##// END OF EJS Templates
Fixed: Wiki headings containing backslash followed by a digit are not displayed correctly (#7846)....
Jean-Philippe Lang -
r4980:7db930f1b44b
parent child
Show More
@@ -1,1193 +1,1195
1 # vim:ts=4:sw=4:
1 # vim:ts=4:sw=4:
2 # = RedCloth - Textile and Markdown Hybrid for Ruby
2 # = RedCloth - Textile and Markdown Hybrid for Ruby
3 #
3 #
4 # Homepage:: http://whytheluckystiff.net/ruby/redcloth/
4 # Homepage:: http://whytheluckystiff.net/ruby/redcloth/
5 # Author:: why the lucky stiff (http://whytheluckystiff.net/)
5 # Author:: why the lucky stiff (http://whytheluckystiff.net/)
6 # Copyright:: (cc) 2004 why the lucky stiff (and his puppet organizations.)
6 # Copyright:: (cc) 2004 why the lucky stiff (and his puppet organizations.)
7 # License:: BSD
7 # License:: BSD
8 #
8 #
9 # (see http://hobix.com/textile/ for a Textile Reference.)
9 # (see http://hobix.com/textile/ for a Textile Reference.)
10 #
10 #
11 # Based on (and also inspired by) both:
11 # Based on (and also inspired by) both:
12 #
12 #
13 # PyTextile: http://diveintomark.org/projects/textile/textile.py.txt
13 # PyTextile: http://diveintomark.org/projects/textile/textile.py.txt
14 # Textism for PHP: http://www.textism.com/tools/textile/
14 # Textism for PHP: http://www.textism.com/tools/textile/
15 #
15 #
16 #
16 #
17
17
18 # = RedCloth
18 # = RedCloth
19 #
19 #
20 # RedCloth is a Ruby library for converting Textile and/or Markdown
20 # RedCloth is a Ruby library for converting Textile and/or Markdown
21 # into HTML. You can use either format, intermingled or separately.
21 # into HTML. You can use either format, intermingled or separately.
22 # You can also extend RedCloth to honor your own custom text stylings.
22 # You can also extend RedCloth to honor your own custom text stylings.
23 #
23 #
24 # RedCloth users are encouraged to use Textile if they are generating
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.
25 # HTML and to use Markdown if others will be viewing the plain text.
26 #
26 #
27 # == What is Textile?
27 # == What is Textile?
28 #
28 #
29 # Textile is a simple formatting style for text
29 # Textile is a simple formatting style for text
30 # documents, loosely based on some HTML conventions.
30 # documents, loosely based on some HTML conventions.
31 #
31 #
32 # == Sample Textile Text
32 # == Sample Textile Text
33 #
33 #
34 # h2. This is a title
34 # h2. This is a title
35 #
35 #
36 # h3. This is a subhead
36 # h3. This is a subhead
37 #
37 #
38 # This is a bit of paragraph.
38 # This is a bit of paragraph.
39 #
39 #
40 # bq. This is a blockquote.
40 # bq. This is a blockquote.
41 #
41 #
42 # = Writing Textile
42 # = Writing Textile
43 #
43 #
44 # A Textile document consists of paragraphs. Paragraphs
44 # A Textile document consists of paragraphs. Paragraphs
45 # can be specially formatted by adding a small instruction
45 # can be specially formatted by adding a small instruction
46 # to the beginning of the paragraph.
46 # to the beginning of the paragraph.
47 #
47 #
48 # h[n]. Header of size [n].
48 # h[n]. Header of size [n].
49 # bq. Blockquote.
49 # bq. Blockquote.
50 # # Numeric list.
50 # # Numeric list.
51 # * Bulleted list.
51 # * Bulleted list.
52 #
52 #
53 # == Quick Phrase Modifiers
53 # == Quick Phrase Modifiers
54 #
54 #
55 # Quick phrase modifiers are also included, to allow formatting
55 # Quick phrase modifiers are also included, to allow formatting
56 # of small portions of text within a paragraph.
56 # of small portions of text within a paragraph.
57 #
57 #
58 # \_emphasis\_
58 # \_emphasis\_
59 # \_\_italicized\_\_
59 # \_\_italicized\_\_
60 # \*strong\*
60 # \*strong\*
61 # \*\*bold\*\*
61 # \*\*bold\*\*
62 # ??citation??
62 # ??citation??
63 # -deleted text-
63 # -deleted text-
64 # +inserted text+
64 # +inserted text+
65 # ^superscript^
65 # ^superscript^
66 # ~subscript~
66 # ~subscript~
67 # @code@
67 # @code@
68 # %(classname)span%
68 # %(classname)span%
69 #
69 #
70 # ==notextile== (leave text alone)
70 # ==notextile== (leave text alone)
71 #
71 #
72 # == Links
72 # == Links
73 #
73 #
74 # To make a hypertext link, put the link text in "quotation
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.
75 # marks" followed immediately by a colon and the URL of the link.
76 #
76 #
77 # Optional: text in (parentheses) following the link text,
77 # Optional: text in (parentheses) following the link text,
78 # but before the closing quotation mark, will become a Title
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.
79 # attribute for the link, visible as a tool tip when a cursor is above it.
80 #
80 #
81 # Example:
81 # Example:
82 #
82 #
83 # "This is a link (This is a title) ":http://www.textism.com
83 # "This is a link (This is a title) ":http://www.textism.com
84 #
84 #
85 # Will become:
85 # Will become:
86 #
86 #
87 # <a href="http://www.textism.com" title="This is a title">This is a link</a>
87 # <a href="http://www.textism.com" title="This is a title">This is a link</a>
88 #
88 #
89 # == Images
89 # == Images
90 #
90 #
91 # To insert an image, put the URL for the image inside exclamation marks.
91 # To insert an image, put the URL for the image inside exclamation marks.
92 #
92 #
93 # Optional: text that immediately follows the URL in (parentheses) will
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
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
95 # have descriptive Alt text for the benefit of readers using non-graphical
96 # browsers.
96 # browsers.
97 #
97 #
98 # Optional: place a colon followed by a URL immediately after the
98 # Optional: place a colon followed by a URL immediately after the
99 # closing ! to make the image into a link.
99 # closing ! to make the image into a link.
100 #
100 #
101 # Example:
101 # Example:
102 #
102 #
103 # !http://www.textism.com/common/textist.gif(Textist)!
103 # !http://www.textism.com/common/textist.gif(Textist)!
104 #
104 #
105 # Will become:
105 # Will become:
106 #
106 #
107 # <img src="http://www.textism.com/common/textist.gif" alt="Textist" />
107 # <img src="http://www.textism.com/common/textist.gif" alt="Textist" />
108 #
108 #
109 # With a link:
109 # With a link:
110 #
110 #
111 # !/common/textist.gif(Textist)!:http://textism.com
111 # !/common/textist.gif(Textist)!:http://textism.com
112 #
112 #
113 # Will become:
113 # Will become:
114 #
114 #
115 # <a href="http://textism.com"><img src="/common/textist.gif" alt="Textist" /></a>
115 # <a href="http://textism.com"><img src="/common/textist.gif" alt="Textist" /></a>
116 #
116 #
117 # == Defining Acronyms
117 # == Defining Acronyms
118 #
118 #
119 # HTML allows authors to define acronyms via the tag. The definition appears as a
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,
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.
121 # this should be used at least once for each acronym in documents where they appear.
122 #
122 #
123 # To quickly define an acronym in Textile, place the full text in (parentheses)
123 # To quickly define an acronym in Textile, place the full text in (parentheses)
124 # immediately following the acronym.
124 # immediately following the acronym.
125 #
125 #
126 # Example:
126 # Example:
127 #
127 #
128 # ACLU(American Civil Liberties Union)
128 # ACLU(American Civil Liberties Union)
129 #
129 #
130 # Will become:
130 # Will become:
131 #
131 #
132 # <acronym title="American Civil Liberties Union">ACLU</acronym>
132 # <acronym title="American Civil Liberties Union">ACLU</acronym>
133 #
133 #
134 # == Adding Tables
134 # == Adding Tables
135 #
135 #
136 # In Textile, simple tables can be added by seperating each column by
136 # In Textile, simple tables can be added by seperating each column by
137 # a pipe.
137 # a pipe.
138 #
138 #
139 # |a|simple|table|row|
139 # |a|simple|table|row|
140 # |And|Another|table|row|
140 # |And|Another|table|row|
141 #
141 #
142 # Attributes are defined by style definitions in parentheses.
142 # Attributes are defined by style definitions in parentheses.
143 #
143 #
144 # table(border:1px solid black).
144 # table(border:1px solid black).
145 # (background:#ddd;color:red). |{}| | | |
145 # (background:#ddd;color:red). |{}| | | |
146 #
146 #
147 # == Using RedCloth
147 # == Using RedCloth
148 #
148 #
149 # RedCloth is simply an extension of the String class, which can handle
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
150 # Textile formatting. Use it like a String and output HTML with its
151 # RedCloth#to_html method.
151 # RedCloth#to_html method.
152 #
152 #
153 # doc = RedCloth.new "
153 # doc = RedCloth.new "
154 #
154 #
155 # h2. Test document
155 # h2. Test document
156 #
156 #
157 # Just a simple test."
157 # Just a simple test."
