html2doc 1.0.4 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 89f05f87cd25266b64654929b7087a07c912fcf8b24beb33f2faceb5123da190
-  data.tar.gz: de12518aafb1ac65bf1f4addf76f16ff246c58aa99561e321dc3d8af7cd0a623
+  metadata.gz: f1d55301cebf98f75ddfd8966237d19c39bae1c9c8b4b01a39f124ca5da8e588
+  data.tar.gz: a84cd4d87e2cf1c9f107300c883b6b6b85496e2067e76d7f0a282feecf6550c1
 SHA512:
-  metadata.gz: 535aa46b12785253d9b5ebacf7f0c70f2f7045a5367dbf3ae2c023188a6005be39f6f003574d7855ca8ca2e757ddcde2f5128602e18156699248281fa6db4502
-  data.tar.gz: 5452e2880c9ba6f231d02039e5f2b2054e5535c903539f34804c65aeb77b63361994fd604f9d3d92ce808244fa5c46e9fc790f747041955606a1a474bdd02449
+  metadata.gz: 2ef21ec975f624420db8ee706ab1997ab7941bb7e9f75bfaecdde056346a08f73842cd151ad6f5c7a9711a0616ccc61177bd24f17693f8c31c6db70e7aa78088
+  data.tar.gz: c204dfcb3f27a24f86908195d47287c74daaaa41d02de5f28fbb99167b17895f7b46e27971ae533c18874f09ed2d32d4205c6dcd9c6ee6b9e045a9dc06656bfa

data/.github/workflows/rake.yml

@@ -0,0 +1,42 @@
+# Auto-generated by Cimas: Do not edit it manually!
+# See https://github.com/metanorma/cimas
+name: rake
+
+on:
+  push:
+    branches: [ master, main ]
+    tags: [ v* ]
+  pull_request:
+
+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: [ '2.7', '2.6', '2.5', '2.4' ]
+        os: [ ubuntu-latest, windows-latest, macos-latest ]
+        experimental: [ false ]
+        include:
+          - ruby: '3.0'
+            os: 'ubuntu-latest'
+            experimental: true
+          - ruby: '3.0'
+            os: 'windows-latest'
+            experimental: true
+          - ruby: '3.0'
+            os: 'macos-latest'
+            experimental: true
+    steps:
+      - uses: actions/checkout@v2
+        with:
+          submodules: true
+
+      - uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: ${{ matrix.ruby }}
+          bundler-cache: true
+
+      - run: bundle exec rake

data/.gitignore

@@ -9,3 +9,5 @@
 
 # rspec failure tracking
 .rspec_status
+
+.rubocop-https--*

data/.rubocop.yml

@@ -1,10 +1,14 @@
 # This project follows the Ribose OSS style guide.
 # https://github.com/riboseinc/oss-guides
 # All project-specific additions and overrides should be specified in this file.
-
 inherit_from:
   - https://raw.githubusercontent.com/riboseinc/oss-guides/master/ci/rubocop.yml
+
+# local repo-specific modifications
+
 AllCops:
-  TargetRubyVersion: 2.3
+  DisplayCopNames: false
+  StyleGuideCopsOnly: false
+  TargetRubyVersion: 2.4
 Rails:
   Enabled: true

data/README.adoc

@@ -3,9 +3,7 @@
 https://github.com/metanorma/html2doc/workflows/main/badge.svg
 
 image:https://img.shields.io/gem/v/html2doc.svg["Gem Version", link="https://rubygems.org/gems/html2doc"]
-image:https://github.com/metanorma/html2doc/workflows/ubuntu/badge.svg["Ubuntu Build Status", link="https://github.com/metanorma/html2doc/actions?query=workflow%3Aubuntu"]
-image:https://github.com/metanorma/html2doc/workflows/macos/badge.svg["OSX Build Status", link="https://github.com/metanorma/html2doc/actions?query=workflow%3Amacos"]
-image:https://github.com/metanorma/html2doc/workflows/windows/badge.svg["Windows Build Status", link="https://github.com/metanorma/html2doc/actions?query=workflow%3Awindows"]
+image:https://github.com/metanorma/html2doc/workflows/rake/badge.svg["Build Status", link="https://github.com/metanorma/html2doc/actions?workflow=rake"]
 image:https://codeclimate.com/github/metanorma/html2doc/badges/gpa.svg["Code Climate", link="https://codeclimate.com/github/metanorma/html2doc"]
 image:https://img.shields.io/github/issues-pr-raw/metanorma/html2doc.svg["Pull Requests", link="https://github.com/metanorma/html2doc/pulls"]
 image:https://img.shields.io/github/commits-since/metanorma/html2doc/latest.svg["Commits since latest",link="https://github.com/metanorma/html2doc/releases"]

