##// END OF EJS Templates
Fixed: Double-dash results in strikethrough text (#5122)....
Jean-Philippe Lang -
r3491:533590c29cf3
parent child
Show More
@@ -1,1191 +1,1193
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\(]) # sta
384 (?!\-\-)
384 (#{QTAGS_JOIN}|) # oqs
385 (#{QTAGS_JOIN}|) # oqs
385 (#{rcq}) # qtag
386 (#{rcq}) # qtag
386 (\w|[^\s].*?[^\s]) # content
387 (\w|[^\s].*?[^\s]) # content
388 (?!\-\-)
387 #{rcq}
389 #{rcq}
388 (#{QTAGS_JOIN}|) # oqa
390 (#{QTAGS_JOIN}|) # oqa
389 (?=[[:punct:]]|\s|\)|$)/x
391 (?=[[:punct:]]|\s|\)|$)/x
390 else
392 else
391 /(#{rcq})
393 /(#{rcq})
392 (#{C})
394 (#{C})
393 (?::(\S+))?
395 (?::(\S+))?
394 (\w|[^\s\-].*?[^\s\-])
396 (\w|[^\s\-].*?[^\s\-])
395 #{rcq}/xm
397 #{rcq}/xm
396 end
398 end
397 [rc, ht, re, rtype]
399 [rc, ht, re, rtype]
398 end
400 end
399
401
400 # Elements to handle
402 # Elements to handle
401 GLYPHS = [
403 GLYPHS = [
402 # [ /([^\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
403 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
404 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
405 # [ /\'/, '&#8216;' ], # single opening
407 # [ /\'/, '&#8216;' ], # single opening
406 # [ /</, '&lt;' ], # less-than
408 # [ /</, '&lt;' ], # less-than
407 # [ />/, '&gt;' ], # greater-than
409 # [ />/, '&gt;' ], # greater-than
408 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
409 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
410 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
411 # [ /"/, '&#8220;' ], # double opening
413 # [ /"/, '&#8220;' ], # double opening
412 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
414 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
413 # [ /\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
414 # [ /(^|[^"][>\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
415 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
417 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
416 # [ /\s->\s/, ' &rarr; ' ], # right arrow
418 # [ /\s->\s/, ' &rarr; ' ], # right arrow
417 # [ /\s-\s/, ' &#8211; ' ], # en dash
419 # [ /\s-\s/, ' &#8211; ' ], # en dash
418 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
420 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
419 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
421 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
420 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
422 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
421 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
423 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
422 ]
424 ]
423
425
424 H_ALGN_VALS = {
426 H_ALGN_VALS = {
425 '<' => 'left',
427 '<' => 'left',
426 '=' => 'center',
428 '=' => 'center',
427 '>' => 'right',
429 '>' => 'right',
428 '<>' => 'justify'
430 '<>' => 'justify'
429 }
431 }
430
432
431 V_ALGN_VALS = {
433 V_ALGN_VALS = {
432 '^' => 'top',
434 '^' => 'top',
433 '-' => 'middle',
435 '-' => 'middle',
434 '~' => 'bottom'
436 '~' => 'bottom'
435 }
437 }
436
438
437 #
439 #
438 # Flexible HTML escaping
440 # Flexible HTML escaping
439 #
441 #
440 def htmlesc( str, mode=:Quotes )
442 def htmlesc( str, mode=:Quotes )
441 if str
443 if str
442 str.gsub!( '&', '&amp;' )
444 str.gsub!( '&', '&amp;' )
443 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
445 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
444 str.gsub!( "'", '&#039;' ) if mode == :Quotes
446 str.gsub!( "'", '&#039;' ) if mode == :Quotes
445 str.gsub!( '<', '&lt;')
447 str.gsub!( '<', '&lt;')
446 str.gsub!( '>', '&gt;')
448 str.gsub!( '>', '&gt;')
447 end
449 end
448 str
450 str
449 end
451 end
450
452
451 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
452 def pgl( text )
454 def pgl( text )
453 #GLYPHS.each do |re, resub, tog|
455 #GLYPHS.each do |re, resub, tog|
454 # next if tog and method( tog ).call
456 # next if tog and method( tog ).call
455 # text.gsub! re, resub
457 # text.gsub! re, resub
456 #end
458 #end
457 text.gsub!(/\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/) do |m|
459 text.gsub!(/\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/) do |m|
458 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
459 end
461 end
460 end
462 end
461
463
462 # Parses Textile attribute lists and builds an HTML attribute string
464 # Parses Textile attribute lists and builds an HTML attribute string
463 def pba( text_in, element = "" )
465 def pba( text_in, element = "" )
464
466
465 return '' unless text_in
467 return '' unless text_in
466
468
467 style = []
469 style = []
468 text = text_in.dup
470 text = text_in.dup
469 if element == 'td'
471 if element == 'td'
470 colspan = $1 if text =~ /\\(\d+)/
472 colspan = $1 if text =~ /\\(\d+)/
471 rowspan = $1 if text =~ /\/(\d+)/
473 rowspan = $1 if text =~ /\/(\d+)/
472 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
473 end
475 end
474
476
475 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
477 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
476
478
477 lang = $1 if
479 lang = $1 if
478 text.sub!( /\[([^)]+?)\]/, '' )
480 text.sub!( /\[([^)]+?)\]/, '' )
479
481
480 cls = $1 if
482 cls = $1 if
481 text.sub!( /\(([^()]+?)\)/, '' )
483 text.sub!( /\(([^()]+?)\)/, '' )
482
484
483 style << "padding-left:#{ $1.length }em;" if
485 style << "padding-left:#{ $1.length }em;" if
484 text.sub!( /([(]+)/, '' )
486 text.sub!( /([(]+)/, '' )
485
487
486 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
488 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
487
489
488 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
490 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
489
491
490 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
492 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
491
493
492 atts = ''
494 atts = ''
493 atts << " style=\"#{ style.join }\"" unless style.empty?
