math_ml 0.9

data/lib/math_ml/string.rb

@@ -0,0 +1,32 @@
+#!/usr/bin/ruby
+#
+# Extension of String class by MathML Library
+#
+# Copyright (C) 2007, KURODA Hiraku <hiraku@hinet.mydns.jp>
+# You can redistribute it and/or modify it under GPL2.
+#
+
+require "math_ml"
+
+module MathML
+	module String
+		@@mathml_latex_parser = nil
+		def self.mathml_latex_parser
+			@@mathml_latex_parser = MathML::LaTeX::Parser.new unless @@mathml_latex_parser
+			@@mathml_latex_parser
+		end
+
+		def self.mathml_latex_parser=(mlp)
+			raise TypeError unless mlp.is_a?(MathML::LaTeX::Parser) || mlp==nil
+			@@mathml_latex_parser = mlp
+		end
+
+		def to_mathml(displaystyle=false)
+			MathML::String.mathml_latex_parser.parse(self, displaystyle)
+		end
+	end
+end
+
+class String
+	include MathML::String
+end

data/lib/math_ml/util.rb

@@ -0,0 +1,350 @@
+#!/usr/bin/ruby
+#
+# Utility for MathML Library
+#
+# Copyright (C) 2006, KURODA Hiraku <hiraku@hinet.mydns.jp>
+# You can redistribute it and/or modify it under GPL2.
+#
+
+require "math_ml"
+
+module MathML::Util
+	ESCAPES = {"<"=>"lt",
+		">"=>"gt",
+		"&"=>"amp",
+		"\""=>"quot",
+		"'"=>"apos"
+	}
+	INVALID_RE = /(?!)/
+	EQNARRAY_RE = /\\begin\s*\{eqnarray\}(#{MathML::LaTeX::MBEC}*?)\\end\s*\{eqnarray\}/
+	SINGLE_COMMAND_RE = /(\\([a-zA-Z]+))[ \t]?/
+
+	def self.escapeXML(s, br=false)
+		r = s.gsub(/[<>&"']/){|m| "&#{ESCAPES[m]};"}
+		br ? r.gsub(/\n/, "<br />\n") : r
+	end
+
+	def escapeXML(s, br=false)
+		MathML::Util.escapeXML(s, br)
+	end
+
+	def self.collect_regexp(a)
+		if a
+			a = [a].flatten
+			a.size>0 ? Regexp.new(a.inject(""){|r, i| i.is_a?(Regexp) ? "#{r}#{i.to_s}|" : r}.chop) : INVALID_RE
+		else
+			INVALID_RE
+		end
+	end
+
+	def collect_regexp(a)
+		MathML::Util.collect_regexp(a)
+	end
+
+	class MathData
+		attr_reader :math_list, :msrc_list, :dmath_list, :dsrc_list, :escape_list, :esrc_list, :user_list, :usrc_list
+		def initialize
+			@math_list = []
+			@msrc_list = []
+			@dmath_list = []
+			@dsrc_list = []
+			@escape_list = []
+			@esrc_list = []
+			@user_list = []
+			@usrc_list = []
+		end
+
+		def update(s)
+			@math_list.concat(s.math_list)
+			@msrc_list.concat(s.msrc_list)
+			@dmath_list.concat(s.dmath_list)
+			@dsrc_list.concat(s.dsrc_list)
+			@escape_list.concat(s.escape_list)
+			@esrc_list.concat(s.esrc_list)
+			@user_list.concat(s.user_list)
+			@usrc_list.concat(s.usrc_list)
+		end
+	end
+
+	class SimpleLaTeX
+		include MathML::Util
+		@@default_latex = nil
+		DEFAULT = {
+			:delimiter=>"\001",
+			:math_env_list=>[
+				/\$((?:\\.|[^\\\$])#{MathML::LaTeX::MBEC}*?)\$/m,
+				/\\\((#{MathML::LaTeX::MBEC}*?)\\\)/m
+			],
+			:dmath_env_list=>[
+				/\$\$(#{MathML::LaTeX::MBEC}*?)\$\$/m,
+				/\\\[(#{MathML::LaTeX::MBEC}*?)\\\]/m
+			],
+			:escape_list=>[
+				/\\(.)/m
+			],
+			:through_list=>[
+			],
+			:escape_any=> false,
+			:without_parse=>false
+		}
+
+		def initialize(options = {})
+			@params = DEFAULT.merge(options)
