##// END OF EJS Templates
Fixed: some textile modifiers combinations don't work, eg. bold inside underline (#5045)....
Jean-Philippe Lang -
r3447:4e5078ebb0fb
parent child
Show More
@@ -0,0 +1,45
1 # Redmine - project management software
2 # Copyright (C) 2006-2010 Jean-Philippe Lang
3 #
4 # This program is free software; you can redistribute it and/or
5 # modify it under the terms of the GNU General Public License
6 # as published by the Free Software Foundation; either version 2
7 # of the License, or (at your option) any later version.
8 #
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU General Public License for more details.
13 #
14 # You should have received a copy of the GNU General Public License
15 # along with this program; if not, write to the Free Software
16 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
17
18 require File.dirname(__FILE__) + '/../../../../test_helper'
19
20 class Redmine::WikiFormatting::TextileFormatterTest < HelperTestCase
21
22 def setup
23 @formatter = Redmine::WikiFormatting::Textile::Formatter
24 end
25
26 MODIFIERS = {
27 "*" => 'strong', # bold
28 "_" => 'em', # italic
29 "+" => 'ins', # underline
30 "-" => 'del', # deleted
31 "^" => 'sup', # superscript
32 "~" => 'sub' # subscript
33 }
34
35 def test_modifiers_combination
36 MODIFIERS.each do |m1, tag1|
37 MODIFIERS.each do |m2, tag2|
38 next if m1 == m2
39 text = "#{m2}#{m1}Phrase modifiers#{m1}#{m2}"
40 html = "<p><#{tag2}><#{tag1}>Phrase modifiers</#{tag1}></#{tag2}></p>"
41 assert_equal html, @formatter.new(text).to_html
42 end
43 end
44 end
45 end
@@ -1,1182 +1,1186
1 1 # vim:ts=4:sw=4:
2 2 # = RedCloth - Textile and Markdown Hybrid for Ruby
3 3 #
4 4 # Homepage:: http://whytheluckystiff.net/ruby/redcloth/
5 5 # Author:: why the lucky stiff (http://whytheluckystiff.net/)
6 6 # Copyright:: (cc) 2004 why the lucky stiff (and his puppet organizations.)
7 7 # License:: BSD
8 8 #
9 9 # (see http://hobix.com/textile/ for a Textile Reference.)
10 10 #
11 11 # Based on (and also inspired by) both:
12 12 #
13 13 # PyTextile: http://diveintomark.org/projects/textile/textile.py.txt
14 14 # Textism for PHP: http://www.textism.com/tools/textile/
15 15 #
16 16 #
17 17
18 18 # = RedCloth
19 19 #
20 20 # RedCloth is a Ruby library for converting Textile and/or Markdown
21 21 # into HTML. You can use either format, intermingled or separately.
22 22 # You can also extend RedCloth to honor your own custom text stylings.
23 23 #
24 24 # RedCloth users are encouraged to use Textile if they are generating
25 25 # HTML and to use Markdown if others will be viewing the plain text.
26 26 #
27 27 # == What is Textile?
28 28 #
29 29 # Textile is a simple formatting style for text
30 30 # documents, loosely based on some HTML conventions.
31 31 #
32 32 # == Sample Textile Text
33 33 #
34 34 # h2. This is a title
35 35 #
36 36 # h3. This is a subhead
37 37 #
38 38 # This is a bit of paragraph.
39 39 #
40 40 # bq. This is a blockquote.
41 41 #
42 42 # = Writing Textile
43 43 #
44 44 # A Textile document consists of paragraphs. Paragraphs
45 45 # can be specially formatted by adding a small instruction
46 46 # to the beginning of the paragraph.
47 47 #
48 48 # h[n]. Header of size [n].
49 49 # bq. Blockquote.
50 50 # # Numeric list.
51 51 # * Bulleted list.
52 52 #
53 53 # == Quick Phrase Modifiers
54 54 #
55 55 # Quick phrase modifiers are also included, to allow formatting
56 56 # of small portions of text within a paragraph.
57 57 #
58 58 # \_emphasis\_
59 59 # \_\_italicized\_\_
60 60 # \*strong\*
61 61 # \*\*bold\*\*
62 62 # ??citation??
63 63 # -deleted text-
64 64 # +inserted text+
65 65 # ^superscript^
66 66 # ~subscript~
67 67 # @code@
68 68 # %(classname)span%
69 69 #
70 70 # ==notextile== (leave text alone)
71 71 #
72 72 # == Links
73 73 #
74 74 # To make a hypertext link, put the link text in "quotation
75 75 # marks" followed immediately by a colon and the URL of the link.
76 76 #
77 77 # Optional: text in (parentheses) following the link text,
78 78 # but before the closing quotation mark, will become a Title
79 79 # attribute for the link, visible as a tool tip when a cursor is above it.
80 80 #
81 81 # Example:
82 82 #
83 83 # "This is a link (This is a title) ":http://www.textism.com
84 84 #
85 85 # Will become:
86 86 #
87 87 # <a href="http://www.textism.com" title="This is a title">This is a link</a>
88 88 #
89 89 # == Images
90 90 #
91 91 # To insert an image, put the URL for the image inside exclamation marks.
92 92 #
93 93 # Optional: text that immediately follows the URL in (parentheses) will
94 94 # be used as the Alt text for the image. Images on the web should always
95 95 # have descriptive Alt text for the benefit of readers using non-graphical
96 96 # browsers.
97 97 #
98 98 # Optional: place a colon followed by a URL immediately after the
99 99 # closing ! to make the image into a link.
