parse5 6.0.1 → 7.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (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 +32 -0
  18. package/dist/cjs/parser/formatting-element-list.js +110 -0
  19. package/dist/cjs/parser/index.d.ts +157 -0
  20. package/dist/cjs/parser/index.js +3049 -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 +250 -0
  26. package/dist/cjs/tokenizer/index.js +2921 -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 +92 -0
  30. package/dist/cjs/tree-adapters/default.js +185 -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 +32 -0
  48. package/dist/parser/formatting-element-list.js +106 -0
  49. package/dist/parser/index.d.ts +157 -0
  50. package/dist/parser/index.js +3054 -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 +250 -0
  56. package/dist/tokenizer/index.js +2917 -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 +92 -0
  60. package/dist/tree-adapters/default.js +182 -0
  61. package/dist/tree-adapters/interface.d.ts +250 -0
  62. package/dist/tree-adapters/interface.js +2 -0
  63. package/package.json +19 -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 +1,2917 @@
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["DECIMAL_CHARACTER_REFERENCE_START"] = 76] = "DECIMAL_CHARACTER_REFERENCE_START";
117
+ State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 77] = "HEXADEMICAL_CHARACTER_REFERENCE";
118
+ State[State["DECIMAL_CHARACTER_REFERENCE"] = 78] = "DECIMAL_CHARACTER_REFERENCE";
119
+ State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 79] = "NUMERIC_CHARACTER_REFERENCE_END";
120
+ })(State || (State = {}));
121
+ //Tokenizer initial states for different modes
122
+ export const TokenizerMode = {
123
+ DATA: State.DATA,
124
+ RCDATA: State.RCDATA,
125
+ RAWTEXT: State.RAWTEXT,
126
+ SCRIPT_DATA: State.SCRIPT_DATA,
127
+ PLAINTEXT: State.PLAINTEXT,
128
+ CDATA_SECTION: State.CDATA_SECTION,
129
+ };
130
+ //Utils
131
+ //OPTIMIZATION: these utility functions should not be moved out of this module. V8 Crankshaft will not inline
132
+ //this functions if they will be situated in another module due to context switch.
133
+ //Always perform inlining check before modifying this functions ('node --trace-inlining').
134
+ function isAsciiDigit(cp) {
135
+ return cp >= $.DIGIT_0 && cp <= $.DIGIT_9;
136
+ }
137
+ function isAsciiUpper(cp) {
138
+ return cp >= $.LATIN_CAPITAL_A && cp <= $.LATIN_CAPITAL_Z;
139
+ }
140
+ function isAsciiLower(cp) {
141
+ return cp >= $.LATIN_SMALL_A && cp <= $.LATIN_SMALL_Z;
142
+ }
143
+ function isAsciiLetter(cp) {
144
+ return isAsciiLower(cp) || isAsciiUpper(cp);
145
+ }
146
+ function isAsciiAlphaNumeric(cp) {
147
+ return isAsciiLetter(cp) || isAsciiDigit(cp);
148
+ }
149
+ function isAsciiUpperHexDigit(cp) {
150
+ return cp >= $.LATIN_CAPITAL_A && cp <= $.LATIN_CAPITAL_F;
151
+ }
152
+ function isAsciiLowerHexDigit(cp) {
153
+ return cp >= $.LATIN_SMALL_A && cp <= $.LATIN_SMALL_F;
154
+ }
155
+ function isAsciiHexDigit(cp) {
156
+ return isAsciiDigit(cp) || isAsciiUpperHexDigit(cp) || isAsciiLowerHexDigit(cp);
157
+ }
158
+ function toAsciiLower(cp) {
159
+ return cp + 32;
160
+ }
161
+ function isWhitespace(cp) {
162
+ return cp === $.SPACE || cp === $.LINE_FEED || cp === $.TABULATION || cp === $.FORM_FEED;
163
+ }
164
+ function isEntityInAttributeInvalidEnd(nextCp) {
165
+ return nextCp === $.EQUALS_SIGN || isAsciiAlphaNumeric(nextCp);
166
+ }
167
+ function isScriptDataDoubleEscapeSequenceEnd(cp) {
168
+ return isWhitespace(cp) || cp === $.SOLIDUS || cp === $.GREATER_THAN_SIGN;
169
+ }
170
+ //Tokenizer
171
+ export class Tokenizer {
172
+ constructor(options, handler) {
173
+ this.options = options;
174
+ this.handler = handler;
175
+ this.paused = false;
176
+ /** Ensures that the parsing loop isn't run multiple times at once. */
177
+ this.inLoop = false;
178
+ /**
179
+ * Indicates that the current adjusted node exists, is not an element in the HTML namespace,
180
+ * and that it is not an integration point for either MathML or HTML.
181
+ *
182
+ * @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
183
+ */
184
+ this.inForeignNode = false;
185
+ this.lastStartTagName = '';
186
+ this.active = false;
187
+ this.state = State.DATA;
188
+ this.returnState = State.DATA;
189
+ this.charRefCode = -1;
190
+ this.consumedAfterSnapshot = -1;
191
+ this.currentCharacterToken = null;
192
+ this.currentToken = null;
193
+ this.currentAttr = { name: '', value: '' };
194
+ this.preprocessor = new Preprocessor(handler);
195
+ this.currentLocation = this.getCurrentLocation(-1);
196
+ }
197
+ //Errors
198
+ _err(code) {
199
+ var _a, _b;
200
+ (_b = (_a = this.handler).onParseError) === null || _b === void 0 ? void 0 : _b.call(_a, this.preprocessor.getError(code));
201
+ }
202
+ // NOTE: `offset` may never run across line boundaries.
203
+ getCurrentLocation(offset) {
204
+ if (!this.options.sourceCodeLocationInfo) {
205
+ return null;
206
+ }
207
+ return {
208
+ startLine: this.preprocessor.line,
209
+ startCol: this.preprocessor.col - offset,
210
+ startOffset: this.preprocessor.offset - offset,
211
+ endLine: -1,
212
+ endCol: -1,
213
+ endOffset: -1,
214
+ };
215
+ }
216
+ _runParsingLoop() {
217
+ if (this.inLoop)
218
+ return;
219
+ this.inLoop = true;
220
+ while (this.active && !this.paused) {
221
+ this.consumedAfterSnapshot = 0;
222
+ const cp = this._consume();
223
+ if (!this._ensureHibernation()) {
224
+ this._callState(cp);
225
+ }
226
+ }
227
+ this.inLoop = false;
228
+ }
229
+ //API
230
+ pause() {
231
+ this.paused = true;
232
+ }
233
+ resume(writeCallback) {
234
+ if (!this.paused) {
235
+ throw new Error('Parser was already resumed');
236
+ }
237
+ this.paused = false;
238
+ // Necessary for synchronous resume.
239
+ if (this.inLoop)
240
+ return;
241
+ this._runParsingLoop();
242
+ if (!this.paused) {
243
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
244
+ }
245
+ }
246
+ write(chunk, isLastChunk, writeCallback) {
247
+ this.active = true;
248
+ this.preprocessor.write(chunk, isLastChunk);
249
+ this._runParsingLoop();
250
+ if (!this.paused) {
251
+ writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
252
+ }
253
+ }
254
+ insertHtmlAtCurrentPos(chunk) {
255
+ this.active = true;
256
+ this.preprocessor.insertHtmlAtCurrentPos(chunk);
257
+ this._runParsingLoop();
258
+ }
259
+ //Hibernation
260
+ _ensureHibernation() {
261
+ if (this.preprocessor.endOfChunkHit) {
262
+ this._unconsume(this.consumedAfterSnapshot);
263
+ this.active = false;
264
+ return true;
265
+ }
266
+ return false;
267
+ }
268
+ //Consumption
269
+ _consume() {
270
+ this.consumedAfterSnapshot++;
271
+ return this.preprocessor.advance();
272
+ }
273
+ _unconsume(count) {
274
+ this.consumedAfterSnapshot -= count;
275
+ this.preprocessor.retreat(count);
276
+ }
277
+ _reconsumeInState(state) {
278
+ this.state = state;
279
+ this._unconsume(1);
280
+ }
281
+ _advanceBy(count) {
282
+ this.consumedAfterSnapshot += count;
283
+ for (let i = 0; i < count; i++) {
284
+ this.preprocessor.advance();
285
+ }
286
+ }
287
+ _consumeSequenceIfMatch(pattern, caseSensitive) {
288
+ if (this.preprocessor.startsWith(pattern, caseSensitive)) {
289
+ // We will already have consumed one character before calling this method.