158 #
158 #
159 # puts doc.to_html
159 # puts doc.to_html
160 #
160 #
161 # By default, RedCloth uses both Textile and Markdown formatting, with
161 # By default, RedCloth uses both Textile and Markdown formatting, with
162 # Textile formatting taking precedence. If you want to turn off Markdown
162 # Textile formatting taking precedence. If you want to turn off Markdown
163 # formatting, to boost speed and limit the processor:
163 # formatting, to boost speed and limit the processor:
164 #
164 #
165 # class RedCloth::Textile.new( str )
165 # class RedCloth::Textile.new( str )
166
166
167 class RedCloth3 < String
167 class RedCloth3 < String
168
168
169 VERSION = '3.0.4'
169 VERSION = '3.0.4'
170 DEFAULT_RULES = [:textile, :markdown]
170 DEFAULT_RULES = [:textile, :markdown]
171
171
172 #
172 #
173 # Two accessor for setting security restrictions.
173 # Two accessor for setting security restrictions.
174 #
174 #
175 # This is a nice thing if you're using RedCloth for
175 # This is a nice thing if you're using RedCloth for
176 # formatting in public places (e.g. Wikis) where you
176 # formatting in public places (e.g. Wikis) where you
177 # don't want users to abuse HTML for bad things.
177 # don't want users to abuse HTML for bad things.
178 #
178 #
179 # If +:filter_html+ is set, HTML which wasn't
179 # If +:filter_html+ is set, HTML which wasn't
180 # created by the Textile processor will be escaped.
180 # created by the Textile processor will be escaped.
181 #
181 #
182 # If +:filter_styles+ is set, it will also disable
182 # If +:filter_styles+ is set, it will also disable
183 # the style markup specifier. ('{color: red}')
183 # the style markup specifier. ('{color: red}')
184 #
184 #
185 attr_accessor :filter_html, :filter_styles
185 attr_accessor :filter_html, :filter_styles
186
186
187 #
187 #
188 # Accessor for toggling hard breaks.
188 # Accessor for toggling hard breaks.
189 #
189 #
190 # If +:hard_breaks+ is set, single newlines will
190 # If +:hard_breaks+ is set, single newlines will
191 # be converted to HTML break tags. This is the
191 # be converted to HTML break tags. This is the
192 # default behavior for traditional RedCloth.
192 # default behavior for traditional RedCloth.
193 #
193 #
194 attr_accessor :hard_breaks
194 attr_accessor :hard_breaks
195
195
196 # Accessor for toggling lite mode.
196 # Accessor for toggling lite mode.
197 #
197 #
198 # In lite mode, block-level rules are ignored. This means
198 # In lite mode, block-level rules are ignored. This means
199 # that tables, paragraphs, lists, and such aren't available.
199 # that tables, paragraphs, lists, and such aren't available.
200 # Only the inline markup for bold, italics, entities and so on.
200 # Only the inline markup for bold, italics, entities and so on.
201 #
201 #
202 # r = RedCloth.new( "And then? She *fell*!", [:lite_mode] )
202 # r = RedCloth.new( "And then? She *fell*!", [:lite_mode] )
203 # r.to_html
203 # r.to_html
204 # #=> "And then? She <strong>fell</strong>!"
204 # #=> "And then? She <strong>fell</strong>!"
205 #
205 #
206 attr_accessor :lite_mode
206 attr_accessor :lite_mode
207
207
208 #
208 #
209 # Accessor for toggling span caps.
209 # Accessor for toggling span caps.
210 #
210 #
211 # Textile places `span' tags around capitalized
211 # Textile places `span' tags around capitalized
212 # words by default, but this wreaks havoc on Wikis.
212 # words by default, but this wreaks havoc on Wikis.
213 # If +:no_span_caps+ is set, this will be
213 # If +:no_span_caps+ is set, this will be
214 # suppressed.
214 # suppressed.
215 #
215 #
216 attr_accessor :no_span_caps
216 attr_accessor :no_span_caps
217
217
218 #
218 #
219 # Establishes the markup predence. Available rules include:
219 # Establishes the markup predence. Available rules include:
220 #
220 #
221 # == Textile Rules
221 # == Textile Rules
222 #
222 #
223 # The following textile rules can be set individually. Or add the complete
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
224 # set of rules with the single :textile rule, which supplies the rule set in
225 # the following precedence:
225 # the following precedence:
226 #
226 #
227 # refs_textile:: Textile references (i.e. [hobix]http://hobix.com/)
227 # refs_textile:: Textile references (i.e. [hobix]http://hobix.com/)
228 # block_textile_table:: Textile table block structures
228 # block_textile_table:: Textile table block structures
229 # block_textile_lists:: Textile list structures
229 # block_textile_lists:: Textile list structures
230 # block_textile_prefix:: Textile blocks with prefixes (i.e. bq., h2., etc.)
230 # block_textile_prefix:: Textile blocks with prefixes (i.e. bq., h2., etc.)
231 # inline_textile_image:: Textile inline images
231 # inline_textile_image:: Textile inline images
232 # inline_textile_link:: Textile inline links
232 # inline_textile_link:: Textile inline links
233 # inline_textile_span:: Textile inline spans
233 # inline_textile_span:: Textile inline spans
234 # glyphs_textile:: Textile entities (such as em-dashes and smart quotes)
234 # glyphs_textile:: Textile entities (such as em-dashes and smart quotes)
235 #
235 #
236 # == Markdown
236 # == Markdown
237 #
237 #
238 # refs_markdown:: Markdown references (for example: [hobix]: http://hobix.com/)
238 # refs_markdown:: Markdown references (for example: [hobix]: http://hobix.com/)
239 # block_markdown_setext:: Markdown setext headers
239 # block_markdown_setext:: Markdown setext headers
240 # block_markdown_atx:: Markdown atx headers
240 # block_markdown_atx:: Markdown atx headers
241 # block_markdown_rule:: Markdown horizontal rules
241 # block_markdown_rule:: Markdown horizontal rules
242 # block_markdown_bq:: Markdown blockquotes
242 # block_markdown_bq:: Markdown blockquotes
243 # block_markdown_lists:: Markdown lists
243 # block_markdown_lists:: Markdown lists
244 # inline_markdown_link:: Markdown links
244 # inline_markdown_link:: Markdown links
245 attr_accessor :rules
245 attr_accessor :rules
246
246
247 # Returns a new RedCloth object, based on _string_ and
247 # Returns a new RedCloth object, based on _string_ and
248 # enforcing all the included _restrictions_.
248 # enforcing all the included _restrictions_.
249 #
249 #
250 # r = RedCloth.new( "h1. A <b>bold</b> man", [:filter_html] )
250 # r = RedCloth.new( "h1. A <b>bold</b> man", [:filter_html] )
251 # r.to_html
251 # r.to_html
252 # #=>"<h1>A &lt;b&gt;bold&lt;/b&gt; man</h1>"
252 # #=>"<h1>A &lt;b&gt;bold&lt;/b&gt; man</h1>"
253 #
253 #
254 def initialize( string, restrictions = [] )
254 def initialize( string, restrictions = [] )
255 restrictions.each { |r| method( "#{ r }=" ).call( true ) }
255 restrictions.each { |r| method( "#{ r }=" ).call( true ) }
256 super( string )
256 super( string )
257 end
257 end
258
258
259 #
259 #
260 # Generates HTML from the Textile contents.
260 # Generates HTML from the Textile contents.
261 #
261 #
262 # r = RedCloth.new( "And then? She *fell*!" )
262 # r = RedCloth.new( "And then? She *fell*!" )
263 # r.to_html( true )
263 # r.to_html( true )
264 # #=>"And then? She <strong>fell</strong>!"
264 # #=>"And then? She <strong>fell</strong>!"
265 #
265 #
266 def to_html( *rules )
266 def to_html( *rules )
267 rules = DEFAULT_RULES if rules.empty?
267 rules = DEFAULT_RULES if rules.empty?
268 # make our working copy
268 # make our working copy
269 text = self.dup
269 text = self.dup
270
270
271 @urlrefs = {}
271 @urlrefs = {}
272 @shelf = []
272 @shelf = []
273 textile_rules = [:block_textile_table, :block_textile_lists,
273 textile_rules = [:block_textile_table, :block_textile_lists,
274 :block_textile_prefix, :inline_textile_image, :inline_textile_link,
274 :block_textile_prefix, :inline_textile_image, :inline_textile_link,
275 :inline_textile_code, :inline_textile_span, :glyphs_textile]
275 :inline_textile_code, :inline_textile_span, :glyphs_textile]
276 markdown_rules = [:refs_markdown, :block_markdown_setext, :block_markdown_atx, :block_markdown_rule,
276 markdown_rules = [:refs_markdown, :block_markdown_setext, :block_markdown_atx, :block_markdown_rule,
277 :block_markdown_bq, :block_markdown_lists,
277 :block_markdown_bq, :block_markdown_lists,
278 :inline_markdown_reflink, :inline_markdown_link]
278 :inline_markdown_reflink, :inline_markdown_link]
279 @rules = rules.collect do |rule|
279 @rules = rules.collect do |rule|
280 case rule
280 case rule
281 when :markdown
281 when :markdown
282 markdown_rules
282 markdown_rules
283 when :textile
283 when :textile
284 textile_rules
284 textile_rules
285 else
285 else
286 rule
286 rule
287 end
287 end
288 end.flatten
288 end.flatten
289
289
290 # standard clean up
290 # standard clean up
291 incoming_entities text
291 incoming_entities text
292 clean_white_space text
292 clean_white_space text
293
293
294 # start processor
294 # start processor
295 @pre_list = []
295 @pre_list = []
296 rip_offtags text
296 rip_offtags text
297 no_textile text
297 no_textile text
298 escape_html_tags text
298 escape_html_tags text
299 hard_break text
299 hard_break text
300 unless @lite_mode
300 unless @lite_mode
301 refs text
301 refs text
302 # need to do this before text is split by #blocks
302 # need to do this before text is split by #blocks
303 block_textile_quotes text
303 block_textile_quotes text
304 blocks text
304 blocks text
305 end
305 end
306 inline text
306 inline text
307 smooth_offtags text
307 smooth_offtags text
308
308
309 retrieve text
309 retrieve text
310
310
311 text.gsub!( /<\/?notextile>/, '' )
311 text.gsub!( /<\/?notextile>/, '' )
312 text.gsub!( /x%x%/, '&#38;' )
312 text.gsub!( /x%x%/, '&#38;' )
313 clean_html text if filter_html
313 clean_html text if filter_html
314 text.strip!
