##// END OF EJS Templates
Fixed: text formatting breaks when using parenthesis inside bold tags (#1334)....
Jean-Philippe Lang -
r3449:24b4c5f48dc0
parent child
Show More
@@ -1,1186 +1,1187
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\(])
383 /(^|[>\s\(]) # sta
384 (#{QTAGS_JOIN}|)
384 (#{QTAGS_JOIN}|) # oqs
385 (#{rcq})
385 (#{rcq}) # qtag
386 (#{C})
386 (\w|[^\s].*?[^\s]) # content
387 (?::(\S+?))?
388 (\w|[^\s].*?[^\s])
389 #{rcq}
387 #{rcq}
390 (#{QTAGS_JOIN}|)
388 (#{QTAGS_JOIN}|) # oqa
391 (?=[[:punct:]]|\s|\)|$)/x
389 (?=[[:punct:]]|\s|\)|$)/x
392 else
390 else
393 /(#{rcq})
391 /(#{rcq})
394 (#{C})
392 (#{C})
395 (?::(\S+))?
393 (?::(\S+))?
396 (\w|[^\s\-].*?[^\s\-])
394 (\w|[^\s\-].*?[^\s\-])
397 #{rcq}/xm
395 #{rcq}/xm
398 end
396 end
399 [rc, ht, re, rtype]
397 [rc, ht, re, rtype]
400 end
398 end
401
399
402 # Elements to handle
400 # Elements to handle
403 GLYPHS = [
401 GLYPHS = [
404 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1&#8217;\2' ], # single closing
402 # [ /([^\s\[{(>])?\'([dmst]\b|ll\b|ve\b|\s|:|$)/, '\1&#8217;\2' ], # single closing
405 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
403 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)\'/, '\1&#8217;' ], # single closing
406 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
404 # [ /\'(?=[#{PUNCT_Q}]*(s\b|[\s#{PUNCT_NOQ}]))/, '&#8217;' ], # single closing
407 # [ /\'/, '&#8216;' ], # single opening
405 # [ /\'/, '&#8216;' ], # single opening
408 # [ /</, '&lt;' ], # less-than
406 # [ /</, '&lt;' ], # less-than
409 # [ />/, '&gt;' ], # greater-than
407 # [ />/, '&gt;' ], # greater-than
410 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
408 # [ /([^\s\[{(])?"(\s|:|$)/, '\1&#8221;\2' ], # double closing
411 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
409 # [ /([^\s\[{(>#{PUNCT_Q}][#{PUNCT_Q}]*)"/, '\1&#8221;' ], # double closing
412 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
410 # [ /"(?=[#{PUNCT_Q}]*[\s#{PUNCT_NOQ}])/, '&#8221;' ], # double closing
413 # [ /"/, '&#8220;' ], # double opening
411 # [ /"/, '&#8220;' ], # double opening
414 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
412 # [ /\b( )?\.{3}/, '\1&#8230;' ], # ellipsis
415 # [ /\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/, '<acronym title="\2">\1</acronym>' ], # 3+ uppercase acronym
413 # [ /\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]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
414 # [ /(^|[^"][>\s])([A-Z][A-Z0-9 ]+[A-Z0-9])([^<A-Za-z0-9]|$)/, '\1<span class="caps">\2</span>\3', :no_span_caps ], # 3+ uppercase caps
417 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
415 # [ /(\.\s)?\s?--\s?/, '\1&#8212;' ], # em dash
418 # [ /\s->\s/, ' &rarr; ' ], # right arrow
416 # [ /\s->\s/, ' &rarr; ' ], # right arrow
419 # [ /\s-\s/, ' &#8211; ' ], # en dash
417 # [ /\s-\s/, ' &#8211; ' ], # en dash
420 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
418 # [ /(\d+) ?x ?(\d+)/, '\1&#215;\2' ], # dimension sign
421 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
419 # [ /\b ?[(\[]TM[\])]/i, '&#8482;' ], # trademark
422 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
420 # [ /\b ?[(\[]R[\])]/i, '&#174;' ], # registered
423 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
421 # [ /\b ?[(\[]C[\])]/i, '&#169;' ] # copyright
424 ]
422 ]
425
423
426 H_ALGN_VALS = {
424 H_ALGN_VALS = {
427 '<' => 'left',
425 '<' => 'left',
428 '=' => 'center',
426 '=' => 'center',
429 '>' => 'right',
427 '>' => 'right',
430 '<>' => 'justify'
428 '<>' => 'justify'
431 }
429 }
432
430
433 V_ALGN_VALS = {
431 V_ALGN_VALS = {
434 '^' => 'top',
432 '^' => 'top',
435 '-' => 'middle',
433 '-' => 'middle',
436 '~' => 'bottom'
434 '~' => 'bottom'
437 }
435 }
438
436
439 #
437 #
440 # Flexible HTML escaping
438 # Flexible HTML escaping
441 #
439 #
442 def htmlesc( str, mode=:Quotes )
440 def htmlesc( str, mode=:Quotes )
443 if str
441 if str
444 str.gsub!( '&', '&amp;' )
442 str.gsub!( '&', '&amp;' )
445 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
443 str.gsub!( '"', '&quot;' ) if mode != :NoQuotes
446 str.gsub!( "'", '&#039;' ) if mode == :Quotes
444 str.gsub!( "'", '&#039;' ) if mode == :Quotes
447 str.gsub!( '<', '&lt;')
445 str.gsub!( '<', '&lt;')
448 str.gsub!( '>', '&gt;')
446 str.gsub!( '>', '&gt;')
449 end
447 end
450 str
448 str
451 end
449 end
452
450
453 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
451 # Search and replace for Textile glyphs (quotes, dashes, other symbols)
454 def pgl( text )
452 def pgl( text )
455 #GLYPHS.each do |re, resub, tog|
453 #GLYPHS.each do |re, resub, tog|
456 # next if tog and method( tog ).call
454 # next if tog and method( tog ).call
457 # text.gsub! re, resub
455 # text.gsub! re, resub
458 #end
456 #end
459 text.gsub!(/\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/) do |m|
457 text.gsub!(/\b([A-Z][A-Z0-9]{2,})\b(?:[(]([^)]*)[)])/) do |m|
460 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
458 "<acronym title=\"#{htmlesc $2}\">#{$1}</acronym>"
461 end
459 end
462 end
460 end
463
461
464 # Parses Textile attribute lists and builds an HTML attribute string
462 # Parses Textile attribute lists and builds an HTML attribute string
465 def pba( text_in, element = "" )
463 def pba( text_in, element = "" )
466
464
467 return '' unless text_in
465 return '' unless text_in
468
466
469 style = []
467 style = []
470 text = text_in.dup
468 text = text_in.dup
471 if element == 'td'
469 if element == 'td'
472 colspan = $1 if text =~ /\\(\d+)/
470 colspan = $1 if text =~ /\\(\d+)/
473 rowspan = $1 if text =~ /\/(\d+)/
471 rowspan = $1 if text =~ /\/(\d+)/
474 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
472 style << "vertical-align:#{ v_align( $& ) };" if text =~ A_VLGN
475 end
473 end
476
474
477 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
475 style << "#{ htmlesc $1 };" if text.sub!( /\{([^}]*)\}/, '' ) && !filter_styles
478
476
479 lang = $1 if
477 lang = $1 if
480 text.sub!( /\[([^)]+?)\]/, '' )
478 text.sub!( /\[([^)]+?)\]/, '' )
481
479
482 cls = $1 if
480 cls = $1 if
483 text.sub!( /\(([^()]+?)\)/, '' )
481 text.sub!( /\(([^()]+?)\)/, '' )
484
482
485 style << "padding-left:#{ $1.length }em;" if
483 style << "padding-left:#{ $1.length }em;" if
486 text.sub!( /([(]+)/, '' )
484 text.sub!( /([(]+)/, '' )
487
485
488 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
486 style << "padding-right:#{ $1.length }em;" if text.sub!( /([)]+)/, '' )
489
487
490 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
488 style << "text-align:#{ h_align( $& ) };" if text =~ A_HLGN
491
489
492 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
490 cls, id = $1, $2 if cls =~ /^(.*?)#(.*)$/
493
491
494 atts = ''
492 atts = ''
495 atts << " style=\"#{ style.join }\"" unless style.empty?