495 atts << " style=\"#{ style.join }\"" unless style.empty?
494 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
496 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
495 atts << " lang=\"#{ lang }\"" if lang
497 atts << " lang=\"#{ lang }\"" if lang
496 atts << " id=\"#{ id }\"" if id
498 atts << " id=\"#{ id }\"" if id
497 atts << " colspan=\"#{ colspan }\"" if colspan
499 atts << " colspan=\"#{ colspan }\"" if colspan
498 atts << " rowspan=\"#{ rowspan }\"" if rowspan
500 atts << " rowspan=\"#{ rowspan }\"" if rowspan
499
501
500 atts
502 atts
501 end
503 end
502
504
503 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
504
506
505 # Parses a Textile table block, building HTML from the result.
507 # Parses a Textile table block, building HTML from the result.
506 def block_textile_table( text )
508 def block_textile_table( text )
507 text.gsub!( TABLE_RE ) do |matches|
509 text.gsub!( TABLE_RE ) do |matches|
508
510
509 tatts, fullrow = $~[1..2]
511 tatts, fullrow = $~[1..2]
510 tatts = pba( tatts, 'table' )
512 tatts = pba( tatts, 'table' )
511 tatts = shelve( tatts ) if tatts
513 tatts = shelve( tatts ) if tatts
512 rows = []
514 rows = []
513
515
514 fullrow.each_line do |row|
516 fullrow.each_line do |row|
515 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
517 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
516 cells = []
518 cells = []
517 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
519 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
518 next if cell == '|'
520 next if cell == '|'
519 ctyp = 'd'
521 ctyp = 'd'
520 ctyp = 'h' if cell =~ /^_/
522 ctyp = 'h' if cell =~ /^_/
521
523
522 catts = ''
524 catts = ''
523 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
525 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
524
526
525 catts = shelve( catts ) if catts
527 catts = shelve( catts ) if catts
526 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
528 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
527 end
529 end
528 ratts = shelve( ratts ) if ratts
530 ratts = shelve( ratts ) if ratts
529 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
531 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
530 end
532 end
531 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
533 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
532 end
534 end
533 end
535 end
534
536
535 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
537 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
536 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
538 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
537
539
538 # Parses Textile lists and generates HTML
540 # Parses Textile lists and generates HTML
539 def block_textile_lists( text )
541 def block_textile_lists( text )
540 text.gsub!( LISTS_RE ) do |match|
542 text.gsub!( LISTS_RE ) do |match|
541 lines = match.split( /\n/ )
543 lines = match.split( /\n/ )
542 last_line = -1
544 last_line = -1
543 depth = []
545 depth = []
544 lines.each_with_index do |line, line_id|
546 lines.each_with_index do |line, line_id|
545 if line =~ LISTS_CONTENT_RE
547 if line =~ LISTS_CONTENT_RE
546 tl,atts,content = $~[1..3]
548 tl,atts,content = $~[1..3]
547 if depth.last
549 if depth.last
548 if depth.last.length > tl.length
550 if depth.last.length > tl.length
549 (depth.length - 1).downto(0) do |i|
551 (depth.length - 1).downto(0) do |i|
550 break if depth[i].length == tl.length
552 break if depth[i].length == tl.length
551 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
553 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
552 depth.pop
554 depth.pop
553 end
555 end
554 end
556 end
555 if depth.last and depth.last.length == tl.length
557 if depth.last and depth.last.length == tl.length
556 lines[line_id - 1] << '</li>'
558 lines[line_id - 1] << '</li>'
557 end
559 end
558 end
560 end
559 unless depth.last == tl
561 unless depth.last == tl
560 depth << tl
562 depth << tl
561 atts = pba( atts )
563 atts = pba( atts )
562 atts = shelve( atts ) if atts
564 atts = shelve( atts ) if atts
563 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
565 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
564 else
566 else
565 lines[line_id] = "\t\t<li>#{ content }"
567 lines[line_id] = "\t\t<li>#{ content }"
566 end
568 end
567 last_line = line_id
569 last_line = line_id
568
570
569 else
571 else
570 last_line = line_id
572 last_line = line_id
571 end
573 end
572 if line_id - last_line > 1 or line_id == lines.length - 1
574 if line_id - last_line > 1 or line_id == lines.length - 1
573 depth.delete_if do |v|
575 depth.delete_if do |v|
574 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
576 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
575 end
577 end
576 end
578 end
577 end
579 end
578 lines.join( "\n" )
580 lines.join( "\n" )
579 end
581 end
580 end
582 end
581
583
582 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
584 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
583 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
585 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
584
586
585 def block_textile_quotes( text )
587 def block_textile_quotes( text )
586 text.gsub!( QUOTES_RE ) do |match|
588 text.gsub!( QUOTES_RE ) do |match|
587 lines = match.split( /\n/ )
589 lines = match.split( /\n/ )
588 quotes = ''
590 quotes = ''
589 indent = 0
591 indent = 0
590 lines.each do |line|
592 lines.each do |line|
591 line =~ QUOTES_CONTENT_RE
593 line =~ QUOTES_CONTENT_RE
592 bq,content = $1, $2
594 bq,content = $1, $2
593 l = bq.count('>')
595 l = bq.count('>')
594 if l != indent
596 if l != indent
595 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")
596 indent = l
598 indent = l
597 end
599 end
598 quotes << (content + "\n")
600 quotes << (content + "\n")
599 end
601 end
600 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
602 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
601 quotes
603 quotes
602 end
604 end
603 end
605 end
604
606
605 CODE_RE = /(\W)
607 CODE_RE = /(\W)
606 @
608 @
607 (?:\|(\w+?)\|)?
