match(obj_txt_in, rgx.inline_link_markup_regular)) { - obj_txt_in = - replaceAll( - obj_txt_in, - rgx.inline_link_markup_regular, - ("$1" - ~ mkup.lnk_o ~ "$2" ~ mkup.lnk_c - ~ mkup.url_o ~ "$3" ~ mkup.url_c - ~ "$4") // ("$1{ $2 }$3$4") - ); + obj_txt_in = (obj_txt_in).replaceAll( + rgx.inline_link_markup_regular, + ("$1" + ~ mkup.lnk_o ~ "$2" ~ mkup.lnk_c + ~ mkup.url_o ~ "$3" ~ mkup.url_c + ~ "$4") // ("$1{ $2 }$3$4") + ); } } return obj_txt_in; @@ -3158,12 +3151,10 @@ template SiSUdocAbstraction() { /+ endnotes (regular) +/ bool flg_notes_reg = false; bool flg_notes_star = false; - obj_txt_in = - replaceAll( - obj_txt_in, - rgx.inline_notes_curly, - (mkup.en_a_o ~ " $1" ~ mkup.en_a_c) - ); + obj_txt_in = (obj_txt_in).replaceAll( + rgx.inline_notes_curly, + (mkup.en_a_o ~ " $1" ~ mkup.en_a_c) + ); if (!(stage_reset_note_numbers) && reset_note_numbers) { stage_reset_note_numbers = true; } @@ -3182,22 +3173,18 @@ template SiSUdocAbstraction() { ++n_foot_sp_asterisk; asterisks_ = "*"; n_foot=n_foot_sp_asterisk; - obj_txt_out ~= - (replaceFirst( - to!string(n.hit), - rgx.inline_al_delimiter_open_symbol_star, - (mkup.en_a_o ~ replicate(asterisks_, n_foot_sp_asterisk) ~ " ") - ) ~ "\n"); + obj_txt_out ~= (to!string(n.hit)).replaceFirst( + rgx.inline_al_delimiter_open_symbol_star, + (mkup.en_a_o ~ replicate(asterisks_, n_foot_sp_asterisk) ~ " ") + ) ~ "\n"; } else if (match(to!string(n.hit), rgx.inline_al_delimiter_open_regular)) { flg_notes_reg = true; ++n_foot_reg; n_foot=n_foot_reg; - obj_txt_out ~= - (replaceFirst( - to!string(n.hit), - rgx.inline_al_delimiter_open_regular, - (mkup.en_a_o ~ to!string(n_foot) ~ " ") - ) ~ "\n"); + obj_txt_out ~= (to!string(n.hit)).replaceFirst( + rgx.inline_al_delimiter_open_regular, + (mkup.en_a_o ~ to!string(n_foot) ~ " ") + ) ~ "\n"; } else { obj_txt_out ~= to!string(n.hit) ~ "\n"; } @@ -3276,9 +3263,9 @@ template SiSUdocAbstraction() { } } body { - obj_txt["munge"]=obj_txt_in; - obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.para_attribs, ""); - obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.obj_cite_number_off_all, ""); + obj_txt["munge"]=(obj_txt_in) + .replaceFirst(rgx.para_attribs, "") + .replaceFirst(rgx.obj_cite_number_off_all, ""); auto t = object_notes_and_links_(obj_txt["munge"]); debug(munge) { writeln(__LINE__); @@ -3295,10 +3282,10 @@ template SiSUdocAbstraction() { } } body { - obj_txt["munge"]=obj_txt_in; - obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.heading, ""); - obj_txt["munge"]=replaceFirst(obj_txt["munge"], rgx.obj_cite_number_off_all, ""); - obj_txt["munge"]=strip(obj_txt["munge"]); + obj_txt["munge"]=(obj_txt_in) + .replaceFirst(rgx.heading, "") + .replaceFirst(rgx.obj_cite_number_off_all, "") + .strip; auto t = object_notes_and_links_(obj_txt["munge"], reset_note_numbers); debug(munge) { writeln(__LINE__); @@ -3310,7 +3297,6 @@ template SiSUdocAbstraction() { } invariant() { } - /+ revisit +/ string code(Ot)(Ot obj_txt_in) in { debug(asserts){ @@ -3490,12 +3476,9 @@ template SiSUdocAbstraction() { static assert(is(typeof(heading_toc_) == char[])); } auto m = matchFirst(cast(char[]) heading_toc_, rgx.heading); - heading_toc_ = - replaceAll( - m.post, - rgx.inline_notes_curly_gen, - ("") - ); + heading_toc_ = (m.post).replaceAll( + rgx.inline_notes_curly_gen, + ""); return heading_toc_; }; auto table_of_contents_gather_headings(O,Ma,Ts,Ta,X,Toc)( @@ -3769,10 +3752,11 @@ template SiSUdocAbstraction() { writeln(heading_number_auto_composite); } if (!empty(heading_number_auto_composite)) { - munge_=replaceFirst(munge_, rgx.heading, - "$1~$2 " ~ heading_number_auto_composite ~ ". "); - munge_=replaceFirst(munge_, rgx.heading_marker_missing_tag, - "$1~" ~ heading_number_auto_composite ~ " "); + munge_=(munge_) + .replaceFirst(rgx.heading, + "$1~$2 " ~ heading_number_auto_composite ~ ". ") + .replaceFirst(rgx.heading_marker_missing_tag, + "$1~" ~ heading_number_auto_composite ~ " "); } } } @@ -3788,16 +3772,19 @@ template SiSUdocAbstraction() { if (!(match(munge_, rgx.heading_anchor_tag))) { // if (anchor_tags_.length == 0) { if (match(munge_, rgx.heading_identify_anchor_tag)) { if (auto m = match(munge_, rgx.heading_extract_named_anchor_tag)) { - munge_=replaceFirst(munge_, rgx.heading_marker_missing_tag, + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, "$1~" ~ toLower(m.captures[1]) ~ "_" ~ m.captures[2] ~ " "); } else if (auto m = match(munge_, rgx.heading_extract_unnamed_anchor_tag)) { - munge_=replaceFirst(munge_, rgx.heading_marker_missing_tag, + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, "$1~" ~ "s" ~ m.captures[1] ~ " "); } } else if (lev_ == "1") { // (if not successful) manufacture a unique anchor tag for lev=="1" static __gshared int heading_num_lev1 = 0; heading_num_lev1 ++; - munge_=replaceFirst(munge_, rgx.heading_marker_missing_tag, + munge_=(munge_).replaceFirst( + rgx.heading_marker_missing_tag, "$1~" ~ "x" ~ to!string(heading_num_lev1) ~ " "); } } @@ -4233,7 +4220,7 @@ template SiSUdocAbstraction() { foreach (mainkey; mainkeys) { write("_0_1 !