bluecloth.rb

来自「用ruby on rails写的一个博客程序,还不错..ruby on rail」· RB 代码 · 共 1,145 行 · 第 1/2 页

#!/usr/bin/ruby
# 
# Bluecloth is a Ruby implementation of Markdown, a text-to-HTML conversion
# tool.
# 
# == Synopsis
# 
#   doc = BlueCloth::new "
#     ## Test document ##
#
#     Just a simple test.
#   "
#
#   puts doc.to_html
# 
# == Authors
# 
# * Michael Granger <[e-mail]>
# 
# == Contributors
#
# * Martin Chase <[e-mail]> - Peer review, helpful suggestions
# * Florian Gross <[e-mail]> - Filter options, suggestions
#
# == Copyright
#
# Original version:
#   Copyright (c) 2003-2004 John Gruber
#   <http://daringfireball.net/>  
#   All rights reserved.
#
# Ruby port:
#   Copyright (c) 2004 The FaerieMUD Consortium.
# 
# BlueCloth is free software; you can redistribute it and/or modify it under the
# terms of the GNU General Public License as published by the Free Software
# Foundation; either version 2 of the License, or (at your option) any later
# version.
# 
# BlueCloth is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
# A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
# 
# == To-do
#
# * Refactor some of the larger uglier methods that have to do their own
#   brute-force scanning because of lack of Perl features in Ruby's Regexp
#   class. Alternately, could add a dependency on 'pcre' and use most Perl
#   regexps.
#
# * Put the StringScanner in the render state for thread-safety.
#
# == Version
#
#  $Id: bluecloth.rb 69 2004-08-25 05:27:15Z ged $
# 

require 'digest/md5'
require 'logger'
require 'strscan'


### BlueCloth is a Ruby implementation of Markdown, a text-to-HTML conversion
### tool.
class BlueCloth < String

	### Exception class for formatting errors.
	class FormatError < RuntimeError

		### Create a new FormatError with the given source +str+ and an optional
		### message about the +specific+ error.
		def initialize( str, specific=nil )
			if specific
				msg = "Bad markdown format near %p: %s" % [ str, specific ]
			else
				msg = "Bad markdown format near %p" % str
			end

			super( msg )
		end
	end


	# Release Version
	Version = '0.0.3'

	# SVN Revision
	SvnRev = %q$Rev: 69 $

	# SVN Id tag
	SvnId = %q$Id: bluecloth.rb 69 2004-08-25 05:27:15Z ged $

	# SVN URL
	SvnUrl = %q$URL: svn+ssh://svn.faeriemud.org/usr/local/svn/BlueCloth/trunk/lib/bluecloth.rb $


	# Rendering state struct. Keeps track of URLs, titles, and HTML blocks
	# midway through a render. I prefer this to the globals of the Perl version
	# because globals make me break out in hives. Or something.
	RenderState = Struct::new( "RenderState", :urls, :titles, :html_blocks, :log )

	# Tab width for #detab! if none is specified
	TabWidth = 4

	# The tag-closing string -- set to '>' for HTML
	EmptyElementSuffix = "/>";

	# Table of MD5 sums for escaped characters
	EscapeTable = {}
	'\\`*_{}[]()#.!'.split(//).each {|char|
		hash = Digest::MD5::hexdigest( char )

		EscapeTable[ char ] = {
 			:md5 => hash,
			:md5re => Regexp::new( hash ),
			:re  => Regexp::new( '\\\\' + Regexp::escape(char) ),
		}
	}


	#################################################################
	###	I N S T A N C E   M E T H O D S
	#################################################################

	### Create a new BlueCloth string.
	def initialize( content="", *restrictions )
		@log = Logger::new( $deferr )
		@log.level = $DEBUG ?
			Logger::DEBUG :
			($VERBOSE ? Logger::INFO : Logger::WARN)
		@scanner = nil

		# Add any restrictions, and set the line-folding attribute to reflect
		# what happens by default.
		@filter_html = nil
		@filter_styles = nil
		restrictions.flatten.each {|r| __send__("#{r}=", true) }
		@fold_lines = true

		super( content )

		@log.debug "String is: %p" % self
	end


	######
	public
	######

	# Filters for controlling what gets output for untrusted input. (But really,
	# you're filtering bad stuff out of untrusted input at submission-time via
	# untainting, aren't you?)
	attr_accessor :filter_html, :filter_styles