290
+ this._advanceBy(pattern.length - 1);
291
+ return true;
292
+ }
293
+ return false;
294
+ }
295
+ //Token creation
296
+ _createStartTagToken() {
297
+ this.currentToken = {
298
+ type: TokenType.START_TAG,
299
+ tagName: '',
300
+ tagID: TAG_ID.UNKNOWN,
301
+ selfClosing: false,
302
+ ackSelfClosing: false,
303
+ attrs: [],
304
+ location: this.getCurrentLocation(1),
305
+ };
306
+ }
307
+ _createEndTagToken() {
308
+ this.currentToken = {
309
+ type: TokenType.END_TAG,
310
+ tagName: '',
311
+ tagID: TAG_ID.UNKNOWN,
312
+ selfClosing: false,
313
+ ackSelfClosing: false,
314
+ attrs: [],
315
+ location: this.getCurrentLocation(2),
316
+ };
317
+ }
318
+ _createCommentToken(offset) {
319
+ this.currentToken = {
320
+ type: TokenType.COMMENT,
321
+ data: '',
322
+ location: this.getCurrentLocation(offset),
323
+ };
324
+ }
325
+ _createDoctypeToken(initialName) {
326
+ this.currentToken = {
327
+ type: TokenType.DOCTYPE,
328
+ name: initialName,
329
+ forceQuirks: false,
330
+ publicId: null,
331
+ systemId: null,
332
+ location: this.currentLocation,
333
+ };
334
+ }
335
+ _createCharacterToken(type, chars) {
336
+ this.currentCharacterToken = {
337
+ type,
338
+ chars,
339
+ location: this.currentLocation,
340
+ };
341
+ }
342
+ //Tag attributes
343
+ _createAttr(attrNameFirstCh) {
344
+ this.currentAttr = {
345
+ name: attrNameFirstCh,
346
+ value: '',
347
+ };
348
+ this.currentLocation = this.getCurrentLocation(0);
349
+ }
350
+ _leaveAttrName() {
351
+ var _a;
352
+ var _b;
353
+ const token = this.currentToken;
354
+ if (getTokenAttr(token, this.currentAttr.name) === null) {
355
+ token.attrs.push(this.currentAttr);
356
+ if (token.location && this.currentLocation) {
357
+ const attrLocations = ((_a = (_b = token.location).attrs) !== null && _a !== void 0 ? _a : (_b.attrs = Object.create(null)));
358
+ attrLocations[this.currentAttr.name] = this.currentLocation;
359
+ // Set end location
360
+ this._leaveAttrValue();
361
+ }
362
+ }
363
+ else {
364
+ this._err(ERR.duplicateAttribute);
365
+ }
366
+ }
367
+ _leaveAttrValue() {
368
+ if (this.currentLocation) {
369
+ this.currentLocation.endLine = this.preprocessor.line;
370
+ this.currentLocation.endCol = this.preprocessor.col;
371
+ this.currentLocation.endOffset = this.preprocessor.offset;
372
+ }
373
+ }
374
+ //Token emission
375
+ prepareToken(ct) {
376
+ this._emitCurrentCharacterToken(ct.location);
377
+ this.currentToken = null;
378
+ if (ct.location) {
379
+ ct.location.endLine = this.preprocessor.line;
380
+ ct.location.endCol = this.preprocessor.col + 1;
381
+ ct.location.endOffset = this.preprocessor.offset + 1;
382
+ }
383
+ this.currentLocation = this.getCurrentLocation(-1);
384
+ }
385
+ emitCurrentTagToken() {
386
+ const ct = this.currentToken;
387
+ this.prepareToken(ct);
388
+ ct.tagID = getTagID(ct.tagName);
389
+ if (ct.type === TokenType.START_TAG) {
390
+ this.lastStartTagName = ct.tagName;
391
+ this.handler.onStartTag(ct);
392
+ }
393
+ else {
394
+ if (ct.attrs.length > 0) {
395
+ this._err(ERR.endTagWithAttributes);
396
+ }
397
+ if (ct.selfClosing) {
398
+ this._err(ERR.endTagWithTrailingSolidus);
399
+ }
400
+ this.handler.onEndTag(ct);
401
+ }
402
+ this.preprocessor.dropParsedChunk();
403
+ }
404
+ emitCurrentComment(ct) {
405
+ this.prepareToken(ct);
406
+ this.handler.onComment(ct);
407
+ this.preprocessor.dropParsedChunk();
408
+ }
409
+ emitCurrentDoctype(ct) {
410
+ this.prepareToken(ct);
411
+ this.handler.onDoctype(ct);
412
+ this.preprocessor.dropParsedChunk();
413
+ }
414
+ _emitCurrentCharacterToken(nextLocation) {
415
+ if (this.currentCharacterToken) {
416
+ //NOTE: if we have a pending character token, make it's end location equal to the
417
+ //current token's start location.
418
+ if (nextLocation && this.currentCharacterToken.location) {
419
+ this.currentCharacterToken.location.endLine = nextLocation.startLine;
420
+ this.currentCharacterToken.location.endCol = nextLocation.startCol;
421
+ this.currentCharacterToken.location.endOffset = nextLocation.startOffset;
422
+ }
423
+ switch (this.currentCharacterToken.type) {
424
+ case TokenType.CHARACTER: {
425
+ this.handler.onCharacter(this.currentCharacterToken);
426
+ break;
427
+ }
428
+ case TokenType.NULL_CHARACTER: {
429
+ this.handler.onNullCharacter(this.currentCharacterToken);
430
+ break;
431
+ }
432
+ case TokenType.WHITESPACE_CHARACTER: {
433
+ this.handler.onWhitespaceCharacter(this.currentCharacterToken);
434
+ break;
435
+ }
436
+ }
437
+ this.currentCharacterToken = null;
438
+ }
439
+ }
440
+ _emitEOFToken() {
441
+ const location = this.getCurrentLocation(0);
442
+ if (location) {
443
+ location.endLine = location.startLine;
444
+ location.endCol = location.startCol;
445
+ location.endOffset = location.startOffset;
446
+ }
447
+ this._emitCurrentCharacterToken(location);
448
+ this.handler.onEof({ type: TokenType.EOF, location });
449
+ this.active = false;
450
+ }
451
+ //Characters emission
452
+ //OPTIMIZATION: specification uses only one type of character tokens (one token per character).
453
+ //This causes a huge memory overhead and a lot of unnecessary parser loops. parse5 uses 3 groups of characters.
454
+ //If we have a sequence of characters that belong to the same group, the parser can process it
455
+ //as a single solid character token.
456
+ //So, there are 3 types of character tokens in parse5:
457
+ //1)TokenType.NULL_CHARACTER - \u0000-character sequences (e.g. '\u0000\u0000\u0000')
458
+ //2)TokenType.WHITESPACE_CHARACTER - any whitespace/new-line character sequences (e.g. '\n \r\t \f')
459
+ //3)TokenType.CHARACTER - any character sequence which don't belong to groups 1 and 2 (e.g. 'abcdef1234@@#$%^')
460
+ _appendCharToCurrentCharacterToken(type, ch) {
461
+ if (this.currentCharacterToken) {
462
+ if (this.currentCharacterToken.type !== type) {
463
+ this.currentLocation = this.getCurrentLocation(0);
464
+ this._emitCurrentCharacterToken(this.currentLocation);
465
+ this.preprocessor.dropParsedChunk();
466
+ }
467
+ else {
468
+ this.currentCharacterToken.chars += ch;
469
+ return;
470
+ }
471
+ }
472
+ this._createCharacterToken(type, ch);
473
+ }
474
+ _emitCodePoint(cp) {
475
+ let type = TokenType.CHARACTER;
476
+ if (isWhitespace(cp)) {
477
+ type = TokenType.WHITESPACE_CHARACTER;
478
+ }
479
+ else if (cp === $.NULL) {
480
+ type = TokenType.NULL_CHARACTER;
481
+ }
482
+ this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
483
+ }
484
+ //NOTE: used when we emit characters explicitly.
485
+ //This is always for non-whitespace and non-null characters, which allows us to avoid additional checks.
486
+ _emitChars(ch) {
487
+ this._appendCharToCurrentCharacterToken(TokenType.CHARACTER, ch);
488
+ }
489
+ // Character reference helpers
490
+ _matchNamedCharacterReference(cp) {
491
+ let result = null;
492
+ let excess = 0;
493
+ let withoutSemicolon = false;
494
+ for (let i = 0, current = htmlDecodeTree[0]; i >= 0; cp = this._consume()) {
495
+ i = determineBranch(htmlDecodeTree, current, i + 1, cp);
496
+ if (i < 0)
497
+ break;
498
+ excess += 1;
499
+ current = htmlDecodeTree[i];
500
+ const masked = current & BinTrieFlags.VALUE_LENGTH;
501
+ // If the branch is a value, store it and continue
502
+ if (masked) {
503
+ // The mask is the number of bytes of the value, including the current byte.
504
+ const valueLength = (masked >> 14) - 1;
505
+ // Attribute values that aren't terminated properly aren't parsed, and shouldn't lead to a parser error.
