html2doc 1.6.0 → 1.7.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 47535bf46876ee49a732b6c136f78b58a9ac009f880b95c5a73c8770293f3735
-  data.tar.gz: a052e0c3ba3ee27ca208b2624d7a832cba67bcc959f1ec5da36f9a7049c26c35
+  metadata.gz: 020bfc8d51718ee0fbc78ba2ce57fe1d306a50e93cb41949e98088228190dee5
+  data.tar.gz: 4fbad0486ed1cc59f7b67be1a5c2629ff840063b2bd8f1ffdbe50a7af082cc4e
 SHA512:
-  metadata.gz: '096dc5a7fe4b35e5afdec632f37b28f9980fcbcba4222e1ec1eb81fe4653a62cc00c5d9b90ed38ae54d4320ea8bf7e0fd0698625045504d371bcb90fb6247a54'
-  data.tar.gz: 7aeebef3892dc2273bc4ab9899624fc113b1989c3af097204b4f73eb250a7d52e8b3cfe62439e84fb179c1e9bbc96563af669d87f56b15dd82b4fc99953a2227
+  metadata.gz: 45e4f66e7ebc8591620a04e35362761ec48a20b51f735dd9977f05de6698e6cdbca67bb0ddf1615596d2924b7ee336b6e4b196f0dbccb00132e59ec660fab425
+  data.tar.gz: 868e01b43b930657accc0861c917116f78a9342512163803d038a70da8926c401ffe58a486d824a80115283cfc4a327743ab8d04931ce02c720a0f4aef5d8207

data/html2doc.gemspec

@@ -28,10 +28,11 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "htmlentities", "~> 4.3.4"
   spec.add_dependency "image_size", ">= 3.2.0"
+  spec.add_dependency "metanorma-utils"
   spec.add_dependency "mime-types"
   spec.add_dependency "nokogiri", "~> 1.14"
   spec.add_dependency "plane1converter", "~> 0.0.1"
-  spec.add_dependency "plurimath"
+  spec.add_dependency "plurimath", "~> 0.5.0"
   spec.add_dependency "thread_safe"
   spec.add_dependency "uuidtools"
 

data/lib/html2doc/base.rb

@@ -141,7 +141,7 @@ class Html2Doc
     docxml.xpath("//*[@id][not(@name)][not(@style = 'mso-element:footnote')]")
       .each do |x|
       (x["id"].empty? ||
-        %w(shapetype v:shapetype shape v:shape).include?(x.name)) and next
+        %w(v:shapetype v:shape v:rect v:line v:group).include?(x.name)) and next
       if x.children.empty? then x.add_child("<a name='#{x['id']}'></a>")
       else x.children.first.previous = "<a name='#{x['id']}'></a>"
       end

data/lib/html2doc/math.rb

@@ -3,6 +3,20 @@ require "plurimath"
 require "htmlentities"
 require "nokogiri"
 require "plane1converter"
+require "metanorma-utils"
+
+module Nokogiri
+  module XML
+    class Node
+      OOXML_NS = "http://schemas.openxmlformats.org/officeDocument/2006/math".freeze
+
+      def ooxml_xpath(path)
+        p = Metanorma::Utils::ns(path).gsub("xmlns:", "m:")
+        xpath(p, "m" => OOXML_NS)
+      end
+    end
+  end
+end
 
 class Html2Doc
   def progress_conv(idx, step, total, threshold, msg)
@@ -20,17 +34,30 @@ class Html2Doc
     doc
   end
 
+  MATHML_NS = "http://www.w3.org/1998/Math/MathML".freeze
+
   # random fixes to MathML input that OOXML needs to render properly
   def ooxml_cleanup(math, docnamespaces)
-    math = unwrap_accents(
-      mathml_preserve_space(
-        mathml_insert_rows(math, docnamespaces), docnamespaces
-      ),
-    )
-    math.add_namespace(nil, "http://www.w3.org/1998/Math/MathML")
+    #encode_math(
+      unwrap_accents(
+        mathml_preserve_space(
+          mathml_insert_rows(math, docnamespaces), docnamespaces
+        ),
+      )
+    #)
+    math.add_namespace(nil, MATHML_NS)
     math
   end
 
+  def encode_math(elem)
+    elem.traverse do |e|
+      e.text? or next
+      e.text.strip.empty? and next
+      e.replace(@c.encode(e.text, :hexadecimal))
+    end
+    elem
+  end
+
   def mathml_insert_rows(math, docnamespaces)
     math.xpath(%w(msup msub msubsup munder mover munderover)
             .map { |m| ".//xmlns:#{m}" }.join(" | "), docnamespaces).each do |x|
@@ -50,47 +77,57 @@ class Html2Doc
 
