isodoc 1.8.3.1 → 2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ee178bd5fb88dd6ee3302731a8325eec443d80aac78adb9b9bd166157ffb4ebb
-  data.tar.gz: 95d464d67887b5ea80dd507a66815107c7e9c99d2e7ab4e11f86467524b5a9f2
+  metadata.gz: 94378642fdbff3fdfd64fc15d3f502fd3ac966926d19b060c5b7bbd4b9f9e545
+  data.tar.gz: eeb1da05b0df09ea66cd55692f3d2f706991904216c5869025b9d936dc93cf12
 SHA512:
-  metadata.gz: 8c5501c94696d11e07b7e7e0689eb3c4ae46917e7494f4c64407b150ba1171fb7d5c8723e59ab88c01f94f733a9880f8a40b6668c5c364f058d9865964957378
-  data.tar.gz: 6a4d747161c3a88b7be6034e82876d5aafd3c500ba1dffdd5f4359382da35285738f2a3fb7c3351892eb5472459054e5f2010dc0e769077052c5e1bd4a7b03ec
+  metadata.gz: b8164bf9d471f1b409ff4c0d933f69b9f834d340548caf3b441d7ca4584460b026420bb752903121561d573f8ddc618e5796d5e22dd09c1a9f3ca429cf567df5
+  data.tar.gz: 75a73a94378f4d0fe6187b343d84961265e60d1417d23c43c38b11c819190cba0838dd0a99ba21d7e49824add339f7a6007c42748f87130a7abd2ce35d14105b

data/.github/workflows/rake.yml

@@ -10,34 +10,6 @@ on:
 
 jobs:
   rake:
-    name: Test on Ruby ${{ matrix.ruby }} ${{ matrix.os }}
-    runs-on: ${{ matrix.os }}
-    continue-on-error: ${{ matrix.experimental }}
-    strategy:
-      fail-fast: false
-      matrix:
-        ruby: [ '3.0', '2.7', '2.6', '2.5' ]
-        os: [ ubuntu-latest, windows-latest, macos-latest ]
-        experimental: [ false ]
-    steps:
-      - uses: actions/checkout@master
-
-      - uses: ruby/setup-ruby@v1
-        with:
-          ruby-version: ${{ matrix.ruby }}
-          bundler-cache: true
-
-      - uses: metanorma/metanorma-build-scripts/inkscape-setup-action@master
-
-      - run: bundle exec rake
-
-  tests-passed:
-    needs: rake
-    runs-on: ubuntu-latest
-    steps:
-      - uses: peter-evans/repository-dispatch@v1
-        with:
-          token: ${{ secrets.METANORMA_CI_PAT_TOKEN || secrets.GITHUB_TOKEN }}
-          repository: ${{ github.repository }}
-          event-type: tests-passed
-          client-payload: '{"ref": "${{ github.ref }}", "sha": "${{ github.sha }}"}'
+    uses: metanorma/metanorma-build-scripts/.github/workflows/inkscape-rake.yml@main
+    secrets:
+      pat_token: ${{ secrets.METANORMA_CI_PAT_TOKEN }}

data/lib/isodoc/base_style/all.css

@@ -68,6 +68,12 @@ code *, pre *, tt *, kbd *, samp * {
   font-family: {{monospacefont}} !important;
   font-variant-ligatures: none; }
 
+p code, dt code, li code, label code, legend code, caption code, th code, td code,
+p tt, dt tt, li tt, label tt, legend tt, caption tt, th tt, td tt,
+p kbd, dt kbd, li kbd, label kbd, legend kbd, caption kbd, th kbd, td kbd,
+p samp, dt samp, li samp, label samp, legend samp, caption samp, th samp, td samp {
+  font-size: {{monospacefontsize}}; }
+
 article, aside, details, figcaption, figure,
 footer, header, hgroup, menu, nav, section {
   display: block; }
@@ -79,6 +85,9 @@ table {
 h1, h2, h3, h4, h5, h6 {
   font-family: {{headerfont}}; }
 
+.h1, .h2, .h3, .h4, .h5, .h6 {
+  font-family: {{headerfont}}; }
+
 blockquote, q {
   quotes: none; }
   blockquote:before, blockquote:after, q:before, q:after {

data/lib/isodoc/base_style/reset.css

@@ -68,6 +68,12 @@ code *, pre *, tt *, kbd *, samp * {
   font-family: {{monospacefont}} !important;
   font-variant-ligatures: none; }
 
