nokogiri 1.11.3 → 1.13.8

Sign up to get free protection for your applications and to get access to all the features.

Potentially problematic release.


This version of nokogiri might be problematic. Click here for more details.

Files changed (179) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +2 -0
  3. data/LICENSE-DEPENDENCIES.md +243 -22
  4. data/LICENSE.md +1 -1
  5. data/README.md +14 -11
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +13 -64
  8. data/ext/nokogiri/depend +35 -34
  9. data/ext/nokogiri/extconf.rb +237 -133
  10. data/ext/nokogiri/gumbo.c +584 -0
  11. data/ext/nokogiri/{html_document.c → html4_document.c} +8 -8
  12. data/ext/nokogiri/{html_element_description.c → html4_element_description.c} +21 -19
  13. data/ext/nokogiri/{html_entity_lookup.c → html4_entity_lookup.c} +7 -7
  14. data/ext/nokogiri/{html_sax_parser_context.c → html4_sax_parser_context.c} +8 -8
  15. data/ext/nokogiri/{html_sax_push_parser.c → html4_sax_push_parser.c} +4 -4
  16. data/ext/nokogiri/libxml2_backwards_compat.c +30 -30
  17. data/ext/nokogiri/nokogiri.c +70 -38
  18. data/ext/nokogiri/nokogiri.h +27 -9
  19. data/ext/nokogiri/xml_attr.c +2 -2
  20. data/ext/nokogiri/xml_attribute_decl.c +3 -3
  21. data/ext/nokogiri/xml_cdata.c +1 -1
  22. data/ext/nokogiri/xml_document.c +50 -50
  23. data/ext/nokogiri/xml_document_fragment.c +0 -2
  24. data/ext/nokogiri/xml_dtd.c +10 -10
  25. data/ext/nokogiri/xml_element_content.c +2 -0
  26. data/ext/nokogiri/xml_element_decl.c +3 -3
  27. data/ext/nokogiri/xml_encoding_handler.c +31 -12
  28. data/ext/nokogiri/xml_entity_decl.c +5 -5
  29. data/ext/nokogiri/xml_namespace.c +4 -2
  30. data/ext/nokogiri/xml_node.c +833 -492
  31. data/ext/nokogiri/xml_node_set.c +24 -24
  32. data/ext/nokogiri/xml_reader.c +90 -11
  33. data/ext/nokogiri/xml_sax_parser.c +6 -6
  34. data/ext/nokogiri/xml_sax_parser_context.c +12 -3
  35. data/ext/nokogiri/xml_schema.c +5 -3
  36. data/ext/nokogiri/xml_text.c +1 -1
  37. data/ext/nokogiri/xml_xpath_context.c +110 -85
  38. data/ext/nokogiri/xslt_stylesheet.c +109 -10
  39. data/gumbo-parser/CHANGES.md +63 -0
  40. data/gumbo-parser/Makefile +101 -0
  41. data/gumbo-parser/THANKS +27 -0
  42. data/gumbo-parser/src/Makefile +34 -0
  43. data/gumbo-parser/src/README.md +41 -0
  44. data/gumbo-parser/src/ascii.c +75 -0
  45. data/gumbo-parser/src/ascii.h +115 -0
  46. data/gumbo-parser/src/attribute.c +42 -0
  47. data/gumbo-parser/src/attribute.h +17 -0
  48. data/gumbo-parser/src/char_ref.c +22225 -0
  49. data/gumbo-parser/src/char_ref.h +29 -0
  50. data/gumbo-parser/src/char_ref.rl +2154 -0
  51. data/gumbo-parser/src/error.c +626 -0
  52. data/gumbo-parser/src/error.h +148 -0
  53. data/gumbo-parser/src/foreign_attrs.c +104 -0
  54. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  55. data/gumbo-parser/src/gumbo.h +943 -0
  56. data/gumbo-parser/src/insertion_mode.h +33 -0
  57. data/gumbo-parser/src/macros.h +91 -0
  58. data/gumbo-parser/src/parser.c +4875 -0
  59. data/gumbo-parser/src/parser.h +41 -0
  60. data/gumbo-parser/src/replacement.h +33 -0
  61. data/gumbo-parser/src/string_buffer.c +103 -0
  62. data/gumbo-parser/src/string_buffer.h +68 -0
  63. data/gumbo-parser/src/string_piece.c +48 -0
  64. data/gumbo-parser/src/svg_attrs.c +174 -0
  65. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  66. data/gumbo-parser/src/svg_tags.c +137 -0
  67. data/gumbo-parser/src/svg_tags.gperf +55 -0
  68. data/gumbo-parser/src/tag.c +222 -0
  69. data/gumbo-parser/src/tag_lookup.c +382 -0
  70. data/gumbo-parser/src/tag_lookup.gperf +169 -0
  71. data/gumbo-parser/src/tag_lookup.h +13 -0
  72. data/gumbo-parser/src/token_buffer.c +79 -0
  73. data/gumbo-parser/src/token_buffer.h +71 -0
  74. data/gumbo-parser/src/token_type.h +17 -0
  75. data/gumbo-parser/src/tokenizer.c +3463 -0
  76. data/gumbo-parser/src/tokenizer.h +112 -0
  77. data/gumbo-parser/src/tokenizer_states.h +339 -0
  78. data/gumbo-parser/src/utf8.c +245 -0
  79. data/gumbo-parser/src/utf8.h +164 -0
  80. data/gumbo-parser/src/util.c +68 -0
  81. data/gumbo-parser/src/util.h +30 -0
  82. data/gumbo-parser/src/vector.c +111 -0
  83. data/gumbo-parser/src/vector.h +45 -0
  84. data/lib/nokogiri/class_resolver.rb +67 -0
  85. data/lib/nokogiri/css/node.rb +9 -8
  86. data/lib/nokogiri/css/parser.rb +361 -342
  87. data/lib/nokogiri/css/parser.y +250 -245
  88. data/lib/nokogiri/css/parser_extras.rb +22 -20
  89. data/lib/nokogiri/css/syntax_error.rb +2 -1
  90. data/lib/nokogiri/css/tokenizer.rb +4 -3
  91. data/lib/nokogiri/css/tokenizer.rex +3 -2
  92. data/lib/nokogiri/css/xpath_visitor.rb +179 -82
  93. data/lib/nokogiri/css.rb +49 -17
  94. data/lib/nokogiri/decorators/slop.rb +8 -7
  95. data/lib/nokogiri/extension.rb +8 -3
  96. data/lib/nokogiri/gumbo.rb +15 -0
  97. data/lib/nokogiri/html.rb +37 -27
  98. data/lib/nokogiri/{html → html4}/builder.rb +3 -2
  99. data/lib/nokogiri/{html → html4}/document.rb +92 -81
  100. data/lib/nokogiri/{html → html4}/document_fragment.rb +13 -9
