To check out this repository please hg clone the following URL, or open the URL using EasyMercurial or your preferred Mercurial client.
root / lib / redcloth3.rb @ 441:cbce1fd3b1b7
History | View | Annotate | Download (37.9 KB)
| 1 | 0:513646585e45 | Chris | # 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 | 441:cbce1fd3b1b7 | Chris | # need to do this before #hard_break and #blocks
|
| 300 | block_textile_quotes text unless @lite_mode |
||
| 301 | 0:513646585e45 | Chris | 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 | 37:94944d00e43c | chris | text.gsub!(/\b([A-Z][A-Z0-9]{1,})\b(?:[(]([^)]*)[)])/) do |m| |
| 460 | 0:513646585e45 | Chris | "<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 | style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles |
||
| 478 | |||
| 479 | lang = $1 if |
||
| 480 | text.sub!( /\[([^)]+?)\]/, '' ) |
||
| 481 | |||
| 482 | cls = $1 if |
||
| 483 | text.sub!( /\(([^()]+?)\)/, '' ) |
||
| 484 | |||
| 485 | style << "padding-left:#{ $1.length }em;" if |
||
| 486 | text.sub!( /([(]+)/, '' ) |
||
| 487 | |||
| 488 | style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' ) |
||
| 489 | |||
| 490 | style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN |
||
| 491 | |||
| 492 | cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/ |
||
| 493 | |||
| 494 | atts = ''
|
||
| 495 | atts << " style=\"#{ style.join }\"" unless style.empty? |
||
| 496 | atts << " class=\"#{ cls }\"" unless cls.to_s.empty? |
||
| 497 | atts << " lang=\"#{ lang }\"" if lang |
||
| 498 | atts << " id=\"#{ id }\"" if id |
||
| 499 | atts << " colspan=\"#{ colspan }\"" if colspan |
||
| 500 | atts << " rowspan=\"#{ rowspan }\"" if rowspan |
||
| 501 | |||
| 502 | atts |
||
| 503 | end
|
||
| 504 | |||
| 505 | TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m |
||
| 506 | |||
| 507 | # Parses a Textile table block, building HTML from the result.
|
||
| 508 | def block_textile_table( text ) |
||
| 509 | text.gsub!( TABLE_RE ) do |matches| |
||
| 510 | |||
| 511 | tatts, fullrow = $~[1..2] |
||
| 512 | tatts = pba( tatts, 'table' )
|
||
| 513 | tatts = shelve( tatts ) if tatts
|
||
| 514 | rows = [] |
||
| 515 | |||
| 516 | fullrow.each_line do |row|
|
||
| 517 | ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m |
||
| 518 | cells = [] |
||
| 519 | row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell| |
||
| 520 | next if cell == '|' |
||
| 521 | ctyp = 'd'
|
||
| 522 | ctyp = 'h' if cell =~ /^_/ |
||
| 523 | |||
| 524 | catts = ''
|
||
| 525 | catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/ |
||
| 526 | |||
| 527 | catts = shelve( catts ) if catts
|
||
| 528 | cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
|
||
| 529 | end
|
||
| 530 | ratts = shelve( ratts ) if ratts
|
||
| 531 | rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
|
||
| 532 | end
|
||
| 533 | "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
|
||
| 534 | end
|
||
| 535 | end
|
||
| 536 | |||
| 537 | LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m |
||
| 538 | LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m |
||
| 539 | |||
| 540 | # Parses Textile lists and generates HTML
|
||
| 541 | def block_textile_lists( text ) |
||
| 542 | text.gsub!( LISTS_RE ) do |match| |
||
| 543 | lines = match.split( /\n/ )
|
||
| 544 | last_line = -1
|
||
| 545 | depth = [] |
||
| 546 | lines.each_with_index do |line, line_id|
|
||
| 547 | if line =~ LISTS_CONTENT_RE |
||
| 548 | tl,atts,content = $~[1..3] |
||
| 549 | if depth.last
|
||
| 550 | if depth.last.length > tl.length
|
||
