isodoc 3.4.5 → 3.4.7

data/lib/isodoc/function/terms.rb

@@ -7,7 +7,7 @@ module IsoDoc
 
       def modification_parse(node, out)
         para = node.at(ns("./p"))
-        para.children.each { |n| parse(n, out) }
+        children_parse(para, out)
       end
 
       def semx_deprecated_term_parse(node, out); end
@@ -16,7 +16,7 @@ module IsoDoc
         node.children.each do |c|
           if c.name == "p"
             out.p class: "DeprecatedTerms", style: "text-align:left;" do |p|
-              c.children.each { |c1| parse(c1, p) }
+              children_parse(c, p)
             end
           else parse(c, out)
           end
@@ -29,7 +29,7 @@ module IsoDoc
         node.children.each do |c|
           if c.name == "p"
             out.p class: "AltTerms", style: "text-align:left;" do |p|
-              c.children.each { |c1| parse(c1, p) }
+              children_parse(c, p)
             end
           else parse(c, out)
           end
@@ -38,7 +38,7 @@ module IsoDoc
 
       def related_parse(node, out)
         out.div class: "RelatedTerms", style: "text-align:left;" do |p|
-          node.children.each { |c1| parse(c1, p) }
+          children_parse(node, p)
         end
       end
 
@@ -49,28 +49,33 @@ module IsoDoc
       def term_parse(node, out)
         node.children.each do |c|
           if c.name == "p"
-            out.p class: "Terms", style: "text-align:left;" do |p|
-              c.children.each { |c1| parse(c1, p) }
-            end
-          else parse(c, out)
+            term_p_parse(c, out)
+          else
+            parse(c, out)
           end
         end
       end
 
+      def term_p_parse(node, out)
+        out.p class: "Terms", style: "text-align:left;" do |p|
+          children_parse(node, p)
+        end
+      end
+
       def termnote_p_class
         nil
       end
 
       def termnote_parse(node, out)
         para = block_body_first_elem(node)
-        out.div **note_attrs(node) do |div|
+        out.div(**note_attrs(node)) do |div|
           termnote_parse1(node, para, div)
           para&.xpath("./following-sibling::*")&.each { |n| parse(n, div) }
         end
       end
 
       def termnote_parse1(node, para, div)
-        div.p **attr_code(class: termnote_p_class) do |p|
+        div.p(**attr_code(class: termnote_p_class)) do |p|
           name = node.at(ns("./fmt-name")) and
             p.span class: "termnote_label" do |s|
               children_parse(name, s)

data/lib/isodoc/function/to_word_html.rb

@@ -22,7 +22,7 @@ module IsoDoc
       end
 
       def make_body(xml, docxml)
-        xml.body **body_attr do |body|
+        xml.body(**body_attr) do |body|
           make_body1(body, docxml)
           make_body2(body, docxml)
           make_body3(body, docxml)

data/lib/isodoc/function/utils.rb

@@ -100,8 +100,7 @@ module IsoDoc
         array.nil? || array.empty? and return ""
         if array.length == 1 then array[0]
         else
-          @i18n.l10n("#{array[0..-2].join(', ')} " \
-                     "#{@i18n.and} #{array.last}",
+          @i18n.l10n("#{array[0..-2].join(', ')} #{@i18n.and} #{array.last}",
                      @lang, @script)
         end
       end
@@ -141,7 +140,7 @@ module IsoDoc
       def liquid(doc)
         # unescape HTML escapes in doc
         doc = doc.split(%r<(\{%|%\})>).each_slice(4).map do |a|
-          a[2] = a[2].gsub(/&lt;/, "<").gsub(/&gt;/, ">") if a.size > 2
+          a[2] = a[2].gsub("&lt;", "<").gsub("&gt;", ">") if a.size > 2
           a.join
         end.join
         Liquid::Template.parse(doc)
@@ -223,8 +222,7 @@ module IsoDoc
             a[0] = c.encode(c.decode(a[0]), :hexadecimal)
             a
           end.join
-        else
-          c.encode(c.decode(text), :hexadecimal)
+        else c.encode(c.decode(text), :hexadecimal)
         end
       end
 