	# RedCloth-compatibility accessor. Line-folding is part of Markdown syntax,
	# so this isn't used by anything.
	attr_accessor :fold_lines


	### Render Markdown-formatted text in this string object as HTML and return
	### it. The parameter is for compatibility with RedCloth, and is currently
	### unused, though that may change in the future.
	def to_html( lite=false )

		# Create a StringScanner we can reuse for various lexing tasks
		@scanner = StringScanner::new( '' )

		# Make a structure to carry around stuff that gets placeholdered out of
		# the source.
		rs = RenderState::new( {}, {}, {} )

		# Make a copy of the string with normalized line endings, tabs turned to
		# spaces, and a couple of guaranteed newlines at the end
		text = self.gsub( /\r\n?/, "\n" ).detab
		text += "\n\n"
		@log.debug "Normalized line-endings: %p" % text

		# Filter HTML if we're asked to do so
		if self.filter_html
			text.gsub!( "<", "&lt;" )
			text.gsub!( ">", "&gt;" )
			@log.debug "Filtered HTML: %p" % text
		end

		# Simplify blank lines
		text.gsub!( /^ +$/, '' )
		@log.debug "Tabs -> spaces/blank lines stripped: %p" % text

		# Replace HTML blocks with placeholders
		text = hide_html_blocks( text, rs )
		@log.debug "Hid HTML blocks: %p" % text
		@log.debug "Render state: %p" % rs

		# Strip link definitions, store in render state
		text = strip_link_definitions( text, rs )
		@log.debug "Stripped link definitions: %p" % text
		@log.debug "Render state: %p" % rs

		# Escape meta-characters
		text = escape_special_chars( text )
		@log.debug "Escaped special characters: %p" % text

		# Transform block-level constructs
		text = apply_block_transforms( text, rs )
		@log.debug "After block-level transforms: %p" % text

		# Now swap back in all the escaped characters
		text = unescape_special_chars( text )
		@log.debug "After unescaping special characters: %p" % text

		return text
	end
	

	### Convert tabs in +str+ to spaces.
	def detab( tabwidth=TabWidth )
		copy = self.dup
		copy.detab!( tabwidth )
		return copy
	end


	### Convert tabs to spaces in place and return self if any were converted.
	def detab!( tabwidth=TabWidth )
		newstr = self.split( /\n/ ).collect {|line|
			line.gsub( /(.*?)\t/ ) do
				$1 + ' ' * (tabwidth - $1.length % tabwidth)
			end
		}.join("\n")
		self.replace( newstr )
	end


	#######
	#private
	#######

	### Do block-level transforms on a copy of +str+ using the specified render
	### state +rs+ and return the results.
	def apply_block_transforms( str, rs )
		# Port: This was called '_runBlockGamut' in the original

		@log.debug "Applying block transforms to:\n  %p" % str
		text = transform_headers( str, rs )
		text = transform_hrules( text, rs )
		text = transform_lists( text, rs )
		text = transform_code_blocks( text, rs )
		text = transform_block_quotes( text, rs )
		text = transform_auto_links( text, rs )
		text = hide_html_blocks( text, rs )

		text = form_paragraphs( text, rs )

		@log.debug "Done with block transforms:\n  %p" % text
		return text
	end


	### Apply Markdown span transforms to a copy of the specified +str+ with the
	### given render state +rs+ and return it.
	def apply_span_transforms( str, rs )
		@log.debug "Applying span transforms to:\n  %p" % str

		str = transform_code_spans( str, rs )
		str = encode_html( str )
		str = transform_images( str, rs )
		str = transform_anchors( str, rs )
		str = transform_italic_and_bold( str, rs )

		# Hard breaks
		str.gsub!( / {2,}\n/, "<br#{EmptyElementSuffix}\n" )

		@log.debug "Done with span transforms:\n  %p" % str
		return str
	end


	# The list of tags which are considered block-level constructs and an
	# alternation pattern suitable for use in regexps made from the list
	StrictBlockTags = %w[ p div h[1-6] blockquote pre table dl ol ul script noscript
		form fieldset iframe math ins del ]
	StrictTagPattern = StrictBlockTags.join('|')

	LooseBlockTags = StrictBlockTags - %w[ins del]
	LooseTagPattern = LooseBlockTags.join('|')