| 551 | (depth.length - 1).downto(0) do |i| |
||
| 552 | break if depth[i].length == tl.length |
||
| 553 | lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t" |
||
| 554 | depth.pop |
||
| 555 | end
|
||
| 556 | end
|
||
| 557 | if depth.last and depth.last.length == tl.length |
||
| 558 | lines[line_id - 1] << '</li>' |
||
| 559 | end
|
||
| 560 | end
|
||
| 561 | unless depth.last == tl
|
||
| 562 | depth << tl |
||
| 563 | atts = pba( atts ) |
||
| 564 | atts = shelve( atts ) if atts
|
||
| 565 | lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
|
||
| 566 | else
|
||
| 567 | lines[line_id] = "\t\t<li>#{ content }"
|
||
| 568 | end
|
||
| 569 | last_line = line_id |
||
| 570 | |||
| 571 | else
|
||
| 572 | last_line = line_id |
||
| 573 | end
|
||
| 574 | if line_id - last_line > 1 or line_id == lines.length - 1 |
||
| 575 | depth.delete_if do |v|
|
||
| 576 | lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
|
||
| 577 | end
|
||
| 578 | end
|
||
| 579 | end
|
||
| 580 | lines.join( "\n" )
|
||
| 581 | end
|
||
| 582 | end
|
||
| 583 | |||
| 584 | QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m |
||
| 585 | QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m |
||
| 586 | |||
| 587 | def block_textile_quotes( text ) |
||
| 588 | text.gsub!( QUOTES_RE ) do |match| |
||
| 589 | lines = match.split( /\n/ )
|
||
| 590 | quotes = ''
|
||
| 591 | indent = 0
|
||
| 592 | lines.each do |line|
|
||
| 593 | line =~ QUOTES_CONTENT_RE
|
||
| 594 | bq,content = $1, $2 |
||
| 595 | l = bq.count('>')
|
||
| 596 | if l != indent
|
||
| 597 | quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n") |
||
| 598 | indent = l |
||
| 599 | end
|
||
| 600 | quotes << (content + "\n")
|
||
| 601 | end
|
||
| 602 | quotes << ("\n" + '</blockquote>' * indent + "\n\n") |
||
| 603 | quotes |
||
| 604 | end
|
||
| 605 | end
|
||
| 606 | |||
| 607 | CODE_RE = /(\W) |
||
| 608 | @
|
||
| 609 | (?:\|(\w+?)\|)?
|
||
| 610 | (.+?)
|
||
| 611 | @
|
||
| 612 | (?=\W)/x
|
||
| 613 | |||
| 614 | def inline_textile_code( text ) |
||
| 615 | text.gsub!( CODE_RE ) do |m| |
||
| 616 | before,lang,code,after = $~[1..4] |
||
| 617 | lang = " lang=\"#{ lang }\"" if lang |
||
| 618 | rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false ) |
||
| 619 | end
|
||
| 620 | end
|
||
| 621 | |||
| 622 | def lT( text ) |
||
| 623 | text =~ /\#$/ ? 'o' : 'u' |
||
| 624 | end
|
||
| 625 | |||
| 626 | def hard_break( text ) |
||
| 627 | text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks |
||
| 628 | end
|
||
| 629 | |||
| 630 | BLOCKS_GROUP_RE = /\n{2,}(?! )/m |
||
| 631 | |||
| 632 | def blocks( text, deep_code = false ) |
||
| 633 | text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk| |
||
| 634 | plain = blk !~ /\A[#*> ]/
|
||
| 635 | |||
| 636 | # skip blocks that are complex HTML
|
||
| 637 | if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1 |
||
| 638 | blk |
||
| 639 | else
|
||
| 640 | # search for indentation levels
|
||
| 641 | blk.strip! |
||
| 642 | if blk.empty?
|
||
| 643 | blk |
||
| 644 | else
|
||
| 645 | code_blk = nil
|
||
| 646 | blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk| |
||
| 647 | flush_left iblk |
||
| 648 | blocks iblk, plain |
||
| 649 | iblk.gsub( /^(\S)/, "\t\\1" ) |
||
| 650 | if plain
|
||
| 651 | code_blk = iblk; ""
|
||
| 652 | else
|
||
| 653 | iblk |
||
| 654 | end
|
||
| 655 | end
|
||
| 656 | |||
| 657 | block_applied = 0
|
||
| 658 | @rules.each do |rule_name| |
||
| 659 | block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) ) |
||
| 660 | end
|
||
| 661 | if block_applied.zero?
