maruku 0.4.0 → 0.4.1

data/lib/maruku/output/to_latex.rb

@@ -18,12 +18,25 @@
 #   Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #++
 
-
 module MaRuKu
-	
+
 class MDDocument
 
-	
+	Latex_preamble_enc_cjk = 
+"\\usepackage[C40]{fontenc}
+\\usepackage[cjkjis]{ucs}
+\\usepackage[utf8x]{inputenc}"
+
+	Latex_preamble_enc_utf8 = 
+"\\usepackage{ucs}
+\\usepackage[utf8x]{inputenc}"
+
+	def latex_require_package(p)
+		if not self.latex_required_packages.include? p
+			self.latex_required_packages.push p
+		end
+	end
+
 	# Render as a LaTeX fragment 
 	def to_latex
 		children_to_latex
@@ -31,25 +44,77 @@ class MDDocument
 
 	# Render as a complete LaTeX document 
 	def to_latex_document
-		header = ""
-		
-		if @doc.attributes[:latex_use_listings]
-			header += "\\usepackage{listings}\n"
-		end
-
 		body = to_latex
 		
 		body += render_latex_signature
 
+		required = 
+		self.latex_required_packages.map {|p|
+			"\\usepackage{#{p}}\n"
+		}.join
+		
+=begin maruku_doc
+Attribute: latex_cjk
+Scope:     document
+Output:    latex
+Summary:   Support for CJK characters.
+
+If the `latex_cjk` attribute is specified, then appropriate headers
+are added to the LaTeX preamble to support Japanese fonts.
+You have to have these fonts installed -- and this can be a pain.
+
+If `latex_cjk` is specified, this is added to the preamble:
+
+<?mrk md_code_block(Maruku::MDDocument::Latex_preamble_enc_cjk) ?>
+
+while the default is to add this:
+
+<?mrk md_code_block(Maruku::MDDocument::Latex_preamble_enc_utf8) ?>
+
+=end
+		encoding = @doc.attributes[:latex_cjk] ? 
+			Latex_preamble_enc_cjk : Latex_preamble_enc_utf8
+
+=begin maruku_doc
+Attribute: latex_preamble
+Scope:     document
+Output:    latex
+Summary:   User-defined preamble.
+
+If the `latex_preamble` attribute is specified, then its value
+will be used as a custom preamble. 
+
+For example:
+
+	Title: My document
+	Latex preamble: preamble.tex
+
+will produce:
+
+	...
+	\input{preamble.tex}
+	...
+
+=end
+	user_preamble = (file = @doc.attributes[:latex_preamble]) ?
+		"\\input{#{file}}\n" : ""
+		
 "\\documentclass{article}
-#{header}
+
+% Packages required to support encoding
+#{encoding}
+
+% Packages required by code
+#{required}
+
+#{user_preamble}
+
 \\usepackage{hyperref}
 \\usepackage{xspace}
 \\usepackage[usenames,dvipsnames]{color}
 \\usepackage[margin=1in]{geometry}
 \\hypersetup{colorlinks=true}
-%\\usepackage{ucs}
-%\\usepackage[utf8x]{inputenc}
+
 \\begin{document} 
 #{body}
 \\end{document}
@@ -74,34 +139,82 @@ module MaRuKu; module Out; module Latex
 	def to_latex_hrule; "\n\\vspace{.5em} \\hrule \\vspace{.5em}\n" end
 	def to_latex_linebreak; "\\linebreak " end
 	
-	def to_latex_paragraph; 
+	def to_latex_paragraph 
 		children_to_latex+"\n\n"
 	end
 
