parse5 6.0.1 → 7.1.1

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 +5 -0
  3. package/dist/cjs/common/doctype.js +120 -0
  4. package/dist/cjs/common/error-codes.d.ts +68 -0
  5. package/dist/cjs/common/error-codes.js +67 -0
  6. package/dist/cjs/common/foreign-content.d.ts +10 -0
  7. package/dist/cjs/common/foreign-content.js +239 -0
  8. package/dist/cjs/common/html.d.ts +288 -0
  9. package/dist/cjs/common/html.js +529 -0
  10. package/dist/cjs/common/token.d.ts +85 -0
  11. package/dist/cjs/common/token.js +25 -0
  12. package/dist/cjs/common/unicode.d.ts +49 -0
  13. package/dist/cjs/common/unicode.js +77 -0
  14. package/dist/cjs/index.d.ts +60 -0
  15. package/dist/cjs/index.js +55 -0
  16. package/dist/cjs/package.json +1 -0
  17. package/dist/cjs/parser/formatting-element-list.d.ts +37 -0
  18. package/dist/cjs/parser/formatting-element-list.js +115 -0
  19. package/dist/cjs/parser/index.d.ts +157 -0
  20. package/dist/cjs/parser/index.js +3053 -0
  21. package/dist/cjs/parser/open-element-stack.d.ts +53 -0
  22. package/dist/cjs/parser/open-element-stack.js +316 -0
  23. package/dist/cjs/serializer/index.d.ts +61 -0
  24. package/dist/cjs/serializer/index.js +173 -0
  25. package/dist/cjs/tokenizer/index.d.ts +248 -0
  26. package/dist/cjs/tokenizer/index.js +2907 -0
  27. package/dist/cjs/tokenizer/preprocessor.d.ts +37 -0
  28. package/dist/cjs/tokenizer/preprocessor.js +198 -0
  29. package/dist/cjs/tree-adapters/default.d.ts +85 -0
  30. package/dist/cjs/tree-adapters/default.js +177 -0
  31. package/dist/cjs/tree-adapters/interface.d.ts +250 -0
  32. package/dist/cjs/tree-adapters/interface.js +3 -0
  33. package/dist/common/doctype.d.ts +5 -0
  34. package/{lib → dist}/common/doctype.js +25 -72
  35. package/dist/common/error-codes.d.ts +68 -0
  36. package/dist/common/error-codes.js +64 -0
  37. package/dist/common/foreign-content.d.ts +10 -0
  38. package/dist/common/foreign-content.js +230 -0
  39. package/dist/common/html.d.ts +288 -0
  40. package/dist/common/html.js +523 -0
  41. package/dist/common/token.d.ts +85 -0
  42. package/dist/common/token.js +21 -0
  43. package/dist/common/unicode.d.ts +49 -0
  44. package/dist/common/unicode.js +69 -0
  45. package/dist/index.d.ts +60 -0
  46. package/dist/index.js +44 -0
  47. package/dist/parser/formatting-element-list.d.ts +37 -0
  48. package/dist/parser/formatting-element-list.js +111 -0
  49. package/dist/parser/index.d.ts +157 -0
  50. package/dist/parser/index.js +3058 -0
  51. package/dist/parser/open-element-stack.d.ts +53 -0
  52. package/dist/parser/open-element-stack.js +312 -0
  53. package/dist/serializer/index.d.ts +61 -0
  54. package/dist/serializer/index.js +168 -0
  55. package/dist/tokenizer/index.d.ts +248 -0
  56. package/dist/tokenizer/index.js +2903 -0
  57. package/dist/tokenizer/preprocessor.d.ts +37 -0
  58. package/dist/tokenizer/preprocessor.js +194 -0
  59. package/dist/tree-adapters/default.d.ts +85 -0
  60. package/dist/tree-adapters/default.js +174 -0
  61. package/dist/tree-adapters/interface.d.ts +250 -0
  62. package/dist/tree-adapters/interface.js +2 -0
  63. package/package.json +20 -5
  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 +1,2903 @@
1
+ import { Preprocessor } from './preprocessor.js';
2
+ import { CODE_POINTS as $, SEQUENCES as $$, REPLACEMENT_CHARACTER, isSurrogate, isUndefinedCodePoint, isControlCodePoint, } from '../common/unicode.js';
3
+ import { TokenType, getTokenAttr, } from '../common/token.js';
4
+ import { htmlDecodeTree, BinTrieFlags, determineBranch } from 'entities/lib/decode.js';
5
+ import { ERR } from '../common/error-codes.js';
6
+ import { TAG_ID, getTagID } from '../common/html.js';
7
+ //C1 Unicode control character reference replacements
8
+ const C1_CONTROLS_REFERENCE_REPLACEMENTS = new Map([
9
+ [0x80, 8364],
10
+ [0x82, 8218],
11
+ [0x83, 402],
12
+ [0x84, 8222],
13
+ [0x85, 8230],
14
+ [0x86, 8224],
15
+ [0x87, 8225],
16
+ [0x88, 710],
17
+ [0x89, 8240],
18
+ [0x8a, 352],
19
+ [0x8b, 8249],
20
+ [0x8c, 338],
21
+ [0x8e, 381],
22
+ [0x91, 8216],
23
+ [0x92, 8217],
24
+ [0x93, 8220],
25
+ [0x94, 8221],
26
+ [0x95, 8226],
27
+ [0x96, 8211],
28
+ [0x97, 8212],
29
+ [0x98, 732],
30
+ [0x99, 8482],
31
+ [0x9a, 353],
32
+ [0x9b, 8250],
33
+ [0x9c, 339],
34
+ [0x9e, 382],
35
+ [0x9f, 376],
36
+ ]);
37
+ //States
38
+ var State;
39
+ (function (State) {
40
+ State[State["DATA"] = 0] = "DATA";
41
+ State[State["RCDATA"] = 1] = "RCDATA";
42
+ State[State["RAWTEXT"] = 2] = "RAWTEXT";
43
+ State[State["SCRIPT_DATA"] = 3] = "SCRIPT_DATA";
44
+ State[State["PLAINTEXT"] = 4] = "PLAINTEXT";
45
+ State[State["TAG_OPEN"] = 5] = "TAG_OPEN";
46
+ State[State["END_TAG_OPEN"] = 6] = "END_TAG_OPEN";
47
+ State[State["TAG_NAME"] = 7] = "TAG_NAME";
48
+ State[State["RCDATA_LESS_THAN_SIGN"] = 8] = "RCDATA_LESS_THAN_SIGN";
49
+ State[State["RCDATA_END_TAG_OPEN"] = 9] = "RCDATA_END_TAG_OPEN";
50
+ State[State["RCDATA_END_TAG_NAME"] = 10] = "RCDATA_END_TAG_NAME";
51
+ State[State["RAWTEXT_LESS_THAN_SIGN"] = 11] = "RAWTEXT_LESS_THAN_SIGN";
52
+ State[State["RAWTEXT_END_TAG_OPEN"] = 12] = "RAWTEXT_END_TAG_OPEN";
53
+ State[State["RAWTEXT_END_TAG_NAME"] = 13] = "RAWTEXT_END_TAG_NAME";
54
+ State[State["SCRIPT_DATA_LESS_THAN_SIGN"] = 14] = "SCRIPT_DATA_LESS_THAN_SIGN";
55
+ State[State["SCRIPT_DATA_END_TAG_OPEN"] = 15] = "SCRIPT_DATA_END_TAG_OPEN";
56
+ State[State["SCRIPT_DATA_END_TAG_NAME"] = 16] = "SCRIPT_DATA_END_TAG_NAME";
57
+ State[State["SCRIPT_DATA_ESCAPE_START"] = 17] = "SCRIPT_DATA_ESCAPE_START";
58
+ State[State["SCRIPT_DATA_ESCAPE_START_DASH"] = 18] = "SCRIPT_DATA_ESCAPE_START_DASH";
59
+ State[State["SCRIPT_DATA_ESCAPED"] = 19] = "SCRIPT_DATA_ESCAPED";
60
+ State[State["SCRIPT_DATA_ESCAPED_DASH"] = 20] = "SCRIPT_DATA_ESCAPED_DASH";
61
+ State[State["SCRIPT_DATA_ESCAPED_DASH_DASH"] = 21] = "SCRIPT_DATA_ESCAPED_DASH_DASH";
62
+ State[State["SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN"] = 22] = "SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN";
63
+ State[State["SCRIPT_DATA_ESCAPED_END_TAG_OPEN"] = 23] = "SCRIPT_DATA_ESCAPED_END_TAG_OPEN";
64
+ State[State["SCRIPT_DATA_ESCAPED_END_TAG_NAME"] = 24] = "SCRIPT_DATA_ESCAPED_END_TAG_NAME";
65
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPE_START"] = 25] = "SCRIPT_DATA_DOUBLE_ESCAPE_START";
66
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED"] = 26] = "SCRIPT_DATA_DOUBLE_ESCAPED";
67
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH"] = 27] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH";
68
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH"] = 28] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH";
69
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN"] = 29] = "SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN";
70
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPE_END"] = 30] = "SCRIPT_DATA_DOUBLE_ESCAPE_END";
71
+ State[State["BEFORE_ATTRIBUTE_NAME"] = 31] = "BEFORE_ATTRIBUTE_NAME";
72
+ State[State["ATTRIBUTE_NAME"] = 32] = "ATTRIBUTE_NAME";
73
+ State[State["AFTER_ATTRIBUTE_NAME"] = 33] = "AFTER_ATTRIBUTE_NAME";
74
+ State[State["BEFORE_ATTRIBUTE_VALUE"] = 34] = "BEFORE_ATTRIBUTE_VALUE";
75
+ State[State["ATTRIBUTE_VALUE_DOUBLE_QUOTED"] = 35] = "ATTRIBUTE_VALUE_DOUBLE_QUOTED";
76
+ State[State["ATTRIBUTE_VALUE_SINGLE_QUOTED"] = 36] = "ATTRIBUTE_VALUE_SINGLE_QUOTED";
77
+ State[State["ATTRIBUTE_VALUE_UNQUOTED"] = 37] = "ATTRIBUTE_VALUE_UNQUOTED";
78
+ State[State["AFTER_ATTRIBUTE_VALUE_QUOTED"] = 38] = "AFTER_ATTRIBUTE_VALUE_QUOTED";
79
+ State[State["SELF_CLOSING_START_TAG"] = 39] = "SELF_CLOSING_START_TAG";
80
+ State[State["BOGUS_COMMENT"] = 40] = "BOGUS_COMMENT";
81
+ State[State["MARKUP_DECLARATION_OPEN"] = 41] = "MARKUP_DECLARATION_OPEN";
82
+ State[State["COMMENT_START"] = 42] = "COMMENT_START";
83
+ State[State["COMMENT_START_DASH"] = 43] = "COMMENT_START_DASH";
84
+ State[State["COMMENT"] = 44] = "COMMENT";
85
+ State[State["COMMENT_LESS_THAN_SIGN"] = 45] = "COMMENT_LESS_THAN_SIGN";
86
+ State[State["COMMENT_LESS_THAN_SIGN_BANG"] = 46] = "COMMENT_LESS_THAN_SIGN_BANG";
87
+ State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH"] = 47] = "COMMENT_LESS_THAN_SIGN_BANG_DASH";
88
+ State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH"] = 48] = "COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH";
89
+ State[State["COMMENT_END_DASH"] = 49] = "COMMENT_END_DASH";
90
+ State[State["COMMENT_END"] = 50] = "COMMENT_END";
91
+ State[State["COMMENT_END_BANG"] = 51] = "COMMENT_END_BANG";
92
+ State[State["DOCTYPE"] = 52] = "DOCTYPE";
93
+ State[State["BEFORE_DOCTYPE_NAME"] = 53] = "BEFORE_DOCTYPE_NAME";
94
+ State[State["DOCTYPE_NAME"] = 54] = "DOCTYPE_NAME";
95
+ State[State["AFTER_DOCTYPE_NAME"] = 55] = "AFTER_DOCTYPE_NAME";
96
+ State[State["AFTER_DOCTYPE_PUBLIC_KEYWORD"] = 56] = "AFTER_DOCTYPE_PUBLIC_KEYWORD";
97
+ State[State["BEFORE_DOCTYPE_PUBLIC_IDENTIFIER"] = 57] = "BEFORE_DOCTYPE_PUBLIC_IDENTIFIER";
98
+ State[State["DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED"] = 58] = "DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED";
99
+ State[State["DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED"] = 59] = "DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED";
100
+ State[State["AFTER_DOCTYPE_PUBLIC_IDENTIFIER"] = 60] = "AFTER_DOCTYPE_PUBLIC_IDENTIFIER";
101
+ State[State["BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS"] = 61] = "BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS";
102
+ State[State["AFTER_DOCTYPE_SYSTEM_KEYWORD"] = 62] = "AFTER_DOCTYPE_SYSTEM_KEYWORD";
103
+ State[State["BEFORE_DOCTYPE_SYSTEM_IDENTIFIER"] = 63] = "BEFORE_DOCTYPE_SYSTEM_IDENTIFIER";
104
+ State[State["DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED"] = 64] = "DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED";
105
+ State[State["DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED"] = 65] = "DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED";
106
+ State[State["AFTER_DOCTYPE_SYSTEM_IDENTIFIER"] = 66] = "AFTER_DOCTYPE_SYSTEM_IDENTIFIER";
107
+ State[State["BOGUS_DOCTYPE"] = 67] = "BOGUS_DOCTYPE";
108
+ State[State["CDATA_SECTION"] = 68] = "CDATA_SECTION";
109
+ State[State["CDATA_SECTION_BRACKET"] = 69] = "CDATA_SECTION_BRACKET";
110
+ State[State["CDATA_SECTION_END"] = 70] = "CDATA_SECTION_END";
111
+ State[State["CHARACTER_REFERENCE"] = 71] = "CHARACTER_REFERENCE";
112
+ State[State["NAMED_CHARACTER_REFERENCE"] = 72] = "NAMED_CHARACTER_REFERENCE";
113
+ State[State["AMBIGUOUS_AMPERSAND"] = 73] = "AMBIGUOUS_AMPERSAND";
114
+ State[State["NUMERIC_CHARACTER_REFERENCE"] = 74] = "NUMERIC_CHARACTER_REFERENCE";
115
+ State[State["HEXADEMICAL_CHARACTER_REFERENCE_START"] = 75] = "HEXADEMICAL_CHARACTER_REFERENCE_START";
116
+ State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 76] = "HEXADEMICAL_CHARACTER_REFERENCE";
117
+ State[State["DECIMAL_CHARACTER_REFERENCE"] = 77] = "DECIMAL_CHARACTER_REFERENCE";
118
+ State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 78] = "NUMERIC_CHARACTER_REFERENCE_END";
119
+ })(State || (State = {}));
120
+ //Tokenizer initial states for different modes
121
+ export const TokenizerMode = {
122
+ DATA: State.DATA,
123
+ RCDATA: State.RCDATA,
124
+ RAWTEXT: State.RAWTEXT,
125
+ SCRIPT_DATA: State.SCRIPT_DATA,
126
+ PLAINTEXT: State.PLAINTEXT,
127
+ CDATA_SECTION: State.CDATA_SECTION,
128
+ };
129
+ //Utils
130
+ //OPTIMIZATION: these utility functions should not be moved out of this module. V8 Crankshaft will not inline
131
+ //this functions if they will be situated in another module due to context switch.