{", mainkey, "}! "); foreach (ref_; bookindex_unordered_hashes[mainkey]["_a"]) { - auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + auto go = (ref_).replaceAll(rgx.book_index_go, "$1"); write(" {", ref_, "}#", go, ", "); } writeln(" \\\\"); @@ -4243,7 +4230,7 @@ template SiSUdocAbstraction() { foreach (subkey; subkeys) { write(" ", subkey, ", "); foreach (ref_; bookindex_unordered_hashes[mainkey][subkey]) { - auto go = replaceAll(ref_, rgx.book_index_go, "$1"); + auto go = (ref_).replaceAll(rgx.book_index_go, "$1"); write(" {", ref_, "}#", go, ", "); } writeln(" \\\\"); @@ -4372,8 +4359,8 @@ template SiSUdocAbstraction() { bi_tmp_seg ~= " \\\\\n "; ++skn; } - bi_tmp_scroll = replaceFirst(bi_tmp_scroll, rgx.trailing_linebreak, ""); - bi_tmp_seg = replaceFirst(bi_tmp_seg, rgx.trailing_linebreak, ""); + bi_tmp_scroll = (bi_tmp_scroll).replaceFirst(rgx.trailing_linebreak, ""); + bi_tmp_seg = (bi_tmp_seg).replaceFirst(rgx.trailing_linebreak, ""); comp_obj_para = comp_obj_para.init; comp_obj_para.use = "backmatter"; comp_obj_para.is_of = "para"; diff --git a/src/sdp/ao_conf_make_meta_sdlang.d b/src/sdp/ao_conf_make_meta_sdlang.d index af76593..04261d0 100644 --- a/src/sdp/ao_conf_make_meta_sdlang.d +++ b/src/sdp/ao_conf_make_meta_sdlang.d @@ -265,7 +265,7 @@ template SiSUheaderExtractSDLang() { string[] authors_arr; auto authors_raw_arr = split(dochead_meta["creator"]["author"], rgx.arr_delimiter); foreach (author_raw; authors_raw_arr) { - authors_arr ~= replace(author_raw, rgx.raw_author_munge, "$2 $1"); + authors_arr ~= (author_raw).replace(rgx.raw_author_munge, "$2 $1"); } dochead_meta["creator"]["author"] = join(authors_arr, ", ").chomp.chomp; auto t = tuple(dochead_make, dochead_meta); diff --git a/src/sdp/ao_rgx.d b/src/sdp/ao_rgx.d index 6d91096..7d47608 100644 --- a/src/sdp/ao_rgx.d +++ b/src/sdp/ao_rgx.d @@ -27,14 +27,6 @@ template SiSUrgxInit() { static levels_markup = ctRegex!(`^[A-D1-4]$`); static levels_numbered = ctRegex!(`^[0-9]$`); static levels_numbered_headings = ctRegex!(`^[0-7]$`); - static src_pth = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[tm])$`); - static src_fn = - ctRegex!(`^([a-zA-Z0-9._-]+/)*(?P(?P[a-zA-Z0-9._-]+)[.](?Pss[tm]))$`); - static src_fn_master = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssm)$`); - static src_fn_text = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]sst)$`); - static src_fn_insert = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssi)$`); - static src_fn_find_inserts = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`); - static insert_src_fn_ssi_or_sst = ctRegex!(`^<<\s*(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`); /+ comments +/ static comment = ctRegex!(`^%+ `); static comments = ctRegex!(`^%+ |^%+$`); @@ -86,7 +78,7 @@ template SiSUrgxInit() { static para_bullet_indent = ctRegex!(`^_([1-9])[*] `); static para_indent = ctRegex!(`^_([1-9]) `); static para_indent_hang = ctRegex!(`^_([0-9])_([0-9]) `); - static para_attribs = ctRegex!(`^_(([0-9])(_([0-9]))?|_([1-9])?[*]) `); + static para_attribs = ctRegex!(`^_(?:(?:[0-9])(?:_([0-9]))?|(?:[1-9])?[*]) `); /+ blocked markup +/ static block_open = ctRegex!("^((code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)[{].*?$)|^`{3} (code([.][a-z][0-9a-z_]+)?|poem|group|block|quote|table)"); static block_poem_open = ctRegex!("^((poem[{].*?$)|`{3} poem)"); @@ -142,8 +134,9 @@ template SiSUrgxInit() { static inline_curly_delimiter_close_regular = ctRegex!(`\s*\}~`, "m"); static inline_text_and_note_curly = ctRegex!(`(?P.+?)(?:(?:[~])[{][*+ ]*)(?P.+?)(?:[}][~])`, "mg"); static note_ref = ctRegex!(`^\S+?noteref_([0-9]+)`, "mg"); // {^{73.}^}#noteref_73 - static inline_url = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg"); - static inline_link_naked_url = ctRegex!(`(?P^|[ ])(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg"); + static inline_url_generic = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]`, "mg"); + static inline_url = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg"); + static inline_link_naked_url = ctRegex!(`(?P^|[ ])(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!'"]?(?:[ ]|$))`, "mg"); static inline_link_markup_regular = ctRegex!(`(?P^|[ ])\{\s*(?P.+?)\s*\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg"); static inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P[.,;:?!]?(?:[ ]|$))`, "mg"); static inline_link_endnote_url_helper = ctRegex!(`\{~\^\s+(?P.+?)\}(?P(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg"); @@ -185,6 +178,14 @@ template SiSUrgxInit() { ctRegex!("(am|bg|bn|br|ca|cs|cy|da|de|el|en|eo|es|et|eu|fi|fr|ga|gl|he|hi|hr|hy|ia|is|it|ja|ko|la|lo|lt|lv|ml|mr|nl|no|nn|oc|pl|pt|pt_BR|ro|ru|sa|se|sk|sl|sq|sr|sv|ta|te|th|tk|tr|uk|ur|vi|zh)"); auto language_code_and_filename = ctRegex!