hpricot 0.5-mswin32 → 0.6-mswin32

Sign up to get free protection for your applications and to get access to all the features.
Files changed (31) hide show
  1. data/CHANGELOG +13 -1
  2. data/README +4 -1
  3. data/Rakefile +111 -75
  4. data/ext/hpricot_scan/HpricotScanService.java +1340 -0
  5. data/ext/hpricot_scan/hpricot_common.rl +76 -0
  6. data/ext/hpricot_scan/hpricot_scan.c +2435 -2181
  7. data/ext/hpricot_scan/hpricot_scan.java.rl +363 -0
  8. data/ext/hpricot_scan/hpricot_scan.rl +3 -70
  9. data/lib/hpricot.rb +1 -0
  10. data/lib/hpricot/blankslate.rb +63 -0
  11. data/lib/hpricot/builder.rb +200 -0
  12. data/lib/hpricot/elements.rb +143 -32
  13. data/lib/hpricot/inspect.rb +7 -3
  14. data/lib/hpricot/parse.rb +128 -101
  15. data/lib/hpricot/tag.rb +23 -15
  16. data/lib/hpricot/tags.rb +164 -0
  17. data/lib/hpricot/traverse.rb +93 -33
  18. data/lib/hpricot/xchar.rb +94 -0
  19. data/lib/i686-linux/hpricot_scan.so +0 -0
  20. data/test/files/pace_application.html +1320 -0
  21. data/test/files/tenderlove.html +16 -0
  22. data/test/files/why.xml +19 -0
  23. data/test/load_files.rb +1 -1
  24. data/test/test_alter.rb +65 -0
  25. data/test/test_builder.rb +24 -0
  26. data/test/test_parser.rb +92 -4
  27. data/test/test_preserved.rb +20 -0
  28. data/test/test_xml.rb +13 -0
  29. metadata +34 -14
  30. data/lib/hpricot/text.rb +0 -115
  31. data/lib/hpricot_scan.so +0 -0
data/test/files/tenderlove.html ADDED
@@ -0,0 +1,16 @@
1
+ <html>
2
+ <HEAD>
3
+ <meta http-equiv="Refresh" content="0; url=http://tenderlovemaking.com">
4
+ <META http-equiv="Refresh" content="0; url=http://tenderlovemaking.com">
5
+ </head>
6
+ <body>
7
+ <a href ="http://tenderlovemaking.com/">My Site!</a>
8
+ <A href ="http://whytheluckystiff.net/">Your Site!</A>
9
+ <MAP>
10
+ <area HREF="http://whytheluckystiff.net/" COORDS="1,2,3,4"></area>
11
+ <AREA HREF="http://tenderlovemaking.com/" COORDS="1,2,3,4">
12
+ </area>
13
+ <AREA HREF="http://tenderlovemaking.com/" COORDS="5,5,10,10" />
14
+ </MAP>
15
+ </body>
16
+ </html>
data/test/files/why.xml ADDED
@@ -0,0 +1,19 @@
1
+ <?xml version='1.0'?><rss xmlns:admin='http://webns.net/mvcb/' version='2.0' xmlns:sy='http://purl.org/rss/1.0/modules/syndication/' xmlns:dc='http://purl.org/dc/elements/1.1/' xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'>
2
+ <channel>
3
+ <title>why the lucky stiff</title>
4
+ <link>http://whytheluckystiff.net</link>
5
+ <description>hex-editing reality to give us infinite grenades!!</description>
6
+ <dc:language>en-us</dc:language>
7
+ <dc:creator/>
8
+ <dc:date>2007-01-16T22:39:04+00:00</dc:date>
9
+ <admin:generatorAgent rdf:resource='http://hobix.com/?v=0.4'/>
10
+ <sy:updatePeriod>hourly</sy:updatePeriod>
11
+ <sy:updateFrequency>1</sy:updateFrequency>
12
+ <sy:updateBase>2000-01-01T12:00+00:00</sy:updateBase>
13
+ <item><title>1.3</title><link>http://whytheluckystiff.net/quatrains/1.3.html</link><guid isPermaLink='false'>quatrains/1.3@http://whytheluckystiff.net</guid><dc:subject>quatrains</dc:subject><dc:subject>quatrains</dc:subject><dc:creator>why the lucky stiff</dc:creator><dc:date>2007-01-14T08:47:05+00:00</dc:date><description>&lt;blockquote&gt;