609 (?:\|(\w+?)\|)?
608 (.+?)
610 (.+?)
609 @
611 @
610 (?=\W)/x
612 (?=\W)/x
611
613
612 def inline_textile_code( text )
614 def inline_textile_code( text )
613 text.gsub!( CODE_RE ) do |m|
615 text.gsub!( CODE_RE ) do |m|
614 before,lang,code,after = $~[1..4]
616 before,lang,code,after = $~[1..4]
615 lang = " lang=\"#{ lang }\"" if lang
617 lang = " lang=\"#{ lang }\"" if lang
616 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false )
618 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }", false )
617 end
619 end
618 end
620 end
619
621
620 def lT( text )
622 def lT( text )
621 text =~ /\#$/ ? 'o' : 'u'
623 text =~ /\#$/ ? 'o' : 'u'
622 end
624 end
623
625
624 def hard_break( text )
626 def hard_break( text )
625 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
627 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
626 end
628 end
627
629
628 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
630 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
629
631
630 def blocks( text, deep_code = false )
632 def blocks( text, deep_code = false )
631 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
633 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
632 plain = blk !~ /\A[#*> ]/
634 plain = blk !~ /\A[#*> ]/
633
635
634 # skip blocks that are complex HTML
636 # skip blocks that are complex HTML
635 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
637 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
636 blk
638 blk
637 else
639 else
638 # search for indentation levels
640 # search for indentation levels
639 blk.strip!
641 blk.strip!
640 if blk.empty?
642 if blk.empty?
641 blk
643 blk
642 else
644 else
643 code_blk = nil
645 code_blk = nil
644 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
646 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
645 flush_left iblk
647 flush_left iblk
646 blocks iblk, plain
648 blocks iblk, plain
647 iblk.gsub( /^(\S)/, "\t\\1" )
649 iblk.gsub( /^(\S)/, "\t\\1" )
648 if plain
650 if plain
649 code_blk = iblk; ""
651 code_blk = iblk; ""
650 else
652 else
651 iblk
653 iblk
652 end
654 end
653 end
655 end
654
656
655 block_applied = 0
657 block_applied = 0
656 @rules.each do |rule_name|
658 @rules.each do |rule_name|
657 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 ) )
658 end
660 end
659 if block_applied.zero?
661 if block_applied.zero?
660 if deep_code
662 if deep_code
661 blk = "\t<pre><code>#{ blk }</code></pre>"
663 blk = "\t<pre><code>#{ blk }</code></pre>"
662 else
664 else
663 blk = "\t<p>#{ blk }</p>"
665 blk = "\t<p>#{ blk }</p>"
664 end
666 end
665 end
667 end
666 # hard_break blk
668 # hard_break blk
667 blk + "\n#{ code_blk }"
669 blk + "\n#{ code_blk }"
668 end
670 end
669 end
671 end
670
672
671 end.join( "\n\n" ) )
673 end.join( "\n\n" ) )
672 end
674 end
673
675
674 def textile_bq( tag, atts, cite, content )
676 def textile_bq( tag, atts, cite, content )
675 cite, cite_title = check_refs( cite )
677 cite, cite_title = check_refs( cite )
676 cite = " cite=\"#{ cite }\"" if cite
678 cite = " cite=\"#{ cite }\"" if cite
677 atts = shelve( atts ) if atts
679 atts = shelve( atts ) if atts
678 "\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>"
679 end
681 end
680
682
681 def textile_p( tag, atts, cite, content )
683 def textile_p( tag, atts, cite, content )
682 atts = shelve( atts ) if atts
684 atts = shelve( atts ) if atts
683 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
685 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
684 end
686 end
685
687
686 alias textile_h1 textile_p
688 alias textile_h1 textile_p
687 alias textile_h2 textile_p
689 alias textile_h2 textile_p
688 alias textile_h3 textile_p
690 alias textile_h3 textile_p
689 alias textile_h4 textile_p
691 alias textile_h4 textile_p
690 alias textile_h5 textile_p
692 alias textile_h5 textile_p
691 alias textile_h6 textile_p
693 alias textile_h6 textile_p
692
694
693 def textile_fn_( tag, num, atts, cite, content )
695 def textile_fn_( tag, num, atts, cite, content )
694 atts << " id=\"fn#{ num }\" class=\"footnote\""
696 atts << " id=\"fn#{ num }\" class=\"footnote\""
695 content = "<sup>#{ num }</sup> #{ content }"
697 content = "<sup>#{ num }</sup> #{ content }"
696 atts = shelve( atts ) if atts
698 atts = shelve( atts ) if atts
697 "\t<p#{ atts }>#{ content }</p>"
699 "\t<p#{ atts }>#{ content }</p>"
698 end
700 end
699
701
700 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
702 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
701
703
702 def block_textile_prefix( text )
704 def block_textile_prefix( text )
703 if text =~ BLOCK_RE
705 if text =~ BLOCK_RE
704 tag,tagpre,num,atts,cite,content = $~[1..6]