+p code, dt code, li code, label code, legend code, caption code, th code, td code,
+p tt, dt tt, li tt, label tt, legend tt, caption tt, th tt, td tt,
+p kbd, dt kbd, li kbd, label kbd, legend kbd, caption kbd, th kbd, td kbd,
+p samp, dt samp, li samp, label samp, legend samp, caption samp, th samp, td samp {
+  font-size: {{monospacefontsize}}; }
+
 article, aside, details, figcaption, figure,
 footer, header, hgroup, menu, nav, section {
   display: block; }
@@ -79,6 +85,9 @@ table {
 h1, h2, h3, h4, h5, h6 {
   font-family: {{headerfont}}; }
 
+.h1, .h2, .h3, .h4, .h5, .h6 {
+  font-family: {{headerfont}}; }
+
 blockquote, q {
   quotes: none; }
   blockquote:before, blockquote:after, q:before, q:after {

data/lib/isodoc/base_style/reset.scss

@@ -74,6 +74,13 @@ code *, pre *, tt *, kbd *, samp * {
   font-variant-ligatures: none;
 }
 
+p code, dt code, li code, label code, legend code, caption code, th code, td code, 
+p tt, dt tt, li tt, label tt, legend tt, caption tt, th tt, td tt, 
+p kbd, dt kbd, li kbd, label kbd, legend kbd, caption kbd, th kbd, td kbd, 
+p samp, dt samp, li samp, label samp, legend samp, caption samp, th samp, td samp {
+  font-size: $monospacefontsize;
+}
+
 article, aside, details, figcaption, figure,
 footer, header, hgroup, menu, nav, section {
   display: block;
@@ -88,6 +95,10 @@ h1, h2, h3, h4, h5, h6 {
   font-family: $headerfont;
 }
 
+.h1, .h2, .h3, .h4, .h5, .h6 {
+  font-family: $headerfont;
+}
+
 blockquote, q {
   quotes: none;
 

data/lib/isodoc/class_utils.rb

@@ -28,25 +28,37 @@ module IsoDoc
       Liquid::Template.parse(doc)
     end
 
-    def case_strict(text, casing, script)
+    def case_strict(text, casing, script, firstonly: true)
       return text unless %w(Latn Cyrl Grek Armn).include?(script)
 
-      letters = text.chars
+      seen = false
+      text.split(/(\s+)/).map do |w|
+        letters = w.chars
+        case_strict1(letters, casing) if !seen || !firstonly
+        seen ||= /\S/.match?(w)
+        letters.join
+      end.join
+    end
+
+    def case_strict1(letters, casing)
+      return letters if letters.empty?
+
       case casing
       when "capital" then letters.first.upcase!
       when "lowercase" then letters.first.downcase!
+      when "allcaps" then letters.map(&:upcase!)
       end
-      letters.join
     end
 
-    def case_with_markup(linkend, casing, script)
+    def case_with_markup(linkend, casing, script, firstonly: true)
       seen = false
       xml = Nokogiri::XML("<root>#{linkend}</root>")
       xml.traverse do |b|
         next unless b.text? && !seen
 
-        b.replace(Common::case_strict(b.text, casing, script))
-        seen = true
+        b.replace(Common::case_strict(b.text, casing, script,
+                                      firstonly: firstonly))
+        seen = true if firstonly
       end
       xml.root.children.to_xml
     end

data/lib/isodoc/convert.rb

@@ -5,6 +5,7 @@ require "fileutils"
 require "tempfile"
 require_relative "i18n"
 require_relative "css"
+require "securerandom"
 
 module IsoDoc
   class Convert < ::IsoDoc::Common
@@ -103,10 +104,16 @@ module IsoDoc
       @fn_bookmarks = {}
       @baseassetpath = options[:baseassetpath]
       @aligncrosselements = options[:aligncrosselements]
+      @tmpimagedir_suffix = tmpimagedir_suffix
+      @tmpfilesdir_suffix = tmpfilesdir_suffix
     end
 