+			@params[:parser] = MathML::LaTeX::Parser.new unless @params[:parser] || @params[:without_parse]
+
+			@params[:math_envs] = collect_regexp(@params[:math_env_list])
+			@params[:dmath_envs] = collect_regexp(@params[:dmath_env_list])
+			@params[:escapes] = collect_regexp(@params[:escape_list])
+			@params[:throughs] = collect_regexp(@params[:through_list])
+			reset_encode_proc
+			reset_rescue_proc
+			reset_decode_proc
+			reset_unencode_proc
+		end
+
+		def reset_encode_proc
+			@encode_proc_re = INVALID_RE
+			@encode_proc = nil
+		end
+
+		def set_encode_proc(*re, &proc)
+			@encode_proc_re = collect_regexp(re)
+			@encode_proc = proc
+		end
+
+		def reset_rescue_proc
+			@rescue_proc = nil
+		end
+
+		def set_rescue_proc(&proc)
+			@rescue_proc = proc
+		end
+
+		def reset_decode_proc
+			@decode_proc = nil
+		end
+
+		def set_decode_proc(&proc)
+			@decode_proc = proc
+		end
+
+		def set_unencode_proc(&proc)
+			@unencode_proc = proc
+		end
+
+		def reset_unencode_proc
+			@unencode_proc = nil
+		end
+
+		def encode(src, *proc_re, &proc)
+			if proc_re.size>0 && proc_re[0].is_a?(MathData)
+				data = proc_re.shift
+			else
+				data = MathData.new
+			end
+
+			proc_re = proc_re.size==0 ? @encode_proc_re : collect_regexp(proc_re)
+			proc = @encode_proc unless proc
+
+			s = StringScanner.new(src)
+			encoded = ""
+
+			until s.eos?
+				if s.scan(/(.*?)(((((#{@params[:throughs]})|#{@params[:dmath_envs]})|#{@params[:math_envs]})|#{proc_re})|#{@params[:escapes]})/m)
+					encoded << s[1]
+					case
+					when s[6]
+						encoded << s[6]
+					when s[5], s[4]
+						env_src = s[5] || s[4]
+						if @params[:dmath_envs]=~env_src
+							encoded << "#{@params[:delimiter]}d#{data.dsrc_list.size}#{@params[:delimiter]}"
+							data.dsrc_list << env_src
+						else
+							encoded << "#{@params[:delimiter]}m#{data.msrc_list.size}#{@params[:delimiter]}"
+							data.msrc_list << env_src
+						end
+					when s[3]
+						size = s[3].size
+						s.pos = left = s.pos-size
+						if r=proc.call(s)
+							right = s.pos
+							encoded << "#{@params[:delimiter]}u#{data.user_list.size}#{@params[:delimiter]}"
+							data.user_list << r
+							data.usrc_list << s.string[left...right]
+						else
+							encoded << s.peek(size)
+							s.pos = s.pos+size
+						end
+					when s[2]
+						encoded << "#{@params[:delimiter]}e#{data.escape_list.size}#{@params[:delimiter]}"
+						@params[:escapes]=~s[2]
+						data.esrc_list << s[2]
+						data.escape_list << escapeXML($+, true)
+					end
+				else
+					encoded << s.rest
+					s.terminate
+				end
+			end
+
+			parse(data, @params[:parser]) unless @params[:without_parse]
+
+			return encoded, data
+		end
+
+		def error_to_html(e)
+			"<br />\n#{escapeXML(e.message)}<br />\n<code>#{escapeXML(e.done).gsub(/\n/, "<br />\n")}<strong>#{escapeXML(e.rest).gsub(/\n/, "<br />\n")}</strong></code><br />"
+		end
+
+		def latex_parser
+			@params[:parser] = MathML::LaTeX::Parser.new unless @params[:parser]
+			@params[:parser]
+		end
+