("(am|bg|bn|br|ca|cs|cy|da|de|el|en|eo|es|et|eu|fi|fr|ga|gl|he|hi|hr|hy|ia|is|it|ja|ko|la|lo|lt|lv|ml|mr|nl|no|nn|oc|pl|pt|pt_BR|ro|ru|sa|se|sk|sl|sq|sr|sv|ta|te|th|tk|tr|uk|ur|vi|zh)/[A-Za-z0-9._-].+?[.](?:sst|ssm)$"); + static src_pth = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[tm])$`); + static src_fn = + ctRegex!(`^([a-zA-Z0-9._-]+/)*(?P(?P[a-zA-Z0-9._-]+)[.](?Pss[tm]))$`); + static src_fn_master = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssm)$`); + static src_fn_text = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]sst)$`); + static src_fn_insert = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssi)$`); + static src_fn_find_inserts = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`); + static insert_src_fn_ssi_or_sst = ctRegex!(`^<<\s*(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`); /+ inline markup footnotes endnotes +/ static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); static inline_notes_al_gen = ctRegex!(`【.+?】`, "m"); @@ -199,27 +200,12 @@ template SiSUrgxInit() { static inline_al_delimiter_open_plus = ctRegex!(`【\+`, "m"); static inline_text_and_note_al = ctRegex!(`(?P.+?)【(?:[*+ ]*)(?P.+?)】`, "mg"); static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|$))`, "mg"); - } -} -/++ - regex: regular expressions used in sisu document parser -+/ -template SiSUoutputRgxInit() { - private import output_defaults; - struct Rgx { /+ inline markup footnotes endnotes +/ - static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); - static inline_notes_al_gen = ctRegex!(`【.+?】`, "m"); - static inline_al_delimiter_open_regular = ctRegex!(`【\s`, "m"); - static inline_al_delimiter_open_symbol_star = ctRegex!(`【[*]\s`, "m"); - static inline_al_delimiter_open_symbol_plus = ctRegex!(`【[+]\s`, "m"); - static inline_al_delimiter_close_regular = ctRegex!(`】`, "m"); - static inline_al_delimiter_open_and_close_regular = ctRegex!(`【|】`, "m"); - static inline_notes_delimiter_al_regular = ctRegex!(`【(.+?)】`, "mg"); - static inline_notes_delimiter_al_regular_number_note = ctRegex!(`【(\d+)\s+(.+?)】`, "mg"); - static inline_al_delimiter_open_asterisk = ctRegex!(`【\*`, "m"); - static inline_al_delimiter_open_plus = ctRegex!(`【\+`, "m"); - static inline_text_and_note_al = ctRegex!(`(?P.+?)【(?:[*+ ]*)(?P.+?)】`, "mg"); - static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|$))`, "mg"); + static inline_link = ctRegex!(`┥(.+?)┝┤(.+?)├`, "mg"); + static inline_a_url = ctRegex!(`(┤)(\S+?)(├)`, "mg"); + static fn_suffix = ctRegex!(`\.fnSuffix`, "mg"); + static inline_link_fn_suffix = ctRegex!(`¤(.+?)(\.fnSuffix)`, "mg"); + static inline_seg_link = ctRegex!(`(¤)(?:.+?)\.fnSuffix`, "mg"); + static mark_internal_site_lnk = ctRegex!(`¤`, "mg"); } } diff --git a/src/sdp/output_epub.d b/src/sdp/output_epub.d index ca11828..c66bf8f 100644 --- a/src/sdp/output_epub.d +++ b/src/sdp/output_epub.d @@ -17,9 +17,10 @@ template outputEPub() { std.typecons, std.uni, std.utf, - defaults; + std.conv : to; import - ao_rgx, + defaults, + output_rgx, output_xhtmls; mixin InternalMarkup; mixin outputXHTMLs; @@ -161,11 +162,12 @@ template outputEPub() { auto return ref const D doc_abstraction, auto return ref I doc_matters, ) { - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; auto xhtml_format = outputXHTMLs(); auto rgx = Rgx(); // string[] toc; string[][string] doc_epub; + string[][string] doc_epub_endnotes; string[] doc; string segment_filename; string[] top_level_headings = ["","","",""]; @@ -173,6 +175,7 @@ template outputEPub() { auto meta_inf_container_xml = epub_container_xml; auto oebps_toc_ncx = epub_oebps_toc(doc_abstraction, doc_matters); auto oebps_content_opf = epub_oebps_content(doc_abstraction, doc_matters); + string suffix = ".xhtml"; foreach (part; doc_matters.keys_seq_seg) { foreach (obj; doc_abstraction[part]) { if (obj.is_a == "heading") { @@ -199,7 +202,8 @@ template outputEPub() { top_level_headings[3] = ""; goto default; default: - top_level_headings[obj.heading_lev_markup] = xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + top_level_headings[obj.heading_lev_markup] = t[0]; break; } break; @@ -209,10 +213,14 @@ template outputEPub() { foreach (top_level_heading; top_level_headings) { doc_epub[segment_filename] ~= top_level_heading; } - doc_epub[segment_filename] ~= xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case 5: .. case 7: - doc_epub[segment_filename] ~= xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case 8: .. case 9: // unused numbers, if remain check if ((doc_matters.opt_action_bool["debug"])) { @@ -254,7 +262,9 @@ template outputEPub() { case "para": switch (obj.is_a) { case "para": - doc_epub[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -280,7 +290,9 @@ template outputEPub() { doc_epub[segment_filename] ~= xhtml_format.nugget(obj); break; case "table": - doc_epub[segment_filename] ~= xhtml_format.para(obj); // + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case "code": doc_epub[segment_filename] ~= xhtml_format.code(obj); @@ -304,19 +316,28 @@ template outputEPub() { case "para": switch (obj.is_a) { case "endnote": - doc_epub[segment_filename] ~= xhtml_format.endnote(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; break; case "glossary": - doc_epub[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case "bibliography": - doc_epub[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case "bookindex": - doc_epub[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; case "blurb": - doc_epub[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub[segment_filename] ~= t[0]; + doc_epub_endnotes[segment_filename] ~= t[1]; break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -346,19 +367,21 @@ template outputEPub() { epub_write_output_files( doc_matters, doc_epub, + doc_epub_endnotes, mimetypes, meta_inf_container_xml, oebps_toc_ncx, oebps_content_opf, ); } - void epub_write_output_files(C,EpD,Mt,Mic,Ot,Oc)( - C doc_matters, - EpD doc_epub, - Mt mimetypes, - Mic meta_inf_container_xml, - Ot oebps_toc_ncx, - Oc oebps_content_opf, + void epub_write_output_files(M,D,E,Mt,Mic,Ot,Oc)( + M doc_matters, + D doc_epub, + E doc_epub_endnotes, + Mt mimetypes, + Mic meta_inf_container_xml, + Ot oebps_toc_ncx, + Oc oebps_content_opf, ) { debug(asserts){ static assert(is(typeof(doc_epub) == string[][string])); @@ -387,6 +410,9 @@ template outputEPub() { foreach (docseg; doc_epub[seg_filename]) { f.writeln(docseg); } + foreach (docseg; doc_epub_endnotes[seg_filename]) { + f.writeln(docseg); + } f.writeln(xhtml_format.tail); // needed for each lev4 } /+ mimetypes +/ diff --git a/src/sdp/output_html.d b/src/sdp/output_html.d index 9b8fc86..d356ee5 100644 --- a/src/sdp/output_html.d +++ b/src/sdp/output_html.d @@ -17,9 +17,10 @@ template outputHTML() { std.typecons, std.uni, std.utf, - defaults; + std.conv : to; import - ao_rgx, + defaults, + output_rgx, output_xhtmls; mixin outputXHTMLs; @@ -27,11 +28,12 @@ template outputHTML() { auto return ref const D doc_abstraction, auto return ref I doc_matters, ) { - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; auto xhtml_format = outputXHTMLs(); auto rgx = Rgx(); string[] doc_html; string[] doc; + string suffix = ".html"; foreach (part; doc_matters.keys_seq_scroll) { foreach (obj; doc_abstraction[part]) { switch (obj.use) { @@ -40,10 +42,10 @@ template outputHTML() { case "para": switch (obj.is_a) { case "heading": - doc_html ~= xhtml_format.heading(obj); + doc_html ~= xhtml_format.heading_scroll(obj, suffix); break; case "toc": - doc_html ~= xhtml_format.toc(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -64,10 +66,10 @@ template outputHTML() { case "para": switch (obj.is_a) { case "heading": - doc_html ~= xhtml_format.heading(obj); + doc_html ~= xhtml_format.heading_scroll(obj, suffix); break; case "para": - doc_html ~= xhtml_format.para(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -93,7 +95,7 @@ template outputHTML() { doc_html ~= xhtml_format.nugget(obj); break; case "table": - doc_html ~= xhtml_format.para(obj); // + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; case "code": doc_html ~= xhtml_format.code(obj); @@ -117,22 +119,22 @@ template outputHTML() { case "para": switch (obj.is_a) { case "heading": - doc_html ~= xhtml_format.heading(obj); + doc_html ~= xhtml_format.heading_scroll(obj, suffix); break; case "endnote": - doc_html ~= xhtml_format.endnote(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; case "glossary": - doc_html ~= xhtml_format.para(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; case "bibliography": - doc_html ~= xhtml_format.para(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; case "bookindex": - doc_html ~= xhtml_format.para(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; case "blurb": - doc_html ~= xhtml_format.para(obj); + doc_html ~= xhtml_format.para_scroll(obj, suffix); break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -153,6 +155,8 @@ template outputHTML() { default: if ((doc_matters.opt_action_bool["debug"])) { writeln(__FILE__, ":", __LINE__, ": ", obj.use); + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a); + writeln(__FILE__, ":", __LINE__, ": ", obj.text); } break; } @@ -188,13 +192,15 @@ template outputHTML() { auto return ref const D doc_abstraction, auto return ref I doc_matters, ) { - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; auto rgx = Rgx(); auto xhtml_format = outputXHTMLs(); string[][string] doc_html; + string[][string] doc_html_endnotes; string[] doc; string segment_filename; string[] top_level_headings = ["","","",""]; + string suffix = ".html"; foreach (part; doc_matters.keys_seq_seg) { foreach (obj; doc_abstraction[part]) { if (obj.is_a == "heading") { // all headings: frontmatter, body & backmatter @@ -221,7 +227,8 @@ template outputHTML() { top_level_headings[3] = ""; goto default; default: - top_level_headings[obj.heading_lev_markup] = xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + top_level_headings[obj.heading_lev_markup] = t[0]; break; } break; @@ -231,10 +238,14 @@ template outputHTML() { foreach (top_level_heading; top_level_headings) { doc_html[segment_filename] ~= top_level_heading; } - doc_html[segment_filename] ~= xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + doc_html[segment_filename] ~= to!string(t[0]); + doc_html_endnotes[segment_filename] ~= t[1]; break; case 5: .. case 7: - doc_html[segment_filename] ~= xhtml_format.heading(obj); + auto t = xhtml_format.heading_seg(obj, suffix); + doc_html[segment_filename] ~= to!string(t[0]); + doc_html_endnotes[segment_filename] ~= t[1]; break; case 8: .. case 9: // unused numbers, if remain check if ((doc_matters.opt_action_bool["debug"])) { @@ -255,7 +266,8 @@ template outputHTML() { case "para": switch (obj.is_a) { case "toc": - doc_html[segment_filename] ~= xhtml_format.toc(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= to!string(t[0]); break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -276,7 +288,9 @@ template outputHTML() { case "para": switch (obj.is_a) { case "para": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= to!string(t[0]); + doc_html_endnotes[segment_filename] ~= t[1]; break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -302,7 +316,9 @@ template outputHTML() { doc_html[segment_filename] ~= xhtml_format.nugget(obj); break; case "table": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; + doc_html_endnotes[segment_filename] ~= t[1]; break; case "code": doc_html[segment_filename] ~= xhtml_format.code(obj); @@ -326,19 +342,28 @@ template outputHTML() { case "para": switch (obj.is_a) { case "endnote": - doc_html[segment_filename] ~= xhtml_format.endnote(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; break; case "glossary": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; + doc_html_endnotes[segment_filename] ~= t[1]; break; case "bibliography": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; + doc_html_endnotes[segment_filename] ~= t[1]; break; case "bookindex": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; + doc_html_endnotes[segment_filename] ~= t[1]; break; case "blurb": - doc_html[segment_filename] ~= xhtml_format.para(obj); + auto t = xhtml_format.para_seg(obj, suffix); + doc_html[segment_filename] ~= t[0]; + doc_html_endnotes[segment_filename] ~= t[1]; break; default: if ((doc_matters.opt_action_bool["debug"])) { @@ -365,16 +390,17 @@ template outputHTML() { } } } - seg_write_output(doc_matters, doc_html); + seg_write_output(doc_matters, doc_html, doc_html_endnotes); } - void seg_write_output(M,C)( + void seg_write_output(M,D,E)( M doc_matters, - C doc_html, + D doc_html, + E doc_html_endnotes, ) { debug(asserts){ static assert(is(typeof(doc_html) == string[][string])); } - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; auto rgx = Rgx(); mixin SiSUpaths; auto pth_html = HtmlPaths(); @@ -389,6 +415,9 @@ template outputHTML() { foreach (docseg; doc_html[seg_filename]) { f.writeln(docseg); } + foreach (docseg; doc_html_endnotes[seg_filename]) { + f.writeln(docseg); + } f.writeln(xhtml_format.tail); // needed for each lev4 } } @@ -517,403 +546,403 @@ template outputHTML() { p.i8 {padding-left: 8em;} p.i9 {padding-left: 9em;} /* hanging indent */ - p.h0i0 { + p[indent=\"h0i0\"] { padding-left: 0em; text-indent: 0em; } - p.h0i1 { + p[indent=\"h0i1\"] { padding-left: 1em; text-indent: -1em; } - p.h0i2 { + p[indent=\"h0i2\"] { padding-left: 2em; text-indent: -2em; } - p.h0i3 { + p[indent=\"h0i3\"] { padding-left: 3em; text-indent: -3em; } - p.h0i4 { + p[indent=\"h0i4\"] { padding-left: 4em; text-indent: -4em; } - p.h0i5 { + p[indent=\"h0i5\"] { padding-left: 5em; text-indent: -5em; } - p.h0i6 { + p[indent=\"h0i6\"] { padding-left: 6em; text-indent: -6em; } - p.h0i7 { + p[indent=\"h0i7\"] { padding-left: 7em; text-indent: -7em; } - p.h0i8 { + p[indent=\"h0i8\"] { padding-left: 8em; text-indent: -8em; } - p.h0i9 { + p[indent=\"h0i9\"] { padding-left: 9em; text-indent: -9em; } - p.h1i0 { + p[indent=\"h1i0\"] { padding-left: 0em; text-indent: 1em; } - p.h1i1 { + p[indent=\"h1i1\"] { padding-left: 1em; text-indent: 0em; } - p.h1i2 { + p[indent=\"h1i2\"] { padding-left: 2em; text-indent: -1em; } - p.h1i3 { + p[indent=\"h1i3\"] { padding-left: 3em; text-indent: -2em; } - p.h1i4 { + p[indent=\"h1i4\"] { padding-left: 4em; text-indent: -3em; } - p.h1i5 { + p[indent=\"h1i5\"] { padding-left: 5em; text-indent: -4em; } - p.h1i6 { + p[indent=\"h1i6\"] { padding-left: 6em; text-indent: -5em; } - p.h1i7 { + p[indent=\"h1i7\"] { padding-left: 7em; text-indent: -6em; } - p.h1i8 { + p[indent=\"h1i8\"] { padding-left: 8em; text-indent: -7em; } - p.h1i9 { + p[indent=\"h1i9\"] { padding-left: 9em; text-indent: -8em; } - p.h2i0 { + p[indent=\"h2i0\"] { padding-left: 0em; text-indent: 