     def tmpimagedir_suffix
-      "_images"
+      "_#{SecureRandom.hex(8)}_images"
+    end
+
+    def tmpfilesdir_suffix
+      "_#{SecureRandom.hex(8)}_files"
     end
 
     def html_doc_path(*file)

data/lib/isodoc/function/blocks.rb

@@ -174,10 +174,13 @@ module IsoDoc
         end
       end
 
-      def para_class(_node)
+      def para_class(node)
         classtype = nil
         classtype = "MsoCommentText" if in_comment
         classtype = "Sourcecode" if @annotation
+        if node["type"] == "floating-title"
+          classtype = "h#{node['depth']}"
+        end
         classtype
       end
 

data/lib/isodoc/function/references.rb

@@ -14,27 +14,28 @@ module IsoDoc
       def nonstd_bibitem(list, bib, ordinal, biblio)
         list.p **attr_code(iso_bibitem_entry_attrs(bib, biblio)) do |ref|
           ids = bibitem_ref_code(bib)
-          identifiers = render_identifier(ids)
-          if biblio then ref_entry_code(ref, ordinal, identifiers, ids)
+          idents = render_identifier(ids)
+          if biblio then ref_entry_code(ref, ordinal, idents, ids)
           else
-            ref << (identifiers[0] || identifiers[1]).to_s
-            ref << ", #{identifiers[1]}" if identifiers[0] && identifiers[1]
+            ref << (idents[:ordinal] || idents[:metanorma] || idents[:sdo]).to_s
+            ref << ", #{idents[sdo]}" if idents[:ordinal] && idents[:sdo]
           end
-          ref << ", " unless biblio && !identifiers[1]
+          ref << ", " unless biblio && !idents[:sdo]
           reference_format(bib, ref)
         end
       end
 
       def std_bibitem_entry(list, bib, ordinal, biblio)
         list.p **attr_code(iso_bibitem_entry_attrs(bib, biblio)) do |ref|
-          identifiers = render_identifier(bibitem_ref_code(bib))
-          if biblio then ref_entry_code(ref, ordinal, identifiers, nil)
+          idents = render_identifier(bibitem_ref_code(bib))
+          if biblio then ref_entry_code(ref, ordinal, idents, nil)
           else
-            ref << (identifiers[0] || identifiers[1]).to_s
-            ref << ", #{identifiers[1]}" if identifiers[0] && identifiers[1]
+            ref << (idents[:ordinal] || idents[:metanorma] || idents[:sdo]).to_s
+            ref << ", #{idents[:sdo]}" if (idents[:ordinal] ||
+                                          idents[:metanorma]) && idents[:sdo]
           end
           date_note_process(bib, ref)
-          ref << ", " unless biblio && !identifiers[1]
+          ref << ", " unless biblio && !idents[:sdo]
           reference_format(bib, ref)
         end
       end
@@ -42,13 +43,15 @@ module IsoDoc
       # if ids is just a number, only use that ([1] Non-Standard)
       # else, use both ordinal, as prefix, and ids
       def ref_entry_code(ref, ordinal, ids, _id)
-        prefix_bracketed_ref(ref, ids[0] || "[#{ordinal}]")
-        ids[1] and ref << (ids[1]).to_s
+        prefix_bracketed_ref(ref, ids[:ordinal] || ids[:metanorma] ||
+                             "[#{ordinal}]")
+        ids[:sdo] and ref << (ids[:sdo]).to_s
       end
 
       def pref_ref_code(bib)
         bib.at(ns("./docidentifier[not(@type = 'DOI' or @type = 'metanorma' "\
-                  "or @type = 'ISSN' or @type = 'ISBN' or "\
+                  "or @type = 'metanorma-ordinal' or "\
+                  "@type = 'ISSN' or @type = 'ISBN' or "\
                   "@type = 'rfc-anchor')]"))
       end
 
@@ -58,11 +61,12 @@ module IsoDoc
         id1 = pref_ref_code(bib)
         id2 = bib.at(ns("./docidentifier[@type = 'DOI' or @type = 'ISSN' or "\
                         "@type = 'ISBN']"))
-        return [id, id1, id2] if id || id1 || id2
+        id3 = bib.at(ns("./docidentifier[@type = 'metanorma-ordinal']"))
+        return [id, id1, id2, id3] if id || id1 || id2 || id3
 
         id = Nokogiri::XML::Node.new("docidentifier", bib.document)
         id << "(NO ID)"
-        [nil, id, nil]
+        [nil, id, nil, nil]
       end
 
       def bracket_if_num(num)
@@ -74,10 +78,15 @@ module IsoDoc
         num
       end
 