706 tag,tagpre,num,atts,cite,content = $~[1..6]
705 atts = pba( atts )
707 atts = pba( atts )
706
708
707 # pass to prefix handler
709 # pass to prefix handler
708 if respond_to? "textile_#{ tag }", true
710 if respond_to? "textile_#{ tag }", true
709 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
711 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
710 elsif respond_to? "textile_#{ tagpre }_", true
712 elsif respond_to? "textile_#{ tagpre }_", true
711 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
713 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
712 end
714 end
713 end
715 end
714 end
716 end
715
717
716 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
718 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
717 def block_markdown_setext( text )
719 def block_markdown_setext( text )
718 if text =~ SETEXT_RE
720 if text =~ SETEXT_RE
719 tag = if $2 == "="; "h1"; else; "h2"; end
721 tag = if $2 == "="; "h1"; else; "h2"; end
720 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
722 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
721 blocks cont
723 blocks cont
722 text.replace( blk + cont )
724 text.replace( blk + cont )
723 end
725 end
724 end
726 end
725
727
726 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
728 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
727 [ ]*
729 [ ]*
728 (.+?) # $2 = Header text
730 (.+?) # $2 = Header text
729 [ ]*
731 [ ]*
730 \#* # optional closing #'s (not counted)
732 \#* # optional closing #'s (not counted)
731 $/x
733 $/x
732 def block_markdown_atx( text )
734 def block_markdown_atx( text )
733 if text =~ ATX_RE
735 if text =~ ATX_RE
734 tag = "h#{ $1.length }"
736 tag = "h#{ $1.length }"
735 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
737 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
736 blocks cont
738 blocks cont
737 text.replace( blk + cont )
739 text.replace( blk + cont )
738 end
740 end
739 end
741 end
740
742
741 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
743 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
742
744
743 def block_markdown_bq( text )
745 def block_markdown_bq( text )
744 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
746 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
745 blk.gsub!( /^ *> ?/, '' )
747 blk.gsub!( /^ *> ?/, '' )
746 flush_left blk
748 flush_left blk
747 blocks blk
749 blocks blk
748 blk.gsub!( /^(\S)/, "\t\\1" )
750 blk.gsub!( /^(\S)/, "\t\\1" )
749 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
751 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
750 end
752 end
751 end
753 end
752
754
753 MARKDOWN_RULE_RE = /^(#{
755 MARKDOWN_RULE_RE = /^(#{
754 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
756 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
755 })$/
757 })$/
756
758
757 def block_markdown_rule( text )
759 def block_markdown_rule( text )
758 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
760 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
759 "<hr />"
761 "<hr />"
760 end
762 end
761 end
763 end
762
764
763 # XXX TODO XXX
765 # XXX TODO XXX
764 def block_markdown_lists( text )
766 def block_markdown_lists( text )
765 end
767 end
766
768
767 def inline_textile_span( text )
769 def inline_textile_span( text )
768 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
770 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
769 text.gsub!( qtag_re ) do |m|
771 text.gsub!( qtag_re ) do |m|
770
772
771 case rtype
773 case rtype
772 when :limit
774 when :limit
773 sta,oqs,qtag,content,oqa = $~[1..6]
775 sta,oqs,qtag,content,oqa = $~[1..6]
774 atts = nil
776 atts = nil
775 if content =~ /^(#{C})(.+)$/
777 if content =~ /^(#{C})(.+)$/
776 atts, content = $~[1..2]
778 atts, content = $~[1..2]
777 end
779 end
778 else
780 else
779 qtag,atts,cite,content = $~[1..4]
781 qtag,atts,cite,content = $~[1..4]
780 sta = ''
782 sta = ''
781 end
783 end
782 atts = pba( atts )
784 atts = pba( atts )
783 atts = shelve( atts ) if atts
785 atts = shelve( atts ) if atts
784
786
785 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
787 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
786
788
787 end
789 end
788 end
790 end
789 end
791 end
790
792
791 LINK_RE = /
793 LINK_RE = /
792 (
794 (
793 ([\s\[{(]|[#{PUNCT}])? # $pre
795 ([\s\[{(]|[#{PUNCT}])? # $pre
794 " # start
796 " # start
795 (#{C}) # $atts
797 (#{C}) # $atts
796 ([^"\n]+?) # $text
798 ([^"\n]+?) # $text
797 \s?
799 \s?
798 (?:\(([^)]+?)\)(?="))? # $title
800 (?:\(([^)]+?)\)(?="))? # $title
799 ":
801 ":
800 ( # $url
802 ( # $url
801 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
803 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
802 [\w\/]\S+?
804 [\w\/]\S+?