+		def parse(data, parser=nil)
+			parser = latex_parser unless parser
+			(data.math_list.size...data.msrc_list.size).each do |i|
+				begin
+					@params[:math_envs]=~data.msrc_list[i]
+					data.math_list[i] = parser.parse($+)
+				rescue MathML::LaTeX::ParseError => e
+					if @rescue_proc
+						data.math_list[i] = @rescue_proc.call(e)
+					else
+						data.math_list[i] = error_to_html(e)
+					end
+				end
+			end
+			(data.dmath_list.size...data.dsrc_list.size).each do |i|
+				begin
+					@params[:dmath_envs]=~data.dsrc_list[i]
+					data.dmath_list[i] = parser.parse($+, true)
+				rescue MathML::LaTeX::ParseError => e
+					if @rescue_proc
+						data.dmath_list[i] = @rescue_proc.call(e)
+					else
+						data.dmath_list[i] = error_to_html(e)
+					end
+				end
+			end
+		end
+
+		def decode(encoded, data, without_parsed = false, &proc)
+			return nil if encoded==nil
+			proc = @decode_proc unless proc
+			encoded.gsub(/#{Regexp.escape(@params[:delimiter])}([demu])(\d+)#{Regexp.escape(@params[:delimiter])}/) do
+				i = $2.to_i
+				t, d, s =
+					case $1
+					when "d"
+						[:dmath, without_parsed ? escapeXML(data.dsrc_list[i], true) : data.dmath_list[i], data.dsrc_list[i]]
+					when "e"
+						[:escape, data.escape_list[i], data.esrc_list[i]]
+					when "m"
+						[:math, without_parsed ? escapeXML(data.msrc_list[i], true) : data.math_list[i], data.msrc_list[i]]
+					when "u"
+						[:user, data.user_list[i], data.usrc_list[i]]
+					end
+				if proc
+					proc.call(d, :type=>t, :index=>i, :src=>s) || d
+				else
+					d
+				end
+			end
+		end
+
+		def decode_partial(type, encoded, data, &proc)
+			return nil if encoded==nil
+			head =
+				case type
+				when :math
+					"m"
+				when :dmath
+					"d"
+				when :escape
+					"e"
+				when :user
+					"u"
+				else
+					return
+				end
+			encoded.gsub(/#{Regexp.escape(@params[:delimiter])}#{head}(\d+)#{Regexp.escape(@params[:delimiter])}/) do
+				i = $1.to_i
+				t, d, s =
+					case head
+					when "d"
+						[:dmath, data.dmath_list[i], data.dsrc_list[i]]
+					when "e"
+						[:escape, data.escape_list[i], data.esrc_list[i]]
+					when "m"
+						[:math, data.math_list[i], data.msrc_list[i]]
+					when "u"
+						[:user, data.user_list[i], data.usrc_list[i]]
+					end
+				if proc
+					proc.call(d, :type=>t, :index=>i, :src=>s) || "#{@params[:delimiter]}#{head}#{i}#{@params[:delimiter]}"
+				else
+					d
+				end
+			end
+		end
+
+		def unencode(encoded, data, without_escape=false, &proc)
+			return nil if encoded==nil
+			proc = @unencode_proc unless proc
+			encoded.gsub(/#{Regexp.escape(@params[:delimiter])}([demu])(\d+)#{Regexp.escape(@params[:delimiter])}/) do
+				i = $2.to_i
+				t, s =
+					case $1
+					when "d"
+						[:dmath, data.dsrc_list[i]]
+					when "e"
+						[:escape, data.esrc_list[i]]
+					when "m"
+						[:math, data.msrc_list[i]]
+					when "u"
+						[:user, data.usrc_list[i]]
+					end
+				s = escapeXML(s, true) unless without_escape
+				if proc