100 100 #
101 101 # Example:
102 102 #
103 103 # !http://www.textism.com/common/textist.gif(Textist)!
104 104 #
105 105 # Will become:
106 106 #
107 107 # <img src="http://www.textism.com/common/textist.gif" alt="Textist" />
108 108 #
109 109 # With a link:
110 110 #
111 111 # !/common/textist.gif(Textist)!:http://textism.com
112 112 #
113 113 # Will become:
114 114 #
115 115 # <a href="http://textism.com"><img src="/common/textist.gif" alt="Textist" /></a>
116 116 #
117 117 # == Defining Acronyms
118 118 #
119 119 # HTML allows authors to define acronyms via the tag. The definition appears as a
120 120 # tool tip when a cursor hovers over the acronym. A crucial aid to clear writing,
121 121 # this should be used at least once for each acronym in documents where they appear.
122 122 #
123 123 # To quickly define an acronym in Textile, place the full text in (parentheses)
124 124 # immediately following the acronym.
125 125 #
126 126 # Example:
127 127 #
128 128 # ACLU(American Civil Liberties Union)
129 129 #
130 130 # Will become:
131 131 #
132 132 # <acronym title="American Civil Liberties Union">ACLU</acronym>
133 133 #
134 134 # == Adding Tables
135 135 #
136 136 # In Textile, simple tables can be added by seperating each column by
137 137 # a pipe.
138 138 #
139 139 # |a|simple|table|row|
140 140 # |And|Another|table|row|
141 141 #
142 142 # Attributes are defined by style definitions in parentheses.
143 143 #
144 144 # table(border:1px solid black).
145 145 # (background:#ddd;color:red). |{}| | | |
146 146 #
147 147 # == Using RedCloth
148 148 #
149 149 # RedCloth is simply an extension of the String class, which can handle
150 150 # Textile formatting. Use it like a String and output HTML with its
151 151 # RedCloth#to_html method.
152 152 #
153 153 # doc = RedCloth.new "
154 154 #
155 155 # h2. Test document
156 156 #
157 157 # Just a simple test."
158 158 #
159 159 # puts doc.to_html
160 160 #
161 161 # By default, RedCloth uses both Textile and Markdown formatting, with
162 162 # Textile formatting taking precedence. If you want to turn off Markdown
163 163 # formatting, to boost speed and limit the processor:
164 164 #
165 165 # class RedCloth::Textile.new( str )
166 166
167 167 class RedCloth3 < String
168 168
169 169 VERSION = '3.0.4'
170 170 DEFAULT_RULES = [:textile, :markdown]
171 171
172 172 #
173 173 # Two accessor for setting security restrictions.
174 174 #
175 175 # This is a nice thing if you're using RedCloth for
176 176 # formatting in public places (e.g. Wikis) where you
177 177 # don't want users to abuse HTML for bad things.
178 178 #
179 179 # If +:filter_html+ is set, HTML which wasn't
180 180 # created by the Textile processor will be escaped.
181 181 #
182 182 # If +:filter_styles+ is set, it will also disable
183 183 # the style markup specifier. ('{color: red}')
184 184 #
185 185 attr_accessor :filter_html, :filter_styles
186 186
187 187 #
188 188 # Accessor for toggling hard breaks.
189 189 #
190 190 # If +:hard_breaks+ is set, single newlines will
191 191 # be converted to HTML break tags. This is the
192 192 # default behavior for traditional RedCloth.
193 193 #
194 194 attr_accessor :hard_breaks
195 195
196 196 # Accessor for toggling lite mode.
197 197 #
198 198 # In lite mode, block-level rules are ignored. This means
199 199 # that tables, paragraphs, lists, and such aren't available.
200 200 # Only the inline markup for bold, italics, entities and so on.
201 201 #
202 202 # r = RedCloth.new( "And then? She *fell*!", [:lite_mode] )
203 203 # r.to_html
204 204 # #=> "And then? She <strong>fell</strong>!"
205 205 #
206 206 attr_accessor :lite_mode
207 207
208 208 #
209 209 # Accessor for toggling span caps.
210 210 #
211 211 # Textile places `span' tags around capitalized
212 212 # words by default, but this wreaks havoc on Wikis.
213 213 # If +:no_span_caps+ is set, this will be
214 214 # suppressed.
215 215 #
216 216 attr_accessor :no_span_caps
217 217
218 218 #
219 219 # Establishes the markup predence. Available rules include:
220 220 #
221 221 # == Textile Rules
222 222 #
223 223 # The following textile rules can be set individually. Or add the complete
224 224 # set of rules with the single :textile rule, which supplies the rule set in
225 225 # the following precedence:
226 226 #
227 227 # refs_textile:: Textile references (i.e. [hobix]http://hobix.com/)
228 228 # block_textile_table:: Textile table block structures
229 229 # block_textile_lists:: Textile list structures
230 230 # block_textile_prefix:: Textile blocks with prefixes (i.e. bq., h2., etc.)
231 231 # inline_textile_image:: Textile inline images
232 232 # inline_textile_link:: Textile inline links
233 233 # inline_textile_span:: Textile inline spans
234 234 # glyphs_textile:: Textile entities (such as em-dashes and smart quotes)
235 235 #
236 236 # == Markdown
237 237 #
238 238 # refs_markdown:: Markdown references (for example: [hobix]: http://hobix.com/)
239 239 # block_markdown_setext:: Markdown setext headers
240 240 # block_markdown_atx:: Markdown atx headers
241 241 # block_markdown_rule:: Markdown horizontal rules
242 242 # block_markdown_bq:: Markdown blockquotes
243 243 # block_markdown_lists:: Markdown lists
244 244 # inline_markdown_link:: Markdown links
245 245 attr_accessor :rules
246 246
247 247 # Returns a new RedCloth object, based on _string_ and
248 248 # enforcing all the included _restrictions_.