493 atts << " style=\"#{ style.join }\"" unless style.empty?
496 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
494 atts << " class=\"#{ cls }\"" unless cls.to_s.empty?
497 atts << " lang=\"#{ lang }\"" if lang
495 atts << " lang=\"#{ lang }\"" if lang
498 atts << " id=\"#{ id }\"" if id
496 atts << " id=\"#{ id }\"" if id
499 atts << " colspan=\"#{ colspan }\"" if colspan
497 atts << " colspan=\"#{ colspan }\"" if colspan
500 atts << " rowspan=\"#{ rowspan }\"" if rowspan
498 atts << " rowspan=\"#{ rowspan }\"" if rowspan
501
499
502 atts
500 atts
503 end
501 end
504
502
505 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
503 TABLE_RE = /^(?:table(_?#{S}#{A}#{C})\. ?\n)?^(#{A}#{C}\.? ?\|.*?\|)(\n\n|\Z)/m
506
504
507 # Parses a Textile table block, building HTML from the result.
505 # Parses a Textile table block, building HTML from the result.
508 def block_textile_table( text )
506 def block_textile_table( text )
509 text.gsub!( TABLE_RE ) do |matches|
507 text.gsub!( TABLE_RE ) do |matches|
510
508
511 tatts, fullrow = $~[1..2]
509 tatts, fullrow = $~[1..2]
512 tatts = pba( tatts, 'table' )
510 tatts = pba( tatts, 'table' )
513 tatts = shelve( tatts ) if tatts
511 tatts = shelve( tatts ) if tatts
514 rows = []
512 rows = []
515
513
516 fullrow.each_line do |row|
514 fullrow.each_line do |row|
517 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
515 ratts, row = pba( $1, 'tr' ), $2 if row =~ /^(#{A}#{C}\. )(.*)/m
518 cells = []
516 cells = []
519 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
517 row.split( /(\|)(?![^\[\|]*\]\])/ )[1..-2].each do |cell|
520 next if cell == '|'
518 next if cell == '|'
521 ctyp = 'd'
519 ctyp = 'd'
522 ctyp = 'h' if cell =~ /^_/
520 ctyp = 'h' if cell =~ /^_/
523
521
524 catts = ''
522 catts = ''
525 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
523 catts, cell = pba( $1, 'td' ), $2 if cell =~ /^(_?#{S}#{A}#{C}\. ?)(.*)/
526
524
527 catts = shelve( catts ) if catts
525 catts = shelve( catts ) if catts
528 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
526 cells << "\t\t\t<t#{ ctyp }#{ catts }>#{ cell }</t#{ ctyp }>"
529 end
527 end
530 ratts = shelve( ratts ) if ratts
528 ratts = shelve( ratts ) if ratts
531 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
529 rows << "\t\t<tr#{ ratts }>\n#{ cells.join( "\n" ) }\n\t\t</tr>"
532 end
530 end
533 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
531 "\t<table#{ tatts }>\n#{ rows.join( "\n" ) }\n\t</table>\n\n"
534 end
532 end
535 end
533 end
536
534
537 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
535 LISTS_RE = /^([#*]+?#{C} .*?)$(?![^#*])/m
538 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
536 LISTS_CONTENT_RE = /^([#*]+)(#{A}#{C}) (.*)$/m
539
537
540 # Parses Textile lists and generates HTML
538 # Parses Textile lists and generates HTML
541 def block_textile_lists( text )
539 def block_textile_lists( text )
542 text.gsub!( LISTS_RE ) do |match|
540 text.gsub!( LISTS_RE ) do |match|
543 lines = match.split( /\n/ )
541 lines = match.split( /\n/ )
544 last_line = -1
542 last_line = -1
545 depth = []
543 depth = []
546 lines.each_with_index do |line, line_id|
544 lines.each_with_index do |line, line_id|
547 if line =~ LISTS_CONTENT_RE
545 if line =~ LISTS_CONTENT_RE
548 tl,atts,content = $~[1..3]
546 tl,atts,content = $~[1..3]
549 if depth.last
547 if depth.last
550 if depth.last.length > tl.length
548 if depth.last.length > tl.length
551 (depth.length - 1).downto(0) do |i|
549 (depth.length - 1).downto(0) do |i|
552 break if depth[i].length == tl.length
550 break if depth[i].length == tl.length
553 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
551 lines[line_id - 1] << "</li>\n\t</#{ lT( depth[i] ) }l>\n\t"
554 depth.pop
552 depth.pop
555 end
553 end
556 end
554 end
557 if depth.last and depth.last.length == tl.length
555 if depth.last and depth.last.length == tl.length
558 lines[line_id - 1] << '</li>'
556 lines[line_id - 1] << '</li>'
559 end
557 end
560 end
558 end
561 unless depth.last == tl
559 unless depth.last == tl
562 depth << tl
560 depth << tl
563 atts = pba( atts )
561 atts = pba( atts )
564 atts = shelve( atts ) if atts
562 atts = shelve( atts ) if atts
565 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
563 lines[line_id] = "\t<#{ lT(tl) }l#{ atts }>\n\t<li>#{ content }"
566 else
564 else
567 lines[line_id] = "\t\t<li>#{ content }"
