1 # vim:ts=4:sw=4:
2 # = RedCloth - Textile and Markdown Hybrid for Ruby
3 #
4 # Homepage:: http://whytheluckystiff.net/ruby/redcloth/
5 # Author:: why the lucky stiff (http://whytheluckystiff.net/)
6 # Copyright:: (cc) 2004 why the lucky stiff (and his puppet organizations.)
7 # License:: BSD
8 #
9 # (see http://hobix.com/textile/ for a Textile Reference.)
10 #
11 # Based on (and also inspired by) both:
12 #
13 # PyTextile: http://diveintomark.org/projects/textile/textile.py.txt
14 # Textism for PHP: http://www.textism.com/tools/textile/
15 #
16 #
17
18 # = RedCloth
19 #
20 # RedCloth is a Ruby library for converting Textile and/or Markdown
21 # into HTML. You can use either format, intermingled or separately.
22 # You can also extend RedCloth to honor your own custom text stylings.
23 #
24 # RedCloth users are encouraged to use Textile if they are generating
25 # HTML and to use Markdown if others will be viewing the plain text.
26 #
27 # == What is Textile?
28 #
29 # Textile is a simple formatting style for text
30 # documents, loosely based on some HTML conventions.
31 #
32 # == Sample Textile Text
33 #
34 # h2. This is a title
35 #
36 # h3. This is a subhead
37 #
38 # This is a bit of paragraph.
39 #
40 # bq. This is a blockquote.
41 #
42 # = Writing Textile
43 #
44 # A Textile document consists of paragraphs. Paragraphs
45 # can be specially formatted by adding a small instruction
46 # to the beginning of the paragraph.
47 #
48 # h[n]. Header of size [n].
49 # bq. Blockquote.
50 # # Numeric list.
51 # * Bulleted list.
52 #
53 # == Quick Phrase Modifiers
54 #
55 # Quick phrase modifiers are also included, to allow formatting
56 # of small portions of text within a paragraph.
57 #
58 # \_emphasis\_
59 # \_\_italicized\_\_
60 # \*strong\*
61 # \*\*bold\*\*
62 # ??citation??
63 # -deleted text-
64 # +inserted text+
65 # ^superscript^
66 # ~subscript~
67 # @code@
68 # %(classname)span%
69 #
70 # ==notextile== (leave text alone)
71 #
72 # == Links
73 #
74 # To make a hypertext link, put the link text in "quotation
75 # marks" followed immediately by a colon and the URL of the link.
76 #
77 # Optional: text in (parentheses) following the link text,
78 # but before the closing quotation mark, will become a Title
79 # attribute for the link, visible as a tool tip when a cursor is above it.
80 #
81 # Example:
82 #
83 # "This is a link (This is a title) ":http://www.textism.com
84 #
85 # Will become:
86 #
87 # <a href="http://www.textism.com" title="This is a title">This is a link</a>
88 #
89 # == Images
90 #
91 # To insert an image, put the URL for the image inside exclamation marks.
92 #
93 # Optional: text that immediately follows the URL in (parentheses) will
94 # be used as the Alt text for the image. Images on the web should always
95 # have descriptive Alt text for the benefit of readers using non-graphical
96 # browsers.
97 #
98 # Optional: place a colon followed by a URL immediately after the
99 # closing ! to make the image into a link.
100 #
101 # Example:
102 #
103 # !http://www.textism.com/common/textist.gif(Textist)!
104 #
105 # Will become:
106 #
107 # <img src="http://www.textism.com/common/textist.gif" alt="Textist" />
108 #
109 # With a link:
110 #
111 # !/common/textist.gif(Textist)!:http://textism.com
112 #
113 # Will become:
114 #
115 # <a href="http://textism.com"><img src="/common/textist.gif" alt="Textist" /></a>
116 #
117 # == Defining Acronyms
118 #
119 # HTML allows authors to define acronyms via the tag. The definition appears as a
120 # tool tip when a cursor hovers over the acronym. A crucial aid to clear writing,
121 # this should be used at least once for each acronym in documents where they appear.
122 #
123 # To quickly define an acronym in Textile, place the full text in (parentheses)
124 # immediately following the acronym.
125 #
126 # Example:
127 #
128 # ACLU(American Civil Liberties Union)
129 #
130 # Will become:
131 #
132 # <acronym title="American Civil Liberties Union">ACLU</acronym>
133 #
134 # == Adding Tables
135 #
136 # In Textile, simple tables can be added by seperating each column by
137 # a pipe.
138 #
139 # |a|simple|table|row|
140 # |And|Another|table|row|
141 #
142 # Attributes are defined by style definitions in parentheses.
143 #
144 # table(border:1px solid black).