132
+ //Always perform inlining check before modifying this functions ('node --trace-inlining').
133
+ function isAsciiDigit(cp) {
134
+ return cp >= $.DIGIT_0 && cp <= $.DIGIT_9;
135
+ }
136
+ function isAsciiUpper(cp) {
137
+ return cp >= $.LATIN_CAPITAL_A && cp <= $.LATIN_CAPITAL_Z;
138
+ }
139
+ function isAsciiLower(cp) {
140
+ return cp >= $.LATIN_SMALL_A && cp <= $.LATIN_SMALL_Z;
141
+ }
142
+ function isAsciiLetter(cp) {
143
+ return isAsciiLower(cp) || isAsciiUpper(cp);
144
+ }
145
+ function isAsciiAlphaNumeric(cp) {
146
+ return isAsciiLetter(cp) || isAsciiDigit(cp);
147
+ }
148
+ function isAsciiUpperHexDigit(cp) {
149
+ return cp >= $.LATIN_CAPITAL_A && cp <= $.LATIN_CAPITAL_F;
150
+ }
151
+ function isAsciiLowerHexDigit(cp) {
152
+ return cp >= $.LATIN_SMALL_A && cp <= $.LATIN_SMALL_F;
153
+ }
154
+ function isAsciiHexDigit(cp) {
155
+ return isAsciiDigit(cp) || isAsciiUpperHexDigit(cp) || isAsciiLowerHexDigit(cp);
156
+ }
157
+ function toAsciiLower(cp) {
158
+ return cp + 32;
159
+ }
160
+ function isWhitespace(cp) {
161
+ return cp === $.SPACE || cp === $.LINE_FEED || cp === $.TABULATION || cp === $.FORM_FEED;
162
+ }
163
+ function isEntityInAttributeInvalidEnd(nextCp) {
164
+ return nextCp === $.EQUALS_SIGN || isAsciiAlphaNumeric(nextCp);
165
+ }
166
+ function isScriptDataDoubleEscapeSequenceEnd(cp) {
167
+ return isWhitespace(cp) || cp === $.SOLIDUS || cp === $.GREATER_THAN_SIGN;
168
+ }
169
+ //Tokenizer
170
+ export class Tokenizer {
171
+ constructor(options, handler) {
172
+ this.options = options;
173
+ this.handler = handler;
174
+ this.paused = false;
175
+ /** Ensures that the parsing loop isn't run multiple times at once. */
176
+ this.inLoop = false;
177
+ /**
178
+ * Indicates that the current adjusted node exists, is not an element in the HTML namespace,
179
+ * and that it is not an integration point for either MathML or HTML.
180
+ *
181
+ * @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
182
+ */
183
+ this.inForeignNode = false;
184
+ this.lastStartTagName = '';
185
+ this.active = false;
186
+ this.state = State.DATA;
187
+ this.returnState = State.DATA;
188
+ this.charRefCode = -1;
189
+ this.consumedAfterSnapshot = -1;
190
+ this.currentCharacterToken = null;
191
+ this.currentToken = null;
192
+ this.currentAttr = { name: '', value: '' };
193
+ this.preprocessor = new Preprocessor(handler);
194
+ this.currentLocation = this.getCurrentLocation(-1);
195
+ }
196
+ //Errors
197
+ _err(code) {
198
+ var _a, _b;
199
+ (_b = (_a = this.handler).onParseError) === null || _b === void 0 ? void 0 : _b.call(_a, this.preprocessor.getError(code));
200
+ }
201
+ // NOTE: `offset` may never run across line boundaries.
202
+ getCurrentLocation(offset) {
203
+ if (!this.options.sourceCodeLocationInfo) {
204
+ return null;
205
+ }
206
+ return {
207
+ startLine: this.preprocessor.line,
208
+ startCol: this.preprocessor.col - offset,
209
+ startOffset: this.preprocessor.offset - offset,
210
+ endLine: -1,
211
+ endCol: -1,
212
+ endOffset: -1,
213
+ };
214
+ }
215
+ _runParsingLoop() {
216
+ if (this.inLoop)
217
+ return;
218
+ this.inLoop = true;
219
+ while (this.active && !this.paused) {
220
+ this.consumedAfterSnapshot = 0;
221
+ const cp = this._consume();
222
+ if (!this._ensureHibernation()) {
223
+ this._callState(cp);
224
+ }
225
+ }
226
+ this.inLoop = false;
227
+ }
228
+ //API
229
+ pause() {
230
+ this.paused = true;
231
+ }
232
+ resume(writeCallback) {
233
+ if (!this.paused) {
234
+ throw new Error('Parser was already resumed');
235
+ }
236
+ this.paused = false;
237
+ // Necessary for synchronous resume.
238
+ if (this.inLoop)
239
+ return;
240
+ this._runParsingLoop();
241
+ if (!this.paused) {
242
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
243
+ }
244
+ }
245
+ write(chunk, isLastChunk, writeCallback) {
246
+ this.active = true;
247
+ this.preprocessor.write(chunk, isLastChunk);
248
+ this._runParsingLoop();
249
+ if (!this.paused) {
250
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
251
+ }
252
+ }
253
+ insertHtmlAtCurrentPos(chunk) {
254
+ this.active = true;
255
+ this.preprocessor.insertHtmlAtCurrentPos(chunk);
256
+ this._runParsingLoop();
257
+ }
258
+ //Hibernation
259
+ _ensureHibernation() {
260
+ if (this.preprocessor.endOfChunkHit) {
261
+ this._unconsume(this.consumedAfterSnapshot);
262
+ this.active = false;
263
+ return true;
264
+ }
265
+ return false;
266
+ }
267
+ //Consumption
268
+ _consume() {
269
+ this.consumedAfterSnapshot++;
270
+ return this.preprocessor.advance();
271
+ }
272
+ _unconsume(count) {
273
+ this.consumedAfterSnapshot -= count;
274
+ this.preprocessor.retreat(count);
275
+ }
276
+ _reconsumeInState(state, cp) {
277
+ this.state = state;
278
+ this._callState(cp);
279
+ }
280
+ _advanceBy(count) {
281
+ this.consumedAfterSnapshot += count;
282
+ for (let i = 0; i < count; i++) {
283
+ this.preprocessor.advance();
284
+ }
285
+ }
286
+ _consumeSequenceIfMatch(pattern, caseSensitive) {
287
+ if (this.preprocessor.startsWith(pattern, caseSensitive)) {
288
+ // We will already have consumed one character before calling this method.
289
+ this._advanceBy(pattern.length - 1);
290
+ return true;
291
+ }
292
+ return false;
293
+ }
294
+ //Token creation
295
+ _createStartTagToken() {
296
+ this.currentToken = {
297
+ type: TokenType.START_TAG,
298
+ tagName: '',
299
+ tagID: TAG_ID.UNKNOWN,
300
+ selfClosing: false,
301
+ ackSelfClosing: false,
302
+ attrs: [],
303
+ location: this.getCurrentLocation(1),
304
+ };
305
+ }
306
+ _createEndTagToken() {
307
+ this.currentToken = {
308
+ type: TokenType.END_TAG,
309
+ tagName: '',
310
+ tagID: TAG_ID.UNKNOWN,
311
+ selfClosing: false,
312
+ ackSelfClosing: false,
313
+ attrs: [],
314
+ location: this.getCurrentLocation(2),
315
+ };
316
+ }
317
+ _createCommentToken(offset) {
318
+ this.currentToken = {
319
+ type: TokenType.COMMENT,
320
+ data: '',
321
+ location: this.getCurrentLocation(offset),
322
+ };
323
+ }
324
+ _createDoctypeToken(initialName) {
325
+ this.currentToken = {
326
+ type: TokenType.DOCTYPE,
327
+ name: initialName,
328
+ forceQuirks: false,
329
+ publicId: null,
330
+ systemId: null,
331
+ location: this.currentLocation,
332
+ };
333
+ }
334
+ _createCharacterToken(type, chars) {
335
+ this.currentCharacterToken = {
336
+ type,
337
+ chars,
338
+ location: this.currentLocation,
339
+ };
340
+ }
341
+ //Tag attributes
342
+ _createAttr(attrNameFirstCh) {
343
+ this.currentAttr = {
344
+ name: attrNameFirstCh,
345
+ value: '',
346
+ };
347
+ this.currentLocation = this.getCurrentLocation(0);
348
+ }
349
+ _leaveAttrName() {
350
+ var _a;
351
+ var _b;
352
+ const token = this.currentToken;
353
+ if (getTokenAttr(token, this.currentAttr.name) === null) {
354
+ token.attrs.push(this.currentAttr);
355
+ if (token.location && this.currentLocation) {
356
+ const attrLocations = ((_a = (_b = token.location).attrs) !== null && _a !== void 0 ? _a : (_b.attrs = Object.create(null)));
357
+ attrLocations[this.currentAttr.name] = this.currentLocation;
358
+ // Set end location
359
+ this._leaveAttrValue();
360
+ }
361
+ }
362
+ else {
363
+ this._err(ERR.duplicateAttribute);
364
+ }
365
+ }
366
+ _leaveAttrValue() {
367
+ if (this.currentLocation) {
368
+ this.currentLocation.endLine = this.preprocessor.line;
369
+ this.currentLocation.endCol = this.preprocessor.col;
370
+ this.currentLocation.endOffset = this.preprocessor.offset;
371
+ }
372
+ }
373
+ //Token emission
374
+ prepareToken(ct) {
375
+ this._emitCurrentCharacterToken(ct.location);
376
+ this.currentToken = null;
377
+ if (ct.location) {
378
+ ct.location.endLine = this.preprocessor.line;
379
+ ct.location.endCol = this.preprocessor.col + 1;
380
+ ct.location.endOffset = this.preprocessor.offset + 1;
381
+ }
382
+ this.currentLocation = this.getCurrentLocation(-1);
383
+ }
384
+ emitCurrentTagToken() {
385
+ const ct = this.currentToken;
386
+ this.prepareToken(ct);
387
+ ct.tagID = getTagID(ct.tagName);
388
+ if (ct.type === TokenType.START_TAG) {
389
+ this.lastStartTagName = ct.tagName;
390
+ this.handler.onStartTag(ct);
391
+ }
392
+ else {
393
+ if (ct.attrs.length > 0) {
394
+ this._err(ERR.endTagWithAttributes);
395
+ }
396
+ if (ct.selfClosing) {
397
+ this._err(ERR.endTagWithTrailingSolidus);
398
+ }
399
+ this.handler.onEndTag(ct);
400
+ }
401
+ this.preprocessor.dropParsedChunk();
402
+ }
403
+ emitCurrentComment(ct) {
404
+ this.prepareToken(ct);
405
+ this.handler.onComment(ct);
406
+ this.preprocessor.dropParsedChunk();
407
+ }
408
+ emitCurrentDoctype(ct) {
409
+ this.prepareToken(ct);
410
+ this.handler.onDoctype(ct);
411
+ this.preprocessor.dropParsedChunk();
412
+ }
413
+ _emitCurrentCharacterToken(nextLocation) {
414
+ if (this.currentCharacterToken) {
415
+ //NOTE: if we have a pending character token, make it's end location equal to the
416
+ //current token's start location.
417
+ if (nextLocation && this.currentCharacterToken.location) {
418
+ this.currentCharacterToken.location.endLine = nextLocation.startLine;
419
+ this.currentCharacterToken.location.endCol = nextLocation.startCol;
420
+ this.currentCharacterToken.location.endOffset = nextLocation.startOffset;
421
+ }
422
+ switch (this.currentCharacterToken.type) {
423
+ case TokenType.CHARACTER: {
424
+ this.handler.onCharacter(this.currentCharacterToken);
425
+ break;
426
+ }
427
+ case TokenType.NULL_CHARACTER: {
428
+ this.handler.onNullCharacter(this.currentCharacterToken);
429
+ break;
430
+ }
431
+ case TokenType.WHITESPACE_CHARACTER: {
432
+ this.handler.onWhitespaceCharacter(this.currentCharacterToken);
433
+ break;
434
+ }
435
+ }
436
+ this.currentCharacterToken = null;
437
+ }
438
+ }
439
+ _emitEOFToken() {
440
+ const location = this.getCurrentLocation(0);
441
+ if (location) {
442
+ location.endLine = location.startLine;
443
+ location.endCol = location.startCol;
444
+ location.endOffset = location.startOffset;
445
+ }
446
+ this._emitCurrentCharacterToken(location);
447
+ this.handler.onEof({ type: TokenType.EOF, location });
448
+ this.active = false;
449
+ }
450
+ //Characters emission
451
+ //OPTIMIZATION: specification uses only one type of character tokens (one token per character).
452
+ //This causes a huge memory overhead and a lot of unnecessary parser loops. parse5 uses 3 groups of characters.
453
+ //If we have a sequence of characters that belong to the same group, the parser can process it
454
+ //as a single solid character token.
455
+ //So, there are 3 types of character tokens in parse5:
456
+ //1)TokenType.NULL_CHARACTER - \u0000-character sequences (e.g. '\u0000\u0000\u0000')
457
+ //2)TokenType.WHITESPACE_CHARACTER - any whitespace/new-line character sequences (e.g. '\n \r\t \f')
458
+ //3)TokenType.CHARACTER - any character sequence which don't belong to groups 1 and 2 (e.g. 'abcdef1234@@#$%^')
459
+ _appendCharToCurrentCharacterToken(type, ch) {
460
+ if (this.currentCharacterToken) {
461
+ if (this.currentCharacterToken.type !== type) {
462
+ this.currentLocation = this.getCurrentLocation(0);
463
+ this._emitCurrentCharacterToken(this.currentLocation);
464
+ this.preprocessor.dropParsedChunk();
465
+ }
466
+ else {
467
+ this.currentCharacterToken.chars += ch;
468
+ return;
469
+ }
470
+ }
471
+ this._createCharacterToken(type, ch);
472
+ }
473
+ _emitCodePoint(cp) {
474
+ const type = isWhitespace(cp)
475
+ ? TokenType.WHITESPACE_CHARACTER
476
+ : cp === $.NULL
477
+ ? TokenType.NULL_CHARACTER
478
+ : TokenType.CHARACTER;
479
+ this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
480
+ }
481
+ //NOTE: used when we emit characters explicitly.