565 lines[line_id] = "\t\t<li>#{ content }"
568 end
566 end
569 last_line = line_id
567 last_line = line_id
570
568
571 else
569 else
572 last_line = line_id
570 last_line = line_id
573 end
571 end
574 if line_id - last_line > 1 or line_id == lines.length - 1
572 if line_id - last_line > 1 or line_id == lines.length - 1
575 depth.delete_if do |v|
573 depth.delete_if do |v|
576 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
574 lines[last_line] << "</li>\n\t</#{ lT( v ) }l>"
577 end
575 end
578 end
576 end
579 end
577 end
580 lines.join( "\n" )
578 lines.join( "\n" )
581 end
579 end
582 end
580 end
583
581
584 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
582 QUOTES_RE = /(^>+([^\n]*?)(\n|$))+/m
585 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
583 QUOTES_CONTENT_RE = /^([> ]+)(.*)$/m
586
584
587 def block_textile_quotes( text )
585 def block_textile_quotes( text )
588 text.gsub!( QUOTES_RE ) do |match|
586 text.gsub!( QUOTES_RE ) do |match|
589 lines = match.split( /\n/ )
587 lines = match.split( /\n/ )
590 quotes = ''
588 quotes = ''
591 indent = 0
589 indent = 0
592 lines.each do |line|
590 lines.each do |line|
593 line =~ QUOTES_CONTENT_RE
591 line =~ QUOTES_CONTENT_RE
594 bq,content = $1, $2
592 bq,content = $1, $2
595 l = bq.count('>')
593 l = bq.count('>')
596 if l != indent
594 if l != indent
597 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
595 quotes << ("\n\n" + (l>indent ? '<blockquote>' * (l-indent) : '</blockquote>' * (indent-l)) + "\n\n")
598 indent = l
596 indent = l
599 end
597 end
600 quotes << (content + "\n")
598 quotes << (content + "\n")
601 end
599 end
602 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
600 quotes << ("\n" + '</blockquote>' * indent + "\n\n")
603 quotes
601 quotes
604 end
602 end
605 end
603 end
606
604
607 CODE_RE = /(\W)
605 CODE_RE = /(\W)
608 @
606 @
609 (?:\|(\w+?)\|)?
607 (?:\|(\w+?)\|)?
610 (.+?)
608 (.+?)
611 @
609 @
612 (?=\W)/x
610 (?=\W)/x
613
611
614 def inline_textile_code( text )
612 def inline_textile_code( text )
615 text.gsub!( CODE_RE ) do |m|
613 text.gsub!( CODE_RE ) do |m|
616 before,lang,code,after = $~[1..4]
614 before,lang,code,after = $~[1..4]
617 lang = " lang=\"#{ lang }\"" if lang
615 lang = " lang=\"#{ lang }\"" if lang
618 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }" )
616 rip_offtags( "#{ before }<code#{ lang }>#{ code }</code>#{ after }" )
619 end
617 end
620 end
618 end
621
619
622 def lT( text )
620 def lT( text )
623 text =~ /\#$/ ? 'o' : 'u'
621 text =~ /\#$/ ? 'o' : 'u'
624 end
622 end
625
623
626 def hard_break( text )
624 def hard_break( text )
627 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
625 text.gsub!( /(.)\n(?!\Z| *([#*=]+(\s|$)|[{|]))/, "\\1<br />" ) if hard_breaks
628 end
626 end
629
627
630 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
628 BLOCKS_GROUP_RE = /\n{2,}(?! )/m
631
629
632 def blocks( text, deep_code = false )
630 def blocks( text, deep_code = false )
633 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
631 text.replace( text.split( BLOCKS_GROUP_RE ).collect do |blk|
634 plain = blk !~ /\A[#*> ]/
632 plain = blk !~ /\A[#*> ]/
635
633
636 # skip blocks that are complex HTML
634 # skip blocks that are complex HTML
637 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
635 if blk =~ /^<\/?(\w+).*>/ and not SIMPLE_HTML_TAGS.include? $1
638 blk
636 blk
639 else
637 else
640 # search for indentation levels
638 # search for indentation levels
641 blk.strip!
639 blk.strip!
642 if blk.empty?
640 if blk.empty?
643 blk
641 blk
644 else
642 else
645 code_blk = nil
643 code_blk = nil
646 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
644 blk.gsub!( /((?:\n(?:\n^ +[^\n]*)+)+)/m ) do |iblk|
647 flush_left iblk
645 flush_left iblk
648 blocks iblk, plain
646 blocks iblk, plain
649 iblk.gsub( /^(\S)/, "\t\\1" )
647 iblk.gsub( /^(\S)/, "\t\\1" )
650 if plain
648 if plain
651 code_blk = iblk; ""
649 code_blk = iblk; ""
652 else
650 else
653 iblk
651 iblk
654 end
652 end
655 end
653 end
656
654
657 block_applied = 0
655 block_applied = 0
658 @rules.each do |rule_name|
656 @rules.each do |rule_name|
659 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
657 block_applied += 1 if ( rule_name.to_s.match /^block_/ and method( rule_name ).call( blk ) )
660 end
658 end
661 if block_applied.zero?
659 if block_applied.zero?