145 # (background:#ddd;color:red). |{}| | | |
146 #
147 # == Using RedCloth
148 #
149 # RedCloth is simply an extension of the String class, which can handle
150 # Textile formatting. Use it like a String and output HTML with its
151 # RedCloth#to_html method.
152 #
153 # doc = RedCloth.new "
154 #
155 # h2. Test document
156 #
157 # Just a simple test."
158 #
159 # puts doc.to_html
160 #
161 # By default, RedCloth uses both Textile and Markdown formatting, with
162 # Textile formatting taking precedence. If you want to turn off Markdown
163 # formatting, to boost speed and limit the processor:
164 #
165 # class RedCloth::Textile.new( str )
166
167 class RedCloth3 < String
168
169 VERSION = '3.0.4'
170 DEFAULT_RULES = [:textile, :markdown]
171
172 #
173 # Two accessor for setting security restrictions.
174 #
175 # This is a nice thing if you're using RedCloth for
176 # formatting in public places (e.g. Wikis) where you
177 # don't want users to abuse HTML for bad things.
178 #
179 # If +:filter_html+ is set, HTML which wasn't
180 # created by the Textile processor will be escaped.
181 #
182 # If +:filter_styles+ is set, it will also disable
183 # the style markup specifier. ('{color: red}')
184 #
185 attr_accessor :filter_html, :filter_styles
186
187 #
188 # Accessor for toggling hard breaks.
189 #
190 # If +:hard_breaks+ is set, single newlines will
191 # be converted to HTML break tags. This is the
192 # default behavior for traditional RedCloth.
193 #
194 attr_accessor :hard_breaks
195
196 # Accessor for toggling lite mode.
197 #
198 # In lite mode, block-level rules are ignored. This means
199 # that tables, paragraphs, lists, and such aren't available.
200 # Only the inline markup for bold, italics, entities and so on.
201 #
202 # r = RedCloth.new( "And then? She *fell*!", [:lite_mode] )
203 # r.to_html
204 # #=> "And then? She <strong>fell</strong>!"
205 #
206 attr_accessor :lite_mode
207
208 #
209 # Accessor for toggling span caps.
210 #
211 # Textile places `span' tags around capitalized
212 # words by default, but this wreaks havoc on Wikis.
213 # If +:no_span_caps+ is set, this will be
214 # suppressed.
215 #
216 attr_accessor :no_span_caps
217
218 #
219 # Establishes the markup predence. Available rules include:
220 #
221 # == Textile Rules
222 #
223 # The following textile rules can be set individually. Or add the complete
224 # set of rules with the single :textile rule, which supplies the rule set in
225 # the following precedence:
226 #
227 # refs_textile:: Textile references (i.e. [hobix]http://hobix.com/)
228 # block_textile_table:: Textile table block structures
229 # block_textile_lists:: Textile list structures
230 # block_textile_prefix:: Textile blocks with prefixes (i.e. bq., h2., etc.)
231 # inline_textile_image:: Textile inline images
232 # inline_textile_link:: Textile inline links
233 # inline_textile_span:: Textile inline spans
234 # glyphs_textile:: Textile entities (such as em-dashes and smart quotes)
235 #
236 # == Markdown
237 #
238 # refs_markdown:: Markdown references (for example: [hobix]: http://hobix.com/)
239 # block_markdown_setext:: Markdown setext headers
240 # block_markdown_atx:: Markdown atx headers
241 # block_markdown_rule:: Markdown horizontal rules
242 # block_markdown_bq:: Markdown blockquotes
243 # block_markdown_lists:: Markdown lists
244 # inline_markdown_link:: Markdown links
245 attr_accessor :rules
246
247 # Returns a new RedCloth object, based on _string_ and
248 # enforcing all the included _restrictions_.
249 #
250 # r = RedCloth.new( "h1. A <b>bold</b> man", [:filter_html] )
251 # r.to_html
252 # #=>"<h1>A <b>bold</b> man</h1>"
253 #
254 def initialize( string, restrictions = [] )
255 restrictions.each { |r| method( "#{ r }=" ).call( true ) }
256 super( string )
257 end
258
259 #
260 # Generates HTML from the Textile contents.
261 #
262 # r = RedCloth.new( "And then? She *fell*!" )
263 # r.to_html( true )
264 # #=>"And then? She <strong>fell</strong>!"
265 #
266 def to_html( *rules )
267 rules = DEFAULT_RULES if rules.empty?