14
+ &lt;p&gt;That cadillac of yours and that driver of yours!&lt;br /&gt;You and your teacups rattling away in the back seat!&lt;br /&gt;You always took the mike, oh, and all those cowboys you shot!&lt;br /&gt;I held your hand! And I&amp;#8217;ll shoot a cowboy one day!&lt;/p&gt;
15
+ &lt;/blockquote&gt;
16
+ &lt;blockquote&gt;
17
+ &lt;p&gt;You said, &amp;#8220;Let&amp;#8217;s run into the woods like kids!&amp;#8221; &lt;br /&gt;You said, &amp;#8220;Let&amp;#8217;s rub our hands together super-hot!&amp;#8221; &lt;br /&gt;And we scalded the trees and left octagons, I think that was you and&lt;br /&gt;You threw parties on the roof!&lt;/p&gt;
18
+ &lt;/blockquote&gt;</description></item></channel>
19
+ </rss>
data/test/load_files.rb CHANGED
@@ -1,6 +1,6 @@
1
1
  module TestFiles
2
2
  Dir.chdir(File.dirname(__FILE__)) do
3
- Dir['files/*.{html,xhtml}'].each do |fname|
3
+ Dir['files/*.{html,xhtml,xml}'].each do |fname|
4
4
  const_set fname[%r!/(\w+)\.\w+$!, 1].upcase, IO.read(fname)
5
5
  end
6
6
  end
data/test/test_alter.rb ADDED
@@ -0,0 +1,65 @@
1
+ #!/usr/bin/env ruby
2
+
3
+ require 'test/unit'
4
+ require 'hpricot'
5
+ require 'load_files'
6
+
7
+ class TestAlter < Test::Unit::TestCase
8
+ def setup
9
+ @basic = Hpricot.parse(TestFiles::BASIC)
10
+ end
11
+
12
+ def test_before
13
+ test0 = "<link rel='stylesheet' href='test0.css' />"
14
+ @basic.at("link").before(test0)
15
+ assert_equal 'test0.css', @basic.at("link").attributes['href']
16
+ end
17
+
18
+ def test_after
19
+ test_inf = "<link rel='stylesheet' href='test_inf.css' />"
20
+ @basic.search("link")[-1].after(test_inf)
21
+ assert_equal 'test_inf.css', @basic.search("link")[-1].attributes['href']
22
+ end
23
+
24
+ def test_wrap
25
+ ohmy = (@basic/"p.ohmy").wrap("<div id='wrapper'></div>")
26
+ assert_equal 'wrapper', ohmy[0].parent['id']
27
+ assert_equal 'ohmy', Hpricot(@basic.to_html).at("#wrapper").children[0]['class']
28
+ end
29
+
30
+ def test_add_class
31
+ first_p = (@basic/"p:first").add_class("testing123")
32
+ assert first_p[0].get_attribute("class").split(" ").include?("testing123")
33
+ assert (Hpricot(@basic.to_html)/"p:first")[0].attributes["class"].split(" ").include?("testing123")
34
+ assert !(Hpricot(@basic.to_html)/"p:gt(0)")[0].attributes["class"].split(" ").include?("testing123")
35
+ end
36
+
37
+ def test_change_attributes
38
+ all_ps = (@basic/"p").attr("title", "Some Title")
39
+ all_as = (@basic/"a").attr("href", "http://my_new_href.com")
40
+ all_lb = (@basic/"link").attr("href") { |e| e.name }
41
+ assert_changed(@basic, "p", all_ps) {|p| p.attributes["title"] == "Some Title"}
42
+ assert_changed(@basic, "a", all_as) {|a| a.attributes["href"] == "http://my_new_href.com"}
43
+ assert_changed(@basic, "link", all_lb) {|a| a.attributes["href"] == "link" }
44
+ end
45
+
46
+ def test_remove_attr
47
+ all_rl = (@basic/"link").remove_attr("href")
48
+ assert_changed(@basic, "link", all_rl) { |link| link['href'].nil? }
49
+ end
50
+
51
