parse5 6.0.1 → 7.3.0

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 (86) hide show
  1. package/README.md +2 -2
  2. package/dist/cjs/common/doctype.d.ts +4 -0
  3. package/dist/cjs/common/doctype.js +118 -0
  4. package/dist/cjs/common/error-codes.d.ts +67 -0
  5. package/dist/cjs/common/error-codes.js +66 -0
  6. package/dist/cjs/common/foreign-content.d.ts +9 -0
  7. package/dist/cjs/common/foreign-content.js +237 -0
  8. package/dist/cjs/common/html.d.ts +289 -0
  9. package/dist/cjs/common/html.js +528 -0
  10. package/dist/cjs/common/token.d.ts +84 -0
  11. package/dist/cjs/common/token.js +24 -0
  12. package/dist/cjs/common/unicode.d.ts +42 -0
  13. package/dist/cjs/common/unicode.js +70 -0
  14. package/dist/cjs/index.d.ts +71 -0
  15. package/dist/cjs/index.js +54 -0
  16. package/dist/cjs/package.json +1 -0
  17. package/dist/cjs/parser/formatting-element-list.d.ts +36 -0
  18. package/dist/cjs/parser/formatting-element-list.js +114 -0
  19. package/dist/cjs/parser/index.d.ts +221 -0
  20. package/dist/cjs/parser/index.js +3240 -0
  21. package/dist/cjs/parser/open-element-stack.d.ts +53 -0
  22. package/dist/cjs/parser/open-element-stack.js +328 -0
  23. package/dist/cjs/serializer/index.d.ts +60 -0
  24. package/dist/cjs/serializer/index.js +172 -0
  25. package/dist/cjs/tokenizer/index.d.ts +247 -0
  26. package/dist/cjs/tokenizer/index.js +2714 -0
  27. package/dist/cjs/tokenizer/preprocessor.d.ts +36 -0
  28. package/dist/cjs/tokenizer/preprocessor.js +200 -0
  29. package/dist/cjs/tree-adapters/default.d.ts +84 -0
  30. package/dist/cjs/tree-adapters/default.js +176 -0
  31. package/dist/cjs/tree-adapters/interface.d.ts +255 -0
  32. package/dist/cjs/tree-adapters/interface.js +2 -0
  33. package/dist/common/doctype.d.ts +4 -0
  34. package/{lib → dist}/common/doctype.js +24 -72
  35. package/dist/common/error-codes.d.ts +67 -0
  36. package/dist/common/error-codes.js +63 -0
  37. package/dist/common/foreign-content.d.ts +9 -0
  38. package/dist/common/foreign-content.js +228 -0
  39. package/dist/common/html.d.ts +289 -0
  40. package/dist/common/html.js +523 -0
  41. package/dist/common/token.d.ts +84 -0
  42. package/dist/common/token.js +20 -0
  43. package/dist/common/unicode.d.ts +42 -0
  44. package/dist/common/unicode.js +62 -0
  45. package/dist/index.d.ts +71 -0
  46. package/dist/index.js +42 -0
  47. package/dist/parser/formatting-element-list.d.ts +36 -0
  48. package/dist/parser/formatting-element-list.js +110 -0
  49. package/dist/parser/index.d.ts +221 -0
  50. package/dist/parser/index.js +3245 -0
  51. package/dist/parser/open-element-stack.d.ts +53 -0
  52. package/dist/parser/open-element-stack.js +324 -0
  53. package/dist/serializer/index.d.ts +60 -0
  54. package/dist/serializer/index.js +168 -0
  55. package/dist/tokenizer/index.d.ts +247 -0
  56. package/dist/tokenizer/index.js +2710 -0
  57. package/dist/tokenizer/preprocessor.d.ts +36 -0
  58. package/dist/tokenizer/preprocessor.js +196 -0
  59. package/dist/tree-adapters/default.d.ts +84 -0
  60. package/dist/tree-adapters/default.js +173 -0
  61. package/dist/tree-adapters/interface.d.ts +255 -0
  62. package/dist/tree-adapters/interface.js +1 -0
  63. package/package.json +21 -6
  64. package/lib/common/error-codes.js +0 -65
  65. package/lib/common/foreign-content.js +0 -265
  66. package/lib/common/html.js +0 -272
  67. package/lib/common/unicode.js +0 -109
  68. package/lib/extensions/error-reporting/mixin-base.js +0 -43
  69. package/lib/extensions/error-reporting/parser-mixin.js +0 -52
  70. package/lib/extensions/error-reporting/preprocessor-mixin.js +0 -24
  71. package/lib/extensions/error-reporting/tokenizer-mixin.js +0 -17
  72. package/lib/extensions/location-info/open-element-stack-mixin.js +0 -35
  73. package/lib/extensions/location-info/parser-mixin.js +0 -223
  74. package/lib/extensions/location-info/tokenizer-mixin.js +0 -146
  75. package/lib/extensions/position-tracking/preprocessor-mixin.js +0 -64
  76. package/lib/index.js +0 -29
  77. package/lib/parser/formatting-element-list.js +0 -181
  78. package/lib/parser/index.js +0 -2956
  79. package/lib/parser/open-element-stack.js +0 -482
  80. package/lib/serializer/index.js +0 -176
  81. package/lib/tokenizer/index.js +0 -2196
  82. package/lib/tokenizer/named-entity-data.js +0 -5
  83. package/lib/tokenizer/preprocessor.js +0 -159
  84. package/lib/tree-adapters/default.js +0 -221
  85. package/lib/utils/merge-options.js +0 -13
  86. package/lib/utils/mixin.js +0 -39
@@ -0,0 +1,523 @@
1
+ /** All valid namespaces in HTML. */
2
+ export var NS;
3
+ (function (NS) {
4
+ NS["HTML"] = "http://www.w3.org/1999/xhtml";
5
+ NS["MATHML"] = "http://www.w3.org/1998/Math/MathML";
6
+ NS["SVG"] = "http://www.w3.org/2000/svg";
7
+ NS["XLINK"] = "http://www.w3.org/1999/xlink";
8
+ NS["XML"] = "http://www.w3.org/XML/1998/namespace";
9
+ NS["XMLNS"] = "http://www.w3.org/2000/xmlns/";
10
+ })(NS || (NS = {}));
11
+ export var ATTRS;
12
+ (function (ATTRS) {
13
+ ATTRS["TYPE"] = "type";
14
+ ATTRS["ACTION"] = "action";
15
+ ATTRS["ENCODING"] = "encoding";
16
+ ATTRS["PROMPT"] = "prompt";
17
+ ATTRS["NAME"] = "name";
18
+ ATTRS["COLOR"] = "color";
19
+ ATTRS["FACE"] = "face";
20
+ ATTRS["SIZE"] = "size";
21
+ })(ATTRS || (ATTRS = {}));
22
+ /**
23
+ * The mode of the document.