268 # make our working copy
269 text = self.dup
270
271 @urlrefs = {}
272 @shelf = []
273 textile_rules = [:block_textile_table, :block_textile_lists,
274 :block_textile_prefix, :inline_textile_image, :inline_textile_link,
275 :inline_textile_code, :inline_textile_span, :glyphs_textile]
276 markdown_rules = [:refs_markdown, :block_markdown_setext, :block_markdown_atx, :block_markdown_rule,
277 :block_markdown_bq, :block_markdown_lists,
278 :inline_markdown_reflink, :inline_markdown_link]
279 @rules = rules.collect do |rule|
280 case rule
281 when :markdown
282 markdown_rules
283 when :textile
284 textile_rules
285 else
286 rule
287 end
288 end.flatten
289
290 # standard clean up
291 incoming_entities text
292 clean_white_space text
293
294 # start processor
295 @pre_list = []
296 rip_offtags text
297 no_textile text
298 escape_html_tags text
299 # need to do this before #hard_break and #blocks
300 block_textile_quotes text unless @lite_mode
301 hard_break text
302 unless @lite_mode
303 refs text
304 blocks text
305 end
306 inline text
307 smooth_offtags text
308
309 retrieve text
310
311 text.gsub!( /<\/?notextile>/, '' )
312 text.gsub!( /x%x%/, '&' )
313 clean_html text if filter_html
314 text.strip!
315 text
316
317 end
318
319 #######
320 private
321 #######
322 #
323 # Mapping of 8-bit ASCII codes to HTML numerical entity equivalents.
324 # (from PyTextile)
325 #
326 TEXTILE_TAGS =
327
328 [[128, 8364], [129, 0], [130, 8218], [131, 402], [132, 8222], [133, 8230],
329 [134, 8224], [135, 8225], [136, 710], [137, 8240], [138, 352], [139, 8249],
330 [140, 338], [141, 0], [142, 0], [143, 0], [144, 0], [145, 8216], [146, 8217],
331 [147, 8220], [148, 8221], [149, 8226], [150, 8211], [151, 8212], [152, 732],
332 [153, 8482], [154, 353], [155, 8250], [156, 339], [157, 0], [158, 0], [159, 376]].
333
334 collect! do |a, b|
335 [a.chr, ( b.zero? and "" or "&#{ b };" )]
336 end
337
338 #
339 # Regular expressions to convert to HTML.
340 #
341 A_HLGN = /(?:(?:<>|<|>|\=|[()]+)+)/
342 A_VLGN = /[\-^~]/
343 C_CLAS = '(?:\([^")]+\))'
344 C_LNGE = '(?:\[[^"\[\]]+\])'
345 C_STYL = '(?:\{[^"}]+\})'
346 S_CSPN = '(?:\\\\\d+)'
347 S_RSPN = '(?:/\d+)'
348 A = "(?:#{A_HLGN}?#{A_VLGN}?|#{A_VLGN}?#{A_HLGN}?)"
349 S = "(?:#{S_CSPN}?#{S_RSPN}|#{S_RSPN}?#{S_CSPN}?)"
350 C = "(?:#{C_CLAS}?#{C_STYL}?#{C_LNGE}?|#{C_STYL}?#{C_LNGE}?#{C_CLAS}?|#{C_LNGE}?#{C_STYL}?#{C_CLAS}?)"
351 # PUNCT = Regexp::quote( '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~' )
352 PUNCT = Regexp::quote( '!"#$%&\'*+,-./:;=?@\\^_`|~' )
353 PUNCT_NOQ = Regexp::quote( '!"#$&\',./:;=?@\\`|' )
354 PUNCT_Q = Regexp::quote( '*-_+^~%' )
355 HYPERLINK = '(\S+?)([^\w\s/;=\?]*?)(?=\s|<|$)'
356
357 # Text markup tags, don't conflict with block tags
358 SIMPLE_HTML_TAGS = [
359 'tt', 'b', 'i', 'big', 'small', 'em', 'strong', 'dfn', 'code',
360 'samp', 'kbd', 'var', 'cite', 'abbr', 'acronym', 'a', 'img', 'br',
361 'br', 'map', 'q', 'sub', 'sup', 'span', 'bdo'
362 ]
363
364 QTAGS = [
365 ['**', 'b', :limit],
366 ['*', 'strong', :limit],
367 ['??', 'cite', :limit],
368 ['-', 'del', :limit],
369 ['__', 'i', :limit],
370 ['_', 'em', :limit],
371 ['%', 'span', :limit],
372 ['+', 'ins', :limit],
373 ['^', 'sup', :limit],
374 ['~', 'sub', :limit]
375 ]
376 QTAGS_JOIN = QTAGS.map {|rc, ht, rtype| Regexp::quote rc}.join('|')
377
378 QTAGS.collect! do |rc, ht, rtype|
379 rcq = Regexp::quote rc
380 re =
381 case rtype
382 when :limit
383 /(^|[>\s\(]) # sta
384 (?!\-\-)
385 (#{QTAGS_JOIN}|) # oqs
386 (#{rcq}) # qtag
387 (\w|[^\s].*?[^\s]) # content
388 (?!\-\-)
389 #{rcq}
390 (#{QTAGS_JOIN}|) # oqa
391 (?=[[:punct:]]|<|\s|\)|$)/x
392 else
393 /(#{rcq})
394 (#{C})
395 (?::(\S+))?
