parse5 0.6.1 → 1.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (80) hide show
  1. package/.npmignore +3 -1
  2. package/LICENSE +1 -1
  3. package/README.md +166 -77
  4. package/index.js +7 -0
  5. package/lib/{html.js → const/html.js} +0 -0
  6. package/lib/{unicode.js → const/unicode.js} +0 -0
  7. package/lib/{named_entity_trie.js → tokenization_stage/named_entity_trie.js} +1 -1
  8. package/lib/tokenization_stage/parser_simulator.js +2 -0
  9. package/lib/{preprocessor.js → tokenization_stage/preprocessor.js} +5 -5
  10. package/lib/{tokenizer.js → tokenization_stage/tokenizer.js} +37 -37
  11. package/lib/{default_tree_adapter.js → tree_adapters/default.js} +47 -0
  12. package/lib/tree_adapters/htmlparser2.js +267 -0
  13. package/lib/{formatting_element_list.js → tree_construction_stage/formatting_element_list.js} +1 -1
  14. package/lib/{open_element_stack.js → tree_construction_stage/open_element_stack.js} +2 -2
  15. package/lib/{parser.js → tree_construction_stage/parser.js} +8 -8
  16. package/lib/tree_construction_stage/tree_serializer.js +165 -0
  17. package/package.json +33 -26
  18. package/test/data/tokenizer/contentModelFlags.test +0 -75
  19. package/test/data/tokenizer/domjs.test +0 -90
  20. package/test/data/tokenizer/entities.test +0 -283
  21. package/test/data/tokenizer/escapeFlag.test +0 -33
  22. package/test/data/tokenizer/namedEntities.test +0 -44189
  23. package/test/data/tokenizer/numericEntities.test +0 -1313
  24. package/test/data/tokenizer/pendingSpecChanges.test +0 -7
  25. package/test/data/tokenizer/test1.test +0 -196
  26. package/test/data/tokenizer/test2.test +0 -179
  27. package/test/data/tokenizer/test3.test +0 -6047
  28. package/test/data/tokenizer/test4.test +0 -344
  29. package/test/data/tokenizer/unicodeChars.test +0 -1295
  30. package/test/data/tokenizer/unicodeCharsProblematic.test +0 -27
  31. package/test/data/tree-construction/adoption01.dat +0 -275
  32. package/test/data/tree-construction/adoption02.dat +0 -31
  33. package/test/data/tree-construction/comments01.dat +0 -146
  34. package/test/data/tree-construction/doctype01.dat +0 -370
  35. package/test/data/tree-construction/domjs-unsafe.dat +0 -0
  36. package/test/data/tree-construction/entities01.dat +0 -603
  37. package/test/data/tree-construction/entities02.dat +0 -249
  38. package/test/data/tree-construction/html5test-com.dat +0 -246
  39. package/test/data/tree-construction/inbody01.dat +0 -43
  40. package/test/data/tree-construction/isindex.dat +0 -40
  41. package/test/data/tree-construction/main-element.dat +0 -29
  42. package/test/data/tree-construction/pending-spec-changes-plain-text-unsafe.dat +0 -0
  43. package/test/data/tree-construction/pending-spec-changes.dat +0 -52
  44. package/test/data/tree-construction/plain-text-unsafe.dat +0 -0
  45. package/test/data/tree-construction/scriptdata01.dat +0 -308
  46. package/test/data/tree-construction/tables01.dat +0 -212
  47. package/test/data/tree-construction/tests1.dat +0 -1952
  48. package/test/data/tree-construction/tests10.dat +0 -799
  49. package/test/data/tree-construction/tests11.dat +0 -482
  50. package/test/data/tree-construction/tests12.dat +0 -62
  51. package/test/data/tree-construction/tests14.dat +0 -74
  52. package/test/data/tree-construction/tests15.dat +0 -208