data/html2doc.gemspec

@@ -28,10 +28,11 @@ Gem::Specification.new do |spec|
   spec.add_dependency "htmlentities", "~> 4.3.4"
   spec.add_dependency "image_size"
   spec.add_dependency "mime-types"
-  spec.add_dependency "nokogiri", ">= 1.10.4"
+  spec.add_dependency "nokogiri", "~> 1.10.4"
   spec.add_dependency "thread_safe"
   spec.add_dependency "uuidtools"
-  spec.add_dependency "asciimath", "~> 2.0.0"
+  spec.add_dependency "asciimath", "~> 2.0.2"
+  spec.add_dependency "plane1converter", "~> 0.0.1"
 
   spec.add_development_dependency "byebug", "~> 9.1"
   spec.add_development_dependency "equivalent-xml", "~> 0.6"
@@ -39,7 +40,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "guard-rspec", "~> 4.7"
   spec.add_development_dependency "rake", "~> 12.0"
   spec.add_development_dependency "rspec", "~> 3.6"
-  spec.add_development_dependency "rubocop", "= 0.54.0"
+  spec.add_development_dependency "rubocop", "~> 1.5.2"
   spec.add_development_dependency "simplecov", "~> 0.15"
   spec.add_development_dependency "timecop", "~> 0.9"
   spec.add_development_dependency "rspec-match_fuzzy", "~> 0.1.3"

data/lib/html2doc/base.rb

@@ -24,11 +24,19 @@ module Html2Doc
     File.open("#{hash[:dir1]}/header.html", "w:UTF-8") { |f| f.write(doc) }
   end
 
+  def self.clear_dir(dir)
+    Dir.foreach(dir) do |f|
+      fn = File.join(dir, f)
+      File.delete(fn) if f != '.' && f != '..'
+    end
+    dir
+  end
+
   def self.create_dir(filename, dir)
-    return dir if dir
+    dir and return clear_dir(dir)
     dir = "#{filename}_files"
     Dir.mkdir(dir) unless File.exists?(dir)
-    dir
+    clear_dir(dir)
   end
 
   def self.process_html(result, hash)
@@ -123,18 +131,19 @@ module Html2Doc
     docxml.xpath("//*[local-name() = 'head']").each do |h|
       h.children.first.add_previous_sibling <<~XML
       #{PRINT_VIEW}
-        <link rel="File-List" href="#{File.basename(dir)}/filelist.xml"/>
+        <link rel="File-List" href="cid:filelist.xml"/>
       XML
     end
   end
 
   def self.filename_substitute(stylesheet, header_filename, filename)
     if header_filename.nil?
-      stylesheet.gsub!(/\n[^\n]*FILENAME[^\n]*i\n/, "\n")
+      stylesheet
     else
-      stylesheet.gsub!(/FILENAME/, File.basename(filename))
+      stylesheet.gsub(/url\("[^"]+"\)/) do |m|
+        /FILENAME/.match(m) ? "url(cid:header.html)" : m
+      end
     end
-    stylesheet
   end
 
   def self.stylesheet(filename, header_filename, fn)

data/lib/html2doc/math.rb

@@ -2,6 +2,7 @@ require "uuidtools"
 require "asciimath"
 require "htmlentities"
 require "nokogiri"
+require "plane1converter"
 
 module Html2Doc
   @xsltemplate =
@@ -9,9 +10,15 @@ module Html2Doc
                              encoding: "utf-8"))
 
   def self.asciimath_to_mathml1(x)
-    AsciiMath::MathMLBuilder.new(:msword => true).append_expression(
-      AsciiMath.parse(HTMLEntities.new.decode(x)).ast).to_s.
-        gsub(/<math>/, "<math xmlns='http://www.w3.org/1998/Math/MathML'>")
+    begin
+      AsciiMath::MathMLBuilder.new(:msword => true).append_expression(
+        AsciiMath.parse(HTMLEntities.new.decode(x)).ast).to_s.
+      gsub(/<math>/, "<math xmlns='http://www.w3.org/1998/Math/MathML'>")
+    rescue StandardError => e
+      puts "parsing: #{x}"
+      puts e.message
+      raise e
+    end
   end
 
   def self.asciimath_to_mathml(doc, delims)
@@ -25,20 +32,93 @@ module Html2Doc
     end.join
   end
 