@@ -239,6 +237,23 @@ module IsoDoc
       def imgfile_suffix(uri, suffix)
         "#{File.join(File.dirname(uri), File.basename(uri, '.*'))}.#{suffix}"
       end
+
+      # Unescape &#x26; to & in href attributes only
+      # This ensures URLs work correctly while preserving &#x26; in text content
+      # This operates on the final string output after all Nokogiri processing
+      def unescape_amp_in_hrefs(html)
+        # Match href="..." and href='...' separately
+        # Note: populate_template converts &amp; to &#x26;, so we replace that
+        html.gsub(/(href\s*=\s*")([^"]*)"|(href\s*=\s*')([^']*)'/) do
+          if Regexp.last_match(1)
+            "#{Regexp.last_match(1)}#{Regexp.last_match(2).gsub('&#x26;',
+                                                                '&')}\""
+          else
+            "#{Regexp.last_match(3)}#{Regexp.last_match(4).gsub('&#x26;',
+                                                                '&')}'"
+          end
+        end
+      end
     end
   end
 end

data/lib/isodoc/html_function/form.rb

@@ -2,53 +2,53 @@ module IsoDoc
   module HtmlFunction
     module Form
       def form_parse(node, out)
-        out.form **attr_code(id: node["id"], name: node["name"],
+        out.form(**attr_code(id: node["id"], name: node["name"],
                              class: node["class"],
-                             action: node["action"]) do |div|
+                             action: node["action"])) do |div|
           children_parse(node, div)
         end
       end
 
       def input_parse(node, out)
-        out.input nil, **attr_code(
+        out.input(nil, **attr_code(
           id: node["id"], name: node["name"], type: node["type"],
           value: node["value"], disabled: node["disabled"],
           readonly: node["readonly"], checked: node["checked"],
           maxlength: node["maxlength"], minlength: node["minlength"]
-        )
+        ))
       end
 
       def select_parse(node, out)
         selected = node.at(ns("./option[@value = '#{node['value']}']"))
         selected and selected["selected"] = true
-        out.select **attr_code(
+        out.select(**attr_code(
           id: node["id"], name: node["name"], size: node["size"],
           disabled: node["disabled"], multiple: node["multiple"]
-        ) do |div|
+        )) do |div|
           children_parse(node, div)
         end
       end
 
       def label_parse(node, out)
-        out.label **attr_code(for: node["for"]) do |div|
+        out.label(**attr_code(for: node["for"])) do |div|
           children_parse(node, div)
         end
       end
 
       def option_parse(node, out)
-        out.option **attr_code(
+        out.option(**attr_code(
           disabled: node["disabled"], selected: node["selected"],
           value: node["value"]
-        ) do |o|
+        )) do |o|
           children_parse(node, o)
         end
       end
 
       def textarea_parse(node, out)
-        out.textarea **attr_code(
+        out.textarea(**attr_code(
           id: node["id"], name: node["name"], rows: node["rows"],
           cols: node["cols"]
-        ) do |div|
+        )) do |div|
           node["value"] and div << node["value"]
         end
       end

data/lib/isodoc/html_function/html.rb

@@ -16,8 +16,7 @@ module IsoDoc
       end
 
       def make_body1(body, _docxml)
-        return if @bare
-
+        @bare and return
         body.div class: "title-section" do |div1|
           div1.p { |p| p << "&#xa0;" } # placeholder
         end
@@ -25,8 +24,7 @@ module IsoDoc
       end
 
       def make_body2(body, _docxml)
-        return if @bare
-
+        @bare and return
         body.div class: "prefatory-section" do |div2|
           div2.p { |p| p << "&#xa0;" } # placeholder
         end
@@ -60,36 +58,68 @@ module IsoDoc
       end
 
       def html_button
-        return "" if @bare
-
+        @bare and return ""
         '<button onclick="topFunction()" id="myBtn" ' \
         'title="Go to top">Top</button>'.freeze
       end
 