+ def test_remove_class
52
+ all_c1 = (@basic/"p[@class*='last']").remove_class("last")
53
+ assert_changed(@basic, "p[@class*='last']", all_c1) { |p| p['class'] == 'final' }
54
+ end
55
+
56
+ def test_remove_all_classes
57
+ all_c2 = (@basic/"p[@class]").remove_class
58
+ assert_changed(@basic, "p[@class]", all_c2) { |p| p['class'].nil? }
59
+ end
60
+
61
+ def assert_changed original, selector, set, &block
62
+ assert set.all?(&block)
63
+ assert Hpricot(original.to_html).search(selector).all?(&block)
64
+ end
65
+ end
data/test/test_builder.rb ADDED
@@ -0,0 +1,24 @@
1
+ #!/usr/bin/env ruby
2
+
3
+ require 'test/unit'
4
+ require 'hpricot'
5
+
6
+ class TestBuilder < Test::Unit::TestCase
7
+ def test_escaping_text
8
+ doc = Hpricot() { b "<a\"b>" }
9
+ assert_equal "<b>&lt;a&quot;b&gt;</b>", doc.to_html
10
+ assert_equal %{<a"b>}, doc.at("text()").to_s
11
+ end
12
+
13
+ def test_no_escaping_text
14
+ doc = Hpricot() { div.test.me! { text "<a\"b>" } }
15
+ assert_equal %{<div class="test" id="me"><a"b></div>}, doc.to_html
16
+ assert_equal %{<a"b>}, doc.at("text()").to_s
17
+ end
18
+
19
+ def test_latin1_entities
20
+ doc = Hpricot() { b "\200\225" }
21
+ assert_equal "<b>&#8364;&#8226;</b>", doc.to_html
22
+ assert_equal "\342\202\254\342\200\242", doc.at("text()").to_s
23
+ end
24
+ end
data/test/test_parser.rb CHANGED
@@ -23,6 +23,19 @@ class TestParser < Test::Unit::TestCase
23
23
  assert_equal 'FOO', Hpricot.make("FOO").first.content
24
24
  end
25
25
 
26
+ def test_filter_by_attr
27
+ @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
28
+
29
+ # this link is escaped in the doc
30
+ link = 'http://www.youtube.com/watch?v=TvSNXyNw26g&search=chris%20ware'
31
+ assert_equal link, @boingboing.at("a[@href='#{link}']")['href']
32
+ end
33
+
34
+ def test_filter_contains
35
+ @basic = Hpricot.parse(TestFiles::BASIC)
36
+ assert_equal '<title>Sample XHTML</title>', @basic.search("title:contains('Sample')").to_s
37
+ end
38
+
26
39
  def test_get_element_by_id
27
40
  @basic = Hpricot.parse(TestFiles::BASIC)
28
41
  assert_equal 'link1', @basic.get_element_by_id('link1')['id']
@@ -84,6 +97,12 @@ class TestParser < Test::Unit::TestCase
84
97
  assert_equal "<p>one</p>", h.search("//div/p:first()").to_s
85
98
  end
86
99
 
100
+ def test_pace
101
+ doc = Hpricot(TestFiles::PACE_APPLICATION)
102
+ assert_equal 'get', doc.at('form[@name=frmSect11]')['method']
103
+ # assert_equal '2', doc.at('#hdnSpouse')['value']
104
+ end
105
+
87
106
  def test_scan_boingboing
88
107
  @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
89
108
  assert_equal 60, (@boingboing/'p.posted').length
@@ -95,7 +114,7 @@ class TestParser < Test::Unit::TestCase
95
114
  assert_equal 60, @boingboing.search("h3").length
96
115
  assert_equal 59, @boingboing.search("h3[text()!='College kids reportedly taking more smart drugs']").length
97
116
  assert_equal 17, @boingboing.search("h3[text()$='s']").length
98
- assert_equal 128, @boingboing.search("p[text()]").length
117
+ assert_equal 129, @boingboing.search("p[text()]").length
99
118
  assert_equal 211, @boingboing.search("p").length