482
+ //This is always for non-whitespace and non-null characters, which allows us to avoid additional checks.
483
+ _emitChars(ch) {
484
+ this._appendCharToCurrentCharacterToken(TokenType.CHARACTER, ch);
485
+ }
486
+ // Character reference helpers
487
+ _matchNamedCharacterReference(cp) {
488
+ let result = null;
489
+ let excess = 0;
490
+ let withoutSemicolon = false;
491
+ for (let i = 0, current = htmlDecodeTree[0]; i >= 0; cp = this._consume()) {
492
+ i = determineBranch(htmlDecodeTree, current, i + 1, cp);
493
+ if (i < 0)
494
+ break;
495
+ excess += 1;
496
+ current = htmlDecodeTree[i];
497
+ const masked = current & BinTrieFlags.VALUE_LENGTH;
498
+ // If the branch is a value, store it and continue
499
+ if (masked) {
500
+ // The mask is the number of bytes of the value, including the current byte.
501
+ const valueLength = (masked >> 14) - 1;
502
+ // Attribute values that aren't terminated properly aren't parsed, and shouldn't lead to a parser error.
503
+ // See the example in https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state
504
+ if (cp !== $.SEMICOLON &&
505
+ this._isCharacterReferenceInAttribute() &&
506
+ isEntityInAttributeInvalidEnd(this.preprocessor.peek(1))) {
507
+ //NOTE: we don't flush all consumed code points here, and instead switch back to the original state after
508
+ //emitting an ampersand. This is fine, as alphanumeric characters won't be parsed differently in attributes.
509
+ result = [$.AMPERSAND];
510
+ // Skip over the value.
511
+ i += valueLength;
512
+ }
513
+ else {
514
+ // If this is a surrogate pair, consume the next two bytes.
515
+ result =
516
+ valueLength === 0
517
+ ? [htmlDecodeTree[i] & ~BinTrieFlags.VALUE_LENGTH]
518
+ : valueLength === 1
519
+ ? [htmlDecodeTree[++i]]
520
+ : [htmlDecodeTree[++i], htmlDecodeTree[++i]];
521
+ excess = 0;
522
+ withoutSemicolon = cp !== $.SEMICOLON;
523
+ }
524
+ if (valueLength === 0) {
525
+ // If the value is zero-length, we're done.
526
+ this._consume();
527
+ break;
528
+ }
529
+ }
530
+ }
531
+ this._unconsume(excess);
532
+ if (withoutSemicolon && !this.preprocessor.endOfChunkHit) {
533
+ this._err(ERR.missingSemicolonAfterCharacterReference);
534
+ }
535
+ // We want to emit the error above on the code point after the entity.
536
+ // We always consume one code point too many in the loop, and we wait to
537
+ // unconsume it until after the error is emitted.
538
+ this._unconsume(1);
539
+ return result;
540
+ }
541
+ _isCharacterReferenceInAttribute() {
542
+ return (this.returnState === State.ATTRIBUTE_VALUE_DOUBLE_QUOTED ||
543
+ this.returnState === State.ATTRIBUTE_VALUE_SINGLE_QUOTED ||
544
+ this.returnState === State.ATTRIBUTE_VALUE_UNQUOTED);
545
+ }
546
+ _flushCodePointConsumedAsCharacterReference(cp) {
547
+ if (this._isCharacterReferenceInAttribute()) {
548
+ this.currentAttr.value += String.fromCodePoint(cp);
549
+ }
550
+ else {
551
+ this._emitCodePoint(cp);
552
+ }
553
+ }
554
+ // Calling states this way turns out to be much faster than any other approach.
555
+ _callState(cp) {
556
+ switch (this.state) {
557
+ case State.DATA: {
558
+ this._stateData(cp);
559
+ break;
560
+ }
561
+ case State.RCDATA: {
562
+ this._stateRcdata(cp);
563
+ break;
564
+ }
565
+ case State.RAWTEXT: {
566
+ this._stateRawtext(cp);
567
+ break;
568
+ }
569
+ case State.SCRIPT_DATA: {
570
+ this._stateScriptData(cp);
571
+ break;
572
+ }
573
+ case State.PLAINTEXT: {
574
+ this._statePlaintext(cp);
575
+ break;
576
+ }
577
+ case State.TAG_OPEN: {
578
+ this._stateTagOpen(cp);
579
+ break;
580
+ }
581
+ case State.END_TAG_OPEN: {
582
+ this._stateEndTagOpen(cp);
583
+ break;
584
+ }
585
+ case State.TAG_NAME: {
586
+ this._stateTagName(cp);
587
+ break;
588
+ }
589
+ case State.RCDATA_LESS_THAN_SIGN: {
590
+ this._stateRcdataLessThanSign(cp);
591
+ break;
592
+ }
593
+ case State.RCDATA_END_TAG_OPEN: {
594
+ this._stateRcdataEndTagOpen(cp);
595
+ break;
596
+ }
597
+ case State.RCDATA_END_TAG_NAME: {
598
+ this._stateRcdataEndTagName(cp);
599
+ break;
600
+ }
601
+ case State.RAWTEXT_LESS_THAN_SIGN: {
602
+ this._stateRawtextLessThanSign(cp);
603
+ break;
604
+ }
605
+ case State.RAWTEXT_END_TAG_OPEN: {
606
+ this._stateRawtextEndTagOpen(cp);
607
+ break;
608
+ }
609
+ case State.RAWTEXT_END_TAG_NAME: {
610
+ this._stateRawtextEndTagName(cp);
611
+ break;
612
+ }
613
+ case State.SCRIPT_DATA_LESS_THAN_SIGN: {
614
+ this._stateScriptDataLessThanSign(cp);
615
+ break;
616
+ }
617
+ case State.SCRIPT_DATA_END_TAG_OPEN: {
618
+ this._stateScriptDataEndTagOpen(cp);
619
+ break;
620
+ }
621
+ case State.SCRIPT_DATA_END_TAG_NAME: {
622
+ this._stateScriptDataEndTagName(cp);
623
+ break;
624
+ }
625
+ case State.SCRIPT_DATA_ESCAPE_START: {
626
+ this._stateScriptDataEscapeStart(cp);
627
+ break;
628
+ }
629
+ case State.SCRIPT_DATA_ESCAPE_START_DASH: {
630
+ this._stateScriptDataEscapeStartDash(cp);
631
+ break;
632
+ }
633
+ case State.SCRIPT_DATA_ESCAPED: {
634
+ this._stateScriptDataEscaped(cp);
635
+ break;
636
+ }
637
+ case State.SCRIPT_DATA_ESCAPED_DASH: {
638
+ this._stateScriptDataEscapedDash(cp);
639
+ break;
640
+ }
641
+ case State.SCRIPT_DATA_ESCAPED_DASH_DASH: {
642
+ this._stateScriptDataEscapedDashDash(cp);
643
+ break;
644
+ }
645
+ case State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN: {
646
+ this._stateScriptDataEscapedLessThanSign(cp);
647
+ break;
648
+ }
649
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN: {
650
+ this._stateScriptDataEscapedEndTagOpen(cp);
651
+ break;
652
+ }
653
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_NAME: {
654
+ this._stateScriptDataEscapedEndTagName(cp);
655
+ break;
656
+ }
657
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_START: {
658
+ this._stateScriptDataDoubleEscapeStart(cp);
659
+ break;
660
+ }
661
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED: {
662
+ this._stateScriptDataDoubleEscaped(cp);
663
+ break;
664
+ }
665
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH: {
666
+ this._stateScriptDataDoubleEscapedDash(cp);
667
+ break;
668
+ }
669
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH: {
670
+ this._stateScriptDataDoubleEscapedDashDash(cp);
671
+ break;
672
+ }
673
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN: {
674
+ this._stateScriptDataDoubleEscapedLessThanSign(cp);
675
+ break;
676
+ }
677
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_END: {
678
+ this._stateScriptDataDoubleEscapeEnd(cp);
679
+ break;
680
+ }
681
+ case State.BEFORE_ATTRIBUTE_NAME: {
682
+ this._stateBeforeAttributeName(cp);
683
+ break;
684
+ }
685
+ case State.ATTRIBUTE_NAME: {
686
+ this._stateAttributeName(cp);
687
+ break;
688
+ }
689
+ case State.AFTER_ATTRIBUTE_NAME: {
690
+ this._stateAfterAttributeName(cp);
691
+ break;
692
+ }
693
+ case State.BEFORE_ATTRIBUTE_VALUE: {
694
+ this._stateBeforeAttributeValue(cp);
695
+ break;
696
+ }
697
+ case State.ATTRIBUTE_VALUE_DOUBLE_QUOTED: {
698
+ this._stateAttributeValueDoubleQuoted(cp);
699
+ break;
700
+ }
701
+ case State.ATTRIBUTE_VALUE_SINGLE_QUOTED: {
702
+ this._stateAttributeValueSingleQuoted(cp);
703
+ break;
704
+ }
705
+ case State.ATTRIBUTE_VALUE_UNQUOTED: {
706
+ this._stateAttributeValueUnquoted(cp);
707
+ break;
708
+ }
709
+ case State.AFTER_ATTRIBUTE_VALUE_QUOTED: {
710
+ this._stateAfterAttributeValueQuoted(cp);
711
+ break;
712
+ }
713
+ case State.SELF_CLOSING_START_TAG: {
714
+ this._stateSelfClosingStartTag(cp);
715
+ break;
716
+ }
717
+ case State.BOGUS_COMMENT: {
718
+ this._stateBogusComment(cp);
719
+ break;
720
+ }
721
+ case State.MARKUP_DECLARATION_OPEN: {
722
+ this._stateMarkupDeclarationOpen(cp);
723
+ break;
724
+ }
725
+ case State.COMMENT_START: {
726
+ this._stateCommentStart(cp);
727
+ break;
728
+ }
729
+ case State.COMMENT_START_DASH: {
730
+ this._stateCommentStartDash(cp);
731
+ break;
732
+ }
733
+ case State.COMMENT: {
734
+ this._stateComment(cp);
735
+ break;
736
+ }
737
+ case State.COMMENT_LESS_THAN_SIGN: {
738
+ this._stateCommentLessThanSign(cp);
739
+ break;
740
+ }
741
+ case State.COMMENT_LESS_THAN_SIGN_BANG: {
742
+ this._stateCommentLessThanSignBang(cp);
743
+ break;
744
+ }
745
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH: {
746
+ this._stateCommentLessThanSignBangDash(cp);
747
+ break;
748
+ }
749
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH: {
750
+ this._stateCommentLessThanSignBangDashDash(cp);
751
+ break;
752
+ }
753
+ case State.COMMENT_END_DASH: {
754
+ this._stateCommentEndDash(cp);
755
+ break;
756
+ }
757
+ case State.COMMENT_END: {
758
+ this._stateCommentEnd(cp);
759
+ break;
760
+ }
761
+ case State.COMMENT_END_BANG: {
762
+ this._stateCommentEndBang(cp);
763
+ break;
764
+ }
765
+ case State.DOCTYPE: {
766
+ this._stateDoctype(cp);
767
+ break;
768
+ }
769
+ case State.BEFORE_DOCTYPE_NAME: {
770
+ this._stateBeforeDoctypeName(cp);
771
+ break;
772
+ }
773
+ case State.DOCTYPE_NAME: {
774
+ this._stateDoctypeName(cp);
775
+ break;
776
+ }
777
+ case State.AFTER_DOCTYPE_NAME: {
778
+ this._stateAfterDoctypeName(cp);
779
+ break;
780
+ }
781
+ case State.AFTER_DOCTYPE_PUBLIC_KEYWORD: {
782
+ this._stateAfterDoctypePublicKeyword(cp);
783
+ break;
784
+ }
785
+ case State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER: {
786
+ this._stateBeforeDoctypePublicIdentifier(cp);
787
+ break;
788
+ }
789
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED: {
790
+ this._stateDoctypePublicIdentifierDoubleQuoted(cp);
791
+ break;
792
+ }
793
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED: {
794
+ this._stateDoctypePublicIdentifierSingleQuoted(cp);
795
+ break;
796
+ }
797
+ case State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER: {
798
+ this._stateAfterDoctypePublicIdentifier(cp);
799
+ break;
800
+ }
801
+ case State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS: {
802
+ this._stateBetweenDoctypePublicAndSystemIdentifiers(cp);
803
+ break;
804
+ }
805
+ case State.AFTER_DOCTYPE_SYSTEM_KEYWORD: {
806
+ this._stateAfterDoctypeSystemKeyword(cp);
807
+ break;
808
+ }
809
+ case State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER: {
810
+ this._stateBeforeDoctypeSystemIdentifier(cp);
811
+ break;
812
+ }
813
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED: {
814
+ this._stateDoctypeSystemIdentifierDoubleQuoted(cp);
815
+ break;
816
+ }
817
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED: {
818
+ this._stateDoctypeSystemIdentifierSingleQuoted(cp);
819
+ break;
820
+ }
821
+ case State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER: {
822
+ this._stateAfterDoctypeSystemIdentifier(cp);
823
+ break;
824
+ }
825
+ case State.BOGUS_DOCTYPE: {
826
+ this._stateBogusDoctype(cp);
827
+ break;
828
+ }
829
+ case State.CDATA_SECTION: {
830
+ this._stateCdataSection(cp);
831
+ break;
832
+ }
833
+ case State.CDATA_SECTION_BRACKET: {
834
+ this._stateCdataSectionBracket(cp);
835
+ break;
836
+ }
837
+ case State.CDATA_SECTION_END: {
838
+ this._stateCdataSectionEnd(cp);
839
+ break;
840
+ }
841
+ case State.CHARACTER_REFERENCE: {
842
+ this._stateCharacterReference(cp);
843
+ break;
844
+ }
845
+ case State.NAMED_CHARACTER_REFERENCE: {
846
+ this._stateNamedCharacterReference(cp);
847
+ break;
848
+ }
849
+ case State.AMBIGUOUS_AMPERSAND: {
850
+ this._stateAmbiguousAmpersand(cp);
851
+ break;
852
+ }
853
+ case State.NUMERIC_CHARACTER_REFERENCE: {
854
+ this._stateNumericCharacterReference(cp);
855
+ break;
856
+ }
857
+ case State.HEXADEMICAL_CHARACTER_REFERENCE_START: {
858
+ this._stateHexademicalCharacterReferenceStart(cp);
859
+ break;
860
+ }
861
+ case State.HEXADEMICAL_CHARACTER_REFERENCE: {
862
+ this._stateHexademicalCharacterReference(cp);
863
+ break;
864
+ }
865
+ case State.DECIMAL_CHARACTER_REFERENCE: {
866
+ this._stateDecimalCharacterReference(cp);
867
+ break;
868
+ }
869
+ case State.NUMERIC_CHARACTER_REFERENCE_END: {
870
+ this._stateNumericCharacterReferenceEnd(cp);
871
+ break;
872
+ }
873
+ default: {
874
+ throw new Error('Unknown state');
875
+ }
876
+ }
877
+ }
878
+ // State machine
879
+ // Data state
880
+ //------------------------------------------------------------------
881
+ _stateData(cp) {
882
+ switch (cp) {
883
+ case $.LESS_THAN_SIGN: {
884