|
||
| 662 | if deep_code
|
||
| 663 | blk = "\t<pre><code>#{ blk }</code></pre>"
|
||
| 664 | else
|
||
| 665 | blk = "\t<p>#{ blk }</p>"
|
||
| 666 | end
|
||
| 667 | end
|
||
| 668 | # hard_break blk
|
||
| 669 | blk + "\n#{ code_blk }"
|
||
| 670 | end
|
||
| 671 | end
|
||
| 672 | |||
| 673 | end.join( "\n\n" ) ) |
||
| 674 | end
|
||
| 675 | |||
| 676 | def textile_bq( tag, atts, cite, content ) |
||
| 677 | cite, cite_title = check_refs( cite ) |
||
| 678 | cite = " cite=\"#{ cite }\"" if cite |
||
| 679 | atts = shelve( atts ) if atts
|
||
| 680 | "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
|
||
| 681 | end
|
||
| 682 | |||
| 683 | def textile_p( tag, atts, cite, content ) |
||
| 684 | atts = shelve( atts ) if atts
|
||
| 685 | "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
|
||
| 686 | end
|
||
| 687 | |||
| 688 | alias textile_h1 textile_p |
||
| 689 | alias textile_h2 textile_p |
||
| 690 | alias textile_h3 textile_p |
||
| 691 | alias textile_h4 textile_p |
||
| 692 | alias textile_h5 textile_p |
||
| 693 | alias textile_h6 textile_p |
||
| 694 | |||
| 695 | def textile_fn_( tag, num, atts, cite, content ) |
||
| 696 | atts << " id=\"fn#{ num }\" class=\"footnote\""
|
||
| 697 | content = "<sup>#{ num }</sup> #{ content }"
|
||
| 698 | atts = shelve( atts ) if atts
|
||
| 699 | "\t<p#{ atts }>#{ content }</p>"
|
||
| 700 | end
|
||
| 701 | |||
| 702 | BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m |
||
| 703 | |||
| 704 | def block_textile_prefix( text ) |
||
| 705 | if text =~ BLOCK_RE |
||
| 706 | tag,tagpre,num,atts,cite,content = $~[1..6] |
||
| 707 | atts = pba( atts ) |
||
| 708 | |||
| 709 | # pass to prefix handler
|
||
| 710 | 441:cbce1fd3b1b7 | Chris | replacement = nil
|
| 711 | 0:513646585e45 | Chris | if respond_to? "textile_#{ tag }", true |
| 712 | 441:cbce1fd3b1b7 | Chris | replacement = method( "textile_#{ tag }" ).call( tag, atts, cite, content )
|
| 713 | 0:513646585e45 | Chris | elsif respond_to? "textile_#{ tagpre }_", true |
| 714 | 441:cbce1fd3b1b7 | Chris | replacement = method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content )
|
| 715 | 0:513646585e45 | Chris | end
|
| 716 | 441:cbce1fd3b1b7 | Chris | text.gsub!( $& ) { replacement } if replacement |
| 717 | 0:513646585e45 | Chris | end
|
| 718 | end
|
||
| 719 | |||
| 720 | SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m |
||
| 721 | def block_markdown_setext( text ) |
||
| 722 | if text =~ SETEXT_RE |
||
| 723 | tag = if $2 == "="; "h1"; else; "h2"; end |
||
| 724 | blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $' |
||
| 725 | blocks cont |
||
| 726 | text.replace( blk + cont ) |
||
| 727 | end
|
||
| 728 | end
|
||
| 729 | |||
| 730 | ATX_RE = /\A(\#{1,6}) # $1 = string of #'s |
||
| 731 | [ ]*
|
||
| 732 | (.+?) # $2 = Header text
|
||
| 733 | [ ]*
|
||
| 734 | \#* # optional closing #'s (not counted)
|
||
| 735 | $/x
|
||
| 736 | def block_markdown_atx( text ) |
||
| 737 | if text =~ ATX_RE |
||
| 738 | tag = "h#{ $1.length }"
|
||
| 739 | blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $' |
||
| 740 | blocks cont |
||
| 741 | text.replace( blk + cont ) |
||
| 742 | end
|
||
| 743 | end
|
||
| 744 | |||
| 745 | MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m |
||
| 746 | |||
| 747 | def block_markdown_bq( text ) |
||
| 748 | text.gsub!( MARKDOWN_BQ_RE ) do |blk| |