+      def unbracket(ident)
+        ident&.text&.sub(/^\[/, "")&.sub(/\]$/, "")
+      end
+
       def render_identifier(ident)
-        [bracket_if_num(ident[0]),
-         ident[1].nil? ? nil : ident[1].text.sub(/^\[/, "").sub(/\]$/, ""),
-         ident[2].nil? ? nil : ident[2].text.sub(/^\[/, "").sub(/\]$/, "")]
+        { metanorma: bracket_if_num(ident[0]),
+          sdo: unbracket(ident[1]),
+          doi: unbracket(ident[2]),
+          ordinal: bracket_if_num(ident[3]) }
       end
 
       def docid_prefix(prefix, docid)
@@ -133,7 +142,7 @@ module IsoDoc
         end
       end
 
-      def is_standard(bib)
+      def standard?(bib)
         ret = false
         drop = %w(metanorma DOI ISSN ISBN)
         bib.xpath(ns("./docidentifier")).each do |id|
@@ -151,7 +160,7 @@ module IsoDoc
             next if implicit_reference(b)
 
             i += 1
-            if is_standard(b) then std_bibitem_entry(div, b, i, biblio)
+            if standard?(b) then std_bibitem_entry(div, b, i, biblio)
             else nonstd_bibitem(div, b, i, biblio)
             end
           else
@@ -166,7 +175,7 @@ module IsoDoc
       end
 
       def norm_ref(isoxml, out, num)
-        f = isoxml.at(ns(norm_ref_xpath)) and f["hidden"] != "true" or
+        (f = isoxml.at(ns(norm_ref_xpath)) and f["hidden"] != "true") or
           return num
         out.div do |div|
           num = num + 1
@@ -186,7 +195,7 @@ module IsoDoc
       end
 
       def bibliography(isoxml, out)
-        f = isoxml.at(ns(bibliography_xpath)) and f["hidden"] != "true" or
+        (f = isoxml.at(ns(bibliography_xpath)) and f["hidden"] != "true") or
           return
         page_break(out)
         out.div do |div|

data/lib/isodoc/function/section_titles.rb

@@ -76,7 +76,12 @@ module IsoDoc
         end
       end
 
-      def variant_title(_node, _out); end
+      def variant_title(node, out)
+        out.p **attr_code(style: "display:none;",
+                          class: "variant-title-#{node['type']}") do |p|
+          node.children.each { |c| parse(c, p) }
+        end
+      end
     end
   end
 end

data/lib/isodoc/function/to_word_html.rb

@@ -30,7 +30,7 @@ module IsoDoc
       end
 
       def init_dir(filename, debug)
-        dir = "#{filename}_files"
+        dir = "#{filename}#{@tmpfilesdir_suffix}"
         unless debug
           Dir.mkdir(dir, 0o777) unless File.exists?(dir)
           FileUtils.rm_rf "#{dir}/*"
@@ -40,11 +40,11 @@ module IsoDoc
 
       # tmp image dir is same directory as @filename
       def tmpimagedir
-        @filename + tmpimagedir_suffix
+        @filename + @tmpimagedir_suffix
       end
 
       def rel_tmpimagedir
-        Pathname.new(@filename).basename.to_s + tmpimagedir_suffix
+        Pathname.new(@filename).basename.to_s + @tmpimagedir_suffix
       end
 
       # isodoc.css overrides any CSS injected by Html2Doc, which

data/lib/isodoc/function/utils.rb

@@ -129,6 +129,7 @@ module IsoDoc
 
       def header_strip(hdr)
         h1 = to_xhtml_fragment(hdr.to_s.gsub(%r{<br\s*/>}, " ")
+          .gsub(%r{</?p(\s[^>]+)?>}, "")
           .gsub(/<\/?h[123456][^>]*>/, "").gsub(/<\/?b[^>]*>/, "").dup)
         h1.traverse do |x|
           if x.name == "span" && /mso-tab-count/.match(x["style"])
@@ -142,9 +143,9 @@ module IsoDoc
 