506
+ // See the example in https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state
507
+ if (cp !== $.SEMICOLON &&
508
+ this._isCharacterReferenceInAttribute() &&
509
+ isEntityInAttributeInvalidEnd(this.preprocessor.peek(1))) {
510
+ //NOTE: we don't flush all consumed code points here, and instead switch back to the original state after
511
+ //emitting an ampersand. This is fine, as alphanumeric characters won't be parsed differently in attributes.
512
+ result = [$.AMPERSAND];
513
+ // Skip over the value.
514
+ i += valueLength;
515
+ }
516
+ else {
517
+ // If this is a surrogate pair, consume the next two bytes.
518
+ result =
519
+ valueLength === 0
520
+ ? [htmlDecodeTree[i] & ~BinTrieFlags.VALUE_LENGTH]
521
+ : valueLength === 1
522
+ ? [htmlDecodeTree[++i]]
523
+ : [htmlDecodeTree[++i], htmlDecodeTree[++i]];
524
+ excess = 0;
525
+ withoutSemicolon = cp !== $.SEMICOLON;
526
+ }
527
+ if (valueLength === 0) {
528
+ // If the value is zero-length, we're done.
529
+ this._consume();
530
+ break;
531
+ }
532
+ }
533
+ }
534
+ this._unconsume(excess);
535
+ if (withoutSemicolon && !this.preprocessor.endOfChunkHit) {
536
+ this._err(ERR.missingSemicolonAfterCharacterReference);
537
+ }
538
+ // We want to emit the error above on the code point after the entity.
539
+ // We always consume one code point too many in the loop, and we wait to
540
+ // unconsume it until after the error is emitted.
541
+ this._unconsume(1);
542
+ return result;
543
+ }
544
+ _isCharacterReferenceInAttribute() {
545
+ return (this.returnState === State.ATTRIBUTE_VALUE_DOUBLE_QUOTED ||
546
+ this.returnState === State.ATTRIBUTE_VALUE_SINGLE_QUOTED ||
547
+ this.returnState === State.ATTRIBUTE_VALUE_UNQUOTED);
548
+ }
549
+ _flushCodePointConsumedAsCharacterReference(cp) {
550
+ if (this._isCharacterReferenceInAttribute()) {
551
+ this.currentAttr.value += String.fromCodePoint(cp);
552
+ }
553
+ else {
554
+ this._emitCodePoint(cp);
555
+ }
556
+ }
557
+ // Calling states this way turns out to be much faster than any other approach.
558
+ _callState(cp) {
559
+ switch (this.state) {
560
+ case State.DATA: {
561
+ this._stateData(cp);
562
+ break;
563
+ }
564
+ case State.RCDATA: {
565
+ this._stateRcdata(cp);
566
+ break;
567
+ }
568
+ case State.RAWTEXT: {
569
+ this._stateRawtext(cp);
570
+ break;
571
+ }
572
+ case State.SCRIPT_DATA: {
573
+ this._stateScriptData(cp);
574
+ break;
575
+ }
576
+ case State.PLAINTEXT: {
577
+ this._statePlaintext(cp);
578
+ break;
579
+ }
580
+ case State.TAG_OPEN: {
581
+ this._stateTagOpen(cp);
582
+ break;
583
+ }
584
+ case State.END_TAG_OPEN: {
585
+ this._stateEndTagOpen(cp);
586
+ break;
587
+ }
588
+ case State.TAG_NAME: {
589
+ this._stateTagName(cp);
590
+ break;
591
+ }
592
+ case State.RCDATA_LESS_THAN_SIGN: {
593
+ this._stateRcdataLessThanSign(cp);
594
+ break;
595
+ }
596
+ case State.RCDATA_END_TAG_OPEN: {
597
+ this._stateRcdataEndTagOpen(cp);
598
+ break;
599
+ }
600
+ case State.RCDATA_END_TAG_NAME: {
601
+ this._stateRcdataEndTagName(cp);
602
+ break;
603
+ }
604
+ case State.RAWTEXT_LESS_THAN_SIGN: {
605
+ this._stateRawtextLessThanSign(cp);
606
+ break;
607
+ }
608
+ case State.RAWTEXT_END_TAG_OPEN: {
609
+ this._stateRawtextEndTagOpen(cp);
610
+ break;
611
+ }
612
+ case State.RAWTEXT_END_TAG_NAME: {
613
+ this._stateRawtextEndTagName(cp);
614
+ break;
615
+ }
616
+ case State.SCRIPT_DATA_LESS_THAN_SIGN: {
617
+ this._stateScriptDataLessThanSign(cp);
618
+ break;
619
+ }
620
+ case State.SCRIPT_DATA_END_TAG_OPEN: {
621
+ this._stateScriptDataEndTagOpen(cp);
622
+ break;
623
+ }
624
+ case State.SCRIPT_DATA_END_TAG_NAME: {
625
+ this._stateScriptDataEndTagName(cp);
626
+ break;
627
+ }
628
+ case State.SCRIPT_DATA_ESCAPE_START: {
629
+ this._stateScriptDataEscapeStart(cp);
630
+ break;
631
+ }
632
+ case State.SCRIPT_DATA_ESCAPE_START_DASH: {
633
+ this._stateScriptDataEscapeStartDash(cp);
634
+ break;
635
+ }
636
+ case State.SCRIPT_DATA_ESCAPED: {
637
+ this._stateScriptDataEscaped(cp);
638
+ break;
639
+ }
640
+ case State.SCRIPT_DATA_ESCAPED_DASH: {
641
+ this._stateScriptDataEscapedDash(cp);
642
+ break;
643
+ }
644
+ case State.SCRIPT_DATA_ESCAPED_DASH_DASH: {
645
+ this._stateScriptDataEscapedDashDash(cp);
646
+ break;
647
+ }
648
+ case State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN: {
649
+ this._stateScriptDataEscapedLessThanSign(cp);
650
+ break;
651
+ }
652
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN: {
653
+ this._stateScriptDataEscapedEndTagOpen(cp);
654
+ break;
655
+ }
656
+ case State.SCRIPT_DATA_ESCAPED_END_TAG_NAME: {
657
+ this._stateScriptDataEscapedEndTagName(cp);
658
+ break;
659
+ }
660
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_START: {
661
+ this._stateScriptDataDoubleEscapeStart(cp);
662
+ break;
663
+ }
664
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED: {
665
+ this._stateScriptDataDoubleEscaped(cp);
666
+ break;
667
+ }
668
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH: {
669
+ this._stateScriptDataDoubleEscapedDash(cp);
670
+ break;
671
+ }
672
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH: {
673
+ this._stateScriptDataDoubleEscapedDashDash(cp);
674
+ break;
675
+ }
676
+ case State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN: {
677
+ this._stateScriptDataDoubleEscapedLessThanSign(cp);
678
+ break;
679
+ }
680
+ case State.SCRIPT_DATA_DOUBLE_ESCAPE_END: {
681
+ this._stateScriptDataDoubleEscapeEnd(cp);
682
+ break;
683
+ }
684
+ case State.BEFORE_ATTRIBUTE_NAME: {
685
+ this._stateBeforeAttributeName(cp);
686
+ break;
687
+ }
688
+ case State.ATTRIBUTE_NAME: {
689
+ this._stateAttributeName(cp);
690
+ break;
691
+ }
692
+ case State.AFTER_ATTRIBUTE_NAME: {
693
+ this._stateAfterAttributeName(cp);
694
+ break;
695
+ }
696
+ case State.BEFORE_ATTRIBUTE_VALUE: {
697
+ this._stateBeforeAttributeValue(cp);
698
+ break;
699
+ }
700
+ case State.ATTRIBUTE_VALUE_DOUBLE_QUOTED: {
701
+ this._stateAttributeValueDoubleQuoted(cp);
702
+ break;
703
+ }
704
+ case State.ATTRIBUTE_VALUE_SINGLE_QUOTED: {
705
+ this._stateAttributeValueSingleQuoted(cp);
706
+ break;
707
+ }
708
+ case State.ATTRIBUTE_VALUE_UNQUOTED: {
709
+ this._stateAttributeValueUnquoted(cp);
710
+ break;
711
+ }
712
+ case State.AFTER_ATTRIBUTE_VALUE_QUOTED: {
713
+ this._stateAfterAttributeValueQuoted(cp);
714
+ break;
715
+ }
716
+ case State.SELF_CLOSING_START_TAG: {
717
+ this._stateSelfClosingStartTag(cp);
718
+ break;
719
+ }
720
+ case State.BOGUS_COMMENT: {
721
+ this._stateBogusComment(cp);
722
+ break;
723
+ }
724
+ case State.MARKUP_DECLARATION_OPEN: {
725
+ this._stateMarkupDeclarationOpen(cp);
726
+ break;
727
+ }
728
+ case State.COMMENT_START: {
729
+ this._stateCommentStart(cp);
730
+ break;
731
+ }
732
+ case State.COMMENT_START_DASH: {
733
+ this._stateCommentStartDash(cp);
734
+ break;
735
+ }
736
+ case State.COMMENT: {
737
+ this._stateComment(cp);
738
+ break;
739
+ }
740
+ case State.COMMENT_LESS_THAN_SIGN: {
741
+ this._stateCommentLessThanSign(cp);
742
+ break;
743
+ }
744
+ case State.COMMENT_LESS_THAN_SIGN_BANG: {
745