249 249 #
250 250 # r = RedCloth.new( "h1. A <b>bold</b> man", [:filter_html] )
251 251 # r.to_html
252 252 # #=>"<h1>A &lt;b&gt;bold&lt;/b&gt; man</h1>"
253 253 #
254 254 def initialize( string, restrictions = [] )
255 255 restrictions.each { |r| method( "#{ r }=" ).call( true ) }
256 256 super( string )
257 257 end
258 258
259 259 #
260 260 # Generates HTML from the Textile contents.
261 261 #
262 262 # r = RedCloth.new( "And then? She *fell*!" )
263 263 # r.to_html( true )
264 264 # #=>"And then? She <strong>fell</strong>!"
265 265 #
266 266 def to_html( *rules )
267 267 rules = DEFAULT_RULES if rules.empty?
268 268 # make our working copy
269 269 text = self.dup
270 270
271 271 @urlrefs = {}
272 272 @shelf = []
273 273 textile_rules = [:block_textile_table, :block_textile_lists,
274 274 :block_textile_prefix, :inline_textile_image, :inline_textile_link,
275 275 :inline_textile_code, :inline_textile_span, :glyphs_textile]
276 276 markdown_rules = [:refs_markdown, :block_markdown_setext, :block_markdown_atx, :block_markdown_rule,
277 277 :block_markdown_bq, :block_markdown_lists,
278 278 :inline_markdown_reflink, :inline_markdown_link]
279 279 @rules = rules.collect do |rule|
280 280 case rule
281 281 when :markdown
282 282 markdown_rules
283 283 when :textile
284 284 textile_rules
285 285 else
286 286 rule
287 287 end
288 288 end.flatten
289 289
290 290 # standard clean up
291 291 incoming_entities text
292 292 clean_white_space text
293 293
294 294 # start processor
295 295 @pre_list = []
296 296 rip_offtags text
297 297 no_textile text
298 298 escape_html_tags text
299 299 hard_break text
300 300 unless @lite_mode
301 301 refs text
302 302 # need to do this before text is split by #blocks
303 303 block_textile_quotes text
304 304 blocks text
305 305 end
306 306 inline text
307 307 smooth_offtags text
308 308
309 309 retrieve text
310 310
311 311 text.gsub!( /<\/?notextile>/, '' )
312 312 text.gsub!( /x%x%/, '&#38;' )
313 313 clean_html text if filter_html
314 314 text.strip!
315 315 text
316 316
317 317 end
318 318
319 319 #######
320 320 private
321 321 #######
322 322 #
323 323 # Mapping of 8-bit ASCII codes to HTML numerical entity equivalents.
324 324 # (from PyTextile)
325 325 #
326 326 TEXTILE_TAGS =
327 327
328 328 [[128, 8364], [129, 0], [130, 8218], [131, 402], [132, 8222], [133, 8230],
329 329 [134, 8224], [135, 8225], [136, 710], [137, 8240], [138, 352], [139, 8249],
330 330 [140, 338], [141, 0], [142, 0], [143, 0], [144, 0], [145, 8216], [146, 8217],
331 331 [147, 8220], [148, 8221], [149, 8226], [150, 8211], [151, 8212], [152, 732],
332 332 [153, 8482], [154, 353], [155, 8250], [156, 339], [157, 0], [158, 0], [159, 376]].
333 333
334 334 collect! do |a, b|
335 335 [a.chr, ( b.zero? and "" or "&#{ b };" )]
336 336 end
337 337
338 338 #
339 339 # Regular expressions to convert to HTML.
340 340 #
341 341 A_HLGN = /(?:(?:<>|<|>|\=|[()]+)+)/
342 342 A_VLGN = /[\-^~]/
343 343 C_CLAS = '(?:\([^)]+\))'
344 344 C_LNGE = '(?:\[[^\[\]]+\])'
345 345 C_STYL = '(?:\{[^}]+\})'
346 346 S_CSPN = '(?:\\\\\d+)'
347 347 S_RSPN = '(?:/\d+)'
348 348 A = "(?:#{A_HLGN}?#{A_VLGN}?|#{A_VLGN}?#{A_HLGN}?)"
349 349 S = "(?:#{S_CSPN}?#{S_RSPN}|#{S_RSPN}?#{S_CSPN}?)"
350 350 C = "(?:#{C_CLAS}?#{C_STYL}?#{C_LNGE}?|#{C_STYL}?#{C_LNGE}?#{C_CLAS}?|#{C_LNGE}?#{C_STYL}?#{C_CLAS}?)"