 	def get_setting(name, default=nil)
 		self.attributes[name] || @doc.attributes[name] || default
 	end
-	
+=begin maruku_doc
+Title: Input format for colors
+Output: latex, html
+Related: code_background_color
+
+Admissible formats:
+
+	green
+	#abc
+	#aabbcc
+=end
+
 	# \color[named]{name} 	
 	# \color[rgb]{1,0.2,0.3} 
 	def latex_color(s, command='color')
 		if s =~ /^\#(\w\w)(\w\w)(\w\w)$/
-			r = $1.hex; g = $2.hex; b=$3.hex				
+			r = $1.hex; g = $2.hex; b=$3.hex
 			# convert from 0-255 to 0.0-1.0
-			r = r / 255.0 
-			g = g / 255.0 
-			b = b / 255.0 
-			
+			r = r / 255.0; g = g / 255.0; b = b / 255.0; 
 			"\\#{command}[rgb]{%0.2f,%0.2f,%0.2f}" % [r,g,b]
-		else
+		elsif s =~ /^\#(\w)(\w)(\w)$/
+			r = $1.hex; g = $2.hex; b=$3.hex
+			# convert from 0-15 to 0.0-1.0
+			r = r / 15.0; g = g / 15.0; b = b / 15.0; 
+			"\\#{command}[rgb]{%0.2f,%0.2f,%0.2f}" % [r,g,b]
+		else	
 			"\\#{command}{#{s}}"
 		end
 	end
 	
 	def to_latex_code;
 		raw_code = self.raw_code
+=begin maruku_doc
+Attribute: latex_use_listings
+Scope: document
+Output: latex
+Summary: Support for `listings` package.
+Related: code_show_spaces, code_background_color, lang, code_lang
+
+If the `latex_use_listings` attribute is specified, then 
+code block are rendered using the `listings` package.
+Otherwise, a standard `verbatim` environment is used.
+
+*	If the `lang` attribute for the code block has been specified,
+	it gets passed to the `listings` package using the `lstset` macro.
+	The default lang for code blocks is specified through 
+	the `code_lang` attribute.
+
+		\lstset{language=ruby}
+
+	Please refer to the documentation of the `listings` package for
+	supported languages.
+	
+	If a language is not supported, the `listings` package will emit
+	a warning during the compilation. Just press enter and nothing
+	wrong will happen.
+
+*	If the `code_show_spaces` is specified, than spaces and tabs will
+	be shown using the macro:
+	
+		\lstset{showspaces=true,showtabs=true}
+	
+*	The background color is given by `code_background_color`.
+
+=end
 		
 		if @doc.attributes[:latex_use_listings]
+			@doc.latex_require_package('listings')
+				
 			s = "\\lstset{columns=fixed,frame=shadowbox}"
 
 			show_spaces = get_setting(:code_show_spaces) 
@@ -203,35 +316,31 @@ module MaRuKu; module Out; module Latex
 		"#{colorspec}{\\tt #{s}}"
 	end
 
-	def to_latex_immediate_link
+	def to_latex_im_link
 		url = self.url
-		return "\\href{#{url}}{#{url.to_latex}}"
+
+		if url[0,1] == '#'
+			url = url[1,url.size]
+			return "\\hyperlink{#{url}}{#{children_to_latex}}"
+		else
+			return "\\href{#{url}}{#{children_to_latex}}"
+		end
 	end
 	
 	def to_latex_link
-		if id = self.ref_id
-			# if empty, use text
-			if id.size == 0
-				id = children.to_s.downcase
-			end
-			
-			ref = @doc.refs[id]
-			if not ref
-				$stderr.puts "Could not find id = '#{id}'"
-				return children_to_latex
-			else
-				url = ref[:url]
-				#title = ref[:title] || 'no title'
-
-				if url[0,1] == '#'
-					url = url[1,url.size]
-					return "\\hyperlink{#{url}}{#{children_to_latex}}"
-				else
-					return "\\href{#{url}}{#{children_to_latex}}"
-				end
-			end
+		id = self.ref_id
+		# if empty, use text
+		if id.size == 0
+			id = children.to_s.downcase
+		end
+		
+		ref = @doc.refs[id]
+		if not ref
+			$stderr.puts "Could not find id = '#{id}'"
+			return children_to_latex
 		else
-			url = self.url
+			url = ref[:url]
+			#title = ref[:title] || 'no title'
 
 			if url[0,1] == '#'
 				url = url[1,url.size]
@@ -240,6 +349,7 @@ module MaRuKu; module Out; module Latex
 				return "\\href{#{url}}{#{children_to_latex}}"
 			end
 		end
+		
 	end
 	
 	def to_latex_email_address
@@ -297,7 +407,8 @@ module MaRuKu; module Out; module Latex
 	end
 	
 	def to_latex_raw_html
-		'{\bf Raw HTML removed in latex version }'
+		#'{\bf Raw HTML removed in latex version }'
+		""
 	end
 	