24
+ *
25
+ * @see {@link https://dom.spec.whatwg.org/#concept-document-limited-quirks}
26
+ */
27
+ export var DOCUMENT_MODE;
28
+ (function (DOCUMENT_MODE) {
29
+ DOCUMENT_MODE["NO_QUIRKS"] = "no-quirks";
30
+ DOCUMENT_MODE["QUIRKS"] = "quirks";
31
+ DOCUMENT_MODE["LIMITED_QUIRKS"] = "limited-quirks";
32
+ })(DOCUMENT_MODE || (DOCUMENT_MODE = {}));
33
+ export var TAG_NAMES;
34
+ (function (TAG_NAMES) {
35
+ TAG_NAMES["A"] = "a";
36
+ TAG_NAMES["ADDRESS"] = "address";
37
+ TAG_NAMES["ANNOTATION_XML"] = "annotation-xml";
38
+ TAG_NAMES["APPLET"] = "applet";
39
+ TAG_NAMES["AREA"] = "area";
40
+ TAG_NAMES["ARTICLE"] = "article";
41
+ TAG_NAMES["ASIDE"] = "aside";
42
+ TAG_NAMES["B"] = "b";
43
+ TAG_NAMES["BASE"] = "base";
44
+ TAG_NAMES["BASEFONT"] = "basefont";
45
+ TAG_NAMES["BGSOUND"] = "bgsound";
46
+ TAG_NAMES["BIG"] = "big";
47
+ TAG_NAMES["BLOCKQUOTE"] = "blockquote";
48
+ TAG_NAMES["BODY"] = "body";
49
+ TAG_NAMES["BR"] = "br";
50
+ TAG_NAMES["BUTTON"] = "button";
51
+ TAG_NAMES["CAPTION"] = "caption";
52
+ TAG_NAMES["CENTER"] = "center";
53
+ TAG_NAMES["CODE"] = "code";
54
+ TAG_NAMES["COL"] = "col";
55
+ TAG_NAMES["COLGROUP"] = "colgroup";
56
+ TAG_NAMES["DD"] = "dd";
57
+ TAG_NAMES["DESC"] = "desc";
58
+ TAG_NAMES["DETAILS"] = "details";
59
+ TAG_NAMES["DIALOG"] = "dialog";
60
+ TAG_NAMES["DIR"] = "dir";
61
+ TAG_NAMES["DIV"] = "div";
62
+ TAG_NAMES["DL"] = "dl";
63
+ TAG_NAMES["DT"] = "dt";
64
+ TAG_NAMES["EM"] = "em";
65
+ TAG_NAMES["EMBED"] = "embed";
66
+ TAG_NAMES["FIELDSET"] = "fieldset";
67
+ TAG_NAMES["FIGCAPTION"] = "figcaption";
68
+ TAG_NAMES["FIGURE"] = "figure";
69
+ TAG_NAMES["FONT"] = "font";
70
+ TAG_NAMES["FOOTER"] = "footer";
71
+ TAG_NAMES["FOREIGN_OBJECT"] = "foreignObject";
72
+ TAG_NAMES["FORM"] = "form";
73
+ TAG_NAMES["FRAME"] = "frame";
74
+ TAG_NAMES["FRAMESET"] = "frameset";
75
+ TAG_NAMES["H1"] = "h1";
76
+ TAG_NAMES["H2"] = "h2";
77
+ TAG_NAMES["H3"] = "h3";
78
+ TAG_NAMES["H4"] = "h4";
79
+ TAG_NAMES["H5"] = "h5";
80
+ TAG_NAMES["H6"] = "h6";
81
+ TAG_NAMES["HEAD"] = "head";
82
+ TAG_NAMES["HEADER"] = "header";
83
+ TAG_NAMES["HGROUP"] = "hgroup";
84
+ TAG_NAMES["HR"] = "hr";
85
+ TAG_NAMES["HTML"] = "html";
86
+ TAG_NAMES["I"] = "i";
87
+ TAG_NAMES["IMG"] = "img";
88
+ TAG_NAMES["IMAGE"] = "image";
89
+ TAG_NAMES["INPUT"] = "input";
90
+ TAG_NAMES["IFRAME"] = "iframe";
91
+ TAG_NAMES["KEYGEN"] = "keygen";
92
+ TAG_NAMES["LABEL"] = "label";
93
+ TAG_NAMES["LI"] = "li";
94
+ TAG_NAMES["LINK"] = "link";
95
+ TAG_NAMES["LISTING"] = "listing";
96
+ TAG_NAMES["MAIN"] = "main";
97
+ TAG_NAMES["MALIGNMARK"] = "malignmark";
98
+ TAG_NAMES["MARQUEE"] = "marquee";
99
+ TAG_NAMES["MATH"] = "math";
100
+ TAG_NAMES["MENU"] = "menu";
101
+ TAG_NAMES["META"] = "meta";
102
+ TAG_NAMES["MGLYPH"] = "mglyph";
103
+ TAG_NAMES["MI"] = "mi";
104
+ TAG_NAMES["MO"] = "mo";
105
+ TAG_NAMES["MN"] = "mn";
106
+ TAG_NAMES["MS"] = "ms";
107
+ TAG_NAMES["MTEXT"] = "mtext";
108
+ TAG_NAMES["NAV"] = "nav";
109
+ TAG_NAMES["NOBR"] = "nobr";
110
+ TAG_NAMES["NOFRAMES"] = "noframes";
111
+ TAG_NAMES["NOEMBED"] = "noembed";
112
+ TAG_NAMES["NOSCRIPT"] = "noscript";