+ this.state = State.TAG_OPEN;
885
+ break;
886
+ }
887
+ case $.AMPERSAND: {
888
+ this.returnState = State.DATA;
889
+ this.state = State.CHARACTER_REFERENCE;
890
+ break;
891
+ }
892
+ case $.NULL: {
893
+ this._err(ERR.unexpectedNullCharacter);
894
+ this._emitCodePoint(cp);
895
+ break;
896
+ }
897
+ case $.EOF: {
898
+ this._emitEOFToken();
899
+ break;
900
+ }
901
+ default: {
902
+ this._emitCodePoint(cp);
903
+ }
904
+ }
905
+ }
906
+ // RCDATA state
907
+ //------------------------------------------------------------------
908
+ _stateRcdata(cp) {
909
+ switch (cp) {
910
+ case $.AMPERSAND: {
911
+ this.returnState = State.RCDATA;
912
+ this.state = State.CHARACTER_REFERENCE;
913
+ break;
914
+ }
915
+ case $.LESS_THAN_SIGN: {
916
+ this.state = State.RCDATA_LESS_THAN_SIGN;
917
+ break;
918
+ }
919
+ case $.NULL: {
920
+ this._err(ERR.unexpectedNullCharacter);
921
+ this._emitChars(REPLACEMENT_CHARACTER);
922
+ break;
923
+ }
924
+ case $.EOF: {
925
+ this._emitEOFToken();
926
+ break;
927
+ }
928
+ default: {
929
+ this._emitCodePoint(cp);
930
+ }
931
+ }
932
+ }
933
+ // RAWTEXT state
934
+ //------------------------------------------------------------------
935
+ _stateRawtext(cp) {
936
+ switch (cp) {
937
+ case $.LESS_THAN_SIGN: {
938
+ this.state = State.RAWTEXT_LESS_THAN_SIGN;
939
+ break;
940
+ }
941
+ case $.NULL: {
942
+ this._err(ERR.unexpectedNullCharacter);
943
+ this._emitChars(REPLACEMENT_CHARACTER);
944
+ break;
945
+ }
946
+ case $.EOF: {
947
+ this._emitEOFToken();
948
+ break;
949
+ }
950
+ default: {
951
+ this._emitCodePoint(cp);
952
+ }
953
+ }
954
+ }
955
+ // Script data state
956
+ //------------------------------------------------------------------
957
+ _stateScriptData(cp) {
958
+ switch (cp) {
959
+ case $.LESS_THAN_SIGN: {
960
+ this.state = State.SCRIPT_DATA_LESS_THAN_SIGN;
961
+ break;
962
+ }
963
+ case $.NULL: {
964
+ this._err(ERR.unexpectedNullCharacter);
965
+ this._emitChars(REPLACEMENT_CHARACTER);
966
+ break;
967
+ }
968
+ case $.EOF: {
969
+ this._emitEOFToken();
970
+ break;
971
+ }
972
+ default: {
973
+ this._emitCodePoint(cp);
974
+ }
975
+ }
976
+ }
977
+ // PLAINTEXT state
978
+ //------------------------------------------------------------------
979
+ _statePlaintext(cp) {
980
+ switch (cp) {
981
+ case $.NULL: {
982
+ this._err(ERR.unexpectedNullCharacter);
983
+ this._emitChars(REPLACEMENT_CHARACTER);
984
+ break;
985
+ }
986
+ case $.EOF: {
987
+ this._emitEOFToken();
988
+ break;
989
+ }
990
+ default: {
991
+ this._emitCodePoint(cp);
992
+ }
993
+ }
994
+ }
995
+ // Tag open state
996
+ //------------------------------------------------------------------
997
+ _stateTagOpen(cp) {
998
+ if (isAsciiLetter(cp)) {
999
+ this._createStartTagToken();
1000
+ this.state = State.TAG_NAME;
1001
+ this._stateTagName(cp);
1002
+ }
1003
+ else
1004
+ switch (cp) {
1005
+ case $.EXCLAMATION_MARK: {
1006
+ this.state = State.MARKUP_DECLARATION_OPEN;
1007
+ break;
1008
+ }
1009
+ case $.SOLIDUS: {
1010
+ this.state = State.END_TAG_OPEN;
1011
+ break;
1012
+ }
1013
+ case $.QUESTION_MARK: {
1014
+ this._err(ERR.unexpectedQuestionMarkInsteadOfTagName);
1015
+ this._createCommentToken(1);
1016
+ this.state = State.BOGUS_COMMENT;
1017
+ this._stateBogusComment(cp);
1018
+ break;
1019
+ }
1020
+ case $.EOF: {
1021
+ this._err(ERR.eofBeforeTagName);
1022
+ this._emitChars('<');
1023
+ this._emitEOFToken();
1024
+ break;
1025
+ }
1026
+ default: {
1027
+ this._err(ERR.invalidFirstCharacterOfTagName);
1028
+ this._emitChars('<');
1029
+ this.state = State.DATA;
1030
+ this._stateData(cp);
1031
+ }
1032
+ }
1033
+ }
1034
+ // End tag open state
1035
+ //------------------------------------------------------------------
1036
+ _stateEndTagOpen(cp) {
1037
+ if (isAsciiLetter(cp)) {
1038
+ this._createEndTagToken();
1039
+ this.state = State.TAG_NAME;
1040
+ this._stateTagName(cp);
1041
+ }
1042
+ else
1043
+ switch (cp) {
1044
+ case $.GREATER_THAN_SIGN: {
1045
+ this._err(ERR.missingEndTagName);
1046
+ this.state = State.DATA;
1047
+ break;
1048
+ }
1049
+ case $.EOF: {
1050
+ this._err(ERR.eofBeforeTagName);
1051
+ this._emitChars('</');
1052
+ this._emitEOFToken();
1053
+ break;
1054
+ }
1055
+ default: {
1056
+ this._err(ERR.invalidFirstCharacterOfTagName);
1057
+ this._createCommentToken(2);
1058
+ this.state = State.BOGUS_COMMENT;
1059
+ this._stateBogusComment(cp);
1060
+ }
1061
+ }
1062
+ }
1063
+ // Tag name state
1064
+ //------------------------------------------------------------------
1065
+ _stateTagName(cp) {
1066
+ const token = this.currentToken;
1067
+ switch (cp) {
1068
+ case $.SPACE:
1069
+ case $.LINE_FEED:
1070
+ case $.TABULATION:
1071
+ case $.FORM_FEED: {
1072
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1073
+ break;
1074
+ }
1075
+ case $.SOLIDUS: {
1076
+ this.state = State.SELF_CLOSING_START_TAG;
1077
+ break;
1078
+ }
1079
+ case $.GREATER_THAN_SIGN: {
1080
+ this.state = State.DATA;
1081
+ this.emitCurrentTagToken();
1082
+ break;
1083
+ }
1084
+ case $.NULL: {
1085
+ this._err(ERR.unexpectedNullCharacter);
1086
+ token.tagName += REPLACEMENT_CHARACTER;
1087
+ break;
1088
+ }
1089
+ case $.EOF: {
1090
+ this._err(ERR.eofInTag);
1091
+ this._emitEOFToken();
1092
+ break;
1093
+ }
1094
+ default: {
1095
+ token.tagName += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1096
+ }
1097
+ }
1098
+ }
1099
+ // RCDATA less-than sign state
1100
+ //------------------------------------------------------------------
1101
+ _stateRcdataLessThanSign(cp) {
1102
+ if (cp === $.SOLIDUS) {
1103
+ this.state = State.RCDATA_END_TAG_OPEN;
1104
+ }
1105
+ else {
1106
+ this._emitChars('<');
1107
+ this.state = State.RCDATA;
1108
+ this._stateRcdata(cp);
1109
+ }
1110
+ }
1111
+ // RCDATA end tag open state
1112
+ //------------------------------------------------------------------
1113
+ _stateRcdataEndTagOpen(cp) {
1114
+ if (isAsciiLetter(cp)) {
1115
+ this.state = State.RCDATA_END_TAG_NAME;
1116
+ this._stateRcdataEndTagName(cp);
1117
+ }
1118
+ else {
1119
+ this._emitChars('</');
1120
+ this.state = State.RCDATA;
1121
+ this._stateRcdata(cp);
1122
+ }
1123
+ }
1124
+ handleSpecialEndTag(_cp) {
1125
+ if (!this.preprocessor.startsWith(this.lastStartTagName, false)) {
1126
+ return !this._ensureHibernation();
1127
+ }
1128
+ this._createEndTagToken();
1129
+ const token = this.currentToken;
1130
+ token.tagName = this.lastStartTagName;
1131
+ const cp = this.preprocessor.peek(this.lastStartTagName.length);
1132
+ switch (cp) {
1133
+ case $.SPACE:
1134
+ case $.LINE_FEED:
1135
+ case $.TABULATION:
1136
+ case $.FORM_FEED: {
1137
+ this._advanceBy(this.lastStartTagName.length);
1138
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1139
+ return false;
1140
+ }
1141
+ case $.SOLIDUS: {
1142
+ this._advanceBy(this.lastStartTagName.length);
1143
+ this.state = State.SELF_CLOSING_START_TAG;
1144
+ return false;
1145
+ }
1146
+ case $.GREATER_THAN_SIGN: {
1147
+ this._advanceBy(this.lastStartTagName.length);
1148
+ this.emitCurrentTagToken();
1149
+ this.state = State.DATA;
1150
+ return false;
1151
+ }
1152
+ default: {
1153
+ return !this._ensureHibernation();
1154
+ }
1155
+ }
1156
+ }
1157
+ // RCDATA end tag name state
1158
+ //------------------------------------------------------------------
1159
+ _stateRcdataEndTagName(cp) {
1160
+ if (this.handleSpecialEndTag(cp)) {
1161
+ this._emitChars('</');
1162
+ this.state = State.RCDATA;
1163
+ this._stateRcdata(cp);
1164
+ }
1165
+ }
1166
+ // RAWTEXT less-than sign state
1167
+ //------------------------------------------------------------------
1168
+ _stateRawtextLessThanSign(cp) {
1169
+ if (cp === $.SOLIDUS) {
1170
+ this.state = State.RAWTEXT_END_TAG_OPEN;
1171
+ }
1172
+ else {
1173
+ this._emitChars('<');
1174
+ this.state = State.RAWTEXT;
1175
+ this._stateRawtext(cp);
1176
+ }
1177
+ }
1178
+ // RAWTEXT end tag open state
1179
+ //------------------------------------------------------------------
1180
+ _stateRawtextEndTagOpen(cp) {
1181
+ if (isAsciiLetter(cp)) {
1182
+ this.state = State.RAWTEXT_END_TAG_NAME;
1183
+ this._stateRawtextEndTagName(cp);
1184
+ }
1185
+ else {
1186
+ this._emitChars('</');
1187
+ this.state = State.RAWTEXT;
1188
+ this._stateRawtext(cp);
1189
+ }
1190
+ }
1191
+ // RAWTEXT end tag name state
1192
+ //------------------------------------------------------------------
1193
+ _stateRawtextEndTagName(cp) {
1194
+ if (this.handleSpecialEndTag(cp)) {
1195
+ this._emitChars('</');
1196
+ this.state = State.RAWTEXT;
1197
+ this._stateRawtext(cp);
1198
+ }
1199
+ }
1200
+ // Script data less-than sign state
1201
+ //------------------------------------------------------------------
1202
+ _stateScriptDataLessThanSign(cp) {
1203
+ switch (cp) {
1204
+ case $.SOLIDUS: {
1205
+ this.state = State.SCRIPT_DATA_END_TAG_OPEN;
1206
+ break;
1207
+ }
1208
+ case $.EXCLAMATION_MARK: {
1209
+ this.state = State.SCRIPT_DATA_ESCAPE_START;
1210
+ this._emitChars('<!');
1211
+ break;
1212
+ }
1213
+ default: {
1214
+ this._emitChars('<');
1215
+ this.state = State.SCRIPT_DATA;
1216
+ this._stateScriptData(cp);
1217
+ }
1218
+ }
1219
+ }
1220
+ // Script data end tag open state
1221
+ //------------------------------------------------------------------
1222
+ _stateScriptDataEndTagOpen(cp) {
1223
+ if (isAsciiLetter(cp)) {
1224
+ this.state = State.SCRIPT_DATA_END_TAG_NAME;
1225
+ this._stateScriptDataEndTagName(cp);
1226
+ }
1227
+ else {
1228
+ this._emitChars('</');
1229
+ this.state = State.SCRIPT_DATA;
1230
+ this._stateScriptData(cp);
1231
+ }
1232
+ }
1233
+ // Script data end tag name state
1234
+ //------------------------------------------------------------------
1235
+ _stateScriptDataEndTagName(cp) {
1236
+ if (this.handleSpecialEndTag(cp)) {
1237
+ this._emitChars('</');
1238
+ this.state = State.SCRIPT_DATA;
1239
+ this._stateScriptData(cp);
1240
+ }
1241
+ }
1242
+ // Script data escape start state
1243
+ //------------------------------------------------------------------
1244
+ _stateScriptDataEscapeStart(cp) {
1245
+ if (cp === $.HYPHEN_MINUS) {
1246
+ this.state = State.SCRIPT_DATA_ESCAPE_START_DASH;
1247
+ this._emitChars('-');
1248
+ }
1249
+ else {
1250
+ this.state = State.SCRIPT_DATA;
1251
+ this._stateScriptData(cp);
1252
+ }
1253
+ }
1254
+ // Script data escape start dash state
1255
+ //------------------------------------------------------------------
1256
+ _stateScriptDataEscapeStartDash(cp) {
1257
+ if (cp === $.HYPHEN_MINUS) {
1258
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1259
+ this._emitChars('-');
1260
+ }
1261
+ else {
1262
+ this.state = State.SCRIPT_DATA;
1263
+ this._stateScriptData(cp);
1264
+ }
1265
+ }
1266
+ // Script data escaped state
1267
+ //------------------------------------------------------------------
1268
+ _stateScriptDataEscaped(cp) {
1269
+ switch (cp) {
1270
+ case $.HYPHEN_MINUS: {
1271
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH;
1272
+ this._emitChars('-');
1273
+ break;
1274
+ }
1275
+ case $.LESS_THAN_SIGN: {
1276
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1277
+ break;
1278
+ }
1279
+ case $.NULL: {
1280
+ this._err(ERR.unexpectedNullCharacter);
1281
+ this._emitChars(REPLACEMENT_CHARACTER);
1282
+ break;
1283
+ }
1284
+ case $.EOF: {
1285
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1286
+ this._emitEOFToken();
1287
+ break;
1288
+ }
1289
+ default: {
1290
+ this._emitCodePoint(cp);
1291
+ }
1292
+ }
1293
+ }
1294
+ // Script data escaped dash state
1295
+ //------------------------------------------------------------------
1296
+ _stateScriptDataEscapedDash(cp) {
1297
+ switch (cp) {
1298
+ case $.HYPHEN_MINUS: {
1299
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1300
+ this._emitChars('-');
1301
+ break;
1302
+ }
1303
+ case $.LESS_THAN_SIGN: {
1304
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1305
+ break;
1306
+ }
1307
+ case $.NULL: {
1308
+ this._err(ERR.unexpectedNullCharacter);
1309
+ this.state = State.SCRIPT_DATA_ESCAPED;
1310
+ this._emitChars(REPLACEMENT_CHARACTER);
1311
+ break;
1312
+ }
1313
+ case $.EOF: {
1314
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1315
+ this._emitEOFToken();
1316
+ break;
1317
+ }