396 (\w|[^\s\-].*?[^\s\-])
397 #{rcq}/xm
398 end
399 [rc, ht, re, rtype]
400 end
401
402 # Elements to handle
403 GLYPHS = [
404 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1’\2' ], # single closing
405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1’' ], # single closing
406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '’' ], # single closing
407 # [ /\'/, '‘' ], # single opening
408 # [ /</, '<' ], # less-than
409 # [ />/, '>' ], # greater-than
410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1”\2' ], # double closing
411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1”' ], # double closing
412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '”' ], # double closing
413 # [ /"/, '“' ], # double opening
414 # [ /\b( )?\.{3}/, '\1…' ], # ellipsis
415 # [ /\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/, '<acronym title="\2">\1</acronym>' ], # 3+ uppercase acronym
416 # [ /(^|[^"][>\s])([A-Z][A-Z0-9 ]+[A-Z0-9])([^<A-Za-z0-9]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
417 # [ /(\.\s)?\s?--\s?/, '\1—' ], # em dash
418 # [ /\s->\s/, ' → ' ], # right arrow
419 # [ /\s-\s/, ' – ' ], # en dash
420 # [ /(\d+) ?x ?(\d+)/, '\1×\2' ], # dimension sign
421 # [ /\b ?[(\[]TM[\])]/i, '™' ], # trademark
422 # [ /\b ?[(\[]R[\])]/i, '®' ], # registered
423 # [ /\b ?[(\[]C[\])]/i, '©' ] # copyright
424 ]
425
426 H_ALGN_VALS = {
427 '<' => 'left',
428 '=' => 'center',
429 '>' => 'right',
430 '<>' => 'justify'
431 }
432
433 V_ALGN_VALS = {
434 '^' => 'top',
435 '-' => 'middle',
436 '~' => 'bottom'
437 }
438
439 #
440 # Flexible HTML escaping
441 #
442 def htmlesc( str, mode=:Quotes )
443 if str
444 str.gsub!( '&', '&' )
445 str.gsub!( '"', '"' ) if mode != :NoQuotes
446 str.gsub!( "'", ''' ) if mode == :Quotes
447 str.gsub!( '<', '<')
448 str.gsub!( '>', '>')
449 end
450 str
451 end
452
453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
454 def pgl( text )
455 #GLYPHS.each do |re, resub, tog|
456 # next if tog and method( tog ).call
457 # text.gsub! re, resub
458 #end
459 text.gsub!(/\b([A-Z][A-Z0-9]{1,})\b(?:[(]([^)]*)[)])/) do |m|
460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
461 end
462 end
463
464 # Parses Textile attribute lists and builds an HTML attribute string
465 def pba( text_in, element = "" )
466
467 return '' unless text_in
468
469 style = []
470 text = text_in.dup
471 if element == 'td'
472 colspan = $1 if text =~ /\\(\d+)/
473 rowspan = $1 if text =~ /\/(\d+)/
474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
475 end
476
477 if text.sub!( /\{([^"}]*)\}/, '' ) && !filter_styles
478 sanitized = sanitize_styles($1)
479 style << "#{ sanitized };" unless sanitized.blank?
480 end
481
482 lang = $1 if
483 text.sub!( /\[([^)]+?)\]/, '' )
484
485 cls = $1 if
486 text.sub!( /\(([^()]+?)\)/, '' )
487
488 style << "padding-left:#{ $1.length }em;" if
489 text.sub!( /([(]+)/, '' )
490
491 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
492
493 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
494
495 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
496
497 atts = ''
498 atts << " style=\"#{ style.join }\"" unless style.empty?
499 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
500 atts << " lang=\"#{ lang }\"" if lang
501 atts << " id=\"#{ id }\"" if id
502 atts << " colspan=\"#{ colspan }\"" if colspan
503 atts << " rowspan=\"#{ rowspan }\"" if rowspan
504
505 atts
506 end
507
508 STYLES_RE = /^(color|width|height|border|background|padding|margin|font|text)(-[a-z]+)*:\s*((\d+%?|\d+px|\d+(\.\d+)?em|#[0-9a-f]+|[a-z]+)\s*)+$/i
509
510 def sanitize_styles(str)
511 styles = str.split(";").map(&:strip)
512 styles.reject! do |style|
513 !style.match(STYLES_RE)
514 end
515 styles.join(";")
516 end
517
518 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
519
520 # Parses a Textile table block, building HTML from the result.