314 text.strip!
315 text
315 text
316
316
317 end
317 end
318
318
319 #######
319 #######
320 private
320 private
321 #######
321 #######
322 #
322 #
323 # Mapping of 8-bit ASCII codes to HTML numerical entity equivalents.
323 # Mapping of 8-bit ASCII codes to HTML numerical entity equivalents.
324 # (from PyTextile)
324 # (from PyTextile)
325 #
325 #
326 TEXTILE_TAGS =
326 TEXTILE_TAGS =
327
327
328 [[128, 8364], [129, 0], [130, 8218], [131, 402], [132, 8222], [133, 8230],
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],
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],
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],
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]].
332 [153, 8482], [154, 353], [155, 8250], [156, 339], [157, 0], [158, 0], [159, 376]].
333
333
334 collect! do |a, b|
334 collect! do |a, b|
335 [a.chr, ( b.zero? and "" or "&#{ b };" )]
335 [a.chr, ( b.zero? and "" or "&#{ b };" )]
336 end
336 end
337
337
338 #
338 #
339 # Regular expressions to convert to HTML.
339 # Regular expressions to convert to HTML.
340 #
340 #
341 A_HLGN = /(?:(?:<>|<|>|\=|[()]+)+)/
341 A_HLGN = /(?:(?:<>|<|>|\=|[()]+)+)/
342 A_VLGN = /[\-^~]/
342 A_VLGN = /[\-^~]/
343 C_CLAS = '(?:\([^)]+\))'
343 C_CLAS = '(?:\([^)]+\))'
344 C_LNGE = '(?:\[[^\[\]]+\])'
344 C_LNGE = '(?:\[[^\[\]]+\])'
345 C_STYL = '(?:\{[^}]+\})'
345 C_STYL = '(?:\{[^}]+\})'
346 S_CSPN = '(?:\\\\\d+)'
346 S_CSPN = '(?:\\\\\d+)'
347 S_RSPN = '(?:/\d+)'
347 S_RSPN = '(?:/\d+)'
348 A = "(?:#{A_HLGN}?#{A_VLGN}?|#{A_VLGN}?#{A_HLGN}?)"
348 A = "(?:#{A_HLGN}?#{A_VLGN}?|#{A_VLGN}?#{A_HLGN}?)"
349 S = "(?:#{S_CSPN}?#{S_RSPN}|#{S_RSPN}?#{S_CSPN}?)"
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}?)"
350 C = "(?:#{C_CLAS}?#{C_STYL}?#{C_LNGE}?|#{C_STYL}?#{C_LNGE}?#{C_CLAS}?|#{C_LNGE}?#{C_STYL}?#{C_CLAS}?)"
351 # PUNCT = Regexp::quote( '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~' )
351 # PUNCT = Regexp::quote( '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~' )
352 PUNCT = Regexp::quote( '!"#$%&\'*+,-./:;=?@\\^_`|~' )
352 PUNCT = Regexp::quote( '!"#$%&\'*+,-./:;=?@\\^_`|~' )
353 PUNCT_NOQ = Regexp::quote( '!"#$&\',./:;=?@\\`|' )
353 PUNCT_NOQ = Regexp::quote( '!"#$&\',./:;=?@\\`|' )
354 PUNCT_Q = Regexp::quote( '*-_+^~%' )
354 PUNCT_Q = Regexp::quote( '*-_+^~%' )
355 HYPERLINK = '(\S+?)([^\w\s/;=\?]*?)(?=\s|<|$)'
355 HYPERLINK = '(\S+?)([^\w\s/;=\?]*?)(?=\s|<|$)'
356
356
357 # Text markup tags, don't conflict with block tags
357 # Text markup tags, don't conflict with block tags
358 SIMPLE_HTML_TAGS = [
358 SIMPLE_HTML_TAGS = [
359 'tt', 'b', 'i', 'big', 'small', 'em', 'strong', 'dfn', 'code',
359 'tt', 'b', 'i', 'big', 'small', 'em', 'strong', 'dfn', 'code',
360 'samp', 'kbd', 'var', 'cite', 'abbr', 'acronym', 'a', 'img', 'br',
360 'samp', 'kbd', 'var', 'cite', 'abbr', 'acronym', 'a', 'img', 'br',
361 'br', 'map', 'q', 'sub', 'sup', 'span', 'bdo'
361 'br', 'map', 'q', 'sub', 'sup', 'span', 'bdo'
362 ]
362 ]
363
363
364 QTAGS = [
364 QTAGS = [
365 ['**', 'b', :limit],
365 ['**', 'b', :limit],
366 ['*', 'strong', :limit],
366 ['*', 'strong', :limit],
367 ['??', 'cite', :limit],
367 ['??', 'cite', :limit],
368 ['-', 'del', :limit],
368 ['-', 'del', :limit],
369 ['__', 'i', :limit],
369 ['__', 'i', :limit],
370 ['_', 'em', :limit],
370 ['_', 'em', :limit],
371 ['%', 'span', :limit],
371 ['%', 'span', :limit],
372 ['+', 'ins', :limit],
372 ['+', 'ins', :limit],
373 ['^', 'sup', :limit],
373 ['^', 'sup', :limit],
374 ['~', 'sub', :limit]
374 ['~', 'sub', :limit]
375 ]
375 ]
376 QTAGS_JOIN = QTAGS.map {|rc, ht, rtype| Regexp::quote rc}.join('|')
376 QTAGS_JOIN = QTAGS.map {|rc, ht, rtype| Regexp::quote rc}.join('|')
377
377
378 QTAGS.collect! do |rc, ht, rtype|
378 QTAGS.collect! do |rc, ht, rtype|
379 rcq = Regexp::quote rc
379 rcq = Regexp::quote rc
380 re =
380 re =
381 case rtype
381 case rtype
382 when :limit
382 when :limit
383 /(^|[>\s\(]) # sta
383 /(^|[>\s\(]) # sta
384 (?!\-\-)
384 (?!\-\-)
385 (#{QTAGS_JOIN}|) # oqs
385 (#{QTAGS_JOIN}|) # oqs
386 (#{rcq}) # qtag
386 (#{rcq}) # qtag
387 (\w|[^\s].*?[^\s]) # content
387 (\w|[^\s].*?[^\s]) # content
388 (?!\-\-)
388 (?!\-\-)
389 #{rcq}
389 #{rcq}
390 (#{QTAGS_JOIN}|) # oqa
390 (#{QTAGS_JOIN}|) # oqa
391 (?=[[:punct:]]|<|\s|\)|$)/x
391 (?=[[:punct:]]|<|\s|\)|$)/x
392 else
392 else
393 /(#{rcq})
393 /(#{rcq})
394 (#{C})
394 (#{C})
395 (?::(\S+))?
395 (?::(\S+))?