113
+ TAG_NAMES["OBJECT"] = "object";
114
+ TAG_NAMES["OL"] = "ol";
115
+ TAG_NAMES["OPTGROUP"] = "optgroup";
116
+ TAG_NAMES["OPTION"] = "option";
117
+ TAG_NAMES["P"] = "p";
118
+ TAG_NAMES["PARAM"] = "param";
119
+ TAG_NAMES["PLAINTEXT"] = "plaintext";
120
+ TAG_NAMES["PRE"] = "pre";
121
+ TAG_NAMES["RB"] = "rb";
122
+ TAG_NAMES["RP"] = "rp";
123
+ TAG_NAMES["RT"] = "rt";
124
+ TAG_NAMES["RTC"] = "rtc";
125
+ TAG_NAMES["RUBY"] = "ruby";
126
+ TAG_NAMES["S"] = "s";
127
+ TAG_NAMES["SCRIPT"] = "script";
128
+ TAG_NAMES["SEARCH"] = "search";
129
+ TAG_NAMES["SECTION"] = "section";
130
+ TAG_NAMES["SELECT"] = "select";
131
+ TAG_NAMES["SOURCE"] = "source";
132
+ TAG_NAMES["SMALL"] = "small";
133
+ TAG_NAMES["SPAN"] = "span";
134
+ TAG_NAMES["STRIKE"] = "strike";
135
+ TAG_NAMES["STRONG"] = "strong";
136
+ TAG_NAMES["STYLE"] = "style";
137
+ TAG_NAMES["SUB"] = "sub";
138
+ TAG_NAMES["SUMMARY"] = "summary";
139
+ TAG_NAMES["SUP"] = "sup";
140
+ TAG_NAMES["TABLE"] = "table";
141
+ TAG_NAMES["TBODY"] = "tbody";
142
+ TAG_NAMES["TEMPLATE"] = "template";
143
+ TAG_NAMES["TEXTAREA"] = "textarea";
144
+ TAG_NAMES["TFOOT"] = "tfoot";
145
+ TAG_NAMES["TD"] = "td";
146
+ TAG_NAMES["TH"] = "th";
147
+ TAG_NAMES["THEAD"] = "thead";
148
+ TAG_NAMES["TITLE"] = "title";
149
+ TAG_NAMES["TR"] = "tr";
150
+ TAG_NAMES["TRACK"] = "track";
151
+ TAG_NAMES["TT"] = "tt";
152
+ TAG_NAMES["U"] = "u";
153
+ TAG_NAMES["UL"] = "ul";
154
+ TAG_NAMES["SVG"] = "svg";
155
+ TAG_NAMES["VAR"] = "var";
156
+ TAG_NAMES["WBR"] = "wbr";
157
+ TAG_NAMES["XMP"] = "xmp";
158
+ })(TAG_NAMES || (TAG_NAMES = {}));
159
+ /**
160
+ * Tag IDs are numeric IDs for known tag names.
161
+ *
162
+ * We use tag IDs to improve the performance of tag name comparisons.
163
+ */
164
+ export var TAG_ID;
165
+ (function (TAG_ID) {
166
+ TAG_ID[TAG_ID["UNKNOWN"] = 0] = "UNKNOWN";
167
+ TAG_ID[TAG_ID["A"] = 1] = "A";
168
+ TAG_ID[TAG_ID["ADDRESS"] = 2] = "ADDRESS";
169
+ TAG_ID[TAG_ID["ANNOTATION_XML"] = 3] = "ANNOTATION_XML";
170
+ TAG_ID[TAG_ID["APPLET"] = 4] = "APPLET";
171
+ TAG_ID[TAG_ID["AREA"] = 5] = "AREA";
172
+ TAG_ID[TAG_ID["ARTICLE"] = 6] = "ARTICLE";
173
+ TAG_ID[TAG_ID["ASIDE"] = 7] = "ASIDE";
174
+ TAG_ID[TAG_ID["B"] = 8] = "B";
175
+ TAG_ID[TAG_ID["BASE"] = 9] = "BASE";
176
+ TAG_ID[TAG_ID["BASEFONT"] = 10] = "BASEFONT";
177
+ TAG_ID[TAG_ID["BGSOUND"] = 11] = "BGSOUND";
178
+ TAG_ID[TAG_ID["BIG"] = 12] = "BIG";
179
+ TAG_ID[TAG_ID["BLOCKQUOTE"] = 13] = "BLOCKQUOTE";
180
+ TAG_ID[TAG_ID["BODY"] = 14] = "BODY";
181
+ TAG_ID[TAG_ID["BR"] = 15] = "BR";
182
+ TAG_ID[TAG_ID["BUTTON"] = 16] = "BUTTON";
183
+ TAG_ID[TAG_ID["CAPTION"] = 17] = "CAPTION";
184
+ TAG_ID[TAG_ID["CENTER"] = 18] = "CENTER";
185
+ TAG_ID[TAG_ID["CODE"] = 19] = "CODE";
186
+ TAG_ID[TAG_ID["COL"] = 20] = "COL";
187
+ TAG_ID[TAG_ID["COLGROUP"] = 21] = "COLGROUP";
188
+ TAG_ID[TAG_ID["DD"] = 22] = "DD";
189
+ TAG_ID[TAG_ID["DESC"] = 23] = "DESC";
190
+ TAG_ID[TAG_ID["DETAILS"] = 24] = "DETAILS";
191
+ TAG_ID[TAG_ID["DIALOG"] = 25] = "DIALOG";
192
+ TAG_ID[TAG_ID["DIR"] = 26] = "DIR";
193
+ TAG_ID[TAG_ID["DIV"] = 27] = "DIV";
194
+ TAG_ID[TAG_ID["DL"] = 28] = "DL";