1318
+ default: {
1319
+ this.state = State.SCRIPT_DATA_ESCAPED;
1320
+ this._emitCodePoint(cp);
1321
+ }
1322
+ }
1323
+ }
1324
+ // Script data escaped dash dash state
1325
+ //------------------------------------------------------------------
1326
+ _stateScriptDataEscapedDashDash(cp) {
1327
+ switch (cp) {
1328
+ case $.HYPHEN_MINUS: {
1329
+ this._emitChars('-');
1330
+ break;
1331
+ }
1332
+ case $.LESS_THAN_SIGN: {
1333
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1334
+ break;
1335
+ }
1336
+ case $.GREATER_THAN_SIGN: {
1337
+ this.state = State.SCRIPT_DATA;
1338
+ this._emitChars('>');
1339
+ break;
1340
+ }
1341
+ case $.NULL: {
1342
+ this._err(ERR.unexpectedNullCharacter);
1343
+ this.state = State.SCRIPT_DATA_ESCAPED;
1344
+ this._emitChars(REPLACEMENT_CHARACTER);
1345
+ break;
1346
+ }
1347
+ case $.EOF: {
1348
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1349
+ this._emitEOFToken();
1350
+ break;
1351
+ }
1352
+ default: {
1353
+ this.state = State.SCRIPT_DATA_ESCAPED;
1354
+ this._emitCodePoint(cp);
1355
+ }
1356
+ }
1357
+ }
1358
+ // Script data escaped less-than sign state
1359
+ //------------------------------------------------------------------
1360
+ _stateScriptDataEscapedLessThanSign(cp) {
1361
+ if (cp === $.SOLIDUS) {
1362
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN;
1363
+ }
1364
+ else if (isAsciiLetter(cp)) {
1365
+ this._emitChars('<');
1366
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_START;
1367
+ this._stateScriptDataDoubleEscapeStart(cp);
1368
+ }
1369
+ else {
1370
+ this._emitChars('<');
1371
+ this.state = State.SCRIPT_DATA_ESCAPED;
1372
+ this._stateScriptDataEscaped(cp);
1373
+ }
1374
+ }
1375
+ // Script data escaped end tag open state
1376
+ //------------------------------------------------------------------
1377
+ _stateScriptDataEscapedEndTagOpen(cp) {
1378
+ if (isAsciiLetter(cp)) {
1379
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_NAME;
1380
+ this._stateScriptDataEscapedEndTagName(cp);
1381
+ }
1382
+ else {
1383
+ this._emitChars('</');
1384
+ this.state = State.SCRIPT_DATA_ESCAPED;
1385
+ this._stateScriptDataEscaped(cp);
1386
+ }
1387
+ }
1388
+ // Script data escaped end tag name state
1389
+ //------------------------------------------------------------------
1390
+ _stateScriptDataEscapedEndTagName(cp) {
1391
+ if (this.handleSpecialEndTag(cp)) {
1392
+ this._emitChars('</');
1393
+ this.state = State.SCRIPT_DATA_ESCAPED;
1394
+ this._stateScriptDataEscaped(cp);
1395
+ }
1396
+ }
1397
+ // Script data double escape start state
1398
+ //------------------------------------------------------------------
1399
+ _stateScriptDataDoubleEscapeStart(cp) {
1400
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1401
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1402
+ this._emitCodePoint(cp);
1403
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1404
+ this._emitCodePoint(this._consume());
1405
+ }
1406
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1407
+ }
1408
+ else if (!this._ensureHibernation()) {
1409
+ this.state = State.SCRIPT_DATA_ESCAPED;
1410
+ this._stateScriptDataEscaped(cp);
1411
+ }
1412
+ }
1413
+ // Script data double escaped state
1414
+ //------------------------------------------------------------------
1415
+ _stateScriptDataDoubleEscaped(cp) {
1416
+ switch (cp) {
1417
+ case $.HYPHEN_MINUS: {
1418
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH;
1419
+ this._emitChars('-');
1420
+ break;
1421
+ }
1422
+ case $.LESS_THAN_SIGN: {
1423
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1424
+ this._emitChars('<');
1425
+ break;
1426
+ }
1427
+ case $.NULL: {
1428
+ this._err(ERR.unexpectedNullCharacter);
1429
+ this._emitChars(REPLACEMENT_CHARACTER);
1430
+ break;
1431
+ }
1432
+ case $.EOF: {
1433
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1434
+ this._emitEOFToken();
1435
+ break;
1436
+ }
1437
+ default: {
1438
+ this._emitCodePoint(cp);
1439
+ }
1440
+ }
1441
+ }
1442
+ // Script data double escaped dash state
1443
+ //------------------------------------------------------------------
1444
+ _stateScriptDataDoubleEscapedDash(cp) {
1445
+ switch (cp) {
1446
+ case $.HYPHEN_MINUS: {
1447
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH;
1448
+ this._emitChars('-');
1449
+ break;
1450
+ }
1451
+ case $.LESS_THAN_SIGN: {
1452
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1453
+ this._emitChars('<');
1454
+ break;
1455
+ }
1456
+ case $.NULL: {
1457
+ this._err(ERR.unexpectedNullCharacter);
1458
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1459
+ this._emitChars(REPLACEMENT_CHARACTER);
1460
+ break;
1461
+ }
1462
+ case $.EOF: {
1463
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1464
+ this._emitEOFToken();
1465
+ break;
1466
+ }
1467
+ default: {
1468
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1469
+ this._emitCodePoint(cp);
1470
+ }
1471
+ }
1472
+ }
1473
+ // Script data double escaped dash dash state
1474
+ //------------------------------------------------------------------
1475
+ _stateScriptDataDoubleEscapedDashDash(cp) {
1476
+ switch (cp) {
1477
+ case $.HYPHEN_MINUS: {
1478
+ this._emitChars('-');
1479
+ break;
1480
+ }
1481
+ case $.LESS_THAN_SIGN: {
1482
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1483
+ this._emitChars('<');
1484
+ break;
1485
+ }
1486
+ case $.GREATER_THAN_SIGN: {
1487
+ this.state = State.SCRIPT_DATA;
1488
+ this._emitChars('>');
1489
+ break;
1490
+ }
1491
+ case $.NULL: {
1492
+ this._err(ERR.unexpectedNullCharacter);
1493
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1494
+ this._emitChars(REPLACEMENT_CHARACTER);
1495
+ break;
1496
+ }
1497
+ case $.EOF: {
1498
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1499
+ this._emitEOFToken();
1500
+ break;
1501
+ }
1502
+ default: {
1503
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1504
+ this._emitCodePoint(cp);
1505
+ }
1506
+ }
1507
+ }
1508
+ // Script data double escaped less-than sign state
1509
+ //------------------------------------------------------------------
1510
+ _stateScriptDataDoubleEscapedLessThanSign(cp) {
1511
+ if (cp === $.SOLIDUS) {
1512
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_END;
1513
+ this._emitChars('/');
1514
+ }
1515
+ else {
1516
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1517
+ this._stateScriptDataDoubleEscaped(cp);
1518
+ }
1519
+ }
1520
+ // Script data double escape end state
1521
+ //------------------------------------------------------------------
1522
+ _stateScriptDataDoubleEscapeEnd(cp) {
1523
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1524
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1525
+ this._emitCodePoint(cp);
1526
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1527
+ this._emitCodePoint(this._consume());
1528
+ }
1529
+ this.state = State.SCRIPT_DATA_ESCAPED;
1530
+ }
1531
+ else if (!this._ensureHibernation()) {
1532
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1533
+ this._stateScriptDataDoubleEscaped(cp);
1534
+ }
1535
+ }
1536
+ // Before attribute name state
1537
+ //------------------------------------------------------------------
1538
+ _stateBeforeAttributeName(cp) {
1539
+ switch (cp) {
1540
+ case $.SPACE:
1541
+ case $.LINE_FEED:
1542
+ case $.TABULATION:
1543
+ case $.FORM_FEED: {
1544
+ // Ignore whitespace
1545
+ break;
1546
+ }
1547
+ case $.SOLIDUS:
1548
+ case $.GREATER_THAN_SIGN:
1549
+ case $.EOF: {
1550
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1551
+ this._stateAfterAttributeName(cp);
1552
+ break;
1553
+ }
1554
+ case $.EQUALS_SIGN: {
1555
+ this._err(ERR.unexpectedEqualsSignBeforeAttributeName);
1556
+ this._createAttr('=');
1557
+ this.state = State.ATTRIBUTE_NAME;
1558
+ break;
1559
+ }
1560
+ default: {
1561
+ this._createAttr('');
1562
+ this.state = State.ATTRIBUTE_NAME;
1563
+ this._stateAttributeName(cp);
1564
+ }
1565
+ }
1566
+ }
1567
+ // Attribute name state
1568
+ //------------------------------------------------------------------
1569
+ _stateAttributeName(cp) {
1570
+ switch (cp) {
1571
+ case $.SPACE:
1572
+ case $.LINE_FEED:
1573
+ case $.TABULATION:
1574
+ case $.FORM_FEED:
1575
+ case $.SOLIDUS:
1576
+ case $.GREATER_THAN_SIGN:
1577
+ case $.EOF: {
1578
+ this._leaveAttrName();
1579
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1580
+ this._stateAfterAttributeName(cp);
1581
+ break;
1582
+ }
1583
+ case $.EQUALS_SIGN: {
1584
+ this._leaveAttrName();
1585
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1586
+ break;
1587
+ }
1588
+ case $.QUOTATION_MARK:
1589
+ case $.APOSTROPHE:
1590
+ case $.LESS_THAN_SIGN: {
1591
+ this._err(ERR.unexpectedCharacterInAttributeName);
1592
+ this.currentAttr.name += String.fromCodePoint(cp);
1593
+ break;
1594
+ }
1595
+ case $.NULL: {
1596
+ this._err(ERR.unexpectedNullCharacter);
1597
+ this.currentAttr.name += REPLACEMENT_CHARACTER;
1598
+ break;
1599
+ }
1600
+ default: {
1601
+ this.currentAttr.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1602
+ }
1603
+ }
1604
+ }
1605
+ // After attribute name state
1606
+ //------------------------------------------------------------------
1607
+ _stateAfterAttributeName(cp) {
1608
+ switch (cp) {
1609
+ case $.SPACE:
1610
+ case $.LINE_FEED:
1611
+ case $.TABULATION:
1612
+ case $.FORM_FEED: {
1613
+ // Ignore whitespace
1614
+ break;
1615
+ }
1616
+ case $.SOLIDUS: {
1617
+ this.state = State.SELF_CLOSING_START_TAG;
1618
+ break;
1619
+ }
1620
+ case $.EQUALS_SIGN: {
1621
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1622
+ break;
1623
+ }
1624
+ case $.GREATER_THAN_SIGN: {
1625
+ this.state = State.DATA;
1626
+ this.emitCurrentTagToken();
1627
+ break;
1628
+ }
1629
+ case $.EOF: {
1630
+ this._err(ERR.eofInTag);
1631
+ this._emitEOFToken();
1632
+ break;
1633
+ }
1634
+ default: {
1635
+ this._createAttr('');
1636
+ this.state = State.ATTRIBUTE_NAME;
1637
+ this._stateAttributeName(cp);
1638
+ }
1639
+ }
1640
+ }
1641
+ // Before attribute value state
1642
+ //------------------------------------------------------------------
1643
+ _stateBeforeAttributeValue(cp) {
1644
+ switch (cp) {
1645
+ case $.SPACE:
1646
+ case $.LINE_FEED:
1647
+ case $.TABULATION:
1648
+ case $.FORM_FEED: {
1649
+ // Ignore whitespace
1650
+ break;
1651
+ }
1652
+ case $.QUOTATION_MARK: {
1653
+ this.state = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
1654
+ break;
1655
+ }
1656
+ case $.APOSTROPHE: {
1657
+ this.state = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
1658
+ break;
1659
+ }
1660
+ case $.GREATER_THAN_SIGN: {
1661
+ this._err(ERR.missingAttributeValue);
1662
+ this.state = State.DATA;
1663
+ this.emitCurrentTagToken();
1664
+ break;
1665
+ }
1666
+ default: {
1667
+ this.state = State.ATTRIBUTE_VALUE_UNQUOTED;
1668
+ this._stateAttributeValueUnquoted(cp);
1669
+ }
1670
+ }
1671
+ }
1672
+ // Attribute value (double-quoted) state
1673
+ //------------------------------------------------------------------
1674
+ _stateAttributeValueDoubleQuoted(cp) {
1675
+ switch (cp) {
1676
+ case $.QUOTATION_MARK: {
1677
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1678
+ break;
1679
+ }
1680
+ case $.AMPERSAND: {
1681
+ this.returnState = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
1682
+ this.state = State.CHARACTER_REFERENCE;
1683
+ break;
1684
+ }
1685
+ case $.NULL: {
1686
+ this._err(ERR.unexpectedNullCharacter);
1687
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1688
+ break;
1689
+ }
1690
+ case $.EOF: {
1691
+ this._err(ERR.eofInTag);
1692
+ this._emitEOFToken();
1693
+ break;
1694
+ }
1695
+ default: {
1696
+ this.currentAttr.value += String.fromCodePoint(cp);
1697
+ }
1698
+ }
1699
+ }
1700
+ // Attribute value (single-quoted) state
1701
+ //------------------------------------------------------------------
1702
+ _stateAttributeValueSingleQuoted(cp) {
1703
+ switch (cp) {
1704
+ case $.APOSTROPHE: {
1705
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1706
+ break;
1707
+ }
1708
+ case $.AMPERSAND: {
1709
+ this.returnState = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
1710
+ this.state = State.CHARACTER_REFERENCE;
1711
+ break;
1712
+ }
1713
+ case $.NULL: {
1714