2em; } - p.h2i1 { + p[indent=\"h2i1\"] { padding-left: 1em; text-indent: 1em; } - p.h2i2 { + p[indent=\"h2i2\"] { padding-left: 2em; text-indent: 0em; } - p.h2i3 { + p[indent=\"h2i3\"] { padding-left: 3em; text-indent: -1em; } - p.h2i4 { + p[indent=\"h2i4\"] { padding-left: 4em; text-indent: -2em; } - p.h2i5 { + p[indent=\"h2i5\"] { padding-left: 5em; text-indent: -3em; } - p.h2i6 { + p[indent=\"h2i6\"] { padding-left: 6em; text-indent: -4em; } - p.h2i7 { + p[indent=\"h2i7\"] { padding-left: 7em; text-indent: -5em; } - p.h2i8 { + p[indent=\"h2i8\"] { padding-left: 8em; text-indent: -6em; } - p.h2i9 { + p[indent=\"h2i9\"] { padding-left: 9em; text-indent: -7em; } - p.h3i0 { + p[indent=\"h3i0\"] { padding-left: 0em; text-indent: 3em; } - p.h3i1 { + p[indent=\"h3i1\"] { padding-left: 1em; text-indent: 2em; } - p.h3i2 { + p[indent=\"h3i2\"] { padding-left: 2em; text-indent: 1em; } - p.h3i3 { + p[indent=\"h3i3\"] { padding-left: 3em; text-indent: 0em; } - p.h3i4 { + p[indent=\"h3i4\"] { padding-left: 4em; text-indent: -1em; } - p.h3i5 { + p[indent=\"h3i5\"] { padding-left: 5em; text-indent: -2em; } - p.h3i6 { + p[indent=\"h3i6\"] { padding-left: 6em; text-indent: -3em; } - p.h3i7 { + p[indent=\"h3i7\"] { padding-left: 7em; text-indent: -4em; } - p.h3i8 { + p[indent=\"h3i8\"] { padding-left: 8em; text-indent: -5em; } - p.h3i9 { + p[indent=\"h3i9\"] { padding-left: 9em; text-indent: -6em; } - p.h4i0 { + p[indent=\"h4i0\"] { padding-left: 0em; text-indent: 4em; } - p.h4i1 { + p[indent=\"h4i1\"] { padding-left: 1em; text-indent: 3em; } - p.h4i2 { + p[indent=\"h4i2\"] { padding-left: 2em; text-indent: 2em; } - p.h4i3 { + p[indent=\"h4i3\"] { padding-left: 3em; text-indent: 1em; } - p.h4i4 { + p[indent=\"h4i4\"] { padding-left: 4em; text-indent: 0em; } - p.h4i5 { + p[indent=\"h4i5\"] { padding-left: 5em; text-indent: -1em; } - p.h4i6 { + p[indent=\"h4i6\"] { padding-left: 6em; text-indent: -2em; } - p.h4i7 { + p[indent=\"h4i7\"] { padding-left: 7em; text-indent: -3em; } - p.h4i8 { + p[indent=\"h4i8\"] { padding-left: 8em; text-indent: -4em; } - p.h4i9 { + p[indent=\"h4i9\"] { padding-left: 9em; text-indent: -5em; } - p.h5i0 { + p[indent=\"h5i0\"] { padding-left: 0em; text-indent: 5em; } - p.h5i1 { + p[indent=\"h5i1\"] { padding-left: 1em; text-indent: 4em; } - p.h5i2 { + p[indent=\"h5i2\"] { padding-left: 2em; text-indent: 3em; } - p.h5i3 { + p[indent=\"h5i3\"] { padding-left: 3em; text-indent: 2em; } - p.h5i4 { + p[indent=\"h5i4\"] { padding-left: 4em; text-indent: 1em; } - p.h5i5 { + p[indent=\"h5i5\"] { padding-left: 5em; text-indent: 0em; } - p.h5i6 { + p[indent=\"h5i6\"] { padding-left: 6em; text-indent: -1em; } - p.h5i7 { + p[indent=\"h5i7\"] { padding-left: 7em; text-indent: -2em; } - p.h5i8 { + p[indent=\"h5i8\"] { padding-left: 8em; text-indent: -3em; } - p.h5i9 { + p[indent=\"h5i9\"] { padding-left: 9em; text-indent: -4em; } - p.h6i0 { + p[indent=\"h6i0\"] { padding-left: 0em; text-indent: 6em; } - p.h6i1 { + p[indent=\"h6i1\"] { padding-left: 1em; text-indent: 5em; } - p.h6i2 { + p[indent=\"h6i2\"] { padding-left: 2em; text-indent: 4em; } - p.h6i3 { + p[indent=\"h6i3\"] { padding-left: 3em; text-indent: 3em; } - p.h6i4 { + p[indent=\"h6i4\"] { padding-left: 4em; text-indent: 2em; } - p.h6i5 { + p[indent=\"h6i5\"] { padding-left: 5em; text-indent: 1em; } - p.h6i6 { + p[indent=\"h6i6\"] { padding-left: 6em; text-indent: 0em; } - p.h6i7 { + p[indent=\"h6i7\"] { padding-left: 7em; text-indent: -1em; } - p.h6i8 { + p[indent=\"h6i8\"] { padding-left: 8em; text-indent: -2em; } - p.h6i9 { + p[indent=\"h6i9\"] { padding-left: 9em; text-indent: -3em; } - p.h7i0 { + p[indent=\"h7i0\"] { padding-left: 0em; text-indent: 7em; } - p.h7i1 { + p[indent=\"h7i1\"] { padding-left: 1em; text-indent: 6em; } - p.h7i2 { + p[indent=\"h7i2\"] { padding-left: 2em; text-indent: 5em; } - p.h7i3 { + p[indent=\"h7i3\"] { padding-left: 3em; text-indent: 4em; } - p.h7i4 { + p[indent=\"h7i4\"] { padding-left: 4em; text-indent: 3em; } - p.h7i5 { + p[indent=\"h7i5\"] { padding-left: 5em; text-indent: 2em; } - p.h7i6 { + p[indent=\"h7i6\"] { padding-left: 6em; text-indent: 1em; } - p.h7i7 { + p[indent=\"h7i7\"] { padding-left: 7em; text-indent: 0em; } - p.h7i8 { + p[indent=\"h7i8\"] { padding-left: 8em; text-indent: -1em; } - p.h7i9 { + p[indent=\"h7i9\"] { padding-left: 9em; text-indent: -2em; } - p.h8i0 { + p[indent=\"h8i0\"] { padding-left: 0em; text-indent: 8em; } - p.h8i1 { + p[indent=\"h8i1\"] { padding-left: 1em; text-indent: 7em; } - p.h8i2 { + p[indent=\"h8i2\"] { padding-left: 2em; text-indent: 6em; } - p.h8i3 { + p[indent=\"h8i3\"] { padding-left: 3em; text-indent: 5em; } - p.h8i4 { + p[indent=\"h8i4\"] { padding-left: 4em; text-indent: 4em; } - p.h8i5 { + p[indent=\"h8i5\"] { padding-left: 5em; text-indent: 3em; } - p.h8i6 { + p[indent=\"h8i6\"] { padding-left: 6em; text-indent: 2em; } - p.h8i7 { + p[indent=\"h8i7\"] { padding-left: 7em; text-indent: 