  101. data/lib/nokogiri/{html → html4}/element_description.rb +2 -1
  102. data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
  103. data/lib/nokogiri/{html → html4}/entity_lookup.rb +3 -2
  104. data/lib/nokogiri/{html → html4}/sax/parser.rb +16 -16
  105. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  106. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +11 -11
  107. data/lib/nokogiri/html4.rb +46 -0
  108. data/lib/nokogiri/html5/document.rb +91 -0
  109. data/lib/nokogiri/html5/document_fragment.rb +83 -0
  110. data/lib/nokogiri/html5/node.rb +100 -0
  111. data/lib/nokogiri/html5.rb +478 -0
  112. data/lib/nokogiri/jruby/dependencies.rb +10 -9
  113. data/lib/nokogiri/syntax_error.rb +1 -0
  114. data/lib/nokogiri/version/constant.rb +2 -1
  115. data/lib/nokogiri/version/info.rb +31 -14
  116. data/lib/nokogiri/version.rb +1 -0
  117. data/lib/nokogiri/xml/attr.rb +5 -3
  118. data/lib/nokogiri/xml/attribute_decl.rb +2 -1
  119. data/lib/nokogiri/xml/builder.rb +71 -31
  120. data/lib/nokogiri/xml/cdata.rb +2 -1
  121. data/lib/nokogiri/xml/character_data.rb +1 -0
  122. data/lib/nokogiri/xml/document.rb +183 -96
  123. data/lib/nokogiri/xml/document_fragment.rb +41 -38
  124. data/lib/nokogiri/xml/dtd.rb +3 -2
  125. data/lib/nokogiri/xml/element_content.rb +1 -0
  126. data/lib/nokogiri/xml/element_decl.rb +2 -1
  127. data/lib/nokogiri/xml/entity_decl.rb +3 -2
  128. data/lib/nokogiri/xml/entity_reference.rb +1 -0
  129. data/lib/nokogiri/xml/namespace.rb +2 -0
  130. data/lib/nokogiri/xml/node/save_options.rb +9 -5
  131. data/lib/nokogiri/xml/node.rb +525 -354
  132. data/lib/nokogiri/xml/node_set.rb +50 -54
  133. data/lib/nokogiri/xml/notation.rb +12 -0
  134. data/lib/nokogiri/xml/parse_options.rb +13 -6
  135. data/lib/nokogiri/xml/pp/character_data.rb +8 -6
  136. data/lib/nokogiri/xml/pp/node.rb +24 -26
  137. data/lib/nokogiri/xml/pp.rb +3 -2
  138. data/lib/nokogiri/xml/processing_instruction.rb +2 -1
  139. data/lib/nokogiri/xml/reader.rb +20 -24
  140. data/lib/nokogiri/xml/relax_ng.rb +1 -0
  141. data/lib/nokogiri/xml/sax/document.rb +44 -49
  142. data/lib/nokogiri/xml/sax/parser.rb +37 -34
  143. data/lib/nokogiri/xml/sax/parser_context.rb +7 -3
  144. data/lib/nokogiri/xml/sax/push_parser.rb +5 -5
  145. data/lib/nokogiri/xml/sax.rb +5 -4
  146. data/lib/nokogiri/xml/schema.rb +7 -6
  147. data/lib/nokogiri/xml/searchable.rb +93 -62
  148. data/lib/nokogiri/xml/syntax_error.rb +5 -4
  149. data/lib/nokogiri/xml/text.rb +1 -0
  150. data/lib/nokogiri/xml/xpath/syntax_error.rb +2 -1
  151. data/lib/nokogiri/xml/xpath.rb +13 -1
  152. data/lib/nokogiri/xml/xpath_context.rb +2 -3
  153. data/lib/nokogiri/xml.rb +37 -37
  154. data/lib/nokogiri/xslt/stylesheet.rb +2 -1
  155. data/lib/nokogiri/xslt.rb +28 -20
  156. data/lib/nokogiri.rb +48 -43
  157. data/lib/xsd/xmlparser/nokogiri.rb +25 -24
  158. data/patches/libxml2/{0002-Remove-script-macro-support.patch → 0001-Remove-script-macro-support.patch} +0 -0
  159. data/patches/libxml2/{0003-Update-entities-to-remove-handling-of-ssi.patch → 0002-Update-entities-to-remove-handling-of-ssi.patch} +0 -0
  160. data/patches/libxml2/{0004-libxml2.la-is-in-top_builddir.patch → 0003-libxml2.la-is-in-top_builddir.patch} +1 -1
  161. data/patches/libxml2/{0008-use-glibc-strlen.patch → 0004-use-glibc-strlen.patch} +3 -3
  162. data/patches/libxml2/{0009-avoid-isnan-isinf.patch → 0005-avoid-isnan-isinf.patch} +4 -4
  163. data/patches/libxml2/0006-update-automake-files-for-arm64.patch +3040 -0
  164. data/patches/libxml2/0008-htmlParseComment-handle-abruptly-closed-comments.patch +61 -0
  165. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  166. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +2445 -1919
  167. data/ports/archives/libxml2-2.9.14.tar.xz +0 -0
  168. data/ports/archives/libxslt-1.1.35.tar.xz +0 -0
  169. metadata +204 -93
  170. data/lib/nokogiri/html/element_description_defaults.rb +0 -672
  171. data/lib/nokogiri/html/sax/parser_context.rb +0 -17
  172. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  173. data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +0 -32
  174. data/patches/libxml2/0006-htmlParseComment-treat-as-if-it-closed-the-comment.patch +0 -73
  175. data/patches/libxml2/0007-use-new-htmlParseLookupCommentEnd-to-find-comment-en.patch +0 -103
  176. data/patches/libxml2/0010-parser.c-shrink-the-input-buffer-when-appropriate.patch +0 -70
  177. data/patches/libxml2/0011-update-automake-files-for-arm64.patch +0 -2511
  178. data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
  179. data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
@@ -1,672 +0,0 @@
1
- # frozen_string_literal: true
2
- module Nokogiri
3
- module HTML
4
- class ElementDescription
5
-
6
- # Methods are defined protected by method_defined? because at
7
- # this point the C-library or Java library is already loaded,
8
- # and we don't want to clobber any methods that have been
9
- # defined there.
