math.rb in html2doc-1.1.2

- old
+ new

@@ -23,17 +23,22 @@
   def self.asciimath_to_mathml(doc, delims)
     return doc if delims.nil? || delims.size < 2
 
     m = doc.split(/(#{Regexp.escape(delims[0])}|#{Regexp.escape(delims[1])})/)
     m.each_slice(4).map.with_index do |(*a), i|
-      i % 500 == 0 && m.size > 1000 && i > 0 and
-        warn "MathML #{i} of #{(m.size / 4).floor}"
+      progress_conv(i, 500, (m.size / 4).floor, 1000, "AsciiMath")
       a[2].nil? || a[2] = asciimath_to_mathml1(a[2])
       a.size > 1 ? a[0] + a[2] : a[0]
     end.join
   end
 
+  def self.progress_conv(idx, step, total, threshold, msg)
+    return unless (idx % step).zero? && total > threshold && idx.positive?
+
+    warn "#{msg} #{idx} of #{total}"
+  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)
@@ -67,22 +72,24 @@
       x.children = x.children.to_xml.gsub(/^\s/, "&#xA0;").gsub(/\s$/, "&#xA0;")
     end
     math
   end
 
+  HTML_NS = 'xmlns="http://www.w3.org/1999/xhtml"'.freeze
+
   def self.unitalic(math)
     math.xpath(".//xmlns:r[xmlns:rPr[not(xmlns:scr)]/xmlns:sty[@m:val = 'p']]").each do |x|
-      x.wrap("<span style='font-style:normal;'></span>")
+      x.wrap("<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 class='nostem' style='font-weight:bold;'><em></em></span>")
+      x.wrap("<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 class='nostem'><em></em></span>")
+      x.wrap("<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 style='font-style:normal;font-weight:bold;'></span>")
+      x.wrap("<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|
       to_plane1(x, :monospace)
     end
     math.xpath(".//xmlns:r[xmlns:rPr/xmlns:scr[@m:val = 'double-struck']]").each do |x|
@@ -126,24 +133,34 @@
 
   def self.mathml_to_ooml(docxml)
     docnamespaces = docxml.collect_namespaces
     m = docxml.xpath("//*[local-name() = 'math']")
     m.each_with_index do |x, i|
-      i % 100 == 0 && m.size > 500 && i > 0 and
-        warn "Math OOXML #{i} of #{m.size}"
-      element = ooxml_cleanup(x, docnamespaces)
-      doc = Nokogiri::XML::Document::new
-      doc.root = element
-      ooxml = unitalic(esc_space(@xsltemplate.transform(doc))).to_s
-        .gsub(/<\?[^>]+>\s*/, "")
-        .gsub(/ xmlns(:[^=]+)?="[^"]+"/, "")
-        .gsub(%r{<(/)?(?!span)(?!em)([a-z])}, "<\\1m:\\2")
-      ooxml = uncenter(x, ooxml)
-      x.swap(ooxml)
+      progress_conv(i, 100, m.size, 500, "Math OOXML")
+      mathml_to_ooml1(x, docnamespaces)
     end
   end
 
+  # We need span and em not to be namespaced. Word can't deal with explicit 
+  # namespaces.
+  # We will end up stripping them out again under Nokogiri 1.11, which correctly
+  # insists on inheriting namespace from parent.
+  def self.ooml_clean(xml)
+    xml.to_s
+      .gsub(/<\?[^>]+>\s*/, "")
+      .gsub(/ xmlns(:[^=]+)?="[^"]+"/, "")
+      .gsub(%r{<(/)?(?!span)(?!em)([a-z])}, "<\\1m:\\2")
+  end
+
+  def self.mathml_to_ooml1(xml, docnamespaces)
+    doc = Nokogiri::XML::Document::new
+    doc.root = ooxml_cleanup(xml, docnamespaces)
+      ooxml = ooml_clean(unitalic(esc_space(@xsltemplate.transform(doc))))
+    ooxml = uncenter(xml, ooxml)
+    xml.swap(ooxml)
+  end
+
   # escape space as &#x32;; we are removing any spaces generated by
   # XML indentation
   def self.esc_space(xml)
     xml.traverse do |n|
       next unless n.text?
@@ -155,10 +172,10 @@
 
   # if oomml has no siblings, by default it is centered; override this with
   # left/right if parent is so tagged
   def self.uncenter(math, ooxml)
     alignnode = math.at(".//ancestor::*[@style][local-name() = 'p' or "\
-                     "local-name() = 'div' or local-name() = 'td']/@style")
+                        "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 "\