       def html_main(docxml)
-        docxml.at("//head").add_child(html_head)
+        docxml.at("//head") << Nokogiri::HTML.fragment(html_head)
         d = docxml.at('//div[@class="main-section"]')
         d.name = "main"
         d.children.empty? or d.children.first.previous = html_button
       end
 
       def sourcecode_parse(node, out)
-        name = node.at(ns("./fmt-name"))
-        tag = node.at(ns(".//sourcecode | .//table")) ? "div" : "pre"
-        n = node.at(ns("./fmt-sourcecode"))
-        s = n || node
+        tag, child_tag = sourcecode_tag(node)
+        s = node.at(ns("./fmt-sourcecode")) || node
         attr = sourcecode_attrs(node).merge(class: "sourcecode")
         out.send tag, **attr do |div|
-          sourcecode_parse1(s, div)
+          sourcecode_pre_wrap(child_tag, s, div)
+          annotation_parse(s, div)
+          sourcecode_name_parse(node, div, node.at(ns("./fmt-name")))
+        end
+      end
+
+      def sourcecode_attrs(node)
+        super.merge(spellcheck: "false", translation: "no")
+      end
+
+      def sourcecode_tag(node)
+        ancestors = node.ancestors.map(&:name)
+          .intersection(%w(sourcecode table))
+        tag = ancestors.empty? ? "figure" : "pre"
+        child_tag = "pre"
+        tag == "pre" || node.at(ns(".//sourcecode | .//table")) and
+          child_tag = "figure"
+        [tag, child_tag]
+      end
+
+      def sourcecode_pre_wrap(tag, node, div)
+        if tag == "pre"
+          div.pre do |pre|
+            sourcecode_parse1(node, pre)
+          end
+        else
+          sourcecode_parse1(node, div)
         end
-        annotation_parse(s, out)
-        sourcecode_name_parse(node, out, name)
+      end
+
+      def sourcecode_name_parse(_node, div, name)
+        name.nil? and return
+        div.figcaption class: "SourceTitle" do |p|
+          children_parse(name, p)
+        end
+      end
+
+      def pseudocode_tag
+        "figure"
       end
 
       def underline_parse(node, out)
         style = node["style"] ? " #{node['style']}" : ""
         attr = { style: "text-decoration: underline#{style}" }
-        out.span **attr do |e|
+        out.span(**attr) do |e|
           node.children.each { |n| parse(n, e) }
         end
       end
@@ -104,24 +134,92 @@ module IsoDoc
       end
 
       def svg_supply_viewbox(svg)
-        svg["viewbox"] and return
-        svg["height"] && svg["width"] or return
+        svg_supply_viewbox?(svg) or return
         h = svg["height"].to_s[/\d+/].to_s
         w = svg["width"].to_s[/\d+/].to_s
-        h.to_i.positive? && w.to_i.positive? or return
+        (h.to_i.positive? && w.to_i.positive?) or return
         svg["viewbox"] = "0 0 #{w} #{h}"
       end
 