+ this._stateCommentLessThanSignBang(cp);
746
+ break;
747
+ }
748
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH: {
749
+ this._stateCommentLessThanSignBangDash(cp);
750
+ break;
751
+ }
752
+ case State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH: {
753
+ this._stateCommentLessThanSignBangDashDash(cp);
754
+ break;
755
+ }
756
+ case State.COMMENT_END_DASH: {
757
+ this._stateCommentEndDash(cp);
758
+ break;
759
+ }
760
+ case State.COMMENT_END: {
761
+ this._stateCommentEnd(cp);
762
+ break;
763
+ }
764
+ case State.COMMENT_END_BANG: {
765
+ this._stateCommentEndBang(cp);
766
+ break;
767
+ }
768
+ case State.DOCTYPE: {
769
+ this._stateDoctype(cp);
770
+ break;
771
+ }
772
+ case State.BEFORE_DOCTYPE_NAME: {
773
+ this._stateBeforeDoctypeName(cp);
774
+ break;
775
+ }
776
+ case State.DOCTYPE_NAME: {
777
+ this._stateDoctypeName(cp);
778
+ break;
779
+ }
780
+ case State.AFTER_DOCTYPE_NAME: {
781
+ this._stateAfterDoctypeName(cp);
782
+ break;
783
+ }
784
+ case State.AFTER_DOCTYPE_PUBLIC_KEYWORD: {
785
+ this._stateAfterDoctypePublicKeyword(cp);
786
+ break;
787
+ }
788
+ case State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER: {
789
+ this._stateBeforeDoctypePublicIdentifier(cp);
790
+ break;
791
+ }
792
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED: {
793
+ this._stateDoctypePublicIdentifierDoubleQuoted(cp);
794
+ break;
795
+ }
796
+ case State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED: {
797
+ this._stateDoctypePublicIdentifierSingleQuoted(cp);
798
+ break;
799
+ }
800
+ case State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER: {
801
+ this._stateAfterDoctypePublicIdentifier(cp);
802
+ break;
803
+ }
804
+ case State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS: {
805
+ this._stateBetweenDoctypePublicAndSystemIdentifiers(cp);
806
+ break;
807
+ }
808
+ case State.AFTER_DOCTYPE_SYSTEM_KEYWORD: {
809
+ this._stateAfterDoctypeSystemKeyword(cp);
810
+ break;
811
+ }
812
+ case State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER: {
813
+ this._stateBeforeDoctypeSystemIdentifier(cp);
814
+ break;
815
+ }
816
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED: {
817
+ this._stateDoctypeSystemIdentifierDoubleQuoted(cp);
818
+ break;
819
+ }
820
+ case State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED: {
821
+ this._stateDoctypeSystemIdentifierSingleQuoted(cp);
822
+ break;
823
+ }
824
+ case State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER: {
825
+ this._stateAfterDoctypeSystemIdentifier(cp);
826
+ break;
827
+ }
828
+ case State.BOGUS_DOCTYPE: {
829
+ this._stateBogusDoctype(cp);
830
+ break;
831
+ }
832
+ case State.CDATA_SECTION: {
833
+ this._stateCdataSection(cp);
834
+ break;
835
+ }
836
+ case State.CDATA_SECTION_BRACKET: {
837
+ this._stateCdataSectionBracket(cp);
838
+ break;
839
+ }
840
+ case State.CDATA_SECTION_END: {
841
+ this._stateCdataSectionEnd(cp);
842
+ break;
843
+ }
844
+ case State.CHARACTER_REFERENCE: {
845
+ this._stateCharacterReference(cp);
846
+ break;
847
+ }
848
+ case State.NAMED_CHARACTER_REFERENCE: {
849
+ this._stateNamedCharacterReference(cp);
850
+ break;
851
+ }
852
+ case State.AMBIGUOUS_AMPERSAND: {
853
+ this._stateAmbiguousAmpersand(cp);
854
+ break;
855
+ }
856
+ case State.NUMERIC_CHARACTER_REFERENCE: {
857
+ this._stateNumericCharacterReference(cp);
858
+ break;
859
+ }
860
+ case State.HEXADEMICAL_CHARACTER_REFERENCE_START: {
861
+ this._stateHexademicalCharacterReferenceStart(cp);
862
+ break;
863
+ }
864
+ case State.DECIMAL_CHARACTER_REFERENCE_START: {
865
+ this._stateDecimalCharacterReferenceStart(cp);
866
+ break;
867
+ }
868
+ case State.HEXADEMICAL_CHARACTER_REFERENCE: {
869
+ this._stateHexademicalCharacterReference(cp);
870
+ break;
871
+ }
872
+ case State.DECIMAL_CHARACTER_REFERENCE: {
873
+ this._stateDecimalCharacterReference(cp);
874
+ break;
875
+ }
876
+ case State.NUMERIC_CHARACTER_REFERENCE_END: {
877
+ this._stateNumericCharacterReferenceEnd();
878
+ break;
879
+ }
880
+ default: {
881
+ throw new Error('Unknown state');
882
+ }
883
+ }
884
+ }
885
+ // State machine
886
+ // Data state
887
+ //------------------------------------------------------------------
888
+ _stateData(cp) {
889
+ switch (cp) {
890
+ case $.LESS_THAN_SIGN: {
891
+ this.state = State.TAG_OPEN;
892
+ break;
893
+ }
894
+ case $.AMPERSAND: {
895
+ this.returnState = State.DATA;
896
+ this.state = State.CHARACTER_REFERENCE;
897
+ break;
898
+ }
899
+ case $.NULL: {
900
+ this._err(ERR.unexpectedNullCharacter);
901
+ this._emitCodePoint(cp);
902
+ break;
903
+ }
904
+ case $.EOF: {
905
+ this._emitEOFToken();
906
+ break;
907
+ }
908
+ default: {
909
+ this._emitCodePoint(cp);
910
+ }
911
+ }
912
+ }
913
+ // RCDATA state
914
+ //------------------------------------------------------------------
915
+ _stateRcdata(cp) {
916
+ switch (cp) {
917
+ case $.AMPERSAND: {
918
+ this.returnState = State.RCDATA;
919
+ this.state = State.CHARACTER_REFERENCE;
920
+ break;
921
+ }
922
+ case $.LESS_THAN_SIGN: {
923
+ this.state = State.RCDATA_LESS_THAN_SIGN;
924
+ break;
925
+ }
926
+ case $.NULL: {
927
+ this._err(ERR.unexpectedNullCharacter);
928
+ this._emitChars(REPLACEMENT_CHARACTER);
929
+ break;
930
+ }
931
+ case $.EOF: {
932
+ this._emitEOFToken();
933
+ break;
934
+ }
935
+ default: {
936
+ this._emitCodePoint(cp);
937
+ }
938
+ }
939
+ }
940
+ // RAWTEXT state
941
+ //------------------------------------------------------------------
942
+ _stateRawtext(cp) {
943
+ switch (cp) {
944
+ case $.LESS_THAN_SIGN: {
945
+ this.state = State.RAWTEXT_LESS_THAN_SIGN;
946
+ break;
947
+ }
948
+ case $.NULL: {
949
+ this._err(ERR.unexpectedNullCharacter);
950
+ this._emitChars(REPLACEMENT_CHARACTER);
951
+ break;
952
+ }
953
+ case $.EOF: {
954
+ this._emitEOFToken();
955
+ break;
956
+ }
957
+ default: {
958
+ this._emitCodePoint(cp);
959
+ }
960
+ }
961
+ }
962
+ // Script data state
963
+ //------------------------------------------------------------------
964
+ _stateScriptData(cp) {
965
+ switch (cp) {
966
+ case $.LESS_THAN_SIGN: {
967
+ this.state = State.SCRIPT_DATA_LESS_THAN_SIGN;
968
+ break;
969
+ }
970
+ case $.NULL: {
971
+ this._err(ERR.unexpectedNullCharacter);
972
+ this._emitChars(REPLACEMENT_CHARACTER);
973
+ break;
974
+ }
975
+ case $.EOF: {
976
+ this._emitEOFToken();
977
+ break;
978
+ }
979
+ default: {
980
+ this._emitCodePoint(cp);
981
+ }
982
+ }
983
+ }
984
+ // PLAINTEXT state
985
+ //------------------------------------------------------------------
986
+ _statePlaintext(cp) {
987
+ switch (cp) {
988
+ case $.NULL: {
989
+ this._err(ERR.unexpectedNullCharacter);
990
+ this._emitChars(REPLACEMENT_CHARACTER);
991
+ break;
992
+ }
993
+ case $.EOF: {
994
+ this._emitEOFToken();
995
+ break;
996
+ }
997
+ default: {
998
+ this._emitCodePoint(cp);
999
+ }
1000
+ }
1001
+ }
1002
+ // Tag open state
1003
+ //------------------------------------------------------------------
1004
+ _stateTagOpen(cp) {
1005
+ if (isAsciiLetter(cp)) {
1006
+ this._createStartTagToken();
1007
+ this.state = State.TAG_NAME;