100
119
  end
101
120
 
@@ -135,8 +154,7 @@ class TestParser < Test::Unit::TestCase
135
154
  assert_equal 60, @boingboing.search("/*/body//p[@class='posted']").length
136
155
  assert_equal 18, @boingboing.search("//script").length
137
156
  divs = @boingboing.search("//script/../div")
138
- assert_equal 2, divs.length
139
- assert_equal 1, divs.search('a').length
157
+ assert_equal 1, divs.length
140
158
  imgs = @boingboing.search('//div/p/a/img')
141
159
  assert_equal 15, imgs.length
142
160
  assert_equal 17, @boingboing.search('//div').search('p/a/img').length
@@ -155,9 +173,16 @@ class TestParser < Test::Unit::TestCase
155
173
  assert_equal 1, @boingboing.search('//input[@checked]').length
156
174
  end
157
175
 
176
+ def test_tag_case
177
+ @tenderlove = Hpricot.parse(TestFiles::TENDERLOVE)
178
+ assert_equal 2, @tenderlove.search('//a').length
179
+ assert_equal 3, @tenderlove.search('//area').length
180
+ assert_equal 2, @tenderlove.search('//meta').length
181
+ end
182
+
158
183
  def test_alt_predicates
159
184
  @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
160
- assert_equal 2, @boingboing.search('//table/tr:last').length
185
+ assert_equal 1, @boingboing.search('//table/tr:last').length
161
186
 
162
187
  @basic = Hpricot.parse(TestFiles::BASIC)
163
188
  assert_equal "<p>The third paragraph</p>",
@@ -167,6 +192,22 @@ class TestParser < Test::Unit::TestCase
167
192
  assert_equal 'last final', @basic.search('//p:last-of-type').first.get_attribute('class')
168
193
  end
169
194
 
195
+ def test_insert_after # ticket #63
196
+ doc = Hpricot('<html><body><div id="a-div"></div></body></html>')
197
+ (doc/'div').each do |element|
198
+ element.after('<p>Paragraph 1</p><p>Paragraph 2</p>')
199
+ end
200
+ assert_equal doc.to_html, '<html><body><div id="a-div"></div><p>Paragraph 1</p><p>Paragraph 2</p></body></html>'
201
+ end
202
+
203
+ def test_insert_before # ticket #61
204
+ doc = Hpricot('<html><body><div id="a-div"></div></body></html>')
205
+ (doc/'div').each do |element|
206
+ element.before('<p>Paragraph 1</p><p>Paragraph 2</p>')
207
+ end
208
+ assert_equal doc.to_html, '<html><body><p>Paragraph 1</p><p>Paragraph 2</p><div id="a-div"></div></body></html>'
209
+ end
210
+
170
211
  def test_many_paths
171
212
  @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
172
213
  assert_equal 62, @boingboing.search('p.posted, link[@rel="alternate"]').length
@@ -264,6 +305,53 @@ class TestParser < Test::Unit::TestCase
264
305
  end
265
306
  end
266
307
 
308
+ def test_youtube_attr
309
+ str = <<-edoc
310
+ <html><body>
311
+ Lorem ipsum. Jolly roger, ding-dong sing-a-long
312
+ <object width="425" height="350">
313
+ <param name="movie" value="http://www.youtube.com/v/NbDQ4M_cuwA"></param>
314
+ <param name="wmode" value="transparent"></param>
315
+ <embed src="http://www.youtube.com/v/NbDQ4M_cuwA"
316
+ type="application/x-shockwave-flash" wmode="transparent" width="425" height="350">
317
+ </embed>
318
+ </object>
319
+ Check out my posting, I have bright mice in large clown cars.