+  def self.unwrap_accents(doc)
+    doc.xpath("//*[@accent = 'true']").each do |x|
+      x.elements.length > 1 or next
+      x.elements[1].name == "mrow" and
+        x.elements[1].replace(x.elements[1].children)
+    end
+    doc
+  end
+
   # random fixes to MathML input that OOXML needs to render properly
   def self.ooxml_cleanup(m, docnamespaces)
+    m = unwrap_accents(mathml_preserve_space(
+      mathml_insert_rows(m, docnamespaces), docnamespaces))
+    m.add_namespace(nil, "http://www.w3.org/1998/Math/MathML")
+    m
+  end
+
+  def self.mathml_insert_rows(m, docnamespaces)
     m.xpath(%w(msup msub msubsup munder mover munderover).
             map { |m| ".//xmlns:#{m}" }.join(" | "), docnamespaces).each do |x|
       next unless x.next_element && x.next_element != "mrow"
       x.next_element.wrap("<mrow/>")
     end
+    m
+  end
+
+  def self.mathml_preserve_space(m, docnamespaces)
     m.xpath(".//xmlns:mtext", docnamespaces).each do |x|
       x.children = x.children.to_xml.gsub(/^\s/, "&#xA0;").gsub(/\s$/, "&#xA0;")
     end
-    m.add_namespace(nil, "http://www.w3.org/1998/Math/MathML")
     m
   end
 
+  def self.unitalic(m)
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'p']]").each do |x|
+      x.wrap("<span style='font-style:normal;'></span>")
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'bi']]").each do |x|
+      x.wrap("<span class='nostem' style='font-weight:bold;'><em></em></span>")
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'i']]").each do |x|
+      x.wrap("<span class='nostem'><em></em></span>")
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'b']]").each do |x|
+      x.wrap("<span style='font-style:normal;font-weight:bold;'></span>")
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr/xmlns:scr[@m:val = 'monospace']]").each do |x|
+      toPlane1(x, :monospace)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr/xmlns:scr[@m:val = 'double-struck']]").each do |x|
+      toPlane1(x, :doublestruck)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'script']]").each do |x|
+      toPlane1(x, :script)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'script']]").each do |x|
+      toPlane1(x, :scriptbold)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'fraktur']]").each do |x|
+      toPlane1(x, :fraktur)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'fraktur']]").each do |x|
+      toPlane1(x, :frakturbold)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+      toPlane1(x, :sans)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+      toPlane1(x, :sansbold)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'i']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+      toPlane1(x, :sansitalic)
+    end
+    m.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'bi']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+      toPlane1(x, :sansbolditalic)
+    end
+    m
+  end
+
+  def self.toPlane1(x, font)
+    x.traverse do |n|
+      next unless n.text?
+      n.replace(Plane1Converter.conv(HTMLEntities.new.decode(n.text), font))
+    end
+    x
+  end
+
   def self.mathml_to_ooml(docxml)
     docnamespaces = docxml.collect_namespaces
     m = docxml.xpath("//*[local-name() = 'math']")
@@ -48,10 +128,10 @@ module Html2Doc
       element = ooxml_cleanup(x, docnamespaces)
       doc = Nokogiri::XML::Document::new()
       doc.root = element
-      ooxml = (esc_space(@xsltemplate.transform(doc))).to_s.
+      ooxml = (unitalic(esc_space(@xsltemplate.transform(doc)))).to_s.
         gsub(/<\?[^>]+>\s*/, "").
         gsub(/ xmlns(:[^=]+)?="[^"]+"/, "").
-        gsub(%r{<(/)?([a-z])}, "<\\1m:\\2")
+        gsub(%r{<(/)?(?!span)(?!em)([a-z])}, "<\\1m:\\2")
       ooxml = uncenter(x, ooxml)
       x.swap(ooxml)
     end

data/lib/html2doc/mime.rb

@@ -11,7 +11,8 @@ module Html2Doc
     Content-Type: multipart/related; boundary="#{boundary}"
 
     --#{boundary}
-    Content-Location: file:///C:/Doc/#{File.basename(filename)}.htm
+    Content-ID: <#{File.basename(filename)}>
+    Content-Disposition: inline; filename="#{File.basename(filename)}"
     Content-Type: text/html; charset="utf-8"
 