 	## Definition lists ###
@@ -371,6 +482,16 @@ module MaRuKu; module Out; module Latex
 				e << h
 			end
 		end
+		
+		# puts a space after commands if needed
+		e.each_index do |i|
+			if e[i] =~ /\\\w+\s*$/ # command
+				if (s=e[i+1]) && s[0] == ?\ # space
+					e[i]  = e[i] + "\\ "
+				end
+			end
+		end
+		
 		e.join(join_char)
 	end
 	

data/lib/maruku/output/to_latex_entities.rb

@@ -25,39 +25,64 @@ module MaRuKu; module Out; module Latex
 	
 	include REXML
 	
+	
 	def to_latex_entity 
 		entity_name = self.entity_name
+	
+		entity = ENTITY_TABLE[entity_name]
+		replace = entity.latex_string
 		
-		MDElement.init_tables
+		entity.latex_packages.each do |p|
+			@doc.latex_require_package p
+		end
+		
+		if replace =~ /^\\/
+			replace = replace + " "
+		end
 		
-		replace = @@entity_to_latex[entity_name]
  		if replace
-#			puts "#{entity_name} -> #{replace.inspect}"
 			return replace
 		else
-			$stderr.puts "Cannot translate #{entity_name}"
+			tell_user "Cannot translate entity #{entity_name.inspect} to LaTeX."
 			return entity_name
 		end
 	end
 	
+	class LatexEntity
+		safe_attr_accessor :html_num, Fixnum
+		safe_attr_accessor :html_entity, String
+		safe_attr_accessor :latex_string, String
+		safe_attr_accessor :latex_packages, Array
+	end
 	
 	# create hash @@entity_to_latex
-	def MDElement.init_tables
-		## why can't I use a class variable?
-		if not $conversion_table_inited
-			doc = Document.new XML_TABLE
-			@@entity_to_latex = {}
-			doc.elements.each("//char") do |c| 
-				num =  c.attributes['num']
-				name =  c.attributes['name']
-				convert =  c.attributes['convertTo']
-				@@entity_to_latex["#{num}"] = convert
-				@@entity_to_latex["#{name}"] = convert
-			end
-#			puts @@entity_to_latex.inspect
-			$conversion_table_inited = true
+	def Latex.init_entity_table
+		doc = Document.new XML_TABLE
+		doc.elements.each("//char") do |c| 
+			num =  c.attributes['num'].to_i
+			name =  c.attributes['name']
+			package =  c.attributes['package']
+			
+			convert =  c.attributes['convertTo']
+			convert.gsub!(/@DOUBLEQUOT/,'"')
+			convert.gsub!(/@QUOT/,"'")
+			convert.gsub!(/@GT/,">")
+			convert.gsub!(/@LT/,"<")
+			convert.gsub!(/@AMP/,"&")
+			convert.freeze
+			
+			e = LatexEntity.new
+			e.html_num = num
+			e.html_entity = name
+			e.latex_string = convert
+			e.latex_packages = package ? package.split : []
+			
+			ENTITY_TABLE[num] = e
+			ENTITY_TABLE[name] = e
 		end
 	end
+	
+	ENTITY_TABLE = {}
 