320
+ <object width="425" height="350">
321
+ <param name="movie" value="http://www.youtube.com/v/foobar"></param>
322
+ <param name="wmode" value="transparent"></param>
323
+ <embed src="http://www.youtube.com/v/foobar"
324
+ type="application/x-shockwave-flash" wmode="transparent" width="425" height="350">
325
+ </embed>
326
+ </object>
327
+ </body></html?
328
+ edoc
329
+ doc = Hpricot(str)
330
+ assert_equal "http://www.youtube.com/v/NbDQ4M_cuwA",
331
+ doc.at("//object/param[@value='http://www.youtube.com/v/NbDQ4M_cuwA']")['value']
332
+ end
333
+
334
+ # ticket #84 by jamezilla
335
+ def test_screwed_xmlns
336
+ doc = Hpricot(<<-edoc)
337
+ <?xml:namespace prefix = cwi />
338
+ <html><body>HAI</body></html>
339
+ edoc
340
+ assert_equal "HAI", doc.at("body").inner_text
341
+ end
342
+
343
+ # Reported by Jonathan Nichols on the Hpricot list (24 May 2007)
344
+ def test_self_closed_form
345
+ doc = Hpricot(<<-edoc)
346
+ <body>
347
+ <form action="/loginRegForm" name="regForm" method="POST" />
348
+ <input type="button">
349
+ </form>
350
+ </body>
351
+ edoc
352
+ assert_equal "button", doc.at("//form/input")['type']
353
+ end
354
+
267
355
  def test_filters
268
356
  @basic = Hpricot.parse(TestFiles::BASIC)
269
357
  assert_equal 0, (@basic/"title:parent").size
data/test/test_preserved.rb CHANGED
@@ -38,9 +38,29 @@ class TestPreserved < Test::Unit::TestCase
38
38
  end
39
39
  end
40
40
 
41
+ def test_escaping_of_contents
42
+ doc = Hpricot(TestFiles::BOINGBOING)
43
+ assert_equal "Fukuda\342\200\231s Automatic Door opens around your body as you pass through it. The idea is to save energy and keep the room clean.", doc.at("img[@alt='200606131240']").next.to_s.strip
44
+ end
45
+
41
46
  def test_files
42
47
  assert_roundtrip TestFiles::BASIC
43
48
  assert_roundtrip TestFiles::BOINGBOING
44
49
  assert_roundtrip TestFiles::CY0
45
50
  end
51
+
52
+ def test_escaping_of_attrs
53
+ # ampersands in URLs
54
+ str = %{<a href="http://google.com/search?q=hpricot&amp;l=en">Google</a>}
55
+ link = (doc = Hpricot(str)).at(:a)
56
+ assert_equal "http://google.com/search?q=hpricot&l=en", link['href']
57
+ assert_equal "http://google.com/search?q=hpricot&l=en", link.attributes['href']
58
+ assert_equal "http://google.com/search?q=hpricot&l=en", link.get_attribute('href')
59
+ assert_equal "http://google.com/search?q=hpricot&amp;l=en", link.raw_attributes['href']
60
+ assert_equal str, doc.to_html
61
+
62
+ # alter the url
63
+ link['href'] = "javascript:alert(\"AGGA-KA-BOO!\")"
64
+ assert_equal %{<a href="javascript:alert(&quot;AGGA-KA-BOO!&quot;)">Google</a>}, doc.to_html
65
+ end
46
66
  end
data/test/test_xml.rb CHANGED
@@ -12,4 +12,17 @@ class TestParser < Test::Unit::TestCase
12
12
  assert_equal "this is title", (doc/:rss/:channel/:title).text
13
13
  assert_equal "http://fake.com", (doc/:rss/:channel/:link).text
14
14
  end
15
+
16
+ # make sure XML doesn't get downcased
17
+ def test_casing
18
+ doc = Hpricot::XML(TestFiles::WHY)
19
+ assert_equal "hourly", (doc.at "sy:updatePeriod").inner_html