1008
+ this._stateTagName(cp);
1009
+ }
1010
+ else
1011
+ switch (cp) {
1012
+ case $.EXCLAMATION_MARK: {
1013
+ this.state = State.MARKUP_DECLARATION_OPEN;
1014
+ break;
1015
+ }
1016
+ case $.SOLIDUS: {
1017
+ this.state = State.END_TAG_OPEN;
1018
+ break;
1019
+ }
1020
+ case $.QUESTION_MARK: {
1021
+ this._err(ERR.unexpectedQuestionMarkInsteadOfTagName);
1022
+ this._createCommentToken(1);
1023
+ this.state = State.BOGUS_COMMENT;
1024
+ this._stateBogusComment(cp);
1025
+ break;
1026
+ }
1027
+ case $.EOF: {
1028
+ this._err(ERR.eofBeforeTagName);
1029
+ this._emitChars('<');
1030
+ this._emitEOFToken();
1031
+ break;
1032
+ }
1033
+ default: {
1034
+ this._err(ERR.invalidFirstCharacterOfTagName);
1035
+ this._emitChars('<');
1036
+ this.state = State.DATA;
1037
+ this._stateData(cp);
1038
+ }
1039
+ }
1040
+ }
1041
+ // End tag open state
1042
+ //------------------------------------------------------------------
1043
+ _stateEndTagOpen(cp) {
1044
+ if (isAsciiLetter(cp)) {
1045
+ this._createEndTagToken();
1046
+ this.state = State.TAG_NAME;
1047
+ this._stateTagName(cp);
1048
+ }
1049
+ else
1050
+ switch (cp) {
1051
+ case $.GREATER_THAN_SIGN: {
1052
+ this._err(ERR.missingEndTagName);
1053
+ this.state = State.DATA;
1054
+ break;
1055
+ }
1056
+ case $.EOF: {
1057
+ this._err(ERR.eofBeforeTagName);
1058
+ this._emitChars('</');
1059
+ this._emitEOFToken();
1060
+ break;
1061
+ }
1062
+ default: {
1063
+ this._err(ERR.invalidFirstCharacterOfTagName);
1064
+ this._createCommentToken(2);
1065
+ this.state = State.BOGUS_COMMENT;
1066
+ this._stateBogusComment(cp);
1067
+ }
1068
+ }
1069
+ }
1070
+ // Tag name state
1071
+ //------------------------------------------------------------------
1072
+ _stateTagName(cp) {
1073
+ const token = this.currentToken;
1074
+ switch (cp) {
1075
+ case $.SPACE:
1076
+ case $.LINE_FEED:
1077
+ case $.TABULATION:
1078
+ case $.FORM_FEED: {
1079
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1080
+ break;
1081
+ }
1082
+ case $.SOLIDUS: {
1083
+ this.state = State.SELF_CLOSING_START_TAG;
1084
+ break;
1085
+ }
1086
+ case $.GREATER_THAN_SIGN: {
1087
+ this.state = State.DATA;
1088
+ this.emitCurrentTagToken();
1089
+ break;
1090
+ }
1091
+ case $.NULL: {
1092
+ this._err(ERR.unexpectedNullCharacter);
1093
+ token.tagName += REPLACEMENT_CHARACTER;
1094
+ break;
1095
+ }
1096
+ case $.EOF: {
1097
+ this._err(ERR.eofInTag);
1098
+ this._emitEOFToken();
1099
+ break;
1100
+ }
1101
+ default: {
1102
+ token.tagName += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1103
+ }
1104
+ }
1105
+ }
1106
+ // RCDATA less-than sign state
1107
+ //------------------------------------------------------------------
1108
+ _stateRcdataLessThanSign(cp) {
1109
+ if (cp === $.SOLIDUS) {
1110
+ this.state = State.RCDATA_END_TAG_OPEN;
1111
+ }
1112
+ else {
1113
+ this._emitChars('<');
1114
+ this.state = State.RCDATA;
1115
+ this._stateRcdata(cp);
1116
+ }
1117
+ }
1118
+ // RCDATA end tag open state
1119
+ //------------------------------------------------------------------
1120
+ _stateRcdataEndTagOpen(cp) {
1121
+ if (isAsciiLetter(cp)) {
1122
+ this.state = State.RCDATA_END_TAG_NAME;
1123
+ this._stateRcdataEndTagName(cp);
1124
+ }
1125
+ else {
1126
+ this._emitChars('</');
1127
+ this.state = State.RCDATA;
1128
+ this._stateRcdata(cp);
1129
+ }
1130
+ }
1131
+ handleSpecialEndTag(_cp) {
1132
+ if (!this.preprocessor.startsWith(this.lastStartTagName, false)) {
1133
+ return !this._ensureHibernation();
1134
+ }
1135
+ this._createEndTagToken();
1136
+ const token = this.currentToken;
1137
+ token.tagName = this.lastStartTagName;
1138
+ const cp = this.preprocessor.peek(this.lastStartTagName.length);
1139
+ switch (cp) {
1140
+ case $.SPACE:
1141
+ case $.LINE_FEED:
1142
+ case $.TABULATION:
1143
+ case $.FORM_FEED: {
1144
+ this._advanceBy(this.lastStartTagName.length);
1145
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1146
+ return false;
1147
+ }
1148
+ case $.SOLIDUS: {
1149
+ this._advanceBy(this.lastStartTagName.length);
1150
+ this.state = State.SELF_CLOSING_START_TAG;
1151
+ return false;
1152
+ }
1153
+ case $.GREATER_THAN_SIGN: {
1154
+ this._advanceBy(this.lastStartTagName.length);
1155
+ this.emitCurrentTagToken();
1156
+ this.state = State.DATA;
1157
+ return false;
1158
+ }
1159
+ default: {
1160
+ return !this._ensureHibernation();
1161
+ }
1162
+ }
1163
+ }
1164
+ // RCDATA end tag name state
1165
+ //------------------------------------------------------------------
1166
+ _stateRcdataEndTagName(cp) {
1167
+ if (this.handleSpecialEndTag(cp)) {
1168
+ this._emitChars('</');
1169
+ this.state = State.RCDATA;
1170
+ this._stateRcdata(cp);
1171
+ }
1172
+ }
1173
+ // RAWTEXT less-than sign state
1174
+ //------------------------------------------------------------------
1175
+ _stateRawtextLessThanSign(cp) {
1176
+ if (cp === $.SOLIDUS) {
1177
+ this.state = State.RAWTEXT_END_TAG_OPEN;
1178
+ }
1179
+ else {
1180
+ this._emitChars('<');
1181
+ this.state = State.RAWTEXT;
1182
+ this._stateRawtext(cp);
1183
+ }
1184
+ }
1185
+ // RAWTEXT end tag open state
1186
+ //------------------------------------------------------------------
1187
+ _stateRawtextEndTagOpen(cp) {
1188
+ if (isAsciiLetter(cp)) {
1189
+ this.state = State.RAWTEXT_END_TAG_NAME;
1190
+ this._stateRawtextEndTagName(cp);
1191
+ }
1192
+ else {
1193
+ this._emitChars('</');
1194
+ this.state = State.RAWTEXT;
1195
+ this._stateRawtext(cp);
1196
+ }
1197
+ }
1198
+ // RAWTEXT end tag name state
1199
+ //------------------------------------------------------------------
1200
+ _stateRawtextEndTagName(cp) {
1201
+ if (this.handleSpecialEndTag(cp)) {
1202
+ this._emitChars('</');
1203
+ this.state = State.RAWTEXT;
1204
+ this._stateRawtext(cp);
1205
+ }
1206
+ }
1207
+ // Script data less-than sign state
1208
+ //------------------------------------------------------------------
1209
+ _stateScriptDataLessThanSign(cp) {
1210
+ switch (cp) {
1211
+ case $.SOLIDUS: {
1212
+ this.state = State.SCRIPT_DATA_END_TAG_OPEN;
1213
+ break;
1214
+ }
1215
+ case $.EXCLAMATION_MARK: {
1216
+ this.state = State.SCRIPT_DATA_ESCAPE_START;
1217
+ this._emitChars('<!');
1218
+ break;
1219
+ }
1220
+ default: {
1221
+ this._emitChars('<');
1222
+ this.state = State.SCRIPT_DATA;
1223
+ this._stateScriptData(cp);
1224
+ }
1225
+ }
1226
+ }
1227
+ // Script data end tag open state
1228
+ //------------------------------------------------------------------
1229
+ _stateScriptDataEndTagOpen(cp) {
1230
+ if (isAsciiLetter(cp)) {
1231
+ this.state = State.SCRIPT_DATA_END_TAG_NAME;
1232
+ this._stateScriptDataEndTagName(cp);
1233
+ }
1234
+ else {
1235
+ this._emitChars('</');
1236
+ this.state = State.SCRIPT_DATA;
1237
+ this._stateScriptData(cp);
1238
+ }
1239
+ }
1240
+ // Script data end tag name state
1241
+ //------------------------------------------------------------------
1242
+ _stateScriptDataEndTagName(cp) {
1243
+ if (this.handleSpecialEndTag(cp)) {
1244
+ this._emitChars('</');
1245
+ this.state = State.SCRIPT_DATA;
1246
+ this._stateScriptData(cp);