10
-
11
- Desc = Struct.new("HTMLElementDescription", :name,
12
- :startTag, :endTag, :saveEndTag,
13
- :empty, :depr, :dtd, :isinline,
14
- :desc,
15
- :subelts, :defaultsubelt,
16
- :attrs_opt, :attrs_depr, :attrs_req)
17
-
18
- # This is filled in down below.
19
- DefaultDescriptions = Hash.new()
20
-
21
- def default_desc
22
- DefaultDescriptions[name.downcase]
23
- end
24
- private :default_desc
25
-
26
- unless method_defined? :implied_start_tag?
27
- def implied_start_tag?
28
- d = default_desc
29
- d ? d.startTag : nil
30
- end
31
- end
32
-
33
- unless method_defined? :implied_end_tag?
34
- def implied_end_tag?
35
- d = default_desc
36
- d ? d.endTag : nil
37
- end
38
- end
39
-
40
- unless method_defined? :save_end_tag?
41
- def save_end_tag?
42
- d = default_desc
43
- d ? d.saveEndTag : nil
44
- end
45
- end
46
-
47
- unless method_defined? :deprecated?
48
- def deprecated?
49
- d = default_desc
50
- d ? d.depr : nil
51
- end
52
- end
53
-
54
- unless method_defined? :description
55
- def description
56
- d = default_desc
57
- d ? d.desc : nil
58
- end
59
- end
60
-
61
- unless method_defined? :default_sub_element
62
- def default_sub_element
63
- d = default_desc
64
- d ? d.defaultsubelt : nil
65
- end
66
- end
67
-
68
- unless method_defined? :optional_attributes
69
- def optional_attributes
70
- d = default_desc
71
- d ? d.attrs_opt : []
72
- end
73
- end
74
-
75
- unless method_defined? :deprecated_attributes
76
- def deprecated_attributes
77
- d = default_desc
78
- d ? d.attrs_depr : []
79
- end
80
- end
81
-
82
- unless method_defined? :required_attributes
83
- def required_attributes
84
- d = default_desc
85
- d ? d.attrs_req : []
86
- end
87
- end
88
-
89
- ###
90
- # Default Element Descriptions (HTML 4.0) copied from
91
- # libxml2/HTMLparser.c and libxml2/include/libxml/HTMLparser.h
92
- #
93
- # The copyright notice for those files and the following list of
94
- # element and attribute descriptions is reproduced here:
95
- #
96
- # Except where otherwise noted in the source code (e.g. the
97
- # files hash.c, list.c and the trio files, which are covered by
98
- # a similar licence but with different Copyright notices) all
99
- # the files are:
100
- #
101
- # Copyright (C) 1998-2003 Daniel Veillard. All Rights Reserved.
102
- #
103
- # Permission is hereby granted, free of charge, to any person
104
- # obtaining a copy of this software and associated documentation
105
- # files (the "Software"), to deal in the Software without
106
- # restriction, including without limitation the rights to use,
107
- # copy, modify, merge, publish, distribute, sublicense, and/or
108
- # sell copies of the Software, and to permit persons to whom the
109
- # Software is fur- nished to do so, subject to the following
110
- # conditions:
111
-
112
- # The above copyright notice and this permission notice shall be
113
- # included in all copies or substantial portions of the
114
- # Software.
115
-
116
- # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
117
- # KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
118
- # WARRANTIES OF MERCHANTABILITY, FIT- NESS FOR A PARTICULAR
119
- # PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE DANIEL
120
- # VEILLARD BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
121
- # WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
122
- # FROM, OUT OF OR IN CON- NECTION WITH THE SOFTWARE OR THE USE
123
- # OR OTHER DEALINGS IN THE SOFTWARE.
124
-
125
- # Except as contained in this notice, the name of Daniel
126
- # Veillard shall not be used in advertising or otherwise to
127
- # promote the sale, use or other deal- ings in this Software
128
- # without prior written authorization from him.