20
+ assert_equal 1, (doc/"guid[@isPermaLink]").length
21
+ end
22
+
23
+ # be sure tags named "text" are ok
24
+ def test_text_tags
25
+ doc = Hpricot::XML("<feed><title>City Poisoned</title><text>Rita Lee has poisoned Brazil.</text></feed>")
26
+ assert_equal "City Poisoned", (doc/"title").text
27
+ end
15
28
  end
metadata CHANGED
@@ -3,10 +3,11 @@ rubygems_version: 0.9.0
3
3
  specification_version: 1
4
4
  name: hpricot
5
5
  version: !ruby/object:Gem::Version
6
- version: "0.5"
7
- date: 2007-01-31 00:00:00 -08:00
6
+ version: "0.6"
7
+ date: 2007-06-15 00:00:00 -07:00
8
8
  summary: a swift, liberal HTML parser with a fantastic library
9
9
  require_paths:
10
+ - lib/i686-linux
10
11
  - lib
11
12
  email: why@ruby-lang.org
12
13
  homepage: http://code.whytheluckystiff.net/hpricot/
@@ -15,7 +16,7 @@ description: a swift, liberal HTML parser with a fantastic library
15
16
  autorequire:
16
17
  default_executable:
17
18
  bindir: bin
18
- has_rdoc: false
19
+ has_rdoc: true
19
20
  required_ruby_version: !ruby/object:Gem::Version::Requirement
20
21
  requirements:
21
22
  - - ">"
@@ -29,41 +30,60 @@ post_install_message:
29
30
  authors:
30
31
  - why the lucky stiff
31
32
  files:
33
+ - CHANGELOG
34
+ - COPYING
35
+ - README
36
+ - Rakefile
37
+ - test/files
32
38
  - test/test_preserved.rb
33
39
  - test/test_paths.rb
34
40
  - test/load_files.rb
35
41
  - test/test_xml.rb
36
42
  - test/test_parser.rb
43
+ - test/test_alter.rb
44
+ - test/test_builder.rb
45
+ - test/files/why.xml
37
46
  - test/files/boingboing.html
38
47
  - test/files/uswebgen.html
39
48
  - test/files/immob.html
40
49
  - test/files/week9.html
41
50
  - test/files/utf8.html
42
- - test/files/cy0.html
43
51
  - test/files/basic.xhtml
52
+ - test/files/cy0.html
53
+ - test/files/tenderlove.html
54
+ - test/files/pace_application.html
55
+ - lib/hpricot
44
56
  - lib/hpricot.rb
57
+ - lib/i686-linux
58
+ - lib/hpricot/builder.rb
45
59
  - lib/hpricot/htmlinfo.rb
46
- - lib/hpricot/text.rb
60
+ - lib/hpricot/xchar.rb
47
61
  - lib/hpricot/inspect.rb
48
62
  - lib/hpricot/modules.rb
49
63
  - lib/hpricot/parse.rb
50
64
  - lib/hpricot/tag.rb
51
65
  - lib/hpricot/traverse.rb
52
66
  - lib/hpricot/elements.rb
67
+ - lib/hpricot/tags.rb
68
+ - lib/hpricot/blankslate.rb
69
+ - extras/mingw-rbconfig.rb
70
+ - ext/hpricot_scan/hpricot_scan.h
71
+ - ext/hpricot_scan/HpricotScanService.java
53
72
  - ext/hpricot_scan/hpricot_scan.c
54
73
  - ext/hpricot_scan/extconf.rb
55
- - ext/hpricot_scan/hpricot_scan.h
74
+ - ext/hpricot_scan/hpricot_common.rl
56
75
  - ext/hpricot_scan/hpricot_scan.rl
57
- - CHANGELOG
58
- - README
59
- - Rakefile
60
- - COPYING
61
- - extras/mingw-rbconfig.rb
62
- - lib/hpricot_scan.so
76
+ - ext/hpricot_scan/hpricot_scan.java.rl
77
+ - lib/i686-linux/hpricot_scan.so
63
78
  test_files: []
64
79
 
65
- rdoc_options: []
66
-
80
+ rdoc_options:
81
+ - --quiet
82
+ - --title
83
+ - The Hpricot Reference
84
+ - --main
85
+ - README
86
+ - --inline-source
67
87
  extra_rdoc_files:
68
88
  - README
69
89