||
| 749 | blk.gsub!( /^ *> ?/, '' ) |
||
| 750 | flush_left blk |
||
| 751 | blocks blk |
||
| 752 | blk.gsub!( /^(\S)/, "\t\\1" ) |
||
| 753 | "<blockquote>\n#{ blk }\n</blockquote>\n\n"
|
||
| 754 | end
|
||
| 755 | end
|
||
| 756 | |||
| 757 | MARKDOWN_RULE_RE = /^(#{ |
||
| 758 | ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
|
||
| 759 | })$/
|
||
| 760 | |||
| 761 | def block_markdown_rule( text ) |
||
| 762 | text.gsub!( MARKDOWN_RULE_RE ) do |blk| |
||
| 763 | "<hr />"
|
||
| 764 | end
|
||
| 765 | end
|
||
| 766 | |||
| 767 | # XXX TODO XXX
|
||
| 768 | def block_markdown_lists( text ) |
||
| 769 | end
|
||
| 770 | |||
| 771 | def inline_textile_span( text ) |
||
| 772 | QTAGS.each do |qtag_rc, ht, qtag_re, rtype| |
||
| 773 | text.gsub!( qtag_re ) do |m|
|
||
| 774 | |||
| 775 | case rtype
|
||
| 776 | when :limit |
||
| 777 | sta,oqs,qtag,content,oqa = $~[1..6] |
||
| 778 | atts = nil
|
||
| 779 | if content =~ /^(#{C})(.+)$/ |
||
| 780 | atts, content = $~[1..2] |
||
| 781 | end
|
||
| 782 | else
|
||
| 783 | qtag,atts,cite,content = $~[1..4] |
||
| 784 | sta = ''
|
||
| 785 | end
|
||
| 786 | atts = pba( atts ) |
||
| 787 | atts = shelve( atts ) if atts
|
||
| 788 | |||
| 789 | "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
|
||
| 790 | |||
| 791 | end
|
||
| 792 | end
|
||
| 793 | end
|
||
| 794 | |||
| 795 | LINK_RE = / |
||
| 796 | (
|
||
| 797 | ([\s\[{(]|[#{PUNCT}])? # $pre
|
||
| 798 | " # start
|
||
| 799 | (#{C}) # $atts
|
||
| 800 | ([^"\n]+?) # $text
|
||
| 801 | \s?
|
||
| 802 | (?:\(([^)]+?)\)(?="))? # $title
|
||
| 803 | ":
|
||
| 804 | ( # $url
|
||
| 805 | (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
|
||
| 806 | [\w\/]\S+?
|
||
| 807 | )
|
||
| 808 | (\/)? # $slash
|
||
| 809 | ([^\w\=\/;\(\)]*?) # $post
|
||
| 810 | )
|
||
| 811 | (?=<|\s|$)
|
||
| 812 | /x
|
||
| 813 | #"
|
||
| 814 | def inline_textile_link( text ) |
||
| 815 | text.gsub!( LINK_RE ) do |m| |
||
| 816 | all,pre,atts,text,title,url,proto,slash,post = $~[1..9] |
||
| 817 | if text.include?('<br />') |
||
| 818 | all |
||
| 819 | else
|
||
| 820 | url, url_title = check_refs( url ) |
||
| 821 | title ||= url_title |
||
| 822 | |||
| 823 | # Idea below : an URL with unbalanced parethesis and
|
||
| 824 | # ending by ')' is put into external parenthesis
|
||
| 825 | if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) ) |
||
| 826 | url=url[0..-2] # discard closing parenth from url |
||
| 827 | post = ")"+post # add closing parenth to post |
||
| 828 | end
|
||
| 829 | atts = pba( atts ) |
||
| 830 | atts = " href=\"#{ htmlesc url }#{ slash }\"#{ atts }"
|
||
| 831 | atts << " title=\"#{ htmlesc title }\"" if title |
||
| 832 | atts = shelve( atts ) if atts
|
||
| 833 | |||
| 834 | external = (url =~ /^https?:\/\//) ? ' class="external"' : '' |
||
| 835 | |||
| 836 | "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
|
||
| 837 | end
|
||
| 838 | end
|
||
| 839 | end
|
||
| 840 | |||
| 841 | MARKDOWN_REFLINK_RE = / |
||
| 842 | \[([^\[\]]+)\] # $text
|
||
| 843 | [ ]? # opt. space
|
||
| 844 | (?:\n[ ]*)? # one optional newline followed by spaces
|
||
| 845 | \[(.*?)\] # $id
|
||
| 846 | /x
|
||
| 847 | |||
| 848 | def inline_markdown_reflink( text ) |
||
| 849 | text.gsub!( MARKDOWN_REFLINK_RE ) do |m| |
||
| 850 | text, id = $~[1..2] |
||
| 851 | |||
| 852 | if id.empty?