+					proc.call(s, :type=>t, :index=>i) || s
+				else
+					s
+				end
+			end
+		end
+
+		def self.encode(src)
+			@@default_latex = self.new unless @@default_latex
+			@@default_latex.encode(src)
+		end
+
+		def self.decode(src, data)
+			@@default_latex.decode(src, data)
+		end
+
+		def parse_eqnarray(src, parser=nil)
+			src = "\\begin{array}{ccc}#{src}\\end{array}"
+			parser = latex_parser unless parser
+			begin
+				parser.parse(src, true)
+			rescue MathML::LaTeX::ParseError => e
+				e = MathML::LaTeX::ParseError.new(e.message,
+					e.rest.sub(/\\end\{array\}\z/, '\end{eqnarray}'),
+					e.done.sub(/\A\\begin\{array\}\{ccc\}/, '\begin{eqnarray}'))
+				@rescue_proc ? @rescue_proc.call(e) : error_to_html(e)
+			end
+		end
+
+		def parse_single_command(src, parser=nil)
+			s = src[SINGLE_COMMAND_RE, 1]
+			parser = latex_parser unless parser
+			begin
+				parser.parse(s)
+			rescue MathML::LaTeX::ParseError => e
+				src[SINGLE_COMMAND_RE, 2]
+			end
+		end
+	end
+end

data/lib/math_ml.rb

@@ -0,0 +1,25 @@
+# MathML Library
+#
+# Copyright (C) 2005, KURODA Hiraku <hiraku@hinet.mydns.jp>
+# You can redistribute it and/or modify it under GPL2.
+
+require "strscan"
+module MathML
+	require "eim_xml"
+
+	class XMLElement < EimXML::Element
+		def pop
+			@contents.pop
+		end
+	end
+
+	def self.pcstring(s, encoded=false)
+		EimXML::PCString.new(s, encoded)
+	end
+
+	class Error < StandardError; end
+end
+
+require "math_ml/element"
+require "math_ml/latex"
+require "math_ml/latex/builtin_commands"

data/spec/math_ml_element_spec.rb

@@ -0,0 +1,32 @@
+require "math_ml"
+
+describe MathML::Element do
+	it "#display_style and #as_display_style" do
+		MathML::Element.new("test").display_style.should == nil
+		e = MathML::Element.new("test")
+		r = e.as_display_style
+		r.should equal(e)
+		e.display_style.should be_true
+	end
+
+	it "#pop" do
+		e = MathML::Element.new("super")
+		s = MathML::Element.new("sub")
+
+		e.pop.should be_nil
+
+		e << s
+		e.pop.should equal(s)
+		e.pop.should be_nil
+
+		e << "text"
+		e.pop.should == "text"
+		e.pop.should be_nil
+	end
+
+	it "#to_s" do
+		e = MathML::Element.new("e")
+		e << "test<"
+		e.to_s.should == "<e>test&lt;</e>"
+	end
+end

data/spec/math_ml_latex_macro_spec.rb

@@ -0,0 +1,122 @@
+require "math_ml"
+require "spec/util"
+
+describe MathML::LaTeX::Macro do
+	include MathML::Spec::Util
+
+	before(:all) do
+		@src = <<'EOT'
+\newcommand{\newcom}{test}
+\newcommand{\paramcom}[2]{param2 #2, param1 #1.}
+\newcommand\ALPHA\alpha
+\newcommand\BETA[1]\beta
+\newcommand{\nothing}{}
+\newenvironment{newenv}{begin_newenv}{end_newenv}
+\newenvironment{paramenv}[2]{begin 1:#1, 2:#2}{end 2:#2 1:#1}
+\newenvironment{nothing}{}{}
+\newenvironment{separated environment}{sep}{env}
+\newenvironment ENV
+EOT
+	end
+
+	before do
+		@m = MathML::LaTeX::Macro.new
+		@m.parse(@src)
+	end
+
+	it "#parse" do
+		@m = MathML::LaTeX::Macro.new
+		lambda{@m.parse(@src)}.should_not raise_error
+
+		@m = MathML::LaTeX::Macro.new