195
+ TAG_ID[TAG_ID["DT"] = 29] = "DT";
196
+ TAG_ID[TAG_ID["EM"] = 30] = "EM";
197
+ TAG_ID[TAG_ID["EMBED"] = 31] = "EMBED";
198
+ TAG_ID[TAG_ID["FIELDSET"] = 32] = "FIELDSET";
199
+ TAG_ID[TAG_ID["FIGCAPTION"] = 33] = "FIGCAPTION";
200
+ TAG_ID[TAG_ID["FIGURE"] = 34] = "FIGURE";
201
+ TAG_ID[TAG_ID["FONT"] = 35] = "FONT";
202
+ TAG_ID[TAG_ID["FOOTER"] = 36] = "FOOTER";
203
+ TAG_ID[TAG_ID["FOREIGN_OBJECT"] = 37] = "FOREIGN_OBJECT";
204
+ TAG_ID[TAG_ID["FORM"] = 38] = "FORM";
205
+ TAG_ID[TAG_ID["FRAME"] = 39] = "FRAME";
206
+ TAG_ID[TAG_ID["FRAMESET"] = 40] = "FRAMESET";
207
+ TAG_ID[TAG_ID["H1"] = 41] = "H1";
208
+ TAG_ID[TAG_ID["H2"] = 42] = "H2";
209
+ TAG_ID[TAG_ID["H3"] = 43] = "H3";
210
+ TAG_ID[TAG_ID["H4"] = 44] = "H4";
211
+ TAG_ID[TAG_ID["H5"] = 45] = "H5";
212
+ TAG_ID[TAG_ID["H6"] = 46] = "H6";
213
+ TAG_ID[TAG_ID["HEAD"] = 47] = "HEAD";
214
+ TAG_ID[TAG_ID["HEADER"] = 48] = "HEADER";
215
+ TAG_ID[TAG_ID["HGROUP"] = 49] = "HGROUP";
216
+ TAG_ID[TAG_ID["HR"] = 50] = "HR";
217
+ TAG_ID[TAG_ID["HTML"] = 51] = "HTML";
218
+ TAG_ID[TAG_ID["I"] = 52] = "I";
219
+ TAG_ID[TAG_ID["IMG"] = 53] = "IMG";
220
+ TAG_ID[TAG_ID["IMAGE"] = 54] = "IMAGE";
221
+ TAG_ID[TAG_ID["INPUT"] = 55] = "INPUT";
222
+ TAG_ID[TAG_ID["IFRAME"] = 56] = "IFRAME";
223
+ TAG_ID[TAG_ID["KEYGEN"] = 57] = "KEYGEN";
224
+ TAG_ID[TAG_ID["LABEL"] = 58] = "LABEL";
225
+ TAG_ID[TAG_ID["LI"] = 59] = "LI";
226
+ TAG_ID[TAG_ID["LINK"] = 60] = "LINK";
227
+ TAG_ID[TAG_ID["LISTING"] = 61] = "LISTING";
228
+ TAG_ID[TAG_ID["MAIN"] = 62] = "MAIN";
229
+ TAG_ID[TAG_ID["MALIGNMARK"] = 63] = "MALIGNMARK";
230
+ TAG_ID[TAG_ID["MARQUEE"] = 64] = "MARQUEE";
231
+ TAG_ID[TAG_ID["MATH"] = 65] = "MATH";
232
+ TAG_ID[TAG_ID["MENU"] = 66] = "MENU";
233
+ TAG_ID[TAG_ID["META"] = 67] = "META";
234
+ TAG_ID[TAG_ID["MGLYPH"] = 68] = "MGLYPH";
235
+ TAG_ID[TAG_ID["MI"] = 69] = "MI";
236
+ TAG_ID[TAG_ID["MO"] = 70] = "MO";
237
+ TAG_ID[TAG_ID["MN"] = 71] = "MN";
238
+ TAG_ID[TAG_ID["MS"] = 72] = "MS";
239
+ TAG_ID[TAG_ID["MTEXT"] = 73] = "MTEXT";
240
+ TAG_ID[TAG_ID["NAV"] = 74] = "NAV";
241
+ TAG_ID[TAG_ID["NOBR"] = 75] = "NOBR";
242
+ TAG_ID[TAG_ID["NOFRAMES"] = 76] = "NOFRAMES";
243
+ TAG_ID[TAG_ID["NOEMBED"] = 77] = "NOEMBED";
244
+ TAG_ID[TAG_ID["NOSCRIPT"] = 78] = "NOSCRIPT";
245
+ TAG_ID[TAG_ID["OBJECT"] = 79] = "OBJECT";
246
+ TAG_ID[TAG_ID["OL"] = 80] = "OL";
247
+ TAG_ID[TAG_ID["OPTGROUP"] = 81] = "OPTGROUP";
248
+ TAG_ID[TAG_ID["OPTION"] = 82] = "OPTION";
249
+ TAG_ID[TAG_ID["P"] = 83] = "P";
250
+ TAG_ID[TAG_ID["PARAM"] = 84] = "PARAM";
251
+ TAG_ID[TAG_ID["PLAINTEXT"] = 85] = "PLAINTEXT";
252
+ TAG_ID[TAG_ID["PRE"] = 86] = "PRE";
253
+ TAG_ID[TAG_ID["RB"] = 87] = "RB";
254
+ TAG_ID[TAG_ID["RP"] = 88] = "RP";
255
+ TAG_ID[TAG_ID["RT"] = 89] = "RT";
256
+ TAG_ID[TAG_ID["RTC"] = 90] = "RTC";
257
+ TAG_ID[TAG_ID["RUBY"] = 91] = "RUBY";
258
+ TAG_ID[TAG_ID["S"] = 92] = "S";
259
+ TAG_ID[TAG_ID["SCRIPT"] = 93] = "SCRIPT";
260
+ TAG_ID[TAG_ID["SEARCH"] = 94] = "SEARCH";
261
+ TAG_ID[TAG_ID["SECTION"] = 95] = "SECTION";
262
+ TAG_ID[TAG_ID["SELECT"] = 96] = "SELECT";
263
+ TAG_ID[TAG_ID["SOURCE"] = 97] = "SOURCE";
