@cjser/parse5 0.0.0-stage → 8.0.1-cjser.2

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.
@@ -0,0 +1,2712 @@
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, EntityDecoder, DecodingMode } from '@cjser/entities/decode';
5
+ import { ERR } from '../common/error-codes.js';
6
+ import { TAG_ID, getTagID } from '../common/html.js';
7
+ //States
8
+ var State;
9
+ (function (State) {
10
+ State[State["DATA"] = 0] = "DATA";
11
+ State[State["RCDATA"] = 1] = "RCDATA";
12
+ State[State["RAWTEXT"] = 2] = "RAWTEXT";
13
+ State[State["SCRIPT_DATA"] = 3] = "SCRIPT_DATA";
14
+ State[State["PLAINTEXT"] = 4] = "PLAINTEXT";
15
+ State[State["TAG_OPEN"] = 5] = "TAG_OPEN";
16
+ State[State["END_TAG_OPEN"] = 6] = "END_TAG_OPEN";
17
+ State[State["TAG_NAME"] = 7] = "TAG_NAME";
18
+ State[State["RCDATA_LESS_THAN_SIGN"] = 8] = "RCDATA_LESS_THAN_SIGN";
19
+ State[State["RCDATA_END_TAG_OPEN"] = 9] = "RCDATA_END_TAG_OPEN";
20
+ State[State["RCDATA_END_TAG_NAME"] = 10] = "RCDATA_END_TAG_NAME";
21
+ State[State["RAWTEXT_LESS_THAN_SIGN"] = 11] = "RAWTEXT_LESS_THAN_SIGN";
22
+ State[State["RAWTEXT_END_TAG_OPEN"] = 12] = "RAWTEXT_END_TAG_OPEN";
23
+ State[State["RAWTEXT_END_TAG_NAME"] = 13] = "RAWTEXT_END_TAG_NAME";
24
+ State[State["SCRIPT_DATA_LESS_THAN_SIGN"] = 14] = "SCRIPT_DATA_LESS_THAN_SIGN";
25
+ State[State["SCRIPT_DATA_END_TAG_OPEN"] = 15] = "SCRIPT_DATA_END_TAG_OPEN";
26
+ State[State["SCRIPT_DATA_END_TAG_NAME"] = 16] = "SCRIPT_DATA_END_TAG_NAME";
27
+ State[State["SCRIPT_DATA_ESCAPE_START"] = 17] = "SCRIPT_DATA_ESCAPE_START";
28
+ State[State["SCRIPT_DATA_ESCAPE_START_DASH"] = 18] = "SCRIPT_DATA_ESCAPE_START_DASH";
29
+ State[State["SCRIPT_DATA_ESCAPED"] = 19] = "SCRIPT_DATA_ESCAPED";
30
+ State[State["SCRIPT_DATA_ESCAPED_DASH"] = 20] = "SCRIPT_DATA_ESCAPED_DASH";
31
+ State[State["SCRIPT_DATA_ESCAPED_DASH_DASH"] = 21] = "SCRIPT_DATA_ESCAPED_DASH_DASH";
32
+ State[State["SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN"] = 22] = "SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN";
33
+ State[State["SCRIPT_DATA_ESCAPED_END_TAG_OPEN"] = 23] = "SCRIPT_DATA_ESCAPED_END_TAG_OPEN";
34
+ State[State["SCRIPT_DATA_ESCAPED_END_TAG_NAME"] = 24] = "SCRIPT_DATA_ESCAPED_END_TAG_NAME";
35
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPE_START"] = 25] = "SCRIPT_DATA_DOUBLE_ESCAPE_START";
36
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED"] = 26] = "SCRIPT_DATA_DOUBLE_ESCAPED";
37
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH"] = 27] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH";
38
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH"] = 28] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH";
39
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN"] = 29] = "SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN";
40
+ State[State["SCRIPT_DATA_DOUBLE_ESCAPE_END"] = 30] = "SCRIPT_DATA_DOUBLE_ESCAPE_END";
41
+ State[State["BEFORE_ATTRIBUTE_NAME"] = 31] = "BEFORE_ATTRIBUTE_NAME";
42
+ State[State["ATTRIBUTE_NAME"] = 32] = "ATTRIBUTE_NAME";
43
+ State[State["AFTER_ATTRIBUTE_NAME"] = 33] = "AFTER_ATTRIBUTE_NAME";
44
+ State[State["BEFORE_ATTRIBUTE_VALUE"] = 34] = "BEFORE_ATTRIBUTE_VALUE";
45
+ State[State["ATTRIBUTE_VALUE_DOUBLE_QUOTED"] = 35] = "ATTRIBUTE_VALUE_DOUBLE_QUOTED";
46
+ State[State["ATTRIBUTE_VALUE_SINGLE_QUOTED"] = 36] = "ATTRIBUTE_VALUE_SINGLE_QUOTED";
47
+ State[State["ATTRIBUTE_VALUE_UNQUOTED"] = 37] = "ATTRIBUTE_VALUE_UNQUOTED";
48
+ State[State["AFTER_ATTRIBUTE_VALUE_QUOTED"] = 38] = "AFTER_ATTRIBUTE_VALUE_QUOTED";
49
+ State[State["SELF_CLOSING_START_TAG"] = 39] = "SELF_CLOSING_START_TAG";
50
+ State[State["BOGUS_COMMENT"] = 40] = "BOGUS_COMMENT";
51
+ State[State["MARKUP_DECLARATION_OPEN"] = 41] = "MARKUP_DECLARATION_OPEN";
52
+ State[State["COMMENT_START"] = 42] = "COMMENT_START";
53
+ State[State["COMMENT_START_DASH"] = 43] = "COMMENT_START_DASH";
54
+ State[State["COMMENT"] = 44] = "COMMENT";
55
+ State[State["COMMENT_LESS_THAN_SIGN"] = 45] = "COMMENT_LESS_THAN_SIGN";
56
+ State[State["COMMENT_LESS_THAN_SIGN_BANG"] = 46] = "COMMENT_LESS_THAN_SIGN_BANG";
57
+ State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH"] = 47] = "COMMENT_LESS_THAN_SIGN_BANG_DASH";
58
+ State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH"] = 48] = "COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH";
59
+ State[State["COMMENT_END_DASH"] = 49] = "COMMENT_END_DASH";
60
+ State[State["COMMENT_END"] = 50] = "COMMENT_END";
61
+ State[State["COMMENT_END_BANG"] = 51] = "COMMENT_END_BANG";
62
+ State[State["DOCTYPE"] = 52] = "DOCTYPE";
63
+ State[State["BEFORE_DOCTYPE_NAME"] = 53] = "BEFORE_DOCTYPE_NAME";
64
+ State[State["DOCTYPE_NAME"] = 54] = "DOCTYPE_NAME";
65
+ State[State["AFTER_DOCTYPE_NAME"] = 55] = "AFTER_DOCTYPE_NAME";
66
+ State[State["AFTER_DOCTYPE_PUBLIC_KEYWORD"] = 56] = "AFTER_DOCTYPE_PUBLIC_KEYWORD";
67
+ State[State["BEFORE_DOCTYPE_PUBLIC_IDENTIFIER"] = 57] = "BEFORE_DOCTYPE_PUBLIC_IDENTIFIER";
68
+ State[State["DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED"] = 58] = "DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED";
69
+ State[State["DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED"] = 59] = "DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED";
70
+ State[State["AFTER_DOCTYPE_PUBLIC_IDENTIFIER"] = 60] = "AFTER_DOCTYPE_PUBLIC_IDENTIFIER";
71
+ State[State["BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS"] = 61] = "BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS";
72
+ State[State["AFTER_DOCTYPE_SYSTEM_KEYWORD"] = 62] = "AFTER_DOCTYPE_SYSTEM_KEYWORD";
73
+ State[State["BEFORE_DOCTYPE_SYSTEM_IDENTIFIER"] = 63] = "BEFORE_DOCTYPE_SYSTEM_IDENTIFIER";
74
+ State[State["DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED"] = 64] = "DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED";
75
+ State[State["DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED"] = 65] = "DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED";
76
+ State[State["AFTER_DOCTYPE_SYSTEM_IDENTIFIER"] = 66] = "AFTER_DOCTYPE_SYSTEM_IDENTIFIER";
77
+ State[State["BOGUS_DOCTYPE"] = 67] = "BOGUS_DOCTYPE";
78
+ State[State["CDATA_SECTION"] = 68] = "CDATA_SECTION";
79
+ State[State["CDATA_SECTION_BRACKET"] = 69] = "CDATA_SECTION_BRACKET";
80
+ State[State["CDATA_SECTION_END"] = 70] = "CDATA_SECTION_END";
81
+ State[State["CHARACTER_REFERENCE"] = 71] = "CHARACTER_REFERENCE";
82
+ State[State["AMBIGUOUS_AMPERSAND"] = 72] = "AMBIGUOUS_AMPERSAND";
83
+ })(State || (State = {}));
84
+ //Tokenizer initial states for different modes
85
+ export const TokenizerMode = {
86
+ DATA: State.DATA,
87
+ RCDATA: State.RCDATA,
88
+ RAWTEXT: State.RAWTEXT,
89
+ SCRIPT_DATA: State.SCRIPT_DATA,
90
+ PLAINTEXT: State.PLAINTEXT,
91
+ CDATA_SECTION: State.CDATA_SECTION,
92
+ };
93
+ //Utils
94
+ //OPTIMIZATION: these utility functions should not be moved out of this module. V8 Crankshaft will not inline
95
+ //this functions if they will be situated in another module due to context switch.
96
+ //Always perform inlining check before modifying this functions ('node --trace-inlining').
97
+ function isAsciiDigit(cp) {
98
+ return cp >= $.DIGIT_0 && cp <= $.DIGIT_9;
99
+ }
100
+ function isAsciiUpper(cp) {
101
+ return cp >= $.LATIN_CAPITAL_A && cp <= $.LATIN_CAPITAL_Z;
102
+ }
103
+ function isAsciiLower(cp) {
104
+ return cp >= $.LATIN_SMALL_A && cp <= $.LATIN_SMALL_Z;
105
+ }
106
+ function isAsciiLetter(cp) {
107
+ return isAsciiLower(cp) || isAsciiUpper(cp);
108
+ }
109
+ function isAsciiAlphaNumeric(cp) {
110
+ return isAsciiLetter(cp) || isAsciiDigit(cp);
111
+ }
112
+ function toAsciiLower(cp) {
113
+ return cp + 32;
114
+ }
115
+ function isWhitespace(cp) {
116
+ return cp === $.SPACE || cp === $.LINE_FEED || cp === $.TABULATION || cp === $.FORM_FEED;
117
+ }
118
+ function isScriptDataDoubleEscapeSequenceEnd(cp) {
119
+ return isWhitespace(cp) || cp === $.SOLIDUS || cp === $.GREATER_THAN_SIGN;
120
+ }
121
+ function getErrorForNumericCharacterReference(code) {
122
+ if (code === $.NULL) {
123
+ return ERR.nullCharacterReference;
124
+ }
125
+ else if (code > 1114111) {
126
+ return ERR.characterReferenceOutsideUnicodeRange;
127
+ }
128
+ else if (isSurrogate(code)) {
129
+ return ERR.surrogateCharacterReference;
130
+ }
131
+ else if (isUndefinedCodePoint(code)) {
132
+ return ERR.noncharacterCharacterReference;
133
+ }
134
+ else if (isControlCodePoint(code) || code === $.CARRIAGE_RETURN) {
135
+ return ERR.controlCharacterReference;
136
+ }
137
+ return null;
138
+ }
139
+ //Tokenizer
140
+ export class Tokenizer {
141
+ constructor(options, handler) {
142
+ this.options = options;
143
+ this.handler = handler;
144
+ this.paused = false;
145
+ /** Ensures that the parsing loop isn't run multiple times at once. */
146
+ this.inLoop = false;
147
+ /**
148
+ * Indicates that the current adjusted node exists, is not an element in the HTML namespace,
149
+ * and that it is not an integration point for either MathML or HTML.
150
+ *
151
+ * @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
152
+ */
153
+ this.inForeignNode = false;
154
+ this.lastStartTagName = '';
155
+ this.active = false;
156
+ this.state = State.DATA;
157
+ this.returnState = State.DATA;
158
+ this.entityStartPos = 0;
159
+ this.consumedAfterSnapshot = -1;
160
+ this.currentCharacterToken = null;
161
+ this.currentToken = null;
162
+ this.currentAttr = { name: '', value: '' };
163
+ this.preprocessor = new Preprocessor(handler);
164
+ this.currentLocation = this.getCurrentLocation(-1);
165
+ this.entityDecoder = new EntityDecoder(htmlDecodeTree, (cp, consumed) => {
166
+ // Note: Set `pos` _before_ flushing, as flushing might drop
167
+ // the current chunk and invalidate `entityStartPos`.