+      def svg_supply_viewbox?(svg)
+        svg["viewbox"] and return false
+        (svg["height"] && svg["width"]) or return false
+        true
+      end
+
+      def image_body_parse_svg(svg, out)
+        svg_supply_viewbox(svg)
+        out.div class: "svg-container" do |div|
+          div.parent.add_child(svg)
+        end
+      end
+
+      def select_altsource?(_altsource, tags)
+        tags.include?("html")
+      end
+
       def image_body_parse(node, attrs, out)
-        if svg = node.at("./m:svg", "m" => "http://www.w3.org/2000/svg")
-          svg_supply_viewbox(svg)
-          out.div class: "svg-container" do |div|
-            div.parent.add_child(svg)
+        svg = node.at(".//m:svg", "m" => "http://www.w3.org/2000/svg")
+        alts = select_altsource(node)
+        if svg && alts.empty?
+          image_body_parse_svg(svg, out)
+        else
+          image_body_parse_picture(node, attrs, alts, out)
+        end
+      end
+
+      def image_body_parse_picture(node, _attrs, alts, out)
+        out.picture do |p|
+          alts.each do |i|
+            if i["tag"] == "default" then next
+            else
+              p.source **attr_code(altsource_attrs(i)
+                .merge({ srcset: svg2datauri(i) }.compact))
+            end
+          end
+          image_picture_img(node, p)
+        end
+      end
+
+      def image_picture_img(node, out)
+        default = node.at(ns("./altsource[@tag = 'default']")) || node
+        out.img **attr_code(image_attrs(default)
+          .merge({ src: svg2datauri(default) }.compact))
+      end
+
+      def altsource_attrs(node)
+        media = node["media"]
+        media.blank? and media = "all"
+        { srcset: node["src"], media: @c.decode(media), type: node["type"] }
+      end
+
+      def svg2datauri(node)
+        svg = node.at(".//m:svg", "m" => "http://www.w3.org/2000/svg") or
+          return nil
+        v = Vectory::Svg.from_node(svg.parent)
+        Vectory::Datauri.from_vector(v).content
+      end
+
+      def semx_parse(node, out)
+        if %w(expression/name letter-symbol/name).include?(node["element"])
+          tag = designation_tag(node)
+          out.send tag do |d|
+            children_parse(node, d)
           end
         else super
         end
       end
 
+      def designation_tag(node)
+        term = node.at("./ancestor::xmlns:term")
+        if node["element"] == "expression/name"
+          e = term.at(".//*[@id = '#{node['source']}']")
+          # <expression type='abbreviation'><name>...
+          e&.parent && e.parent["type"] == "abbreviation" and return "abbr"
+        end
+        "dfn"
+      end
+
       def in_comment
         @in_comment
       end

data/lib/isodoc/html_function/postprocess.rb

@@ -8,15 +8,19 @@ module IsoDoc
     module Html
       def postprocess(result, filename, _dir)
         result = from_xhtml(cleanup(to_xhtml(textcleanup(result))))
-        toHTML(result, filename)
+        result = toHTML(result, filename)
         @files_to_delete.each { |f| FileUtils.rm_rf f }
+        result
       end
 
       def toHTML(result, filename)
         result = from_xhtml(html_cleanup(to_xhtml(result)))
         result = from_xhtml(move_images(resize_images(to_xhtml(result))))
         result = html5(script_cdata(inject_script(result)))
+        # Unescape & to & in href attributes after all Nokogiri processing
+        result = unescape_amp_in_hrefs(result)
         File.open(filename, "w:UTF-8") { |f| f.write(result) }
+        result
       end
 
       def html5(doc)
@@ -31,7 +35,7 @@ module IsoDoc
       end
 
       def heading_anchors(html)
-        html.xpath("//h1 | //h2 | //h3 | //h4 | //h5 | //h6 | //h7 | //h8 "\
+        html.xpath("//h1 | //h2 | //h3 | //h4 | //h5 | //h6 | //h7 | //h8 | " \
                    "//span[@class = 'inline-header']").each do |h|
           h.at("./ancestor::div[@id='toc']") and next
           div = h.xpath("./ancestor::div[@id]")
@@ -49,7 +53,7 @@ module IsoDoc
 
       def sourcecode_cleanup(html)
         ann = ".//div[@class = 'annotation']"
-        html.xpath("//pre[#{ann}] | //div[@class = 'sourcecode'][#{ann}]")
+        html.xpath("//pre[#{ann}] | //figure[@class = 'sourcecode'][#{ann}]")
           .each do |p|
           ins = p.after("<pre class='sourcecode'/>").next_element
           p.xpath(ann).each do |d|

data/lib/isodoc/presentation_function/designations.rb

@@ -104,7 +104,8 @@ module IsoDoc
     end
 
     def designation_annotate(desgn, name, orig)
-      designation_boldface(desgn)
+      # designation_boldface(desgn)
+      designation_expression(desgn, name, orig)
       designation_field(desgn, name, orig)
       designation_grammar(desgn, name)
       designation_localization(desgn, name, orig)
@@ -113,12 +114,32 @@ module IsoDoc
       desgn.children = name.children
     end
 