803 )
805 )
804 (\/)? # $slash
806 (\/)? # $slash
805 ([^\w\=\/;\(\)]*?) # $post
807 ([^\w\=\/;\(\)]*?) # $post
806 )
808 )
807 (?=<|\s|$)
809 (?=<|\s|$)
808 /x
810 /x
809 #"
811 #"
810 def inline_textile_link( text )
812 def inline_textile_link( text )
811 text.gsub!( LINK_RE ) do |m|
813 text.gsub!( LINK_RE ) do |m|
812 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
814 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
813 if text.include?('<br />')
815 if text.include?('<br />')
814 all
816 all
815 else
817 else
816 url, url_title = check_refs( url )
818 url, url_title = check_refs( url )
817 title ||= url_title
819 title ||= url_title
818
820
819 # Idea below : an URL with unbalanced parethesis and
821 # Idea below : an URL with unbalanced parethesis and
820 # ending by ')' is put into external parenthesis
822 # ending by ')' is put into external parenthesis
821 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
823 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
822 url=url[0..-2] # discard closing parenth from url
824 url=url[0..-2] # discard closing parenth from url
823 post = ")"+post # add closing parenth to post
825 post = ")"+post # add closing parenth to post
824 end
826 end
825 atts = pba( atts )
827 atts = pba( atts )
826 atts = " href=\"#{ url }#{ slash }\"#{ atts }"
828 atts = " href=\"#{ url }#{ slash }\"#{ atts }"
827 atts << " title=\"#{ htmlesc title }\"" if title
829 atts << " title=\"#{ htmlesc title }\"" if title
828 atts = shelve( atts ) if atts
830 atts = shelve( atts ) if atts
829
831
830 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
832 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
831
833
832 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
834 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
833 end
835 end
834 end
836 end
835 end
837 end
836
838
837 MARKDOWN_REFLINK_RE = /
839 MARKDOWN_REFLINK_RE = /
838 \[([^\[\]]+)\] # $text
840 \[([^\[\]]+)\] # $text
839 [ ]? # opt. space
841 [ ]? # opt. space
840 (?:\n[ ]*)? # one optional newline followed by spaces
842 (?:\n[ ]*)? # one optional newline followed by spaces
841 \[(.*?)\] # $id
843 \[(.*?)\] # $id
842 /x
844 /x
843
845
844 def inline_markdown_reflink( text )
846 def inline_markdown_reflink( text )
845 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
847 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
846 text, id = $~[1..2]
848 text, id = $~[1..2]
847
849
848 if id.empty?
850 if id.empty?
849 url, title = check_refs( text )
851 url, title = check_refs( text )
850 else
852 else
851 url, title = check_refs( id )
853 url, title = check_refs( id )
852 end
854 end
853
855
854 atts = " href=\"#{ url }\""
856 atts = " href=\"#{ url }\""
855 atts << " title=\"#{ title }\"" if title
857 atts << " title=\"#{ title }\"" if title
856 atts = shelve( atts )
858 atts = shelve( atts )
857
859
858 "<a#{ atts }>#{ text }</a>"
860 "<a#{ atts }>#{ text }</a>"
859 end
861 end
860 end
862 end
861
863
862 MARKDOWN_LINK_RE = /
864 MARKDOWN_LINK_RE = /
863 \[([^\[\]]+)\] # $text
865 \[([^\[\]]+)\] # $text
864 \( # open paren
866 \( # open paren
865 [ \t]* # opt space
867 [ \t]* # opt space
866 <?(.+?)>? # $href
868 <?(.+?)>? # $href
867 [ \t]* # opt space
869 [ \t]* # opt space
868 (?: # whole title
870 (?: # whole title
869 (['"]) # $quote
871 (['"]) # $quote
870 (.*?) # $title
872 (.*?) # $title
871 \3 # matching quote
873 \3 # matching quote
872 )? # title is optional
874 )? # title is optional
873 \)
875 \)
874 /x
876 /x
875
877
876 def inline_markdown_link( text )
878 def inline_markdown_link( text )
877 text.gsub!( MARKDOWN_LINK_RE ) do |m|
879 text.gsub!( MARKDOWN_LINK_RE ) do |m|
878 text, url, quote, title = $~[1..4]
880 text, url, quote, title = $~[1..4]
879
881
880 atts = " href=\"#{ url }\""
882 atts = " href=\"#{ url }\""
881 atts << " title=\"#{ title }\"" if title
883 atts << " title=\"#{ title }\"" if title
882 atts = shelve( atts )
884 atts = shelve( atts )
883
885
884 "<a#{ atts }>#{ text }</a>"
886 "<a#{ atts }>#{ text }</a>"
885 end
887 end
886 end
888 end
887
889
888 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
890 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
889 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
891 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
890
892
891 def refs( text )
893 def refs( text )
892 @rules.each do |rule_name|
894 @rules.each do |rule_name|
893 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
895 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
894 end
896 end
895 end
897 end
896
898
897 def refs_textile( text )
899 def refs_textile( text )
898 text.gsub!( TEXTILE_REFS_RE ) do |m|
900 text.gsub!( TEXTILE_REFS_RE ) do |m|
899 flag, url = $~[2..3]
901 flag, url = $~[2..3]
900 @urlrefs[flag.downcase] = [url, nil]
902 @urlrefs[flag.downcase] = [url, nil]
901 nil
903 nil
902 end
904 end
903 end
905 end
904
906
905 def refs_markdown( text )
907 def refs_markdown( text )
906 text.gsub!( MARKDOWN_REFS_RE ) do |m|
908 text.gsub!( MARKDOWN_REFS_RE ) do |m|
907 flag, url = $~[2..3]
909 flag, url = $~[2..3]
908 title = $~[6]
910 title = $~[6]
909 @urlrefs[flag.downcase] = [url, title]
911 @urlrefs[flag.downcase] = [url, title]
910 nil
912 nil
911 end
913 end
912 end
914 end
913
915
914 def check_refs( text )
916 def check_refs( text )
915 ret = @urlrefs[text.downcase] if text
917 ret = @urlrefs[text.downcase] if text
916 ret || [text, nil]
918 ret || [text, nil]
917 end
919 end
918
920
919 IMAGE_RE = /
921 IMAGE_RE = /
920 (>|\s|^) # start of line?
922 (>|\s|^) # start of line?