|
||
| 853 | url, title = check_refs( text ) |
||
| 854 | else
|
||
| 855 | url, title = check_refs( id ) |
||
| 856 | end
|
||
| 857 | |||
| 858 | atts = " href=\"#{ url }\""
|
||
| 859 | atts << " title=\"#{ title }\"" if title |
||
| 860 | atts = shelve( atts ) |
||
| 861 | |||
| 862 | "<a#{ atts }>#{ text }</a>"
|
||
| 863 | end
|
||
| 864 | end
|
||
| 865 | |||
| 866 | MARKDOWN_LINK_RE = / |
||
| 867 | \[([^\[\]]+)\] # $text
|
||
| 868 | \( # open paren
|
||
| 869 | [ \t]* # opt space
|
||
| 870 | <?(.+?)>? # $href
|
||
| 871 | [ \t]* # opt space
|
||
| 872 | (?: # whole title
|
||
| 873 | (['"]) # $quote
|
||
| 874 | (.*?) # $title
|
||
| 875 | \3 # matching quote
|
||
| 876 | )? # title is optional
|
||
| 877 | \)
|
||
| 878 | /x
|
||
| 879 | |||
| 880 | def inline_markdown_link( text ) |
||
| 881 | text.gsub!( MARKDOWN_LINK_RE ) do |m| |
||
| 882 | text, url, quote, title = $~[1..4] |
||
| 883 | |||
| 884 | atts = " href=\"#{ url }\""
|
||
| 885 | atts << " title=\"#{ title }\"" if title |
||
| 886 | atts = shelve( atts ) |
||
| 887 | |||
| 888 | "<a#{ atts }>#{ text }</a>"
|
||
| 889 | end
|
||
| 890 | end
|
||
| 891 | |||
| 892 | TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/ |
||
| 893 | MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m |
||
| 894 | |||
| 895 | def refs( text ) |
||
| 896 | @rules.each do |rule_name| |
||
| 897 | method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/ |
||
| 898 | end
|
||
| 899 | end
|
||
| 900 | |||
| 901 | def refs_textile( text ) |
||
| 902 | text.gsub!( TEXTILE_REFS_RE ) do |m| |
||
| 903 | flag, url = $~[2..3] |
||
| 904 | @urlrefs[flag.downcase] = [url, nil] |
||
| 905 | nil
|
||
| 906 | end
|
||
| 907 | end
|
||
| 908 | |||
| 909 | def refs_markdown( text ) |
||
| 910 | text.gsub!( MARKDOWN_REFS_RE ) do |m| |
||
| 911 | flag, url = $~[2..3] |
||
| 912 | title = $~[6] |
||
| 913 | @urlrefs[flag.downcase] = [url, title]
|
||
| 914 | nil
|
||
| 915 | end
|
||
| 916 | end
|
||
| 917 | |||
| 918 | def check_refs( text ) |
||
| 919 | ret = @urlrefs[text.downcase] if text |
||
| 920 | ret || [text, nil]
|
||
| 921 | end
|
||
| 922 | |||
| 923 | IMAGE_RE = / |
||
| 924 | (>|\s|^) # start of line?