396 (\w|[^\s\-].*?[^\s\-])
396 (\w|[^\s\-].*?[^\s\-])
397 #{rcq}/xm
397 #{rcq}/xm
398 end
398 end
399 [rc, ht, re, rtype]
399 [rc, ht, re, rtype]
400 end
400 end
401
401
402 # Elements to handle
402 # Elements to handle
403 GLYPHS = [
403 GLYPHS = [
404 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1&#8217;\2' ], # single closing
404 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1&#8217;\2' ], # single closing
405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
407 # [ /\'/, '&#8216;' ], # single opening
407 # [ /\'/, '&#8216;' ], # single opening
408 # [ /</, '&lt;' ], # less-than
408 # [ /</, '&lt;' ], # less-than
409 # [ />/, '&gt;' ], # greater-than
409 # [ />/, '&gt;' ], # greater-than
410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
413 # [ /"/, '&#8220;' ], # double opening
413 # [ /"/, '&#8220;' ], # double opening
414 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
414 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
415 # [ /\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/, '<acronym title="\2">\1</acronym>' ], # 3+ uppercase acronym
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]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
416 # [ /(^|[^"][>\s])([A-Z][A-Z0-9 ]+[A-Z0-9])([^<A-Za-z0-9]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
417 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
417 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
418 # [ /\s->\s/, ' &rarr; ' ], # right arrow
418 # [ /\s->\s/, ' &rarr; ' ], # right arrow
419 # [ /\s-\s/, ' &#8211; ' ], # en dash
419 # [ /\s-\s/, ' &#8211; ' ], # en dash
420 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
420 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
421 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
421 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
422 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
422 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
423 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
423 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
424 ]
424 ]
425
425
426 H_ALGN_VALS = {
426 H_ALGN_VALS = {
427 '<' => 'left',
427 '<' => 'left',
428 '=' => 'center',
428 '=' => 'center',
429 '>' => 'right',
429 '>' => 'right',
430 '<>' => 'justify'
430 '<>' => 'justify'
431 }
431 }
432
432
433 V_ALGN_VALS = {
433 V_ALGN_VALS = {
434 '^' => 'top',
434 '^' => 'top',
435 '-' => 'middle',
435 '-' => 'middle',
436 '~' => 'bottom'
436 '~' => 'bottom'
437 }
437 }
438
438
439 #
439 #
440 # Flexible HTML escaping
440 # Flexible HTML escaping
441 #
441 #
442 def htmlesc( str, mode=:Quotes )
442 def htmlesc( str, mode=:Quotes )
443 if str
443 if str
444 str.gsub!( '&', '&amp;' )
444 str.gsub!( '&', '&amp;' )
445 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
445 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
446 str.gsub!( "'", '&#039;' ) if mode == :Quotes
446 str.gsub!( "'", '&#039;' ) if mode == :Quotes
447 str.gsub!( '<', '&lt;')
447 str.gsub!( '<', '&lt;')
448 str.gsub!( '>', '&gt;')
448 str.gsub!( '>', '&gt;')
449 end
449 end
450 str
450 str
451 end
451 end
452
452
453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
454 def pgl( text )
454 def pgl( text )
455 #GLYPHS.each do |re, resub, tog|
455 #GLYPHS.each do |re, resub, tog|
456 # next if tog and method( tog ).call
456 # next if tog and method( tog ).call
457 # text.gsub! re, resub
457 # text.gsub! re, resub
458 #end
458 #end
459 text.gsub!(/\b([A-Z][A-Z0-9]{1,})\b(?:[(]([^)]*)[)])/) do |m|
459 text.gsub!(/\b([A-Z][A-Z0-9]{1,})\b(?:[(]([^)]*)[)])/) do |m|
460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
461 end
461 end
462 end
462 end
463
463
464 # Parses Textile attribute lists and builds an HTML attribute string
464 # Parses Textile attribute lists and builds an HTML attribute string
465 def pba( text_in, element = "" )
465 def pba( text_in, element = "" )
466
466
467 return '' unless text_in
467 return '' unless text_in
468
468
469 style = []
469 style = []
470 text = text_in.dup
470 text = text_in.dup
471 if element == 'td'
471 if element == 'td'
472 colspan = $1 if text =~ /\\(\d+)/
472 colspan = $1 if text =~ /\\(\d+)/
473 rowspan = $1 if text =~ /\/(\d+)/
473 rowspan = $1 if text =~ /\/(\d+)/
474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
475 end
475 end
476
476
477 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
477 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
478
478
479 lang = $1 if
479 lang = $1 if
480 text.sub!( /\[([^)]+?)\]/, '' )
480 text.sub!( /\[([^)]+?)\]/, '' )
481
481
482 cls = $1 if
482 cls = $1 if
483 text.sub!( /\(([^()]+?)\)/, '' )
483 text.sub!( /\(([^()]+?)\)/, '' )
484
484
485 style << "padding-left:#{ $1.length }em;" if
485 style << "padding-left:#{ $1.length }em;" if
486 text.sub!( /([(]+)/, '' )
486 text.sub!( /([(]+)/, '' )
487
487
488 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
488 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
489
489
490 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
490 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
491
491
492 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
492 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
493
493
494 atts = ''
494 atts = ''
495 atts << " style=\"#{ style.join }\"" unless style.empty?
495 atts << " style=\"#{ style.join }\"" unless style.empty?
496 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
496 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
497 atts << " lang=\"#{ lang }\"" if lang
497 atts << " lang=\"#{ lang }\"" if lang
498 atts << " id=\"#{ id }\"" if id
498 atts << " id=\"#{ id }\"" if id
499 atts << " colspan=\"#{ colspan }\"" if colspan
499 atts << " colspan=\"#{ colspan }\"" if colspan
500 atts << " rowspan=\"#{ rowspan }\"" if rowspan
500 atts << " rowspan=\"#{ rowspan }\"" if rowspan
501
501
502 atts
502 atts
503 end
503 end
504
504
505 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
505 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
506
506
507 # Parses a Textile table block, building HTML from the result.
507 # Parses a Textile table block, building HTML from the result.
508 def block_textile_table( text )
508 def block_textile_table( text )
509 text.gsub!( TABLE_RE ) do |matches|
509 text.gsub!( TABLE_RE ) do |matches|
510
510
511 tatts, fullrow = $~[1..2]
511 tatts, fullrow = $~[1..2]
512 tatts = pba( tatts, 'table' )
512 tatts = pba( tatts, 'table' )
513 tatts = shelve( tatts ) if tatts
513 tatts = shelve( tatts ) if tatts
514 rows = []
514 rows = []
515
515
516 fullrow.each_line do |row|
516 fullrow.each_line do |row|
517 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
517 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
518 cells = []
518 cells = []
519 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
519 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
520 next if cell == '|'
520 next if cell == '|'
521 ctyp = 'd'
521 ctyp = 'd'
522 ctyp = 'h' if cell =~ /^_/
522 ctyp = 'h' if cell =~ /^_/
523
523
524 catts = ''
524 catts = ''
525 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
525 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
526
526
527 catts = shelve( catts ) if catts
527 catts = shelve( catts ) if catts
528 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
528 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
529 end
529 end
530 ratts = shelve( ratts ) if ratts
530 ratts = shelve( ratts ) if ratts
531 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
531 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
532 end
532 end
533 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
533 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
534 end
534 end
535 end
535 end
536
536
537 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
537 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
538 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
538 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
539
539
540 # Parses Textile lists and generates HTML
540 # Parses Textile lists and generates HTML
541 def block_textile_lists( text )
541 def block_textile_lists( text )
542 text.gsub!( LISTS_RE ) do |match|
542 text.gsub!( LISTS_RE ) do |match|
543 lines = match.split( /\n/ )
543 lines = match.split( /\n/ )
544 last_line = -1
544 last_line = -1
545 depth = []
545 depth = []
546 lines.each_with_index do |line, line_id|
546 lines.each_with_index do |line, line_id|
547 if line =~ LISTS_CONTENT_RE
547 if line =~ LISTS_CONTENT_RE
548 tl,atts,content = $~[1..3]
548 tl,atts,content = $~[1..3]
549 if depth.last
549 if depth.last
550 if depth.last.length > tl.length
550 if depth.last.length > tl.length
551 (depth.length - 1).downto(0) do |i|
551 (depth.length - 1).downto(0) do |i|
552 break if depth[i].length == tl.length
552 break if depth[i].length == tl.length
553 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
553 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
554 depth.pop
554 depth.pop
555 end
555 end
556 end
556 end
557 if depth.last and depth.last.length == tl.length
557 if depth.last and depth.last.length == tl.length
558 lines[line_id - 1] << '</li>'
558 lines[line_id - 1] << '</li>'
559 end
559 end
560 end
560 end
561 unless depth.last == tl
561 unless depth.last == tl
562 depth << tl
562 depth << tl
563 atts = pba( atts )
563 atts = pba( atts )
564 atts = shelve( atts ) if atts
564 atts = shelve( atts ) if atts
565 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
565 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
566 else
566 else
567 lines[line_id] = "\t\t<li>#{ content }"
567 lines[line_id] = "\t\t<li>#{ content }"
568 end
568 end
569 last_line = line_id
569 last_line = line_id
570
570
571 else
571 else
572 last_line = line_id
572 last_line = line_id
573 end
573 end
574 if line_id - last_line > 1 or line_id == lines.length - 1
574 if line_id - last_line > 1 or line_id == lines.length - 1
575 depth.delete_if do |v|
575 depth.delete_if do |v|
576 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
576 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
577 end
577 end
578 end
578 end
579 end
579 end
580 lines.join( "\n" )
580 lines.join( "\n" )
581 end
581 end
582 end
582 end
583
583
584 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
584 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
585 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
585 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
586
586
587 def block_textile_quotes( text )
587 def block_textile_quotes( text )
588 text.gsub!( QUOTES_RE ) do |match|
588 text.gsub!( QUOTES_RE ) do |match|
589 lines = match.split( /\n/ )
589 lines = match.split( /\n/ )
590 quotes = ''
590 quotes = ''
591 indent = 0
591 indent = 0
592 lines.each do |line|
592 lines.each do |line|
593 line =~ QUOTES_CONTENT_RE
593 line =~ QUOTES_CONTENT_RE
594 bq,content = $1, $2
594 bq,content = $1, $2
595 l = bq.count('>')
595 l = bq.count('>')
596 if l != indent
596 if l != indent
597 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
597 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
598 indent = l
598 indent = l
599 end
599 end
600 quotes << (content + "\n")
600 quotes << (content + "\n")
601 end
601 end
602 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
602 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
603 quotes
603 quotes
604 end
604 end
605 end
605 end
606
606
607 CODE_RE = /(\W)
607 CODE_RE = /(\W)
608 @
608 @
609 (?:\|(\w+?)\|)?
609 (?:\|(\w+?)\|)?
610 (.+?)
610 (.+?)
611 @
611 @
612 (?=\W)/x
612 (?=\W)/x
613
613
614 def inline_textile_code( text )
614 def inline_textile_code( text )
615 text.gsub!( CODE_RE ) do |m|
615 text.gsub!( CODE_RE ) do |m|
616 before,lang,code,after = $~[1..4]
616 before,lang,code,after = $~[1..4]
617 lang = " lang=\"#{ lang }\"" if lang
617 lang = " lang=\"#{ lang }\"" if lang
618 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false )
618 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false )
619 end
619 end
620 end
620 end
621
621
622 def lT( text )
622 def lT( text )
623 text =~ /\#$/ ? 'o' : 'u'
623 text =~ /\#$/ ? 'o' : 'u'
624 end
624 end
625
625
626 def hard_break( text )
626 def hard_break( text )
627 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
627 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
628 end
628 end
629
629
630 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
630 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
631
631
632 def blocks( text, deep_code = false )
632 def blocks( text, deep_code = false )
633 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
633 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
634 plain = blk !~ /\A[#*> ]/
634 plain = blk !~ /\A[#*> ]/
635
635
636 # skip blocks that are complex HTML
636 # skip blocks that are complex HTML
637 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
637 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
638 blk
638 blk
639 else
639 else
640 # search for indentation levels
640 # search for indentation levels
641 blk.strip!