129
-
130
- # Attributes defined and categorized
131
- FONTSTYLE = ["tt", "i", "b", "u", "s", "strike", "big", "small"]
132
- PHRASE = ['em', 'strong', 'dfn', 'code', 'samp',
133
- 'kbd', 'var', 'cite', 'abbr', 'acronym']
134
- SPECIAL = ['a', 'img', 'applet', 'embed', 'object', 'font','basefont',
135
- 'br', 'script', 'map', 'q', 'sub', 'sup', 'span', 'bdo',
136
- 'iframe']
137
- PCDATA = []
138
- HEADING = ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']
139
- LIST = ['ul', 'ol', 'dir', 'menu']
140
- FORMCTRL = ['input', 'select', 'textarea', 'label', 'button']
141
- BLOCK = [HEADING, LIST, 'pre', 'p', 'dl', 'div', 'center', 'noscript',
142
- 'noframes', 'blockquote', 'form', 'isindex', 'hr', 'table',
143
- 'fieldset', 'address']
144
- INLINE = [PCDATA, FONTSTYLE, PHRASE, SPECIAL, FORMCTRL]
145
- FLOW = [BLOCK, INLINE]
146
- MODIFIER = []
147
- EMPTY = []
148
-
149
- HTML_FLOW = FLOW
150
- HTML_INLINE = INLINE
151
- HTML_PCDATA = PCDATA
152
- HTML_CDATA = HTML_PCDATA
153
-
154
- COREATTRS = ['id', 'class', 'style', 'title']
155
- I18N = ['lang', 'dir']
156
- EVENTS = ['onclick', 'ondblclick', 'onmousedown', 'onmouseup',
157
- 'onmouseover', 'onmouseout', 'onkeypress', 'onkeydown',
158
- 'onkeyup']
159
- ATTRS = [COREATTRS, I18N,EVENTS]
160
- CELLHALIGN = ['align', 'char', 'charoff']
161
- CELLVALIGN = ['valign']
162
-
163
- HTML_ATTRS = ATTRS
164
- CORE_I18N_ATTRS = [COREATTRS, I18N]
165
- CORE_ATTRS = COREATTRS
166
- I18N_ATTRS = I18N
167
-
168
-
169
- A_ATTRS = [ATTRS, 'charset', 'type', 'name',
170
- 'href', 'hreflang', 'rel', 'rev', 'accesskey', 'shape',
171
- 'coords', 'tabindex', 'onfocus', 'onblur']
172
- TARGET_ATTR = ['target']
173
- ROWS_COLS_ATTR = ['rows', 'cols']
174
- ALT_ATTR = ['alt']
175
- SRC_ALT_ATTRS = ['src', 'alt']
176
- HREF_ATTRS = ['href']
177
- CLEAR_ATTRS = ['clear']
178
- INLINE_P = [INLINE, 'p']
179
-
180
- FLOW_PARAM = [FLOW, 'param']
181
- APPLET_ATTRS = [COREATTRS , 'codebase',
182
- 'archive', 'alt', 'name', 'height', 'width', 'align',
183
- 'hspace', 'vspace']
184
- AREA_ATTRS = ['shape', 'coords', 'href', 'nohref',
185
- 'tabindex', 'accesskey', 'onfocus', 'onblur']
186
- BASEFONT_ATTRS = ['id', 'size', 'color', 'face']
187
- QUOTE_ATTRS = [ATTRS, 'cite']
188
- BODY_CONTENTS = [FLOW, 'ins', 'del']
189
- BODY_ATTRS = [ATTRS, 'onload', 'onunload']
190
- BODY_DEPR = ['background', 'bgcolor', 'text',
191
- 'link', 'vlink', 'alink']
192
- BUTTON_ATTRS = [ATTRS, 'name', 'value', 'type',
193
- 'disabled', 'tabindex', 'accesskey', 'onfocus', 'onblur']
194
-
195
-
196
- COL_ATTRS = [ATTRS, 'span', 'width', CELLHALIGN, CELLVALIGN]
197
- COL_ELT = ['col']
198
- EDIT_ATTRS = [ATTRS, 'datetime', 'cite']
199
- COMPACT_ATTRS = [ATTRS, 'compact']
200
- DL_CONTENTS = ['dt', 'dd']
201
- COMPACT_ATTR = ['compact']
202
- LABEL_ATTR = ['label']
203
- FIELDSET_CONTENTS = [FLOW, 'legend' ]
204
- FONT_ATTRS = [COREATTRS, I18N, 'size', 'color', 'face' ]
205
- FORM_CONTENTS = [HEADING, LIST, INLINE, 'pre', 'p', 'div', 'center',
206
- 'noscript', 'noframes', 'blockquote', 'isindex', 'hr',
207
- 'table', 'fieldset', 'address']
208
- FORM_ATTRS = [ATTRS, 'method', 'enctype', 'accept', 'name', 'onsubmit',
209
- 'onreset', 'accept-charset']
210
- FRAME_ATTRS = [COREATTRS, 'longdesc', 'name', 'src', 'frameborder',
211
- 'marginwidth', 'marginheight', 'noresize', 'scrolling' ]
212
- FRAMESET_ATTRS = [COREATTRS, 'rows', 'cols', 'onload', 'onunload']
213
- FRAMESET_CONTENTS = ['frameset', 'frame', 'noframes']
214
- HEAD_ATTRS = [I18N, 'profile']
215
- HEAD_CONTENTS = ['title', 'isindex', 'base', 'script', 'style', 'meta',
216
- 'link', 'object']
217
- HR_DEPR = ['align', 'noshade', 'size', 'width']
218
- VERSION_ATTR = ['version']
219
- HTML_CONTENT = ['head', 'body', 'frameset']
220
- IFRAME_ATTRS = [COREATTRS, 'longdesc', 'name', 'src', 'frameborder',
221
- 'marginwidth', 'marginheight', 'scrolling', 'align',
222
- 'height', 'width']
223
- IMG_ATTRS = [ATTRS, 'longdesc', 'name', 'height', 'width', 'usemap',
224
- 'ismap']
225
- EMBED_ATTRS = [COREATTRS, 'align', 'alt', 'border', 'code', 'codebase',
226
- 'frameborder', 'height', 'hidden', 'hspace', 'name',
227
- 'palette', 'pluginspace', 'pluginurl', 'src', 'type',
228
- 'units', 'vspace', 'width']
229