  53. package/test/data/tree-construction/tests16.dat +0 -2299
  54. package/test/data/tree-construction/tests17.dat +0 -153
  55. package/test/data/tree-construction/tests18.dat +0 -269
  56. package/test/data/tree-construction/tests19.dat +0 -1249
  57. package/test/data/tree-construction/tests2.dat +0 -763
  58. package/test/data/tree-construction/tests20.dat +0 -455
  59. package/test/data/tree-construction/tests21.dat +0 -257
  60. package/test/data/tree-construction/tests22.dat +0 -157
  61. package/test/data/tree-construction/tests23.dat +0 -155
  62. package/test/data/tree-construction/tests24.dat +0 -79
  63. package/test/data/tree-construction/tests25.dat +0 -219
  64. package/test/data/tree-construction/tests26.dat +0 -313
  65. package/test/data/tree-construction/tests3.dat +0 -305
  66. package/test/data/tree-construction/tests4.dat +0 -59
  67. package/test/data/tree-construction/tests5.dat +0 -191
  68. package/test/data/tree-construction/tests6.dat +0 -663
  69. package/test/data/tree-construction/tests7.dat +0 -390
  70. package/test/data/tree-construction/tests8.dat +0 -148
  71. package/test/data/tree-construction/tests9.dat +0 -457
  72. package/test/data/tree-construction/tests_innerHTML_1.dat +0 -741
  73. package/test/data/tree-construction/tricky01.dat +0 -261
  74. package/test/data/tree-construction/webkit01.dat +0 -594
  75. package/test/data/tree-construction/webkit02.dat +0 -94
  76. package/test/fixtures/formatting_element_list_test.js +0 -267
  77. package/test/fixtures/open_element_stack_test.js +0 -495
  78. package/test/fixtures/parser_test.js +0 -160
  79. package/test/fixtures/tokenizer_test.js +0 -178
  80. package/test/run_tests.js +0 -10
package/test/data/tree-construction/tricky01.dat DELETED
@@ -1,261 +0,0 @@
1
- #data
2
- <b><p>Bold </b> Not bold</p>
3
- Also not bold.
4
- #errors
5
- #document
6
- | <html>
7
- | <head>
8
- | <body>
9
- | <b>
10
- | <p>
11
- | <b>
12
- | "Bold "
13
- | " Not bold"
14
- | "
15
- Also not bold."
16
-
17
- #data
18
- <html>
19
- <font color=red><i>Italic and Red<p>Italic and Red </font> Just italic.</p> Italic only.</i> Plain
20
- <p>I should not be red. <font color=red>Red. <i>Italic and red.</p>
21
- <p>Italic and red. </i> Red.</font> I should not be red.</p>
22
- <b>Bold <i>Bold and italic</b> Only Italic </i> Plain
23
- #errors
24
- #document
25
- | <html>
26
- | <head>
27
- | <body>
28
- | <font>
29
- | color="red"
30
- | <i>
31
- | "Italic and Red"
32
- | <i>
33
- | <p>
34
- | <font>
35
- | color="red"
36
- | "Italic and Red "
37
- | " Just italic."
38
- | " Italic only."
39
- | " Plain
40
- "
41
- | <p>
42
- | "I should not be red. "
43
- | <font>
44
- | color="red"
45
- | "Red. "
46
- | <i>
47
- | "Italic and red."
48
- | <font>
49
- | color="red"
50
- | <i>
51
- | "
52
- "
53
- | <p>
54
- | <font>
55
- | color="red"
56
- | <i>
57
- | "Italic and red. "
58
- | " Red."
59
- | " I should not be red."
60
- | "
61
- "
62
- | <b>
63
- | "Bold "
64
- | <i>
65
- | "Bold and italic"
66
- | <i>
67
- | " Only Italic "
68
- | " Plain"
69
-
70
- #data
71
- <html><body>
72
- <p><font size="7">First paragraph.</p>
73
- <p>Second paragraph.</p></font>
74
- <b><p><i>Bold and Italic</b> Italic</p>
75
- #errors
76
- #document
77
- | <html>
78
- | <head>
79
- | <body>
80
- | "
81
- "
82
- | <p>
83
- | <font>
84
- | size="7"
85
- | "First paragraph."