|
||
| 925 | \! # opening
|
||
| 926 | (\<|\=|\>)? # optional alignment atts
|
||
| 927 | (#{C}) # optional style,class atts
|
||
| 928 | (?:\. )? # optional dot-space
|
||
| 929 | ([^\s(!]+?) # presume this is the src
|
||
| 930 | \s? # optional space
|
||
| 931 | (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
|
||
| 932 | \! # closing
|
||
| 933 | (?::#{ HYPERLINK })? # optional href
|
||
| 934 | /x
|
||
| 935 | |||
| 936 | def inline_textile_image( text ) |
||
| 937 | text.gsub!( IMAGE_RE ) do |m| |
||
| 938 | stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8] |
||
| 939 | htmlesc title |
||
| 940 | atts = pba( atts ) |
||
| 941 | atts = " src=\"#{ url }\"#{ atts }"
|
||
| 942 | atts << " title=\"#{ title }\"" if title |
||
| 943 | atts << " alt=\"#{ title }\""
|
||
| 944 | # size = @getimagesize($url);
|
||
| 945 | # if($size) $atts.= " $size[3]";
|
||
| 946 | |||
| 947 | href, alt_title = check_refs( href ) if href
|
||
| 948 | url, url_title = check_refs( url ) |
||
| 949 | |||
| 950 | out = ''
|
||
| 951 | out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href |
||
| 952 | out << "<img#{ shelve( atts ) } />"
|
||
| 953 | out << "</a>#{ href_a1 }#{ href_a2 }" if href |
||
| 954 | |||
| 955 | if algn
|
||
| 956 | algn = h_align( algn ) |
||
| 957 | if stln == "<p>" |
||
| 958 | out = "<p style=\"float:#{ algn }\">#{ out }"
|
||
| 959 | else
|
||
| 960 | out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
|
||
| 961 | end
|
||
| 962 | else
|
||
| 963 | out = stln + out |
||
| 964 | end
|
||
| 965 | |||
| 966 | out |
||
| 967 | end
|
||
| 968 | end
|
||
| 969 | |||
| 970 | def shelve( val ) |
||
| 971 | @shelf << val
|
||
| 972 | " :redsh##{ @shelf.length }:"
|
||
| 973 | end
|
||
| 974 | |||
| 975 | def retrieve( text ) |
||
| 976 | @shelf.each_with_index do |r, i| |
||
| 977 | text.gsub!( " :redsh##{ i + 1 }:", r )
|
||
| 978 | end
|
||
| 979 | end
|
||
| 980 | |||
| 981 | def incoming_entities( text ) |
||
| 982 | ## turn any incoming ampersands into a dummy character for now.
|
||
| 983 | ## This uses a negative lookahead for alphanumerics followed by a semicolon,
|
||
| 984 | ## implying an incoming html entity, to be skipped
|
||
| 985 | |||
| 986 | text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" ) |
||
| 987 | end
|
||
| 988 | |||
| 989 | def no_textile( text ) |
||
| 990 | text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
|
||
| 991 | '\1<notextile>\2</notextile>\3' )
|
||
| 992 | text.gsub!( /^ *==([^=]+.*?)==/m,
|
||
| 993 | '\1<notextile>\2</notextile>\3' )
|
||
| 994 | end
|
||
| 995 | |||
| 996 | def clean_white_space( text ) |
||
| 997 | # normalize line breaks
|
||
| 998 | text.gsub!( /\r\n/, "\n" ) |
||
| 999 | text.gsub!( /\r/, "\n" ) |
||
| 1000 | text.gsub!( /\t/, ' ' ) |
||
| 1001 | text.gsub!( /^ +$/, '' ) |
||
| 1002 | text.gsub!( /\n{3,}/, "\n\n" ) |
||
| 1003 | text.gsub!( /"$/, "\" " ) |
||
| 1004 | |||
| 1005 | # if entire document is indented, flush
|
||
| 1006 | # to the left side
|
||
| 1007 | flush_left text |
||
| 1008 | end
|
||
| 1009 | |||
| 1010 | def flush_left( text ) |
||
| 1011 | indt = 0
|
||
| 1012 | if text =~ /^ / |
||
| 1013 | while text !~ /^ {#{indt}}\S/ |
||
| 1014 | indt += 1
|
||
| 1015 | end unless text.empty? |
||
| 1016 | if indt.nonzero?