- INPUT_ATTRS = [ATTRS, 'type', 'name', 'value', 'checked', 'disabled',
230
- 'readonly', 'size', 'maxlength', 'src', 'alt', 'usemap',
231
- 'ismap', 'tabindex', 'accesskey', 'onfocus', 'onblur',
232
- 'onselect', 'onchange', 'accept']
233
- PROMPT_ATTRS = [COREATTRS, I18N, 'prompt']
234
- LABEL_ATTRS = [ATTRS, 'for', 'accesskey', 'onfocus', 'onblur']
235
- LEGEND_ATTRS = [ATTRS, 'accesskey']
236
- ALIGN_ATTR = ['align']
237
- LINK_ATTRS = [ATTRS, 'charset', 'href', 'hreflang', 'type', 'rel', 'rev',
238
- 'media']
239
- MAP_CONTENTS = [BLOCK, 'area']
240
- NAME_ATTR = ['name']
241
- ACTION_ATTR = ['action']
242
- BLOCKLI_ELT = [BLOCK, 'li']
243
- META_ATTRS = [I18N, 'http-equiv', 'name', 'scheme']
244
- CONTENT_ATTR = ['content']
245
- TYPE_ATTR = ['type']
246
- NOFRAMES_CONTENT = ['body', FLOW, MODIFIER]
247
- OBJECT_CONTENTS = [FLOW, 'param']
248
- OBJECT_ATTRS = [ATTRS, 'declare', 'classid', 'codebase', 'data', 'type',
249
- 'codetype', 'archive', 'standby', 'height', 'width',
250
- 'usemap', 'name', 'tabindex']
251
- OBJECT_DEPR = ['align', 'border', 'hspace', 'vspace']
252
- OL_ATTRS = ['type', 'compact', 'start']
253
- OPTION_ELT = ['option']
254
- OPTGROUP_ATTRS = [ATTRS, 'disabled']
255
- OPTION_ATTRS = [ATTRS, 'disabled', 'label', 'selected', 'value']
256
- PARAM_ATTRS = ['id', 'value', 'valuetype', 'type']
257
- WIDTH_ATTR = ['width']
258
- PRE_CONTENT = [PHRASE, 'tt', 'i', 'b', 'u', 's', 'strike', 'a', 'br',
259
- 'script', 'map', 'q', 'span', 'bdo', 'iframe']
260
- SCRIPT_ATTRS = ['charset', 'src', 'defer', 'event', 'for']
261
- LANGUAGE_ATTR = ['language']
262
- SELECT_CONTENT = ['optgroup', 'option']
263
- SELECT_ATTRS = [ATTRS, 'name', 'size', 'multiple', 'disabled', 'tabindex',
264
- 'onfocus', 'onblur', 'onchange']
265
- STYLE_ATTRS = [I18N, 'media', 'title']
266
- TABLE_ATTRS = [ATTRS, 'summary', 'width', 'border', 'frame', 'rules',
267
- 'cellspacing', 'cellpadding', 'datapagesize']
268
- TABLE_DEPR = ['align', 'bgcolor']
269
- TABLE_CONTENTS = ['caption', 'col', 'colgroup', 'thead', 'tfoot', 'tbody',
270
- 'tr']
271
- TR_ELT = ['tr']
272
- TALIGN_ATTRS = [ATTRS, CELLHALIGN, CELLVALIGN]
273
- TH_TD_DEPR = ['nowrap', 'bgcolor', 'width', 'height']
274
- TH_TD_ATTR = [ATTRS, 'abbr', 'axis', 'headers', 'scope', 'rowspan',
275
- 'colspan', CELLHALIGN, CELLVALIGN]
276
- TEXTAREA_ATTRS = [ATTRS, 'name', 'disabled', 'readonly', 'tabindex',
277
- 'accesskey', 'onfocus', 'onblur', 'onselect',
278
- 'onchange']
279
- TR_CONTENTS = ['th', 'td']
280
- BGCOLOR_ATTR = ['bgcolor']
281
- LI_ELT = ['li']
282
- UL_DEPR = ['type', 'compact']
283
- DIR_ATTR = ['dir']
284
-
285
- [
286
- ['a', false, false, false, false, false, :any, true,
287
- 'anchor ',
288
- HTML_INLINE, nil, A_ATTRS, TARGET_ATTR, []
289
- ],
290
- ['abbr', false, false, false, false, false, :any, true,
291
- 'abbreviated form',
292
- HTML_INLINE, nil, HTML_ATTRS, [], []
293
- ],
294
- ['acronym', false, false, false, false, false, :any, true, '',
295
- HTML_INLINE, nil, HTML_ATTRS, [], []
296
- ],
297
- ['address', false, false, false, false, false, :any, false,
298
- 'information on author',
299
- INLINE_P , nil, HTML_ATTRS, [], []
300
- ],
301
- ['applet', false, false, false, false, true, :loose, true,
302
- 'java applet ',
303
- FLOW_PARAM, nil, [], APPLET_ATTRS, []
304
- ],
305
- ['area', false, true, true, true, false, :any, false,
306
- 'client-side image map area ',
307
- EMPTY, nil, AREA_ATTRS, TARGET_ATTR, ALT_ATTR
308
- ],
309
- ['b', false, true, false, false, false, :any, true,
310
- 'bold text style',
311
- HTML_INLINE, nil, HTML_ATTRS, [], []
312
- ],
313
- ['base', false, true, true, true, false, :any, false,
314
- 'document base uri ',
315
- EMPTY, nil, [], TARGET_ATTR, HREF_ATTRS
316
- ],
317
- ['basefont', false, true, true, true, true, :loose, true,
318
- 'base font size ',
319
- EMPTY, nil, [], BASEFONT_ATTRS, []
320
- ],
321
- ['bdo', false, false, false, false, false, :any, true,
322
- 'i18n bidi over-ride ',
323
- HTML_INLINE, nil, CORE_I18N_ATTRS, [], DIR_ATTR
324
- ],
325
- ['big', false, true, false, false, false, :any, true,
326
- 'large text style',
327
- HTML_INLINE, nil, HTML_ATTRS, [], []
328
- ],
329
- ['blockquote', false, false, false, false, false, :any, false,
330
- 'long quotation ',
331
- HTML_FLOW, nil, QUOTE_ATTRS, [], []