86
- | <font>
87
- | size="7"
88
- | "
89
- "
90
- | <p>
91
- | "Second paragraph."
92
- | "
93
- "
94
- | <b>
95
- | <p>
96
- | <b>
97
- | <i>
98
- | "Bold and Italic"
99
- | <i>
100
- | " Italic"
101
-
102
- #data
103
- <html>
104
- <dl>
105
- <dt><b>Boo
106
- <dd>Goo?
107
- </dl>
108
- </html>
109
- #errors
110
- #document
111
- | <html>
112
- | <head>
113
- | <body>
114
- | <dl>
115
- | "
116
- "
117
- | <dt>
118
- | <b>
119
- | "Boo
120
- "
121
- | <dd>
122
- | <b>
123
- | "Goo?
124
- "
125
- | <b>
126
- | "
127
- "
128
-
129
- #data
130
- <html><body>
131
- <label><a><div>Hello<div>World</div></a></label>
132
- </body></html>
133
- #errors
134
- #document
135
- | <html>
136
- | <head>
137
- | <body>
138
- | "
139
- "
140
- | <label>
141
- | <a>
142
- | <div>
143
- | <a>
144
- | "Hello"
145
- | <div>
146
- | "World"
147
- | "
148
- "
149
-
150
- #data
151
- <table><center> <font>a</center> <img> <tr><td> </td> </tr> </table>
152
- #errors
153
- #document
154
- | <html>
155
- | <head>
156
- | <body>
157
- | <center>
158
- | " "
159
- | <font>
160
- | "a"
161
- | <font>
162
- | <img>
163
- | " "
164
- | <table>
165
- | " "
166
- | <tbody>
167
- | <tr>
168
- | <td>
169
- | " "
170
- | " "
171
- | " "
172
-
173
- #data
174
- <table><tr><p><a><p>You should see this text.
175
- #errors
176
- #document
177
- | <html>
178
- | <head>
179
- | <body>
180
- | <p>
181
- | <a>
182
- | <p>
183
- | <a>
184
- | "You should see this text."
185
- | <table>
186
- | <tbody>
187
- | <tr>
188
-
189
- #data
190
- <TABLE>
191
- <TR>
192
- <CENTER><CENTER><TD></TD></TR><TR>
193
- <FONT>
194
- <TABLE><tr></tr></TABLE>
195
- </P>
196
- <a></font><font></a>
197
- This page contains an insanely badly-nested tag sequence.
198
- #errors
199
- #document
200
- | <html>
201
- | <head>
202
- | <body>
203
- | <center>
204
- | <center>
205
- | <font>
206
- | "
207
- "
208
- | <table>
209
- | "
210
- "
211
- | <tbody>
212
- | <tr>
213
- | "
214
- "
215
- | <td>
216
- | <tr>
217
- | "
218
- "
219
- | <table>
220
- | <tbody>
221
- | <tr>
222
- | <font>
223
- | "
224
- "
225
- | <p>
226
- | "
227
- "
228
- | <a>
229
- | <a>
230
- | <font>
231
- | <font>
232
- | "
233
- This page contains an insanely badly-nested tag sequence."
234
-
235
- #data
236
- <html>
237
- <body>
238
- <b><nobr><div>This text is in a div inside a nobr</nobr>More text that should not be in the nobr, i.e., the
239
- nobr should have closed the div inside it implicitly. </b><pre>A pre tag outside everything else.</pre>
240
- </body>
241
- </html>
242
- #errors
243
- #document
244
- | <html>
245
- | <head>
246
- | <body>
247
- | "
248
- "
249
- | <b>
250
- | <nobr>
251
- | <div>
252
- | <b>
253
- | <nobr>
254
- | "This text is in a div inside a nobr"
255
- | "More text that should not be in the nobr, i.e., the
256
- nobr should have closed the div inside it implicitly. "
257
- | <pre>
258
- | "A pre tag outside everything else."
259
- | "
260
-
261
- "