-    def designation_boldface(desgn)
+    # KILL
+    def designation_boldfacex(desgn)
       desgn["element"] == "preferred" or return
       name = desgn.at(ns("./expression/name | ./letter-symbol/name")) or return
       name.children = "<strong>#{name.children}</strong>"
     end
 
+    def designation_expression(desgn, name, orig)
+      if origname = orig.at(ns("./expression/name"))
+        element = "expression/name"
+      elsif origname = orig.at(ns("./letter-symbol/name"))
+        element = "letter-symbol/name"
+      elsif origname = orig.at(ns("./graphical-symbol"))
+        element = "graphical-symbol"
+      end
+      name.children = <<~XML.strip
+        <semx element='#{element}' source='#{origname['id'] || origname['original-id']}'>#{to_xml(name.children).strip}</semx>
+      XML
+      designation_expression_boldface(desgn, name)
+    end
+
+    def designation_expression_boldface(desgn, name)
+      name.name == "name" && desgn["element"] == "preferred" and
+        name.children = "<strong>#{to_xml(name.children)}</strong>"
+    end
+
     def designation_field(_desgn, name, orig)
       f = orig.xpath(ns("./field-of-application | ./usage-info"))
         &.map { |u| to_xml(semx_fmt_dup(u)) }&.join(", ")

data/lib/isodoc/presentation_function/ids.rb

@@ -99,7 +99,9 @@ module IsoDoc
         "//annotation | //floating-title | //li | //executivesummary | " \
         "//preface/abstract | //foreword | //introduction | //annex | " \
         "//acknowledgements | //clause | //references | //terms | " \
-        "//preferred | //deprecates | //admitted | //related")).each do |s|
+        "//preferred | //deprecates | //admitted | //related | " \
+        "//expression/name | //letter-symbol/name | //graphical-symbol"))
+        .each do |s|
         add_id(s)
       end
     end

data/lib/isodoc/presentation_function/image.rb

@@ -185,8 +185,9 @@ module IsoDoc
     end
 
     # used downstream to load svg images inline
+    # directory can be string, or array
     def svg_load(directory, filename)
-      dir = File.join(@libdir, "html", directory)
+      dir = File.join(@libdir, "html", *directory)
       filename = File.join(dir, filename)
       File.exist?(filename) or return
       file = File.read(filename) or return

data/lib/isodoc/version.rb

@@ -1,3 +1,3 @@
 module IsoDoc
-  VERSION = "3.4.5".freeze
+  VERSION = "3.4.7".freeze
 end

data/lib/isodoc/word_function/body.rb

@@ -69,7 +69,7 @@ module IsoDoc
       end
 
       def para_parse(node, out)
-        out.p **attr_code(para_attrs(node)) do |p|
+        out.p(**attr_code(para_attrs(node))) do |p|
           unless @termdomain.empty?
             p << "&#x3c;#{@termdomain}&#x3e; "
             @termdomain = ""
@@ -87,6 +87,10 @@ module IsoDoc
         "Note"
       end
 
+      def admonition_subclass(_node)
+        ""
+      end
+
       def para_attrs(node)
         attrs = { class: para_class(node), id: node["id"], style: "" }
         unless node["align"].nil?
@@ -109,7 +113,7 @@ module IsoDoc
       end
 
       def formula_parse1(node, out)
-        out.div **attr_code(class: "formula") do |div|
+        out.div(**attr_code(class: "formula")) do |div|
           div.p do |_p|
             parse(node.at(ns("./fmt-stem")), div)
             insert_tab(div, 1)
@@ -151,7 +155,7 @@ module IsoDoc
 
       def table_of_contents(clause, out)
         page_break(out)
-        out.div **attr_code(preface_attrs(clause)) do |div|
+        out.div(**attr_code(preface_attrs(clause))) do |div|
           div.p class: "zzContents" do |p|
             clause.at(ns("./fmt-title"))&.children&.each { |c| parse(c, p) }
           end
@@ -162,7 +166,7 @@ module IsoDoc
       end
 
       def figure_parse1(node, out)
-        out.div **figure_attrs(node) do |div|
+        out.div(**figure_attrs(node)) do |div|
           node.children.each do |n|
             parse(n, div) unless n.name == "fmt-name"
           end