264
+ TAG_ID[TAG_ID["SMALL"] = 98] = "SMALL";
265
+ TAG_ID[TAG_ID["SPAN"] = 99] = "SPAN";
266
+ TAG_ID[TAG_ID["STRIKE"] = 100] = "STRIKE";
267
+ TAG_ID[TAG_ID["STRONG"] = 101] = "STRONG";
268
+ TAG_ID[TAG_ID["STYLE"] = 102] = "STYLE";
269
+ TAG_ID[TAG_ID["SUB"] = 103] = "SUB";
270
+ TAG_ID[TAG_ID["SUMMARY"] = 104] = "SUMMARY";
271
+ TAG_ID[TAG_ID["SUP"] = 105] = "SUP";
272
+ TAG_ID[TAG_ID["TABLE"] = 106] = "TABLE";
273
+ TAG_ID[TAG_ID["TBODY"] = 107] = "TBODY";
274
+ TAG_ID[TAG_ID["TEMPLATE"] = 108] = "TEMPLATE";
275
+ TAG_ID[TAG_ID["TEXTAREA"] = 109] = "TEXTAREA";
276
+ TAG_ID[TAG_ID["TFOOT"] = 110] = "TFOOT";
277
+ TAG_ID[TAG_ID["TD"] = 111] = "TD";
278
+ TAG_ID[TAG_ID["TH"] = 112] = "TH";
279
+ TAG_ID[TAG_ID["THEAD"] = 113] = "THEAD";
280
+ TAG_ID[TAG_ID["TITLE"] = 114] = "TITLE";
281
+ TAG_ID[TAG_ID["TR"] = 115] = "TR";
282
+ TAG_ID[TAG_ID["TRACK"] = 116] = "TRACK";
283
+ TAG_ID[TAG_ID["TT"] = 117] = "TT";
284
+ TAG_ID[TAG_ID["U"] = 118] = "U";
285
+ TAG_ID[TAG_ID["UL"] = 119] = "UL";
286
+ TAG_ID[TAG_ID["SVG"] = 120] = "SVG";
287
+ TAG_ID[TAG_ID["VAR"] = 121] = "VAR";
288
+ TAG_ID[TAG_ID["WBR"] = 122] = "WBR";
289
+ TAG_ID[TAG_ID["XMP"] = 123] = "XMP";
290
+ })(TAG_ID || (TAG_ID = {}));
291
+ const TAG_NAME_TO_ID = new Map([
292
+ [TAG_NAMES.A, TAG_ID.A],
293
+ [TAG_NAMES.ADDRESS, TAG_ID.ADDRESS],
294
+ [TAG_NAMES.ANNOTATION_XML, TAG_ID.ANNOTATION_XML],
295
+ [TAG_NAMES.APPLET, TAG_ID.APPLET],
296
+ [TAG_NAMES.AREA, TAG_ID.AREA],
297
+ [TAG_NAMES.ARTICLE, TAG_ID.ARTICLE],
298
+ [TAG_NAMES.ASIDE, TAG_ID.ASIDE],
299
+ [TAG_NAMES.B, TAG_ID.B],
300
+ [TAG_NAMES.BASE, TAG_ID.BASE],
301
+ [TAG_NAMES.BASEFONT, TAG_ID.BASEFONT],
302
+ [TAG_NAMES.BGSOUND, TAG_ID.BGSOUND],
303
+ [TAG_NAMES.BIG, TAG_ID.BIG],
304
+ [TAG_NAMES.BLOCKQUOTE, TAG_ID.BLOCKQUOTE],
305
+ [TAG_NAMES.BODY, TAG_ID.BODY],
306
+ [TAG_NAMES.BR, TAG_ID.BR],
307
+ [TAG_NAMES.BUTTON, TAG_ID.BUTTON],
308
+ [TAG_NAMES.CAPTION, TAG_ID.CAPTION],
309
+ [TAG_NAMES.CENTER, TAG_ID.CENTER],
310
+ [TAG_NAMES.CODE, TAG_ID.CODE],
311
+ [TAG_NAMES.COL, TAG_ID.COL],
312
+ [TAG_NAMES.COLGROUP, TAG_ID.COLGROUP],
313
+ [TAG_NAMES.DD, TAG_ID.DD],
314
+ [TAG_NAMES.DESC, TAG_ID.DESC],
315
+ [TAG_NAMES.DETAILS, TAG_ID.DETAILS],
316
+ [TAG_NAMES.DIALOG, TAG_ID.DIALOG],
317
+ [TAG_NAMES.DIR, TAG_ID.DIR],
318
+ [TAG_NAMES.DIV, TAG_ID.DIV],
319
+ [TAG_NAMES.DL, TAG_ID.DL],
320
+ [TAG_NAMES.DT, TAG_ID.DT],
321
+ [TAG_NAMES.EM, TAG_ID.EM],
322
+ [TAG_NAMES.EMBED, TAG_ID.EMBED],
323
+ [TAG_NAMES.FIELDSET, TAG_ID.FIELDSET],
324
+ [TAG_NAMES.FIGCAPTION, TAG_ID.FIGCAPTION],
325
+ [TAG_NAMES.FIGURE, TAG_ID.FIGURE],
326
+ [TAG_NAMES.FONT, TAG_ID.FONT],
327
+ [TAG_NAMES.FOOTER, TAG_ID.FOOTER],
328
+ [TAG_NAMES.FOREIGN_OBJECT, TAG_ID.FOREIGN_OBJECT],
329
+ [TAG_NAMES.FORM, TAG_ID.FORM],
330
+ [TAG_NAMES.FRAME, TAG_ID.FRAME],
331
+ [TAG_NAMES.FRAMESET, TAG_ID.FRAMESET],
332
+ [TAG_NAMES.H1, TAG_ID.H1],
333
+ [TAG_NAMES.H2, TAG_ID.H2],
334
+ [TAG_NAMES.H3, TAG_ID.H3],
335
+ [TAG_NAMES.H4, TAG_ID.H4],
336
+ [TAG_NAMES.H5, TAG_ID.H5],
337
+ [TAG_NAMES.H6, TAG_ID.H6],
338
+ [TAG_NAMES.HEAD, TAG_ID.HEAD],
339
+ [TAG_NAMES.HEADER, TAG_ID.HEADER],