351 351 # PUNCT = Regexp::quote( '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~' )
352 352 PUNCT = Regexp::quote( '!"#$%&\'*+,-./:;=?@\\^_`|~' )
353 353 PUNCT_NOQ = Regexp::quote( '!"#$&\',./:;=?@\\`|' )
354 354 PUNCT_Q = Regexp::quote( '*-_+^~%' )
355 355 HYPERLINK = '(\S+?)([^\w\s/;=\?]*?)(?=\s|<|$)'
356 356
357 357 # Text markup tags, don't conflict with block tags
358 358 SIMPLE_HTML_TAGS = [
359 359 'tt', 'b', 'i', 'big', 'small', 'em', 'strong', 'dfn', 'code',
360 360 'samp', 'kbd', 'var', 'cite', 'abbr', 'acronym', 'a', 'img', 'br',
361 361 'br', 'map', 'q', 'sub', 'sup', 'span', 'bdo'
362 362 ]
363 363
364 364 QTAGS = [
365 365 ['**', 'b', :limit],
366 366 ['*', 'strong', :limit],
367 367 ['??', 'cite', :limit],
368 368 ['-', 'del', :limit],
369 369 ['__', 'i', :limit],
370 370 ['_', 'em', :limit],
371 371 ['%', 'span', :limit],
372 372 ['+', 'ins', :limit],
373 373 ['^', 'sup', :limit],
374 374 ['~', 'sub', :limit]
375 375 ]
376 QTAGS_JOIN = QTAGS.map {|rc, ht, rtype| Regexp::quote rc}.join('|')
377
376 378 QTAGS.collect! do |rc, ht, rtype|
377 379 rcq = Regexp::quote rc
378 380 re =
379 381 case rtype
380 382 when :limit
381 383 /(^|[>\s\(])
384 (#{QTAGS_JOIN}|)
382 385 (#{rcq})
383 386 (#{C})
384 387 (?::(\S+?))?
385 (\w|[^\s\-].*?[^\s\-])
388 (\w|[^\s].*?[^\s])
386 389 #{rcq}
390 (#{QTAGS_JOIN}|)
387 391 (?=[[:punct:]]|\s|\)|$)/x
388 392 else
389 393 /(#{rcq})
390 394 (#{C})
391 395 (?::(\S+))?
392 396 (\w|[^\s\-].*?[^\s\-])
393 397 #{rcq}/xm
394 398 end
395 399 [rc, ht, re, rtype]
396 400 end
397 401
398 402 # Elements to handle
399 403 GLYPHS = [
400 404 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1&#8217;\2' ], # single closing
401 405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
402 406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
403 407 # [ /\'/, '&#8216;' ], # single opening
404 408 # [ /</, '&lt;' ], # less-than
405 409 # [ />/, '&gt;' ], # greater-than
406 410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
407 411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
408 412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
409 413 # [ /"/, '&#8220;' ], # double opening
410 414 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
411 415 # [ /\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/, '<acronym title="\2">\1</acronym>' ], # 3+ uppercase acronym
412 416 # [ /(^|[^"][>\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
413 417 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
414 418 # [ /\s->\s/, ' &rarr; ' ], # right arrow
415 419 # [ /\s-\s/, ' &#8211; ' ], # en dash
416 420 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
417 421 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
418 422 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
419 423 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
420 424 ]
421 425
422 426 H_ALGN_VALS = {
423 427 '<' => 'left',
424 428 '=' => 'center',
425 429 '>' => 'right',
426 430 '<>' => 'justify'
427 431 }
428 432
429 433 V_ALGN_VALS = {
430 434 '^' => 'top',
431 435 '-' => 'middle',
432 436 '~' => 'bottom'
433 437 }
434 438
435 439 #
436 440 # Flexible HTML escaping
437 441 #
438 442 def htmlesc( str, mode=:Quotes )
439 443 if str
440 444 str.gsub!( '&', '&amp;' )
441 445 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
442 446 str.gsub!( "'", '&#039;' ) if mode == :Quotes
443 447 str.gsub!( '<', '&lt;')
444 448 str.gsub!( '>', '&gt;')
445 449 end
446 450 str
447 451 end
448 452
449 453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
450 454 def pgl( text )
451 455 #GLYPHS.each do |re, resub, tog|
452 456 # next if tog and method( tog ).call
453 457 # text.gsub! re, resub
454 458 #end
455 459 text.gsub!(/\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/) do |m|
456 460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
457 461 end
458 462 end
459 463
460 464 # Parses Textile attribute lists and builds an HTML attribute string
461 465 def pba( text_in, element = "" )
462 466
463 467 return '' unless text_in
464 468
465 469 style = []
466 470 text = text_in.dup
467 471 if element == 'td'
468 472 colspan = $1 if text =~ /\\(\d+)/
469 473 rowspan = $1 if text =~ /\/(\d+)/
470 474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
471 475 end
472 476
473 477 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
474 478
475 479 lang = $1 if
476 480 text.sub!( /\[([^)]+?)\]/, '' )
477 481
478 482 cls = $1 if
479 483 text.sub!( /\(([^()]+?)\)/, '' )
480 484
481 485 style << "padding-left:#{ $1.length }em;" if
482 486 text.sub!( /([(]+)/, '' )
483 487
484 488 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
485 489
486 490 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
487 491
488 492 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
489 493
490 494 atts = ''
491 495 atts << " style=\"#{ style.join }\"" unless style.empty?
492 496 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
493 497 atts << " lang=\"#{ lang }\"" if lang
494 498 atts << " id=\"#{ id }\"" if id
495 499 atts << " colspan=\"#{ colspan }\"" if colspan
496 500 atts << " rowspan=\"#{ rowspan }\"" if rowspan
497 501
498 502 atts
499 503 end
500 504
501 505 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
502 506
503 507 # Parses a Textile table block, building HTML from the result.
