From da0f051e1aced4fa1fd5cd13c0548279bf04b2a0 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Wed, 12 Apr 2017 15:23:23 -0400 Subject: 0.13.9 sisupod & epub3, zipped output --- src/sdp/output_epub3.d | 717 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 717 insertions(+) create mode 100644 src/sdp/output_epub3.d (limited to 'src/sdp/output_epub3.d') diff --git a/src/sdp/output_epub3.d b/src/sdp/output_epub3.d new file mode 100644 index 0000000..f0ac2c3 --- /dev/null +++ b/src/sdp/output_epub3.d @@ -0,0 +1,717 @@ +template outputEPub3() { + private import + std.algorithm, + std.array, + std.container, + std.digest.sha, + std.exception, + std.file, + std.getopt, + std.json, + std.outbuffer, + std.path, + std.process, + std.range, + std.regex, + std.stdio, + std.string, + std.traits, + std.typecons, + std.uni, + std.utf, + std.zip, + std.conv : to; + import + create_zip_file, + defaults, + output_rgx, + output_xhtmls; + mixin InternalMarkup; + mixin outputXHTMLs; + string epub3_mimetypes() { + string o; + o = format(q"¶application/epub+zip¶") ~ "\n"; + return o; + } + string epub3_container_xml() { + string o; + o = format(q"¶¶") ~ "\n"; + o ~= format(q"¶ + + + ¶") ~ "\n\n"; + return o; + } + string epub3_oebps_content(D,I,P)(D doc_abstraction, I doc_matters, P parts) { + string uuid = "18275d951861c77f78acd05672c9906924c59f18a2e0ba06dad95959693e9bd8"; // TODO sort uuid in doc_matters! + string content = format(q"¶ + + + + %s + main + %s + subtitle + %s + %s + %s + Copyright: %s + %s + urn:uuid:%s + + + + + + + + + + ¶", + uuid, + doc_matters.dochead_meta["title"]["full"], + doc_matters.dochead_meta["title"]["main"], + (doc_matters.dochead_meta["title"]["sub"].empty) + ? "" : doc_matters.dochead_meta["title"]["sub"], + (doc_matters.dochead_meta["creator"]["author"].empty) + ? "" : doc_matters.dochead_meta["creator"]["author"], + (doc_matters.dochead_meta["creator"]["author"].empty) + ? "" : doc_matters.dochead_meta["creator"]["author"], + doc_matters.language, + (doc_matters.dochead_meta["date"]["published"].empty) + ? "" : doc_matters.dochead_meta["date"]["published"], + (doc_matters.dochead_meta["rights"]["copyright"].empty) + ? "" : doc_matters.dochead_meta["rights"]["copyright"], + uuid, + uuid, + uuid, + ); + content ~= " " ~ "" ~ "\n "; + content ~= parts["manifest_documents"]; + // TODO sort jpg & png + content ~= " " ~ "" ~ "\n "; + foreach (image; doc_matters.image_list) { + content ~= format(q"¶ + ¶", + image, // strip image type, remove .png .jpg suffix, use in media-type="image/" + image, + ); + } + content ~= " " ~ "" ~ "\n "; + content ~= " " ~ "" ~ "\n "; + content ~= parts["spine"]; + content ~= " " ~ "" ~ "\n "; + content ~= " " ~ "" ~ "\n "; + content ~= parts["guide"]; + content ~= " " ~ "" ~ "\n "; + content ~= "" ~ ""; + return content; + } + string epub3_oebps_toc_nav_xhtml(D,I)(D doc_abstraction, I doc_matters) { + enum DomTags { none, open, close, close_and_open, open_still, } + auto markup = InlineMarkup(); + string toc ="\n"; + return toc; + } + string epub2_oebps_toc_ncx(D,I)(D doc_abstraction, I doc_matters) { + int counter = 0; + string uuid = "18275d951861c77f78acd05672c9906924c59f18a2e0ba06dad95959693e9bd8"; // TODO shared elsewhere + auto markup = InlineMarkup(); + enum DomTags { none, open, close, close_and_open, open_still, } + string toc = format(q"¶ + + + + %s%s + + + + + + + + + %s + + + %s + + ¶", + doc_matters.dochead_meta["title"]["full"], // title + (doc_matters.dochead_meta["creator"]["author"].empty) ? "" : " by " ~ doc_matters.dochead_meta["creator"]["author"], // author + uuid, // uuid + "3", // content depth + doc_matters.dochead_meta["title"]["full"], // title + (doc_matters.dochead_meta["creator"]["author"].empty) ? "" : doc_matters.dochead_meta["creator"]["author"], // author + ); + foreach (sect; doc_matters.keys_seq_seg) { + foreach (obj; doc_abstraction[sect]) { + if (obj.is_a == "heading") { + foreach_reverse (k; 0 .. 