340
+ [TAG_NAMES.HGROUP, TAG_ID.HGROUP],
341
+ [TAG_NAMES.HR, TAG_ID.HR],
342
+ [TAG_NAMES.HTML, TAG_ID.HTML],
343
+ [TAG_NAMES.I, TAG_ID.I],
344
+ [TAG_NAMES.IMG, TAG_ID.IMG],
345
+ [TAG_NAMES.IMAGE, TAG_ID.IMAGE],
346
+ [TAG_NAMES.INPUT, TAG_ID.INPUT],
347
+ [TAG_NAMES.IFRAME, TAG_ID.IFRAME],
348
+ [TAG_NAMES.KEYGEN, TAG_ID.KEYGEN],
349
+ [TAG_NAMES.LABEL, TAG_ID.LABEL],
350
+ [TAG_NAMES.LI, TAG_ID.LI],
351
+ [TAG_NAMES.LINK, TAG_ID.LINK],
352
+ [TAG_NAMES.LISTING, TAG_ID.LISTING],
353
+ [TAG_NAMES.MAIN, TAG_ID.MAIN],
354
+ [TAG_NAMES.MALIGNMARK, TAG_ID.MALIGNMARK],
355
+ [TAG_NAMES.MARQUEE, TAG_ID.MARQUEE],
356
+ [TAG_NAMES.MATH, TAG_ID.MATH],
357
+ [TAG_NAMES.MENU, TAG_ID.MENU],
358
+ [TAG_NAMES.META, TAG_ID.META],
359
+ [TAG_NAMES.MGLYPH, TAG_ID.MGLYPH],
360
+ [TAG_NAMES.MI, TAG_ID.MI],
361
+ [TAG_NAMES.MO, TAG_ID.MO],
362
+ [TAG_NAMES.MN, TAG_ID.MN],
363
+ [TAG_NAMES.MS, TAG_ID.MS],
364
+ [TAG_NAMES.MTEXT, TAG_ID.MTEXT],
365
+ [TAG_NAMES.NAV, TAG_ID.NAV],
366
+ [TAG_NAMES.NOBR, TAG_ID.NOBR],
367
+ [TAG_NAMES.NOFRAMES, TAG_ID.NOFRAMES],
368
+ [TAG_NAMES.NOEMBED, TAG_ID.NOEMBED],
369
+ [TAG_NAMES.NOSCRIPT, TAG_ID.NOSCRIPT],
370
+ [TAG_NAMES.OBJECT, TAG_ID.OBJECT],
371
+ [TAG_NAMES.OL, TAG_ID.OL],
372
+ [TAG_NAMES.OPTGROUP, TAG_ID.OPTGROUP],
373
+ [TAG_NAMES.OPTION, TAG_ID.OPTION],
374
+ [TAG_NAMES.P, TAG_ID.P],
375
+ [TAG_NAMES.PARAM, TAG_ID.PARAM],
376
+ [TAG_NAMES.PLAINTEXT, TAG_ID.PLAINTEXT],
377
+ [TAG_NAMES.PRE, TAG_ID.PRE],
378
+ [TAG_NAMES.RB, TAG_ID.RB],
379
+ [TAG_NAMES.RP, TAG_ID.RP],
380
+ [TAG_NAMES.RT, TAG_ID.RT],
381
+ [TAG_NAMES.RTC, TAG_ID.RTC],
382
+ [TAG_NAMES.RUBY, TAG_ID.RUBY],
383
+ [TAG_NAMES.S, TAG_ID.S],
384
+ [TAG_NAMES.SCRIPT, TAG_ID.SCRIPT],
385
+ [TAG_NAMES.SEARCH, TAG_ID.SEARCH],
386
+ [TAG_NAMES.SECTION, TAG_ID.SECTION],
387
+ [TAG_NAMES.SELECT, TAG_ID.SELECT],
388
+ [TAG_NAMES.SOURCE, TAG_ID.SOURCE],
389
+ [TAG_NAMES.SMALL, TAG_ID.SMALL],
390
+ [TAG_NAMES.SPAN, TAG_ID.SPAN],
391
+ [TAG_NAMES.STRIKE, TAG_ID.STRIKE],
392
+ [TAG_NAMES.STRONG, TAG_ID.STRONG],
393
+ [TAG_NAMES.STYLE, TAG_ID.STYLE],
394
+ [TAG_NAMES.SUB, TAG_ID.SUB],
395
+ [TAG_NAMES.SUMMARY, TAG_ID.SUMMARY],
396
+ [TAG_NAMES.SUP, TAG_ID.SUP],
397
+ [TAG_NAMES.TABLE, TAG_ID.TABLE],
398
+ [TAG_NAMES.TBODY, TAG_ID.TBODY],
399
+ [TAG_NAMES.TEMPLATE, TAG_ID.TEMPLATE],
400
+ [TAG_NAMES.TEXTAREA, TAG_ID.TEXTAREA],
401
+ [TAG_NAMES.TFOOT, TAG_ID.TFOOT],
402
+ [TAG_NAMES.TD, TAG_ID.TD],
403
+ [TAG_NAMES.TH, TAG_ID.TH],
404
+ [TAG_NAMES.THEAD, TAG_ID.THEAD],
405
+ [TAG_NAMES.TITLE, TAG_ID.TITLE],
406
+ [TAG_NAMES.TR, TAG_ID.TR],
407
+ [TAG_NAMES.TRACK, TAG_ID.TRACK],
408
+ [TAG_NAMES.TT, TAG_ID.TT],
409
+ [TAG_NAMES.U, TAG_ID.U],
410
+ [TAG_NAMES.UL, TAG_ID.UL],
411
+ [TAG_NAMES.SVG, TAG_ID.SVG],
412
+ [TAG_NAMES.VAR, TAG_ID.VAR],
413
+ [TAG_NAMES.WBR, TAG_ID.WBR],
414
+ [TAG_NAMES.XMP, TAG_ID.XMP],
415
+ ]);
416
+ export function getTagID(tagName) {
417
+ var _a;
418
+ return (_a = TAG_NAME_TO_ID.get(tagName)) !== null && _a !== void 0 ? _a : TAG_ID.UNKNOWN;
419
+ }
420
+ const $ = TAG_ID;
421
+ export const SPECIAL_ELEMENTS = {
422
+ [NS.HTML]: new Set([
423
+ $.ADDRESS,