504 508 def block_textile_table( text )
505 509 text.gsub!( TABLE_RE ) do |matches|
506 510
507 511 tatts, fullrow = $~[1..2]
508 512 tatts = pba( tatts, 'table' )
509 513 tatts = shelve( tatts ) if tatts
510 514 rows = []
511 515
512 516 fullrow.each_line do |row|
513 517 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
514 518 cells = []
515 519 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
516 520 next if cell == '|'
517 521 ctyp = 'd'
518 522 ctyp = 'h' if cell =~ /^_/
519 523
520 524 catts = ''
521 525 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
522 526
523 527 catts = shelve( catts ) if catts
524 528 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
525 529 end
526 530 ratts = shelve( ratts ) if ratts
527 531 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
528 532 end
529 533 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
530 534 end
531 535 end
532 536
533 537 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
534 538 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
535 539
536 540 # Parses Textile lists and generates HTML
537 541 def block_textile_lists( text )
538 542 text.gsub!( LISTS_RE ) do |match|
539 543 lines = match.split( /\n/ )
540 544 last_line = -1
541 545 depth = []
542 546 lines.each_with_index do |line, line_id|
543 547 if line =~ LISTS_CONTENT_RE
544 548 tl,atts,content = $~[1..3]
545 549 if depth.last
546 550 if depth.last.length > tl.length
547 551 (depth.length - 1).downto(0) do |i|
548 552 break if depth[i].length == tl.length
549 553 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
550 554 depth.pop
551 555 end
552 556 end
553 557 if depth.last and depth.last.length == tl.length
554 558 lines[line_id - 1] << '</li>'
555 559 end
556 560 end
557 561 unless depth.last == tl
558 562 depth << tl
559 563 atts = pba( atts )
560 564 atts = shelve( atts ) if atts
561 565 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
562 566 else
563 567 lines[line_id] = "\t\t<li>#{ content }"
564 568 end
565 569 last_line = line_id
566 570
567 571 else
568 572 last_line = line_id
569 573 end
570 574 if line_id - last_line > 1 or line_id == lines.length - 1
571 575 depth.delete_if do |v|
572 576 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
573 577 end
574 578 end
575 579 end
576 580 lines.join( "\n" )
577 581 end
578 582 end
579 583
580 584 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
581 585 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
582 586
583 587 def block_textile_quotes( text )
584 588 text.gsub!( QUOTES_RE ) do |match|
585 589 lines = match.split( /\n/ )
586 590 quotes = ''
587 591 indent = 0
588 592 lines.each do |line|
589 593 line =~ QUOTES_CONTENT_RE
590 594 bq,content = $1, $2
591 595 l = bq.count('>')
592 596 if l != indent
593 597 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
594 598 indent = l
595 599 end
596 600 quotes << (content + "\n")
597 601 end
598 602 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
599 603 quotes
600 604 end
601 605 end
602 606
603 607 CODE_RE = /(\W)
604 608 @
605 609 (?:\|(\w+?)\|)?
606 610 (.+?)
607 611 @
608 612 (?=\W)/x
609 613
610 614 def inline_textile_code( text )
611 615 text.gsub!( CODE_RE ) do |m|
612 616 before,lang,code,after = $~[1..4]
613 617 lang = " lang=\"#{ lang }\"" if lang
614 618 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }" )
615 619 end
616 620 end
617 621
618 622 def lT( text )
619 623 text =~ /\#$/ ? 'o' : 'u'
620 624 end
621 625
622 626 def hard_break( text )
623 627 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
624 628 end
625 629
626 630 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
627 631
628 632 def blocks( text, deep_code = false )
629 633 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
630 634 plain = blk !~ /\A[#*> ]/
631 635
632 636 # skip blocks that are complex HTML
633 637 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
634 638 blk
635 639 else
636 640 # search for indentation levels
637 641 blk.strip!
638 642 if blk.empty?
639 643 blk
640 644 else
641 645 code_blk = nil
642 646 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
643 647 flush_left iblk
644 648 blocks iblk, plain
645 649 iblk.gsub( /^(\S)/, "\t\\1" )
646 650 if plain
647 651 code_blk = iblk; ""
648 652 else
649 653 iblk
650 654 end
651 655 end
652 656
653 657 block_applied = 0
654 658 @rules.each do |rule_name|
655 659 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
656 660 end
657 661 if block_applied.zero?