168
+ this.preprocessor.pos = this.entityStartPos + consumed - 1;
169
+ this._flushCodePointConsumedAsCharacterReference(cp);
170
+ }, handler.onParseError
171
+ ? {
172
+ missingSemicolonAfterCharacterReference: () => {
173
+ this._err(ERR.missingSemicolonAfterCharacterReference, 1);
174
+ },
175
+ absenceOfDigitsInNumericCharacterReference: (consumed) => {
176
+ this._err(ERR.absenceOfDigitsInNumericCharacterReference, this.entityStartPos - this.preprocessor.pos + consumed);
177
+ },
178
+ validateNumericCharacterReference: (code) => {
179
+ const error = getErrorForNumericCharacterReference(code);
180
+ if (error)
181
+ this._err(error, 1);
182
+ },
183
+ }
184
+ : undefined);
185
+ }
186
+ //Errors
187
+ _err(code, cpOffset = 0) {
188
+ var _a, _b;
189
+ (_b = (_a = this.handler).onParseError) === null || _b === void 0 ? void 0 : _b.call(_a, this.preprocessor.getError(code, cpOffset));
190
+ }
191
+ // NOTE: `offset` may never run across line boundaries.
192
+ getCurrentLocation(offset) {
193
+ if (!this.options.sourceCodeLocationInfo) {
194
+ return null;
195
+ }
196
+ return {
197
+ startLine: this.preprocessor.line,
198
+ startCol: this.preprocessor.col - offset,
199
+ startOffset: this.preprocessor.offset - offset,
200
+ endLine: -1,
201
+ endCol: -1,
202
+ endOffset: -1,
203
+ };
204
+ }
205
+ _runParsingLoop() {
206
+ if (this.inLoop)
207
+ return;
208
+ this.inLoop = true;
209
+ while (this.active && !this.paused) {
210
+ this.consumedAfterSnapshot = 0;
211
+ const cp = this._consume();
212
+ if (!this._ensureHibernation()) {
213
+ this._callState(cp);
214
+ }
215
+ }
216
+ this.inLoop = false;
217
+ }
218
+ //API
219
+ pause() {
220
+ this.paused = true;
221
+ }
222
+ resume(writeCallback) {
223
+ if (!this.paused) {
224
+ throw new Error('Parser was already resumed');
225
+ }
226
+ this.paused = false;
227
+ // Necessary for synchronous resume.
228
+ if (this.inLoop)
229
+ return;
230
+ this._runParsingLoop();
231
+ if (!this.paused) {
232
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
233
+ }
234
+ }
235
+ write(chunk, isLastChunk, writeCallback) {
236
+ this.active = true;
237
+ this.preprocessor.write(chunk, isLastChunk);
238
+ this._runParsingLoop();
239
+ if (!this.paused) {
240
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
241
+ }
242
+ }
243
+ insertHtmlAtCurrentPos(chunk) {
244
+ this.active = true;
245
+ this.preprocessor.insertHtmlAtCurrentPos(chunk);
246
+ this._runParsingLoop();
247
+ }
248
+ //Hibernation
249
+ _ensureHibernation() {
250
+ if (this.preprocessor.endOfChunkHit) {
251
+ this.preprocessor.retreat(this.consumedAfterSnapshot);
252
+ this.consumedAfterSnapshot = 0;
253
+ this.active = false;
254
+ return true;
255
+ }
256
+ return false;
257
+ }
258
+ //Consumption
259
+ _consume() {
260
+ this.consumedAfterSnapshot++;
261
+ return this.preprocessor.advance();
262
+ }
263
+ _advanceBy(count) {
264
+ this.consumedAfterSnapshot += count;
265
+ for (let i = 0; i < count; i++) {
266
+ this.preprocessor.advance();
267
+ }
268
+ }
269
+ _consumeSequenceIfMatch(pattern, caseSensitive) {
270
+ if (this.preprocessor.startsWith(pattern, caseSensitive)) {
271
+ // We will already have consumed one character before calling this method.
272
+ this._advanceBy(pattern.length - 1);
273
+ return true;
274
+ }
275
+ return false;
276
+ }
277
+ //Token creation
278
+ _createStartTagToken() {
279
+ this.currentToken = {
280
+ type: TokenType.START_TAG,
281
+ tagName: '',
282
+ tagID: TAG_ID.UNKNOWN,
283
+ selfClosing: false,
284
+ ackSelfClosing: false,
285
+ attrs: [],
286
+ location: this.getCurrentLocation(1),
287
+ };
288
+ }
289
+ _createEndTagToken() {
290
+ this.currentToken = {
291
+ type: TokenType.END_TAG,
292
+ tagName: '',
293
+ tagID: TAG_ID.UNKNOWN,
294
+ selfClosing: false,
295
+ ackSelfClosing: false,
296
+ attrs: [],
297
+ location: this.getCurrentLocation(2),
298
+ };
299
+ }
300
+ _createCommentToken(offset) {
301
+ this.currentToken = {
302
+ type: TokenType.COMMENT,
303
+ data: '',
304
+ location: this.getCurrentLocation(offset),
305
+ };
306
+ }
307
+ _createDoctypeToken(initialName) {
308
+ this.currentToken = {
309
+ type: TokenType.DOCTYPE,
310
+ name: initialName,
311
+ forceQuirks: false,
312
+ publicId: null,
313
+ systemId: null,
314
+ location: this.currentLocation,
315
+ };
316
+ }
317
+ _createCharacterToken(type, chars) {
318
+ this.currentCharacterToken = {
319
+ type,
320
+ chars,
321
+ location: this.currentLocation,
322
+ };
323
+ }
324
+ //Tag attributes
325
+ _createAttr(attrNameFirstCh) {
326
+ this.currentAttr = {
327
+ name: attrNameFirstCh,
328
+ value: '',
329
+ };
330
+ this.currentLocation = this.getCurrentLocation(0);
331
+ }
332
+ _leaveAttrName() {
333
+ var _a;
334
+ var _b;
335
+ const token = this.currentToken;
336
+ if (getTokenAttr(token, this.currentAttr.name) === null) {
337
+ token.attrs.push(this.currentAttr);
338
+ if (token.location && this.currentLocation) {
339
+ const attrLocations = ((_a = (_b = token.location).attrs) !== null && _a !== void 0 ? _a : (_b.attrs = Object.create(null)));
340
+ attrLocations[this.currentAttr.name] = this.currentLocation;
341
+ // Set end location
342
+ this._leaveAttrValue();
343
+ }
344
+ }
345
+ else {
346
+ this._err(ERR.duplicateAttribute);
347
+ }
348
+ }
349
+ _leaveAttrValue() {
350
+ if (this.currentLocation) {
351
+ this.currentLocation.endLine = this.preprocessor.line;
352
+ this.currentLocation.endCol = this.preprocessor.col;
353
+ this.currentLocation.endOffset = this.preprocessor.offset;
354
+ }
355
+ }
356
+ //Token emission
357
+ prepareToken(ct) {
358
+ this._emitCurrentCharacterToken(ct.location);
359
+ this.currentToken = null;
360
+ if (ct.location) {
361
+ ct.location.endLine = this.preprocessor.line;
362
+ ct.location.endCol = this.preprocessor.col + 1;
363
+ ct.location.endOffset = this.preprocessor.offset + 1;
364
+ }
365
+ this.currentLocation = this.getCurrentLocation(-1);
366
+ }
367
+ emitCurrentTagToken() {
368
+ const ct = this.currentToken;
369
+ this.prepareToken(ct);
370
+ ct.tagID = getTagID(ct.tagName);
371
+ if (ct.type === TokenType.START_TAG) {
372
+ this.lastStartTagName = ct.tagName;
373
+ this.handler.onStartTag(ct);
374
+ }
375
+ else {
376
+ if (ct.attrs.length > 0) {
377
+ this._err(ERR.endTagWithAttributes);
378
+ }
379
+ if (ct.selfClosing) {
380
+ this._err(ERR.endTagWithTrailingSolidus);
381
+ }
382
+ this.handler.onEndTag(ct);
383
+ }
384
+ this.preprocessor.dropParsedChunk();
385
+ }
386
+ emitCurrentComment(ct) {
387
+ this.prepareToken(ct);
388
+ this.handler.onComment(ct);
389
+ this.preprocessor.dropParsedChunk();
390
+ }
391
+ emitCurrentDoctype(ct) {
392
+ this.prepareToken(ct);
393
+ this.handler.onDoctype(ct);
394
+ this.preprocessor.dropParsedChunk();
395
+ }
396
+ _emitCurrentCharacterToken(nextLocation) {
397
+ if (this.currentCharacterToken) {
398
+ //NOTE: if we have a pending character token, make it's end location equal to the
399
+ //current token's start location.
400
+ if (nextLocation && this.currentCharacterToken.location) {
401
+ this.currentCharacterToken.location.endLine = nextLocation.startLine;
402
+ this.currentCharacterToken.location.endCol = nextLocation.startCol;
403
+ this.currentCharacterToken.location.endOffset = nextLocation.startOffset;
404
+ }
405
+ switch (this.currentCharacterToken.type) {
406
+ case TokenType.CHARACTER: {
407
+ this.handler.onCharacter(this.currentCharacterToken);
408
+ break;
409
+ }
410
+ case TokenType.NULL_CHARACTER: {
411
+ this.handler.onNullCharacter(this.currentCharacterToken);
412
+ break;
413
+ }
414
+ case TokenType.WHITESPACE_CHARACTER: {
415
+ this.handler.onWhitespaceCharacter(this.currentCharacterToken);
416
+ break;
417
+ }
418
+ }
419
+ this.currentCharacterToken = null;
420
+ }
421
+ }
422
+ _emitEOFToken() {
423
+ const location = this.getCurrentLocation(0);
424
+ if (location) {
425
+ location.endLine = location.startLine;
426
+ location.endCol = location.startCol;
427
+ location.endOffset = location.startOffset;
428
+ }
429
+ this._emitCurrentCharacterToken(location);
430
+ this.handler.onEof({ type: TokenType.EOF, location });
431
+ this.active = false;
432
+ }
433
+ //Characters emission
434
+ //OPTIMIZATION: The specification uses only one type of character token (one token per character).
435
+ //This causes a huge memory overhead and a lot of unnecessary parser loops. parse5 uses 3 groups of characters.
436
+ //If we have a sequence of characters that belong to the same group, the parser can process it
437
+ //as a single solid character token.
438
+ //So, there are 3 types of character tokens in parse5:
439
+ //1)TokenType.NULL_CHARACTER - \u0000-character sequences (e.g. '\u0000\u0000\u0000')
440
+ //2)TokenType.WHITESPACE_CHARACTER - any whitespace/new-line character sequences (e.g. '\n \r\t \f')
441
+ //3)TokenType.CHARACTER - any character sequence which don't belong to groups 1 and 2 (e.g. 'abcdef1234@@#$%^')
442
+ _appendCharToCurrentCharacterToken(type, ch) {
443
+ if (this.currentCharacterToken) {
444
+ if (this.currentCharacterToken.type === type) {
445
+ this.currentCharacterToken.chars += ch;
446
+ return;
447
+ }
448
+ else {
449
+ this.currentLocation = this.getCurrentLocation(0);
450
+ this._emitCurrentCharacterToken(this.currentLocation);
451
+ this.preprocessor.dropParsedChunk();
452
+ }
453
+ }
454
+ this._createCharacterToken(type, ch);
455
+ }
456
+ _emitCodePoint(cp) {
457
+ const type = isWhitespace(cp)
458
+ ? TokenType.WHITESPACE_CHARACTER
459
+ : cp === $.NULL
460
+ ? TokenType.NULL_CHARACTER
461
+ : TokenType.CHARACTER;
462
+ // OPTIMIZATION: Use String.fromCharCode for BMP characters (< 0x10000) which is faster
463
+ // than String.fromCodePoint. Characters outside BMP are rare in HTML.
464
+ this._appendCharToCurrentCharacterToken(type, cp < 65536 ? String.fromCharCode(cp) : String.fromCodePoint(cp));
465
+ }
466
+ //NOTE: used when we emit characters explicitly.
467
+ //This is always for non-whitespace and non-null characters, which allows us to avoid additional checks.
468
+ _emitChars(ch) {
469
+ this._appendCharToCurrentCharacterToken(TokenType.CHARACTER, ch);
470
+ }
471
+ // Character reference helpers
472
+ _startCharacterReference() {
473
+ this.returnState = this.state;
474
+ this.state = State.CHARACTER_REFERENCE;
475
+ this.entityStartPos = this.preprocessor.pos;
476
+ this.entityDecoder.startEntity(this._isCharacterReferenceInAttribute() ? DecodingMode.Attribute : DecodingMode.Legacy);
477
+ }
478
+ _isCharacterReferenceInAttribute() {
479
+ return (this.returnState === State.ATTRIBUTE_VALUE_DOUBLE_QUOTED ||
480
+ this.returnState === State.ATTRIBUTE_VALUE_SINGLE_QUOTED ||
481
+ this.returnState === State.ATTRIBUTE_VALUE_UNQUOTED);
482
+ }
483
+ _flushCodePointConsumedAsCharacterReference(cp) {
484
+ if (this._isCharacterReferenceInAttribute()) {
485
+ this.currentAttr.value += String.fromCodePoint(cp);
486
+ }
487
+ else {
488
+ this._emitCodePoint(cp);
489
+ }
490
+ }
491
+ // Calling states this way turns out to be much faster than any other approach.