 # The following is a conversion chart for html elements, courtesy of 
 # text2html 
@@ -164,7 +189,7 @@ module MaRuKu; module Out; module Latex
 	  <char num='8745' name='cup' convertTo='$\\cup$' />
 	  <char num='8746' name='cap' convertTo='$\\cap$' />
 	  <char num='8747' name='int' convertTo='$\\int$' />
-	  <char num='8756' name='there4' convertTo='$\\therefore$' /> <!-- only AMS -->
+	  <char num='8756' name='there4' convertTo='$\\therefore$' package='amssymb' /> <!-- only AMS -->
 	  <char num='8764' name='sim' convertTo='$\\sim$' />
 	  <char num='8776' name='asymp' convertTo='$\\approx$' />
 	  <char num='8773' name='cong' convertTo='$\\cong$' />
@@ -175,9 +200,9 @@ module MaRuKu; module Out; module Latex
 	  <char num='8805' name='ge' convertTo='$\\geq$' />
 	  <char num='8834' name='sub' convertTo='$\\subset$' />
 	  <char num='8835' name='sup' convertTo='$\\supset$' />
-	  <char num='8838' name='sube' convertTo='$\\subseteq$' />
+<!--	  <char num='8838' name='sube' convertTo='$\\subseteq$' />-->
 	  <char num='8839' name='supe' convertTo='$\\supseteq$' />
-	  <char num='8836' name='nsub' convertTo='$\\nsubset$' /> <!-- only AMS -->
+<!--	  <char num='8836' name='nsub' convertTo='$\\nsubset$'  /> <!-- only AMS -->
 
 	  <char num='8853' name='oplus' convertTo='$\\oplus$' />
 	  <char num='8855' name='otimes' convertTo='$\\otimes$' />
@@ -190,13 +215,15 @@ module MaRuKu; module Out; module Latex
 	  <char num='9001' name='rang' convertTo='$\\rangle$' />
 
 	  <char num='9002' name='lang' convertTo='$\\langle$' />
-	  <char num='9674' name='loz' convertTo='$\\lozenge$' /> <!-- only AMS -->
+	  <char num='9674' name='loz' convertTo='$\\lozenge$' package='amssymb' /> <!-- only AMS -->
 	  <char num='9824' name='spades' convertTo='$\\spadesuit$' />
 	  <char num='9827' name='clubs' convertTo='$\\clubsuit$' />
 	  <char num='9829' name='hearts' convertTo='$\\heartsuit$' />
 	  <char num='9830' name='diams' convertTo='$\\diamondsuit$' />                                     
 	  <char num='38' name='amp' convertTo='\\@AMP' />    
-	  <char num='34' name='quot' convertTo='@DOUBLEQUOT' />
+<!--	  <char num='34' name='quot' convertTo='\\@DOUBLEQUOT' />  XXX -->
+	<char num='34' name='quot' convertTo='\"' /> 
+	<char num='39' name='apos' convertTo=\"'\" /> 
 	  <char num='169' name='copy' convertTo='\\copyright' />
 
 	  <char num='60' name='lt' convertTo='$@LT$' />
@@ -212,31 +239,32 @@ module MaRuKu; module Out; module Latex
 	  <char num='8211' name='ndash' convertTo='--' />
 	  <char num='8212' name='mdash' convertTo='---' />
 	  <char num='8216' name='lsquo' convertTo='`' />
-	  <char num='8217' name='rsquo' convertTo='@QUOT' />
+	  <char num='8217' name='rsquo' convertTo=\"'\" /> <!-- XXXX -->
 	  <char num='8220' name='ldquo' convertTo='``' />
-	  <char num='8221' name='rdquo' convertTo='@QUOT@QUOT' />
+	  <char num='8221' name='rdquo' convertTo=\"''\" /> <!-- XXXX -->
 	  <char num='8224' name='dagger' convertTo='\\dag' />
 	  <char num='8225' name='Dagger' convertTo='\\ddag' />
-	  <char num='8240' name='permil' convertTo='\\permil' /> <!-- wasysym package -->
+	  <char num='8240' name='permil' convertTo='\\permil' package='wasysym' /> <!-- wasysym package -->
 