521 def block_textile_table( text )
522 text.gsub!( TABLE_RE ) do |matches|
523
524 tatts, fullrow = $~[1..2]
525 tatts = pba( tatts, 'table' )
526 tatts = shelve( tatts ) if tatts
527 rows = []
528
529 fullrow.each_line do |row|
530 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
531 cells = []
532 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
533 next if cell == '|'
534 ctyp = 'd'
535 ctyp = 'h' if cell =~ /^_/
536
537 catts = ''
538 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
539
540 catts = shelve( catts ) if catts
541 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
542 end
543 ratts = shelve( ratts ) if ratts
544 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
545 end
546 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
547 end
548 end
549
550 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
551 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
552
553 # Parses Textile lists and generates HTML
554 def block_textile_lists( text )
555 text.gsub!( LISTS_RE ) do |match|
556 lines = match.split( /\n/ )
557 last_line = -1
558 depth = []
559 lines.each_with_index do |line, line_id|
560 if line =~ LISTS_CONTENT_RE
561 tl,atts,content = $~[1..3]
562 if depth.last
563 if depth.last.length > tl.length
564 (depth.length - 1).downto(0) do |i|
565 break if depth[i].length == tl.length
566 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
567 depth.pop
568 end
569 end
570 if depth.last and depth.last.length == tl.length
571 lines[line_id - 1] << '</li>'
572 end
573 end
574 unless depth.last == tl
575 depth << tl
576 atts = pba( atts )
577 atts = shelve( atts ) if atts
578 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
579 else
580 lines[line_id] = "\t\t<li>#{ content }"
581 end
582 last_line = line_id
583
584 else
585 last_line = line_id
586 end
587 if line_id - last_line > 1 or line_id == lines.length - 1
588 depth.delete_if do |v|
589 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
590 end
591 end
592 end
593 lines.join( "\n" )
594 end
595 end
596
597 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
598 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
599
600 def block_textile_quotes( text )
601 text.gsub!( QUOTES_RE ) do |match|
602 lines = match.split( /\n/ )
603 quotes = ''
604 indent = 0
605 lines.each do |line|
606 line =~ QUOTES_CONTENT_RE
607 bq,content = $1, $2
608 l = bq.count('>')
609 if l != indent
610 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
611 indent = l
612 end
613 quotes << (content + "\n")
614 end
615 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
616 quotes
617 end
618 end
619
620 CODE_RE = /(\W)
621 @
622 (?:\|(\w+?)\|)?
623 (.+?)
624 @
625 (?=\W)/x
626
627 def inline_textile_code( text )
628 text.gsub!( CODE_RE ) do |m|
629 before,lang,code,after = $~[1..4]
630 lang = " lang=\"#{ lang }\"" if lang
631 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false )
632 end
633 end
634
635 def lT( text )
636 text =~ /\#$/ ? 'o' : 'u'
637 end
638
639 def hard_break( text )
640 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
641 end
642
643 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
644
645 def blocks( text, deep_code = false )
646 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
647 plain = blk !~ /\A[#*> ]/
648
649 # skip blocks that are complex HTML
650 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
651 blk
652 else
653 # search for indentation levels
654 blk.strip!
655 if blk.empty?
656 blk
657 else
658 code_blk = nil
659 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
660 flush_left iblk
661 blocks iblk, plain
662 iblk.gsub( /^(\S)/, "\t\\1" )
663 if plain
664 code_blk = iblk; ""
665 else
666 iblk
667 end
668 end
669
670 block_applied = 0
671 @rules.each do |rule_name|
672 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
673 end
674 if block_applied.zero?