   HTML_NS = 'xmlns="http://www.w3.org/1999/xhtml"'.freeze
 
+  def wrap_text(elem, wrapper)
+    elem.traverse do |e|
+      e.text? or next
+      e.text.strip.empty? and next
+      e.wrap(wrapper)
+    end
+  end
+
   def unitalic(math)
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'p']]").each do |x|
-      x.wrap("<span #{HTML_NS} style='font-style:normal;'></span>")
+    math.ooxml_xpath(".//r[rPr[not(m:scr)]/sty[@m:val = 'p']]").each do |x|
+      wrap_text(x, "<span #{HTML_NS} style='font-style:normal;'></span>")
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'bi']]").each do |x|
-      x.wrap("<span #{HTML_NS} class='nostem' style='font-weight:bold;'><em></em></span>")
+    math.ooxml_xpath(".//r[rPr[not(m:scr)]/sty[@m:val = 'bi']]").each do |x|
+      wrap_text(x,
+                "<span #{HTML_NS} class='nostem' style='font-weight:bold;'><em></em></span>")
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'i']]").each do |x|
-      x.wrap("<span #{HTML_NS} class='nostem'><em></em></span>")
+    math.ooxml_xpath(".//r[rPr[not(m:scr)]/sty[@m:val = 'i']]").each do |x|
+      wrap_text(x, "<span #{HTML_NS} class='nostem'><em></em></span>")
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'b']]").each do |x|
-      x.wrap("<span #{HTML_NS} style='font-style:normal;font-weight:bold;'></span>")
+    math.ooxml_xpath(".//r[rPr[not(m:scr)]/sty[@m:val = 'b']]").each do |x|
+      wrap_text(x,
+                "<span #{HTML_NS} style='font-style:normal;font-weight:bold;'></span>")
     end
-    math.xpath(".//xmlns:r[xmlns:rPr/xmlns:scr[@m:val = 'monospace']]").each do |x|
+    math.ooxml_xpath(".//r[rPr/scr[@m:val = 'monospace']]").each do |x|
       to_plane1(x, :monospace)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr/xmlns:scr[@m:val = 'double-struck']]").each do |x|
+    math.ooxml_xpath(".//r[rPr/scr[@m:val = 'double-struck']]").each do |x|
       to_plane1(x, :doublestruck)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'script']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[not(m:sty) or m:sty/@m:val = 'p']/scr[@m:val = 'script']]").each do |x|
       to_plane1(x, :script)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'script']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[m:sty/@m:val = 'b']/scr[@m:val = 'script']]").each do |x|
       to_plane1(x, :scriptbold)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'fraktur']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[not(m:sty) or m:sty/@m:val = 'p']/scr[@m:val = 'fraktur']]").each do |x|
       to_plane1(x, :fraktur)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'fraktur']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[m:sty/@m:val = 'b']/scr[@m:val = 'fraktur']]").each do |x|
       to_plane1(x, :frakturbold)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:sty) or xmlns:sty/@m:val = 'p']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[not(m:sty) or m:sty/@m:val = 'p']/scr[@m:val = 'sans-serif']]").each do |x|
       to_plane1(x, :sans)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[m:sty/@m:val = 'b']/scr[@m:val = 'sans-serif']]").each do |x|
       to_plane1(x, :sansbold)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'i']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[m:sty/@m:val = 'i']/scr[@m:val = 'sans-serif']]").each do |x|
       to_plane1(x, :sansitalic)
     end
-    math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'bi']/xmlns:scr[@m:val = 'sans-serif']]").each do |x|
+    math.ooxml_xpath(".//r[rPr[m:sty/@m:val = 'bi']/scr[@m:val = 'sans-serif']]").each do |x|
       to_plane1(x, :sansbolditalic)
     end
     math
@@ -119,22 +156,26 @@ class Html2Doc
   # We will end up stripping them out again under Nokogiri 1.11, which correctly
   # insists on inheriting namespace from parent.
   def ooml_clean(xml)
-    xml.to_s
+    xml.to_xml(indent: 0)
       .gsub(/<\?[^>]+>\s*/, "")
       .gsub(/ xmlns(:[^=]+)?="[^"]+"/, "")
-      .gsub(%r{<(/)?(?!span)(?!em)([a-z])}, "<\\1m:\\2")
+    # .gsub(%r{<(/)?(?!span)(?!em)([a-z])}, "<\\1m:\\2")
   end
 