921 \! # opening
923 \! # opening
922 (\<|\=|\>)? # optional alignment atts
924 (\<|\=|\>)? # optional alignment atts
923 (#{C}) # optional style,class atts
925 (#{C}) # optional style,class atts
924 (?:\. )? # optional dot-space
926 (?:\. )? # optional dot-space
925 ([^\s(!]+?) # presume this is the src
927 ([^\s(!]+?) # presume this is the src
926 \s? # optional space
928 \s? # optional space
927 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
929 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
928 \! # closing
930 \! # closing
929 (?::#{ HYPERLINK })? # optional href
931 (?::#{ HYPERLINK })? # optional href
930 /x
932 /x
931
933
932 def inline_textile_image( text )
934 def inline_textile_image( text )
933 text.gsub!( IMAGE_RE ) do |m|
935 text.gsub!( IMAGE_RE ) do |m|
934 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
936 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
935 htmlesc title
937 htmlesc title
936 atts = pba( atts )
938 atts = pba( atts )
937 atts = " src=\"#{ url }\"#{ atts }"
939 atts = " src=\"#{ url }\"#{ atts }"
938 atts << " title=\"#{ title }\"" if title
940 atts << " title=\"#{ title }\"" if title
939 atts << " alt=\"#{ title }\""
941 atts << " alt=\"#{ title }\""
940 # size = @getimagesize($url);
942 # size = @getimagesize($url);
941 # if($size) $atts.= " $size[3]";
943 # if($size) $atts.= " $size[3]";
942
944
943 href, alt_title = check_refs( href ) if href
945 href, alt_title = check_refs( href ) if href
944 url, url_title = check_refs( url )
946 url, url_title = check_refs( url )
945
947
946 out = ''
948 out = ''
947 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
949 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
948 out << "<img#{ shelve( atts ) } />"
950 out << "<img#{ shelve( atts ) } />"
949 out << "</a>#{ href_a1 }#{ href_a2 }" if href
951 out << "</a>#{ href_a1 }#{ href_a2 }" if href
950
952
951 if algn
953 if algn
952 algn = h_align( algn )
954 algn = h_align( algn )
953 if stln == "<p>"
955 if stln == "<p>"
954 out = "<p style=\"float:#{ algn }\">#{ out }"
956 out = "<p style=\"float:#{ algn }\">#{ out }"
955 else
957 else
956 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
958 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
957 end
959 end
958 else
960 else
959 out = stln + out
961 out = stln + out
960 end
962 end
961
963
962 out
964 out
963 end
965 end
964 end
966 end
965
967
966 def shelve( val )
968 def shelve( val )
967 @shelf << val
969 @shelf << val
968 " :redsh##{ @shelf.length }:"
970 " :redsh##{ @shelf.length }:"
969 end
971 end
970
972
971 def retrieve( text )
973 def retrieve( text )
972 @shelf.each_with_index do |r, i|
974 @shelf.each_with_index do |r, i|
973 text.gsub!( " :redsh##{ i + 1 }:", r )
975 text.gsub!( " :redsh##{ i + 1 }:", r )
974 end
976 end
975 end
977 end
976
978
977 def incoming_entities( text )
979 def incoming_entities( text )
978 ## turn any incoming ampersands into a dummy character for now.
980 ## turn any incoming ampersands into a dummy character for now.
979 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
981 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
980 ## implying an incoming html entity, to be skipped
982 ## implying an incoming html entity, to be skipped
981
983
982 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
984 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
983 end
985 end
984
986
985 def no_textile( text )
987 def no_textile( text )
986 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
988 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
987 '\1<notextile>\2</notextile>\3' )
989 '\1<notextile>\2</notextile>\3' )
988 text.gsub!( /^ *==([^=]+.*?)==/m,
990 text.gsub!( /^ *==([^=]+.*?)==/m,
989 '\1<notextile>\2</notextile>\3' )
991 '\1<notextile>\2</notextile>\3' )
990 end
992 end
991
993
992 def clean_white_space( text )
994 def clean_white_space( text )
993 # normalize line breaks
995 # normalize line breaks
994 text.gsub!( /\r\n/, "\n" )
996 text.gsub!( /\r\n/, "\n" )
995 text.gsub!( /\r/, "\n" )
997 text.gsub!( /\r/, "\n" )
996 text.gsub!( /\t/, ' ' )
998 text.gsub!( /\t/, ' ' )
997 text.gsub!( /^ +$/, '' )
999 text.gsub!( /^ +$/, '' )
998 text.gsub!( /\n{3,}/, "\n\n" )
1000 text.gsub!( /\n{3,}/, "\n\n" )
999 text.gsub!( /"$/, "\" " )
1001 text.gsub!( /"$/, "\" " )
1000
1002
1001 # if entire document is indented, flush
1003 # if entire document is indented, flush
1002 # to the left side
1004 # to the left side
1003 flush_left text
1005 flush_left text
1004 end
1006 end
1005
1007
1006 def flush_left( text )
1008 def flush_left( text )
1007 indt = 0
1009 indt = 0
1008 if text =~ /^ /
1010 if text =~ /^ /
1009 while text !~ /^ {#{indt}}\S/
1011 while text !~ /^ {#{indt}}\S/
1010 indt += 1
1012 indt += 1
1011 end unless text.empty?
1013 end unless text.empty?
1012 if indt.nonzero?
1014 if indt.nonzero?