1247
+ }
1248
+ }
1249
+ // Script data escape start state
1250
+ //------------------------------------------------------------------
1251
+ _stateScriptDataEscapeStart(cp) {
1252
+ if (cp === $.HYPHEN_MINUS) {
1253
+ this.state = State.SCRIPT_DATA_ESCAPE_START_DASH;
1254
+ this._emitChars('-');
1255
+ }
1256
+ else {
1257
+ this.state = State.SCRIPT_DATA;
1258
+ this._stateScriptData(cp);
1259
+ }
1260
+ }
1261
+ // Script data escape start dash state
1262
+ //------------------------------------------------------------------
1263
+ _stateScriptDataEscapeStartDash(cp) {
1264
+ if (cp === $.HYPHEN_MINUS) {
1265
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1266
+ this._emitChars('-');
1267
+ }
1268
+ else {
1269
+ this.state = State.SCRIPT_DATA;
1270
+ this._stateScriptData(cp);
1271
+ }
1272
+ }
1273
+ // Script data escaped state
1274
+ //------------------------------------------------------------------
1275
+ _stateScriptDataEscaped(cp) {
1276
+ switch (cp) {
1277
+ case $.HYPHEN_MINUS: {
1278
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH;
1279
+ this._emitChars('-');
1280
+ break;
1281
+ }
1282
+ case $.LESS_THAN_SIGN: {
1283
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1284
+ break;
1285
+ }
1286
+ case $.NULL: {
1287
+ this._err(ERR.unexpectedNullCharacter);
1288
+ this._emitChars(REPLACEMENT_CHARACTER);
1289
+ break;
1290
+ }
1291
+ case $.EOF: {
1292
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1293
+ this._emitEOFToken();
1294
+ break;
1295
+ }
1296
+ default: {
1297
+ this._emitCodePoint(cp);
1298
+ }
1299
+ }
1300
+ }
1301
+ // Script data escaped dash state
1302
+ //------------------------------------------------------------------
1303
+ _stateScriptDataEscapedDash(cp) {
1304
+ switch (cp) {
1305
+ case $.HYPHEN_MINUS: {
1306
+ this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
1307
+ this._emitChars('-');
1308
+ break;
1309
+ }
1310
+ case $.LESS_THAN_SIGN: {
1311
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1312
+ break;
1313
+ }
1314
+ case $.NULL: {
1315
+ this._err(ERR.unexpectedNullCharacter);
1316
+ this.state = State.SCRIPT_DATA_ESCAPED;
1317
+ this._emitChars(REPLACEMENT_CHARACTER);
1318
+ break;
1319
+ }
1320
+ case $.EOF: {
1321
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1322
+ this._emitEOFToken();
1323
+ break;
1324
+ }
1325
+ default: {
1326
+ this.state = State.SCRIPT_DATA_ESCAPED;
1327
+ this._emitCodePoint(cp);
1328
+ }
1329
+ }
1330
+ }
1331
+ // Script data escaped dash dash state
1332
+ //------------------------------------------------------------------
1333
+ _stateScriptDataEscapedDashDash(cp) {
1334
+ switch (cp) {
1335
+ case $.HYPHEN_MINUS: {
1336
+ this._emitChars('-');
1337
+ break;
1338
+ }
1339
+ case $.LESS_THAN_SIGN: {
1340
+ this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
1341
+ break;
1342
+ }
1343
+ case $.GREATER_THAN_SIGN: {
1344
+ this.state = State.SCRIPT_DATA;
1345
+ this._emitChars('>');
1346
+ break;
1347
+ }
1348
+ case $.NULL: {
1349
+ this._err(ERR.unexpectedNullCharacter);
1350
+ this.state = State.SCRIPT_DATA_ESCAPED;
1351
+ this._emitChars(REPLACEMENT_CHARACTER);
1352
+ break;
1353
+ }
1354
+ case $.EOF: {
1355
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1356
+ this._emitEOFToken();
1357
+ break;
1358
+ }
1359
+ default: {
1360
+ this.state = State.SCRIPT_DATA_ESCAPED;
1361
+ this._emitCodePoint(cp);
1362
+ }
1363
+ }
1364
+ }
1365
+ // Script data escaped less-than sign state
1366
+ //------------------------------------------------------------------
1367
+ _stateScriptDataEscapedLessThanSign(cp) {
1368
+ if (cp === $.SOLIDUS) {
1369
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN;
1370
+ }
1371
+ else if (isAsciiLetter(cp)) {
1372
+ this._emitChars('<');
1373
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_START;
1374
+ this._stateScriptDataDoubleEscapeStart(cp);
1375
+ }
1376
+ else {
1377
+ this._emitChars('<');
1378
+ this.state = State.SCRIPT_DATA_ESCAPED;
1379
+ this._stateScriptDataEscaped(cp);
1380
+ }
1381
+ }
1382
+ // Script data escaped end tag open state
1383
+ //------------------------------------------------------------------
1384
+ _stateScriptDataEscapedEndTagOpen(cp) {
1385
+ if (isAsciiLetter(cp)) {
1386
+ this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_NAME;
1387
+ this._stateScriptDataEscapedEndTagName(cp);
1388
+ }
1389
+ else {
1390
+ this._emitChars('</');
1391
+ this.state = State.SCRIPT_DATA_ESCAPED;
1392
+ this._stateScriptDataEscaped(cp);
1393
+ }
1394
+ }
1395
+ // Script data escaped end tag name state
1396
+ //------------------------------------------------------------------
1397
+ _stateScriptDataEscapedEndTagName(cp) {
1398
+ if (this.handleSpecialEndTag(cp)) {
1399
+ this._emitChars('</');
1400
+ this.state = State.SCRIPT_DATA_ESCAPED;
1401
+ this._stateScriptDataEscaped(cp);
1402
+ }
1403
+ }
1404
+ // Script data double escape start state
1405
+ //------------------------------------------------------------------
1406
+ _stateScriptDataDoubleEscapeStart(cp) {
1407
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1408
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1409
+ this._emitCodePoint(cp);
1410
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1411
+ this._emitCodePoint(this._consume());
1412
+ }
1413
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1414
+ }
1415
+ else if (!this._ensureHibernation()) {
1416
+ this.state = State.SCRIPT_DATA_ESCAPED;
1417
+ this._stateScriptDataEscaped(cp);
1418
+ }
1419
+ }
1420
+ // Script data double escaped state
1421
+ //------------------------------------------------------------------
1422
+ _stateScriptDataDoubleEscaped(cp) {
1423
+ switch (cp) {
1424
+ case $.HYPHEN_MINUS: {
1425
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH;
1426
+ this._emitChars('-');
1427
+ break;
1428
+ }
1429
+ case $.LESS_THAN_SIGN: {
1430
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1431
+ this._emitChars('<');
1432
+ break;
1433
+ }
1434
+ case $.NULL: {
1435
+ this._err(ERR.unexpectedNullCharacter);
1436
+ this._emitChars(REPLACEMENT_CHARACTER);
1437
+ break;
1438
+ }
1439
+ case $.EOF: {
1440
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1441
+ this._emitEOFToken();
1442
+ break;
1443
+ }
1444
+ default: {
1445
+ this._emitCodePoint(cp);
1446
+ }
1447
+ }
1448
+ }
1449
+ // Script data double escaped dash state
1450
+ //------------------------------------------------------------------
1451
+ _stateScriptDataDoubleEscapedDash(cp) {
1452
+ switch (cp) {
1453
+ case $.HYPHEN_MINUS: {
1454
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH;
1455
+ this._emitChars('-');
1456
+ break;
1457
+ }
1458
+ case $.LESS_THAN_SIGN: {
1459
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1460
+ this._emitChars('<');
1461
+ break;
1462
+ }
1463
+ case $.NULL: {
1464
+ this._err(ERR.unexpectedNullCharacter);
1465
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1466
+ this._emitChars(REPLACEMENT_CHARACTER);
1467
+ break;
1468
+ }
1469
+ case $.EOF: {
1470
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1471