	# Nested blocks:
	# 	<div>
	# 		<div>
	# 		tags for inner block must be indented.
	# 		</div>
	# 	</div>
	StrictBlockRegex = %r{
		^						# Start of line
		<(#{StrictTagPattern})	# Start tag: \2
		\b						# word break
		(.*\n)*?				# Any number of lines, minimal match
		</\1>					# Matching end tag
		[ ]*					# trailing spaces
		$						# End of line or document
	  }ix

	# More-liberal block-matching
	LooseBlockRegex = %r{
		^						# Start of line
		<(#{LooseTagPattern})	# start tag: \2
		\b						# word break
		(.*\n)*?				# Any number of lines, minimal match
		.*</\1>					# Anything + Matching end tag
		[ ]*					# trailing spaces
		$						# End of line or document
	  }ix

	# Special case for <hr />.
	HruleBlockRegex = %r{
		(						# $1
			\A\n?				# Start of doc + optional \n
			|					# or
			.*\n\n				# anything + blank line
		)
		(						# save in $2
			[ ]*				# Any spaces
			<hr					# Tag open
			\b					# Word break
			([^<>])*?			# Attributes
			/?>					# Tag close
			$					# followed by a blank line or end of document
		)
	  }ix

	### Replace all blocks of HTML in +str+ that start in the left margin with
	### tokens.
	def hide_html_blocks( str, rs )
		@log.debug "Hiding HTML blocks in %p" % str
		
		# Tokenizer proc to pass to gsub
		tokenize = lambda {|match|
			key = Digest::MD5::hexdigest( match )
			rs.html_blocks[ key ] = match
			@log.debug "Replacing %p with %p" % [ match, key ]
			"\n\n#{key}\n\n"
		}

		rval = str.dup

		@log.debug "Finding blocks with the strict regex..."
		rval.gsub!( StrictBlockRegex, &tokenize )

		@log.debug "Finding blocks with the loose regex..."
		rval.gsub!( LooseBlockRegex, &tokenize )

		@log.debug "Finding hrules..."
		rval.gsub!( HruleBlockRegex ) {|match| $1 + tokenize[$2] }

		return rval
	end


	# Link defs are in the form: ^[id]: url "optional title"
	LinkRegex = %r{
		^[ ]*\[(.+)\]:		# id = $1
		  [ ]*
		  \n?				# maybe *one* newline
		  [ ]*
		<?(\S+?)>?				# url = $2
		  [ ]*
		  \n?				# maybe one newline
		  [ ]*
		(?:
			# Titles are delimited by "quotes" or (parens).
			["(]
			(.+?)			# title = $3
			[")]			# Matching ) or "
			[ ]*
		)?	# title is optional
		(?:\n+|\Z)
	  }x

	### Strip link definitions from +str+, storing them in the given RenderState
	### +rs+.
	def strip_link_definitions( str, rs )
		str.gsub( LinkRegex ) {|match|
			id, url, title = $1, $2, $3

			rs.urls[ id.downcase ] = encode_html( url )
			unless title.nil?
				rs.titles[ id.downcase ] = title.gsub( /"/, "&quot;" )
			end
			""
		}
	end


	### Escape special characters in the given +str+
	def escape_special_chars( str )
		@log.debug "  Escaping special characters"
		text = ''

		# The original Markdown source has something called '$tags_to_skip'
		# declared here, but it's never used, so I don't define it.

		tokenize_html( str ) {|token, str|
			@log.debug "   Adding %p token %p" % [ token, str ]
			case token

			# Within tags, encode * and _
			when :tag
				text += str.
					gsub( /\*/, EscapeTable['*'][:md5] ).
					gsub( /_/, EscapeTable['_'][:md5] )

			# Encode backslashed stuff in regular text
			when :text
				text += encode_backslash_escapes( str )
			else
				raise TypeError, "Unknown token type %p" % token
			end
		}

		@log.debug "  Text with escapes is now: %p" % text
		return text
	end


	### Swap escaped special characters in a copy of the given +str+ and return
	### it.
	def unescape_special_chars( str )
		EscapeTable.each {|char, hash|
			@log.debug "Unescaping escaped %p with %p" % [ char, hash[:md5re] ]
			str.gsub!( hash[:md5re], char )
		}

		return str
	end


	### Return a copy of the given +str+ with any backslashed special character
	### in it replaced with MD5 placeholders.
	def encode_backslash_escapes( str )
		# Make a copy with any double-escaped backslashes encoded
		text = str.gsub( /\\\\/, EscapeTable['\\'][:md5] )
		