1013 text.gsub!( /^ {#{indt}}/, '' )
1015 text.gsub!( /^ {#{indt}}/, '' )
1014 end
1016 end
1015 end
1017 end
1016 end
1018 end
1017
1019
1018 def footnote_ref( text )
1020 def footnote_ref( text )
1019 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1021 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1020 '<sup><a href="#fn\1">\1</a></sup>\2' )
1022 '<sup><a href="#fn\1">\1</a></sup>\2' )
1021 end
1023 end
1022
1024
1023 OFFTAGS = /(code|pre|kbd|notextile)/
1025 OFFTAGS = /(code|pre|kbd|notextile)/
1024 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1026 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1025 OFFTAG_OPEN = /<#{ OFFTAGS }/
1027 OFFTAG_OPEN = /<#{ OFFTAGS }/
1026 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1028 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1027 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1029 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1028 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1030 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1029
1031
1030 def glyphs_textile( text, level = 0 )
1032 def glyphs_textile( text, level = 0 )
1031 if text !~ HASTAG_MATCH
1033 if text !~ HASTAG_MATCH
1032 pgl text
1034 pgl text
1033 footnote_ref text
1035 footnote_ref text
1034 else
1036 else
1035 codepre = 0
1037 codepre = 0
1036 text.gsub!( ALLTAG_MATCH ) do |line|
1038 text.gsub!( ALLTAG_MATCH ) do |line|
1037 ## matches are off if we're between <code>, <pre> etc.
1039 ## matches are off if we're between <code>, <pre> etc.
1038 if $1
1040 if $1
1039 if line =~ OFFTAG_OPEN
1041 if line =~ OFFTAG_OPEN
1040 codepre += 1
1042 codepre += 1
1041 elsif line =~ OFFTAG_CLOSE
1043 elsif line =~ OFFTAG_CLOSE
1042 codepre -= 1
1044 codepre -= 1
1043 codepre = 0 if codepre < 0
1045 codepre = 0 if codepre < 0
1044 end
1046 end
1045 elsif codepre.zero?
1047 elsif codepre.zero?
1046 glyphs_textile( line, level + 1 )
1048 glyphs_textile( line, level + 1 )
1047 else
1049 else
1048 htmlesc( line, :NoQuotes )
1050 htmlesc( line, :NoQuotes )
1049 end
1051 end
1050 # p [level, codepre, line]
1052 # p [level, codepre, line]
1051
1053
1052 line
1054 line
1053 end
1055 end
1054 end
1056 end
1055 end
1057 end
1056
1058
1057 def rip_offtags( text, escape_aftertag=true )
1059 def rip_offtags( text, escape_aftertag=true )
1058 if text =~ /<.*>/
1060 if text =~ /<.*>/
1059 ## strip and encode <pre> content
1061 ## strip and encode <pre> content
1060 codepre, used_offtags = 0, {}
1062 codepre, used_offtags = 0, {}
1061 text.gsub!( OFFTAG_MATCH ) do |line|
1063 text.gsub!( OFFTAG_MATCH ) do |line|
1062 if $3
1064 if $3
1063 first, offtag, aftertag = $3, $4, $5
1065 first, offtag, aftertag = $3, $4, $5
1064 codepre += 1
1066 codepre += 1
1065 used_offtags[offtag] = true
1067 used_offtags[offtag] = true
1066 if codepre - used_offtags.length > 0
1068 if codepre - used_offtags.length > 0
1067 htmlesc( line, :NoQuotes )
1069 htmlesc( line, :NoQuotes )
1068 @pre_list.last << line
1070 @pre_list.last << line
1069 line = ""
1071 line = ""
1070 else
1072 else
1071 ### htmlesc is disabled between CODE tags which will be parsed with highlighter
1073 ### htmlesc is disabled between CODE tags which will be parsed with highlighter
1072 ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
1074 ### Regexp in formatter.rb is : /<code\s+class="(\w+)">\s?(.+)/m
1073 ### NB: some changes were made not to use $N variables, because we use "match"
1075 ### NB: some changes were made not to use $N variables, because we use "match"
1074 ### and it breaks following lines
1076 ### and it breaks following lines
1075 htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/)
1077 htmlesc( aftertag, :NoQuotes ) if aftertag && escape_aftertag && !first.match(/<code\s+class="(\w+)">/)
1076 line = "<redpre##{ @pre_list.length }>"
1078 line = "<redpre##{ @pre_list.length }>"
1077 first.match(/<#{ OFFTAGS }([^>]*)>/)
1079 first.match(/<#{ OFFTAGS }([^>]*)>/)
1078 tag = $1
1080 tag = $1
1079 $2.to_s.match(/(class\=\S+)/i)
1081 $2.to_s.match(/(class\=\S+)/i)
1080 tag << " #{$1}" if $1
1082 tag << " #{$1}" if $1
1081 @pre_list << "<#{ tag }>#{ aftertag }"
1083 @pre_list << "<#{ tag }>#{ aftertag }"
1082 end
1084 end
1083 elsif $1 and codepre > 0
1085 elsif $1 and codepre > 0
1084 if codepre - used_offtags.length > 0
1086 if codepre - used_offtags.length > 0
1085 htmlesc( line, :NoQuotes )
1087 htmlesc( line, :NoQuotes )
1086 @pre_list.last << line
1088 @pre_list.last << line
1087 line = ""
1089 line = ""
1088 end
1090 end
1089 codepre -= 1 unless codepre.zero?
1091 codepre -= 1 unless codepre.zero?
1090 used_offtags = {} if codepre.zero?
1092 used_offtags = {} if codepre.zero?
1091 end
1093 end
1092 line
1094 line
1093 end
1095 end
1094 end
1096 end
1095 text
1097 text
1096 end
1098 end
1097
1099
1098 def smooth_offtags( text )
1100 def smooth_offtags( text )
1099 unless @pre_list.empty?
1101 unless @pre_list.empty?