662 if deep_code
660 if deep_code
663 blk = "\t<pre><code>#{ blk }</code></pre>"
661 blk = "\t<pre><code>#{ blk }</code></pre>"
664 else
662 else
665 blk = "\t<p>#{ blk }</p>"
663 blk = "\t<p>#{ blk }</p>"
666 end
664 end
667 end
665 end
668 # hard_break blk
666 # hard_break blk
669 blk + "\n#{ code_blk }"
667 blk + "\n#{ code_blk }"
670 end
668 end
671 end
669 end
672
670
673 end.join( "\n\n" ) )
671 end.join( "\n\n" ) )
674 end
672 end
675
673
676 def textile_bq( tag, atts, cite, content )
674 def textile_bq( tag, atts, cite, content )
677 cite, cite_title = check_refs( cite )
675 cite, cite_title = check_refs( cite )
678 cite = " cite=\"#{ cite }\"" if cite
676 cite = " cite=\"#{ cite }\"" if cite
679 atts = shelve( atts ) if atts
677 atts = shelve( atts ) if atts
680 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
678 "\t<blockquote#{ cite }>\n\t\t<p#{ atts }>#{ content }</p>\n\t</blockquote>"
681 end
679 end
682
680
683 def textile_p( tag, atts, cite, content )
681 def textile_p( tag, atts, cite, content )
684 atts = shelve( atts ) if atts
682 atts = shelve( atts ) if atts
685 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
683 "\t<#{ tag }#{ atts }>#{ content }</#{ tag }>"
686 end
684 end
687
685
688 alias textile_h1 textile_p
686 alias textile_h1 textile_p
689 alias textile_h2 textile_p
687 alias textile_h2 textile_p
690 alias textile_h3 textile_p
688 alias textile_h3 textile_p
691 alias textile_h4 textile_p
689 alias textile_h4 textile_p
692 alias textile_h5 textile_p
690 alias textile_h5 textile_p
693 alias textile_h6 textile_p
691 alias textile_h6 textile_p
694
692
695 def textile_fn_( tag, num, atts, cite, content )
693 def textile_fn_( tag, num, atts, cite, content )
696 atts << " id=\"fn#{ num }\" class=\"footnote\""
694 atts << " id=\"fn#{ num }\" class=\"footnote\""
697 content = "<sup>#{ num }</sup> #{ content }"
695 content = "<sup>#{ num }</sup> #{ content }"
698 atts = shelve( atts ) if atts
696 atts = shelve( atts ) if atts
699 "\t<p#{ atts }>#{ content }</p>"
697 "\t<p#{ atts }>#{ content }</p>"
700 end
698 end
701
699
702 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
700 BLOCK_RE = /^(([a-z]+)(\d*))(#{A}#{C})\.(?::(\S+))? (.*)$/m
703
701
704 def block_textile_prefix( text )
702 def block_textile_prefix( text )
705 if text =~ BLOCK_RE
703 if text =~ BLOCK_RE
706 tag,tagpre,num,atts,cite,content = $~[1..6]
704 tag,tagpre,num,atts,cite,content = $~[1..6]
707 atts = pba( atts )
705 atts = pba( atts )
708
706
709 # pass to prefix handler
707 # pass to prefix handler
710 if respond_to? "textile_#{ tag }", true
708 if respond_to? "textile_#{ tag }", true
711 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
709 text.gsub!( $&, method( "textile_#{ tag }" ).call( tag, atts, cite, content ) )
712 elsif respond_to? "textile_#{ tagpre }_", true
710 elsif respond_to? "textile_#{ tagpre }_", true
713 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
711 text.gsub!( $&, method( "textile_#{ tagpre }_" ).call( tagpre, num, atts, cite, content ) )
714 end
712 end
715 end
713 end
716 end
714 end
717
715
718 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
716 SETEXT_RE = /\A(.+?)\n([=-])[=-]* *$/m
719 def block_markdown_setext( text )
717 def block_markdown_setext( text )
720 if text =~ SETEXT_RE
718 if text =~ SETEXT_RE
721 tag = if $2 == "="; "h1"; else; "h2"; end
719 tag = if $2 == "="; "h1"; else; "h2"; end
722 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
720 blk, cont = "<#{ tag }>#{ $1 }</#{ tag }>", $'
723 blocks cont
721 blocks cont
724 text.replace( blk + cont )
722 text.replace( blk + cont )
725 end
723 end
726 end
724 end
727
725
728 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
726 ATX_RE = /\A(\#{1,6}) # $1 = string of #'s
729 [ ]*
727 [ ]*
730 (.+?) # $2 = Header text
728 (.+?) # $2 = Header text
731 [ ]*
729 [ ]*
732 \#* # optional closing #'s (not counted)
730 \#* # optional closing #'s (not counted)
733 $/x
731 $/x
734 def block_markdown_atx( text )
732 def block_markdown_atx( text )
735 if text =~ ATX_RE
733 if text =~ ATX_RE
736 tag = "h#{ $1.length }"
734 tag = "h#{ $1.length }"
737 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
735 blk, cont = "<#{ tag }>#{ $2 }</#{ tag }>\n\n", $'
738 blocks cont
736 blocks cont
739 text.replace( blk + cont )
737 text.replace( blk + cont )
740 end
738 end
741 end
739 end
742
740
743 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
741 MARKDOWN_BQ_RE = /\A(^ *> ?.+$(.+\n)*\n*)+/m
744
742
745 def block_markdown_bq( text )
743 def block_markdown_bq( text )
746 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
744 text.gsub!( MARKDOWN_BQ_RE ) do |blk|
747 blk.gsub!( /^ *> ?/, '' )
745 blk.gsub!( /^ *> ?/, '' )
748 flush_left blk
746 flush_left blk
749 blocks blk
747 blocks blk
750 blk.gsub!( /^(\S)/, "\t\\1" )
748 blk.gsub!( /^(\S)/, "\t\\1" )
751 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
749 "<blockquote>\n#{ blk }\n</blockquote>\n\n"
752 end
750 end
753 end
751 end
754
752
755 MARKDOWN_RULE_RE = /^(#{
753 MARKDOWN_RULE_RE = /^(#{
756 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
754 ['*', '-', '_'].collect { |ch| ' ?(' + Regexp::quote( ch ) + ' ?){3,}' }.join( '|' )
757 })$/
755 })$/
758
756
759 def block_markdown_rule( text )
757 def block_markdown_rule( text )
760 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
758 text.gsub!( MARKDOWN_RULE_RE ) do |blk|
761 "<hr />"
759 "<hr />"
762 end
760 end
763 end
761 end
764
762
765 # XXX TODO XXX
763 # XXX TODO XXX
766 def block_markdown_lists( text )
764 def block_markdown_lists( text )
767 end
765 end
768
766
769 def inline_textile_span( text )
767 def inline_textile_span( text )
770 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
768 QTAGS.each do |qtag_rc, ht, qtag_re, rtype|
771 text.gsub!( qtag_re ) do |m|
769 text.gsub!( qtag_re ) do |m|
772
770
773 case rtype
771 case rtype
774 when :limit
772 when :limit
775 sta,oqs,qtag,atts,cite,content,oqa = $~[1..7]
773 sta,oqs,qtag,content,oqa = $~[1..6]
774 atts = nil
775 if content =~ /^(#{C})(.+)$/
776 atts, content = $~[1..2]
777 end
776 else
778 else
777 qtag,atts,cite,content = $~[1..4]
779 qtag,atts,cite,content = $~[1..4]
778 sta = ''
780 sta = ''
779 end
781 end
780 atts = pba( atts )
782 atts = pba( atts )
781 atts << " cite=\"#{ cite }\"" if cite
782 atts = shelve( atts ) if atts
783 atts = shelve( atts ) if atts
783
784
784 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
785 "#{ sta }#{ oqs }<#{ ht }#{ atts }>#{ content }</#{ ht }>#{ oqa }"
785
786
786 end
787 end
787 end
788 end
788 end
789 end
789
790
790 LINK_RE = /
791 LINK_RE = /
791 (
792 (
792 ([\s\[{(]|[#{PUNCT}])? # $pre
793 ([\s\[{(]|[#{PUNCT}])? # $pre
793 " # start
794 " # start
794 (#{C}) # $atts
795 (#{C}) # $atts
795 ([^"\n]+?) # $text
796 ([^"\n]+?) # $text
796 \s?