-	  <char num='8364' name='euro' convertTo='\\euro' /> <!-- eurosym package -->
-	  <char num='8249' name='lsaquo' convertTo='\\guilsinglleft' />
-	  <char num='8250' name='rsaquo' convertTo='\\guilsinglright' />
+	  <char num='8364' name='euro' convertTo='\\euro' package='eurosym' /> <!-- eurosym package -->
+	  <char num='8249' name='lsaquo' convertTo='\\guilsinglleft' package='aeguill'/>
+	  <char num='8250' name='rsaquo' convertTo='\\guilsinglright' package='aeguill' />
 <!--	  <char num='160' name='nbsp' convertTo='\\nolinebreak' />-->
 	  <char num='160' name='nbsp' convertTo='~' />
 	  <char num='161' name='iexcl' convertTo='\\textexclamdown' />
 	  <char num='163' name='pound' convertTo='\\pounds' />
-	  <char num='164' name='curren' convertTo='\\currency' /> <!-- wasysym package -->
-	  <char num='165' name='yen' convertTo='\\textyen' /> <!-- textcomp -->
+	  <char num='164' name='curren' convertTo='\\currency' package='wasysym' /> <!-- wasysym package -->
+	  <char num='165' name='yen' convertTo='\\textyen' package='textcomp'/> <!-- textcomp -->
 
 	  <char num='166' name='brvbar' convertTo='\\brokenvert' /> <!-- wasysym -->
 	  <char num='167' name='sect' convertTo='\\S' />
-	  <char num='171' name='laquo' convertTo='\\guillemotleft' />
-	  <char num='187' name='raquo' convertTo='\\guillemotright' />
+	  <char num='171' name='laquo' convertTo='\\guillemotleft' package='aeguill'/>
+	  <char num='187' name='raquo' convertTo='\\guillemotright' package='aeguill'/>
 	  <char num='174' name='reg' convertTo='\\textregistered' />
 	  <char num='170' name='ordf' convertTo='\\textordfeminine' />
 	  <char num='172' name='not' convertTo='$\\neg$' />
-	  <char num='176' name='deg' convertTo='$\\degree$' /> <!-- mathabx -->
+	<!--  <char num='176' name='deg' convertTo='$\\degree$' /> <!-- mathabx -->
+	  <char num='176' name='deg' convertTo='\\textdegree' package='textcomp'/>
 
 	  <char num='177' name='plusmn' convertTo='$\\pm$' />
 	  <char num='180' name='acute' convertTo='@QUOT' />
@@ -244,14 +272,14 @@ module MaRuKu; module Out; module Latex
 	  <char num='182' name='para' convertTo='\\P' />
 	  <char num='183' name='middot' convertTo='$\\cdot$' />
 	  <char num='186' name='ordm' convertTo='\\textordmasculine' />
-	  <char num='162' name='cent' convertTo='\\cent' /> <!-- wasysym -->
+	  <char num='162' name='cent' convertTo='\\cent' package='wasysym' /> 
 	  <char num='185' name='sup1' convertTo='$^1$' />
 