492
+ _callState(cp) {
493
+ switch (this.state) {
494
+ case State.DATA: {
495
+ this._stateData(cp);
496
+ break;
497
+ }
498
+ case State.RCDATA: {
499
+ this._stateRcdata(cp);
500
+ break;
501
+ }
502
+ case State.RAWTEXT: {
503
+ this._stateRawtext(cp);
504
+ break;
505
+ }
506
+ case State.SCRIPT_DATA: {
507
+ this._stateScriptData(cp);
508
+ break;
509
+ }
510
+ case State.PLAINTEXT: {
511
+ this._statePlaintext(cp);
512
+ break;
513
+ }
514
+ case State.TAG_OPEN: {
515
+ this._stateTagOpen(cp);
516
+ break;
517
+ }
518
+ case State.END_TAG_OPEN: {
519
+ this._stateEndTagOpen(cp);
520
+ break;
521
+ }
522
+ case State.TAG_NAME: {
523
+ this._stateTagName(cp);
524
+ break;
525
+ }
526
+ case State.RCDATA_LESS_THAN_SIGN: {
527
+ this._stateRcdataLessThanSign(cp);
528
+ break;
529
+ }
530
+ case State.RCDATA_END_TAG_OPEN: {
531
+ this._stateRcdataEndTagOpen(cp);
532
+ break;
533
+ }
534
+ case State.RCDATA_END_TAG_NAME: {
535
+ this._stateRcdataEndTagName(cp);
536
+ break;
537
+ }
538
+ case State.RAWTEXT_LESS_THAN_SIGN: {
539
+ this._stateRawtextLessThanSign(cp);
540
+ break;
541
+ }
542
+ case State.RAWTEXT_END_TAG_OPEN: {
543
+ this._stateRawtextEndTagOpen(cp);
544
+ break;
545
+ }
546
+ case State.RAWTEXT_END_TAG_NAME: {
547
+ this._stateRawtextEndTagName(cp);
548
+ break;
549
+ }
550
+ case State.SCRIPT_DATA_LESS_THAN_SIGN: {
551
+ this._stateScriptDataLessThanSign(cp);
552
+ break;
553
+ }
554
+ case State.SCRIPT_DATA_END_TAG_OPEN: {
555
+ this._stateScriptDataEndTagOpen(cp);
556
+ break;
557
+ }
558
+ case State.SCRIPT_DATA_END_TAG_NAME: {
559
+ this._stateScriptDataEndTagName(cp);
560
+ break;
561
+ }
562
+ case State.SCRIPT_DATA_ESCAPE_START: {
563
+ this._stateScriptDataEscapeStart(cp);
564
+ break;
565
+ }
566
+ case State.SCRIPT_DATA_ESCAPE_START_DASH: {
567
+ this._stateScriptDataEscapeStartDash(cp);
568
+ break;
569
+ }
570
+ case State.SCRIPT_DATA_ESCAPED: {
571
+ this._stateScriptDataEscaped(cp);
572
+ break;
573
+ }
574
+ case State.SCRIPT_DATA_ESCAPED_DASH: {
575
+ this._stateScriptDataEscapedDash(cp);
576
+ break;
577
+ }
578
+ case State.SCRIPT_DATA_ESCAPED_DASH_DASH: {
579
+ this._stateScriptDataEscapedDashDash(cp);
580
+ break;
581
+ }
582
+ case State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN: {
583
+ this._stateScriptDataEscapedLessThanSign(cp);
584
+ break;
585
+ }
586
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN: {
587
+ this._stateScriptDataEscapedEndTagOpen(cp);
588
+ break;
589
+ }
590
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_NAME: {
591
+ this._stateScriptDataEscapedEndTagName(cp);
592
+ break;
593
+ }
594
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_START: {
595
+ this._stateScriptDataDoubleEscapeStart(cp);
596
+ break;
597
+ }
598
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED: {
599
+ this._stateScriptDataDoubleEscaped(cp);
600
+ break;
601
+ }
602
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH: {
603
+ this._stateScriptDataDoubleEscapedDash(cp);
604
+ break;
605
+ }
606
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH: {
607
+ this._stateScriptDataDoubleEscapedDashDash(cp);
608
+ break;
609
+ }
610
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN: {
611
+ this._stateScriptDataDoubleEscapedLessThanSign(cp);
612
+ break;
613
+ }
614
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_END: {
615
+ this._stateScriptDataDoubleEscapeEnd(cp);
616
+ break;
617
+ }
618
+ case State.BEFORE_ATTRIBUTE_NAME: {
619
+ this._stateBeforeAttributeName(cp);
620
+ break;
621
+ }
622
+ case State.ATTRIBUTE_NAME: {
623
+ this._stateAttributeName(cp);
624
+ break;
625
+ }
626
+ case State.AFTER_ATTRIBUTE_NAME: {
627
+ this._stateAfterAttributeName(cp);
628
+ break;
629
+ }
630
+ case State.BEFORE_ATTRIBUTE_VALUE: {
631
+ this._stateBeforeAttributeValue(cp);
632
+ break;
633
+ }
634
+ case State.ATTRIBUTE_VALUE_DOUBLE_QUOTED: {
635
+ this._stateAttributeValueDoubleQuoted(cp);
636
+ break;
637
+ }
638
+ case State.ATTRIBUTE_VALUE_SINGLE_QUOTED: {
639
+ this._stateAttributeValueSingleQuoted(cp);
640
+ break;
641
+ }
642
+ case State.ATTRIBUTE_VALUE_UNQUOTED: {
643
+ this._stateAttributeValueUnquoted(cp);
644
+ break;
645
+ }
646
+ case State.AFTER_ATTRIBUTE_VALUE_QUOTED: {
647
+ this._stateAfterAttributeValueQuoted(cp);
648
+ break;
649
+ }
650
+ case State.SELF_CLOSING_START_TAG: {
651
+ this._stateSelfClosingStartTag(cp);
652
+ break;
653
+ }
654
+ case State.BOGUS_COMMENT: {
655
+ this._stateBogusComment(cp);
656
+ break;
657
+ }
658
+ case State.MARKUP_DECLARATION_OPEN: {
659
+ this._stateMarkupDeclarationOpen(cp);
660
+ break;
661
+ }
662
+ case State.COMMENT_START: {
663
+ this._stateCommentStart(cp);
664
+ break;
665
+ }
666
+ case State.COMMENT_START_DASH: {
667
+ this._stateCommentStartDash(cp);
668
+ break;
669
+ }
670
+ case State.COMMENT: {
671
+ this._stateComment(cp);
672
+ break;
673
+ }
674
+ case State.COMMENT_LESS_THAN_SIGN: {
675
+ this._stateCommentLessThanSign(cp);
676
+ break;
677
+ }
678
+ case State.COMMENT_LESS_THAN_SIGN_BANG: {
679
+ this._stateCommentLessThanSignBang(cp);
680
+ break;
681
+ }
682
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH: {
683
+ this._stateCommentLessThanSignBangDash(cp);
684
+ break;
685
+ }
686
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH: {
687
+ this._stateCommentLessThanSignBangDashDash(cp);
688
+ break;
689
+ }
690
+ case State.COMMENT_END_DASH: {
691
+ this._stateCommentEndDash(cp);
692
+ break;
693
+ }
694
+ case State.COMMENT_END: {
695
+ this._stateCommentEnd(cp);
696
+ break;
697
+ }
698
+ case State.COMMENT_END_BANG: {
699
+ this._stateCommentEndBang(cp);
700
+ break;
701
+ }
702
+ case State.DOCTYPE: {
703
+ this._stateDoctype(cp);
704
+ break;
705
+ }
706
+ case State.BEFORE_DOCTYPE_NAME: {
707
+ this._stateBeforeDoctypeName(cp);
708
+ break;
709
+ }
710
+ case State.DOCTYPE_NAME: {
711
+ this._stateDoctypeName(cp);
712
+ break;
713
+ }
714
+ case State.AFTER_DOCTYPE_NAME: {
715
+ this._stateAfterDoctypeName(cp);
716
+ break;
717
+ }
718
+ case State.AFTER_DOCTYPE_PUBLIC_KEYWORD: {
719
+ this._stateAfterDoctypePublicKeyword(cp);
720
+ break;
721
+ }
722
+ case State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER: {
723
+ this._stateBeforeDoctypePublicIdentifier(cp);
724
+ break;
725
+ }
726
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED: {
727
+ this._stateDoctypePublicIdentifierDoubleQuoted(cp);
728
+ break;
729
+ }
730
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED: {
731
+ this._stateDoctypePublicIdentifierSingleQuoted(cp);
732
+ break;
733
+ }
734
+ case State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER: {
735
+ this._stateAfterDoctypePublicIdentifier(cp);
736
+ break;
737
+ }
738
+ case State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS: {
739
+ this._stateBetweenDoctypePublicAndSystemIdentifiers(cp);
740
+ break;
741
+ }
742
+ case State.AFTER_DOCTYPE_SYSTEM_KEYWORD: {
743
+ this._stateAfterDoctypeSystemKeyword(cp);
744
+ break;
745
+ }
746
+ case State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER: {
747
+ this._stateBeforeDoctypeSystemIdentifier(cp);
748
+ break;
749
+ }
750
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED: {
751
+ this._stateDoctypeSystemIdentifierDoubleQuoted(cp);
752
+ break;
753
+ }
754
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED: {
755
+ this._stateDoctypeSystemIdentifierSingleQuoted(cp);
756
+ break;
757
+ }
758
+ case State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER: {
759
+ this._stateAfterDoctypeSystemIdentifier(cp);
760
+ break;
761
+ }
762
+ case State.BOGUS_DOCTYPE: {
763
+ this._stateBogusDoctype(cp);
764
+ break;
765
+ }
766
+ case State.CDATA_SECTION: {
767
+ this._stateCdataSection(cp);
768
+ break;
769
+ }
770
+ case State.CDATA_SECTION_BRACKET: {
771
+ this._stateCdataSectionBracket(cp);
772
+ break;
773
+ }
774
+ case State.CDATA_SECTION_END: {
775
+ this._stateCdataSectionEnd(cp);
776
+ break;
777
+ }
778
+ case State.CHARACTER_REFERENCE: {
779
+ this._stateCharacterReference();
780
+ break;
781
+ }
782
+ case State.AMBIGUOUS_AMPERSAND: {
783
+ this._stateAmbiguousAmpersand(cp);
784
+ break;
785
+ }
786
+ default: {
787
+ throw new Error('Unknown state');
788
+ }
789
+ }
790
+ }
791
+ // State machine
792
+ // Data state
793
+ //------------------------------------------------------------------
794
+ _stateData(cp) {
795
+ switch (cp) {
796
+ case $.LESS_THAN_SIGN: {
797
+ this.state = State.TAG_OPEN;
798
+ break;
799
+ }
800
+ case $.AMPERSAND: {
801
+ this._startCharacterReference();
802
+ break;
803
+ }
804
+ case $.NULL: {
805
+ this._err(ERR.unexpectedNullCharacter);
806
+ this._emitCodePoint(cp);
807
+ break;
808
+ }
809
+ case $.EOF: {
810
+ this._emitEOFToken();
811
+ break;
812
+ }
813
+ default: {
814
+ this._emitCodePoint(cp);
815
+ }
816
+ }
817
+ }
818
+ // RCDATA state
819
+ //------------------------------------------------------------------
820
+ _stateRcdata(cp) {
821
+ switch (cp) {
822
+ case $.AMPERSAND: {
823
+ this._startCharacterReference();
824
+ break;
825
+ }
826
+ case $.LESS_THAN_SIGN: {
827
+ this.state = State.RCDATA_LESS_THAN_SIGN;
828
+ break;
829
+ }
830
+ case $.NULL: {
831
+ this._err(ERR.unexpectedNullCharacter);
832
+ this._emitChars(REPLACEMENT_CHARACTER);
833
+ break;
834
+ }
835
+ case $.EOF: {
836
+ this._emitEOFToken();
837
+ break;
838
+ }
839
+ default: {
840
+ this._emitCodePoint(cp);
841
+ }
842
+ }
843
+ }
844
+ // RAWTEXT state
845
+ //------------------------------------------------------------------
846
+ _stateRawtext(cp) {
847
+ switch (cp) {
848
+ case $.LESS_THAN_SIGN: {
849
+ this.state = State.RAWTEXT_LESS_THAN_SIGN;
850
+ break;
851
+ }
852
+ case $.NULL: {
853
+ this._err(ERR.unexpectedNullCharacter);
854
+ this._emitChars(REPLACEMENT_CHARACTER);
855
+ break;
856
+ }
857
+ case $.EOF: {
858
+ this._emitEOFToken();
859
+ break;
860
+ }
861
+ default: {
862
+ this._emitCodePoint(cp);
863
+ }
864
+ }
865
+ }
866
+ // Script data state
867
+ //------------------------------------------------------------------
868
+ _stateScriptData(cp) {
869
+ switch (cp) {
870
+ case $.LESS_THAN_SIGN: {
871
+ this.state = State.SCRIPT_DATA_LESS_THAN_SIGN;
872
+ break;
873
+ }
874
+ case $.NULL: {
875
+ this._err(ERR.unexpectedNullCharacter);
876
+ this._emitChars(REPLACEMENT_CHARACTER);
877
+ break;
878
+ }
879
+ case $.EOF: {
880
+ this._emitEOFToken();
881
+ break;
882
+ }
883
+ default: {
884
+ this._emitCodePoint(cp);
885
+ }
886
+ }
887
+ }
888
+ // PLAINTEXT state
889
+ //------------------------------------------------------------------
890
+ _statePlaintext(cp) {
891
+ switch (cp) {
892
+ case $.NULL: {
893
+ this._err(ERR.unexpectedNullCharacter);
894
+ this._emitChars(REPLACEMENT_CHARACTER);
895
+ break;
896
+ }
897
+ case $.EOF: {
898
+ this._emitEOFToken();