424
+ $.APPLET,
425
+ $.AREA,
426
+ $.ARTICLE,
427
+ $.ASIDE,
428
+ $.BASE,
429
+ $.BASEFONT,
430
+ $.BGSOUND,
431
+ $.BLOCKQUOTE,
432
+ $.BODY,
433
+ $.BR,
434
+ $.BUTTON,
435
+ $.CAPTION,
436
+ $.CENTER,
437
+ $.COL,
438
+ $.COLGROUP,
439
+ $.DD,
440
+ $.DETAILS,
441
+ $.DIR,
442
+ $.DIV,
443
+ $.DL,
444
+ $.DT,
445
+ $.EMBED,
446
+ $.FIELDSET,
447
+ $.FIGCAPTION,
448
+ $.FIGURE,
449
+ $.FOOTER,
450
+ $.FORM,
451
+ $.FRAME,
452
+ $.FRAMESET,
453
+ $.H1,
454
+ $.H2,
455
+ $.H3,
456
+ $.H4,
457
+ $.H5,
458
+ $.H6,
459
+ $.HEAD,
460
+ $.HEADER,
461
+ $.HGROUP,
462
+ $.HR,
463
+ $.HTML,
464
+ $.IFRAME,
465
+ $.IMG,
466
+ $.INPUT,
467
+ $.LI,
468
+ $.LINK,
469
+ $.LISTING,
470
+ $.MAIN,
471
+ $.MARQUEE,
472
+ $.MENU,
473
+ $.META,
474
+ $.NAV,
475
+ $.NOEMBED,
476
+ $.NOFRAMES,
477
+ $.NOSCRIPT,
478
+ $.OBJECT,
479
+ $.OL,
480
+ $.P,
481
+ $.PARAM,
482
+ $.PLAINTEXT,
483
+ $.PRE,
484
+ $.SCRIPT,
485
+ $.SECTION,
486
+ $.SELECT,
487
+ $.SOURCE,
488
+ $.STYLE,
489
+ $.SUMMARY,
490
+ $.TABLE,
491
+ $.TBODY,
492
+ $.TD,
493
+ $.TEMPLATE,
494
+ $.TEXTAREA,
495
+ $.TFOOT,
496
+ $.TH,
497
+ $.THEAD,
498
+ $.TITLE,
499
+ $.TR,
500
+ $.TRACK,
501
+ $.UL,
502
+ $.WBR,
503
+ $.XMP,
504
+ ]),
505
+ [NS.MATHML]: new Set([$.MI, $.MO, $.MN, $.MS, $.MTEXT, $.ANNOTATION_XML]),
506
+ [NS.SVG]: new Set([$.TITLE, $.FOREIGN_OBJECT, $.DESC]),
507
+ [NS.XLINK]: new Set(),
508
+ [NS.XML]: new Set(),
509
+ [NS.XMLNS]: new Set(),
510
+ };
511
+ export const NUMBERED_HEADERS = new Set([$.H1, $.H2, $.H3, $.H4, $.H5, $.H6]);
512
+ const UNESCAPED_TEXT = new Set([
513
+ TAG_NAMES.STYLE,
514
+ TAG_NAMES.SCRIPT,
515
+ TAG_NAMES.XMP,
516
+ TAG_NAMES.IFRAME,
517
+ TAG_NAMES.NOEMBED,
518
+ TAG_NAMES.NOFRAMES,
519
+ TAG_NAMES.PLAINTEXT,
520
+ ]);
521
+ export function hasUnescapedText(tn, scriptingEnabled) {
522
+ return UNESCAPED_TEXT.has(tn) || (scriptingEnabled && tn === TAG_NAMES.NOSCRIPT);
523
+ }
@@ -0,0 +1,84 @@
1
+ import type { TAG_ID } from './html.js';
2
+ export declare enum TokenType {
3
+ CHARACTER = 0,
4
+ NULL_CHARACTER = 1,
5
+ WHITESPACE_CHARACTER = 2,
6
+ START_TAG = 3,
7
+ END_TAG = 4,
8
+ COMMENT = 5,
9
+ DOCTYPE = 6,
10
+ EOF = 7,
11
+ HIBERNATION = 8
12
+ }
13
+ export interface Location {
14
+ /** One-based line index of the first character. */
15
+ startLine: number;
16
+ /** One-based column index of the first character. */
17
+ startCol: number;
18
+ /** Zero-based first character index. */
19
+ startOffset: number;
20
+ /** One-based line index of the last character. */
21
+ endLine: number;
22
+ /** One-based column index of the last character. Points directly *after* the last character. */
23
+ endCol: number;
24
+ /** Zero-based last character index. Points directly *after* the last character. */
25
+ endOffset: number;
26
+ }
27
+ export interface LocationWithAttributes extends Location {
28
+ /** Start tag attributes' location info. */
29
+ attrs?: Record<string, Location>;
30
+ }
31
+ export interface ElementLocation extends LocationWithAttributes {
32
+ /** Element's start tag location info. */
33
+ startTag?: Location;
34
+ /**
35
+ * Element's end tag location info.