658 662 if deep_code
659 663 blk = "\t<pre><code>#{ blk }</code></pre>"
660 664 else
661 665 blk = "\t<p>#{ blk }</p>"
662 666 end
663 667 end
664 668 # hard_break blk
665 669 blk + "\n#{ code_blk }"
666 670 end
667 671 end
668 672
669 673 end.join( "\n\n" ) )
670 674 end
671 675
672 676 def textile_bq( tag, atts, cite, content )
673 677 cite, cite_title = check_refs( cite )
674 678 cite = " cite=\"#{ cite }\"" if cite
675 679 atts = shelve( atts ) if atts
676 680 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
677 681 end
678 682
679 683 def textile_p( tag, atts, cite, content )
680 684 atts = shelve( atts ) if atts
681 685 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
682 686 end
683 687
684 688 alias textile_h1 textile_p
685 689 alias textile_h2 textile_p
686 690 alias textile_h3 textile_p
687 691 alias textile_h4 textile_p
688 692 alias textile_h5 textile_p
689 693 alias textile_h6 textile_p
690 694
691 695 def textile_fn_( tag, num, atts, cite, content )
692 696 atts << " id=\"fn#{ num }\" class=\"footnote\""
693 697 content = "<sup>#{ num }</sup> #{ content }"
694 698 atts = shelve( atts ) if atts
695 699 "\t<p#{ atts }>#{ content }</p>"
696 700 end
697 701
698 702 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
699 703
700 704 def block_textile_prefix( text )
701 705 if text =~ BLOCK_RE
702 706 tag,tagpre,num,atts,cite,content = $~[1..6]
703 707 atts = pba( atts )
704 708
705 709 # pass to prefix handler
706 710 if respond_to? "textile_#{ tag }", true
707 711 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
708 712 elsif respond_to? "textile_#{ tagpre }_", true
709 713 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
710 714 end
711 715 end
712 716 end
713 717
714 718 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
715 719 def block_markdown_setext( text )
716 720 if text =~ SETEXT_RE
717 721 tag = if $2 == "="; "h1"; else; "h2"; end
718 722 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
719 723 blocks cont
720 724 text.replace( blk + cont )
721 725 end
722 726 end
723 727
724 728 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
725 729 [ ]*
726 730 (.+?) # $2 = Header text
727 731 [ ]*
728 732 \#* # optional closing #'s (not counted)
729 733 $/x
730 734 def block_markdown_atx( text )
731 735 if text =~ ATX_RE
732 736 tag = "h#{ $1.length }"
733 737 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
734 738 blocks cont
735 739 text.replace( blk + cont )
736 740 end
737 741 end
738 742
739 743 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
740 744
741 745 def block_markdown_bq( text )
742 746 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
743 747 blk.gsub!( /^ *> ?/, '' )
744 748 flush_left blk
745 749 blocks blk
746 750 blk.gsub!( /^(\S)/, "\t\\1" )
747 751 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
748 752 end
749 753 end
750 754
751 755 MARKDOWN_RULE_RE = /^(#{
752 756 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
753 757 })$/
754 758
755 759 def block_markdown_rule( text )
756 760 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
757 761 "<hr />"
758 762 end
759 763 end
760 764
761 765 # XXX TODO XXX
762 766 def block_markdown_lists( text )
763 767 end
764 768
765 769 def inline_textile_span( text )
766 770 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
767 771 text.gsub!( qtag_re ) do |m|
768 772
769 773 case rtype
770 774 when :limit
771 sta,qtag,atts,cite,content = $~[1..5]
775 sta,oqs,qtag,atts,cite,content,oqa = $~[1..7]
772 776 else
773 777 qtag,atts,cite,content = $~[1..4]
774 778 sta = ''
775 779 end
776 780 atts = pba( atts )
777 781 atts << " cite=\"#{ cite }\"" if cite
778 782 atts = shelve( atts ) if atts
779 783
780 "#{ sta }<#{ ht }#{ atts }>#{ content }</#{ ht }>"
784 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
781 785
782 786 end
783 787 end
784 788 end
785 789
786 790 LINK_RE = /
787 791 (
788 792 ([\s\[{(]|[#{PUNCT}])? # $pre
789 793 " # start
790 794 (#{C}) # $atts
791 795 ([^"\n]+?) # $text
792 796 \s?
793 797 (?:\(([^)]+?)\)(?="))? # $title
794 798 ":
795 799 ( # $url
796 800 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
797 801 [\w\/]\S+?
798 802 )
799 803 (\/)? # $slash
800 804 ([^\w\=\/;\(\)]*?) # $post
801 805 )
802 806 (?=<|\s|$)
803 807 /x
804 808 #"
805 809 def inline_textile_link( text )
806 810 text.gsub!( LINK_RE ) do |m|
807 811 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
808 812 if text.include?('<br />')
809 813 all
810 814 else
811 815 url, url_title = check_refs( url )
812 816 title ||= url_title
813 817
814 818 # Idea below : an URL with unbalanced parethesis and
815 819 # ending by ')' is put into external parenthesis
816 820 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
817 821 url=url[0..-2] # discard closing parenth from url
818 822 post = ")"+post # add closing parenth to post
819 823 end
820 824 atts = pba( atts )
821 825 atts = " href=\"#{ url }#{ slash }\"#{ atts }"
822 826 atts << " title=\"#{ htmlesc title }\"" if title
823 827 atts = shelve( atts ) if atts
824 828
825 829 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
826 830
827 831 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
828 832 end
829 833 end
830 834 end
831 835
832 836 MARKDOWN_REFLINK_RE = /
833 837 \[([^\[\]]+)\] # $text
834 838 [ ]? # opt. space
835 839 (?:\n[ ]*)? # one optional newline followed by spaces
836 840 \[(.*?)\] # $id
837 841 /x
838 842
839 843 def inline_markdown_reflink( text )
840 844 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
841 845 text, id = $~[1..2]
842 846
843 847 if id.empty?