899
+ break;
900
+ }
901
+ default: {
902
+ this._emitCodePoint(cp);
903
+ }
904
+ }
905
+ }
906
+ // Tag open state
907
+ //------------------------------------------------------------------
908
+ _stateTagOpen(cp) {
909
+ if (isAsciiLetter(cp)) {
910
+ this._createStartTagToken();
911
+ this.state = State.TAG_NAME;
912
+ this._stateTagName(cp);
913
+ }
914
+ else
915
+ switch (cp) {
916
+ case $.EXCLAMATION_MARK: {
917
+ this.state = State.MARKUP_DECLARATION_OPEN;
918
+ break;
919
+ }
920
+ case $.SOLIDUS: {
921
+ this.state = State.END_TAG_OPEN;
922
+ break;
923
+ }
924
+ case $.QUESTION_MARK: {
925
+ this._err(ERR.unexpectedQuestionMarkInsteadOfTagName);
926
+ this._createCommentToken(1);
927
+ this.state = State.BOGUS_COMMENT;
928
+ this._stateBogusComment(cp);
929
+ break;
930
+ }
931
+ case $.EOF: {
932
+ this._err(ERR.eofBeforeTagName);
933
+ this._emitChars('<');
934
+ this._emitEOFToken();
935
+ break;
936
+ }
937
+ default: {
938
+ this._err(ERR.invalidFirstCharacterOfTagName);
939
+ this._emitChars('<');
940
+ this.state = State.DATA;
941
+ this._stateData(cp);
942
+ }
943
+ }
944
+ }
945
+ // End tag open state
946
+ //------------------------------------------------------------------
947
+ _stateEndTagOpen(cp) {
948
+ if (isAsciiLetter(cp)) {
949
+ this._createEndTagToken();
950
+ this.state = State.TAG_NAME;
951
+ this._stateTagName(cp);
952
+ }
953
+ else
954
+ switch (cp) {
955
+ case $.GREATER_THAN_SIGN: {
956
+ this._err(ERR.missingEndTagName);
957
+ this.state = State.DATA;
958
+ break;
959
+ }
960
+ case $.EOF: {
961
+ this._err(ERR.eofBeforeTagName);
962
+ this._emitChars('</');
963
+ this._emitEOFToken();
964
+ break;
965
+ }
966
+ default: {
967
+ this._err(ERR.invalidFirstCharacterOfTagName);
968
+ this._createCommentToken(2);
969
+ this.state = State.BOGUS_COMMENT;
970
+ this._stateBogusComment(cp);
971
+ }
972
+ }
973
+ }
974
+ // Tag name state
975
+ //------------------------------------------------------------------
976
+ _stateTagName(cp) {
977
+ const token = this.currentToken;
978
+ switch (cp) {
979
+ case $.SPACE:
980
+ case $.LINE_FEED:
981
+ case $.TABULATION:
982
+ case $.FORM_FEED: {
983
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
984
+ break;
985
+ }
986
+ case $.SOLIDUS: {
987
+ this.state = State.SELF_CLOSING_START_TAG;
988
+ break;
989
+ }
990
+ case $.GREATER_THAN_SIGN: {
991
+ this.state = State.DATA;
992
+ this.emitCurrentTagToken();
993
+ break;
994
+ }
995
+ case $.NULL: {
996
+ this._err(ERR.unexpectedNullCharacter);
997
+ token.tagName += REPLACEMENT_CHARACTER;
998
+ break;
999
+ }
1000
+ case $.EOF: {
1001
+ this._err(ERR.eofInTag);
1002
+ this._emitEOFToken();
1003
+ break;
1004
+ }
1005
+ default: {
1006
+ token.tagName += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1007
+ }
1008
+ }
1009
+ }
1010
+ // RCDATA less-than sign state
1011
+ //------------------------------------------------------------------
1012
+ _stateRcdataLessThanSign(cp) {
1013
+ if (cp === $.SOLIDUS) {
1014
+ this.state = State.RCDATA_END_TAG_OPEN;
1015
+ }
1016
+ else {
1017
+ this._emitChars('<');
1018
+ this.state = State.RCDATA;
1019
+ this._stateRcdata(cp);
1020
+ }
1021
+ }
1022
+ // RCDATA end tag open state
1023
+ //------------------------------------------------------------------
1024
+ _stateRcdataEndTagOpen(cp) {
1025
+ if (isAsciiLetter(cp)) {
1026
+ this.state = State.RCDATA_END_TAG_NAME;
1027
+ this._stateRcdataEndTagName(cp);
1028
+ }
1029
+ else {
1030
+ this._emitChars('</');
1031
+ this.state = State.RCDATA;
1032
+ this._stateRcdata(cp);
1033
+ }
1034
+ }
1035
+ handleSpecialEndTag(_cp) {
1036
+ if (!this.preprocessor.startsWith(this.lastStartTagName, false)) {
1037
+ return !this._ensureHibernation();
1038
+ }
1039
+ this._createEndTagToken();
1040
+ const token = this.currentToken;
1041
+ token.tagName = this.lastStartTagName;
1042
+ const cp = this.preprocessor.peek(this.lastStartTagName.length);
1043
+ switch (cp) {
1044
+ case $.SPACE:
1045
+ case $.LINE_FEED:
1046
+ case $.TABULATION:
1047
+ case $.FORM_FEED: {
1048
+ this._advanceBy(this.lastStartTagName.length);
1049
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1050
+ return false;
1051
+ }
1052
+ case $.SOLIDUS: {
1053
+ this._advanceBy(this.lastStartTagName.length);
1054
+ this.state = State.SELF_CLOSING_START_TAG;
1055
+ return false;
1056
+ }
1057
+ case $.GREATER_THAN_SIGN: {
1058
+ this._advanceBy(this.lastStartTagName.length);
1059
+ this.emitCurrentTagToken();
1060
+ this.state = State.DATA;
1061
+ return false;
1062
+ }
1063
+ default: {
1064
+ return !this._ensureHibernation();
1065
+ }
1066
+ }
1067
+ }
1068
+ // RCDATA end tag name state
1069
+ //------------------------------------------------------------------
1070
+ _stateRcdataEndTagName(cp) {
1071
+ if (this.handleSpecialEndTag(cp)) {
1072
+ this._emitChars('</');
1073
+ this.state = State.RCDATA;
1074
+ this._stateRcdata(cp);
1075
+ }
1076
+ }
1077
+ // RAWTEXT less-than sign state
1078
+ //------------------------------------------------------------------
1079
+ _stateRawtextLessThanSign(cp) {
1080
+ if (cp === $.SOLIDUS) {
1081
+ this.state = State.RAWTEXT_END_TAG_OPEN;
1082
+ }
1083
+ else {
1084
+ this._emitChars('<');
1085
+ this.state = State.RAWTEXT;
1086
+ this._stateRawtext(cp);
1087
+ }
1088
+ }
1089
+ // RAWTEXT end tag open state
1090
+ //------------------------------------------------------------------
1091
+ _stateRawtextEndTagOpen(cp) {
1092
+ if (isAsciiLetter(cp)) {
1093
+ this.state = State.RAWTEXT_END_TAG_NAME;
1094
+ this._stateRawtextEndTagName(cp);
1095
+ }
1096
+ else {
1097
+ this._emitChars('</');
1098
+ this.state = State.RAWTEXT;
1099
+ this._stateRawtext(cp);
1100
+ }
1101
+ }
1102
+ // RAWTEXT end tag name state
1103
+ //------------------------------------------------------------------
1104
+ _stateRawtextEndTagName(cp) {
1105
+ if (this.handleSpecialEndTag(cp)) {
1106
+ this._emitChars('</');
1107
+ this.state = State.RAWTEXT;
1108
+ this._stateRawtext(cp);
1109
+ }
1110
+ }
1111
+ // Script data less-than sign state
1112
+ //------------------------------------------------------------------
1113
+ _stateScriptDataLessThanSign(cp) {
1114
+ switch (cp) {
1115
+ case $.SOLIDUS: {
1116
+ this.state = State.SCRIPT_DATA_END_TAG_OPEN;
1117
+ break;
1118
+ }
1119
+ case $.EXCLAMATION_MARK: {
1120
+ this.state = State.SCRIPT_DATA_ESCAPE_START;
1121
+ this._emitChars('<!');
1122
+ break;
1123
+ }
1124
+ default: {
1125
+ this._emitChars('<');
1126
+ this.state = State.SCRIPT_DATA;
1127
+ this._stateScriptData(cp);
1128
+ }
1129
+ }
1130
+ }
1131
+ // Script data end tag open state
1132
+ //------------------------------------------------------------------
1133
+ _stateScriptDataEndTagOpen(cp) {
1134
+ if (isAsciiLetter(cp)) {
1135
+ this.state = State.SCRIPT_DATA_END_TAG_NAME;
1136
+ this._stateScriptDataEndTagName(cp);
1137
+ }
1138
+ else {
1139
+ this._emitChars('</');
1140
+ this.state = State.SCRIPT_DATA;
1141
+ this._stateScriptData(cp);
1142
+ }
1143
+ }
1144
+ // Script data end tag name state
1145
+ //------------------------------------------------------------------
1146
+ _stateScriptDataEndTagName(cp) {
1147
+ if (this.handleSpecialEndTag(cp)) {
1148
+ this._emitChars('</');
1149
+ this.state = State.SCRIPT_DATA;
1150
+ this._stateScriptData(cp);
1151
+ }
1152
+ }
1153
+ // Script data escape start state
1154
+ //------------------------------------------------------------------
1155
+ _stateScriptDataEscapeStart(cp) {
1156
+ if (cp === $.HYPHEN_MINUS) {
1157
+ this.state = State.SCRIPT_DATA_ESCAPE_START_DASH;
1158
+ this._emitChars('-');
1159
+ }
1160
+ else {
1161
+ this.state = State.SCRIPT_DATA;
1162
+ this._stateScriptData(cp);
1163
+ }
1164
+ }
1165
+ // Script data escape start dash state
1166
+ //------------------------------------------------------------------
1167
+ _stateScriptDataEscapeStartDash(cp) {
1168
+ if (cp === $.HYPHEN_MINUS) {
1169
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1170
+ this._emitChars('-');
1171
+ }
1172
+ else {
1173
+ this.state = State.SCRIPT_DATA;
1174
+ this._stateScriptData(cp);
1175
+ }
1176
+ }
1177
+ // Script data escaped state
1178
+ //------------------------------------------------------------------
1179
+ _stateScriptDataEscaped(cp) {
1180
+ switch (cp) {
1181
+ case $.HYPHEN_MINUS: {
1182
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH;
1183
+ this._emitChars('-');
1184
+ break;
1185
+ }
1186
+ case $.LESS_THAN_SIGN: {
1187
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1188
+ break;
1189
+ }
1190
+ case $.NULL: {
1191
+ this._err(ERR.unexpectedNullCharacter);
1192
+ this._emitChars(REPLACEMENT_CHARACTER);
1193
+ break;
1194
+ }
1195
+ case $.EOF: {
1196
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1197
+ this._emitEOFToken();
1198
+ break;
1199
+ }
1200
+ default: {
1201
+ this._emitCodePoint(cp);
1202
+ }
1203
+ }
1204
+ }
1205
+ // Script data escaped dash state
1206
+ //------------------------------------------------------------------
1207
+ _stateScriptDataEscapedDash(cp) {
1208
+ switch (cp) {
1209
+ case $.HYPHEN_MINUS: {
1210
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1211
+ this._emitChars('-');
1212
+ break;
1213
+ }
1214
+ case $.LESS_THAN_SIGN: {
1215
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1216
+ break;
1217
+ }
1218
+ case $.NULL: {
1219
+ this._err(ERR.unexpectedNullCharacter);
1220
+ this.state = State.SCRIPT_DATA_ESCAPED;
1221
+ this._emitChars(REPLACEMENT_CHARACTER);
1222
+ break;
1223
+ }
1224
+ case $.EOF: {
1225
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1226
+ this._emitEOFToken();
1227
+ break;
1228
+ }
1229
+ default: {
1230
+ this.state = State.SCRIPT_DATA_ESCAPED;
1231
+ this._emitCodePoint(cp);
1232
+ }
1233
+ }
1234
+ }
1235
+ // Script data escaped dash dash state
1236
+ //------------------------------------------------------------------
1237
+ _stateScriptDataEscapedDashDash(cp) {
1238
+ switch (cp) {
1239
+ case $.HYPHEN_MINUS: {
1240
+ this._emitChars('-');
1241
+ break;
1242
+ }
1243
+ case $.LESS_THAN_SIGN: {
1244
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1245
+ break;
1246
+ }
1247
+ case $.GREATER_THAN_SIGN: {
1248
+ this.state = State.SCRIPT_DATA;
1249
+ this._emitChars('>');
1250
+ break;
1251
+ }
1252
+ case $.NULL: {
1253
+ this._err(ERR.unexpectedNullCharacter);
1254
+ this.state = State.SCRIPT_DATA_ESCAPED;
1255
+ this._emitChars(REPLACEMENT_CHARACTER);
1256
+ break;
1257
+ }
1258
+ case $.EOF: {
1259
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1260
+ this._emitEOFToken();
1261
+ break;
1262
+ }
1263
+ default: {
1264
+ this.state = State.SCRIPT_DATA_ESCAPED;
1265
+ this._emitCodePoint(cp);
1266
+ }
1267
+ }
1268
+ }
1269
+ // Script data escaped less-than sign state
1270