		EscapeTable.each_pair {|char, esc|
			next if char == '\\'
			text.gsub!( esc[:re], esc[:md5] )
		}

		return text
	end


	### Transform any Markdown-style horizontal rules in a copy of the specified
	### +str+ and return it.
	def transform_hrules( str, rs )
		@log.debug " Transforming horizontal rules"
		str.gsub( /^( ?[\-\*_] ?){3,}$/, "\n<hr#{EmptyElementSuffix}\n" )
	end



	# Patterns to match and transform lists
	ListMarkerOl = %r{\d+\.}
	ListMarkerUl = %r{[*+-]}
	ListMarkerAny = Regexp::union( ListMarkerOl, ListMarkerUl )

	ListRegexp = %r{
		  (?:
			^[ ]{0,#{TabWidth - 1}}		# Indent < tab width
			(#{ListMarkerAny})			# unordered or ordered ($1)
			[ ]+						# At least one space
		  )
		  (?m:.+?)						# item content (include newlines)
		  (?:
			  \z						# Either EOF
			|							#  or
			  \n{2,}					# Blank line...
			  (?=\S)					# ...followed by non-space
			  (?![ ]*					# ...but not another item
				(#{ListMarkerAny})
			   [ ]+)
		  )
	  }x

	### Transform Markdown-style lists in a copy of the specified +str+ and
	### return it.
	def transform_lists( str, rs )
		@log.debug " Transforming lists at %p" % (str[0,100] + '...')

		str.gsub( ListRegexp ) {|list|
			@log.debug "  Found list %p" % list
			bullet = $1
			list_type = (ListMarkerUl.match(bullet) ? "ul" : "ol")
			list.gsub!( /\n{2,}/, "\n\n\n" )

			%{<%s>\n%s</%s>\n} % [
				list_type,
				transform_list_items( list, rs ),
				list_type,
			]
		}
	end


	# Pattern for transforming list items
	ListItemRegexp = %r{
		(\n)?							# leading line = $1
		(^[ ]*)							# leading whitespace = $2
		(#{ListMarkerAny}) [ ]+			# list marker = $3
		((?m:.+?)						# list item text   = $4
		(\n{1,2}))
		(?= \n* (\z | \2 (#{ListMarkerAny}) [ ]+))
	  }x

	### Transform list items in a copy of the given +str+ and return it.
	def transform_list_items( str, rs )
		@log.debug " Transforming list items"

		# Trim trailing blank lines
		str = str.sub( /\n{2,}\z/, "\n" )

		str.gsub( ListItemRegexp ) {|line|
			@log.debug "  Found item line %p" % line
			leading_line, item = $1, $4

			if leading_line or /\n{2,}/.match( item )
				@log.debug "   Found leading line or item has a blank"
				item = apply_block_transforms( outdent(item), rs )
			else
				# Recursion for sub-lists
				@log.debug "   Recursing for sublist"
				item = transform_lists( outdent(item), rs ).chomp
				item = apply_span_transforms( item, rs )
			end

			%{<li>%s</li>\n} % item
		}
	end


	# Pattern for matching codeblocks
	CodeBlockRegexp = %r{
		(?:\n\n|\A)
		(									# $1 = the code block
		  (?:
			(?:[ ]{#{TabWidth}} | \t)		# a tab or tab-width of spaces
			.*\n+
		  )+
		)
		(^[ ]{0,#{TabWidth - 1}}\S|\Z)		# Lookahead for non-space at
											# line-start, or end of doc
	  }x

	### Transform Markdown-style codeblocks in a copy of the specified +str+ and
	### return it.
	def transform_code_blocks( str, rs )
		@log.debug " Transforming code blocks"

		str.gsub( CodeBlockRegexp ) {|block|
			codeblock = $1
			remainder = $2

			# Generate the codeblock
			%{\n\n<pre><code>%s\n</code></pre>\n\n%s} %
				[ encode_code( outdent(codeblock), rs ).rstrip, remainder ]
		}
	end


	# Pattern for matching Markdown blockquote blocks
	BlockQuoteRegexp = %r{
		  (?:
			^[ ]*>[ ]?		# '>' at the start of a line
			  .+\n			# rest of the first line