844 848 url, title = check_refs( text )
845 849 else
846 850 url, title = check_refs( id )
847 851 end
848 852
849 853 atts = " href=\"#{ url }\""
850 854 atts << " title=\"#{ title }\"" if title
851 855 atts = shelve( atts )
852 856
853 857 "<a#{ atts }>#{ text }</a>"
854 858 end
855 859 end
856 860
857 861 MARKDOWN_LINK_RE = /
858 862 \[([^\[\]]+)\] # $text
859 863 \( # open paren
860 864 [ \t]* # opt space
861 865 <?(.+?)>? # $href
862 866 [ \t]* # opt space
863 867 (?: # whole title
864 868 (['"]) # $quote
865 869 (.*?) # $title
866 870 \3 # matching quote
867 871 )? # title is optional
868 872 \)
869 873 /x
870 874
871 875 def inline_markdown_link( text )
872 876 text.gsub!( MARKDOWN_LINK_RE ) do |m|
873 877 text, url, quote, title = $~[1..4]
874 878
875 879 atts = " href=\"#{ url }\""
876 880 atts << " title=\"#{ title }\"" if title
877 881 atts = shelve( atts )
878 882
879 883 "<a#{ atts }>#{ text }</a>"
880 884 end
881 885 end
882 886
883 887 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
884 888 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
885 889
886 890 def refs( text )
887 891 @rules.each do |rule_name|
888 892 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
889 893 end
890 894 end
891 895
892 896 def refs_textile( text )
893 897 text.gsub!( TEXTILE_REFS_RE ) do |m|
894 898 flag, url = $~[2..3]
895 899 @urlrefs[flag.downcase] = [url, nil]
896 900 nil
897 901 end
898 902 end
899 903
900 904 def refs_markdown( text )
901 905 text.gsub!( MARKDOWN_REFS_RE ) do |m|
902 906 flag, url = $~[2..3]
903 907 title = $~[6]
904 908 @urlrefs[flag.downcase] = [url, title]
905 909 nil
906 910 end
907 911 end
908 912
909 913 def check_refs( text )
910 914 ret = @urlrefs[text.downcase] if text
911 915 ret || [text, nil]
912 916 end
913 917
914 918 IMAGE_RE = /
915 919 (>|\s|^) # start of line?
916 920 \! # opening
917 921 (\<|\=|\>)? # optional alignment atts
918 922 (#{C}) # optional style,class atts
919 923 (?:\. )? # optional dot-space
920 924 ([^\s(!]+?) # presume this is the src
921 925 \s? # optional space
922 926 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
923 927 \! # closing
924 928 (?::#{ HYPERLINK })? # optional href
925 929 /x
926 930
927 931 def inline_textile_image( text )
928 932 text.gsub!( IMAGE_RE ) do |m|
929 933 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
930 934 htmlesc title
931 935 atts = pba( atts )
932 936 atts = " src=\"#{ url }\"#{ atts }"
933 937 atts << " title=\"#{ title }\"" if title
934 938 atts << " alt=\"#{ title }\""
935 939 # size = @getimagesize($url);
936 940 # if($size) $atts.= " $size[3]";
937 941
938 942 href, alt_title = check_refs( href ) if href
939 943 url, url_title = check_refs( url )
940 944
941 945 out = ''
942 946 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
943 947 out << "<img#{ shelve( atts ) } />"
944 948 out << "</a>#{ href_a1 }#{ href_a2 }" if href
945 949
946 950 if algn
947 951 algn = h_align( algn )
948 952 if stln == "<p>"
949 953 out = "<p style=\"float:#{ algn }\">#{ out }"
950 954 else
951 955 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
952 956 end
953 957 else
954 958 out = stln + out
955 959 end
956 960
957 961 out
958 962 end
959 963 end
960 964
961 965 def shelve( val )
962 966 @shelf << val
963 967 " :redsh##{ @shelf.length }:"
964 968 end
965 969
966 970 def retrieve( text )
967 971 @shelf.each_with_index do |r, i|
968 972 text.gsub!( " :redsh##{ i + 1 }:", r )
969 973 end
970 974 end
971 975
972 976 def incoming_entities( text )
973 977 ## turn any incoming ampersands into a dummy character for now.
974 978 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
975 979 ## implying an incoming html entity, to be skipped
976 980
977 981 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
978 982 end
979 983
980 984 def no_textile( text )
981 985 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
982 986 '\1<notextile>\2</notextile>\3' )
983 987 text.gsub!( /^ *==([^=]+.*?)==/m,
984 988 '\1<notextile>\2</notextile>\3' )
985 989 end
986 990
987 991 def clean_white_space( text )
988 992 # normalize line breaks
989 993 text.gsub!( /\r\n/, "\n" )
990 994 text.gsub!( /\r/, "\n" )
991 995 text.gsub!( /\t/, ' ' )
992 996 text.gsub!( /^ +$/, '' )
993 997 text.gsub!( /\n{3,}/, "\n\n" )
994 998 text.gsub!( /"$/, "\" " )
995 999
996 1000 # if entire document is indented, flush
997 1001 # to the left side
998 1002 flush_left text
999 1003 end
1000 1004
1001 1005 def flush_left( text )
1002 1006 indt = 0
1003 1007 if text =~ /^ /
1004 1008 while text !~ /^ {#{indt}}\S/
1005 1009 indt += 1
1006 1010 end unless text.empty?
1007 1011 if indt.nonzero?
1008 1012 text.gsub!( /^ {#{indt}}/, '' )
1009 1013 end
1010 1014 end
1011 1015 end
1012 1016
1013 1017 def footnote_ref( text )
1014 1018 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1015 1019 '<sup><a href="#fn\1">\1</a></sup>\2' )
1016 1020 end
1017 1021
1018 1022 OFFTAGS = /(code|pre|kbd|notextile)/
1019 1023 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1020 1024 OFFTAG_OPEN = /<#{ OFFTAGS }/
1021 1025 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1022 1026 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1023 1027 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1024 1028
1025 1029 def glyphs_textile( text, level = 0 )
1026 1030 if text !~ HASTAG_MATCH
1027 1031 pgl text
1028 1032 footnote_ref text
1029 1033 else
1030 1034 codepre = 0
1031 1035 text.gsub!( ALLTAG_MATCH ) do |line|
1032 1036 ## matches are off if we're between <code>, <pre> etc.