+ //------------------------------------------------------------------
1271
+ _stateScriptDataEscapedLessThanSign(cp) {
1272
+ if (cp === $.SOLIDUS) {
1273
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN;
1274
+ }
1275
+ else if (isAsciiLetter(cp)) {
1276
+ this._emitChars('<');
1277
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_START;
1278
+ this._stateScriptDataDoubleEscapeStart(cp);
1279
+ }
1280
+ else {
1281
+ this._emitChars('<');
1282
+ this.state = State.SCRIPT_DATA_ESCAPED;
1283
+ this._stateScriptDataEscaped(cp);
1284
+ }
1285
+ }
1286
+ // Script data escaped end tag open state
1287
+ //------------------------------------------------------------------
1288
+ _stateScriptDataEscapedEndTagOpen(cp) {
1289
+ if (isAsciiLetter(cp)) {
1290
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_NAME;
1291
+ this._stateScriptDataEscapedEndTagName(cp);
1292
+ }
1293
+ else {
1294
+ this._emitChars('</');
1295
+ this.state = State.SCRIPT_DATA_ESCAPED;
1296
+ this._stateScriptDataEscaped(cp);
1297
+ }
1298
+ }
1299
+ // Script data escaped end tag name state
1300
+ //------------------------------------------------------------------
1301
+ _stateScriptDataEscapedEndTagName(cp) {
1302
+ if (this.handleSpecialEndTag(cp)) {
1303
+ this._emitChars('</');
1304
+ this.state = State.SCRIPT_DATA_ESCAPED;
1305
+ this._stateScriptDataEscaped(cp);
1306
+ }
1307
+ }
1308
+ // Script data double escape start state
1309
+ //------------------------------------------------------------------
1310
+ _stateScriptDataDoubleEscapeStart(cp) {
1311
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1312
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1313
+ this._emitCodePoint(cp);
1314
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1315
+ this._emitCodePoint(this._consume());
1316
+ }
1317
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1318
+ }
1319
+ else if (!this._ensureHibernation()) {
1320
+ this.state = State.SCRIPT_DATA_ESCAPED;
1321
+ this._stateScriptDataEscaped(cp);
1322
+ }
1323
+ }
1324
+ // Script data double escaped state
1325
+ //------------------------------------------------------------------
1326
+ _stateScriptDataDoubleEscaped(cp) {
1327
+ switch (cp) {
1328
+ case $.HYPHEN_MINUS: {
1329
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH;
1330
+ this._emitChars('-');
1331
+ break;
1332
+ }
1333
+ case $.LESS_THAN_SIGN: {
1334
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1335
+ this._emitChars('<');
1336
+ break;
1337
+ }
1338
+ case $.NULL: {
1339
+ this._err(ERR.unexpectedNullCharacter);
1340
+ this._emitChars(REPLACEMENT_CHARACTER);
1341
+ break;
1342
+ }
1343
+ case $.EOF: {
1344
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1345
+ this._emitEOFToken();
1346
+ break;
1347
+ }
1348
+ default: {
1349
+ this._emitCodePoint(cp);
1350
+ }
1351
+ }
1352
+ }
1353
+ // Script data double escaped dash state
1354
+ //------------------------------------------------------------------
1355
+ _stateScriptDataDoubleEscapedDash(cp) {
1356
+ switch (cp) {
1357
+ case $.HYPHEN_MINUS: {
1358
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH;
1359
+ this._emitChars('-');
1360
+ break;
1361
+ }
1362
+ case $.LESS_THAN_SIGN: {
1363
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1364
+ this._emitChars('<');
1365
+ break;
1366
+ }
1367
+ case $.NULL: {
1368
+ this._err(ERR.unexpectedNullCharacter);
1369
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1370
+ this._emitChars(REPLACEMENT_CHARACTER);
1371
+ break;
1372
+ }
1373
+ case $.EOF: {
1374
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1375
+ this._emitEOFToken();
1376
+ break;
1377
+ }
1378
+ default: {
1379
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1380
+ this._emitCodePoint(cp);
1381
+ }
1382
+ }
1383
+ }
1384
+ // Script data double escaped dash dash state
1385
+ //------------------------------------------------------------------
1386
+ _stateScriptDataDoubleEscapedDashDash(cp) {
1387
+ switch (cp) {
1388
+ case $.HYPHEN_MINUS: {
1389
+ this._emitChars('-');
1390
+ break;
1391
+ }
1392
+ case $.LESS_THAN_SIGN: {
1393
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1394
+ this._emitChars('<');
1395
+ break;
1396
+ }
1397
+ case $.GREATER_THAN_SIGN: {
1398
+ this.state = State.SCRIPT_DATA;
1399
+ this._emitChars('>');
1400
+ break;
1401
+ }
1402
+ case $.NULL: {
1403
+ this._err(ERR.unexpectedNullCharacter);
1404
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1405
+ this._emitChars(REPLACEMENT_CHARACTER);
1406
+ break;
1407
+ }
1408
+ case $.EOF: {
1409
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1410
+ this._emitEOFToken();
1411
+ break;
1412
+ }
1413
+ default: {
1414
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1415
+ this._emitCodePoint(cp);
1416
+ }
1417
+ }
1418
+ }
1419
+ // Script data double escaped less-than sign state
1420
+ //------------------------------------------------------------------
1421
+ _stateScriptDataDoubleEscapedLessThanSign(cp) {
1422
+ if (cp === $.SOLIDUS) {
1423
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_END;
1424
+ this._emitChars('/');
1425
+ }
1426
+ else {
1427
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1428
+ this._stateScriptDataDoubleEscaped(cp);
1429
+ }
1430
+ }
1431
+ // Script data double escape end state
1432
+ //------------------------------------------------------------------
1433
+ _stateScriptDataDoubleEscapeEnd(cp) {
1434
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1435
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1436
+ this._emitCodePoint(cp);
1437
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1438
+ this._emitCodePoint(this._consume());
1439
+ }
1440
+ this.state = State.SCRIPT_DATA_ESCAPED;
1441
+ }
1442
+ else if (!this._ensureHibernation()) {
1443
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1444
+ this._stateScriptDataDoubleEscaped(cp);
1445
+ }
1446
+ }
1447
+ // Before attribute name state
1448
+ //------------------------------------------------------------------
1449
+ _stateBeforeAttributeName(cp) {
1450
+ switch (cp) {
1451
+ case $.SPACE:
1452
+ case $.LINE_FEED:
1453
+ case $.TABULATION:
1454
+ case $.FORM_FEED: {
1455
+ // Ignore whitespace
1456
+ break;
1457
+ }
1458
+ case $.SOLIDUS:
1459
+ case $.GREATER_THAN_SIGN:
1460
+ case $.EOF: {
1461
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1462
+ this._stateAfterAttributeName(cp);
1463
+ break;
1464
+ }
1465
+ case $.EQUALS_SIGN: {
1466
+ this._err(ERR.unexpectedEqualsSignBeforeAttributeName);
1467
+ this._createAttr('=');
1468
+ this.state = State.ATTRIBUTE_NAME;
1469
+ break;
1470
+ }
1471
+ default: {
1472
+ this._createAttr('');
1473
+ this.state = State.ATTRIBUTE_NAME;
1474
+ this._stateAttributeName(cp);
1475
+ }
1476
+ }
1477
+ }
1478
+ // Attribute name state
1479
+ //------------------------------------------------------------------
1480
+ _stateAttributeName(cp) {
1481
+ switch (cp) {
1482
+ case $.SPACE:
1483
+ case $.LINE_FEED:
1484
+ case $.TABULATION:
1485
+ case $.FORM_FEED:
1486
+ case $.SOLIDUS:
1487
+ case $.GREATER_THAN_SIGN:
1488
+ case $.EOF: {
1489
+ this._leaveAttrName();
1490
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1491
+ this._stateAfterAttributeName(cp);
1492
+ break;
1493
+ }
1494
+ case $.EQUALS_SIGN: {
1495
+ this._leaveAttrName();
1496
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1497
+ break;
1498
+ }
1499
+ case $.QUOTATION_MARK:
1500
+ case $.APOSTROPHE:
1501
+ case $.LESS_THAN_SIGN: {
1502
+ this._err(ERR.unexpectedCharacterInAttributeName);
1503
+ this.currentAttr.name += String.fromCodePoint(cp);
1504
+ break;
1505
+ }
1506
+ case $.NULL: {
1507
+ this._err(ERR.unexpectedNullCharacter);
1508
+ this.currentAttr.name += REPLACEMENT_CHARACTER;
1509
+ break;
1510
+ }
1511
+ default: {
1512
+ this.currentAttr.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1513
+ }
1514
+ }
1515
+ }
1516
+ // After attribute name state
1517
+ //------------------------------------------------------------------
1518
+ _stateAfterAttributeName(cp) {
1519
+ switch (cp) {
1520
+ case $.SPACE:
1521
+ case $.LINE_FEED:
1522
+ case $.TABULATION:
1523
+ case $.FORM_FEED: {
1524
+ // Ignore whitespace
1525
+ break;
1526
+ }
1527
+ case $.SOLIDUS: {
1528
+ this.state = State.SELF_CLOSING_START_TAG;
1529
+ break;
1530
+ }
1531
+ case $.EQUALS_SIGN: {
1532
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1533
+ break;
1534
+ }
1535
+ case $.GREATER_THAN_SIGN: {
1536
+ this.state = State.DATA;
1537
+ this.emitCurrentTagToken();
1538
+ break;
1539
+ }
1540
+ case $.EOF: {
1541
+ this._err(ERR.eofInTag);
1542
+ this._emitEOFToken();
1543
+ break;
1544
+ }
1545
+ default: {
1546
+ this._createAttr('');
1547
+ this.state = State.ATTRIBUTE_NAME;
1548
+ this._stateAttributeName(cp);
1549
+ }
1550
+ }
1551
+ }
1552
+ // Before attribute value state
1553
+ //------------------------------------------------------------------
1554
+ _stateBeforeAttributeValue(cp) {
1555
+ switch (cp) {
1556
+ case $.SPACE:
1557
+ case $.LINE_FEED:
1558
+ case $.TABULATION:
1559
+ case $.FORM_FEED: {
1560
+ // Ignore whitespace
1561
+ break;
1562
+ }
1563
+ case $.QUOTATION_MARK: {
1564
+ this.state = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
1565
+ break;
1566
+ }
1567
+ case $.APOSTROPHE: {
1568
+ this.state = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
1569
+ break;
1570
+ }
1571
+ case $.GREATER_THAN_SIGN: {
1572
+ this._err(ERR.missingAttributeValue);
1573
+ this.state = State.DATA;
1574
+ this.emitCurrentTagToken();
1575
+ break;
1576
+ }
1577
+ default: {
1578
+ this.state = State.ATTRIBUTE_VALUE_UNQUOTED;
1579
+ this._stateAttributeValueUnquoted(cp);
1580
+ }
1581
+ }
1582
+ }
1583
+ // Attribute value (double-quoted) state
1584
+ //------------------------------------------------------------------
1585
+ _stateAttributeValueDoubleQuoted(cp) {
1586
+ switch (cp) {
1587
+ case $.QUOTATION_MARK: {
1588
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1589
+ break;
1590
+ }
1591
+ case $.AMPERSAND: {
1592
+ this._startCharacterReference();
1593
+ break;
1594
+ }
1595
+ case $.NULL: {
1596
+ this._err(ERR.unexpectedNullCharacter);
1597
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1598
+ break;
1599
+ }
1600
+ case $.EOF: {
1601
+ this._err(ERR.eofInTag);
1602
+ this._emitEOFToken();
1603
+ break;
1604
+ }
1605
+ default: {
1606
+ this.currentAttr.value += String.fromCodePoint(cp);
1607
+ }
1608
+ }
1609
+ }
1610
+ // Attribute value (single-quoted) state
1611
+ //------------------------------------------------------------------
1612
+ _stateAttributeValueSingleQuoted(cp) {
1613
+ switch (cp) {
1614
+ case $.APOSTROPHE: {
1615
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1616
+ break;
1617
+ }
1618
+ case $.AMPERSAND: {
1619
+ this._startCharacterReference();