36
+ * This property is undefined, if the element has no closing tag.
37
+ */
38
+ endTag?: Location;
39
+ }
40
+ interface TokenBase {
41
+ readonly type: TokenType;
42
+ location: Location | null;
43
+ }
44
+ export interface DoctypeToken extends TokenBase {
45
+ readonly type: TokenType.DOCTYPE;
46
+ name: string | null;
47
+ forceQuirks: boolean;
48
+ publicId: string | null;
49
+ systemId: string | null;
50
+ }
51
+ export interface Attribute {
52
+ /** The name of the attribute. */
53
+ name: string;
54
+ /** The namespace of the attribute. */
55
+ namespace?: string;
56
+ /** The namespace-related prefix of the attribute. */
57
+ prefix?: string;
58
+ /** The value of the attribute. */
59
+ value: string;
60
+ }
61
+ export interface TagToken extends TokenBase {
62
+ readonly type: TokenType.START_TAG | TokenType.END_TAG;
63
+ tagName: string;
64
+ /** Used to cache the ID of the tag name. */
65
+ tagID: TAG_ID;
66
+ selfClosing: boolean;
67
+ ackSelfClosing: boolean;
68
+ attrs: Attribute[];
69
+ location: LocationWithAttributes | null;
70
+ }
71
+ export declare function getTokenAttr(token: TagToken, attrName: string): string | null;
72
+ export interface CommentToken extends TokenBase {
73
+ readonly type: TokenType.COMMENT;
74
+ data: string;
75
+ }
76
+ export interface EOFToken extends TokenBase {
77
+ readonly type: TokenType.EOF;
78
+ }
79
+ export interface CharacterToken extends TokenBase {
80
+ type: TokenType.CHARACTER | TokenType.NULL_CHARACTER | TokenType.WHITESPACE_CHARACTER;
81
+ chars: string;
82
+ }
83
+ export type Token = DoctypeToken | TagToken | CommentToken | EOFToken | CharacterToken;
84
+ export {};
@@ -0,0 +1,20 @@
1
+ export var TokenType;
2
+ (function (TokenType) {
3
+ TokenType[TokenType["CHARACTER"] = 0] = "CHARACTER";
4
+ TokenType[TokenType["NULL_CHARACTER"] = 1] = "NULL_CHARACTER";
5
+ TokenType[TokenType["WHITESPACE_CHARACTER"] = 2] = "WHITESPACE_CHARACTER";
6
+ TokenType[TokenType["START_TAG"] = 3] = "START_TAG";
7
+ TokenType[TokenType["END_TAG"] = 4] = "END_TAG";
8
+ TokenType[TokenType["COMMENT"] = 5] = "COMMENT";
9
+ TokenType[TokenType["DOCTYPE"] = 6] = "DOCTYPE";
10
+ TokenType[TokenType["EOF"] = 7] = "EOF";
11
+ TokenType[TokenType["HIBERNATION"] = 8] = "HIBERNATION";
12
+ })(TokenType || (TokenType = {}));
13
+ export function getTokenAttr(token, attrName) {
14
+ for (let i = token.attrs.length - 1; i >= 0; i--) {
15
+ if (token.attrs[i].name === attrName) {
16
+ return token.attrs[i].value;
17
+ }
18
+ }
19
+ return null;
20
+ }
@@ -0,0 +1,42 @@
1
+ export declare const REPLACEMENT_CHARACTER = "\uFFFD";
2
+ export declare enum CODE_POINTS {
3
+ EOF = -1,
4
+ NULL = 0,
5
+ TABULATION = 9,
6
+ CARRIAGE_RETURN = 13,
7
+ LINE_FEED = 10,
8
+ FORM_FEED = 12,
9
+ SPACE = 32,
10
+ EXCLAMATION_MARK = 33,
11
+ QUOTATION_MARK = 34,
12
+ AMPERSAND = 38,
13
+ APOSTROPHE = 39,
14
+ HYPHEN_MINUS = 45,
15
+ SOLIDUS = 47,
16
+ DIGIT_0 = 48,
17
+ DIGIT_9 = 57,
18
+ SEMICOLON = 59,
19
+ LESS_THAN_SIGN = 60,
20
+ EQUALS_SIGN = 61,
21
+ GREATER_THAN_SIGN = 62,
22
+ QUESTION_MARK = 63,
23
+ LATIN_CAPITAL_A = 65,
24
+ LATIN_CAPITAL_Z = 90,
25
+ RIGHT_SQUARE_BRACKET = 93,
26
+ GRAVE_ACCENT = 96,
27
+ LATIN_SMALL_A = 97,
28
+ LATIN_SMALL_Z = 122
29
+ }
30
+ export declare const SEQUENCES: {
31
+ DASH_DASH: string;
32
+ CDATA_START: string;
33
+ DOCTYPE: string;
34
+ SCRIPT: string;
35
+ PUBLIC: string;
36
+ SYSTEM: string;
37
+ };
38
+ export declare function isSurrogate(cp: number): boolean;
39
+ export declare function isSurrogatePair(cp: number): boolean;
40
+ export declare function getSurrogatePairCodePoint(cp1: number, cp2: number): number;
41
+ export declare function isControlCodePoint(cp: number): boolean;
42
+ export declare function isUndefinedCodePoint(cp: number): boolean;