332
- ],
333
- ['body', true, true, false, false, false, :any, false,
334
- 'document body ',
335
- BODY_CONTENTS, 'div', BODY_ATTRS, BODY_DEPR, []
336
- ],
337
- ['br', false, true, true, true, false, :any, true,
338
- 'forced line break ',
339
- EMPTY, nil, CORE_ATTRS, CLEAR_ATTRS, []
340
- ],
341
- ['button', false, false, false, false, false, :any, true,
342
- 'push button ',
343
- [HTML_FLOW, MODIFIER], nil, BUTTON_ATTRS, [], []
344
- ],
345
- ['caption', false, false, false, false, false, :any, false,
346
- 'table caption ',
347
- HTML_INLINE, nil, HTML_ATTRS, [], []
348
- ],
349
- ['center', false, true, false, false, true, :loose, false,
350
- 'shorthand for div align=center ',
351
- HTML_FLOW, nil, [], HTML_ATTRS, []
352
- ],
353
- ['cite', false, false, false, false, false, :any, true, 'citation',
354
- HTML_INLINE, nil, HTML_ATTRS, [], []
355
- ],
356
- ['code', false, false, false, false, false, :any, true,
357
- 'computer code fragment',
358
- HTML_INLINE, nil, HTML_ATTRS, [], []
359
- ],
360
- ['col', false, true, true, true, false, :any, false, 'table column ',
361
- EMPTY, nil, COL_ATTRS, [], []
362
- ],
363
- ['colgroup', false, true, false, false, false, :any, false,
364
- 'table column group ',
365
- COL_ELT, 'col', COL_ATTRS, [], []
366
- ],
367
- ['dd', false, true, false, false, false, :any, false,
368
- 'definition description ',
369
- HTML_FLOW, nil, HTML_ATTRS, [], []
370
- ],
371
- ['del', false, false, false, false, false, :any, true,
372
- 'deleted text ',
373
- HTML_FLOW, nil, EDIT_ATTRS, [], []
374
- ],
375
- ['dfn', false, false, false, false, false, :any, true,
376
- 'instance definition',
377
- HTML_INLINE, nil, HTML_ATTRS, [], []
378
- ],
379
- ['dir', false, false, false, false, true, :loose, false,
380
- 'directory list',
381
- BLOCKLI_ELT, 'li', [], COMPACT_ATTRS, []
382
- ],
383
- ['div', false, false, false, false, false, :any, false,
384
- 'generic language/style container',
385
- HTML_FLOW, nil, HTML_ATTRS, ALIGN_ATTR, []
386
- ],
387
- ['dl', false, false, false, false, false, :any, false,
388
- 'definition list ',
389
- DL_CONTENTS, 'dd', HTML_ATTRS, COMPACT_ATTR, []
390
- ],
391
- ['dt', false, true, false, false, false, :any, false,
392
- 'definition term ',
393
- HTML_INLINE, nil, HTML_ATTRS, [], []
394
- ],
395
- ['em', false, true, false, false, false, :any, true,
396
- 'emphasis',
397
- HTML_INLINE, nil, HTML_ATTRS, [], []
398
- ],
399
- ['embed', false, true, false, false, true, :loose, true,
400
- 'generic embedded object ',
401
- EMPTY, nil, EMBED_ATTRS, [], []
402
- ],
403
- ['fieldset', false, false, false, false, false, :any, false,
404
- 'form control group ',
405
- FIELDSET_CONTENTS, nil, HTML_ATTRS, [], []
406
- ],
407
- ['font', false, true, false, false, true, :loose, true,
408
- 'local change to font ',
409
- HTML_INLINE, nil, [], FONT_ATTRS, []
410
- ],
411
- ['form', false, false, false, false, false, :any, false,
412
- 'interactive form ',
413
- FORM_CONTENTS, 'fieldset', FORM_ATTRS, TARGET_ATTR, ACTION_ATTR
414
- ],
415
- ['frame', false, true, true, true, false, :frameset, false,
416
- 'subwindow ',
417
- EMPTY, nil, [], FRAME_ATTRS, []
418
- ],
419
- ['frameset', false, false, false, false, false, :frameset, false,
420
- 'window subdivision',
421
- FRAMESET_CONTENTS, 'noframes', [], FRAMESET_ATTRS, []
422
- ],
423
- ['htrue', false, false, false, false, false, :any, false,
424
- 'heading ',
425
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
426
- ],
427
- ['htrue', false, false, false, false, false, :any, false,
428
- 'heading ',
429
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
430
- ],
431
- ['htrue', false, false, false, false, false, :any, false,
432
- 'heading ',
433
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
434
- ],
435
- ['h4', false, false, false, false, false, :any, false,
436
- 'heading ',
437
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
438
- ],
439
- ['h5', false, false, false, false, false, :any, false,
440
- 'heading ',
441
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
442
- ],
443
- ['h6', false, false, false, false, false, :any, false,
444
- 'heading ',
445
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
446
- ],
447
- ['head', true, true, false, false, false, :any, false,