1em; } - p.h8i8 { + p[indent=\"h8i8\"] { padding-left: 8em; text-indent: 0em; } - p.h8i9 { + p[indent=\"h8i9\"] { padding-left: 9em; text-indent: -1em; } - p.h9i0 { + p[indent=\"h9i0\"] { padding-left: 0em; text-indent: 9em; } - p.h9i1 { + p[indent=\"h9i1\"] { padding-left: 1em; text-indent: 8em; } - p.h9i2 { + p[indent=\"h9i2\"] { padding-left: 2em; text-indent: 7em; } - p.h9i3 { + p[indent=\"h9i3\"] { padding-left: 3em; text-indent: 6em; } - p.h9i4 { + p[indent=\"h9i4\"] { padding-left: 4em; text-indent: 5em; } - p.h9i5 { + p[indent=\"h9i5\"] { padding-left: 5em; text-indent: 4em; } - p.h9i6 { + p[indent=\"h9i6\"] { padding-left: 6em; text-indent: 3em; } - p.h9i7 { + p[indent=\"h9i7\"] { padding-left: 7em; text-indent: 2em; } - p.h9i8 { + p[indent=\"h9i8\"] { padding-left: 8em; text-indent: 1em; } - p.h9i9 { + p[indent=\"h9i9\"] { padding-left: 9em; text-indent: 0em; } diff --git a/src/sdp/output_hub.d b/src/sdp/output_hub.d index a1182a6..67275e0 100644 --- a/src/sdp/output_hub.d +++ b/src/sdp/output_hub.d @@ -27,10 +27,10 @@ template outputHub() { output_xhtmls, source_sisupod; import - ao_rgx, + output_rgx, output_xhtmls; void outputHub(D,I)(D doc_abstraction, I doc_matters) { - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; auto rgx = Rgx(); if ((doc_matters.opt_action_bool["verbose"])) { writeln(doc_matters.keys_seq_seg); diff --git a/src/sdp/output_rgx.d b/src/sdp/output_rgx.d new file mode 100644 index 0000000..5a9a47e --- /dev/null +++ b/src/sdp/output_rgx.d @@ -0,0 +1,37 @@ +/++ + regex: regular expressions used in sisu document parser ++/ +template SiSUoutputRgxInit() { + private import defaults; + struct Rgx { + static src_pth = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[tm])$`); + static src_fn = + ctRegex!(`^([a-zA-Z0-9._-]+/)*(?P(?P[a-zA-Z0-9._-]+)[.](?Pss[tm]))$`); + static src_fn_master = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssm)$`); + static src_fn_text = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]sst)$`); + static src_fn_insert = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ssi)$`); + static src_fn_find_inserts = ctRegex!(`^(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[im])$`); + static insert_src_fn_ssi_or_sst = ctRegex!(`^<<\s*(?P[a-zA-Z0-9._-]+/)*(?P[a-zA-Z0-9._-]+[.]ss[ti])$`); + /+ inline markup footnotes endnotes +/ + static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); + static inline_notes_al_gen = ctRegex!(`【.+?】`, "m"); + static inline_al_delimiter_open_regular = ctRegex!(`【\s`, "m"); + static inline_al_delimiter_open_symbol_star = ctRegex!(`【[*]\s`, "m"); + static inline_al_delimiter_open_symbol_plus = ctRegex!(`【[+]\s`, "m"); + static inline_al_delimiter_close_regular = ctRegex!(`】`, "m"); + static inline_al_delimiter_open_and_close_regular = ctRegex!(`【|】`, "m"); + static inline_notes_delimiter_al_regular = ctRegex!(`【(.+?)】`, "mg"); + static inline_notes_delimiter_al_regular_number_note = ctRegex!(`【(\d+)\s+(.+?)】`, "mg"); + static inline_al_delimiter_open_asterisk = ctRegex!(`【\*`, "m"); + static inline_al_delimiter_open_plus = ctRegex!(`【\+`, "m"); + static inline_text_and_note_al = ctRegex!(`(?P.+?)【(?:[*+ ]*)(?P.+?)】`, "mg"); + static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+]*\s+.+?】|$))`, "mg"); + /+ inline markup footnotes endnotes +/ + static inline_link = ctRegex!(`┥(.+?)┝┤(.+?)├`, "mg"); + static inline_a_url = ctRegex!(`(┤)(\S+?)(├)`, "mg"); + static fn_suffix = ctRegex!(`\.fnSuffix`, "mg"); + static inline_link_fn_suffix = ctRegex!(`¤(.+?)(\.fnSuffix)`, "mg"); + static inline_seg_link = ctRegex!(`(¤)(?:.+?)\.fnSuffix`, "mg"); + static mark_internal_site_lnk = ctRegex!(`¤`, "mg"); + } +} diff --git a/src/sdp/output_xhtmls.d b/src/sdp/output_xhtmls.d index 6a4c3a3..de5236d 100644 --- a/src/sdp/output_xhtmls.d +++ b/src/sdp/output_xhtmls.d @@ -1,5 +1,30 @@ template outputXHTMLs() { + private import + std.algorithm, + std.array, + std.container, + std.exception, + std.file, + std.getopt, + std.json, + std.process, + std.stdio, + std.path, + std.range, + std.regex, + std.string, + std.traits, + std.typecons, + std.uni, + std.utf, + std.conv : to; + import + defaults, + output_rgx, + output_xhtmls; + mixin SiSUoutputRgxInit; struct outputXHTMLs { + auto rgx = Rgx(); string _xhtml_anchor_tags(const(string[]) anchor_tags) { string tags=""; if (anchor_tags.length > 0) { @@ -98,6 +123,137 @@ template outputXHTMLs() { ¶"); return o; } + auto inline_links(O)( + auto return ref const O obj, + string _txt, + string _suffix = ".html", + string seg_scroll = "seg", + ) { + if (obj.inline_links) { + if ((seg_scroll == "scroll") + && match(_txt, rgx.mark_internal_site_lnk)) { + _txt = (_txt).replaceAll( + rgx.inline_seg_link, + "$1"); + } + _txt = (_txt).replaceAll( + rgx.inline_link_fn_suffix, + ("$1" ~ _suffix)); + _txt = (_txt).replaceAll( + rgx.inline_link, + ("$1")); + _txt = (_txt).replaceAll( + rgx.mark_internal_site_lnk, + "" + ); + } + debug(markup_links) { + if ( match(_txt, rgx.inline_link)) { + writeln(__LINE__, + " (missed) markup link identified (", + obj.inline_links, + "): ", obj.is_a, ": ", + obj.text + ); + } + } + debug(markup) { + if ( match(_txt, rgx.inline_link)) { + writeln(__LINE__, + " (missed) markup link identified (", + obj.inline_links, + "): ", obj.is_a, ": ", + obj.text + ); + } + } + return _txt; + } + auto inline_notes_scroll(O)( + auto return ref const O obj, + string _txt, + ) { + if (obj.inline_notes_reg) { + _txt = (_txt).replaceAll( + rgx.inline_notes_delimiter_al_regular_number_note, + (" $1 ") + ); + } + debug(markup_endnotes) { + if (match(_txt, rgx.inline_notes_delimiter_al_regular_number_note)) { + writeln(__LINE__, " (missed) markup endnote: ", obj.is_a, ": ", obj.text); + } + } + debug(markup) { + if (match(_txt, rgx.inline_notes_delimiter_al_regular_number_note)) { + writeln(__LINE__, " (missed) markup endnote: ", obj.is_a, ": ", obj.text); + } + } + // if (obj.inline_notes_star) { + // _txt = replaceAll( + // _txt, + // rgx.inline_notes_delimiter_al_regular_number_note, + // (" $1 ") + // ); + // } + return _txt; + } + auto inline_notes_seg(O)( + auto return ref const O obj, + string _txt, + ) { + string[] _endnotes; + if (obj.inline_notes_reg) { + /+ need markup for text, and separated footnote +/ + foreach(m; matchAll(_txt, rgx.inline_notes_delimiter_al_regular_number_note)) { + _endnotes ~= format( + "%s%s%s%s\n %s%s%s%s%s\n %s\n%s", + "

", + "", + " ", + m.captures[1], + ".", + m.captures[2], + "

" + ); + } + _txt = (_txt).replaceAll( + rgx.inline_notes_delimiter_al_regular_number_note, + (" $1 ") + ); + } else if (match(_txt, rgx.inline_notes_delimiter_al_regular_number_note)) { + debug(markup) { + writeln(__LINE__, " endnote: ", obj.is_a, ": ", obj.text); + } + } + auto t = tuple( + _txt, + _endnotes, + ); + return t; + } + auto inline_markup_scroll(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + string _txt = obj.text; + _txt = inline_links(obj, _txt, _suffix, "scroll"); + _txt = inline_notes_scroll(obj, _txt); + return _txt; + } + auto inline_markup_seg(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + string _txt = obj.text; + _txt = inline_links(obj, _txt, _suffix, "seg"); + auto t = inline_notes_seg(obj, _txt); + return t; + } auto toc(O)( auto return ref const O obj, ) { @@ -116,6 +272,7 @@ template outputXHTMLs() { } auto heading(O)( auto return ref const O obj, + string _txt, ) { auto tags = _xhtml_anchor_tags(obj.anchor_tags); string o; @@ -129,7 +286,7 @@ template outputXHTMLs() { obj.heading_lev_markup, obj.is_a, tags, - obj.text, + _txt, obj.heading_lev_markup, ); } else { @@ -147,14 +304,38 @@ template outputXHTMLs() { obj.obj_cite_number, obj.obj_cite_number, tags, - obj.text, + _txt, obj.heading_lev_markup, ); } return o; } + auto heading_scroll(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + auto tags = _xhtml_anchor_tags(obj.anchor_tags); + string _txt = inline_markup_scroll(obj, _suffix); // issue + string o = heading(obj, _txt); + return o; + } + auto heading_seg(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + auto t = inline_markup_seg(obj, _suffix); + string _txt = t[0]; + string[] _endnotes = t[1]; + string o = heading(obj, _txt); + auto u = tuple( + o, + _endnotes, + ); + return u; + } auto para(O)( auto return ref const O obj, + string _txt, ) { auto tags = _xhtml_anchor_tags(obj.anchor_tags); string o; @@ -168,7 +349,7 @@ template outputXHTMLs() { obj.indent_hang, obj.indent_base, tags, - obj.text + _txt ); } else { o = format(q"¶
@@ -184,11 +365,34 @@ template outputXHTMLs() { obj.indent_base, obj.obj_cite_number, tags, - obj.text + _txt ); } return o; } + auto para_scroll(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + auto tags = _xhtml_anchor_tags(obj.anchor_tags); + string _txt = inline_markup_scroll(obj, _suffix); // issue + string o = para(obj, _txt); + return o; + } + auto para_seg(O)( + auto return ref const O obj, + string _suffix = ".html", + ) { + auto t = inline_markup_seg(obj, _suffix); + string _txt = to!string(t[0]); + string[] _endnotes = t[1]; + string o = para(obj, _txt); + auto u = tuple( + o, + _endnotes, + ); + return u; + } auto nugget(O)( auto return ref const O obj, ) { diff --git a/src/sdp/source_sisupod.d b/src/sdp/source_sisupod.d index aa91b6d..ce2bfaf 100644 --- a/src/sdp/source_sisupod.d +++ b/src/sdp/source_sisupod.d @@ -17,15 +17,16 @@ template SiSUpod() { std.typecons, std.uni, std.utf, - defaults; + std.conv : to; import - ao_rgx, + defaults, + output_rgx, output_xhtmls; void SiSUpod(T)(T doc_matters) { debug(asserts){ // static assert(is(typeof(doc_matters) == tuple)); } - mixin SiSUrgxInit; + mixin SiSUoutputRgxInit; mixin SiSUpaths; auto pth_sisupod = SiSUpodPaths(); mixin SiSUlanguageCodes; @@ -103,6 +104,7 @@ template SiSUpod() { ); } } + } catch (ErrnoException ex) { // Handle error -- cgit v1.2.3