641 blk.strip!
642 if blk.empty?
642 if blk.empty?
643 blk
643 blk
644 else
644 else
645 code_blk = nil
645 code_blk = nil
646 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
646 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
647 flush_left iblk
647 flush_left iblk
648 blocks iblk, plain
648 blocks iblk, plain
649 iblk.gsub( /^(\S)/, "\t\\1" )
649 iblk.gsub( /^(\S)/, "\t\\1" )
650 if plain
650 if plain
651 code_blk = iblk; ""
651 code_blk = iblk; ""
652 else
652 else
653 iblk
653 iblk
654 end
654 end
655 end
655 end
656
656
657 block_applied = 0
657 block_applied = 0
658 @rules.each do |rule_name|
658 @rules.each do |rule_name|
659 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
659 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
660 end
660 end
661 if block_applied.zero?
661 if block_applied.zero?
662 if deep_code
662 if deep_code
663 blk = "\t<pre><code>#{ blk }</code></pre>"
663 blk = "\t<pre><code>#{ blk }</code></pre>"
664 else
664 else
665 blk = "\t<p>#{ blk }</p>"
665 blk = "\t<p>#{ blk }</p>"
666 end
666 end
667 end
667 end
668 # hard_break blk
668 # hard_break blk
669 blk + "\n#{ code_blk }"
669 blk + "\n#{ code_blk }"
670 end
670 end
671 end
671 end
672
672
673 end.join( "\n\n" ) )
673 end.join( "\n\n" ) )
674 end
674 end
675
675
676 def textile_bq( tag, atts, cite, content )
676 def textile_bq( tag, atts, cite, content )
677 cite, cite_title = check_refs( cite )
677 cite, cite_title = check_refs( cite )
678 cite = " cite=\"#{ cite }\"" if cite
678 cite = " cite=\"#{ cite }\"" if cite
679 atts = shelve( atts ) if atts
679 atts = shelve( atts ) if atts
680 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
680 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
681 end
681 end
682
682
683 def textile_p( tag, atts, cite, content )
683 def textile_p( tag, atts, cite, content )
684 atts = shelve( atts ) if atts
684 atts = shelve( atts ) if atts
685 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
685 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
686 end
686 end
687
687
688 alias textile_h1 textile_p
688 alias textile_h1 textile_p
689 alias textile_h2 textile_p
689 alias textile_h2 textile_p
690 alias textile_h3 textile_p
690 alias textile_h3 textile_p
691 alias textile_h4 textile_p
691 alias textile_h4 textile_p
692 alias textile_h5 textile_p
692 alias textile_h5 textile_p
693 alias textile_h6 textile_p
693 alias textile_h6 textile_p
694
694
695 def textile_fn_( tag, num, atts, cite, content )
695 def textile_fn_( tag, num, atts, cite, content )
696 atts << " id=\"fn#{ num }\" class=\"footnote\""
696 atts << " id=\"fn#{ num }\" class=\"footnote\""
697 content = "<sup>#{ num }</sup> #{ content }"
697 content = "<sup>#{ num }</sup> #{ content }"
698 atts = shelve( atts ) if atts
698 atts = shelve( atts ) if atts
699 "\t<p#{ atts }>#{ content }</p>"
699 "\t<p#{ atts }>#{ content }</p>"
700 end
700 end
701
701
702 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
702 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
703
703
704 def block_textile_prefix( text )
704 def block_textile_prefix( text )
705 if text =~ BLOCK_RE
705 if text =~ BLOCK_RE
706 tag,tagpre,num,atts,cite,content = $~[1..6]
706 tag,tagpre,num,atts,cite,content = $~[1..6]
707 atts = pba( atts )
707 atts = pba( atts )
708
708
709 # pass to prefix handler
709 # pass to prefix handler
710 replacement = nil
710 if respond_to? "textile_#{ tag }", true
711 if respond_to? "textile_#{ tag }", true
711 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
712 replacement = method( "textile_#{ tag }" ).call( tag, atts, cite, content )
712 elsif respond_to? "textile_#{ tagpre }_", true
713 elsif respond_to? "textile_#{ tagpre }_", true
713 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
714 replacement = method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content )
714 end
715 end
716 text.gsub!( $& ) { replacement } if replacement
715 end
717 end
716 end
718 end
717
719
718 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
720 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
719 def block_markdown_setext( text )
721 def block_markdown_setext( text )
720 if text =~ SETEXT_RE
722 if text =~ SETEXT_RE
721 tag = if $2 == "="; "h1"; else; "h2"; end
723 tag = if $2 == "="; "h1"; else; "h2"; end
722 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
724 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
723 blocks cont
725 blocks cont
724 text.replace( blk + cont )
726 text.replace( blk + cont )
725 end
727 end
726 end
728 end
727
729
728 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
730 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
729 [ ]*
731 [ ]*
730 (.+?) # $2 = Header text
732 (.+?) # $2 = Header text
731 [ ]*
733 [ ]*
732 \#* # optional closing #'s (not counted)
734 \#* # optional closing #'s (not counted)
733 $/x
735 $/x
734 def block_markdown_atx( text )
736 def block_markdown_atx( text )
735 if text =~ ATX_RE
737 if text =~ ATX_RE
736 tag = "h#{ $1.length }"
738 tag = "h#{ $1.length }"
737 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
739 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
738 blocks cont
740 blocks cont
739 text.replace( blk + cont )
741 text.replace( blk + cont )
740 end
742 end
741 end
743 end
742
744
743 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
745 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
744
746
745 def block_markdown_bq( text )
747 def block_markdown_bq( text )
746 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
748 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
747 blk.gsub!( /^ *> ?/, '' )
749 blk.gsub!( /^ *> ?/, '' )
748 flush_left blk
750 flush_left blk
749 blocks blk
751 blocks blk
750 blk.gsub!( /^(\S)/, "\t\\1" )
752 blk.gsub!( /^(\S)/, "\t\\1" )
751 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
753 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
752 end
754 end
753 end
755 end
754
756
755 MARKDOWN_RULE_RE = /^(#{
757 MARKDOWN_RULE_RE = /^(#{
756 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
758 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
757 })$/
759 })$/
758
760
759 def block_markdown_rule( text )
761 def block_markdown_rule( text )
760 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
762 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
761 "<hr />"
763 "<hr />"
762 end
764 end
763 end
765 end
764
766
765 # XXX TODO XXX
767 # XXX TODO XXX
766 def block_markdown_lists( text )
768 def block_markdown_lists( text )
767 end
769 end
768
770
769 def inline_textile_span( text )
771 def inline_textile_span( text )
770 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
772 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
771 text.gsub!( qtag_re ) do |m|
773 text.gsub!( qtag_re ) do |m|
772
774
773 case rtype
775 case rtype
774 when :limit
776 when :limit
775 sta,oqs,qtag,content,oqa = $~[1..6]
777 sta,oqs,qtag,content,oqa = $~[1..6]
776 atts = nil
778 atts = nil
777 if content =~ /^(#{C})(.+)$/
779 if content =~ /^(#{C})(.+)$/
778 atts, content = $~[1..2]
780 atts, content = $~[1..2]
779 end
781 end
780 else
782 else
781 qtag,atts,cite,content = $~[1..4]
783 qtag,atts,cite,content = $~[1..4]
782 sta = ''
784 sta = ''
783 end
785 end
784 atts = pba( atts )
786 atts = pba( atts )
785 atts = shelve( atts ) if atts
787 atts = shelve( atts ) if atts
786
788
787 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
789 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
788
790
789 end
791 end
790 end
792 end
791 end
793 end
792
794
793 LINK_RE = /
795 LINK_RE = /
794 (
796 (
795 ([\s\[{(]|[#{PUNCT}])? # $pre
797 ([\s\[{(]|[#{PUNCT}])? # $pre
796 " # start
798 " # start
797 (#{C}) # $atts
799 (#{C}) # $atts
798 ([^"\n]+?) # $text
800 ([^"\n]+?) # $text
799 \s?
801 \s?
800 (?:\(([^)]+?)\)(?="))? # $title
802 (?:\(([^)]+?)\)(?="))? # $title
801 ":
803 ":
802 ( # $url
804 ( # $url
803 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
805 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
804 [\w\/]\S+?
806 [\w\/]\S+?