       def header_strip_elem?(elem)
         elem.name == "img" ||
-          elem.name == "span" && elem["class"] == "MsoCommentReference" ||
-          elem.name == "a" && elem["class"] == "FootnoteRef" ||
-          elem.name == "span" && /mso-bookmark/.match(elem["style"])
+          (elem.name == "span" && elem["class"] == "MsoCommentReference") ||
+          (elem.name == "a" && elem["class"] == "FootnoteRef") ||
+          (elem.name == "span" && /mso-bookmark/.match(elem["style"]))
       end
 
       def liquid(doc)
@@ -192,12 +193,6 @@ module IsoDoc
           save_dataimage(img["src"], false)
         when %r{^([A-Z]:)?/}
           img["src"]
-=begin
-        if /^data:/.match? img["src"]
-          save_dataimage(img["src"], false)
-        elsif %r{^([A-Z]:)?/}.match? img["src"]
-          img["src"]
-=end
         else
           File.join(@localdir, img["src"])
         end
@@ -213,12 +208,16 @@ module IsoDoc
            application/x-msmetafile image/x-xbitmap).include? type
       end
 
-      def cleanup_entities(text)
+      def cleanup_entities(text, is_xml: true)
         c = HTMLEntities.new
-        text.split(/([<>])/).each_slice(4).map do |a|
-          a[0] = c.encode(c.decode(a[0]), :hexadecimal)
-          a
-        end.join
+        if is_xml
+          text.split(/([<>])/).each_slice(4).map do |a|
+            a[0] = c.encode(c.decode(a[0]), :hexadecimal)
+            a
+          end.join
+        else
+          c.encode(c.decode(text), :hexadecimal)
+        end
       end
 
       def external_path(path)

data/lib/isodoc/headlesshtml_convert.rb

@@ -10,7 +10,7 @@ module IsoDoc
     include HtmlFunction::Html
 
     def tmpimagedir_suffix
-      "_headlessimages"
+      "_#{SecureRandom.hex(8)}_headlessimages"
     end
 
     def initialize(options)

data/lib/isodoc/html_convert.rb

@@ -13,7 +13,7 @@ module IsoDoc
     include HtmlFunction::Html
 
     def tmpimagedir_suffix
-      "_htmlimages"
+      "_#{SecureRandom.hex(8)}_htmlimages"
     end
 
     def initialize(options)

data/lib/isodoc/html_function/postprocess.rb

@@ -122,8 +122,10 @@ module IsoDoc
       end
 
       def html_toc_entry(level, header)
+        content = header.at("./following-sibling::p"\
+                            "[@class = 'variant-title-toc']") || header
         %(<li class="#{level}"><a href="##{header['id']}">\
-      #{header_strip(header)}</a></li>)
+      #{header_strip(content)}</a></li>)
       end
 
       def toclevel_classes
@@ -144,7 +146,7 @@ module IsoDoc
         idx = docxml.at("//div[@id = 'toc']") or return docxml
         toc = "<ul>"
         path = toclevel_classes.map do |l|
-          "//main//#{l}[not(@class = 'TermNum')][not(@class = 'noTOC')][text()]"
+          "//main//#{l}#{toc_exclude_class}"
         end
         docxml.xpath(path.join(" | ")).each_with_index do |h, tocidx|
           h["id"] ||= "toc#{tocidx}"
@@ -154,6 +156,11 @@ module IsoDoc
         docxml
       end
 
+      def toc_exclude_class
+        "[not(@class = 'TermNum')][not(@class = 'noTOC')]"\
+          "[string-length(normalize-space(.))>0]"
+      end
+
       # presupposes that the image source is local
       def move_images(docxml)
         FileUtils.rm_rf tmpimagedir

data/lib/isodoc/i18n.rb

@@ -1,4 +1,5 @@
 require "yaml"
+require_relative "function/utils"
 
 module IsoDoc
   class I18n
@@ -17,7 +18,7 @@ module IsoDoc
         end
         ret
       when Array then ret.map { |n| normalise_hash(n) }
-      when String then ret.unicode_normalize(:nfc)
+      when String then cleanup_entities(ret.unicode_normalize(:nfc))
       else ret
       end
     end
@@ -96,8 +97,9 @@ module IsoDoc
       xml.traverse do |n|
         next unless n.text?
 