+ this._emitEOFToken();
1472
+ break;
1473
+ }
1474
+ default: {
1475
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1476
+ this._emitCodePoint(cp);
1477
+ }
1478
+ }
1479
+ }
1480
+ // Script data double escaped dash dash state
1481
+ //------------------------------------------------------------------
1482
+ _stateScriptDataDoubleEscapedDashDash(cp) {
1483
+ switch (cp) {
1484
+ case $.HYPHEN_MINUS: {
1485
+ this._emitChars('-');
1486
+ break;
1487
+ }
1488
+ case $.LESS_THAN_SIGN: {
1489
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
1490
+ this._emitChars('<');
1491
+ break;
1492
+ }
1493
+ case $.GREATER_THAN_SIGN: {
1494
+ this.state = State.SCRIPT_DATA;
1495
+ this._emitChars('>');
1496
+ break;
1497
+ }
1498
+ case $.NULL: {
1499
+ this._err(ERR.unexpectedNullCharacter);
1500
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1501
+ this._emitChars(REPLACEMENT_CHARACTER);
1502
+ break;
1503
+ }
1504
+ case $.EOF: {
1505
+ this._err(ERR.eofInScriptHtmlCommentLikeText);
1506
+ this._emitEOFToken();
1507
+ break;
1508
+ }
1509
+ default: {
1510
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1511
+ this._emitCodePoint(cp);
1512
+ }
1513
+ }
1514
+ }
1515
+ // Script data double escaped less-than sign state
1516
+ //------------------------------------------------------------------
1517
+ _stateScriptDataDoubleEscapedLessThanSign(cp) {
1518
+ if (cp === $.SOLIDUS) {
1519
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_END;
1520
+ this._emitChars('/');
1521
+ }
1522
+ else {
1523
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1524
+ this._stateScriptDataDoubleEscaped(cp);
1525
+ }
1526
+ }
1527
+ // Script data double escape end state
1528
+ //------------------------------------------------------------------
1529
+ _stateScriptDataDoubleEscapeEnd(cp) {
1530
+ if (this.preprocessor.startsWith($$.SCRIPT, false) &&
1531
+ isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek($$.SCRIPT.length))) {
1532
+ this._emitCodePoint(cp);
1533
+ for (let i = 0; i < $$.SCRIPT.length; i++) {
1534
+ this._emitCodePoint(this._consume());
1535
+ }
1536
+ this.state = State.SCRIPT_DATA_ESCAPED;
1537
+ }
1538
+ else if (!this._ensureHibernation()) {
1539
+ this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
1540
+ this._stateScriptDataDoubleEscaped(cp);
1541
+ }
1542
+ }
1543
+ // Before attribute name state
1544
+ //------------------------------------------------------------------
1545
+ _stateBeforeAttributeName(cp) {
1546
+ switch (cp) {
1547
+ case $.SPACE:
1548
+ case $.LINE_FEED:
1549
+ case $.TABULATION:
1550
+ case $.FORM_FEED: {
1551
+ // Ignore whitespace
1552
+ break;
1553
+ }
1554
+ case $.SOLIDUS:
1555
+ case $.GREATER_THAN_SIGN:
1556
+ case $.EOF: {
1557
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1558
+ this._stateAfterAttributeName(cp);
1559
+ break;
1560
+ }
1561
+ case $.EQUALS_SIGN: {
1562
+ this._err(ERR.unexpectedEqualsSignBeforeAttributeName);
1563
+ this._createAttr('=');
1564
+ this.state = State.ATTRIBUTE_NAME;
1565
+ break;
1566
+ }
1567
+ default: {
1568
+ this._createAttr('');
1569
+ this.state = State.ATTRIBUTE_NAME;
1570
+ this._stateAttributeName(cp);
1571
+ }
1572
+ }
1573
+ }
1574
+ // Attribute name state
1575
+ //------------------------------------------------------------------
1576
+ _stateAttributeName(cp) {
1577
+ switch (cp) {
1578
+ case $.SPACE:
1579
+ case $.LINE_FEED:
1580
+ case $.TABULATION:
1581
+ case $.FORM_FEED:
1582
+ case $.SOLIDUS:
1583
+ case $.GREATER_THAN_SIGN:
1584
+ case $.EOF: {
1585
+ this._leaveAttrName();
1586
+ this.state = State.AFTER_ATTRIBUTE_NAME;
1587
+ this._stateAfterAttributeName(cp);
1588
+ break;
1589
+ }
1590
+ case $.EQUALS_SIGN: {
1591
+ this._leaveAttrName();
1592
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1593
+ break;
1594
+ }
1595
+ case $.QUOTATION_MARK:
1596
+ case $.APOSTROPHE:
1597
+ case $.LESS_THAN_SIGN: {
1598
+ this._err(ERR.unexpectedCharacterInAttributeName);
1599
+ this.currentAttr.name += String.fromCodePoint(cp);
1600
+ break;
1601
+ }
1602
+ case $.NULL: {
1603
+ this._err(ERR.unexpectedNullCharacter);
1604
+ this.currentAttr.name += REPLACEMENT_CHARACTER;
1605
+ break;
1606
+ }
1607
+ default: {
1608
+ this.currentAttr.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
1609
+ }
1610
+ }
1611
+ }
1612
+ // After attribute name state
1613
+ //------------------------------------------------------------------
1614
+ _stateAfterAttributeName(cp) {
1615
+ switch (cp) {
1616
+ case $.SPACE:
1617
+ case $.LINE_FEED:
1618
+ case $.TABULATION:
1619
+ case $.FORM_FEED: {
1620
+ // Ignore whitespace
1621
+ break;
1622
+ }
1623
+ case $.SOLIDUS: {
1624
+ this.state = State.SELF_CLOSING_START_TAG;
1625
+ break;
1626
+ }
1627
+ case $.EQUALS_SIGN: {
1628
+ this.state = State.BEFORE_ATTRIBUTE_VALUE;
1629
+ break;
1630
+ }
1631
+ case $.GREATER_THAN_SIGN: {
1632
+ this.state = State.DATA;
1633
+ this.emitCurrentTagToken();
1634
+ break;
1635
+ }
1636
+ case $.EOF: {
1637
+ this._err(ERR.eofInTag);
1638
+ this._emitEOFToken();
1639
+ break;
1640
+ }
1641
+ default: {
1642
+ this._createAttr('');
1643
+ this.state = State.ATTRIBUTE_NAME;
1644
+ this._stateAttributeName(cp);
1645
+ }
1646
+ }
1647
+ }
1648
+ // Before attribute value state
1649
+ //------------------------------------------------------------------
1650
+ _stateBeforeAttributeValue(cp) {
1651
+ switch (cp) {
1652
+ case $.SPACE:
1653
+ case $.LINE_FEED:
1654
+ case $.TABULATION:
1655
+ case $.FORM_FEED: {
1656
+ // Ignore whitespace
1657
+ break;
1658
+ }
1659
+ case $.QUOTATION_MARK: {
1660
+ this.state = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
1661
+ break;
1662
+ }
1663
+ case $.APOSTROPHE: {
1664
+ this.state = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
1665
+ break;
1666
+ }
1667
+ case $.GREATER_THAN_SIGN: {
1668
+ this._err(ERR.missingAttributeValue);
1669
+ this.state = State.DATA;
1670
+ this.emitCurrentTagToken();
1671
+ break;
1672
+ }
1673
+ default: {
1674
+ this.state = State.ATTRIBUTE_VALUE_UNQUOTED;
1675
+ this._stateAttributeValueUnquoted(cp);
1676
+ }
1677
+ }
1678
+ }
1679
+ // Attribute value (double-quoted) state
1680
+ //------------------------------------------------------------------
1681
+ _stateAttributeValueDoubleQuoted(cp) {
1682
+ switch (cp) {
1683
+ case $.QUOTATION_MARK: {
1684
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1685
+ break;
1686
+ }
1687
+ case $.AMPERSAND: {
1688
+ this.returnState = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
1689
+ this.state = State.CHARACTER_REFERENCE;
1690
+ break;
1691
+ }
1692
+ case $.NULL: {
1693
+ this._err(ERR.unexpectedNullCharacter);
1694
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1695
+ break;
1696
+ }
1697
+ case $.EOF: {
1698
+ this._err(ERR.eofInTag);
1699
+ this._emitEOFToken();
1700
+ break;
1701
+ }
1702
+ default: {
1703
+ this.currentAttr.value += String.fromCodePoint(cp);
1704
+ }
1705
+ }
1706
+ }
1707
+ // Attribute value (single-quoted) state
1708
+ //------------------------------------------------------------------