805 )
807 )
806 (\/)? # $slash
808 (\/)? # $slash
807 ([^\w\=\/;\(\)]*?) # $post
809 ([^\w\=\/;\(\)]*?) # $post
808 )
810 )
809 (?=<|\s|$)
811 (?=<|\s|$)
810 /x
812 /x
811 #"
813 #"
812 def inline_textile_link( text )
814 def inline_textile_link( text )
813 text.gsub!( LINK_RE ) do |m|
815 text.gsub!( LINK_RE ) do |m|
814 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
816 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
815 if text.include?('<br />')
817 if text.include?('<br />')
816 all
818 all
817 else
819 else
818 url, url_title = check_refs( url )
820 url, url_title = check_refs( url )
819 title ||= url_title
821 title ||= url_title
820
822
821 # Idea below : an URL with unbalanced parethesis and
823 # Idea below : an URL with unbalanced parethesis and
822 # ending by ')' is put into external parenthesis
824 # ending by ')' is put into external parenthesis
823 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
825 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
824 url=url[0..-2] # discard closing parenth from url
826 url=url[0..-2] # discard closing parenth from url
825 post = ")"+post # add closing parenth to post
827 post = ")"+post # add closing parenth to post
826 end
828 end
827 atts = pba( atts )
829 atts = pba( atts )
828 atts = " href=\"#{ htmlesc url }#{ slash }\"#{ atts }"
830 atts = " href=\"#{ htmlesc url }#{ slash }\"#{ atts }"
829 atts << " title=\"#{ htmlesc title }\"" if title
831 atts << " title=\"#{ htmlesc title }\"" if title
830 atts = shelve( atts ) if atts
832 atts = shelve( atts ) if atts
831
833
832 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
834 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
833
835
834 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
836 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
835 end
837 end
836 end
838 end
837 end
839 end
838
840
839 MARKDOWN_REFLINK_RE = /
841 MARKDOWN_REFLINK_RE = /
840 \[([^\[\]]+)\] # $text
842 \[([^\[\]]+)\] # $text
841 [ ]? # opt. space
843 [ ]? # opt. space
842 (?:\n[ ]*)? # one optional newline followed by spaces
844 (?:\n[ ]*)? # one optional newline followed by spaces
843 \[(.*?)\] # $id
845 \[(.*?)\] # $id
844 /x
846 /x
845
847
846 def inline_markdown_reflink( text )
848 def inline_markdown_reflink( text )
847 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
849 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
848 text, id = $~[1..2]
850 text, id = $~[1..2]
849
851
850 if id.empty?
852 if id.empty?
851 url, title = check_refs( text )
853 url, title = check_refs( text )
852 else
854 else
853 url, title = check_refs( id )
855 url, title = check_refs( id )
854 end
856 end
855
857
856 atts = " href=\"#{ url }\""
858 atts = " href=\"#{ url }\""
857 atts << " title=\"#{ title }\"" if title
859 atts << " title=\"#{ title }\"" if title
858 atts = shelve( atts )
860 atts = shelve( atts )
859
861
860 "<a#{ atts }>#{ text }</a>"
862 "<a#{ atts }>#{ text }</a>"
861 end
863 end
862 end
864 end
863
865
864 MARKDOWN_LINK_RE = /
866 MARKDOWN_LINK_RE = /
865 \[([^\[\]]+)\] # $text
867 \[([^\[\]]+)\] # $text
866 \( # open paren
868 \( # open paren
867 [ \t]* # opt space
869 [ \t]* # opt space
868 <?(.+?)>? # $href
870 <?(.+?)>? # $href
869 [ \t]* # opt space
871 [ \t]* # opt space
870 (?: # whole title
872 (?: # whole title
871 (['"]) # $quote
873 (['"]) # $quote
872 (.*?) # $title
874 (.*?) # $title
873 \3 # matching quote
875 \3 # matching quote
874 )? # title is optional
876 )? # title is optional
875 \)
877 \)
876 /x
878 /x
877
879
878 def inline_markdown_link( text )
880 def inline_markdown_link( text )
879 text.gsub!( MARKDOWN_LINK_RE ) do |m|
881 text.gsub!( MARKDOWN_LINK_RE ) do |m|
880 text, url, quote, title = $~[1..4]
882 text, url, quote, title = $~[1..4]
881
883
882 atts = " href=\"#{ url }\""
884 atts = " href=\"#{ url }\""
883 atts << " title=\"#{ title }\"" if title
885 atts << " title=\"#{ title }\"" if title
884 atts = shelve( atts )
886 atts = shelve( atts )
885
887
886 "<a#{ atts }>#{ text }</a>"
888 "<a#{ atts }>#{ text }</a>"
887 end
889 end
888 end
890 end
889
891
890 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
892 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
891 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
893 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
892
894
893 def refs( text )
895 def refs( text )
894 @rules.each do |rule_name|
896 @rules.each do |rule_name|
895 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
897 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
896 end
898 end
897 end
899 end
898
900
899 def refs_textile( text )
901 def refs_textile( text )
900 text.gsub!( TEXTILE_REFS_RE ) do |m|
902 text.gsub!( TEXTILE_REFS_RE ) do |m|
901 flag, url = $~[2..3]
903 flag, url = $~[2..3]
902 @urlrefs[flag.downcase] = [url, nil]
904 @urlrefs[flag.downcase] = [url, nil]
903 nil
905 nil
904 end
906 end
905 end
907 end
906
908
907 def refs_markdown( text )
909 def refs_markdown( text )
908 text.gsub!( MARKDOWN_REFS_RE ) do |m|
910 text.gsub!( MARKDOWN_REFS_RE ) do |m|
909 flag, url = $~[2..3]
911 flag, url = $~[2..3]
910 title = $~[6]
912 title = $~[6]
911 @urlrefs[flag.downcase] = [url, title]
913 @urlrefs[flag.downcase] = [url, title]
912 nil
914 nil
913 end
915 end
914 end
916 end
915
917
916 def check_refs( text )
918 def check_refs( text )
917 ret = @urlrefs[text.downcase] if text
919 ret = @urlrefs[text.downcase] if text
918 ret || [text, nil]
920 ret || [text, nil]
919 end
921 end
920
922
921 IMAGE_RE = /
923 IMAGE_RE = /
922 (>|\s|^) # start of line?
924 (>|\s|^) # start of line?
923 \! # opening
925 \! # opening
924 (\<|\=|\>)? # optional alignment atts
926 (\<|\=|\>)? # optional alignment atts
925 (#{C}) # optional style,class atts
927 (#{C}) # optional style,class atts
926 (?:\. )? # optional dot-space
928 (?:\. )? # optional dot-space
927 ([^\s(!]+?) # presume this is the src
929 ([^\s(!]+?) # presume this is the src
928 \s? # optional space
930 \s? # optional space
929 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
931 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
930 \! # closing
932 \! # closing
931 (?::#{ HYPERLINK })? # optional href
933 (?::#{ HYPERLINK })? # optional href
932 /x
934 /x
933
935
934 def inline_textile_image( text )
936 def inline_textile_image( text )
935 text.gsub!( IMAGE_RE ) do |m|
937 text.gsub!( IMAGE_RE ) do |m|
936 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
938 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
937 htmlesc title
939 htmlesc title
938 atts = pba( atts )
940 atts = pba( atts )
939 atts = " src=\"#{ url }\"#{ atts }"
941 atts = " src=\"#{ url }\"#{ atts }"
940 atts << " title=\"#{ title }\"" if title
942 atts << " title=\"#{ title }\"" if title
941 atts << " alt=\"#{ title }\""
943 atts << " alt=\"#{ title }\""
942 # size = @getimagesize($url);
944 # size = @getimagesize($url);
943 # if($size) $atts.= " $size[3]";
945 # if($size) $atts.= " $size[3]";
944
946
945 href, alt_title = check_refs( href ) if href
947 href, alt_title = check_refs( href ) if href
946 url, url_title = check_refs( url )
948 url, url_title = check_refs( url )
947
949
948 out = ''
950 out = ''
949 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
951 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
950 out << "<img#{ shelve( atts ) } />"
952 out << "<img#{ shelve( atts ) } />"
951 out << "</a>#{ href_a1 }#{ href_a2 }" if href
953 out << "</a>#{ href_a1 }#{ href_a2 }" if href
952
954
953 if algn
955 if algn
954 algn = h_align( algn )
956 algn = h_align( algn )
955 if stln == "<p>"
957 if stln == "<p>"
956 out = "<p style=\"float:#{ algn }\">#{ out }"
958 out = "<p style=\"float:#{ algn }\">#{ out }"
957 else
959 else
958 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
960 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
959 end
961 end
960 else
962 else
961 out = stln + out
963 out = stln + out
962 end
964 end
963
965
964 out
966 out
965 end
967 end
966 end
968 end
967
969
968 def shelve( val )
970 def shelve( val )
969 @shelf << val
971 @shelf << val
970 " :redsh##{ @shelf.length }:"
972 " :redsh##{ @shelf.length }:"
971 end
973 end
972
974
973 def retrieve( text )
975 def retrieve( text )
974 @shelf.each_with_index do |r, i|
976 @shelf.each_with_index do |r, i|
975 text.gsub!( " :redsh##{ i + 1 }:", r )
977 text.gsub!( " :redsh##{ i + 1 }:", r )
976 end
978 end
977 end
979 end
978
980
979 def incoming_entities( text )
981 def incoming_entities( text )
980 ## turn any incoming ampersands into a dummy character for now.
982 ## turn any incoming ampersands into a dummy character for now.