675 if deep_code
676 blk = "\t<pre><code>#{ blk }</code></pre>"
677 else
678 blk = "\t<p>#{ blk }</p>"
679 end
680 end
681 # hard_break blk
682 blk + "\n#{ code_blk }"
683 end
684 end
685
686 end.join( "\n\n" ) )
687 end
688
689 def textile_bq( tag, atts, cite, content )
690 cite, cite_title = check_refs( cite )
691 cite = " cite=\"#{ cite }\"" if cite
692 atts = shelve( atts ) if atts
693 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
694 end
695
696 def textile_p( tag, atts, cite, content )
697 atts = shelve( atts ) if atts
698 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
699 end
700
701 alias textile_h1 textile_p
702 alias textile_h2 textile_p
703 alias textile_h3 textile_p
704 alias textile_h4 textile_p
705 alias textile_h5 textile_p
706 alias textile_h6 textile_p
707
708 def textile_fn_( tag, num, atts, cite, content )
709 atts << " id=\"fn#{ num }\" class=\"footnote\""
710 content = "<sup>#{ num }</sup> #{ content }"
711 atts = shelve( atts ) if atts
712 "\t<p#{ atts }>#{ content }</p>"
713 end
714
715 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
716
717 def block_textile_prefix( text )
718 if text =~ BLOCK_RE
719 tag,tagpre,num,atts,cite,content = $~[1..6]
720 atts = pba( atts )
721
722 # pass to prefix handler
723 replacement = nil
724 if respond_to? "textile_#{ tag }", true
725 replacement = method( "textile_#{ tag }" ).call( tag, atts, cite, content )
726 elsif respond_to? "textile_#{ tagpre }_", true
727 replacement = method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content )
728 end
729 text.gsub!( $& ) { replacement } if replacement
730 end
731 end
732
733 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
734 def block_markdown_setext( text )
735 if text =~ SETEXT_RE
736 tag = if $2 == "="; "h1"; else; "h2"; end
737 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
738 blocks cont
739 text.replace( blk + cont )
740 end
741 end
742
743 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
744 [ ]*
745 (.+?) # $2 = Header text
746 [ ]*
747 \#* # optional closing #'s (not counted)
748 $/x
749 def block_markdown_atx( text )
750 if text =~ ATX_RE
751 tag = "h#{ $1.length }"
752 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
753 blocks cont
754 text.replace( blk + cont )
755 end
756 end
757
758 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
759
760 def block_markdown_bq( text )
761 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
762 blk.gsub!( /^ *> ?/, '' )
763 flush_left blk
764 blocks blk
765 blk.gsub!( /^(\S)/, "\t\\1" )
766 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
767 end
768 end
769
770 MARKDOWN_RULE_RE = /^(#{
771 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
772 })$/
773
774 def block_markdown_rule( text )
775 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
776 "<hr />"
777 end
778 end
779
780 # XXX TODO XXX
781 def block_markdown_lists( text )
782 end
783
784 def inline_textile_span( text )
785 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
786 text.gsub!( qtag_re ) do |m|
787
788 case rtype
789 when :limit
790 sta,oqs,qtag,content,oqa = $~[1..6]
791 atts = nil
792 if content =~ /^(#{C})(.+)$/
793 atts, content = $~[1..2]
794 end
795 else
796 qtag,atts,cite,content = $~[1..4]
797 sta = ''
798 end
799 atts = pba( atts )
800 atts = shelve( atts ) if atts
801
802 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
803
804 end
805 end
806 end
807
808 LINK_RE = /
809 (
810 ([\s\[{(]|[#{PUNCT}])? # $pre
811 " # start
812 (#{C}) # $atts
813 ([^"\n]+?) # $text
814 \s?
815 (?:\(([^)]+?)\)(?="))? # $title
816 ":
817 ( # $url
818 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
819 [\w\/]\S+?
820 )
821 (\/)? # $slash
822 ([^\w\=\/;\(\)]*?) # $post
823 )
824 (?=<|\s|$)
825 /x
826 #"
827 def inline_textile_link( text )
828 text.gsub!( LINK_RE ) do |m|
829 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
830 if text.include?('<br />')
831 all
832 else
833 url, url_title = check_refs( url )
834 title ||= url_title
835
836 # Idea below : an URL with unbalanced parethesis and
837 # ending by ')' is put into external parenthesis
838 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
839 url=url[0..-2] # discard closing parenth from url
840 post = ")"+post # add closing parenth to post
841 end
842 atts = pba( atts )
843 atts = " href=\"#{ htmlesc url }#{ slash }\"#{ atts }"
844 atts << " title=\"#{ htmlesc title }\"" if title
845 atts = shelve( atts ) if atts
846
847 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
848
849 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
850 end
851 end
852 end
853
854 MARKDOWN_REFLINK_RE = /
855 \[([^\[\]]+)\] # $text
856 [ ]? # opt. space
857 (?:\n[ ]*)? # one optional newline followed by spaces
858 \[(.*?)\] # $id
859 /x
860
861 def inline_markdown_reflink( text )
862 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
863 text, id = $~[1..2]
864
865 if id.empty?