797 \s?
797 (?:\(([^)]+?)\)(?="))? # $title
798 (?:\(([^)]+?)\)(?="))? # $title
798 ":
799 ":
799 ( # $url
800 ( # $url
800 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
801 (\/|[a-zA-Z]+:\/\/|www\.|mailto:) # $proto
801 [\w\/]\S+?
802 [\w\/]\S+?
802 )
803 )
803 (\/)? # $slash
804 (\/)? # $slash
804 ([^\w\=\/;\(\)]*?) # $post
805 ([^\w\=\/;\(\)]*?) # $post
805 )
806 )
806 (?=<|\s|$)
807 (?=<|\s|$)
807 /x
808 /x
808 #"
809 #"
809 def inline_textile_link( text )
810 def inline_textile_link( text )
810 text.gsub!( LINK_RE ) do |m|
811 text.gsub!( LINK_RE ) do |m|
811 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
812 all,pre,atts,text,title,url,proto,slash,post = $~[1..9]
812 if text.include?('<br />')
813 if text.include?('<br />')
813 all
814 all
814 else
815 else
815 url, url_title = check_refs( url )
816 url, url_title = check_refs( url )
816 title ||= url_title
817 title ||= url_title
817
818
818 # Idea below : an URL with unbalanced parethesis and
819 # Idea below : an URL with unbalanced parethesis and
819 # ending by ')' is put into external parenthesis
820 # ending by ')' is put into external parenthesis
820 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
821 if ( url[-1]==?) and ((url.count("(") - url.count(")")) < 0 ) )
821 url=url[0..-2] # discard closing parenth from url
822 url=url[0..-2] # discard closing parenth from url
822 post = ")"+post # add closing parenth to post
823 post = ")"+post # add closing parenth to post
823 end
824 end
824 atts = pba( atts )
825 atts = pba( atts )
825 atts = " href=\"#{ url }#{ slash }\"#{ atts }"
826 atts = " href=\"#{ url }#{ slash }\"#{ atts }"
826 atts << " title=\"#{ htmlesc title }\"" if title
827 atts << " title=\"#{ htmlesc title }\"" if title
827 atts = shelve( atts ) if atts
828 atts = shelve( atts ) if atts
828
829
829 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
830 external = (url =~ /^https?:\/\//) ? ' class="external"' : ''
830
831
831 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
832 "#{ pre }<a#{ atts }#{ external }>#{ text }</a>#{ post }"
832 end
833 end
833 end
834 end
834 end
835 end
835
836
836 MARKDOWN_REFLINK_RE = /
837 MARKDOWN_REFLINK_RE = /
837 \[([^\[\]]+)\] # $text
838 \[([^\[\]]+)\] # $text
838 [ ]? # opt. space
839 [ ]? # opt. space
839 (?:\n[ ]*)? # one optional newline followed by spaces
840 (?:\n[ ]*)? # one optional newline followed by spaces
840 \[(.*?)\] # $id
841 \[(.*?)\] # $id
841 /x
842 /x
842
843
843 def inline_markdown_reflink( text )
844 def inline_markdown_reflink( text )
844 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
845 text.gsub!( MARKDOWN_REFLINK_RE ) do |m|
845 text, id = $~[1..2]
846 text, id = $~[1..2]
846
847
847 if id.empty?
848 if id.empty?
848 url, title = check_refs( text )
849 url, title = check_refs( text )
849 else
850 else
850 url, title = check_refs( id )
851 url, title = check_refs( id )
851 end
852 end
852
853
853 atts = " href=\"#{ url }\""
854 atts = " href=\"#{ url }\""
854 atts << " title=\"#{ title }\"" if title
855 atts << " title=\"#{ title }\"" if title
855 atts = shelve( atts )
856 atts = shelve( atts )
856
857
857 "<a#{ atts }>#{ text }</a>"
858 "<a#{ atts }>#{ text }</a>"
858 end
859 end
859 end
860 end
860
861
861 MARKDOWN_LINK_RE = /
862 MARKDOWN_LINK_RE = /
862 \[([^\[\]]+)\] # $text
863 \[([^\[\]]+)\] # $text
863 \( # open paren
864 \( # open paren
864 [ \t]* # opt space
865 [ \t]* # opt space
865 <?(.+?)>? # $href
866 <?(.+?)>? # $href
866 [ \t]* # opt space
867 [ \t]* # opt space
867 (?: # whole title
868 (?: # whole title
868 (['"]) # $quote
869 (['"]) # $quote
869 (.*?) # $title
870 (.*?) # $title
870 \3 # matching quote
871 \3 # matching quote
871 )? # title is optional
872 )? # title is optional
872 \)
873 \)
873 /x
874 /x
874
875
875 def inline_markdown_link( text )
876 def inline_markdown_link( text )
876 text.gsub!( MARKDOWN_LINK_RE ) do |m|
877 text.gsub!( MARKDOWN_LINK_RE ) do |m|
877 text, url, quote, title = $~[1..4]
878 text, url, quote, title = $~[1..4]
878
879
879 atts = " href=\"#{ url }\""
880 atts = " href=\"#{ url }\""
880 atts << " title=\"#{ title }\"" if title
881 atts << " title=\"#{ title }\"" if title
881 atts = shelve( atts )
882 atts = shelve( atts )
882
883
883 "<a#{ atts }>#{ text }</a>"
884 "<a#{ atts }>#{ text }</a>"
884 end
885 end
885 end
886 end
886
887
887 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
888 TEXTILE_REFS_RE = /(^ *)\[([^\[\n]+?)\](#{HYPERLINK})(?=\s|$)/
888 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
889 MARKDOWN_REFS_RE = /(^ *)\[([^\n]+?)\]:\s+<?(#{HYPERLINK})>?(?:\s+"((?:[^"]|\\")+)")?(?=\s|$)/m
889
890
890 def refs( text )
891 def refs( text )
891 @rules.each do |rule_name|
892 @rules.each do |rule_name|
892 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
893 method( rule_name ).call( text ) if rule_name.to_s.match /^refs_/
893 end
894 end
894 end
895 end
895
896
896 def refs_textile( text )
897 def refs_textile( text )
897 text.gsub!( TEXTILE_REFS_RE ) do |m|
898 text.gsub!( TEXTILE_REFS_RE ) do |m|
898 flag, url = $~[2..3]
899 flag, url = $~[2..3]
899 @urlrefs[flag.downcase] = [url, nil]
900 @urlrefs[flag.downcase] = [url, nil]
900 nil
901 nil
901 end
902 end
902 end
903 end
903
904
904 def refs_markdown( text )
905 def refs_markdown( text )
905 text.gsub!( MARKDOWN_REFS_RE ) do |m|
906 text.gsub!( MARKDOWN_REFS_RE ) do |m|
906 flag, url = $~[2..3]
907 flag, url = $~[2..3]
907 title = $~[6]
908 title = $~[6]
908 @urlrefs[flag.downcase] = [url, title]
909 @urlrefs[flag.downcase] = [url, title]
909 nil
910 nil
910 end
911 end
911 end
912 end
912
913
913 def check_refs( text )
914 def check_refs( text )
914 ret = @urlrefs[text.downcase] if text
915 ret = @urlrefs[text.downcase] if text
915 ret || [text, nil]
916 ret || [text, nil]
916 end
917 end
917
918
918 IMAGE_RE = /
919 IMAGE_RE = /
919 (>|\s|^) # start of line?