981 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
983 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
982 ## implying an incoming html entity, to be skipped
984 ## implying an incoming html entity, to be skipped
983
985
984 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
986 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
985 end
987 end
986
988
987 def no_textile( text )
989 def no_textile( text )
988 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
990 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
989 '\1<notextile>\2</notextile>\3' )
991 '\1<notextile>\2</notextile>\3' )
990 text.gsub!( /^ *==([^=]+.*?)==/m,
992 text.gsub!( /^ *==([^=]+.*?)==/m,
991 '\1<notextile>\2</notextile>\3' )
993 '\1<notextile>\2</notextile>\3' )
992 end
994 end
993
995
994 def clean_white_space( text )
996 def clean_white_space( text )
995 # normalize line breaks
997 # normalize line breaks
996 text.gsub!( /\r\n/, "\n" )
998 text.gsub!( /\r\n/, "\n" )
997 text.gsub!( /\r/, "\n" )
999 text.gsub!( /\r/, "\n" )
998 text.gsub!( /\t/, ' ' )
1000 text.gsub!( /\t/, ' ' )
999 text.gsub!( /^ +$/, '' )
1001 text.gsub!( /^ +$/, '' )
1000 text.gsub!( /\n{3,}/, "\n\n" )
1002 text.gsub!( /\n{3,}/, "\n\n" )
1001 text.gsub!( /"$/, "\" " )
1003 text.gsub!( /"$/, "\" " )
1002
1004
1003 # if entire document is indented, flush
1005 # if entire document is indented, flush
1004 # to the left side
1006 # to the left side
1005 flush_left text
1007 flush_left text
1006 end
1008 end
1007
1009
1008 def flush_left( text )
1010 def flush_left( text )
1009 indt = 0
1011 indt = 0
1010 if text =~ /^ /
1012 if text =~ /^ /
1011 while text !~ /^ {#{indt}}\S/
1013 while text !~ /^ {#{indt}}\S/
1012 indt += 1
1014 indt += 1
1013 end unless text.empty?
1015 end unless text.empty?
1014 if indt.nonzero?
1016 if indt.nonzero?
1015 text.gsub!( /^ {#{indt}}/, '' )
1017 text.gsub!( /^ {#{indt}}/, '' )
1016 end
1018 end
1017 end
1019 end
1018 end
1020 end
1019
1021
1020 def footnote_ref( text )
1022 def footnote_ref( text )
1021 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1023 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1022 '<sup><a href="#fn\1">\1</a></sup>\2' )
1024 '<sup><a href="#fn\1">\1</a></sup>\2' )
1023 end
1025 end
1024
1026
1025 OFFTAGS = /(code|pre|kbd|notextile)/
1027 OFFTAGS = /(code|pre|kbd|notextile)/
1026 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1028 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1027 OFFTAG_OPEN = /<#{ OFFTAGS }/
1029 OFFTAG_OPEN = /<#{ OFFTAGS }/
1028 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1030 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1029 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1031 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1030 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1032 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1031
1033
1032 def glyphs_textile( text, level = 0 )
1034 def glyphs_textile( text, level = 0 )
1033 if text !~ HASTAG_MATCH
1035 if text !~ HASTAG_MATCH
1034 pgl text
1036 pgl text
1035 footnote_ref text
1037 footnote_ref text
1036 else
1038 else
1037 codepre = 0
1039 codepre = 0
1038 text.gsub!( ALLTAG_MATCH ) do |line|
1040 text.gsub!( ALLTAG_MATCH ) do |line|
1039 ## matches are off if we're between <code>, <pre> etc.
1041 ## matches are off if we're between <code>, <pre> etc.
1040 if $1
1042 if $1
1041 if line =~ OFFTAG_OPEN
1043 if line =~ OFFTAG_OPEN
1042 codepre += 1
1044 codepre += 1
1043 elsif line =~ OFFTAG_CLOSE
1045 elsif line =~ OFFTAG_CLOSE
1044 codepre -= 1
1046 codepre -= 1
1045 codepre = 0 if codepre < 0
1047 codepre = 0 if codepre < 0
1046 end
1048 end
1047 elsif codepre.zero?
1049 elsif codepre.zero?
1048 glyphs_textile( line, level + 1 )
1050 glyphs_textile( line, level + 1 )
1049 else
1051 else
1050 htmlesc( line, :NoQuotes )
1052 htmlesc( line, :NoQuotes )
1051 end
1053 end
1052 # p [level, codepre, line]
1054 # p [level, codepre, line]
1053
1055
1054 line
1056 line
1055 end
1057 end
1056 end
1058 end
1057 end
1059 end
1058
1060
1059 def rip_offtags( text, escape_aftertag=true )
1061 def rip_offtags( text, escape_aftertag=true )
1060 if text =~ /<.*>/
1062 if text =~ /<.*>/
1061 ## strip and encode <pre> content
1063 ## strip and encode <pre> content
1062 codepre, used_offtags = 0, {}
1064 codepre, used_offtags = 0, {}
1063 text.gsub!( OFFTAG_MATCH ) do |line|
1065 text.gsub!( OFFTAG_MATCH ) do |line|
1064 if $3
1066 if $3
1065 first, offtag, aftertag = $3, $4, $5
1067 first, offtag, aftertag = $3, $4, $5
1066 codepre += 1
1068 codepre += 1
1067 used_offtags[offtag] = true
1069 used_offtags[offtag] = true
1068 if codepre - used_offtags.length > 0
1070 if codepre - used_offtags.length > 0
1069 htmlesc( line, :NoQuotes )
1071 htmlesc( line, :NoQuotes )
1070 @pre_list.last << line
1072 @pre_list.last << line
1071 line = ""
1073 line = ""
1072 else
1074 else
1073 ### htmlesc is disabled between CODE tags which will be parsed with highlighter
1075 ### htmlesc is disabled between CODE tags which will be parsed with highlighter
1074 ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
1076 ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
1075 ### NB: some changes were made not to use $N variables, because we use "match"
1077 ### NB: some changes were made not to use $N variables, because we use "match"
1076 ### and it breaks following lines
1078 ### and it breaks following lines
1077 htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/)
1079 htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/)
1078 line = "<redpre##{ @pre_list.length }>"
1080 line = "<redpre##{ @pre_list.length }>"
1079 first.match(/<#{ OFFTAGS }([^>]*)>/)
1081 first.match(/<#{ OFFTAGS }([^>]*)>/)
1080 tag = $1
1082 tag = $1
1081 $2.to_s.match(/(class\=("[^"]+"|'[^']+'))/i)
1083 $2.to_s.match(/(class\=("[^"]+"|'[^']+'))/i)
1082 tag << " #{$1}" if $1
1084 tag << " #{$1}" if $1
1083 @pre_list << "<#{ tag }>#{ aftertag }"
1085 @pre_list << "<#{ tag }>#{ aftertag }"
1084 end
1086 end
1085 elsif $1 and codepre > 0
1087 elsif $1 and codepre > 0
1086 if codepre - used_offtags.length > 0
1088 if codepre - used_offtags.length > 0
1087 htmlesc( line, :NoQuotes )
1089 htmlesc( line, :NoQuotes )
1088 @pre_list.last << line
1090 @pre_list.last << line
1089 line = ""
1091 line = ""
1090 end
1092 end
1091 codepre -= 1 unless codepre.zero?
1093 codepre -= 1 unless codepre.zero?
1092 used_offtags = {} if codepre.zero?
1094 used_offtags = {} if codepre.zero?
1093 end
1095 end
1094 line
1096 line
1095 end
1097 end
1096 end
1098 end
1097 text
1099 text
1098 end
1100 end
1099
1101
1100 def smooth_offtags( text )
1102 def smooth_offtags( text )
1101 unless @pre_list.empty?
1103 unless @pre_list.empty?