|
||
| 1017 | text.gsub!( /^ {#{indt}}/, '' ) |
||
| 1018 | end
|
||
| 1019 | end
|
||
| 1020 | end
|
||
| 1021 | |||
| 1022 | def footnote_ref( text ) |
||
| 1023 | text.gsub!( /\b\[([0-9]+?)\](\s)?/,
|
||
| 1024 | '<sup><a href="#fn\1">\1</a></sup>\2' )
|
||
| 1025 | end
|
||
| 1026 | |||
| 1027 | OFFTAGS = /(code|pre|kbd|notextile)/ |
||
| 1028 | OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi |
||
| 1029 | OFFTAG_OPEN = /<#{ OFFTAGS }/ |
||
| 1030 | OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/ |
||
| 1031 | HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m |
||
| 1032 | ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m |
||
| 1033 | |||
| 1034 | def glyphs_textile( text, level = 0 ) |
||
| 1035 | if text !~ HASTAG_MATCH |
||
| 1036 | pgl text |
||
| 1037 | footnote_ref text |
||
| 1038 | else
|
||
| 1039 | codepre = 0
|
||
| 1040 | text.gsub!( ALLTAG_MATCH ) do |line| |
||
| 1041 | ## matches are off if we're between <code>, <pre> etc.
|
||
| 1042 | if $1 |
||
| 1043 | if line =~ OFFTAG_OPEN |
||
| 1044 | codepre += 1
|
||
| 1045 | elsif line =~ OFFTAG_CLOSE |
||
| 1046 | codepre -= 1
|
||
| 1047 | codepre = 0 if codepre < 0 |
||
| 1048 | end
|
||
| 1049 | elsif codepre.zero?
|
||
| 1050 | glyphs_textile( line, level + 1 )
|
||
| 1051 | else
|
||
| 1052 | htmlesc( line, :NoQuotes )
|
||
| 1053 | end
|
||
| 1054 | # p [level, codepre, line]
|
||
| 1055 | |||
| 1056 | line |
||
| 1057 | end
|
||
| 1058 | end
|
||
| 1059 | end
|
||
| 1060 | |||
| 1061 | def rip_offtags( text, escape_aftertag=true ) |
||
| 1062 | if text =~ /<.*>/ |
||
| 1063 | ## strip and encode <pre> content
|
||
| 1064 | codepre, used_offtags = 0, {}
|
||
| 1065 | text.gsub!( OFFTAG_MATCH ) do |line| |
||
| 1066 | if $3 |
||
| 1067 | first, offtag, aftertag = $3, $4, $5 |
||
| 1068 | codepre += 1
|
||
| 1069 | used_offtags[offtag] = true
|
||
| 1070 | if codepre - used_offtags.length > 0 |
||
| 1071 | htmlesc( line, :NoQuotes )
|
||
| 1072 | @pre_list.last << line
|
||
| 1073 | line = ""
|
||
| 1074 | else
|
||
| 1075 | ### htmlesc is disabled between CODE tags which will be parsed with highlighter
|
||
| 1076 | ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
|
||
| 1077 | ### NB: some changes were made not to use $N variables, because we use "match"
|
||
| 1078 | ### and it breaks following lines
|
||
| 1079 | htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/) |
||
| 1080 | line = "<redpre##{ @pre_list.length }>"
|
||
| 1081 | first.match(/<#{ OFFTAGS }([^>]*)>/)
|
||
| 1082 | tag = $1
|
||
| 1083 | 119:8661b858af72 | Chris | $2.to_s.match(/(class\=("[^"]+"|'[^']+'))/i) |
| 1084 | 0:513646585e45 | Chris | tag << " #{$1}" if $1 |
| 1085 | @pre_list << "<#{ tag }>#{ aftertag }" |
||
| 1086 | end
|
||
| 1087 | elsif $1 and codepre > 0 |
||
| 1088 | if codepre - used_offtags.length > 0 |
||
| 1089 | htmlesc( line, :NoQuotes )
|
||
| 1090 | @pre_list.last << line
|
||
| 1091 | line = ""
|
||
| 1092 | end
|
||
| 1093 | codepre -= 1 unless codepre.zero? |
||
| 1094 | used_offtags = {} if codepre.zero?