1033 1037 if $1
1034 1038 if line =~ OFFTAG_OPEN
1035 1039 codepre += 1
1036 1040 elsif line =~ OFFTAG_CLOSE
1037 1041 codepre -= 1
1038 1042 codepre = 0 if codepre < 0
1039 1043 end
1040 1044 elsif codepre.zero?
1041 1045 glyphs_textile( line, level + 1 )
1042 1046 else
1043 1047 htmlesc( line, :NoQuotes )
1044 1048 end
1045 1049 # p [level, codepre, line]
1046 1050
1047 1051 line
1048 1052 end
1049 1053 end
1050 1054 end
1051 1055
1052 1056 def rip_offtags( text )
1053 1057 if text =~ /<.*>/
1054 1058 ## strip and encode <pre> content
1055 1059 codepre, used_offtags = 0, {}
1056 1060 text.gsub!( OFFTAG_MATCH ) do |line|
1057 1061 if $3
1058 1062 offtag, aftertag = $4, $5
1059 1063 codepre += 1
1060 1064 used_offtags[offtag] = true
1061 1065 if codepre - used_offtags.length > 0
1062 1066 htmlesc( line, :NoQuotes )
1063 1067 @pre_list.last << line
1064 1068 line = ""
1065 1069 else
1066 1070 htmlesc( aftertag, :NoQuotes ) if aftertag
1067 1071 line = "<redpre##{ @pre_list.length }>"
1068 1072 $3.match(/<#{ OFFTAGS }([^>]*)>/)
1069 1073 tag = $1
1070 1074 $2.to_s.match(/(class\=\S+)/i)
1071 1075 tag << " #{$1}" if $1
1072 1076 @pre_list << "<#{ tag }>#{ aftertag }"
1073 1077 end
1074 1078 elsif $1 and codepre > 0
1075 1079 if codepre - used_offtags.length > 0
1076 1080 htmlesc( line, :NoQuotes )
1077 1081 @pre_list.last << line
1078 1082 line = ""
1079 1083 end
1080 1084 codepre -= 1 unless codepre.zero?
1081 1085 used_offtags = {} if codepre.zero?
1082 1086 end
1083 1087 line
1084 1088 end
1085 1089 end
1086 1090 text
1087 1091 end
1088 1092
1089 1093 def smooth_offtags( text )
1090 1094 unless @pre_list.empty?
1091 1095 ## replace <pre> content
1092 1096 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1093 1097 end
1094 1098 end
1095 1099
1096 1100 def inline( text )
1097 1101 [/^inline_/, /^glyphs_/].each do |meth_re|
1098 1102 @rules.each do |rule_name|
1099 1103 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1100 1104 end
1101 1105 end
1102 1106 end
1103 1107
1104 1108 def h_align( text )
1105 1109 H_ALGN_VALS[text]
1106 1110 end
1107 1111
1108 1112 def v_align( text )
1109 1113 V_ALGN_VALS[text]
1110 1114 end
1111 1115
1112 1116 def textile_popup_help( name, windowW, windowH )
1113 1117 ' <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 />'
1114 1118 end
1115 1119
1116 1120 # HTML cleansing stuff
1117 1121 BASIC_TAGS = {
1118 1122 'a' => ['href', 'title'],
1119 1123 'img' => ['src', 'alt', 'title'],
1120 1124 'br' => [],
1121 1125 'i' => nil,
1122 1126 'u' => nil,
1123 1127 'b' => nil,
1124 1128 'pre' => nil,
1125 1129 'kbd' => nil,
1126 1130 'code' => ['lang'],
1127 1131 'cite' => nil,
1128 1132 'strong' => nil,
1129 1133 'em' => nil,
1130 1134 'ins' => nil,
1131 1135 'sup' => nil,
1132 1136 'sub' => nil,
1133 1137 'del' => nil,
1134 1138 'table' => nil,
1135 1139 'tr' => nil,
1136 1140 'td' => ['colspan', 'rowspan'],
1137 1141 'th' => nil,
1138 1142 'ol' => nil,
1139 1143 'ul' => nil,
1140 1144 'li' => nil,
1141 1145 'p' => nil,
1142 1146 'h1' => nil,
1143 1147 'h2' => nil,
1144 1148 'h3' => nil,
1145 1149 'h4' => nil,
1146 1150 'h5' => nil,
1147 1151 'h6' => nil,
1148 1152 'blockquote' => ['cite']
1149 1153 }
1150 1154
1151 1155 def clean_html( text, tags = BASIC_TAGS )
1152 1156 text.gsub!( /<!\[CDATA\[/, '' )
1153 1157 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1154 1158 raw = $~
1155 1159 tag = raw[2].downcase
1156 1160 if tags.has_key? tag
1157 1161 pcs = [tag]
1158 1162 tags[tag].each do |prop|
1159 1163 ['"', "'", ''].each do |q|
1160 1164 q2 = ( q != '' ? q : '\s' )
1161 1165 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1162 1166 attrv = $1
1163 1167 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1164 1168 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1165 1169 break
1166 1170 end
1167 1171 end
1168 1172 end if tags[tag]
1169 1173 "<#{raw[1]}#{pcs.join " "}>"
1170 1174 else
1171 1175 " "
1172 1176 end
1173 1177 end
1174 1178 end
1175 1179
1176 1180 ALLOWED_TAGS = %w(redpre pre code notextile)
1177 1181
1178 1182 def escape_html_tags(text)
1179 1183 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1180 1184 end
1181 1185 end
1182 1186
General Comments 0
You need to be logged in to leave comments. Login now