1102 ## replace <pre> content
1104 ## replace <pre> content
1103 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1105 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1104 end
1106 end
1105 end
1107 end
1106
1108
1107 def inline( text )
1109 def inline( text )
1108 [/^inline_/, /^glyphs_/].each do |meth_re|
1110 [/^inline_/, /^glyphs_/].each do |meth_re|
1109 @rules.each do |rule_name|
1111 @rules.each do |rule_name|
1110 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1112 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1111 end
1113 end
1112 end
1114 end
1113 end
1115 end
1114
1116
1115 def h_align( text )
1117 def h_align( text )
1116 H_ALGN_VALS[text]
1118 H_ALGN_VALS[text]
1117 end
1119 end
1118
1120
1119 def v_align( text )
1121 def v_align( text )
1120 V_ALGN_VALS[text]
1122 V_ALGN_VALS[text]
1121 end
1123 end
1122
1124
1123 def textile_popup_help( name, windowW, windowH )
1125 def textile_popup_help( name, windowW, windowH )
1124 ' <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 />'
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 />'
1125 end
1127 end
1126
1128
1127 # HTML cleansing stuff
1129 # HTML cleansing stuff
1128 BASIC_TAGS = {
1130 BASIC_TAGS = {
1129 'a' => ['href', 'title'],
1131 'a' => ['href', 'title'],
1130 'img' => ['src', 'alt', 'title'],
1132 'img' => ['src', 'alt', 'title'],
1131 'br' => [],
1133 'br' => [],
1132 'i' => nil,
1134 'i' => nil,
1133 'u' => nil,
1135 'u' => nil,
1134 'b' => nil,
1136 'b' => nil,
1135 'pre' => nil,
1137 'pre' => nil,
1136 'kbd' => nil,
1138 'kbd' => nil,
1137 'code' => ['lang'],
1139 'code' => ['lang'],
1138 'cite' => nil,
1140 'cite' => nil,
1139 'strong' => nil,
1141 'strong' => nil,
1140 'em' => nil,
1142 'em' => nil,
1141 'ins' => nil,
1143 'ins' => nil,
1142 'sup' => nil,
1144 'sup' => nil,
1143 'sub' => nil,
1145 'sub' => nil,
1144 'del' => nil,
1146 'del' => nil,
1145 'table' => nil,
1147 'table' => nil,
1146 'tr' => nil,
1148 'tr' => nil,
1147 'td' => ['colspan', 'rowspan'],
1149 'td' => ['colspan', 'rowspan'],
1148 'th' => nil,
1150 'th' => nil,
1149 'ol' => nil,
1151 'ol' => nil,
1150 'ul' => nil,
1152 'ul' => nil,
1151 'li' => nil,
1153 'li' => nil,
1152 'p' => nil,
1154 'p' => nil,
1153 'h1' => nil,
1155 'h1' => nil,
1154 'h2' => nil,
1156 'h2' => nil,
1155 'h3' => nil,
1157 'h3' => nil,
1156 'h4' => nil,
1158 'h4' => nil,
1157 'h5' => nil,
1159 'h5' => nil,
1158 'h6' => nil,
1160 'h6' => nil,
1159 'blockquote' => ['cite']
1161 'blockquote' => ['cite']
1160 }
1162 }
1161
1163
1162 def clean_html( text, tags = BASIC_TAGS )
1164 def clean_html( text, tags = BASIC_TAGS )
1163 text.gsub!( /<!\[CDATA\[/, '' )
1165 text.gsub!( /<!\[CDATA\[/, '' )
1164 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1166 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1165 raw = $~
1167 raw = $~
1166 tag = raw[2].downcase
1168 tag = raw[2].downcase
1167 if tags.has_key? tag
1169 if tags.has_key? tag
1168 pcs = [tag]
1170 pcs = [tag]
1169 tags[tag].each do |prop|
1171 tags[tag].each do |prop|
1170 ['"', "'", ''].each do |q|
1172 ['"', "'", ''].each do |q|
1171 q2 = ( q != '' ? q : '\s' )
1173 q2 = ( q != '' ? q : '\s' )
1172 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1174 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1173 attrv = $1
1175 attrv = $1
1174 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1176 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1175 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1177 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1176 break
1178 break
1177 end
1179 end
1178 end
1180 end
1179 end if tags[tag]
1181 end if tags[tag]
1180 "<#{raw[1]}#{pcs.join " "}>"
1182 "<#{raw[1]}#{pcs.join " "}>"
1181 else
1183 else
1182 " "
1184 " "
1183 end
1185 end
1184 end
1186 end
1185 end
1187 end
1186
1188
1187 ALLOWED_TAGS = %w(redpre pre code notextile)
1189 ALLOWED_TAGS = %w(redpre pre code notextile)
1188
1190
1189 def escape_html_tags(text)
1191 def escape_html_tags(text)
1190 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1192 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1191 end
1193 end
1192 end
1194 end
1193
1195
@@ -1,96 +1,102
1 # Redmine - project management software
1 # Redmine - project management software
2 # Copyright (C) 2006-2010 Jean-Philippe Lang
2 # Copyright (C) 2006-2010 Jean-Philippe Lang
3 #
3 #
4 # This program is free software; you can redistribute it and/or
4 # This program is free software; you can redistribute it and/or
5 # modify it under the terms of the GNU General Public License
5 # modify it under the terms of the GNU General Public License
6 # as published by the Free Software Foundation; either version 2
6 # as published by the Free Software Foundation; either version 2
7 # of the License, or (at your option) any later version.
7 # of the License, or (at your option) any later version.
8 #
8 #
9 # This program is distributed in the hope that it will be useful,
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU General Public License for more details.
12 # GNU General Public License for more details.
13 #
13 #
14 # You should have received a copy of the GNU General Public License
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
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.
16 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
17
17
18 require File.expand_path('../../../../../test_helper', __FILE__)
18 require File.expand_path('../../../../../test_helper', __FILE__)
19
19
20 class Redmine::WikiFormatting::TextileFormatterTest < HelperTestCase
20 class Redmine::WikiFormatting::TextileFormatterTest < HelperTestCase
21
21
22 def setup
22 def setup
23 @formatter = Redmine::WikiFormatting::Textile::Formatter
23 @formatter = Redmine::WikiFormatting::Textile::Formatter
24 end
24 end
25
25
26 MODIFIERS = {
26 MODIFIERS = {
27 "*" => 'strong', # bold
27 "*" => 'strong', # bold
28 "_" => 'em', # italic
28 "_" => 'em', # italic
29 "+" => 'ins', # underline
29 "+" => 'ins', # underline
30 "-" => 'del', # deleted
30 "-" => 'del', # deleted
31 "^" => 'sup', # superscript
31 "^" => 'sup', # superscript
32 "~" => 'sub' # subscript
32 "~" => 'sub' # subscript
33 }
33 }
34
34
35 def test_modifiers
35 def test_modifiers
36 assert_html_output(
36 assert_html_output(
37 '*bold*' => '<strong>bold</strong>',
37 '*bold*' => '<strong>bold</strong>',
38 'before *bold*' => 'before <strong>bold</strong>',
38 'before *bold*' => 'before <strong>bold</strong>',
39 '*bold* after' => '<strong>bold</strong> after',
39 '*bold* after' => '<strong>bold</strong> after',
40 '*two words*' => '<strong>two words</strong>',
40 '*two words*' => '<strong>two words</strong>',
41 '*two*words*' => '<strong>two*words</strong>',
41 '*two*words*' => '<strong>two*words</strong>',
42 '*two * words*' => '<strong>two * words</strong>',
42 '*two * words*' => '<strong>two * words</strong>',
43 '*two* *words*' => '<strong>two</strong> <strong>words</strong>',
43 '*two* *words*' => '<strong>two</strong> <strong>words</strong>',
44 '*(two)* *(words)*' => '<strong>(two)</strong> <strong>(words)</strong>',
44 '*(two)* *(words)*' => '<strong>(two)</strong> <strong>(words)</strong>',
45 # with class
45 # with class
46 '*(foo)two words*' => '<strong class="foo">two words</strong>'
46 '*(foo)two words*' => '<strong class="foo">two words</strong>'
47 )
47 )
48 end
48 end
49
49
50 def test_modifiers_combination
50 def test_modifiers_combination
51 MODIFIERS.each do |m1, tag1|
51 MODIFIERS.each do |m1, tag1|
52 MODIFIERS.each do |m2, tag2|
52 MODIFIERS.each do |m2, tag2|
53 next if m1 == m2
53 next if m1 == m2
54 text = "#{m2}#{m1}Phrase modifiers#{m1}#{m2}"
54 text = "#{m2}#{m1}Phrase modifiers#{m1}#{m2}"
55 html = "<#{tag2}><#{tag1}>Phrase modifiers</#{tag1}></#{tag2}>"
55 html = "<#{tag2}><#{tag1}>Phrase modifiers</#{tag1}></#{tag2}>"
56 assert_html_output text => html
56 assert_html_output text => html
57 end
57 end
58 end
58 end
59 end
59 end
60
60
61 def test_inline_code
61 def test_inline_code
62 assert_html_output(
62 assert_html_output(
63 'this is @some code@' => 'this is <code>some code</code>',
63 'this is @some code@' => 'this is <code>some code</code>',
64 '@<Location /redmine>@' => '<code>&lt;Location /redmine&gt;</code>'
64 '@<Location /redmine>@' => '<code>&lt;Location /redmine&gt;</code>'
65 )
65 )
66 end
66 end
67
67
68 def test_escaping
68 def test_escaping
69 assert_html_output(
69 assert_html_output(
70 'this is a <script>' => 'this is a &lt;script&gt;'
70 'this is a <script>' => 'this is a &lt;script&gt;'
71 )
71 )
72 end
72 end
73
74 def test_use_of_backslashes_followed_by_numbers_in_headers
75 assert_html_output({
76 'h1. 2009\02\09' => '<h1>2009\02\09</h1>'
77 }, false)
78 end
73
79
74 def test_double_dashes_should_not_strikethrough
80 def test_double_dashes_should_not_strikethrough
75 assert_html_output(
81 assert_html_output(
76 'double -- dashes -- test' => 'double -- dashes -- test',
82 'double -- dashes -- test' => 'double -- dashes -- test',
77 'double -- *dashes* -- test' => 'double -- <strong>dashes</strong> -- test'
83 'double -- *dashes* -- test' => 'double -- <strong>dashes</strong> -- test'
78 )
84 )
79 end
85 end
80
86
81 def test_acronyms
87 def test_acronyms
82 assert_html_output(
88 assert_html_output(
83 'this is an acronym: GPL(General Public License)' => 'this is an acronym: <acronym title="General Public License">GPL</acronym>',
89 'this is an acronym: GPL(General Public License)' => 'this is an acronym: <acronym title="General Public License">GPL</acronym>',
84 '2 letters JP(Jean-Philippe) acronym' => '2 letters <acronym title="Jean-Philippe">JP</acronym> acronym',
90 '2 letters JP(Jean-Philippe) acronym' => '2 letters <acronym title="Jean-Philippe">JP</acronym> acronym',
85 'GPL(This is a double-quoted "title")' => '<acronym title="This is a double-quoted &quot;title&quot;">GPL</acronym>'
91 'GPL(This is a double-quoted "title")' => '<acronym title="This is a double-quoted &quot;title&quot;">GPL</acronym>'
86 )
92 )
87 end
93 end
88
94
89 private
95 private
90
96
91 def assert_html_output(to_test)
97 def assert_html_output(to_test, expect_paragraph = true)
92 to_test.each do |text, expected|
98 to_test.each do |text, expected|
93 assert_equal "<p>#{expected}</p>", @formatter.new(text).to_html, "Formatting the following text failed:\n===\n#{text}\n===\n"
99 assert_equal(( expect_paragraph ? "<p>#{expected}</p>" : expected ), @formatter.new(text).to_html, "Formatting the following text failed:\n===\n#{text}\n===\n")
94 end
100 end
95 end
101 end
96 end
102 end
General Comments 0
You need to be logged in to leave comments. Login now