|
||
| 1095 | end
|
||
| 1096 | line |
||
| 1097 | end
|
||
| 1098 | end
|
||
| 1099 | text |
||
| 1100 | end
|
||
| 1101 | |||
| 1102 | def smooth_offtags( text ) |
||
| 1103 | unless @pre_list.empty? |
||
| 1104 | ## replace <pre> content
|
||
| 1105 | text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] } |
||
| 1106 | end
|
||
| 1107 | end
|
||
| 1108 | |||
| 1109 | def inline( text ) |
||
| 1110 | [/^inline_/, /^glyphs_/].each do |meth_re| |
||
| 1111 | @rules.each do |rule_name| |
||
| 1112 | method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
|
||
| 1113 | end
|
||
| 1114 | end
|
||
| 1115 | end
|
||
| 1116 | |||
| 1117 | def h_align( text ) |
||
| 1118 | H_ALGN_VALS[text]
|
||
| 1119 | end
|
||
| 1120 | |||
| 1121 | def v_align( text ) |
||
| 1122 | V_ALGN_VALS[text]
|
||
| 1123 | end
|
||
| 1124 | |||
| 1125 | def textile_popup_help( name, windowW, windowH ) |
||
| 1126 | ' <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 />' |
||
| 1127 | end
|
||
| 1128 | |||
| 1129 | # HTML cleansing stuff
|
||
| 1130 | BASIC_TAGS = {
|
||
| 1131 | 'a' => ['href', 'title'], |
||
| 1132 | 'img' => ['src', 'alt', 'title'], |
||
| 1133 | 'br' => [],
|
||
| 1134 | 'i' => nil, |
||
| 1135 | 'u' => nil, |
||
| 1136 | 'b' => nil, |
||
| 1137 | 'pre' => nil, |
||
| 1138 | 'kbd' => nil, |
||
| 1139 | 'code' => ['lang'], |
||
| 1140 | 'cite' => nil, |
||
| 1141 | 'strong' => nil, |
||
| 1142 | 'em' => nil, |
||
| 1143 | 'ins' => nil, |
||
| 1144 | 'sup' => nil, |
||
| 1145 | 'sub' => nil, |
||
| 1146 | 'del' => nil, |
||
| 1147 | 'table' => nil, |
||
| 1148 | 'tr' => nil, |
||
| 1149 | 'td' => ['colspan', 'rowspan'], |
||
| 1150 | 'th' => nil, |
||
| 1151 | 'ol' => nil, |
||
| 1152 | 'ul' => nil, |
||
| 1153 | 'li' => nil, |
||
| 1154 | 'p' => nil, |
||
| 1155 | 'h1' => nil, |
||
| 1156 | 'h2' => nil, |
||
| 1157 | 'h3' => nil, |
||
| 1158 | 'h4' => nil, |
||
| 1159 | 'h5' => nil, |
||
| 1160 | 'h6' => nil, |
||
| 1161 | 'blockquote' => ['cite'] |
||
| 1162 | } |
||
| 1163 | |||
| 1164 | def clean_html( text, tags = BASIC_TAGS ) |
||
| 1165 | text.gsub!( /<!\[CDATA\[/, '' ) |
||
| 1166 | text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do |
||
| 1167 | raw = $~
|
||
| 1168 | tag = raw[2].downcase
|
||
| 1169 | if tags.has_key? tag
|
||
| 1170 | pcs = [tag] |
||
| 1171 | tags[tag].each do |prop|
|
||
| 1172 | ['"', "'", ''].each do |q| |
||
| 1173 | q2 = ( q != '' ? q : '\s' ) |
||
| 1174 | if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i |
||
| 1175 | attrv = $1
|
||
| 1176 | next if prop == 'src' and attrv =~ %r{^(?!http)\w+:} |
||
| 1177 | pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
|
||
| 1178 | break
|
||
| 1179 | end
|
||
| 1180 | end
|
||
| 1181 | end if tags[tag] |
||
| 1182 | "<#{raw[1]}#{pcs.join " "}>"
|
||
| 1183 | else
|
||
| 1184 | " "
|
||
| 1185 | end
|
||
| 1186 | end
|
||
| 1187 | end
|
||
| 1188 | |||
| 1189 | ALLOWED_TAGS = %w(redpre pre code notextile) |
||
| 1190 | |||
| 1191 | def escape_html_tags(text) |
||
| 1192 | text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "<#{$1}#{'>' unless $3.blank?}" } |
||
| 1193 | end
|
||
| 1194 | end
|