+ this._err(ERR.unexpectedNullCharacter);
1715
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1716
+ break;
1717
+ }
1718
+ case $.EOF: {
1719
+ this._err(ERR.eofInTag);
1720
+ this._emitEOFToken();
1721
+ break;
1722
+ }
1723
+ default: {
1724
+ this.currentAttr.value += String.fromCodePoint(cp);
1725
+ }
1726
+ }
1727
+ }
1728
+ // Attribute value (unquoted) state
1729
+ //------------------------------------------------------------------
1730
+ _stateAttributeValueUnquoted(cp) {
1731
+ switch (cp) {
1732
+ case $.SPACE:
1733
+ case $.LINE_FEED:
1734
+ case $.TABULATION:
1735
+ case $.FORM_FEED: {
1736
+ this._leaveAttrValue();
1737
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1738
+ break;
1739
+ }
1740
+ case $.AMPERSAND: {
1741
+ this.returnState = State.ATTRIBUTE_VALUE_UNQUOTED;
1742
+ this.state = State.CHARACTER_REFERENCE;
1743
+ break;
1744
+ }
1745
+ case $.GREATER_THAN_SIGN: {
1746
+ this._leaveAttrValue();
1747
+ this.state = State.DATA;
1748
+ this.emitCurrentTagToken();
1749
+ break;
1750
+ }
1751
+ case $.NULL: {
1752
+ this._err(ERR.unexpectedNullCharacter);
1753
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1754
+ break;
1755
+ }
1756
+ case $.QUOTATION_MARK:
1757
+ case $.APOSTROPHE:
1758
+ case $.LESS_THAN_SIGN:
1759
+ case $.EQUALS_SIGN:
1760
+ case $.GRAVE_ACCENT: {
1761
+ this._err(ERR.unexpectedCharacterInUnquotedAttributeValue);
1762
+ this.currentAttr.value += String.fromCodePoint(cp);
1763
+ break;
1764
+ }
1765
+ case $.EOF: {
1766
+ this._err(ERR.eofInTag);
1767
+ this._emitEOFToken();
1768
+ break;
1769
+ }
1770
+ default: {
1771
+ this.currentAttr.value += String.fromCodePoint(cp);
1772
+ }
1773
+ }
1774
+ }
1775
+ // After attribute value (quoted) state
1776
+ //------------------------------------------------------------------
1777
+ _stateAfterAttributeValueQuoted(cp) {
1778
+ switch (cp) {
1779
+ case $.SPACE:
1780
+ case $.LINE_FEED:
1781
+ case $.TABULATION:
1782
+ case $.FORM_FEED: {
1783
+ this._leaveAttrValue();
1784
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1785
+ break;
1786
+ }
1787
+ case $.SOLIDUS: {
1788
+ this._leaveAttrValue();
1789
+ this.state = State.SELF_CLOSING_START_TAG;
1790
+ break;
1791
+ }
1792
+ case $.GREATER_THAN_SIGN: {
1793
+ this._leaveAttrValue();
1794
+ this.state = State.DATA;
1795
+ this.emitCurrentTagToken();
1796
+ break;
1797
+ }
1798
+ case $.EOF: {
1799
+ this._err(ERR.eofInTag);
1800
+ this._emitEOFToken();
1801
+ break;
1802
+ }
1803
+ default: {
1804
+ this._err(ERR.missingWhitespaceBetweenAttributes);
1805
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1806
+ this._stateBeforeAttributeName(cp);
1807
+ }
1808
+ }
1809
+ }
1810
+ // Self-closing start tag state
1811
+ //------------------------------------------------------------------
1812
+ _stateSelfClosingStartTag(cp) {
1813
+ switch (cp) {
1814
+ case $.GREATER_THAN_SIGN: {
1815
+ const token = this.currentToken;
1816
+ token.selfClosing = true;
1817
+ this.state = State.DATA;
1818
+ this.emitCurrentTagToken();
1819
+ break;
1820
+ }
1821
+ case $.EOF: {
1822
+ this._err(ERR.eofInTag);
1823
+ this._emitEOFToken();
1824
+ break;
1825
+ }
1826
+ default: {
1827
+ this._err(ERR.unexpectedSolidusInTag);
1828
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1829
+ this._stateBeforeAttributeName(cp);
1830
+ }
1831
+ }
1832
+ }
1833
+ // Bogus comment state
1834
+ //------------------------------------------------------------------
1835
+ _stateBogusComment(cp) {
1836
+ const token = this.currentToken;
1837
+ switch (cp) {
1838
+ case $.GREATER_THAN_SIGN: {
1839
+ this.state = State.DATA;
1840
+ this.emitCurrentComment(token);
1841
+ break;
1842
+ }
1843
+ case $.EOF: {
1844
+ this.emitCurrentComment(token);
1845
+ this._emitEOFToken();
1846
+ break;
1847
+ }
1848
+ case $.NULL: {
1849
+ this._err(ERR.unexpectedNullCharacter);
1850
+ token.data += REPLACEMENT_CHARACTER;
1851
+ break;
1852
+ }
1853
+ default: {
1854
+ token.data += String.fromCodePoint(cp);
1855
+ }
1856
+ }
1857
+ }
1858
+ // Markup declaration open state
1859
+ //------------------------------------------------------------------
1860
+ _stateMarkupDeclarationOpen(cp) {
1861
+ if (this._consumeSequenceIfMatch($$.DASH_DASH, true)) {
1862
+ this._createCommentToken($$.DASH_DASH.length + 1);
1863
+ this.state = State.COMMENT_START;
1864
+ }
1865
+ else if (this._consumeSequenceIfMatch($$.DOCTYPE, false)) {
1866
+ // NOTE: Doctypes tokens are created without fixed offsets. We keep track of the moment a doctype *might* start here.
1867
+ this.currentLocation = this.getCurrentLocation($$.DOCTYPE.length + 1);
1868
+ this.state = State.DOCTYPE;
1869
+ }
1870
+ else if (this._consumeSequenceIfMatch($$.CDATA_START, true)) {
1871
+ if (this.inForeignNode) {
1872
+ this.state = State.CDATA_SECTION;
1873
+ }
1874
+ else {
1875
+ this._err(ERR.cdataInHtmlContent);
1876
+ this._createCommentToken($$.CDATA_START.length + 1);
1877
+ this.currentToken.data = '[CDATA[';
1878
+ this.state = State.BOGUS_COMMENT;
1879
+ }
1880
+ }
1881
+ //NOTE: Sequence lookups can be abrupted by hibernation. In that case, lookup
1882
+ //results are no longer valid and we will need to start over.
1883
+ else if (!this._ensureHibernation()) {
1884
+ this._err(ERR.incorrectlyOpenedComment);
1885
+ this._createCommentToken(2);
1886
+ this.state = State.BOGUS_COMMENT;
1887
+ this._stateBogusComment(cp);
1888
+ }
1889
+ }
1890
+ // Comment start state
1891
+ //------------------------------------------------------------------
1892
+ _stateCommentStart(cp) {
1893
+ switch (cp) {
1894
+ case $.HYPHEN_MINUS: {
1895
+ this.state = State.COMMENT_START_DASH;
1896
+ break;
1897
+ }
1898
+ case $.GREATER_THAN_SIGN: {
1899
+ this._err(ERR.abruptClosingOfEmptyComment);
1900
+ this.state = State.DATA;
1901
+ const token = this.currentToken;
1902
+ this.emitCurrentComment(token);
1903
+ break;
1904
+ }
1905
+ default: {
1906
+ this.state = State.COMMENT;
1907
+ this._stateComment(cp);
1908
+ }
1909
+ }
1910
+ }
1911
+ // Comment start dash state
1912
+ //------------------------------------------------------------------
1913
+ _stateCommentStartDash(cp) {
1914
+ const token = this.currentToken;
1915
+ switch (cp) {
1916
+ case $.HYPHEN_MINUS: {
1917
+ this.state = State.COMMENT_END;
1918
+ break;
1919
+ }
1920
+ case $.GREATER_THAN_SIGN: {
1921
+ this._err(ERR.abruptClosingOfEmptyComment);
1922
+ this.state = State.DATA;
1923
+ this.emitCurrentComment(token);
1924
+ break;
1925
+ }
1926
+ case $.EOF: {
1927
+ this._err(ERR.eofInComment);
1928
+ this.emitCurrentComment(token);
1929
+ this._emitEOFToken();
1930
+ break;
1931
+ }
1932
+ default: {
1933
+ token.data += '-';
1934
+ this.state = State.COMMENT;
1935
+ this._stateComment(cp);
1936
+ }
1937
+ }
1938
+ }
1939
+ // Comment state
1940
+ //------------------------------------------------------------------
1941
+ _stateComment(cp) {
1942
+ const token = this.currentToken;
1943
+ switch (cp) {
1944
+ case $.HYPHEN_MINUS: {
1945
+ this.state = State.COMMENT_END_DASH;
1946
+ break;
1947
+ }
1948
+ case $.LESS_THAN_SIGN: {
1949
+ token.data += '<';
1950
+ this.state = State.COMMENT_LESS_THAN_SIGN;
1951
+ break;
1952
+ }
1953
+ case $.NULL: {
1954
+ this._err(ERR.unexpectedNullCharacter);
1955
+ token.data += REPLACEMENT_CHARACTER;
1956
+ break;
1957
+ }
1958
+ case $.EOF: {
1959
+ this._err(ERR.eofInComment);
1960
+ this.emitCurrentComment(token);
1961
+ this._emitEOFToken();
1962
+ break;
1963
+ }
1964
+ default: {
1965
+ token.data += String.fromCodePoint(cp);
1966
+ }
1967
+ }
1968
+ }
1969
+ // Comment less-than sign state
1970
+ //------------------------------------------------------------------
1971
+ _stateCommentLessThanSign(cp) {
1972
+ const token = this.currentToken;
1973
+ switch (cp) {
1974
+ case $.EXCLAMATION_MARK: {
1975
+ token.data += '!';
1976
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG;
1977
+ break;
1978
+ }
1979
+ case $.LESS_THAN_SIGN: {
1980
+ token.data += '<';
1981
+ break;
1982
+ }
1983
+ default: {
1984
+ this.state = State.COMMENT;
1985
+ this._stateComment(cp);
1986
+ }
1987
+ }
1988
+ }
1989
+ // Comment less-than sign bang state
1990
+ //------------------------------------------------------------------
1991
+ _stateCommentLessThanSignBang(cp) {
1992
+ if (cp === $.HYPHEN_MINUS) {
1993
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH;
1994
+ }
1995
+ else {
1996
+ this.state = State.COMMENT;
1997
+ this._stateComment(cp);
1998
+ }
1999
+ }
2000
+ // Comment less-than sign bang dash state
2001
+ //------------------------------------------------------------------
2002
+ _stateCommentLessThanSignBangDash(cp) {
2003
+ if (cp === $.HYPHEN_MINUS) {
2004
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH;
2005
+ }
2006
+ else {
2007
+ this.state = State.COMMENT_END_DASH;
2008
+ this._stateCommentEndDash(cp);
2009
+ }
2010
+ }
2011
+ // Comment less-than sign bang dash dash state
2012
+ //------------------------------------------------------------------
2013
+ _stateCommentLessThanSignBangDashDash(cp) {
2014
+ if (cp !== $.GREATER_THAN_SIGN && cp !== $.EOF) {
2015
+ this._err(ERR.nestedComment);
2016
+ }
2017
+ this.state = State.COMMENT_END;
2018
+ this._stateCommentEnd(cp);
2019
+ }
2020
+ // Comment end dash state
2021
+ //------------------------------------------------------------------
2022
+ _stateCommentEndDash(cp) {
2023
+ const token = this.currentToken;
2024
+ switch (cp) {
2025
+ case $.HYPHEN_MINUS: {
2026
+ this.state = State.COMMENT_END;
2027
+ break;
2028
+ }
2029
+ case $.EOF: {
2030
+ this._err(ERR.eofInComment);
2031
+ this.emitCurrentComment(token);
2032
+ this._emitEOFToken();
2033
+ break;
2034
+ }
2035
+ default: {
2036
+ token.data += '-';
2037
+ this.state = State.COMMENT;
2038
+ this._stateComment(cp);
2039
+ }
2040
+ }
2041
+ }
2042
+ // Comment end state
2043
+ //------------------------------------------------------------------
2044
+ _stateCommentEnd(cp) {
2045
+ const token = this.currentToken;
2046
+ switch (cp) {
2047
+ case $.GREATER_THAN_SIGN: {
2048
+ this.state = State.DATA;
2049
+ this.emitCurrentComment(token);
2050
+ break;
2051
+ }
2052
+ case $.EXCLAMATION_MARK: {
2053
+ this.state = State.COMMENT_END_BANG;
2054
+ break;
2055
+ }
2056
+ case $.HYPHEN_MINUS: {
2057
+ token.data += '-';
2058
+ break;
2059
+ }
2060
+ case $.EOF: {
2061
+ this._err(ERR.eofInComment);
2062
+ this.emitCurrentComment(token);
2063
+ this._emitEOFToken();
2064
+ break;
2065
+ }
2066
+ default: {
2067
+ token.data += '--';
2068
+ this.state = State.COMMENT;
2069
+ this._stateComment(cp);
2070
+ }
2071
+ }
2072
+ }
2073
+ // Comment end bang state
2074
+ //------------------------------------------------------------------
2075
+ _stateCommentEndBang(cp) {
2076
+ const token = this.currentToken;
2077
+ switch (cp) {
2078
+ case $.HYPHEN_MINUS: {
2079
+ token.data += '--!';
2080
+ this.state = State.COMMENT_END_DASH;
2081
+ break;
2082
+ }
2083
+ case $.GREATER_THAN_SIGN: {
2084
+ this._err(ERR.incorrectlyClosedComment);
2085
+ this.state = State.DATA;
2086
+ this.emitCurrentComment(token);
2087
+ break;
2088
+ }
2089
+ case $.EOF: {
2090
+ this._err(ERR.eofInComment);
2091
+ this.emitCurrentComment(token);
2092
+ this._emitEOFToken();
2093
+ break;
2094
+ }
2095
+ default: {
2096
+ token.data += '--!';
2097
+ this.state = State.COMMENT;
2098
+ this._stateComment(cp);
2099
+ }
2100
+ }
2101
+ }
2102
+ // DOCTYPE state
2103
+ //------------------------------------------------------------------
2104
+ _stateDoctype(cp) {
2105
+ switch (cp) {
2106
+ case $.SPACE:
2107
+ case $.LINE_FEED:
2108
+ case $.TABULATION:
2109
+ case $.FORM_FEED: {
2110
+ this.state = State.BEFORE_DOCTYPE_NAME;
2111
+ break;
2112
+ }
2113
+ case $.GREATER_THAN_SIGN: {
2114
+ this.state = State.BEFORE_DOCTYPE_NAME;
2115
+ this._stateBeforeDoctypeName(cp);
2116
+ break;
2117
+ }
2118
+ case $.EOF: {
2119
+ this._err(ERR.eofInDoctype);
2120
+ this._createDoctypeToken(null);
2121
+ const token = this.currentToken;
2122
+ token.forceQuirks = true;
2123
+ this.emitCurrentDoctype(token);
2124
+ this._emitEOFToken();
2125
+ break;
2126
+ }
2127
+ default: {
2128
+ this._err(ERR.missingWhitespaceBeforeDoctypeName);
2129
+ this.state = State.BEFORE_DOCTYPE_NAME;