7) { + switch (obj.dom_markedup[k]) { + case DomTags.close : + toc ~= ""; + break; + case DomTags.close_and_open : + ++counter; + toc ~= ""; + toc ~= format(q"¶ + + %s + + ¶", + counter, + obj.text, + obj.segment_anchor_tag, // lev < 4 [no link]; lev == 4 [filename] markup.xhtml; lev > 4 [filename#ocn] (links done in segment_anchor_tag) + ); + break; + case DomTags.open : + ++counter; + toc ~= format(q"¶ + + %s + + ¶", + counter, + obj.text, + obj.segment_anchor_tag, // lev < 4 [no link]; lev == 4 [filename] markup.xhtml; lev > 4 [filename#ocn] (fix links in segment_anchor_tag) + ); + break; + default : + break; + } + } + } + } + } + toc ~= format(q"¶ + ¶"); + return toc; + } + + void outputEPub3(D,I)( + auto return ref const D doc_abstraction, + auto return ref I doc_matters, + ) { + mixin SiSUoutputRgxInit; + auto xhtml_format = outputXHTMLs(); + auto rgx = Rgx(); + string[][string] doc_epub3; + string[][string] doc_epub3_endnotes; + string[] doc; + string segment_filename; + string[] top_level_headings = ["","","",""]; + string[string] oepbs_content_parts; + string suffix = ".xhtml"; + foreach (part; doc_matters.keys_seq_seg) { + foreach (obj; doc_abstraction[part]) { + if (obj.is_a == "heading") { + switch (obj.heading_lev_markup) { + case 0: .. case 3: + /+ fill buffer, and replace with new levels from 1 to 3 +/ + switch (obj.heading_lev_markup) { + case 0: + top_level_headings[0] = ""; + top_level_headings[1] = ""; + top_level_headings[2] = ""; + top_level_headings[3] = ""; + goto default; + case 1: + top_level_headings[1] = ""; + top_level_headings[2] = ""; + top_level_headings[3] = ""; + goto default; + case 2: + top_level_headings[2] = ""; + top_level_headings[3] = ""; + goto default; + case 3: + top_level_headings[3] = ""; + goto default; + default: + auto t = xhtml_format.heading_seg(obj, suffix); + top_level_headings[obj.heading_lev_markup] = t[0]; + break; + } + break; + case 4: + segment_filename = obj.segment_anchor_tag; + doc_epub3[segment_filename] ~= xhtml_format.seg_head(doc_matters.dochead_meta); + foreach (top_level_heading; top_level_headings) { + doc_epub3[segment_filename] ~= top_level_heading; + } + auto t = xhtml_format.heading_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case 5: .. case 7: + auto t = xhtml_format.heading_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case 8: .. case 9: // unused numbers, if remain check + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a, ": ", obj.heading_lev_markup); + writeln(__FILE__, ":", __LINE__, ": ", obj.text); // check + } + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a, ": ", obj.heading_lev_markup); + } + break; + } + } else { + switch (obj.use) { + case "frontmatter": + switch (obj.is_of) { + case "para": + switch (obj.is_a) { + case "toc": + doc_epub3[segment_filename] ~= xhtml_format.toc(obj); + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_of); + } + break; + } + break; + case "body": + switch (obj.is_of) { + case "para": + switch (obj.is_a) { + case "para": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a); + } + break; + } + break; + case "block": + switch (obj.is_a) { + case "quote": + auto t = xhtml_format.quote_seg(obj, suffix); + doc_epub3[segment_filename] ~= to!string(t[0]); + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "group": + auto t = xhtml_format.group_seg(obj, suffix); + doc_epub3[segment_filename] ~= to!string(t[0]); + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "block": + auto t = xhtml_format.block_seg(obj, suffix); + doc_epub3[segment_filename] ~= to!string(t[0]); + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "poem": + break; + case "verse": + auto t = xhtml_format.verse_seg(obj, suffix); + doc_epub3[segment_filename] ~= to!string(t[0]); + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "code": + doc_epub3[segment_filename] ~= xhtml_format.code(obj); + break; + case "table": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_of); + } + break; + } + break; + case "backmatter": + switch (obj.is_of) { + case "para": + switch (obj.is_a) { + case "endnote": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + break; + case "glossary": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "bibliography": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "bookindex": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + case "blurb": + auto t = xhtml_format.para_seg(obj, suffix); + doc_epub3[segment_filename] ~= t[0]; + doc_epub3_endnotes[segment_filename] ~= t[1]; + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_a); + } + break; + } + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.is_of); + } + break; + } + break; + case "comment": + break; + default: + if ((doc_matters.opt_action_bool["debug"])) { + writeln(__FILE__, ":", __LINE__, ": ", obj.use); + } + break; + } + } + if (obj.is_a == "heading") { + if (obj.heading_lev_markup == 4) { + oepbs_content_parts["manifest_documents"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + obj.segment_anchor_tag, + ); + oepbs_content_parts["spine"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + ); + oepbs_content_parts["guide"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + obj.segment_anchor_tag, + ); + } else if (obj.heading_lev_markup > 4) { + oepbs_content_parts["manifest_documents"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + obj.obj_cite_number, + obj.segment_anchor_tag, + obj.obj_cite_number, + ); + oepbs_content_parts["spine"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + obj.obj_cite_number, + ); + oepbs_content_parts["guide"] ~= format(q"¶ + ¶", + obj.segment_anchor_tag, + obj.obj_cite_number, + obj.segment_anchor_tag, + obj.obj_cite_number, + ); + } + } + } + } + /+ epub specific documents +/ + auto mimetypes = epub3_mimetypes; + auto meta_inf_container_xml = epub3_container_xml; + auto oebps_toc_ncx = epub2_oebps_toc_ncx(doc_abstraction, doc_matters); + auto oebps_toc_nav_xhtml = epub3_oebps_toc_nav_xhtml(doc_abstraction, doc_matters); + auto oebps_content_opf = epub3_oebps_content(doc_abstraction, doc_matters, oepbs_content_parts); + epub3_write_output_files( + doc_matters, + doc_epub3, + doc_epub3_endnotes, + mimetypes, + meta_inf_container_xml, + oebps_toc_nav_xhtml, + oebps_toc_ncx, + oebps_content_opf, + ); + } + void epub3_write_output_files(M,D,E,Mt,Mic,Otnx,Otn,Oc)( + M doc_matters, + D doc_epub3, + E doc_epub3_endnotes, + Mt mimetypes, + Mic meta_inf_container_xml, + Otnx oebps_toc_nav_xhtml, + Otn oebps_toc_ncx, + Oc oebps_content_opf, + ) { + debug(asserts) { + static assert(is(typeof(doc_epub3) == string[][string])); + static assert(is(typeof(mimetypes) == string)); + static assert(is(typeof(meta_inf_container_xml) == string)); + static assert(is(typeof(oebps_toc_nav_xhtml) == string)); + static assert(is(typeof(oebps_toc_ncx) == string)); + static assert(is(typeof(oebps_content_opf) == string)); + } + mixin SiSUpaths; + auto pth_epub3 = Epub3paths(); + auto xhtml_format = outputXHTMLs(); + /+ zip file +/ + auto fn_epub = pth_epub3.epub_file(doc_matters.source_filename); + auto zip = new ZipArchive(); // ZipArchive zip = new ZipArchive(); + /+ zip archive member files +/ + try { + if (!exists(pth_epub3.doc_meta_inf(doc_matters.source_filename))) { + pth_epub3.doc_meta_inf(doc_matters.source_filename).mkdirRecurse; + } + if (!exists(pth_epub3.doc_oebps_css(doc_matters.source_filename))) { + pth_epub3.doc_oebps_css(doc_matters.source_filename).mkdirRecurse; + } + if (!exists(pth_epub3.doc_oebps_image(doc_matters.source_filename))) { + pth_epub3.doc_oebps_image(doc_matters.source_filename).mkdirRecurse; + } + { /+ OEBPS/[segments].xhtml (the document contents) +/ + foreach (seg_filename; doc_matters.segnames) { + string fn = pth_epub3.fn_oebps_content_xhtml(doc_matters.source_filename, seg_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add seg fn to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + debug(epub_output) { + string fn_dbg = pth_epub3.dbg_fn_oebps_content_xhtml(doc_matters.source_filename, seg_filename); + auto f = File(fn_dbg, "w"); + } + /+ // f.writeln(seg_head); // not needed built and inserted earlier +/ + foreach (docseg; doc_epub3[seg_filename]) { + debug(epub_output) { f.writeln(docseg); } + zip_data.write(docseg.dup); // cast as: char[] + } + foreach (docseg; doc_epub3_endnotes[seg_filename]) { + debug(epub_output) { f.writeln(docseg); } + zip_data.write(docseg.dup); // cast as: char[] + } + debug(epub_output) { f.writeln(xhtml_format.tail); } // needed for each lev4 + zip_data.write(xhtml_format.tail.