920 (>|\s|^) # start of line?
920 \! # opening
921 \! # opening
921 (\<|\=|\>)? # optional alignment atts
922 (\<|\=|\>)? # optional alignment atts
922 (#{C}) # optional style,class atts
923 (#{C}) # optional style,class atts
923 (?:\. )? # optional dot-space
924 (?:\. )? # optional dot-space
924 ([^\s(!]+?) # presume this is the src
925 ([^\s(!]+?) # presume this is the src
925 \s? # optional space
926 \s? # optional space
926 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
927 (?:\(((?:[^\(\)]|\([^\)]+\))+?)\))? # optional title
927 \! # closing
928 \! # closing
928 (?::#{ HYPERLINK })? # optional href
929 (?::#{ HYPERLINK })? # optional href
929 /x
930 /x
930
931
931 def inline_textile_image( text )
932 def inline_textile_image( text )
932 text.gsub!( IMAGE_RE ) do |m|
933 text.gsub!( IMAGE_RE ) do |m|
933 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
934 stln,algn,atts,url,title,href,href_a1,href_a2 = $~[1..8]
934 htmlesc title
935 htmlesc title
935 atts = pba( atts )
936 atts = pba( atts )
936 atts = " src=\"#{ url }\"#{ atts }"
937 atts = " src=\"#{ url }\"#{ atts }"
937 atts << " title=\"#{ title }\"" if title
938 atts << " title=\"#{ title }\"" if title
938 atts << " alt=\"#{ title }\""
939 atts << " alt=\"#{ title }\""
939 # size = @getimagesize($url);
940 # size = @getimagesize($url);
940 # if($size) $atts.= " $size[3]";
941 # if($size) $atts.= " $size[3]";
941
942
942 href, alt_title = check_refs( href ) if href
943 href, alt_title = check_refs( href ) if href
943 url, url_title = check_refs( url )
944 url, url_title = check_refs( url )
944
945
945 out = ''
946 out = ''
946 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
947 out << "<a#{ shelve( " href=\"#{ href }\"" ) }>" if href
947 out << "<img#{ shelve( atts ) } />"
948 out << "<img#{ shelve( atts ) } />"
948 out << "</a>#{ href_a1 }#{ href_a2 }" if href
949 out << "</a>#{ href_a1 }#{ href_a2 }" if href
949
950
950 if algn
951 if algn
951 algn = h_align( algn )
952 algn = h_align( algn )
952 if stln == "<p>"
953 if stln == "<p>"
953 out = "<p style=\"float:#{ algn }\">#{ out }"
954 out = "<p style=\"float:#{ algn }\">#{ out }"
954 else
955 else
955 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
956 out = "#{ stln }<div style=\"float:#{ algn }\">#{ out }</div>"
956 end
957 end
957 else
958 else
958 out = stln + out
959 out = stln + out
959 end
960 end
960
961
961 out
962 out
962 end
963 end
963 end
964 end
964
965
965 def shelve( val )
966 def shelve( val )
966 @shelf << val
967 @shelf << val
967 " :redsh##{ @shelf.length }:"
968 " :redsh##{ @shelf.length }:"
968 end
969 end
969
970
970 def retrieve( text )
971 def retrieve( text )
971 @shelf.each_with_index do |r, i|
972 @shelf.each_with_index do |r, i|
972 text.gsub!( " :redsh##{ i + 1 }:", r )
973 text.gsub!( " :redsh##{ i + 1 }:", r )
973 end
974 end
974 end
975 end
975
976
976 def incoming_entities( text )
977 def incoming_entities( text )
977 ## turn any incoming ampersands into a dummy character for now.
978 ## turn any incoming ampersands into a dummy character for now.
978 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
979 ## This uses a negative lookahead for alphanumerics followed by a semicolon,
979 ## implying an incoming html entity, to be skipped
980 ## implying an incoming html entity, to be skipped
980
981
981 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
982 text.gsub!( /&(?![#a-z0-9]+;)/i, "x%x%" )
982 end
983 end
983
984
984 def no_textile( text )
985 def no_textile( text )
985 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
986 text.gsub!( /(^|\s)==([^=]+.*?)==(\s|$)?/,
986 '\1<notextile>\2</notextile>\3' )
987 '\1<notextile>\2</notextile>\3' )
987 text.gsub!( /^ *==([^=]+.*?)==/m,
988 text.gsub!( /^ *==([^=]+.*?)==/m,
988 '\1<notextile>\2</notextile>\3' )
989 '\1<notextile>\2</notextile>\3' )
989 end
990 end
990
991
991 def clean_white_space( text )
992 def clean_white_space( text )
992 # normalize line breaks
993 # normalize line breaks
993 text.gsub!( /\r\n/, "\n" )
994 text.gsub!( /\r\n/, "\n" )
994 text.gsub!( /\r/, "\n" )
995 text.gsub!( /\r/, "\n" )
995 text.gsub!( /\t/, ' ' )
996 text.gsub!( /\t/, ' ' )
996 text.gsub!( /^ +$/, '' )
997 text.gsub!( /^ +$/, '' )
997 text.gsub!( /\n{3,}/, "\n\n" )
998 text.gsub!( /\n{3,}/, "\n\n" )
998 text.gsub!( /"$/, "\" " )
999 text.gsub!( /"$/, "\" " )
999
1000
1000 # if entire document is indented, flush
1001 # if entire document is indented, flush
1001 # to the left side
1002 # to the left side
1002 flush_left text
1003 flush_left text
1003 end
1004 end
1004
1005
1005 def flush_left( text )
1006 def flush_left( text )
1006 indt = 0
1007 indt = 0
1007 if text =~ /^ /
1008 if text =~ /^ /
1008 while text !~ /^ {#{indt}}\S/
1009 while text !~ /^ {#{indt}}\S/
1009 indt += 1
1010 indt += 1
1010 end unless text.empty?
1011 end unless text.empty?
1011 if indt.nonzero?
1012 if indt.nonzero?