2130
+ this._stateBeforeDoctypeName(cp);
2131
+ }
2132
+ }
2133
+ }
2134
+ // Before DOCTYPE name state
2135
+ //------------------------------------------------------------------
2136
+ _stateBeforeDoctypeName(cp) {
2137
+ if (isAsciiUpper(cp)) {
2138
+ this._createDoctypeToken(String.fromCharCode(toAsciiLower(cp)));
2139
+ this.state = State.DOCTYPE_NAME;
2140
+ }
2141
+ else
2142
+ switch (cp) {
2143
+ case $.SPACE:
2144
+ case $.LINE_FEED:
2145
+ case $.TABULATION:
2146
+ case $.FORM_FEED: {
2147
+ // Ignore whitespace
2148
+ break;
2149
+ }
2150
+ case $.NULL: {
2151
+ this._err(ERR.unexpectedNullCharacter);
2152
+ this._createDoctypeToken(REPLACEMENT_CHARACTER);
2153
+ this.state = State.DOCTYPE_NAME;
2154
+ break;
2155
+ }
2156
+ case $.GREATER_THAN_SIGN: {
2157
+ this._err(ERR.missingDoctypeName);
2158
+ this._createDoctypeToken(null);
2159
+ const token = this.currentToken;
2160
+ token.forceQuirks = true;
2161
+ this.emitCurrentDoctype(token);
2162
+ this.state = State.DATA;
2163
+ break;
2164
+ }
2165
+ case $.EOF: {
2166
+ this._err(ERR.eofInDoctype);
2167
+ this._createDoctypeToken(null);
2168
+ const token = this.currentToken;
2169
+ token.forceQuirks = true;
2170
+ this.emitCurrentDoctype(token);
2171
+ this._emitEOFToken();
2172
+ break;
2173
+ }
2174
+ default: {
2175
+ this._createDoctypeToken(String.fromCodePoint(cp));
2176
+ this.state = State.DOCTYPE_NAME;
2177
+ }
2178
+ }
2179
+ }
2180
+ // DOCTYPE name state
2181
+ //------------------------------------------------------------------
2182
+ _stateDoctypeName(cp) {
2183
+ const token = this.currentToken;
2184
+ switch (cp) {
2185
+ case $.SPACE:
2186
+ case $.LINE_FEED:
2187
+ case $.TABULATION:
2188
+ case $.FORM_FEED: {
2189
+ this.state = State.AFTER_DOCTYPE_NAME;
2190
+ break;
2191
+ }
2192
+ case $.GREATER_THAN_SIGN: {
2193
+ this.state = State.DATA;
2194
+ this.emitCurrentDoctype(token);
2195
+ break;
2196
+ }
2197
+ case $.NULL: {
2198
+ this._err(ERR.unexpectedNullCharacter);
2199
+ token.name += REPLACEMENT_CHARACTER;
2200
+ break;
2201
+ }
2202
+ case $.EOF: {
2203
+ this._err(ERR.eofInDoctype);
2204
+ token.forceQuirks = true;
2205
+ this.emitCurrentDoctype(token);
2206
+ this._emitEOFToken();
2207
+ break;
2208
+ }
2209
+ default: {
2210
+ token.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
2211
+ }
2212
+ }
2213
+ }
2214
+ // After DOCTYPE name state
2215
+ //------------------------------------------------------------------
2216
+ _stateAfterDoctypeName(cp) {
2217
+ const token = this.currentToken;
2218
+ switch (cp) {
2219
+ case $.SPACE:
2220
+ case $.LINE_FEED:
2221
+ case $.TABULATION:
2222
+ case $.FORM_FEED: {
2223
+ // Ignore whitespace
2224
+ break;
2225
+ }
2226
+ case $.GREATER_THAN_SIGN: {
2227
+ this.state = State.DATA;
2228
+ this.emitCurrentDoctype(token);
2229
+ break;
2230
+ }
2231
+ case $.EOF: {
2232
+ this._err(ERR.eofInDoctype);
2233
+ token.forceQuirks = true;
2234
+ this.emitCurrentDoctype(token);
2235
+ this._emitEOFToken();
2236
+ break;
2237
+ }
2238
+ default:
2239
+ if (this._consumeSequenceIfMatch($$.PUBLIC, false)) {
2240
+ this.state = State.AFTER_DOCTYPE_PUBLIC_KEYWORD;
2241
+ }
2242
+ else if (this._consumeSequenceIfMatch($$.SYSTEM, false)) {
2243
+ this.state = State.AFTER_DOCTYPE_SYSTEM_KEYWORD;
2244
+ }
2245
+ //NOTE: sequence lookup can be abrupted by hibernation. In that case lookup
2246
+ //results are no longer valid and we will need to start over.
2247
+ else if (!this._ensureHibernation()) {
2248
+ this._err(ERR.invalidCharacterSequenceAfterDoctypeName);
2249
+ token.forceQuirks = true;
2250
+ this.state = State.BOGUS_DOCTYPE;
2251
+ this._stateBogusDoctype(cp);
2252
+ }
2253
+ }
2254
+ }
2255
+ // After DOCTYPE public keyword state
2256
+ //------------------------------------------------------------------
2257
+ _stateAfterDoctypePublicKeyword(cp) {
2258
+ const token = this.currentToken;
2259
+ switch (cp) {
2260
+ case $.SPACE:
2261
+ case $.LINE_FEED:
2262
+ case $.TABULATION:
2263
+ case $.FORM_FEED: {
2264
+ this.state = State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER;
2265
+ break;
2266
+ }
2267
+ case $.QUOTATION_MARK: {
2268
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2269
+ token.publicId = '';
2270
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2271
+ break;
2272
+ }
2273
+ case $.APOSTROPHE: {
2274
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2275
+ token.publicId = '';
2276
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2277
+ break;
2278
+ }
2279
+ case $.GREATER_THAN_SIGN: {
2280
+ this._err(ERR.missingDoctypePublicIdentifier);
2281
+ token.forceQuirks = true;
2282
+ this.state = State.DATA;
2283
+ this.emitCurrentDoctype(token);
2284
+ break;
2285
+ }
2286
+ case $.EOF: {
2287
+ this._err(ERR.eofInDoctype);
2288
+ token.forceQuirks = true;
2289
+ this.emitCurrentDoctype(token);
2290
+ this._emitEOFToken();
2291
+ break;
2292
+ }
2293
+ default: {
2294
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2295
+ token.forceQuirks = true;
2296
+ this.state = State.BOGUS_DOCTYPE;
2297
+ this._stateBogusDoctype(cp);
2298
+ }
2299
+ }
2300
+ }
2301
+ // Before DOCTYPE public identifier state
2302
+ //------------------------------------------------------------------
2303
+ _stateBeforeDoctypePublicIdentifier(cp) {
2304
+ const token = this.currentToken;
2305
+ switch (cp) {
2306
+ case $.SPACE:
2307
+ case $.LINE_FEED:
2308
+ case $.TABULATION:
2309
+ case $.FORM_FEED: {
2310
+ // Ignore whitespace
2311
+ break;
2312
+ }
2313
+ case $.QUOTATION_MARK: {
2314
+ token.publicId = '';
2315
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2316
+ break;
2317
+ }
2318
+ case $.APOSTROPHE: {
2319
+ token.publicId = '';
2320
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2321
+ break;
2322
+ }
2323
+ case $.GREATER_THAN_SIGN: {
2324
+ this._err(ERR.missingDoctypePublicIdentifier);
2325
+ token.forceQuirks = true;
2326
+ this.state = State.DATA;
2327
+ this.emitCurrentDoctype(token);
2328
+ break;
2329
+ }
2330
+ case $.EOF: {
2331
+ this._err(ERR.eofInDoctype);
2332
+ token.forceQuirks = true;
2333
+ this.emitCurrentDoctype(token);
2334
+ this._emitEOFToken();
2335
+ break;
2336
+ }
2337
+ default: {
2338
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2339
+ token.forceQuirks = true;
2340
+ this.state = State.BOGUS_DOCTYPE;
2341
+ this._stateBogusDoctype(cp);
2342
+ }
2343
+ }
2344
+ }
2345
+ // DOCTYPE public identifier (double-quoted) state
2346
+ //------------------------------------------------------------------
2347
+ _stateDoctypePublicIdentifierDoubleQuoted(cp) {
2348
+ const token = this.currentToken;
2349
+ switch (cp) {
2350
+ case $.QUOTATION_MARK: {
2351
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2352
+ break;
2353
+ }
2354
+ case $.NULL: {
2355
+ this._err(ERR.unexpectedNullCharacter);
2356
+ token.publicId += REPLACEMENT_CHARACTER;
2357
+ break;
2358
+ }
2359
+ case $.GREATER_THAN_SIGN: {
2360
+ this._err(ERR.abruptDoctypePublicIdentifier);
2361
+ token.forceQuirks = true;
2362
+ this.emitCurrentDoctype(token);
2363
+ this.state = State.DATA;
2364
+ break;
2365
+ }
2366
+ case $.EOF: {
2367
+ this._err(ERR.eofInDoctype);
2368
+ token.forceQuirks = true;
2369
+ this.emitCurrentDoctype(token);
2370
+ this._emitEOFToken();
2371
+ break;
2372
+ }
2373
+ default: {
2374
+ token.publicId += String.fromCodePoint(cp);
2375
+ }
2376
+ }
2377
+ }
2378
+ // DOCTYPE public identifier (single-quoted) state
2379
+ //------------------------------------------------------------------
2380
+ _stateDoctypePublicIdentifierSingleQuoted(cp) {
2381
+ const token = this.currentToken;
2382
+ switch (cp) {
2383
+ case $.APOSTROPHE: {
2384
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2385
+ break;
2386
+ }
2387
+ case $.NULL: {
2388
+ this._err(ERR.unexpectedNullCharacter);
2389
+ token.publicId += REPLACEMENT_CHARACTER;
2390
+ break;
2391
+ }
2392
+ case $.GREATER_THAN_SIGN: {
2393
+ this._err(ERR.abruptDoctypePublicIdentifier);
2394
+ token.forceQuirks = true;
2395
+ this.emitCurrentDoctype(token);
2396
+ this.state = State.DATA;
2397
+ break;
2398
+ }
2399
+ case $.EOF: {
2400
+ this._err(ERR.eofInDoctype);
2401
+ token.forceQuirks = true;
2402
+ this.emitCurrentDoctype(token);
2403
+ this._emitEOFToken();
2404
+ break;
2405
+ }
2406
+ default: {
2407
+ token.publicId += String.fromCodePoint(cp);
2408
+ }
2409
+ }
2410
+ }
2411
+ // After DOCTYPE public identifier state
2412
+ //------------------------------------------------------------------
2413
+ _stateAfterDoctypePublicIdentifier(cp) {
2414
+ const token = this.currentToken;
2415
+ switch (cp) {
2416
+ case $.SPACE:
2417
+ case $.LINE_FEED:
2418
+ case $.TABULATION:
2419
+ case $.FORM_FEED: {
2420
+ this.state = State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS;
2421
+ break;
2422
+ }
2423
+ case $.GREATER_THAN_SIGN: {
2424
+ this.state = State.DATA;
2425
+ this.emitCurrentDoctype(token);
2426
+ break;
2427
+ }
2428
+ case $.QUOTATION_MARK: {
2429
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2430
+ token.systemId = '';
2431
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2432
+ break;
2433
+ }
2434
+ case $.APOSTROPHE: {
2435
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2436
+ token.systemId = '';
2437
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2438
+ break;
2439
+ }
2440
+ case $.EOF: {
2441
+ this._err(ERR.eofInDoctype);
2442
+ token.forceQuirks = true;
2443
+ this.emitCurrentDoctype(token);
2444
+ this._emitEOFToken();
2445
+ break;
2446
+ }
2447
+ default: {
2448
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2449
+ token.forceQuirks = true;
2450
+ this.state = State.BOGUS_DOCTYPE;
2451
+ this._stateBogusDoctype(cp);
2452
+ }
2453
+ }
2454
+ }
2455
+ // Between DOCTYPE public and system identifiers state
2456
+ //------------------------------------------------------------------
2457
+ _stateBetweenDoctypePublicAndSystemIdentifiers(cp) {
2458
+ const token = this.currentToken;
2459
+ switch (cp) {
2460
+ case $.SPACE:
2461
+ case $.LINE_FEED:
2462
+ case $.TABULATION:
2463
+ case $.FORM_FEED: {
2464
+ // Ignore whitespace
2465
+ break;
2466
+ }
2467
+ case $.GREATER_THAN_SIGN: {
2468
+ this.emitCurrentDoctype(token);
2469
+ this.state = State.DATA;
2470
+ break;
2471
+ }
2472
+ case $.QUOTATION_MARK: {
2473
+ token.systemId = '';
2474
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2475
+ break;
2476
+ }
2477
+ case $.APOSTROPHE: {
2478
+ token.systemId = '';
2479
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2480
+ break;
2481
+ }
2482
+ case $.EOF: {
2483
+ this._err(ERR.eofInDoctype);
2484
+ token.forceQuirks = true;
2485
+ this.emitCurrentDoctype(token);
2486
+ this._emitEOFToken();
2487
+ break;
2488
+ }
2489
+ default: {
2490
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2491
+ token.forceQuirks = true;
2492
+ this.state = State.BOGUS_DOCTYPE;
2493
+ this._stateBogusDoctype(cp);
2494
+ }
2495
+ }
2496
+ }
2497
+ // After DOCTYPE system keyword state
2498
+ //------------------------------------------------------------------
2499
+ _stateAfterDoctypeSystemKeyword(cp) {
2500
+ const token = this.currentToken;
2501
+ switch (cp) {
2502
+ case $.SPACE:
2503
+ case $.LINE_FEED:
2504
+ case $.TABULATION:
2505
+ case $.FORM_FEED: {
2506
+ this.state = State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER;
2507
+ break;
2508
+ }
2509
+ case $.QUOTATION_MARK: {
2510
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2511
+ token.systemId = '';
2512
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2513
+ break;
2514
+ }
2515
+ case $.APOSTROPHE: {
2516
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2517
+ token.systemId = '';
2518
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2519
+ break;
2520
+ }
2521
+ case $.GREATER_THAN_SIGN: {
2522
+ this._err(ERR.missingDoctypeSystemIdentifier);
2523
+ token.forceQuirks = true;
2524
+ this.state = State.DATA;
2525
+ this.emitCurrentDoctype(token);
2526
+ break;
2527
+ }
2528
+ case $.EOF: {
2529
+ this._err(ERR.eofInDoctype);
2530
+ token.forceQuirks = true;
2531
+ this.emitCurrentDoctype(token);
2532
+ this._emitEOFToken();
2533
+ break;
2534
+ }
2535
+ default: {