1620
+ break;
1621
+ }
1622
+ case $.NULL: {
1623
+ this._err(ERR.unexpectedNullCharacter);
1624
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1625
+ break;
1626
+ }
1627
+ case $.EOF: {
1628
+ this._err(ERR.eofInTag);
1629
+ this._emitEOFToken();
1630
+ break;
1631
+ }
1632
+ default: {
1633
+ this.currentAttr.value += String.fromCodePoint(cp);
1634
+ }
1635
+ }
1636
+ }
1637
+ // Attribute value (unquoted) state
1638
+ //------------------------------------------------------------------
1639
+ _stateAttributeValueUnquoted(cp) {
1640
+ switch (cp) {
1641
+ case $.SPACE:
1642
+ case $.LINE_FEED:
1643
+ case $.TABULATION:
1644
+ case $.FORM_FEED: {
1645
+ this._leaveAttrValue();
1646
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1647
+ break;
1648
+ }
1649
+ case $.AMPERSAND: {
1650
+ this._startCharacterReference();
1651
+ break;
1652
+ }
1653
+ case $.GREATER_THAN_SIGN: {
1654
+ this._leaveAttrValue();
1655
+ this.state = State.DATA;
1656
+ this.emitCurrentTagToken();
1657
+ break;
1658
+ }
1659
+ case $.NULL: {
1660
+ this._err(ERR.unexpectedNullCharacter);
1661
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1662
+ break;
1663
+ }
1664
+ case $.QUOTATION_MARK:
1665
+ case $.APOSTROPHE:
1666
+ case $.LESS_THAN_SIGN:
1667
+ case $.EQUALS_SIGN:
1668
+ case $.GRAVE_ACCENT: {
1669
+ this._err(ERR.unexpectedCharacterInUnquotedAttributeValue);
1670
+ this.currentAttr.value += String.fromCodePoint(cp);
1671
+ break;
1672
+ }
1673
+ case $.EOF: {
1674
+ this._err(ERR.eofInTag);
1675
+ this._emitEOFToken();
1676
+ break;
1677
+ }
1678
+ default: {
1679
+ this.currentAttr.value += String.fromCodePoint(cp);
1680
+ }
1681
+ }
1682
+ }
1683
+ // After attribute value (quoted) state
1684
+ //------------------------------------------------------------------
1685
+ _stateAfterAttributeValueQuoted(cp) {
1686
+ switch (cp) {
1687
+ case $.SPACE:
1688
+ case $.LINE_FEED:
1689
+ case $.TABULATION:
1690
+ case $.FORM_FEED: {
1691
+ this._leaveAttrValue();
1692
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1693
+ break;
1694
+ }
1695
+ case $.SOLIDUS: {
1696
+ this._leaveAttrValue();
1697
+ this.state = State.SELF_CLOSING_START_TAG;
1698
+ break;
1699
+ }
1700
+ case $.GREATER_THAN_SIGN: {
1701
+ this._leaveAttrValue();
1702
+ this.state = State.DATA;
1703
+ this.emitCurrentTagToken();
1704
+ break;
1705
+ }
1706
+ case $.EOF: {
1707
+ this._err(ERR.eofInTag);
1708
+ this._emitEOFToken();
1709
+ break;
1710
+ }
1711
+ default: {
1712
+ this._err(ERR.missingWhitespaceBetweenAttributes);
1713
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1714
+ this._stateBeforeAttributeName(cp);
1715
+ }
1716
+ }
1717
+ }
1718
+ // Self-closing start tag state
1719
+ //------------------------------------------------------------------
1720
+ _stateSelfClosingStartTag(cp) {
1721
+ switch (cp) {
1722
+ case $.GREATER_THAN_SIGN: {
1723
+ const token = this.currentToken;
1724
+ token.selfClosing = true;
1725
+ this.state = State.DATA;
1726
+ this.emitCurrentTagToken();
1727
+ break;
1728
+ }
1729
+ case $.EOF: {
1730
+ this._err(ERR.eofInTag);
1731
+ this._emitEOFToken();
1732
+ break;
1733
+ }
1734
+ default: {
1735
+ this._err(ERR.unexpectedSolidusInTag);
1736
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1737
+ this._stateBeforeAttributeName(cp);
1738
+ }
1739
+ }
1740
+ }
1741
+ // Bogus comment state
1742
+ //------------------------------------------------------------------
1743
+ _stateBogusComment(cp) {
1744
+ const token = this.currentToken;
1745
+ switch (cp) {
1746
+ case $.GREATER_THAN_SIGN: {
1747
+ this.state = State.DATA;
1748
+ this.emitCurrentComment(token);
1749
+ break;
1750
+ }
1751
+ case $.EOF: {
1752
+ this.emitCurrentComment(token);
1753
+ this._emitEOFToken();
1754
+ break;
1755
+ }
1756
+ case $.NULL: {
1757
+ this._err(ERR.unexpectedNullCharacter);
1758
+ token.data += REPLACEMENT_CHARACTER;
1759
+ break;
1760
+ }
1761
+ default: {
1762
+ token.data += String.fromCodePoint(cp);
1763
+ }
1764
+ }
1765
+ }
1766
+ // Markup declaration open state
1767
+ //------------------------------------------------------------------
1768
+ _stateMarkupDeclarationOpen(cp) {
1769
+ if (this._consumeSequenceIfMatch($$.DASH_DASH, true)) {
1770
+ this._createCommentToken($$.DASH_DASH.length + 1);
1771
+ this.state = State.COMMENT_START;
1772
+ }
1773
+ else if (this._consumeSequenceIfMatch($$.DOCTYPE, false)) {
1774
+ // NOTE: Doctypes tokens are created without fixed offsets. We keep track of the moment a doctype *might* start here.
1775
+ this.currentLocation = this.getCurrentLocation($$.DOCTYPE.length + 1);
1776
+ this.state = State.DOCTYPE;
1777
+ }
1778
+ else if (this._consumeSequenceIfMatch($$.CDATA_START, true)) {
1779
+ if (this.inForeignNode) {
1780
+ this.state = State.CDATA_SECTION;
1781
+ }
1782
+ else {
1783
+ this._err(ERR.cdataInHtmlContent);
1784
+ this._createCommentToken($$.CDATA_START.length + 1);
1785
+ this.currentToken.data = '[CDATA[';
1786
+ this.state = State.BOGUS_COMMENT;
1787
+ }
1788
+ }
1789
+ //NOTE: Sequence lookups can be abrupted by hibernation. In that case, lookup
1790
+ //results are no longer valid and we will need to start over.
1791
+ else if (!this._ensureHibernation()) {
1792
+ this._err(ERR.incorrectlyOpenedComment);
1793
+ this._createCommentToken(2);
1794
+ this.state = State.BOGUS_COMMENT;
1795
+ this._stateBogusComment(cp);
1796
+ }
1797
+ }
1798
+ // Comment start state
1799
+ //------------------------------------------------------------------
1800
+ _stateCommentStart(cp) {
1801
+ switch (cp) {
1802
+ case $.HYPHEN_MINUS: {
1803
+ this.state = State.COMMENT_START_DASH;
1804
+ break;
1805
+ }
1806
+ case $.GREATER_THAN_SIGN: {
1807
+ this._err(ERR.abruptClosingOfEmptyComment);
1808
+ this.state = State.DATA;
1809
+ const token = this.currentToken;
1810
+ this.emitCurrentComment(token);
1811
+ break;
1812
+ }
1813
+ default: {
1814
+ this.state = State.COMMENT;
1815
+ this._stateComment(cp);
1816
+ }
1817
+ }
1818
+ }
1819
+ // Comment start dash state
1820
+ //------------------------------------------------------------------
1821
+ _stateCommentStartDash(cp) {
1822
+ const token = this.currentToken;
1823
+ switch (cp) {
1824
+ case $.HYPHEN_MINUS: {
1825
+ this.state = State.COMMENT_END;
1826
+ break;
1827
+ }
1828
+ case $.GREATER_THAN_SIGN: {
1829
+ this._err(ERR.abruptClosingOfEmptyComment);
1830
+ this.state = State.DATA;
1831
+ this.emitCurrentComment(token);
1832
+ break;
1833
+ }
1834
+ case $.EOF: {
1835
+ this._err(ERR.eofInComment);
1836
+ this.emitCurrentComment(token);
1837
+ this._emitEOFToken();
1838
+ break;
1839
+ }
1840
+ default: {
1841
+ token.data += '-';
1842
+ this.state = State.COMMENT;
1843
+ this._stateComment(cp);
1844
+ }
1845
+ }
1846
+ }
1847
+ // Comment state
1848
+ //------------------------------------------------------------------
1849
+ _stateComment(cp) {
1850
+ const token = this.currentToken;
1851
+ switch (cp) {
1852
+ case $.HYPHEN_MINUS: {
1853
+ this.state = State.COMMENT_END_DASH;
1854
+ break;
1855
+ }
1856
+ case $.LESS_THAN_SIGN: {
1857
+ token.data += '<';
1858
+ this.state = State.COMMENT_LESS_THAN_SIGN;
1859
+ break;
1860
+ }
1861
+ case $.NULL: {
1862
+ this._err(ERR.unexpectedNullCharacter);
1863
+ token.data += REPLACEMENT_CHARACTER;
1864
+ break;
1865
+ }
1866
+ case $.EOF: {
1867
+ this._err(ERR.eofInComment);
1868
+ this.emitCurrentComment(token);
1869
+ this._emitEOFToken();
1870
+ break;
1871
+ }
1872
+ default: {
1873
+ token.data += String.fromCodePoint(cp);
1874
+ }
1875
+ }
1876
+ }
1877
+ // Comment less-than sign state
1878
+ //------------------------------------------------------------------
1879
+ _stateCommentLessThanSign(cp) {
1880
+ const token = this.currentToken;
1881
+ switch (cp) {
1882
+ case $.EXCLAMATION_MARK: {
1883
+ token.data += '!';
1884
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG;
1885
+ break;
1886
+ }
1887
+ case $.LESS_THAN_SIGN: {
1888
+ token.data += '<';
1889
+ break;
1890
+ }
1891
+ default: {
1892
+ this.state = State.COMMENT;
1893
+ this._stateComment(cp);
1894
+ }
1895
+ }
1896
+ }
1897
+ // Comment less-than sign bang state
1898
+ //------------------------------------------------------------------
1899
+ _stateCommentLessThanSignBang(cp) {
1900
+ if (cp === $.HYPHEN_MINUS) {
1901
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH;
1902
+ }
1903
+ else {
1904
+ this.state = State.COMMENT;
1905
+ this._stateComment(cp);
1906
+ }
1907
+ }
1908
+ // Comment less-than sign bang dash state
1909
+ //------------------------------------------------------------------
1910
+ _stateCommentLessThanSignBangDash(cp) {
1911
+ if (cp === $.HYPHEN_MINUS) {
1912
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH;
1913
+ }
1914
+ else {
1915
+ this.state = State.COMMENT_END_DASH;
1916
+ this._stateCommentEndDash(cp);
1917
+ }
1918
+ }
1919
+ // Comment less-than sign bang dash dash state
1920
+ //------------------------------------------------------------------
1921
+ _stateCommentLessThanSignBangDashDash(cp) {
1922
+ if (cp !== $.GREATER_THAN_SIGN && cp !== $.EOF) {
1923
+ this._err(ERR.nestedComment);
1924
+ }
1925
+ this.state = State.COMMENT_END;
1926
+ this._stateCommentEnd(cp);
1927
+ }
1928
+ // Comment end dash state
1929
+ //------------------------------------------------------------------
1930
+ _stateCommentEndDash(cp) {
1931
+ const token = this.currentToken;
1932
+ switch (cp) {
1933
+ case $.HYPHEN_MINUS: {
1934
+ this.state = State.COMMENT_END;
1935
+ break;
1936
+ }
1937
+ case $.EOF: {
1938
+ this._err(ERR.eofInComment);
1939
+ this.emitCurrentComment(token);
1940
+ this._emitEOFToken();
1941
+ break;
1942
+ }
1943
+ default: {
1944
+ token.data += '-';
1945
+ this.state = State.COMMENT;
1946
+ this._stateComment(cp);
1947
+ }
1948
+ }
1949
+ }
1950
+ // Comment end state
1951
+ //------------------------------------------------------------------
1952
+ _stateCommentEnd(cp) {
1953
+ const token = this.currentToken;
1954
+ switch (cp) {
1955
+ case $.GREATER_THAN_SIGN: {
1956
+ this.state = State.DATA;
1957
+ this.emitCurrentComment(token);
1958
+ break;
1959
+ }
1960
+ case $.EXCLAMATION_MARK: {
1961
+ this.state = State.COMMENT_END_BANG;
1962
+ break;
1963
+ }
1964
+ case $.HYPHEN_MINUS: {
1965
+ token.data += '-';
1966
+ break;
1967
+ }
1968
+ case $.EOF: {
1969
+ this._err(ERR.eofInComment);
1970
+ this.emitCurrentComment(token);
1971
+ this._emitEOFToken();
1972
+ break;
1973
+ }
1974
+ default: {
1975
+ token.data += '--';
1976
+ this.state = State.COMMENT;
1977
+ this._stateComment(cp);
1978
+ }
1979
+ }
1980
+ }
1981
+ // Comment end bang state
1982
+ //------------------------------------------------------------------
1983
+ _stateCommentEndBang(cp) {
1984
+ const token = this.currentToken;
1985
+ switch (cp) {
1986