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + } + string fn; + debug(epub_output) { string fn_dbg; } + File f; + { /+ mimetypes (identify zip file type) +/ + debug(epub_output) { + fn_dbg = pth_epub3.dbg_fn_mimetypes(doc_matters.source_filename); + File(fn_dbg, "w").writeln(mimetypes); + } + fn = pth_epub3.fn_mimetypes(doc_matters.source_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add mimetypes to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + zip_data.write(mimetypes.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + { /+ META-INF/container.xml (identify doc root) +/ + debug(epub_output) { + fn_dbg = pth_epub3.dbg_fn_dmi_container_xml(doc_matters.source_filename); + File(fn_dbg, "w").writeln(meta_inf_container_xml); + } + fn = pth_epub3.fn_dmi_container_xml(doc_matters.source_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add META-INF/container.xml to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + zip_data.write(meta_inf_container_xml.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + { /+ OEBPS/toc_nav.xhtml (navigation toc epub3) +/ + debug(epub_output) { + fn_dbg = pth_epub3.dbg_fn_oebps_toc_nav_xhtml(doc_matters.source_filename); + File(fn_dbg, "w").writeln(oebps_toc_nav_xhtml); + } + fn = pth_epub3.fn_oebps_toc_nav_xhtml(doc_matters.source_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add OEBPS/toc_nav.xhtml to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + zip_data.write(oebps_toc_nav_xhtml.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + { /+ TODO OEBPS/toc.ncx (navigation toc epub2) +/ + debug(epub_output) { + fn_dbg = pth_epub3.dbg_fn_oebps_toc_ncx(doc_matters.source_filename); + File(fn_dbg, "w").writeln(oebps_toc_ncx); + } + fn = pth_epub3.fn_oebps_toc_ncx(doc_matters.source_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add OEBPS/toc.ncx to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + zip_data.write(oebps_toc_ncx.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + { /+ TODO OEBPS/content.opf (doc manifest) +/ + debug(epub_output) { + fn_dbg = pth_epub3.dbg_fn_oebps_content_opf(doc_matters.source_filename); + File(fn_dbg, "w").writeln(oebps_content_opf); + } + fn = pth_epub3.fn_oebps_content_opf(doc_matters.source_filename); + /+ add zip archive file members (with their content) +/ + auto zip_arc_member_file = new ArchiveMember(); + // add OEBPS/content.opf to zip archive + zip_arc_member_file.name = fn; + auto zip_data = new OutBuffer(); + zip_data.write(oebps_content_opf.dup); // cast as: char[] + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + /+ create the zip file +/ + createZipFile!()(fn_epub, zip.build()); + } + { /+ OEBPS/_sisu/image (images) +/ + foreach (image; doc_matters.image_list) { + if (exists("_sisu/image/"~ image)) { + ("_sisu/image/"~ image) + .copy((pth_epub3.doc_oebps_image(doc_matters.source_filename)) ~ "/" ~ image); + } + } + foreach (image; doc_matters.image_list) { + debug(epub_images) { + writeln( + "_sisu/image/", image, " -> ", + pth_epub3.doc_oebps_image(doc_matters.source_filename), "/", image + ); + } + auto fn_src = "_sisu/image/"~ image; + auto fn_out = pth_epub3.doc_oebps_image(doc_matters.source_filename).to!string ~ "/" ~ image; + if (exists(fn_src)) { + { + auto zip_arc_member_file = new ArchiveMember(); + zip_arc_member_file.name = fn_out; + auto zip_data = new OutBuffer(); + zip_data.write(cast(char[]) ((fn_src).read)); + zip_arc_member_file.expandedData = zip_data.toBytes(); + zip.addMember(zip_arc_member_file); + createZipFile!()(fn_epub, zip.build()); + } + } + } + } + } + catch (ErrnoException ex) { + // Handle error + } + debug(epub_archive) { + if (exists(fn_epub)) { + try { + auto zipped = new ZipArchive((fn_epub).read); + foreach (filename, member; zipped.directory) { + auto data = zipped.expand(member); + writeln(filename, " length ", data.length); // member.name + // Use data + } + } + catch (ZipException ex) { + // Handle errors + } + } + } + } + +} -- cgit v1.2.3