+		lambda{@m.parse('\newcommand{notcommand}{}')}.should raise_parse_error("Need newcommand.", '\\newcommand{', "notcommand}{}")
+		lambda{@m.parse('\newcommand{\separated command}{}')}.should raise_parse_error("Syntax error.", '\newcommand{\separated', " command}{}")
+		lambda{@m.parse('\newcommand{\nobody}')}.should raise_parse_error("Need parameter.", '\newcommand{\nobody}', "")
+		lambda{@m.parse('\newcommand{\noparam}{#1}')}.should raise_parse_error("Parameter \# too large.", '\newcommand{\noparam}{#', "1}")
+		lambda{@m.parse('\newcommand{\overopt}[1]{#1#2}')}.should raise_parse_error("Parameter \# too large.", '\newcommand{\overopt}[1]{#1#', "2}")
+		lambda{@m.parse('\newcommand{\strangeopt}[-1]')}.should raise_parse_error("Need positive number.", '\newcommand{\strangeopt}[', "-1]")
+		lambda{@m.parse('\newcommand{\strangeopt}[a]')}.should raise_parse_error("Need positive number.", '\newcommand{\strangeopt}[', "a]")
+
+		lambda{@m.parse('\newenvironment{\command}{}{}')}.should raise_parse_error("Syntax error.", '\newenvironment{', '\command}{}{}')
+		lambda{@m.parse('\newenvironment{nobegin}')}.should raise_parse_error("Need begin block.", '\newenvironment{nobegin}', "")
+		lambda{@m.parse('\newenvironment{noend}{}')}.should raise_parse_error("Need end block.", '\newenvironment{noend}{}', "")
+		lambda{@m.parse('\newenvironment{noparam}{#1}{}')}.should raise_parse_error("Parameter \# too large.", '\newenvironment{noparam}{#', "1}{}")
+		lambda{@m.parse('\newenvironment{overparam}[1]{#1#2}{}')}.should raise_parse_error("Parameter \# too large.", '\newenvironment{overparam}[1]{#1#', "2}{}")
+		lambda{@m.parse('\newenvironment{strangeparam}[-1]{}{}')}.should raise_parse_error("Need positive number.", '\newenvironment{strangeparam}[', "-1]{}{}")
+		lambda{@m.parse('\newenvironment{strangeparam}[a]{}{}')}.should raise_parse_error("Need positive number.", '\newenvironment{strangeparam}[', "a]{}{}")
+
+		lambda{@m.parse('\newcommand{\valid}{OK} \invalid{\test}{NG}')}.should raise_parse_error("Syntax error.", '\newcommand{\valid}{OK} ', '\invalid{\test}{NG}')
+		lambda{@m.parse('\newcommand{\valid}{OK} invalid{\test}{NG}')}.should raise_parse_error("Syntax error.", '\newcommand{\valid}{OK} ', 'invalid{\test}{NG}')
+
+		lambda{@m.parse('\newcommand{\newcom}[test')}.should raise_parse_error("Option not closed.", '\newcommand{\newcom}', '[test')
+		lambda{@m.parse('\newcommand{\newcom}[1][test')}.should raise_parse_error("Option not closed.", '\newcommand{\newcom}[1]', '[test')
+		lambda{@m.parse('\newcommand{\newcom}[1][]{#1#2}')}.should raise_parse_error("Parameter \# too large.", '\newcommand{\newcom}[1][]{#1#', '2}')
+		lambda{@m.parse('\newenvironment{newenv}[1][test')}.should raise_parse_error("Option not closed.", '\newenvironment{newenv}[1]', '[test')
+		lambda{@m.parse('\newenvironment{newenv}[1][test')}.should raise_parse_error("Option not closed.", '\newenvironment{newenv}[1]', '[test')
+