+ case $.HYPHEN_MINUS: {
1987
+ token.data += '--!';
1988
+ this.state = State.COMMENT_END_DASH;
1989
+ break;
1990
+ }
1991
+ case $.GREATER_THAN_SIGN: {
1992
+ this._err(ERR.incorrectlyClosedComment);
1993
+ this.state = State.DATA;
1994
+ this.emitCurrentComment(token);
1995
+ break;
1996
+ }
1997
+ case $.EOF: {
1998
+ this._err(ERR.eofInComment);
1999
+ this.emitCurrentComment(token);
2000
+ this._emitEOFToken();
2001
+ break;
2002
+ }
2003
+ default: {
2004
+ token.data += '--!';
2005
+ this.state = State.COMMENT;
2006
+ this._stateComment(cp);
2007
+ }
2008
+ }
2009
+ }
2010
+ // DOCTYPE state
2011
+ //------------------------------------------------------------------
2012
+ _stateDoctype(cp) {
2013
+ switch (cp) {
2014
+ case $.SPACE:
2015
+ case $.LINE_FEED:
2016
+ case $.TABULATION:
2017
+ case $.FORM_FEED: {
2018
+ this.state = State.BEFORE_DOCTYPE_NAME;
2019
+ break;
2020
+ }
2021
+ case $.GREATER_THAN_SIGN: {
2022
+ this.state = State.BEFORE_DOCTYPE_NAME;
2023
+ this._stateBeforeDoctypeName(cp);
2024
+ break;
2025
+ }
2026
+ case $.EOF: {
2027
+ this._err(ERR.eofInDoctype);
2028
+ this._createDoctypeToken(null);
2029
+ const token = this.currentToken;
2030
+ token.forceQuirks = true;
2031
+ this.emitCurrentDoctype(token);
2032
+ this._emitEOFToken();
2033
+ break;
2034
+ }
2035
+ default: {
2036
+ this._err(ERR.missingWhitespaceBeforeDoctypeName);
2037
+ this.state = State.BEFORE_DOCTYPE_NAME;
2038
+ this._stateBeforeDoctypeName(cp);
2039
+ }
2040
+ }
2041
+ }
2042
+ // Before DOCTYPE name state
2043
+ //------------------------------------------------------------------
2044
+ _stateBeforeDoctypeName(cp) {
2045
+ if (isAsciiUpper(cp)) {
2046
+ this._createDoctypeToken(String.fromCharCode(toAsciiLower(cp)));
2047
+ this.state = State.DOCTYPE_NAME;
2048
+ }
2049
+ else
2050
+ switch (cp) {
2051
+ case $.SPACE:
2052
+ case $.LINE_FEED:
2053
+ case $.TABULATION:
2054
+ case $.FORM_FEED: {
2055
+ // Ignore whitespace
2056
+ break;
2057
+ }
2058
+ case $.NULL: {
2059
+ this._err(ERR.unexpectedNullCharacter);
2060
+ this._createDoctypeToken(REPLACEMENT_CHARACTER);
2061
+ this.state = State.DOCTYPE_NAME;
2062
+ break;
2063
+ }
2064
+ case $.GREATER_THAN_SIGN: {
2065
+ this._err(ERR.missingDoctypeName);
2066
+ this._createDoctypeToken(null);
2067
+ const token = this.currentToken;
2068
+ token.forceQuirks = true;
2069
+ this.emitCurrentDoctype(token);
2070
+ this.state = State.DATA;
2071
+ break;
2072
+ }
2073
+ case $.EOF: {
2074
+ this._err(ERR.eofInDoctype);
2075
+ this._createDoctypeToken(null);
2076
+ const token = this.currentToken;
2077
+ token.forceQuirks = true;
2078
+ this.emitCurrentDoctype(token);
2079
+ this._emitEOFToken();
2080
+ break;
2081
+ }
2082
+ default: {
2083
+ this._createDoctypeToken(String.fromCodePoint(cp));
2084
+ this.state = State.DOCTYPE_NAME;
2085
+ }
2086
+ }
2087
+ }
2088
+ // DOCTYPE name state
2089
+ //------------------------------------------------------------------
2090
+ _stateDoctypeName(cp) {
2091
+ const token = this.currentToken;
2092
+ switch (cp) {
2093
+ case $.SPACE:
2094
+ case $.LINE_FEED:
2095
+ case $.TABULATION:
2096
+ case $.FORM_FEED: {
2097
+ this.state = State.AFTER_DOCTYPE_NAME;
2098
+ break;
2099
+ }
2100
+ case $.GREATER_THAN_SIGN: {
2101
+ this.state = State.DATA;
2102
+ this.emitCurrentDoctype(token);
2103
+ break;
2104
+ }
2105
+ case $.NULL: {
2106
+ this._err(ERR.unexpectedNullCharacter);
2107
+ token.name += REPLACEMENT_CHARACTER;
2108
+ break;
2109
+ }
2110
+ case $.EOF: {
2111
+ this._err(ERR.eofInDoctype);
2112
+ token.forceQuirks = true;
2113
+ this.emitCurrentDoctype(token);
2114
+ this._emitEOFToken();
2115
+ break;
2116
+ }
2117
+ default: {
2118
+ token.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
2119
+ }
2120
+ }
2121
+ }
2122
+ // After DOCTYPE name state
2123
+ //------------------------------------------------------------------
2124
+ _stateAfterDoctypeName(cp) {
2125
+ const token = this.currentToken;
2126
+ switch (cp) {
2127
+ case $.SPACE:
2128
+ case $.LINE_FEED:
2129
+ case $.TABULATION:
2130
+ case $.FORM_FEED: {
2131
+ // Ignore whitespace
2132
+ break;
2133
+ }
2134
+ case $.GREATER_THAN_SIGN: {
2135
+ this.state = State.DATA;
2136
+ this.emitCurrentDoctype(token);
2137
+ break;
2138
+ }
2139
+ case $.EOF: {
2140
+ this._err(ERR.eofInDoctype);
2141
+ token.forceQuirks = true;
2142
+ this.emitCurrentDoctype(token);
2143
+ this._emitEOFToken();
2144
+ break;
2145
+ }
2146
+ default: {
2147
+ if (this._consumeSequenceIfMatch($$.PUBLIC, false)) {
2148
+ this.state = State.AFTER_DOCTYPE_PUBLIC_KEYWORD;
2149
+ }
2150
+ else if (this._consumeSequenceIfMatch($$.SYSTEM, false)) {
2151
+ this.state = State.AFTER_DOCTYPE_SYSTEM_KEYWORD;
2152
+ }
2153
+ //NOTE: sequence lookup can be abrupted by hibernation. In that case lookup
2154
+ //results are no longer valid and we will need to start over.
2155
+ else if (!this._ensureHibernation()) {
2156
+ this._err(ERR.invalidCharacterSequenceAfterDoctypeName);
2157
+ token.forceQuirks = true;
2158
+ this.state = State.BOGUS_DOCTYPE;
2159
+ this._stateBogusDoctype(cp);
2160
+ }
2161
+ }
2162
+ }
2163
+ }
2164
+ // After DOCTYPE public keyword state
2165
+ //------------------------------------------------------------------
2166
+ _stateAfterDoctypePublicKeyword(cp) {
2167
+ const token = this.currentToken;
2168
+ switch (cp) {
2169
+ case $.SPACE:
2170
+ case $.LINE_FEED:
2171
+ case $.TABULATION:
2172
+ case $.FORM_FEED: {
2173
+ this.state = State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER;
2174
+ break;
2175
+ }
2176
+ case $.QUOTATION_MARK: {
2177
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2178
+ token.publicId = '';
2179
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2180
+ break;
2181
+ }
2182
+ case $.APOSTROPHE: {
2183
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2184
+ token.publicId = '';
2185
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2186
+ break;
2187
+ }
2188
+ case $.GREATER_THAN_SIGN: {
2189
+ this._err(ERR.missingDoctypePublicIdentifier);
2190
+ token.forceQuirks = true;
2191
+ this.state = State.DATA;
2192
+ this.emitCurrentDoctype(token);
2193
+ break;
2194
+ }
2195
+ case $.EOF: {
2196
+ this._err(ERR.eofInDoctype);
2197
+ token.forceQuirks = true;
2198
+ this.emitCurrentDoctype(token);
2199
+ this._emitEOFToken();
2200
+ break;
2201
+ }
2202
+ default: {
2203
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2204
+ token.forceQuirks = true;
2205
+ this.state = State.BOGUS_DOCTYPE;
2206
+ this._stateBogusDoctype(cp);
2207
+ }
2208
+ }
2209
+ }
2210
+ // Before DOCTYPE public identifier state
2211
+ //------------------------------------------------------------------
2212
+ _stateBeforeDoctypePublicIdentifier(cp) {
2213
+ const token = this.currentToken;
2214
+ switch (cp) {
2215
+ case $.SPACE:
2216
+ case $.LINE_FEED:
2217
+ case $.TABULATION:
2218
+ case $.FORM_FEED: {
2219
+ // Ignore whitespace
2220
+ break;
2221
+ }
2222
+ case $.QUOTATION_MARK: {
2223
+ token.publicId = '';
2224
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2225
+ break;
2226
+ }
2227
+ case $.APOSTROPHE: {
2228
+ token.publicId = '';
2229
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2230
+ break;
2231
+ }
2232
+ case $.GREATER_THAN_SIGN: {
2233
+ this._err(ERR.missingDoctypePublicIdentifier);
2234
+ token.forceQuirks = true;
2235
+ this.state = State.DATA;
2236
+ this.emitCurrentDoctype(token);
2237
+ break;
2238
+ }
2239
+ case $.EOF: {
2240
+ this._err(ERR.eofInDoctype);
2241
+ token.forceQuirks = true;
2242
+ this.emitCurrentDoctype(token);
2243
+ this._emitEOFToken();
2244
+ break;
2245
+ }
2246
+ default: {
2247
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2248
+ token.forceQuirks = true;
2249
+ this.state = State.BOGUS_DOCTYPE;
2250
+ this._stateBogusDoctype(cp);
2251
+ }
2252
+ }
2253
+ }
2254
+ // DOCTYPE public identifier (double-quoted) state
2255
+ //------------------------------------------------------------------
2256
+ _stateDoctypePublicIdentifierDoubleQuoted(cp) {
2257
+ const token = this.currentToken;
2258
+ switch (cp) {
2259
+ case $.QUOTATION_MARK: {
2260
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2261
+ break;
2262
+ }
2263
+ case $.NULL: {
2264
+ this._err(ERR.unexpectedNullCharacter);
2265
+ token.publicId += REPLACEMENT_CHARACTER;
2266
+ break;
2267
+ }
2268
+ case $.GREATER_THAN_SIGN: {
2269
+ this._err(ERR.abruptDoctypePublicIdentifier);
2270
+ token.forceQuirks = true;
2271
+ this.emitCurrentDoctype(token);
2272
+ this.state = State.DATA;
2273
+ break;
2274
+ }
2275
+ case $.EOF: {
2276
+ this._err(ERR.eofInDoctype);
2277
+ token.forceQuirks = true;
2278
+ this.emitCurrentDoctype(token);
2279
+ this._emitEOFToken();
2280
+ break;
2281
+ }
2282
+ default: {
2283
+ token.publicId += String.fromCodePoint(cp);
2284
+ }
2285
+ }
2286
+ }
2287
+ // DOCTYPE public identifier (single-quoted) state
2288
+ //------------------------------------------------------------------
2289
+ _stateDoctypePublicIdentifierSingleQuoted(cp) {
2290
+ const token = this.currentToken;
2291
+ switch (cp) {
2292
+ case $.APOSTROPHE: {
2293
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2294
+ break;
2295
+ }
2296
+ case $.NULL: {
2297
+ this._err(ERR.unexpectedNullCharacter);
2298
+ token.publicId += REPLACEMENT_CHARACTER;
2299
+ break;
2300
+ }
2301
+ case $.GREATER_THAN_SIGN: {
2302
+ this._err(ERR.abruptDoctypePublicIdentifier);
2303
+ token.forceQuirks = true;
2304
+ this.emitCurrentDoctype(token);
2305
+ this.state = State.DATA;
2306
+ break;
2307
+ }
2308
+ case $.EOF: {
2309
+ this._err(ERR.eofInDoctype);
2310
+ token.forceQuirks = true;
2311
+ this.emitCurrentDoctype(token);
2312
+ this._emitEOFToken();
2313
+ break;
2314
+ }
2315
+ default: {
2316
+ token.publicId += String.fromCodePoint(cp);
2317
+ }
2318
+ }
2319
+ }
2320
+ // After DOCTYPE public identifier state
2321
+ //------------------------------------------------------------------
2322
+ _stateAfterDoctypePublicIdentifier(cp) {
2323
+ const token = this.currentToken;
2324
+ switch (cp) {
2325
+ case $.SPACE:
2326
+ case $.LINE_FEED:
2327
+ case $.TABULATION:
2328
+ case $.FORM_FEED: {
2329
+ this.state = State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS;
2330
+ break;
2331
+ }
2332
+ case $.GREATER_THAN_SIGN: {
2333
+ this.state = State.DATA;
2334
+ this.emitCurrentDoctype(token);
2335
+ break;
2336
+ }
2337
+ case $.QUOTATION_MARK: {
2338
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2339
+ token.systemId = '';
2340
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2341
+ break;
2342
+ }
2343
+ case $.APOSTROPHE: {
2344
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2345
+ token.systemId = '';
2346
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2347
+ break;
2348
+ }
2349
+ case $.EOF: {