-        n.replace(n.text.gsub(/ /, "").gsub(/:/, "：").gsub(/,/, "、")
-          .gsub(/\(/, "（").gsub(/\)/, "）").gsub(/\[/, "【").gsub(/\]/, "】"))
+        n.replace(cleanup_entities(n.text.gsub(/ /, "").gsub(/:/, "：")
+          .gsub(/,/, "、").gsub(/\(/, "（").gsub(/\)/, "）")
+          .gsub(/\[/, "【").gsub(/\]/, "】"), is_xml: false))
       end
       xml.to_xml.gsub(/<b>/, "").gsub("</b>", "").gsub(/<\?[^>]+>/, "")
     end
@@ -111,6 +113,7 @@ module IsoDoc
       l10n(names[0..-2].join(", ") + " #{andword} #{names[-1]}", @lang, @script)
     end
 
+    include Function::Utils
     # module_function :l10n
   end
 end

data/lib/isodoc/pdf_convert.rb

@@ -20,7 +20,7 @@ module IsoDoc
     end
 
     def tmpimagedir_suffix
-      "_pdfimages"
+      "_#{SecureRandom.hex(8)}_pdfimages"
     end
 
     def convert(input_filename, file = nil, debug = false, output_filename = nil)

data/lib/isodoc/presentation_function/bibdata.rb

@@ -10,12 +10,6 @@ module IsoDoc
         "</localized-strings>"
     end
 
-    def docid_prefixes(docxml)
-      docxml.xpath(ns("//references/bibitem/docidentifier")).each do |i|
-        i.children = @xrefs.klass.docid_prefix(i["type"], i.text)
-      end
-    end
-
     def address_precompose(bib)
       bib.xpath(ns("//bibdata//address")).each do |b|
         next if b.at(ns("./formattedAddress"))

data/lib/isodoc/presentation_function/section.rb

@@ -19,6 +19,22 @@ module IsoDoc
       prefix_name(elem, "<tab/>", "#{lbl}#{clausedelim}", "title")
     end
 
+    def floattitle(docxml)
+      docxml.xpath(ns("//clause | //annex | //appendix | //introduction | "\
+                      "//foreword | //preface/abstract | //acknowledgements | "\
+                      "//terms | //definitions | //references"))
+        .each do |f|
+        floattitle1(f)
+      end
+    end
+
+    def floattitle1(elem)
+      elem.xpath(ns(".//floating-title")).each do |p|
+        p.name = "p"
+        p["type"] = "floating-title"
+      end
+    end
+
     def annex(docxml)
       docxml.xpath(ns("//annex")).each do |f|
         annex1(f)
@@ -44,7 +60,29 @@ module IsoDoc
       prefix_name(elem, "", "#{lbl}#{clausedelim}", "name")
     end
 
-    def references(docxml); end
+    def references(docxml)
+      bibliography_bibitem_number(docxml)
+    end
+
+    def bibliography_bibitem_number_skip(bibitem)
+      @xrefs.klass.implicit_reference(bibitem) ||
+        bibitem.at(ns(".//docidentifier[@type = 'metanorma']")) ||
+        bibitem.at(ns(".//docidentifier[@type = 'metanorma-ordinal']"))
+      # || @xrefs.klass.standard?(bibitem)
+    end
+
+    def bibliography_bibitem_number(docxml)
+      i = 0
+      docxml.xpath(ns("//references[@normative = 'false']/bibitem")).each do |b|
+        next if bibliography_bibitem_number_skip(b)
+
+        i += 1
+        id = b.at(ns(".//docidentifier"))
+        id.previous =
+          "<docidentifier type='metanorma-ordinal'>[#{i}]</docidentifier>"
+      end
+      @xrefs.references docxml
+    end
 
     def docid_prefixes(docxml)
       docxml.xpath(ns("//references/bibitem/docidentifier")).each do |i|
@@ -86,9 +124,7 @@ module IsoDoc
       display_order_xpath(docxml, "//indexsect", i)
     end
 
-    def clausetitle(docxml)
-      docxml.xpath(ns("//variant-title[@type = 'toc']")).each(&:remove)
-    end
+    def clausetitle(docxml); end
 
     def toc(docxml)
       docxml.xpath(ns("//toc//xref[text()]")).each do |x|