2536
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2537
+ token.forceQuirks = true;
2538
+ this.state = State.BOGUS_DOCTYPE;
2539
+ this._stateBogusDoctype(cp);
2540
+ }
2541
+ }
2542
+ }
2543
+ // Before DOCTYPE system identifier state
2544
+ //------------------------------------------------------------------
2545
+ _stateBeforeDoctypeSystemIdentifier(cp) {
2546
+ const token = this.currentToken;
2547
+ switch (cp) {
2548
+ case $.SPACE:
2549
+ case $.LINE_FEED:
2550
+ case $.TABULATION:
2551
+ case $.FORM_FEED: {
2552
+ // Ignore whitespace
2553
+ break;
2554
+ }
2555
+ case $.QUOTATION_MARK: {
2556
+ token.systemId = '';
2557
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2558
+ break;
2559
+ }
2560
+ case $.APOSTROPHE: {
2561
+ token.systemId = '';
2562
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2563
+ break;
2564
+ }
2565
+ case $.GREATER_THAN_SIGN: {
2566
+ this._err(ERR.missingDoctypeSystemIdentifier);
2567
+ token.forceQuirks = true;
2568
+ this.state = State.DATA;
2569
+ this.emitCurrentDoctype(token);
2570
+ break;
2571
+ }
2572
+ case $.EOF: {
2573
+ this._err(ERR.eofInDoctype);
2574
+ token.forceQuirks = true;
2575
+ this.emitCurrentDoctype(token);
2576
+ this._emitEOFToken();
2577
+ break;
2578
+ }
2579
+ default: {
2580
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2581
+ token.forceQuirks = true;
2582
+ this.state = State.BOGUS_DOCTYPE;
2583
+ this._stateBogusDoctype(cp);
2584
+ }
2585
+ }
2586
+ }
2587
+ // DOCTYPE system identifier (double-quoted) state
2588
+ //------------------------------------------------------------------
2589
+ _stateDoctypeSystemIdentifierDoubleQuoted(cp) {
2590
+ const token = this.currentToken;
2591
+ switch (cp) {
2592
+ case $.QUOTATION_MARK: {
2593
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2594
+ break;
2595
+ }
2596
+ case $.NULL: {
2597
+ this._err(ERR.unexpectedNullCharacter);
2598
+ token.systemId += REPLACEMENT_CHARACTER;
2599
+ break;
2600
+ }
2601
+ case $.GREATER_THAN_SIGN: {
2602
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2603
+ token.forceQuirks = true;
2604
+ this.emitCurrentDoctype(token);
2605
+ this.state = State.DATA;
2606
+ break;
2607
+ }
2608
+ case $.EOF: {
2609
+ this._err(ERR.eofInDoctype);
2610
+ token.forceQuirks = true;
2611
+ this.emitCurrentDoctype(token);
2612
+ this._emitEOFToken();
2613
+ break;
2614
+ }
2615
+ default: {
2616
+ token.systemId += String.fromCodePoint(cp);
2617
+ }
2618
+ }
2619
+ }
2620
+ // DOCTYPE system identifier (single-quoted) state
2621
+ //------------------------------------------------------------------
2622
+ _stateDoctypeSystemIdentifierSingleQuoted(cp) {
2623
+ const token = this.currentToken;
2624
+ switch (cp) {
2625
+ case $.APOSTROPHE: {
2626
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2627
+ break;
2628
+ }
2629
+ case $.NULL: {
2630
+ this._err(ERR.unexpectedNullCharacter);
2631
+ token.systemId += REPLACEMENT_CHARACTER;
2632
+ break;
2633
+ }
2634
+ case $.GREATER_THAN_SIGN: {
2635
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2636
+ token.forceQuirks = true;
2637
+ this.emitCurrentDoctype(token);
2638
+ this.state = State.DATA;
2639
+ break;
2640
+ }
2641
+ case $.EOF: {
2642
+ this._err(ERR.eofInDoctype);
2643
+ token.forceQuirks = true;
2644
+ this.emitCurrentDoctype(token);
2645
+ this._emitEOFToken();
2646
+ break;
2647
+ }
2648
+ default: {
2649
+ token.systemId += String.fromCodePoint(cp);
2650
+ }
2651
+ }
2652
+ }
2653
+ // After DOCTYPE system identifier state
2654
+ //------------------------------------------------------------------
2655
+ _stateAfterDoctypeSystemIdentifier(cp) {
2656
+ const token = this.currentToken;
2657
+ switch (cp) {
2658
+ case $.SPACE:
2659
+ case $.LINE_FEED:
2660
+ case $.TABULATION:
2661
+ case $.FORM_FEED: {
2662
+ // Ignore whitespace
2663
+ break;
2664
+ }
2665
+ case $.GREATER_THAN_SIGN: {
2666
+ this.emitCurrentDoctype(token);
2667
+ this.state = State.DATA;
2668
+ break;
2669
+ }
2670
+ case $.EOF: {
2671
+ this._err(ERR.eofInDoctype);
2672
+ token.forceQuirks = true;
2673
+ this.emitCurrentDoctype(token);
2674
+ this._emitEOFToken();
2675
+ break;
2676
+ }
2677
+ default: {
2678
+ this._err(ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
2679
+ this.state = State.BOGUS_DOCTYPE;
2680
+ this._stateBogusDoctype(cp);
2681
+ }
2682
+ }
2683
+ }
2684
+ // Bogus DOCTYPE state
2685
+ //------------------------------------------------------------------
2686
+ _stateBogusDoctype(cp) {
2687
+ const token = this.currentToken;
2688
+ switch (cp) {
2689
+ case $.GREATER_THAN_SIGN: {
2690
+ this.emitCurrentDoctype(token);
2691
+ this.state = State.DATA;
2692
+ break;
2693
+ }
2694
+ case $.NULL: {
2695
+ this._err(ERR.unexpectedNullCharacter);
2696
+ break;
2697
+ }
2698
+ case $.EOF: {
2699
+ this.emitCurrentDoctype(token);
2700
+ this._emitEOFToken();
2701
+ break;
2702
+ }
2703
+ default:
2704
+ // Do nothing
2705
+ }
2706
+ }
2707
+ // CDATA section state
2708
+ //------------------------------------------------------------------
2709
+ _stateCdataSection(cp) {
2710
+ switch (cp) {
2711
+ case $.RIGHT_SQUARE_BRACKET: {
2712
+ this.state = State.CDATA_SECTION_BRACKET;
2713
+ break;
2714
+ }
2715
+ case $.EOF: {
2716
+ this._err(ERR.eofInCdata);
2717
+ this._emitEOFToken();
2718
+ break;
2719
+ }
2720
+ default: {
2721
+ this._emitCodePoint(cp);
2722
+ }
2723
+ }
2724
+ }
2725
+ // CDATA section bracket state
2726
+ //------------------------------------------------------------------
2727
+ _stateCdataSectionBracket(cp) {
2728
+ if (cp === $.RIGHT_SQUARE_BRACKET) {
2729
+ this.state = State.CDATA_SECTION_END;
2730
+ }
2731
+ else {
2732
+ this._emitChars(']');
2733
+ this.state = State.CDATA_SECTION;
2734
+ this._stateCdataSection(cp);
2735
+ }
2736
+ }
2737
+ // CDATA section end state
2738
+ //------------------------------------------------------------------
2739
+ _stateCdataSectionEnd(cp) {
2740
+ switch (cp) {
2741
+ case $.GREATER_THAN_SIGN: {
2742
+ this.state = State.DATA;
2743
+ break;
2744
+ }
2745
+ case $.RIGHT_SQUARE_BRACKET: {
2746
+ this._emitChars(']');
2747
+ break;
2748
+ }
2749
+ default: {
2750
+ this._emitChars(']]');
2751
+ this.state = State.CDATA_SECTION;
2752
+ this._stateCdataSection(cp);
2753
+ }
2754
+ }
2755
+ }
2756
+ // Character reference state
2757
+ //------------------------------------------------------------------
2758
+ _stateCharacterReference(cp) {
2759
+ if (cp === $.NUMBER_SIGN) {
2760
+ this.state = State.NUMERIC_CHARACTER_REFERENCE;
2761
+ }
2762
+ else if (isAsciiAlphaNumeric(cp)) {
2763
+ this.state = State.NAMED_CHARACTER_REFERENCE;
2764
+ this._stateNamedCharacterReference(cp);
2765
+ }
2766
+ else {
2767
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2768
+ this._reconsumeInState(this.returnState, cp);
2769
+ }
2770
+ }
2771
+ // Named character reference state
2772
+ //------------------------------------------------------------------
2773
+ _stateNamedCharacterReference(cp) {
2774
+ const matchResult = this._matchNamedCharacterReference(cp);
2775
+ //NOTE: Matching can be abrupted by hibernation. In that case, match
2776
+ //results are no longer valid and we will need to start over.
2777
+ if (this._ensureHibernation()) {
2778
+ // Stay in the state, try again.
2779
+ }
2780
+ else if (matchResult) {
2781
+ for (let i = 0; i < matchResult.length; i++) {
2782
+ this._flushCodePointConsumedAsCharacterReference(matchResult[i]);
2783
+ }
2784
+ this.state = this.returnState;
2785
+ }
2786
+ else {
2787
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2788
+ this.state = State.AMBIGUOUS_AMPERSAND;
2789
+ }
2790
+ }
2791
+ // Ambiguos ampersand state
2792
+ //------------------------------------------------------------------
2793
+ _stateAmbiguousAmpersand(cp) {
2794
+ if (isAsciiAlphaNumeric(cp)) {
2795
+ this._flushCodePointConsumedAsCharacterReference(cp);
2796
+ }
2797
+ else {
2798
+ if (cp === $.SEMICOLON) {
2799
+ this._err(ERR.unknownNamedCharacterReference);
2800
+ }
2801
+ this._reconsumeInState(this.returnState, cp);
2802
+ }
2803
+ }
2804
+ // Numeric character reference state
2805
+ //------------------------------------------------------------------
2806
+ _stateNumericCharacterReference(cp) {
2807
+ this.charRefCode = 0;
2808
+ if (cp === $.LATIN_SMALL_X || cp === $.LATIN_CAPITAL_X) {
2809
+ this.state = State.HEXADEMICAL_CHARACTER_REFERENCE_START;
2810
+ }
2811
+ // Inlined decimal character reference start state
2812
+ else if (isAsciiDigit(cp)) {
2813
+ this.state = State.DECIMAL_CHARACTER_REFERENCE;
2814
+ this._stateDecimalCharacterReference(cp);
2815
+ }
2816
+ else {
2817
+ this._err(ERR.absenceOfDigitsInNumericCharacterReference);
2818
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2819
+ this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
2820
+ this._reconsumeInState(this.returnState, cp);
2821
+ }
2822
+ }
2823
+ // Hexademical character reference start state
2824
+ //------------------------------------------------------------------
2825
+ _stateHexademicalCharacterReferenceStart(cp) {
2826
+ if (isAsciiHexDigit(cp)) {
2827
+ this.state = State.HEXADEMICAL_CHARACTER_REFERENCE;
2828
+ this._stateHexademicalCharacterReference(cp);
2829
+ }
2830
+ else {
2831
+ this._err(ERR.absenceOfDigitsInNumericCharacterReference);
2832
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2833
+ this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
2834
+ this._unconsume(2);
2835
+ this.state = this.returnState;
2836
+ }
2837
+ }
2838
+ // Hexademical character reference state
2839
+ //------------------------------------------------------------------
2840
+ _stateHexademicalCharacterReference(cp) {
2841
+ if (isAsciiUpperHexDigit(cp)) {
2842
+ this.charRefCode = this.charRefCode * 16 + cp - 0x37;
2843
+ }
2844
+ else if (isAsciiLowerHexDigit(cp)) {
2845
+ this.charRefCode = this.charRefCode * 16 + cp - 0x57;
2846
+ }
2847
+ else if (isAsciiDigit(cp)) {
2848
+ this.charRefCode = this.charRefCode * 16 + cp - 0x30;
2849
+ }
2850
+ else if (cp === $.SEMICOLON) {
2851
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2852
+ }
2853
+ else {
2854
+ this._err(ERR.missingSemicolonAfterCharacterReference);
2855
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2856
+ this._stateNumericCharacterReferenceEnd(cp);
2857
+ }
2858
+ }
2859
+ // Decimal character reference state
2860
+ //------------------------------------------------------------------
2861
+ _stateDecimalCharacterReference(cp) {
2862
+ if (isAsciiDigit(cp)) {
2863
+ this.charRefCode = this.charRefCode * 10 + cp - 0x30;
2864
+ }
2865
+ else if (cp === $.SEMICOLON) {
2866
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2867
+ }
2868
+ else {
2869
+ this._err(ERR.missingSemicolonAfterCharacterReference);
2870
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2871
+ this._stateNumericCharacterReferenceEnd(cp);
2872
+ }
2873
+ }
2874
+ // Numeric character reference end state
2875
+ //------------------------------------------------------------------
2876
+ _stateNumericCharacterReferenceEnd(cp) {
2877
+ if (this.charRefCode === $.NULL) {
2878
+ this._err(ERR.nullCharacterReference);
2879
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2880
+ }
2881
+ else if (this.charRefCode > 1114111) {
2882
+ this._err(ERR.characterReferenceOutsideUnicodeRange);
2883
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2884
+ }
2885
+ else if (isSurrogate(this.charRefCode)) {
2886
+ this._err(ERR.surrogateCharacterReference);
2887
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2888
+ }
2889
+ else if (isUndefinedCodePoint(this.charRefCode)) {
2890
+ this._err(ERR.noncharacterCharacterReference);
2891
+ }
2892
+ else if (isControlCodePoint(this.charRefCode) || this.charRefCode === $.CARRIAGE_RETURN) {
2893
+ this._err(ERR.controlCharacterReference);
2894
+ const replacement = C1_CONTROLS_REFERENCE_REPLACEMENTS.get(this.charRefCode);
2895
+ if (replacement !== undefined) {
2896
+ this.charRefCode = replacement;
2897
+ }
2898
+ }
2899
+ this._flushCodePointConsumedAsCharacterReference(this.charRefCode);
2900
+ this._reconsumeInState(this.returnState, cp);
2901
+ }
2902
+ }
2903
+ //# sourceMappingURL=index.js.map