1100 ## replace <pre> content
1102 ## replace <pre> content
1101 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1103 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1102 end
1104 end
1103 end
1105 end
1104
1106
1105 def inline( text )
1107 def inline( text )
1106 [/^inline_/, /^glyphs_/].each do |meth_re|
1108 [/^inline_/, /^glyphs_/].each do |meth_re|
1107 @rules.each do |rule_name|
1109 @rules.each do |rule_name|
1108 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1110 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1109 end
1111 end
1110 end
1112 end
1111 end
1113 end
1112
1114
1113 def h_align( text )
1115 def h_align( text )
1114 H_ALGN_VALS[text]
1116 H_ALGN_VALS[text]
1115 end
1117 end
1116
1118
1117 def v_align( text )
1119 def v_align( text )
1118 V_ALGN_VALS[text]
1120 V_ALGN_VALS[text]
1119 end
1121 end
1120
1122
1121 def textile_popup_help( name, windowW, windowH )
1123 def textile_popup_help( name, windowW, windowH )
1122 ' <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 />'
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 />'
1123 end
1125 end
1124
1126
1125 # HTML cleansing stuff
1127 # HTML cleansing stuff
1126 BASIC_TAGS = {
1128 BASIC_TAGS = {
1127 'a' => ['href', 'title'],
1129 'a' => ['href', 'title'],
1128 'img' => ['src', 'alt', 'title'],
1130 'img' => ['src', 'alt', 'title'],
1129 'br' => [],
1131 'br' => [],
1130 'i' => nil,
1132 'i' => nil,
1131 'u' => nil,
1133 'u' => nil,
1132 'b' => nil,
1134 'b' => nil,
1133 'pre' => nil,
1135 'pre' => nil,
1134 'kbd' => nil,
1136 'kbd' => nil,
1135 'code' => ['lang'],
1137 'code' => ['lang'],
1136 'cite' => nil,
1138 'cite' => nil,
1137 'strong' => nil,
1139 'strong' => nil,
1138 'em' => nil,
1140 'em' => nil,
1139 'ins' => nil,
1141 'ins' => nil,
1140 'sup' => nil,
1142 'sup' => nil,
1141 'sub' => nil,
1143 'sub' => nil,
1142 'del' => nil,
1144 'del' => nil,
1143 'table' => nil,
1145 'table' => nil,
1144 'tr' => nil,
1146 'tr' => nil,
1145 'td' => ['colspan', 'rowspan'],
1147 'td' => ['colspan', 'rowspan'],
1146 'th' => nil,
1148 'th' => nil,
1147 'ol' => nil,
1149 'ol' => nil,
1148 'ul' => nil,
1150 'ul' => nil,
1149 'li' => nil,
1151 'li' => nil,
1150 'p' => nil,
1152 'p' => nil,
1151 'h1' => nil,
1153 'h1' => nil,
1152 'h2' => nil,
1154 'h2' => nil,
1153 'h3' => nil,
1155 'h3' => nil,
1154 'h4' => nil,
1156 'h4' => nil,
1155 'h5' => nil,
1157 'h5' => nil,
1156 'h6' => nil,
1158 'h6' => nil,
1157 'blockquote' => ['cite']
1159 'blockquote' => ['cite']
1158 }
1160 }
1159
1161
1160 def clean_html( text, tags = BASIC_TAGS )
1162 def clean_html( text, tags = BASIC_TAGS )
1161 text.gsub!( /<!\[CDATA\[/, '' )
1163 text.gsub!( /<!\[CDATA\[/, '' )
1162 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1164 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1163 raw = $~
1165 raw = $~
1164 tag = raw[2].downcase
1166 tag = raw[2].downcase
1165 if tags.has_key? tag
1167 if tags.has_key? tag
1166 pcs = [tag]
1168 pcs = [tag]
1167 tags[tag].each do |prop|
1169 tags[tag].each do |prop|
1168 ['"', "'", ''].each do |q|
1170 ['"', "'", ''].each do |q|
1169 q2 = ( q != '' ? q : '\s' )
1171 q2 = ( q != '' ? q : '\s' )
1170 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1172 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1171 attrv = $1
1173 attrv = $1
1172 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1174 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1173 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1175 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1174 break
1176 break
1175 end
1177 end
1176 end
1178 end
1177 end if tags[tag]
1179 end if tags[tag]
1178 "<#{raw[1]}#{pcs.join " "}>"
1180 "<#{raw[1]}#{pcs.join " "}>"
1179 else
1181 else
1180 " "
1182 " "
1181 end
1183 end
1182 end
1184 end
1183 end
1185 end
1184
1186
1185 ALLOWED_TAGS = %w(redpre pre code notextile)
1187 ALLOWED_TAGS = %w(redpre pre code notextile)
1186
1188
1187 def escape_html_tags(text)
1189 def escape_html_tags(text)
1188 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1190 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1189 end
1191 end
1190 end
1192 end
1191
1193
@@ -1,81 +1,88
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.dirname(__FILE__) + '/../../../../test_helper'
18 require File.dirname(__FILE__) + '/../../../../test_helper'
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
73
74 def test_double_dashes_should_not_strikethrough
75 assert_html_output(
76 'double -- dashes -- test' => 'double -- dashes -- test',
77 'double -- *dashes* -- test' => 'double -- <strong>dashes</strong> -- test'
78 )
79 end
80
74 private
81 private
75
82
76 def assert_html_output(to_test)
83 def assert_html_output(to_test)
77 to_test.each do |text, expected|
84 to_test.each do |text, expected|
78 assert_equal "<p>#{expected}</p>", @formatter.new(text).to_html, "Formatting the following text failed:\n===\n#{text}\n===\n"
85 assert_equal "<p>#{expected}</p>", @formatter.new(text).to_html, "Formatting the following text failed:\n===\n#{text}\n===\n"
79 end
86 end
80 end
87 end
81 end
88 end
General Comments 0
You need to be logged in to leave comments. Login now