     #{result}
@@ -29,7 +30,8 @@ module Html2Doc
     encoded_file = Base64.strict_encode64(content).gsub(/(.{76})/, "\\1\n")
     <<~"FILE"
     --#{boundary}
-    Content-Location: file:///C:/Doc/#{File.basename(filename)}_files/#{item}
+    Content-ID: <#{File.basename(item)}>
+    Content-Disposition: inline; filename="#{File.basename(item)}"
     Content-Transfer-Encoding: base64
     Content-Type: #{content_type}
 
@@ -52,15 +54,22 @@ module Html2Doc
 
   def self.mime_package(result, filename, dir)
     boundary = mime_boundary
-    mhtml = mime_preamble(boundary, filename, result)
-    mhtml += mime_attachment(boundary, filename, "filelist.xml", dir)
+    mhtml = mime_preamble(boundary, "#{filename}.htm", result)
+    mhtml += mime_attachment(boundary, "#{filename}.htm", "filelist.xml", dir)
     Dir.foreach(dir) do |item|
       next if item == "." || item == ".." || /^\./.match(item) ||
         item == "filelist.xml"
-      mhtml += mime_attachment(boundary, filename, item, dir)
+      mhtml += mime_attachment(boundary, "#{filename}.htm", item, dir)
     end
     mhtml += "--#{boundary}--"
-    File.open("#{filename}.doc", "w:UTF-8") { |f| f.write mhtml }
+    File.open("#{filename}.doc", "w:UTF-8") { |f| f.write contentid(mhtml) }
+  end
+
+  def self.contentid(mhtml)
+    mhtml.gsub %r{(<img[^>]*?src=")([^\"']+)(['"])}m do |m|
+      repl = "#{$1}cid:#{File.basename($2)}#{$3}"
+      /^data:|^https?:/.match($2) ? m : repl
+    end
   end
 
   # max width for Word document is 400, max height is 680
@@ -68,7 +77,7 @@ module Html2Doc
     realSize = ImageSize.path(path).size
     s = [i["width"].to_i, i["height"].to_i]
     s = realSize if s[0].zero? && s[1].zero?
-    return [nil, nil] if realSize[0].nil? || realSize[1].nil?
+    return [nil, nil] if realSize.nil? || realSize[0].nil? || realSize[1].nil?
     s[1] = s[0] * realSize[1] / realSize[0] if s[1].zero? && !s[0].zero?
     s[0] = s[1] * realSize[0] / realSize[1] if s[0].zero? && !s[1].zero?
     s = [(s[0] * maxheight / s[1]).ceil, maxheight] if s[1] > maxheight
@@ -92,7 +101,7 @@ module Html2Doc
       next unless i.element? && %w(img v:imagedata).include?(i.name)
       #warnsvg(i["src"])
       next if /^http/.match i["src"]
-      next if %r{^data:image/[^;]+;base64}.match i["src"]
+      next if %r{^data:(image|application)/[^;]+;base64}.match i["src"]
       local_filename = %r{^([A-Z]:)?/}.match(i["src"]) ? i["src"] :
         File.join(localdir, i["src"])
       new_filename = "#{mkuuid}#{File.extname(i["src"])}"
@@ -113,14 +122,14 @@ module Html2Doc
 
   def self.header_image_cleanup1(a, dir, filename, localdir)
     if a.size == 2 && !(/ src="https?:/.match a[1]) &&
-        !(%r{ src="data:image/[^;]+;base64}.match a[1])
+        !(%r{ src="data:(image|application)/[^;]+;base64}.match a[1])
       m = / src=['"](?<src>[^"']+)['"]/.match a[1]
       #warnsvg(m[:src])
       m2 = /\.(?<suffix>[a-zA-Z_0-9]+)$/.match m[:src]
       new_filename = "#{mkuuid}.#{m2[:suffix]}"
       old_filename = %r{^([A-Z]:)?/}.match(m[:src]) ? m[:src] : File.join(localdir, m[:src])
       FileUtils.cp old_filename, File.join(dir, new_filename)
-      a[1].sub!(%r{ src=['"](?<src>[^"']+)['"]}, " src='file:///C:/Doc/#{filename}_files/#{new_filename}'")
+      a[1].sub!(%r{ src=['"](?<src>[^"']+)['"]}, " src='cid:#{new_filename}'")
     end
     a.join
   end