2350
+ this._err(ERR.eofInDoctype);
2351
+ token.forceQuirks = true;
2352
+ this.emitCurrentDoctype(token);
2353
+ this._emitEOFToken();
2354
+ break;
2355
+ }
2356
+ default: {
2357
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2358
+ token.forceQuirks = true;
2359
+ this.state = State.BOGUS_DOCTYPE;
2360
+ this._stateBogusDoctype(cp);
2361
+ }
2362
+ }
2363
+ }
2364
+ // Between DOCTYPE public and system identifiers state
2365
+ //------------------------------------------------------------------
2366
+ _stateBetweenDoctypePublicAndSystemIdentifiers(cp) {
2367
+ const token = this.currentToken;
2368
+ switch (cp) {
2369
+ case $.SPACE:
2370
+ case $.LINE_FEED:
2371
+ case $.TABULATION:
2372
+ case $.FORM_FEED: {
2373
+ // Ignore whitespace
2374
+ break;
2375
+ }
2376
+ case $.GREATER_THAN_SIGN: {
2377
+ this.emitCurrentDoctype(token);
2378
+ this.state = State.DATA;
2379
+ break;
2380
+ }
2381
+ case $.QUOTATION_MARK: {
2382
+ token.systemId = '';
2383
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2384
+ break;
2385
+ }
2386
+ case $.APOSTROPHE: {
2387
+ token.systemId = '';
2388
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2389
+ break;
2390
+ }
2391
+ case $.EOF: {
2392
+ this._err(ERR.eofInDoctype);
2393
+ token.forceQuirks = true;
2394
+ this.emitCurrentDoctype(token);
2395
+ this._emitEOFToken();
2396
+ break;
2397
+ }
2398
+ default: {
2399
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2400
+ token.forceQuirks = true;
2401
+ this.state = State.BOGUS_DOCTYPE;
2402
+ this._stateBogusDoctype(cp);
2403
+ }
2404
+ }
2405
+ }
2406
+ // After DOCTYPE system keyword state
2407
+ //------------------------------------------------------------------
2408
+ _stateAfterDoctypeSystemKeyword(cp) {
2409
+ const token = this.currentToken;
2410
+ switch (cp) {
2411
+ case $.SPACE:
2412
+ case $.LINE_FEED:
2413
+ case $.TABULATION:
2414
+ case $.FORM_FEED: {
2415
+ this.state = State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER;
2416
+ break;
2417
+ }
2418
+ case $.QUOTATION_MARK: {
2419
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2420
+ token.systemId = '';
2421
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2422
+ break;
2423
+ }
2424
+ case $.APOSTROPHE: {
2425
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2426
+ token.systemId = '';
2427
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2428
+ break;
2429
+ }
2430
+ case $.GREATER_THAN_SIGN: {
2431
+ this._err(ERR.missingDoctypeSystemIdentifier);
2432
+ token.forceQuirks = true;
2433
+ this.state = State.DATA;
2434
+ this.emitCurrentDoctype(token);
2435
+ break;
2436
+ }
2437
+ case $.EOF: {
2438
+ this._err(ERR.eofInDoctype);
2439
+ token.forceQuirks = true;
2440
+ this.emitCurrentDoctype(token);
2441
+ this._emitEOFToken();
2442
+ break;
2443
+ }
2444
+ default: {
2445
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2446
+ token.forceQuirks = true;
2447
+ this.state = State.BOGUS_DOCTYPE;
2448
+ this._stateBogusDoctype(cp);
2449
+ }
2450
+ }
2451
+ }
2452
+ // Before DOCTYPE system identifier state
2453
+ //------------------------------------------------------------------
2454
+ _stateBeforeDoctypeSystemIdentifier(cp) {
2455
+ const token = this.currentToken;
2456
+ switch (cp) {
2457
+ case $.SPACE:
2458
+ case $.LINE_FEED:
2459
+ case $.TABULATION:
2460
+ case $.FORM_FEED: {
2461
+ // Ignore whitespace
2462
+ break;
2463
+ }
2464
+ case $.QUOTATION_MARK: {
2465
+ token.systemId = '';
2466
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2467
+ break;
2468
+ }
2469
+ case $.APOSTROPHE: {
2470
+ token.systemId = '';
2471
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2472
+ break;
2473
+ }
2474
+ case $.GREATER_THAN_SIGN: {
2475
+ this._err(ERR.missingDoctypeSystemIdentifier);
2476
+ token.forceQuirks = true;
2477
+ this.state = State.DATA;
2478
+ this.emitCurrentDoctype(token);
2479
+ break;
2480
+ }
2481
+ case $.EOF: {
2482
+ this._err(ERR.eofInDoctype);
2483
+ token.forceQuirks = true;
2484
+ this.emitCurrentDoctype(token);
2485
+ this._emitEOFToken();
2486
+ break;
2487
+ }
2488
+ default: {
2489
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2490
+ token.forceQuirks = true;
2491
+ this.state = State.BOGUS_DOCTYPE;
2492
+ this._stateBogusDoctype(cp);
2493
+ }
2494
+ }
2495
+ }
2496
+ // DOCTYPE system identifier (double-quoted) state
2497
+ //------------------------------------------------------------------
2498
+ _stateDoctypeSystemIdentifierDoubleQuoted(cp) {
2499
+ const token = this.currentToken;
2500
+ switch (cp) {
2501
+ case $.QUOTATION_MARK: {
2502
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2503
+ break;
2504
+ }
2505
+ case $.NULL: {
2506
+ this._err(ERR.unexpectedNullCharacter);
2507
+ token.systemId += REPLACEMENT_CHARACTER;
2508
+ break;
2509
+ }
2510
+ case $.GREATER_THAN_SIGN: {
2511
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2512
+ token.forceQuirks = true;
2513
+ this.emitCurrentDoctype(token);
2514
+ this.state = State.DATA;
2515
+ break;
2516
+ }
2517
+ case $.EOF: {
2518
+ this._err(ERR.eofInDoctype);
2519
+ token.forceQuirks = true;
2520
+ this.emitCurrentDoctype(token);
2521
+ this._emitEOFToken();
2522
+ break;
2523
+ }
2524
+ default: {
2525
+ token.systemId += String.fromCodePoint(cp);
2526
+ }
2527
+ }
2528
+ }
2529
+ // DOCTYPE system identifier (single-quoted) state
2530
+ //------------------------------------------------------------------
2531
+ _stateDoctypeSystemIdentifierSingleQuoted(cp) {
2532
+ const token = this.currentToken;
2533
+ switch (cp) {
2534
+ case $.APOSTROPHE: {
2535
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2536
+ break;
2537
+ }
2538
+ case $.NULL: {
2539
+ this._err(ERR.unexpectedNullCharacter);
2540
+ token.systemId += REPLACEMENT_CHARACTER;
2541
+ break;
2542
+ }
2543
+ case $.GREATER_THAN_SIGN: {
2544
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2545
+ token.forceQuirks = true;
2546
+ this.emitCurrentDoctype(token);
2547
+ this.state = State.DATA;
2548
+ break;
2549
+ }
2550
+ case $.EOF: {
2551
+ this._err(ERR.eofInDoctype);
2552
+ token.forceQuirks = true;
2553
+ this.emitCurrentDoctype(token);
2554
+ this._emitEOFToken();
2555
+ break;
2556
+ }
2557
+ default: {
2558
+ token.systemId += String.fromCodePoint(cp);
2559
+ }
2560
+ }
2561
+ }
2562
+ // After DOCTYPE system identifier state
2563
+ //------------------------------------------------------------------
2564
+ _stateAfterDoctypeSystemIdentifier(cp) {
2565
+ const token = this.currentToken;
2566
+ switch (cp) {
2567
+ case $.SPACE:
2568
+ case $.LINE_FEED:
2569
+ case $.TABULATION:
2570
+ case $.FORM_FEED: {
2571
+ // Ignore whitespace
2572
+ break;
2573
+ }
2574
+ case $.GREATER_THAN_SIGN: {
2575
+ this.emitCurrentDoctype(token);
2576
+ this.state = State.DATA;
2577
+ break;
2578
+ }
2579
+ case $.EOF: {
2580
+ this._err(ERR.eofInDoctype);
2581
+ token.forceQuirks = true;
2582
+ this.emitCurrentDoctype(token);
2583
+ this._emitEOFToken();
2584
+ break;
2585
+ }
2586
+ default: {
2587
+ this._err(ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
2588
+ this.state = State.BOGUS_DOCTYPE;
2589
+ this._stateBogusDoctype(cp);
2590
+ }
2591
+ }
2592
+ }
2593
+ // Bogus DOCTYPE state
2594
+ //------------------------------------------------------------------
2595
+ _stateBogusDoctype(cp) {
2596
+ const token = this.currentToken;
2597
+ switch (cp) {
2598
+ case $.GREATER_THAN_SIGN: {
2599
+ this.emitCurrentDoctype(token);
2600
+ this.state = State.DATA;
2601
+ break;
2602
+ }
2603
+ case $.NULL: {
2604
+ this._err(ERR.unexpectedNullCharacter);
2605
+ break;
2606
+ }
2607
+ case $.EOF: {
2608
+ this.emitCurrentDoctype(token);
2609
+ this._emitEOFToken();
2610
+ break;
2611
+ }
2612
+ default:
2613
+ // Do nothing
2614
+ }
2615
+ }
2616
+ // CDATA section state
2617
+ //------------------------------------------------------------------
2618
+ _stateCdataSection(cp) {
2619
+ switch (cp) {
2620
+ case $.RIGHT_SQUARE_BRACKET: {
2621
+ this.state = State.CDATA_SECTION_BRACKET;
2622
+ break;
2623
+ }
2624
+ case $.EOF: {
2625
+ this._err(ERR.eofInCdata);
2626
+ this._emitEOFToken();
2627
+ break;
2628
+ }
2629
+ default: {
2630
+ this._emitCodePoint(cp);
2631
+ }
2632
+ }
2633
+ }
2634
+ // CDATA section bracket state
2635
+ //------------------------------------------------------------------
2636
+ _stateCdataSectionBracket(cp) {
2637
+ if (cp === $.RIGHT_SQUARE_BRACKET) {
2638
+ this.state = State.CDATA_SECTION_END;
2639
+ }
2640
+ else {
2641
+ this._emitChars(']');
2642
+ this.state = State.CDATA_SECTION;
2643
+ this._stateCdataSection(cp);
2644
+ }
2645
+ }
2646
+ // CDATA section end state
2647
+ //------------------------------------------------------------------
2648
+ _stateCdataSectionEnd(cp) {
2649
+ switch (cp) {
2650
+ case $.GREATER_THAN_SIGN: {
2651
+ this.state = State.DATA;
2652
+ break;
2653
+ }
2654
+ case $.RIGHT_SQUARE_BRACKET: {
2655
+ this._emitChars(']');
2656
+ break;
2657
+ }
2658
+ default: {
2659
+ this._emitChars(']]');
2660
+ this.state = State.CDATA_SECTION;
2661
+ this._stateCdataSection(cp);
2662
+ }
2663
+ }
2664
+ }
2665
+ // Character reference state
2666
+ //------------------------------------------------------------------
2667
+ _stateCharacterReference() {
2668
+ let length = this.entityDecoder.write(this.preprocessor.html, this.preprocessor.pos);
2669
+ if (length < 0) {
2670
+ if (this.preprocessor.lastChunkWritten) {
2671
+ length = this.entityDecoder.end();
2672
+ }
2673
+ else {
2674
+ // Wait for the rest of the entity.
2675
+ this.active = false;
2676
+ // Mark the entire buffer as read.
2677
+ this.preprocessor.pos = this.preprocessor.html.length - 1;
2678
+ this.consumedAfterSnapshot = 0;
2679
+ this.preprocessor.endOfChunkHit = true;
2680
+ return;
2681
+ }
2682
+ }
2683
+ if (length === 0) {
2684
+ // This was not a valid entity. Go back to the beginning, and
2685
+ // figure out what to do.
2686
+ this.preprocessor.pos = this.entityStartPos;
2687
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2688
+ this.state =
2689
+ !this._isCharacterReferenceInAttribute() && isAsciiAlphaNumeric(this.preprocessor.peek(1))
2690
+ ? State.AMBIGUOUS_AMPERSAND
2691
+ : this.returnState;
2692
+ }
2693
+ else {
2694
+ // We successfully parsed an entity. Switch to the return state.
2695
+ this.state = this.returnState;
2696
+ }
2697
+ }
2698
+ // Ambiguos ampersand state
2699
+ //------------------------------------------------------------------
2700
+ _stateAmbiguousAmpersand(cp) {
2701
+ if (isAsciiAlphaNumeric(cp)) {
2702
+ this._flushCodePointConsumedAsCharacterReference(cp);
2703
+ }
2704
+ else {
2705
+ if (cp === $.SEMICOLON) {
2706
+ this._err(ERR.unknownNamedCharacterReference);
2707
+ }
2708
+ this.state = this.returnState;
2709
+ this._callState(cp);
2710
+ }
2711
+ }
2712
+ }