1012 text.gsub!( /^ {#{indt}}/, '' )
1013 text.gsub!( /^ {#{indt}}/, '' )
1013 end
1014 end
1014 end
1015 end
1015 end
1016 end
1016
1017
1017 def footnote_ref( text )
1018 def footnote_ref( text )
1018 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1019 text.gsub!( /\b\[([0-9]+?)\](\s)?/,
1019 '<sup><a href="#fn\1">\1</a></sup>\2' )
1020 '<sup><a href="#fn\1">\1</a></sup>\2' )
1020 end
1021 end
1021
1022
1022 OFFTAGS = /(code|pre|kbd|notextile)/
1023 OFFTAGS = /(code|pre|kbd|notextile)/
1023 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1024 OFFTAG_MATCH = /(?:(<\/#{ OFFTAGS }>)|(<#{ OFFTAGS }[^>]*>))(.*?)(?=<\/?#{ OFFTAGS }\W|\Z)/mi
1024 OFFTAG_OPEN = /<#{ OFFTAGS }/
1025 OFFTAG_OPEN = /<#{ OFFTAGS }/
1025 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1026 OFFTAG_CLOSE = /<\/?#{ OFFTAGS }/
1026 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1027 HASTAG_MATCH = /(<\/?\w[^\n]*?>)/m
1027 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1028 ALLTAG_MATCH = /(<\/?\w[^\n]*?>)|.*?(?=<\/?\w[^\n]*?>|$)/m
1028
1029
1029 def glyphs_textile( text, level = 0 )
1030 def glyphs_textile( text, level = 0 )
1030 if text !~ HASTAG_MATCH
1031 if text !~ HASTAG_MATCH
1031 pgl text
1032 pgl text
1032 footnote_ref text
1033 footnote_ref text
1033 else
1034 else
1034 codepre = 0
1035 codepre = 0
1035 text.gsub!( ALLTAG_MATCH ) do |line|
1036 text.gsub!( ALLTAG_MATCH ) do |line|
1036 ## matches are off if we're between <code>, <pre> etc.
1037 ## matches are off if we're between <code>, <pre> etc.
1037 if $1
1038 if $1
1038 if line =~ OFFTAG_OPEN
1039 if line =~ OFFTAG_OPEN
1039 codepre += 1
1040 codepre += 1
1040 elsif line =~ OFFTAG_CLOSE
1041 elsif line =~ OFFTAG_CLOSE
1041 codepre -= 1
1042 codepre -= 1
1042 codepre = 0 if codepre < 0
1043 codepre = 0 if codepre < 0
1043 end
1044 end
1044 elsif codepre.zero?
1045 elsif codepre.zero?
1045 glyphs_textile( line, level + 1 )
1046 glyphs_textile( line, level + 1 )
1046 else
1047 else
1047 htmlesc( line, :NoQuotes )
1048 htmlesc( line, :NoQuotes )
1048 end
1049 end
1049 # p [level, codepre, line]
1050 # p [level, codepre, line]
1050
1051
1051 line
1052 line
1052 end
1053 end
1053 end
1054 end
1054 end
1055 end
1055
1056
1056 def rip_offtags( text )
1057 def rip_offtags( text )
1057 if text =~ /<.*>/
1058 if text =~ /<.*>/
1058 ## strip and encode <pre> content
1059 ## strip and encode <pre> content
1059 codepre, used_offtags = 0, {}
1060 codepre, used_offtags = 0, {}
1060 text.gsub!( OFFTAG_MATCH ) do |line|
1061 text.gsub!( OFFTAG_MATCH ) do |line|
1061 if $3
1062 if $3
1062 offtag, aftertag = $4, $5
1063 offtag, aftertag = $4, $5
1063 codepre += 1
1064 codepre += 1
1064 used_offtags[offtag] = true
1065 used_offtags[offtag] = true
1065 if codepre - used_offtags.length > 0
1066 if codepre - used_offtags.length > 0
1066 htmlesc( line, :NoQuotes )
1067 htmlesc( line, :NoQuotes )
1067 @pre_list.last << line
1068 @pre_list.last << line
1068 line = ""
1069 line = ""
1069 else
1070 else
1070 htmlesc( aftertag, :NoQuotes ) if aftertag
1071 htmlesc( aftertag, :NoQuotes ) if aftertag
1071 line = "<redpre##{ @pre_list.length }>"
1072 line = "<redpre##{ @pre_list.length }>"
1072 $3.match(/<#{ OFFTAGS }([^>]*)>/)
1073 $3.match(/<#{ OFFTAGS }([^>]*)>/)
1073 tag = $1
1074 tag = $1
1074 $2.to_s.match(/(class\=\S+)/i)
1075 $2.to_s.match(/(class\=\S+)/i)
1075 tag << " #{$1}" if $1
1076 tag << " #{$1}" if $1
1076 @pre_list << "<#{ tag }>#{ aftertag }"
1077 @pre_list << "<#{ tag }>#{ aftertag }"
1077 end
1078 end
1078 elsif $1 and codepre > 0
1079 elsif $1 and codepre > 0
1079 if codepre - used_offtags.length > 0
1080 if codepre - used_offtags.length > 0
1080 htmlesc( line, :NoQuotes )
1081 htmlesc( line, :NoQuotes )
1081 @pre_list.last << line
1082 @pre_list.last << line
1082 line = ""
1083 line = ""
1083 end
1084 end
1084 codepre -= 1 unless codepre.zero?
1085 codepre -= 1 unless codepre.zero?
1085 used_offtags = {} if codepre.zero?
1086 used_offtags = {} if codepre.zero?
1086 end
1087 end
1087 line
1088 line
1088 end
1089 end
1089 end
1090 end
1090 text
1091 text
1091 end
1092 end
1092
1093
1093 def smooth_offtags( text )
1094 def smooth_offtags( text )
1094 unless @pre_list.empty?
1095 unless @pre_list.empty?