data/lib/html2doc/mml2omml.xsl

@@ -878,9 +878,21 @@
                </xsl:otherwise>
             </xsl:choose>
          </xsl:if>
-      </xsl:variable>
+       </xsl:variable>
+       <!-- NN 20200831 https://github.com/metanorma/metanorma-nist/issues/155 Read in ancestor mstyle -->
+       <xsl:variable name="mathvariant2">
+         <xsl:choose>
+           <xsl:when test="$mathvariant=''">
+             <xsl:value-of select="$ndCur/ancestor::mml:mstyle[@mathvariant][1]/@mathvariant"/>
+           </xsl:when>
+           <xsl:otherwise>
+             <xsl:value-of select="$mathvariant"/>
+           </xsl:otherwise>
+         </xsl:choose>
+       </xsl:variable>
+       <!-- END NN -->
       <xsl:call-template name="CreateMathRPR">
-         <xsl:with-param name="mathvariant" select="$mathvariant"/>
+         <xsl:with-param name="mathvariant" select="$mathvariant2"/>
          <xsl:with-param name="fontstyle" select="$fontstyle"/>
          <xsl:with-param name="fontweight" select="$fontweight"/>
          <xsl:with-param name="ndCur" select="$ndCur"/>
@@ -905,7 +917,7 @@
             <xsl:with-param name="fontweight" select="$fontweight"/>
             <xsl:with-param name="ndCur" select="$ndCur"/>
          </xsl:call-template>
-      </xsl:variable>
+       </xsl:variable>
       <xsl:if test="$fLit=1 or $fNor=1 or ($sFontCur!='italic' and $sFontCur!='')">
          <rPr>
             <xsl:if test="$fNor=1">
@@ -919,7 +931,7 @@
                <xsl:with-param name="fNor" select="$fNor"/>
             </xsl:call-template>
          </rPr>
-      </xsl:if>
+       </xsl:if>
   </xsl:template>
 
   <!-- %%Template: GetFontCur
@@ -935,8 +947,10 @@
          </xsl:when>
          <xsl:when test="not($ndCur)">
             <xsl:value-of select="'italic'"/>
-         </xsl:when>
-         <xsl:when test="$ndCur/self::mml:mi and (string-length(normalize-space($ndCur)) &lt;= 1)               or $ndCur/self::mml:mn and string(number($ndCur/text()))!='NaN'               or $ndCur/self::mml:mo">
+          </xsl:when>
+         <!--<xsl:when test="$ndCur/self::mml:mi and (string-length(normalize-space($ndCur)) &lt;= 1)               or $ndCur/self::mml:mn and string(number($ndCur/text()))!='NaN'               or $ndCur/self::mml:mo">-->
+           <!-- https://github.com/metanorma/metanorma-nist/issues/155 : DO NOT italicise mml:mo -->
+           <xsl:when test="$ndCur/self::mml:mi and (string-length(normalize-space($ndCur)) &lt;= 1)               or $ndCur/self::mml:mn and string(number($ndCur/text()))!='NaN'">
 
 	   <!-- The default for the above three cases is fontstyle=italic fontweight=normal.-->
 	   <xsl:choose>
@@ -1000,6 +1014,12 @@
             </sty>
          </xsl:when>
          <xsl:when test="$font='italic'">
+           <!-- NN 20200901 https://github.com/metanorma/metanorma-itu/issues/173 italic needs to be added explicitly to mtext -->
+           <xsl:if test="$fNor=1">
+            <sty>
+               <xsl:attribute name="m:val">i</xsl:attribute>
+            </sty>
+           </xsl:if>
       </xsl:when>
          <xsl:when test="$font='script'">
             <scr>
@@ -1067,8 +1087,16 @@
                <xsl:attribute name="m:val">bi</xsl:attribute>
             </sty>
          </xsl:when>
-         <xsl:when test="$font='monospace'"/>
+         <xsl:when test="$font='monospace'">
          <!-- We can't do monospace, so leave empty -->
+         <!-- NN 2020 https://github.com/metanorma/html2doc/issues/47 no, we will -->
+         <scr>
+               <xsl:attribute name="m:val">monospace</xsl:attribute>
+            </scr>
+            <sty>
+               <xsl:attribute name="m:val">p</xsl:attribute>
+             </sty>
+             </xsl:when>
       <xsl:when test="$font='bold'">
             <sty>
                <xsl:attribute name="m:val">b</xsl:attribute>