866 url, title = check_refs( text )
867 else
868 url, title = check_refs( id )
869 end
870
871 atts = " href=\"#{ url }\""
872 atts << " title=\"#{ title }\"" if title
873 atts = shelve( atts )
874
875 "<a#{ atts }>#{ text }</a>"
876 end
877 end
878
879 MARKDOWN_LINK_RE = /
880 \[([^\[\]]+)\] # $text
881 \( # open paren
882 [ \t]* # opt space
883 <?(.+?)>? # $href
884 [ \t]* # opt space
885 (?: # whole title
886 (['"]) # $quote
887 (.*?) # $title
888 \3 # matching quote
889 )? # title is optional
890 \)
891 /x
892
893 def inline_markdown_link( text )
894 text.gsub!( MARKDOWN_LINK_RE ) do |m|
895 text, url, quote, title = $~[1..4]
896
897 atts = " href=\"#{ url }\""
898 atts << " title=\"#{ title }\"" if title
899 atts = shelve( atts )
900
901 "<a#{ atts }>#{ text }</a>"
902 end
903 end
904
905 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
906 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
907
908 def refs( text )
909 @rules.each do |rule_name|
910 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
911 end
912 end
913
914 def refs_textile( text )
915 text.gsub!( TEXTILE_REFS_RE ) do |m|
916 flag, url = $~[2..3]
917 @urlrefs[flag.downcase] = [url, nil]
918 nil
919 end
920 end
921
922 def refs_markdown( text )
923 text.gsub!( MARKDOWN_REFS_RE ) do |m|
924 flag, url = $~[2..3]
925 title = $~[6]
926 @urlrefs[flag.downcase] = [url, title]
927 nil
928 end
929 end
930
931 def check_refs( text )
932 ret = @urlrefs[text.downcase] if text
933 ret || [text, nil]
934 end
935
936 IMAGE_RE = /
937 (>|\s|^) # start of line?
938 \! # opening
939 (\<|\=|\>)? # optional alignment atts
940 (#{C}) # optional style,class atts
941 (?:\. )? # optional dot-space
942 ([^\s(!]+?) # presume this is the src
943 \s? # optional space
944 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
945 \! # closing
946 (?::#{ HYPERLINK })? # optional href
947 /x
948
949 def inline_textile_image( text )
950 text.gsub!( IMAGE_RE ) do |m|
951 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
952 htmlesc title
953 atts = pba( atts )
954 atts = " src=\"#{ htmlesc url.dup }\"#{ atts }"
955 atts << " title=\"#{ title }\"" if title
956 atts << " alt=\"#{ title }\""
957 # size = @getimagesize($url);
958 # if($size) $atts.= " $size[3]";
959
960 href, alt_title = check_refs( href ) if href
961 url, url_title = check_refs( url )
962
963 out = ''
964 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
965 out << "<img#{ shelve( atts ) } />"
966 out << "</a>#{ href_a1 }#{ href_a2 }" if href
967
968 if algn
969 algn = h_align( algn )
970 if stln == "<p>"
971 out = "<p style=\"float:#{ algn }\">#{ out }"
972 else
973 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
974 end
975 else
976 out = stln + out
977 end
978
979 out
980 end
981 end
982
983 def shelve( val )
984 @shelf << val
985 " :redsh##{ @shelf.length }:"
986 end
987
988 def retrieve( text )
989 @shelf.each_with_index do |r, i|
990 text.gsub!( " :redsh##{ i + 1 }:", r )
991 end
992 end
993
994 def incoming_entities( text )
995 ## turn any incoming ampersands into a dummy character for now.
996 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
997 ## implying an incoming html entity, to be skipped
998
999 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
1000 end
1001
1002 def no_textile( text )
1003 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
1004 '\1<notextile>\2</notextile>\3' )
1005 text.gsub!( /^ *==([^=]+.*?)==/m,
1006 '\1<notextile>\2</notextile>\3' )
1007 end
1008
1009 def clean_white_space( text )
1010 # normalize line breaks
1011 text.gsub!( /\r\n/, "\n" )
1012 text.gsub!( /\r/, "\n" )
1013 text.gsub!( /\t/, ' ' )
1014 text.gsub!( /^ +$/, '' )
1015 text.gsub!( /\n{3,}/, "\n\n" )
1016 text.gsub!( /"$/, "\" " )
1017
1018 # if entire document is indented, flush
1019 # to the left side
1020 flush_left text
1021 end
1022
1023 def flush_left( text )
1024 indt = 0
1025 if text =~ /^ /
1026 while text !~ /^ {#{indt}}\S/
1027 indt += 1
1028 end unless text.empty?
1029 if indt.nonzero?
1030 text.gsub!( /^ {#{indt}}/, '' )
1031 end
1032 end
1033 end
1034
1035 def footnote_ref( text )
1036 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1037 '<sup><a href="#fn\1">\1</a></sup>\2' )
1038 end
1039
1040 OFFTAGS = /(code|pre|kbd|notextile)/
1041 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1042 OFFTAG_OPEN = /<#{ OFFTAGS }/
1043 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1044 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1045 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1046
1047 def glyphs_textile( text, level = 0 )
1048 if text !~ HASTAG_MATCH
1049 pgl text
1050 footnote_ref text
1051 else
1052 codepre = 0
1053 text.gsub!( ALLTAG_MATCH ) do |line|
1054 ## matches are off if we're between <code>, <pre> etc.
1055 if $1
1056 if line =~ OFFTAG_OPEN
1057 codepre += 1
1058 elsif line =~ OFFTAG_CLOSE
1059 codepre -= 1
1060 codepre = 0 if codepre < 0
1061 end
1062 elsif codepre.zero?