+		lambda{@m.parse('\newcommand{\newcom')}.should raise_parse_error("Block not closed.", '\newcommand', '{\newcom')
+		lambda{@m.parse('\newcommand{\newcom}{test1{test2}{test3')}.should raise_parse_error("Block not closed.", '\newcommand{\newcom}', '{test1{test2}{test3')
+
+		lambda{@m.parse('\newenvironment{newenv}[1][]{#1 #2}')}.should raise_parse_error("Parameter \# too large.", '\newenvironment{newenv}[1][]{#1 #', '2}')
+	end
+
+	it "#commands" do
+		@m.commands("newcom").num.should == 0
+		@m.commands("paramcom").num.should == 2
+		@m.commands("no").should == nil
+	end
+
+	it "#expand_command" do
+		@m.expand_command("not coommand", []).should == nil
+
+		@m.expand_command("newcom", []).should == "test"
+		@m.expand_command("newcom", ["dummy_param"]).should == "test"
+		@m.expand_command("paramcom", ["1", "2"]).should == "param2 2, param1 1."
+		@m.expand_command("paramcom", ["12", "34"]).should == "param2 34, param1 12."
+		lambda{@m.expand_command("paramcom", ["12"])}.should raise_parse_error("Need more parameter.", "", "")
+		lambda{@m.expand_command("paramcom", [])}.should raise_parse_error("Need more parameter.", "", "")
+	end
+
+	it "#environments" do
+		@m.environments("newenv").num.should == 0
+		@m.environments("paramenv").num.should == 2
+		@m.environments("not_env").should == nil
+		@m.environments("separated environment").num.should == 0
+	end
+
+	it "#expand_environment" do
+		@m.expand_environment('notregistered', "dummy", []).should == nil
+		@m.expand_environment("newenv", "body", []).should == ' begin_newenv body end_newenv '
+		@m.expand_environment("paramenv", "body", ["1", "2"]).should == ' begin 1:1, 2:2 body end 2:2 1:1 '
+		@m.expand_environment("paramenv", "body", ["12", "34"]).should == ' begin 1:12, 2:34 body end 2:34 1:12 '
+		lambda{@m.expand_environment("paramenv", "body", ["1"])}.should raise_parse_error("Need more parameter.", "", "")
+		lambda{@m.expand_environment("paramenv", "body", [])}.should raise_parse_error("Need more parameter.", "", "")
+		@m.expand_environment("nothing", "body", []).should == '  body  '
+		@m.expand_environment("separated environment", "body", []).should == ' sep body env '
+		@m.expand_environment("E", "body", []).should == ' N body V '
+	end
+
+	it "#expand_with_options" do
+		src = <<'EOT'
+\newcommand{\opt}[1][x]{#1}
+\newcommand{\optparam}[2][]{#1#2}
+\newenvironment{newenv}[1][x]{s:#1}{e:#1}
+\newenvironment{optenv}[2][]{s:#1}{e:#2}
+EOT
+
+		m = MathML::LaTeX::Macro.new
+		m.parse(src)
+
+		m.expand_command("opt", []).should == 'x'
+		m.expand_command("opt", [], "1").should == '1'
+
+		m.expand_command("optparam", ["1"]).should == '1'
+		m.expand_command("optparam", ["1"], "2").should == '21'
+
+		m.expand_environment("newenv", "test", []).should == " s:x test e:x "
+		m.expand_environment("newenv", "test", [], "1").should == " s:1 test e:1 "
+
+		m.expand_environment("optenv", "test", ["1"]).should == " s: test e:1 "
+		m.expand_environment("optenv", "test", ["1"], "2").should == " s:2 test e:1 "
+	end
+end