  - CHANGELOG
data/lib/hpricot/text.rb DELETED
@@ -1,115 +0,0 @@
1
- require 'hpricot/modules'
2
- require 'hpricot/raw_string'
3
- require 'hpricot/htmlinfo'
4
- require 'hpricot/encoder'
5
- require 'hpricot/fstr'
6
- require 'iconv'
7
-
8
- module Hpricot
9
- class Text
10
- # :stopdoc:
11
- class << self
12
- alias new_internal new
13
- end
14
- # :startdoc:
15
-
16
- def Text.new(arg)
17
- arg = arg.to_node if Hpricot::Location === arg
18
- if Text === arg
19
- new_internal arg.rcdata, arg.normalized_rcdata
20
- elsif String === arg
21
- arg2 = arg.gsub(/&/, '&amp;')
22
- arg = arg2.freeze if arg != arg2
23
- new_internal arg
24
- else
25
- raise TypeError, "cannot initialize Text with #{arg.inspect}"
26
- end
27
- end
28
-
29
- def initialize(rcdata, normalized_rcdata=internal_normalize(rcdata)) # :notnew:
30
- init_raw_string
31
- @rcdata = rcdata && Hpricot.frozen_string(rcdata)
32
- @normalized_rcdata = @rcdata == normalized_rcdata ? @rcdata : normalized_rcdata
33
- end
34
- attr_reader :rcdata, :normalized_rcdata
35
-
36
- def internal_normalize(rcdata)
37
- # - character references are decoded as much as possible.
38
- # - undecodable character references are converted to decimal numeric character refereces.
39
- result = rcdata.gsub(/&(?:#([0-9]+)|#x([0-9a-fA-F]+)|([A-Za-z][A-Za-z0-9]*));/o) {|s|
40
- u = nil
41
- if $1
42
- u = $1.to_i
43
- elsif $2
44
- u = $2.hex
45
- elsif $3
46
- u = NamedCharacters[$3]
47
- end
48
- if !u || u < 0 || 0x7fffffff < u
49
- '?'
50
- elsif u == 38 # '&' character.
51
- '&#38;'
52
- elsif u <= 0x7f
53
- [u].pack("C")
54
- else
55
- begin
56
- Iconv.conv(Encoder.internal_charset, 'UTF-8', [u].pack("U"))
57
- rescue Iconv::Failure
58
- "&##{u};"
59
- end
60
- end
61
- }
62
- Hpricot.frozen_string(result)
63
- end
64
- private :internal_normalize
65
-
66
- # Hpricot::Text#to_s converts the text to a string.
67
- # - character references are decoded as much as possible.
68
- # - undecodable character reference are converted to `?' character.
69
- def to_s
70
- @normalized_rcdata.gsub(/&(?:#([0-9]+));/o) {|s|
71
- u = $1.to_i
72
- if 0 <= u && u <= 0x7f
73
- [u].pack("C")
74
- else
75
- '?'
76
- end
77
- }
78
- end
79
-
80
- def empty?
81
- @normalized_rcdata.empty?
82
- end
83
-
84
- def strip
85
- rcdata = @normalized_rcdata.dup
86
- rcdata.sub!(/\A(?:\s|&nbsp;)+/, '')
87
- rcdata.sub!(/(?:\s|&nbsp;)+\z/, '')
88
- if rcdata == @normalized_rcdata
89
- self
90
- else
91
- rcdata.freeze
92
- Text.new_internal(rcdata, rcdata)
93
- end
94
- end
95
-
96
- # Hpricot::Text.concat returns a text which is concatenation of arguments.
97
- #
98
- # An argument should be one of follows.
99
- # - String
100
- # - Hpricot::Text
101
- # - Hpricot::Location which points Hpricot::Text
102
- def Text.concat(*args)
103
- rcdata = ''
104
- args.each {|arg|
105
- arg = arg.to_node if Hpricot::Location === arg
106
- if Text === arg
107
- rcdata << arg.rcdata
108
- else
109
- rcdata << arg.gsub(/&/, '&amp;')
110
- end
111
- }
112
- new_internal rcdata
113
- end
114
- end
115
- end