lib/html2doc/math.rb
require "uuidtools"
require "plurimath"
require "htmlentities"
require "nokogiri"
require "plane1converter"
class Html2Doc
def asciimath_to_mathml1(expr, retain_asciimath)
warn expr
ret = Plurimath::Math.parse(HTMLEntities.new.decode(expr), "asciimath").to_mathml
.gsub(/<math>/, "<math xmlns='http://www.w3.org/1998/Math/MathML'>")
retain_asciimath and
ret += "<asciimath>#{@c.encode(@c.decode(expr), :basic)}</asciimath>"
ret
rescue StandardError => e
puts "parsing: #{expr}"
puts e.message
raise e
end
def asciimath_to_mathml(doc, delims, retain_asciimath: false)
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|
progress_conv(i, 500, (m.size / 4).floor, 1000, "AsciiMath")
a[2].nil? or a[2] = asciimath_to_mathml1(a[2], retain_asciimath)
a.size > 1 ? a[0] + a[2] : a[0]
end.join
end
def progress_conv(idx, step, total, threshold, msg)
return unless (idx % step).zero? && total > threshold && idx.positive?
warn "#{msg} #{idx} of #{total}"
end
def 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 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")
math
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|
next unless x.next_element && x.next_element != "mrow"
x.next_element.wrap("<mrow/>")
end
math
end
def mathml_preserve_space(math, docnamespaces)
math.xpath(".//xmlns:mtext", docnamespaces).each do |x|
x.children = x.children.to_xml.gsub(/^\s/, " ").gsub(/\s$/, " ")
end
math
end
HTML_NS = 'xmlns="http://www.w3.org/1999/xhtml"'.freeze
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>")
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>")
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>")
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>")
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|
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|
to_plane1(x, :script)
end
math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns: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|
to_plane1(x, :fraktur)
end
math.xpath(".//xmlns:r[xmlns:rPr[xmlns:sty/@m:val = 'b']/xmlns: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|
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|
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|
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|
to_plane1(x, :sansbolditalic)
end
math
end
def to_plane1(xml, font)
xml.traverse do |n|
next unless n.text?
n.replace(Plane1Converter.conv(@c.decode(n.text), font))
end
xml
end
def mathml_to_ooml(docxml)
docnamespaces = docxml.collect_namespaces
m = docxml.xpath("//*[local-name() = 'math']")
m.each_with_index do |x, i|
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 ooml_clean(xml)
xml.to_s
.gsub(/<\?[^>]+>\s*/, "")
.gsub(/ xmlns(:[^=]+)?="[^"]+"/, "")
.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)
xml.swap(ooxml)
end
def accent_tr(xml)
xml.xpath(".//*[local-name()='accPr']/*[local-name()='chr']").each do |x|
x["m:val"] &&= accent_tr1(x["m:val"])
x["val"] &&= accent_tr1(x["val"])
end
xml
end
def accent_tr1(accent)
case accent
when "\u2192" then "\u20D7"
when "^" then "\u0302"
when "~" then "\u0303"
else accent
end
end
# escape space as 2; we are removing any spaces generated by
# XML indentation
def esc_space(xml)
xml.traverse do |n|
next unless n.text?
n = n.text.gsub(/ /, "2")
end
xml
end
# if oomml has no siblings, by default it is centered; override this with
# left/right if parent is so tagged
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
ooxml
end
end