   def mathml_to_ooml1(xml, docnamespaces)
     doc = Nokogiri::XML::Document::new
     doc.root = ooxml_cleanup(xml, docnamespaces)
-    ooxml = ooml_clean(unitalic(esc_space(accent_tr(@xsltemplate.transform(doc)))))
-    ooxml = uncenter(xml, ooxml)
+    # ooxml = @xsltemplate.transform(doc)
+    d = xml.parent["block"] != "false" # display_style
+    ooxml = Nokogiri::XML(Plurimath::Math.parse(doc.to_xml(indent: 0),
+                                                :mathml).to_omml)
+    ooxml = unitalic(accent_tr(ooxml))
+    ooxml = ooml_clean(uncenter(xml, ooxml))
     xml.swap(ooxml)
   end
 
   def accent_tr(xml)
-    xml.xpath(".//*[local-name()='accPr']/*[local-name()='chr']").each do |x|
+    xml.ooxml_xpath(".//accPr/chr").each do |x|
       x["m:val"] &&= accent_tr1(x["m:val"])
       x["val"] &&= accent_tr1(x["val"])
     end
@@ -150,30 +191,48 @@ class Html2Doc
     end
   end
 
-  # escape space as &#x32;; we are removing any spaces generated by
-  # XML indentation
-  def esc_space(xml)
-    xml.traverse do |n|
-      next unless n.text?
+  OOXML_NS = "http://schemas.openxmlformats.org/officeDocument/2006/math".freeze
 
-      n = n.text.gsub(/ /, "&#x32;")
-    end
-    xml
+  def math_only_para?(node)
+    x = node.dup
+    x.xpath(".//m:math", "m" => MATHML_NS).each(&:remove)
+    x.xpath(".//m:oMathPara | .//m:oMath", "m" => OOXML_NS).each(&:remove)
+    x.xpath(".//m:oMathPara | .//m:oMath").each(&:remove)
+    # namespace can go missing during processing
+    x.text.strip.empty?
+  end
+
+  def math_block?(_ooxml, mathml)
+    # ooxml.name == "oMathPara" || mathml["displaystyle"] == "true"
+    mathml["displaystyle"] == "true"
   end
 
+  STYLE_BEARING_NODE =
+    %w(p div td th li).map { |x| ".//ancestor::#{x}" }.join(" | ").freeze
+
   # if oomml has no siblings, by default it is centered; override this with
   # left/right if parent is so tagged
+  # also if ooml has mathPara already, or is in para with only oMath content
   def uncenter(math, ooxml)
-    alignnode = math.at(".//ancestor::*[@style][local-name() = 'p' or "\
-                        "local-name() = 'div' or local-name() = 'td']/@style")
-    return ooxml unless alignnode && (math.next == nil && math.previous == nil)
-
-    %w(left right).each do |dir|
-      if alignnode.text.include? ("text-align:#{dir}")
-        ooxml = "<m:oMathPara><m:oMathParaPr><m:jc "\
-                "m:val='#{dir}'/></m:oMathParaPr>#{ooxml}</m:oMathPara>"
-      end
-    end
+    alignnode = math.xpath(STYLE_BEARING_NODE).last
+    ooxml.document? and ooxml = ooxml.root
+    ret = uncenter_unneeded(math, ooxml, alignnode) and return ret
+    dir = "left"
+    alignnode["style"]&.include?("text-align:right") and dir = "right"
+    ooxml.name == "oMathPara" or
+      ooxml.wrap("<m:oMathPara></m:oMathPara>")
+    ooxml.elements.first.previous =
+      "<m:oMathParaPr><m:jc m:val='#{dir}'/></m:oMathParaPr>"
     ooxml
   end
+
+  def uncenter_unneeded(math, ooxml, alignnode)
+    (math_block?(ooxml, math) || !alignnode) and return ooxml
+    if !math_only_para?(alignnode)
+      ooxml.name == "oMathPara" and
+        ooxml = ooxml.elements.detect { |x| x.name == "oMath" }
+      return ooxml
+    end
+    nil
+  end
 end

data/lib/html2doc/version.rb

@@ -1,3 +1,3 @@
 class Html2Doc
-  VERSION = "1.6.0".freeze
+  VERSION = "1.7.0".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html2doc
 version: !ruby/object:Gem::Version
-  version: 1.6.0
+  version: 1.7.0
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-08-07 00:00:00.000000000 Z
+date: 2023-09-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: htmlentities
@@ -38,6 +38,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: 3.2.0
+- !ruby/object:Gem::Dependency
+  name: metanorma-utils
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 - !ruby/object:Gem::Dependency
   name: mime-types
   requirement: !ruby/object:Gem::Requirement
@@ -84,16 +98,16 @@ dependencies:
   name: plurimath
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 0.5.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 0.5.0
 - !ruby/object:Gem::Dependency
   name: thread_safe
   requirement: !ruby/object:Gem::Requirement