data/lib/isodoc/word_function/inline.rb

@@ -2,7 +2,7 @@ module IsoDoc
   module WordFunction
     module Body
       def insert_tab(out, count)
-        out.span **attr_code(style: "mso-tab-count:#{count}") do |span|
+        out.span(**attr_code(style: "mso-tab-count:#{count}")) do |span|
           [1..count].each { span << "&#xA0; " }
         end
       end
@@ -46,7 +46,7 @@ module IsoDoc
         attrs = { src: imgsrc(node),
                   height: node["height"], alt: node["alt"],
                   title: node["title"], width: node["width"] }
-        out.img **attr_code(attrs)
+        image_body_parse(node, attrs, out)
       end
 
       def emf_attributes(node)
@@ -89,7 +89,7 @@ module IsoDoc
       end
 
       def rt_parse(node, out)
-        out.rt **{ style: "font-size: 6pt;" } do |e|
+        out.rt(**{ style: "font-size: 6pt;" }) do |e|
           children_parse(node, e)
         end
       end

data/lib/isodoc/word_function/lists.rb

@@ -8,7 +8,7 @@ module IsoDoc
                         style: "margin-left:0pt;text-align:left;" }.freeze
 
       def dt_parse(dterm, term)
-        term.p **attr_code(WORD_DT_ATTRS) do |p|
+        term.p(**attr_code(WORD_DT_ATTRS)) do |p|
           if dterm.elements.empty?
             p << dterm.text
           else
@@ -24,7 +24,7 @@ module IsoDoc
       end
 
       def dl_parse_nontable(node, out)
-        out.div **attr_code(class: "figdl") do |div|
+        out.div(**attr_code(class: "figdl")) do |div|
           node["id"] and bookmark_parse(node, div)
           list_title_parse(node, div)
           node.elements.select { |n| dt_dd?(n) }
@@ -39,7 +39,7 @@ module IsoDoc
         "text-indent: -2.0cm; margin-left: 2.0cm; tab-stops: 2.0cm;".freeze
 
       def dl_parse_nontable1(out, dterm, ddef)
-        out.p **attr_code(style: WORD_EMBED_DL_ATTRS, id: dterm["id"]) do |p|
+        out.p(**attr_code(style: WORD_EMBED_DL_ATTRS, id: dterm["id"])) do |p|
           dterm.children.each { |n| parse(n, p) }
           insert_tab(p, 1)
           ddef["id"] and bookmark_parse(ddef, out)
@@ -70,9 +70,9 @@ module IsoDoc
       end
 
       def dl_parse_table(node, out)
-        out.div **attr_code(align: "left") do |div|
+        out.div(**attr_code(align: "left")) do |div|
           list_title_parse(node, div)
-          div.table **attr_code(dl_table_attrs(node)) do |v|
+          div.table(**attr_code(dl_table_attrs(node))) do |v|
             node.elements.select { |n| dt_dd?(n) }
               .each_slice(2) do |dt, dd|
               dl_parse_table1(v, dt, dd)

data/lib/isodoc/word_function/postprocess.rb

@@ -17,8 +17,10 @@ module IsoDoc
 
       def postprocess_cleanup(result)
         result = cleanup(to_xhtml(textcleanup(result)))
-        from_xhtml(word_cleanup(result))
+        result = from_xhtml(word_cleanup(result))
           .gsub("-DOUBLE_HYPHEN_ESCAPE-", "--")
+        # Unescape & to & in href attributes after all Nokogiri processing
+        unescape_amp_in_hrefs(result)
       end
 
       def toWord(result, filename, dir, header)

data/lib/isodoc/word_function/table.rb

@@ -115,7 +115,7 @@ module IsoDoc
         @in_table = true
         table_title_parse(node, out)
         out.div align: "center", class: "table_container" do |div|
-          div.table **table_attrs(node) do |t|
+          div.table(**table_attrs(node)) do |t|
             table_parse_core(node, t)
             table_parse_tail(node, t)
           end