448
- 'document head ',
449
- HEAD_CONTENTS, nil, HEAD_ATTRS, [], []
450
- ],
451
- ['hr', false, true, true, true, false, :any, false,
452
- 'horizontal rule ',
453
- EMPTY, nil, HTML_ATTRS, HR_DEPR, []
454
- ],
455
- ['html', true, true, false, false, false, :any, false,
456
- 'document root element ',
457
- HTML_CONTENT, nil, I18N_ATTRS, VERSION_ATTR, []
458
- ],
459
- ['i', false, true, false, false, false, :any, true,
460
- 'italic text style',
461
- HTML_INLINE, nil, HTML_ATTRS, [], []
462
- ],
463
- ['iframe', false, false, false, false, false, :any, true,
464
- 'inline subwindow ',
465
- HTML_FLOW, nil, [], IFRAME_ATTRS, []
466
- ],
467
- ['img', false, true, true, true, false, :any, true,
468
- 'embedded image ',
469
- EMPTY, nil, IMG_ATTRS, ALIGN_ATTR, SRC_ALT_ATTRS
470
- ],
471
- ['input', false, true, true, true, false, :any, true,
472
- 'form control ',
473
- EMPTY, nil, INPUT_ATTRS, ALIGN_ATTR, []
474
- ],
475
- ['ins', false, false, false, false, false, :any, true,
476
- 'inserted text',
477
- HTML_FLOW, nil, EDIT_ATTRS, [], []
478
- ],
479
- ['isindex', false, true, true, true, true, :loose, false,
480
- 'single line prompt ',
481
- EMPTY, nil, [], PROMPT_ATTRS, []
482
- ],
483
- ['kbd', false, false, false, false, false, :any, true,
484
- 'text to be entered by the user',
485
- HTML_INLINE, nil, HTML_ATTRS, [], []
486
- ],
487
- ['label', false, false, false, false, false, :any, true,
488
- 'form field label text ',
489
- [HTML_INLINE, MODIFIER], nil, LABEL_ATTRS, [], []
490
- ],
491
- ['legend', false, false, false, false, false, :any, false,
492
- 'fieldset legend ',
493
- HTML_INLINE, nil, LEGEND_ATTRS, ALIGN_ATTR, []
494
- ],
495
- ['li', false, true, true, false, false, :any, false,
496
- 'list item ',
497
- HTML_FLOW, nil, HTML_ATTRS, [], []
498
- ],
499
- ['link', false, true, true, true, false, :any, false,
500
- 'a media-independent link ',
501
- EMPTY, nil, LINK_ATTRS, TARGET_ATTR, []
502
- ],
503
- ['map', false, false, false, false, false, :any, true,
504
- 'client-side image map ',
505
- MAP_CONTENTS, nil, HTML_ATTRS, [], NAME_ATTR
506
- ],
507
- ['menu', false, false, false, false, true, :loose, false,
508
- 'menu list ',
509
- BLOCKLI_ELT, nil, [], COMPACT_ATTRS, []
510
- ],
511
- ['meta', false, true, true, true, false, :any, false,
512
- 'generic metainformation ',
513
- EMPTY, nil, META_ATTRS, [], CONTENT_ATTR
514
- ],
515
- ['noframes', false, false, false, false, false, :frameset, false,
516
- 'alternate content container for non frame-based rendering ',
517
- NOFRAMES_CONTENT, 'body', HTML_ATTRS, [], []
518
- ],
519
- ['noscript', false, false, false, false, false, :any, false,
520
- 'alternate content container for non script-based rendering ',
521
- HTML_FLOW, 'div', HTML_ATTRS, [], []
522
- ],
523
- ['object', false, false, false, false, false, :any, true,
524
- 'generic embedded object ',
525
- OBJECT_CONTENTS, 'div', OBJECT_ATTRS, OBJECT_DEPR, []
526
- ],
527
- ['ol', false, false, false, false, false, :any, false,
528
- 'ordered list ',
529
- LI_ELT, 'li', HTML_ATTRS, OL_ATTRS, []
530
- ],
531
- ['optgroup', false, false, false, false, false, :any, false,
532
- 'option group ',
533
- OPTION_ELT, 'option', OPTGROUP_ATTRS, [], LABEL_ATTR
534
- ],
535
- ['option', false, true, false, false, false, :any, false,
536
- 'selectable choice ',
537
- HTML_PCDATA, nil, OPTION_ATTRS, [], []
538
- ],
539
- ['p', false, true, false, false, false, :any, false,
540
- 'paragraph ',
541
- HTML_INLINE, nil, HTML_ATTRS, ALIGN_ATTR, []
542
- ],
543
- ['param', false, true, true, true, false, :any, false,
544
- 'named property value ',
545
- EMPTY, nil, PARAM_ATTRS, [], NAME_ATTR
546
- ],
547
- ['pre', false, false, false, false, false, :any, false,
548
- 'preformatted text ',
549
- PRE_CONTENT, nil, HTML_ATTRS, WIDTH_ATTR, []
550
- ],
551
- ['q', false, false, false, false, false, :any, true,
552
- 'short inline quotation ',
553
- HTML_INLINE, nil, QUOTE_ATTRS, [], []
554
- ],
555
- ['s', false, true, false, false, true, :loose, true,
556
- 'strike-through text style',
557
- HTML_INLINE, nil, [], HTML_ATTRS, []
558
- ],
559
- ['samp', false, false, false, false, false, :any, true,
560
- 'sample program output, scripts, etc.',
561
- HTML_INLINE, nil, HTML_ATTRS, [], []