1095 ## replace <pre> content
1096 ## replace <pre> content
1096 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1097 text.gsub!( /<redpre#(\d+)>/ ) { @pre_list[$1.to_i] }
1097 end
1098 end
1098 end
1099 end
1099
1100
1100 def inline( text )
1101 def inline( text )
1101 [/^inline_/, /^glyphs_/].each do |meth_re|
1102 [/^inline_/, /^glyphs_/].each do |meth_re|
1102 @rules.each do |rule_name|
1103 @rules.each do |rule_name|
1103 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1104 method( rule_name ).call( text ) if rule_name.to_s.match( meth_re )
1104 end
1105 end
1105 end
1106 end
1106 end
1107 end
1107
1108
1108 def h_align( text )
1109 def h_align( text )
1109 H_ALGN_VALS[text]
1110 H_ALGN_VALS[text]
1110 end
1111 end
1111
1112
1112 def v_align( text )
1113 def v_align( text )
1113 V_ALGN_VALS[text]
1114 V_ALGN_VALS[text]
1114 end
1115 end
1115
1116
1116 def textile_popup_help( name, windowW, windowH )
1117 def textile_popup_help( name, windowW, windowH )
1117 ' <a target="_blank" href="http://hobix.com/textile/#' + helpvar + '" onclick="window.open(this.href, \'popupwindow\', \'width=' + windowW + ',height=' + windowH + ',scrollbars,resizable\'); return false;">' + name + '</a><br />'
1118 ' <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 />'
1118 end
1119 end
1119
1120
1120 # HTML cleansing stuff
1121 # HTML cleansing stuff
1121 BASIC_TAGS = {
1122 BASIC_TAGS = {
1122 'a' => ['href', 'title'],
1123 'a' => ['href', 'title'],
1123 'img' => ['src', 'alt', 'title'],
1124 'img' => ['src', 'alt', 'title'],
1124 'br' => [],
1125 'br' => [],
1125 'i' => nil,
1126 'i' => nil,
1126 'u' => nil,
1127 'u' => nil,
1127 'b' => nil,
1128 'b' => nil,
1128 'pre' => nil,
1129 'pre' => nil,
1129 'kbd' => nil,
1130 'kbd' => nil,
1130 'code' => ['lang'],
1131 'code' => ['lang'],
1131 'cite' => nil,
1132 'cite' => nil,
1132 'strong' => nil,
1133 'strong' => nil,
1133 'em' => nil,
1134 'em' => nil,
1134 'ins' => nil,
1135 'ins' => nil,
1135 'sup' => nil,
1136 'sup' => nil,
1136 'sub' => nil,
1137 'sub' => nil,
1137 'del' => nil,
1138 'del' => nil,
1138 'table' => nil,
1139 'table' => nil,
1139 'tr' => nil,
1140 'tr' => nil,
1140 'td' => ['colspan', 'rowspan'],
1141 'td' => ['colspan', 'rowspan'],
1141 'th' => nil,
1142 'th' => nil,
1142 'ol' => nil,
1143 'ol' => nil,
1143 'ul' => nil,
1144 'ul' => nil,
1144 'li' => nil,
1145 'li' => nil,
1145 'p' => nil,
1146 'p' => nil,
1146 'h1' => nil,
1147 'h1' => nil,
1147 'h2' => nil,
1148 'h2' => nil,
1148 'h3' => nil,
1149 'h3' => nil,
1149 'h4' => nil,
1150 'h4' => nil,
1150 'h5' => nil,
1151 'h5' => nil,
1151 'h6' => nil,
1152 'h6' => nil,
1152 'blockquote' => ['cite']
1153 'blockquote' => ['cite']
1153 }
1154 }
1154
1155
1155 def clean_html( text, tags = BASIC_TAGS )
1156 def clean_html( text, tags = BASIC_TAGS )
1156 text.gsub!( /<!\[CDATA\[/, '' )
1157 text.gsub!( /<!\[CDATA\[/, '' )
1157 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1158 text.gsub!( /<(\/*)(\w+)([^>]*)>/ ) do
1158 raw = $~
1159 raw = $~
1159 tag = raw[2].downcase
1160 tag = raw[2].downcase
1160 if tags.has_key? tag
1161 if tags.has_key? tag
1161 pcs = [tag]
1162 pcs = [tag]
1162 tags[tag].each do |prop|
1163 tags[tag].each do |prop|
1163 ['"', "'", ''].each do |q|
1164 ['"', "'", ''].each do |q|
1164 q2 = ( q != '' ? q : '\s' )
1165 q2 = ( q != '' ? q : '\s' )
1165 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1166 if raw[3] =~ /#{prop}\s*=\s*#{q}([^#{q2}]+)#{q}/i
1166 attrv = $1
1167 attrv = $1
1167 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1168 next if prop == 'src' and attrv =~ %r{^(?!http)\w+:}
1168 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1169 pcs << "#{prop}=\"#{$1.gsub('"', '\\"')}\""
1169 break
1170 break
1170 end
1171 end
1171 end
1172 end
1172 end if tags[tag]
1173 end if tags[tag]
1173 "<#{raw[1]}#{pcs.join " "}>"
1174 "<#{raw[1]}#{pcs.join " "}>"
1174 else
1175 else
1175 " "
1176 " "
1176 end
1177 end
1177 end
1178 end
1178 end
1179 end
1179
1180
1180 ALLOWED_TAGS = %w(redpre pre code notextile)
1181 ALLOWED_TAGS = %w(redpre pre code notextile)
1181
1182
1182 def escape_html_tags(text)
1183 def escape_html_tags(text)
1183 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1184 text.gsub!(%r{<(\/?([!\w]+)[^<>\n]*)(>?)}) {|m| ALLOWED_TAGS.include?($2) ? "<#{$1}#{$3}" : "&lt;#{$1}#{'&gt;' unless $3.blank?}" }
1184 end
1185 end
1185 end
1186 end
1186
1187
@@ -1,45 +1,63
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
36 to_test = {
37 '*bold*' => '<strong>bold</strong>',
38 'before *bold*' => 'before <strong>bold</strong>',
39 '*bold* after' => '<strong>bold</strong> after',
40 '*two words*' => '<strong>two words</strong>',
41 '*two*words*' => '<strong>two*words</strong>',
42 '*two * words*' => '<strong>two * words</strong>',
43 '*two* *words*' => '<strong>two</strong> <strong>words</strong>',
44 '*(two)* *(words)*' => '<strong>(two)</strong> <strong>(words)</strong>',
45 # with class
46 '*(foo)two words*' => '<strong class="foo">two words</strong>',
47 }
48 to_test.each do |text, expected|
49 assert_equal "<p>#{expected}</p>", @formatter.new(text).to_html
50 end
51 end
52
35 def test_modifiers_combination
53 def test_modifiers_combination
36 MODIFIERS.each do |m1, tag1|
54 MODIFIERS.each do |m1, tag1|
37 MODIFIERS.each do |m2, tag2|
55 MODIFIERS.each do |m2, tag2|
38 next if m1 == m2
56 next if m1 == m2
39 text = "#{m2}#{m1}Phrase modifiers#{m1}#{m2}"
57 text = "#{m2}#{m1}Phrase modifiers#{m1}#{m2}"
40 html = "<p><#{tag2}><#{tag1}>Phrase modifiers</#{tag1}></#{tag2}></p>"
58 html = "<p><#{tag2}><#{tag1}>Phrase modifiers</#{tag1}></#{tag2}></p>"
41 assert_equal html, @formatter.new(text).to_html
59 assert_equal html, @formatter.new(text).to_html
42 end
60 end
43 end
61 end
44 end
62 end
45 end
63 end
General Comments 0
You need to be logged in to leave comments. Login now