1063 glyphs_textile( line, level + 1 )
1064 else
1065 htmlesc( line, :NoQuotes )
1066 end
1067 # p [level, codepre, line]
1068
1069 line
1070 end
1071 end
1072 end
1073
1074 def rip_offtags( text, escape_aftertag=true, escape_line=true )
1075 if text =~ /<.*>/
1076 ## strip and encode <pre> content
1077 codepre, used_offtags = 0, {}
1078 text.gsub!( OFFTAG_MATCH ) do |line|
1079 if $3
1080 first, offtag, aftertag = $3, $4, $5
1081 codepre += 1
1082 used_offtags[offtag] = true
1083 if codepre - used_offtags.length > 0
1084 htmlesc( line, :NoQuotes ) if escape_line
1085 @pre_list.last << line
1086 line = ""
1087 else
1088 ### htmlesc is disabled between CODE tags which will be parsed with highlighter
1089 ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
1090 ### NB: some changes were made not to use $N variables, because we use "match"
1091 ### and it breaks following lines
1092 htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/)
1093 line = "<redpre##{ @pre_list.length }>"
1094 first.match(/<#{ OFFTAGS }([^>]*)>/)
1095 tag = $1
1096 $2.to_s.match(/(class\=("[^"]+"|'[^']+'))/i)
1097 tag << " #{$1}" if $1
1098 @pre_list << "<#{ tag }>#{ aftertag }"
1099 end
1100 elsif $1 and codepre > 0
1101 if codepre - used_offtags.length > 0
1102 htmlesc( line, :NoQuotes ) if escape_line
1103 @pre_list.last << line
1104 line = ""
1105 end
1106 codepre -= 1 unless codepre.zero?
1107 used_offtags = {} if codepre.zero?
1108 end
1109 line
1110 end
1111 end
1112 text
1113 end
1114
1115 def smooth_offtags( text )
1116 unless @pre_list.empty?
1117 ## replace <pre> content
1118 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1119 end
1120 end
1121
1122 def inline( text )
1123 [/^inline_/, /^glyphs_/].each do |meth_re|
1124 @rules.each do |rule_name|
1125 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1126 end
1127 end
1128 end
1129
1130 def h_align( text )
1131 H_ALGN_VALS[text]
1132 end
1133
1134 def v_align( text )
1135 V_ALGN_VALS[text]
1136 end
1137
1138 def textile_popup_help( name, windowW, windowH )
1139 ' <a target="_blank" href="http://hobix.com/textile/#' + helpvar + '" onclick="window.open(this.href, \'popupwindow\', \'width=' + windowW + ',height=' + windowH + ',scrollbars,resizable\'); return false;">' + name + '</a><br />'
1140 end
1141
1142 # HTML cleansing stuff
1143 BASIC_TAGS = {
1144 'a' => ['href', 'title'],
1145 'img' => ['src', 'alt', 'title'],
1146 'br' => [],
1147 'i' => nil,
1148 'u' => nil,
1149 'b' => nil,
1150 'pre' => nil,
1151 'kbd' => nil,
1152 'code' => ['lang'],
1153 'cite' => nil,
1154 'strong' => nil,
1155 'em' => nil,
1156 'ins' => nil,
1157 'sup' => nil,
1158 'sub' => nil,
1159 'del' => nil,
1160 'table' => nil,
1161 'tr' => nil,
1162 'td' => ['colspan', 'rowspan'],
1163 'th' => nil,
1164 'ol' => nil,
1165 'ul' => nil,
1166 'li' => nil,
1167 'p' => nil,
1168 'h1' => nil,
1169 'h2' => nil,
1170 'h3' => nil,
1171 'h4' => nil,
1172 'h5' => nil,
1173 'h6' => nil,
1174 'blockquote' => ['cite']
1175 }
1176
1177 def clean_html( text, tags = BASIC_TAGS )
1178 text.gsub!( /<!\[CDATA\[/, '' )
1179 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1180 raw = $~
1181 tag = raw[2].downcase
1182 if tags.has_key? tag
1183 pcs = [tag]
1184 tags[tag].each do |prop|
1185 ['"', "'", ''].each do |q|
1186 q2 = ( q != '' ? q : '\s' )
1187 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1188 attrv = $1
1189 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1190 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1191 break
1192 end
1193 end
1194 end if tags[tag]
1195 "<#{raw[1]}#{pcs.join " "}>"
1196 else
1197 " "
1198 end
1199 end
1200 end
1201
1202 ALLOWED_TAGS = %w(redpre pre code notextile)
1203
1204 def escape_html_tags(text)
1205 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "<#{$1}#{'>' unless $3.blank?}" }
1206 end
1207 end
1208