562
- ],
563
- ['script', false, false, false, false, false, :any, true,
564
- 'script statements ',
565
- HTML_CDATA, nil, SCRIPT_ATTRS, LANGUAGE_ATTR, TYPE_ATTR
566
- ],
567
- ['select', false, false, false, false, false, :any, true,
568
- 'option selector ',
569
- SELECT_CONTENT, nil, SELECT_ATTRS, [], []
570
- ],
571
- ['small', false, true, false, false, false, :any, true,
572
- 'small text style',
573
- HTML_INLINE, nil, HTML_ATTRS, [], []
574
- ],
575
- ['span', false, false, false, false, false, :any, true,
576
- 'generic language/style container ',
577
- HTML_INLINE, nil, HTML_ATTRS, [], []
578
- ],
579
- ['strike', false, true, false, false, true, :loose, true,
580
- 'strike-through text',
581
- HTML_INLINE, nil, [], HTML_ATTRS, []
582
- ],
583
- ['strong', false, true, false, false, false, :any, true,
584
- 'strong emphasis',
585
- HTML_INLINE, nil, HTML_ATTRS, [], []
586
- ],
587
- ['style', false, false, false, false, false, :any, false,
588
- 'style info ',
589
- HTML_CDATA, nil, STYLE_ATTRS, [], TYPE_ATTR
590
- ],
591
- ['sub', false, true, false, false, false, :any, true,
592
- 'subscript',
593
- HTML_INLINE, nil, HTML_ATTRS, [], []
594
- ],
595
- ['sup', false, true, false, false, false, :any, true,
596
- 'superscript ',
597
- HTML_INLINE, nil, HTML_ATTRS, [], []
598
- ],
599
- ['table', false, false, false, false, false, :any, false,
600
- '',
601
- TABLE_CONTENTS, 'tr', TABLE_ATTRS, TABLE_DEPR, []
602
- ],
603
- ['tbody', true, false, false, false, false, :any, false,
604
- 'table body ',
605
- TR_ELT, 'tr', TALIGN_ATTRS, [], []
606
- ],
607
- ['td', false, false, false, false, false, :any, false,
608
- 'table data cell',
609
- HTML_FLOW, nil, TH_TD_ATTR, TH_TD_DEPR, []
610
- ],
611
- ['textarea', false, false, false, false, false, :any, true,
612
- 'multi-line text field ',
613
- HTML_PCDATA, nil, TEXTAREA_ATTRS, [], ROWS_COLS_ATTR
614
- ],
615
- ['tfoot', false, true, false, false, false, :any, false,
616
- 'table footer ',
617
- TR_ELT, 'tr', TALIGN_ATTRS, [], []
618
- ],
619
- ['th', false, true, false, false, false, :any, false,
620
- 'table header cell',
621
- HTML_FLOW, nil, TH_TD_ATTR, TH_TD_DEPR, []
622
- ],
623
- ['thead', false, true, false, false, false, :any, false,
624
- 'table header ',
625
- TR_ELT, 'tr', TALIGN_ATTRS, [], []
626
- ],
627
- ['title', false, false, false, false, false, :any, false,
628
- 'document title ',
629
- HTML_PCDATA, nil, I18N_ATTRS, [], []
630
- ],
631
- ['tr', false, false, false, false, false, :any, false,
632
- 'table row ',
633
- TR_CONTENTS, 'td', TALIGN_ATTRS, BGCOLOR_ATTR, []
634
- ],
635
- ['tt', false, true, false, false, false, :any, true,
636
- 'teletype or monospaced text style',
637
- HTML_INLINE, nil, HTML_ATTRS, [], []
638
- ],
639
- ['u', false, true, false, false, true, :loose, true,
640
- 'underlined text style',
641
- HTML_INLINE, nil, [], HTML_ATTRS, []
642
- ],
643
- ['ul', false, false, false, false, false, :any, false,
644
- 'unordered list ',
645
- LI_ELT, 'li', HTML_ATTRS, UL_DEPR, []
646
- ],
647
- ['var', false, false, false, false, false, :any, true,
648
- 'instance of a variable or program argument',
649
- HTML_INLINE, nil, HTML_ATTRS, [], []
650
- ]
651
- ].each do |descriptor|
652
- name = descriptor[0]
653
-
654
- begin
655
- d = Desc.new(*descriptor)
656
-
657
- # flatten all the attribute lists (Ruby1.9, *[a,b,c] can be
658
- # used to flatten a literal list, but not in Ruby1.8).
659
- d[:subelts] = d[:subelts].flatten
660
- d[:attrs_opt] = d[:attrs_opt].flatten
661
- d[:attrs_depr] = d[:attrs_depr].flatten
662
- d[:attrs_req] = d[:attrs_req].flatten
663
- rescue => e
664
- p name
665
- raise e
666
- end
667
-
668
- DefaultDescriptions[name] = d
669
- end
670
- end
671
- end
672
- end
@@ -1,17 +0,0 @@
1
- # frozen_string_literal: true
2
- module Nokogiri
3
- module HTML
4
- module SAX
5
- ###
6
- # Context for HTML SAX parsers. This class is usually not instantiated
7
- # by the user. Instead, you should be looking at
8
- # Nokogiri::HTML::SAX::Parser
9
- class ParserContext < Nokogiri::XML::SAX::ParserContext
10
- def self.new thing, encoding = 'UTF-8'
11
- [:read, :close].all? { |x| thing.respond_to?(x) } ? super :
12
- memory(thing, encoding)
13
- end
14
- end
15
- end
16
- end
17
- end