to_ascii_latex 0.0.19 → 0.1.0

data/Gemfile

@@ -7,3 +7,4 @@ source 'https://rubygems.org'
 gemspec
 
 gem 'to_latex'
+gem 'nokogiri'

data/Rakefile

@@ -1 +1,10 @@
 require "bundler/gem_tasks"
+
+require 'rake/testtask'
+
+Rake::TestTask.new do |t|
+  t.libs << 'test'
+end
+
+desc "Run tests"
+task :default => :test

data/lib/to_ascii_latex/version.rb

@@ -1,3 +1,3 @@
 module ToAsciiLatex
-  VERSION = "0.0.19"
+  VERSION = "0.1.0"
 end

data/lib/to_ascii_latex.rb

@@ -1,5 +1,6 @@
 require "to_ascii_latex/version"
 require 'to_latex'
+require 'nokogiri'
 
 module ToAsciiLatex
   @@smartquotes = false
@@ -23,35 +24,28 @@ module ToAsciiLatex
   # http://www.fileformat.info/info/unicode/char/fb01/index.htm
   # http://www.w3.org/Math/characters/unicode.xml
   # http://www.johndcook.com/unicode_latex.html
+
   NBSP = "\u00A0"
-  MAPPING = { "\u2014"  => "--",
-              "\u2018"  => "`",
-              "\u2019"  => "'",
-              "\u201E"  => "``",
-              "\u201C"  => "``",
-              "\u201D"  => "''",
-              "\u00A0"  => ' ',
-              "\u00E6"  => '\ae{}',
-              "\u00E0"  => '\\\\`{a}',
-              "\u0097"  => '--',
-              "\u0092"  => "'",
-              "\u2026"  => '{}\ldots{}',
-              "\uFB01"  => 'fi',
-              "\uFB00"  => 'ff',
-              "\u2013"  => '-',
-              "\u25A0"  => '\ensuremath{\blacksquare}',
-              "\u2022"  => '\ensuremath{\bullet}',
-              "\u2248"  => '\ensuremath{\approx}',
-              "\u0B0C"  => '?',
-              "\u00B0"  => '\ensuremath{^{\circ}}',
-              "\u00A7"  => '\S{}',
-              "\u20AC"  => '\euro{}',
-              "\u00A2"  => '\textcent{}',
-              "\u2227"  => '\ensuremath{\land}',
-              "\u00AC"  => '\ensuremath{\neg}',
-              "\u00F6"  => '\"{o}',
-            }
-  MAPPING_RE = /#{MAPPING.keys.join('|')}/
+  MAPPING = {}
+  RE_SPECIALS = %w{. | ( ) [ ] { } \ ^ $ + * ?}
+
+  mapping = Nokogiri::XML(open(File.join(File.dirname(__FILE__), 'unicode.xml')))
+  mapping.xpath('//character[@id and @mode and latex]').each{|char|
+    id = char['id'].to_s
+    next if id =~ /^U[-0-9A-F]+$/ && id =~ /-/
+    raise "Unexpected char #{id.inspect}" unless id =~ /^U[0-9A-F]+$/i
+    id.gsub!(/^U/, '')
+    id.gsub!(/^0/, '') if id.size == 5 && id =~ /^0/
+
+    key = [id.gsub(/^U/i, '').hex].pack('U')
+    value = char.at('.//latex').inner_text.strip
+    next if key == value || value == "\\space"
+
+    value = "\\ensuremath{#{value}}" if char['mode'] == 'math'
+    value = "{#{value}}" if value !~ /^\\/ || value !~ /}$/
+    MAPPING[key] = value
+  }
+  RE = /#{MAPPING.keys.collect{|c| RE_SPECIALS.include?(c) ? "\\#{c}" : c}.join('|')}/
 
   # Add a hook - whenever a class or module calls `extend ToAsciiLatex`,
   # run this code
@@ -64,7 +58,7 @@ module ToAsciiLatex
       # Define this new method
       def self.new_escape(s)
         s = s.gsub(NBSP, ' ') if @@replace_nbsp
-        x = old_escape(s).gsub(MAPPING_RE){|c| MAPPING[c]}.gsub('\backslash{}', '\ensuremath{\backslash}')
+        x = s.gsub(RE){|c| MAPPING[c]} # .gsub('\backslash{}', '\ensuremath{\backslash}')
         warn "Unicode in  #{x.inspect}" if x.delete("^\u{0000}-\u{007F}") != x
 
         if @@smartquotes