 	  <char num='178' name='sup2' convertTo='$^2$' />
 	  <char num='179' name='sup3' convertTo='$^3$' />
 	  <char num='189' name='frac12' convertTo='$\\frac{1}{2}$' />
 	  <char num='188' name='frac14' convertTo='$\\frac{1}{4}$' />
-	  <char num='190' name='frac34' convertTo='$\\frac{3}{4}' />
+	  <char num='190' name='frac34' convertTo='$\\frac{3}{4}$' />
 	  <char num='192' name='Agrave' convertTo='\\`A' />
 	  <char num='193' name='Aacute' convertTo='\\@QUOTA' />
 	  <char num='194' name='Acirc' convertTo='\\^A' />
@@ -260,7 +288,7 @@ module MaRuKu; module Out; module Latex
 	  <char num='196' name='Auml' convertTo='\\@DOUBLEQUOTA' />
 	  <char num='197' name='Aring' convertTo='\\AA' />
 	  <char num='198' name='AElig' convertTo='\\AE' />
-	  <char num='199' name='Ccedil' convertTo='\\cC' />
+	  <char num='199' name='Ccedil' convertTo='\\c{C}' />
 	  <char num='200' name='Egrave' convertTo='\\`E' />
 	  <char num='201' name='Eacute' convertTo='\\@QUOTE' />
 	  <char num='202' name='Ecirc' convertTo='\\^E' />
@@ -272,7 +300,7 @@ module MaRuKu; module Out; module Latex
 	  <char num='208' name='ETH' convertTo='$\\eth$' /> <!-- AMS -->
 	  <char num='209' name='Ntilde' convertTo='\\~N' />    
 	  <char num='210' name='Ograve' convertTo='\\`O' />
-	  <char num='211' name='Oacute' convertTo='\\@QUOTO' />
+	  <char num='211' name='Oacute' convertTo='\\@QUOT O' />
 	  <char num='212' name='Ocirc' convertTo='\\^O' />
 	  <char num='213' name='Otilde' convertTo='\\~O' />
 	  <char num='214' name='Ouml' convertTo='\\@DOUBLEQUOTO' />
@@ -283,7 +311,6 @@ module MaRuKu; module Out; module Latex
 	  <char num='219' name='Ucirc' convertTo='\\^U' />
 	  <char num='220' name='Uuml' convertTo='\\@DOUBLEQUOTU' />
 	  <char num='221' name='Yacute' convertTo='\\@QUOTY' />
-	  <char num='222' name='THORN' convertTo='\\Thorn' />    <!-- wasysym -->
 	  <char num='223' name='szlig' convertTo='\\ss' />
 	  <char num='224' name='agrave' convertTo='\\`a' />
 	  <char num='225' name='aacute' convertTo='\\@QUOTa' />
@@ -292,7 +319,7 @@ module MaRuKu; module Out; module Latex
 	  <char num='228' name='auml' convertTo='\\@DOUBLEQUOTa' />
 	  <char num='229' name='aring' convertTo='\\aa' />
 	  <char num='230' name='aelig' convertTo='\\ae' />
-	  <char num='231' name='ccedil' convertTo='\\cc' />
+	  <char num='231' name='ccedil' convertTo='\\c{c}' />
 	  <char num='232' name='egrave' convertTo='\\`e' />
 	  <char num='233' name='eacute' convertTo='\\@QUOTe' />
 	  <char num='234' name='ecirc' convertTo='\\^e' />
@@ -301,23 +328,28 @@ module MaRuKu; module Out; module Latex
 	  <char num='237' name='iacute' convertTo='\\@QUOTi' />
 	  <char num='238' name='icirc' convertTo='\\^i' />
 	  <char num='239' name='iuml' convertTo='\\@DOUBLEQUOTi' />
-	  <char num='240' name='eth' convertTo='$\\eth$' /> <!-- -->
+	  <char num='240' name='eth' convertTo='$\\eth$' package='amssymb'/> <!-- -->
 	  <char num='241' name='ntilde' convertTo='\\~n' />
 	  <char num='242' name='ograve' convertTo='\\`o' />
 	  <char num='243' name='oacute' convertTo='\\@QUOTo' />
 	  <char num='244' name='ocirc' convertTo='\\^o' />
 	  <char num='245' name='otilde' convertTo='\\~o' />
 	  <char num='246' name='ouml' convertTo='\\@DOUBLEQUOTo' />
-	  <char num='247' name='divide' convertTo='$\\divide$' />       
+<!--   <char num='247' name='divide' convertTo='$\\divide$' />       -->
 	  <char num='248' name='oslash' convertTo='\\o' /> 
 	  <char num='249' name='ugrave' convertTo='\\`u' />
 	  <char num='250' name='uacute' convertTo='\\@QUOTu' />
 	  <char num='251' name='ucirc' convertTo='\\^u' />
 	  <char num='252' name='uuml' convertTo='\\@DOUBLEQUOTu' />
 	  <char num='253' name='yacute' convertTo='\\@QUOTy' />
-	  <char num='254' name='thorn' convertTo='\\thorn' /> <!-- wasysym -->
+
 	  <char num='255' name='yuml' convertTo='\\@DOUBLEQUOTy' />                  
+
+		<char num='222' name='THORN' convertTo='\\Thorn' package='wasysym' /> 
+		<char num='254' name='thorn' convertTo='\\thorn' package='wasysym' />
 	</chars>"
+	
+	init_entity_table
 
 end end end