1709
+ _stateAttributeValueSingleQuoted(cp) {
1710
+ switch (cp) {
1711
+ case $.APOSTROPHE: {
1712
+ this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
1713
+ break;
1714
+ }
1715
+ case $.AMPERSAND: {
1716
+ this.returnState = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
1717
+ this.state = State.CHARACTER_REFERENCE;
1718
+ break;
1719
+ }
1720
+ case $.NULL: {
1721
+ this._err(ERR.unexpectedNullCharacter);
1722
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1723
+ break;
1724
+ }
1725
+ case $.EOF: {
1726
+ this._err(ERR.eofInTag);
1727
+ this._emitEOFToken();
1728
+ break;
1729
+ }
1730
+ default: {
1731
+ this.currentAttr.value += String.fromCodePoint(cp);
1732
+ }
1733
+ }
1734
+ }
1735
+ // Attribute value (unquoted) state
1736
+ //------------------------------------------------------------------
1737
+ _stateAttributeValueUnquoted(cp) {
1738
+ switch (cp) {
1739
+ case $.SPACE:
1740
+ case $.LINE_FEED:
1741
+ case $.TABULATION:
1742
+ case $.FORM_FEED: {
1743
+ this._leaveAttrValue();
1744
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1745
+ break;
1746
+ }
1747
+ case $.AMPERSAND: {
1748
+ this.returnState = State.ATTRIBUTE_VALUE_UNQUOTED;
1749
+ this.state = State.CHARACTER_REFERENCE;
1750
+ break;
1751
+ }
1752
+ case $.GREATER_THAN_SIGN: {
1753
+ this._leaveAttrValue();
1754
+ this.state = State.DATA;
1755
+ this.emitCurrentTagToken();
1756
+ break;
1757
+ }
1758
+ case $.NULL: {
1759
+ this._err(ERR.unexpectedNullCharacter);
1760
+ this.currentAttr.value += REPLACEMENT_CHARACTER;
1761
+ break;
1762
+ }
1763
+ case $.QUOTATION_MARK:
1764
+ case $.APOSTROPHE:
1765
+ case $.LESS_THAN_SIGN:
1766
+ case $.EQUALS_SIGN:
1767
+ case $.GRAVE_ACCENT: {
1768
+ this._err(ERR.unexpectedCharacterInUnquotedAttributeValue);
1769
+ this.currentAttr.value += String.fromCodePoint(cp);
1770
+ break;
1771
+ }
1772
+ case $.EOF: {
1773
+ this._err(ERR.eofInTag);
1774
+ this._emitEOFToken();
1775
+ break;
1776
+ }
1777
+ default: {
1778
+ this.currentAttr.value += String.fromCodePoint(cp);
1779
+ }
1780
+ }
1781
+ }
1782
+ // After attribute value (quoted) state
1783
+ //------------------------------------------------------------------
1784
+ _stateAfterAttributeValueQuoted(cp) {
1785
+ switch (cp) {
1786
+ case $.SPACE:
1787
+ case $.LINE_FEED:
1788
+ case $.TABULATION:
1789
+ case $.FORM_FEED: {
1790
+ this._leaveAttrValue();
1791
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1792
+ break;
1793
+ }
1794
+ case $.SOLIDUS: {
1795
+ this._leaveAttrValue();
1796
+ this.state = State.SELF_CLOSING_START_TAG;
1797
+ break;
1798
+ }
1799
+ case $.GREATER_THAN_SIGN: {
1800
+ this._leaveAttrValue();
1801
+ this.state = State.DATA;
1802
+ this.emitCurrentTagToken();
1803
+ break;
1804
+ }
1805
+ case $.EOF: {
1806
+ this._err(ERR.eofInTag);
1807
+ this._emitEOFToken();
1808
+ break;
1809
+ }
1810
+ default: {
1811
+ this._err(ERR.missingWhitespaceBetweenAttributes);
1812
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1813
+ this._stateBeforeAttributeName(cp);
1814
+ }
1815
+ }
1816
+ }
1817
+ // Self-closing start tag state
1818
+ //------------------------------------------------------------------
1819
+ _stateSelfClosingStartTag(cp) {
1820
+ switch (cp) {
1821
+ case $.GREATER_THAN_SIGN: {
1822
+ const token = this.currentToken;
1823
+ token.selfClosing = true;
1824
+ this.state = State.DATA;
1825
+ this.emitCurrentTagToken();
1826
+ break;
1827
+ }
1828
+ case $.EOF: {
1829
+ this._err(ERR.eofInTag);
1830
+ this._emitEOFToken();
1831
+ break;
1832
+ }
1833
+ default: {
1834
+ this._err(ERR.unexpectedSolidusInTag);
1835
+ this.state = State.BEFORE_ATTRIBUTE_NAME;
1836
+ this._stateBeforeAttributeName(cp);
1837
+ }
1838
+ }
1839
+ }
1840
+ // Bogus comment state
1841
+ //------------------------------------------------------------------
1842
+ _stateBogusComment(cp) {
1843
+ const token = this.currentToken;
1844
+ switch (cp) {
1845
+ case $.GREATER_THAN_SIGN: {
1846
+ this.state = State.DATA;
1847
+ this.emitCurrentComment(token);
1848
+ break;
1849
+ }
1850
+ case $.EOF: {
1851
+ this.emitCurrentComment(token);
1852
+ this._emitEOFToken();
1853
+ break;
1854
+ }
1855
+ case $.NULL: {
1856
+ this._err(ERR.unexpectedNullCharacter);
1857
+ token.data += REPLACEMENT_CHARACTER;
1858
+ break;
1859
+ }
1860
+ default: {
1861
+ token.data += String.fromCodePoint(cp);
1862
+ }
1863
+ }
1864
+ }
1865
+ // Markup declaration open state
1866
+ //------------------------------------------------------------------
1867
+ _stateMarkupDeclarationOpen(cp) {
1868
+ if (this._consumeSequenceIfMatch($$.DASH_DASH, true)) {
1869
+ this._createCommentToken($$.DASH_DASH.length + 1);
1870
+ this.state = State.COMMENT_START;
1871
+ }
1872
+ else if (this._consumeSequenceIfMatch($$.DOCTYPE, false)) {
1873
+ // NOTE: Doctypes tokens are created without fixed offsets. We keep track of the moment a doctype *might* start here.
1874
+ this.currentLocation = this.getCurrentLocation($$.DOCTYPE.length + 1);
1875
+ this.state = State.DOCTYPE;
1876
+ }
1877
+ else if (this._consumeSequenceIfMatch($$.CDATA_START, true)) {
1878
+ if (this.inForeignNode) {
1879
+ this.state = State.CDATA_SECTION;
1880
+ }
1881
+ else {
1882
+ this._err(ERR.cdataInHtmlContent);
1883
+ this._createCommentToken($$.CDATA_START.length + 1);
1884
+ this.currentToken.data = '[CDATA[';
1885
+ this.state = State.BOGUS_COMMENT;
1886
+ }
1887
+ }
1888
+ //NOTE: Sequence lookups can be abrupted by hibernation. In that case, lookup
1889
+ //results are no longer valid and we will need to start over.
1890
+ else if (!this._ensureHibernation()) {
1891
+ this._err(ERR.incorrectlyOpenedComment);
1892
+ this._createCommentToken(2);
1893
+ this.state = State.BOGUS_COMMENT;
1894
+ this._stateBogusComment(cp);
1895
+ }
1896
+ }
1897
+ // Comment start state
1898
+ //------------------------------------------------------------------
1899
+ _stateCommentStart(cp) {
1900
+ switch (cp) {
1901
+ case $.HYPHEN_MINUS: {
1902
+ this.state = State.COMMENT_START_DASH;
1903
+ break;
1904
+ }
1905
+ case $.GREATER_THAN_SIGN: {
1906
+ this._err(ERR.abruptClosingOfEmptyComment);
1907
+ this.state = State.DATA;
1908
+ const token = this.currentToken;
1909
+ this.emitCurrentComment(token);
1910
+ break;
1911
+ }
1912
+ default: {
1913
+ this.state = State.COMMENT;
1914
+ this._stateComment(cp);
1915
+ }
1916
+ }
1917
+ }
1918
+ // Comment start dash state
1919
+ //------------------------------------------------------------------
1920
+ _stateCommentStartDash(cp) {
1921
+ const token = this.currentToken;
1922
+ switch (cp) {
1923
+ case $.HYPHEN_MINUS: {
1924
+ this.state = State.COMMENT_END;
1925
+ break;
1926
+ }
1927
+ case $.GREATER_THAN_SIGN: {
1928
+ this._err(ERR.abruptClosingOfEmptyComment);
1929
+ this.state = State.DATA;
1930
+ this.emitCurrentComment(token);
1931
+ break;
1932
+ }
1933
+ case $.EOF: {
1934
+ this._err(ERR.eofInComment);
1935
+ this.emitCurrentComment(token);
1936
+ this._emitEOFToken();
1937
+ break;
1938
+ }
1939
+ default: {
1940
+ token.data += '-';
1941
+ this.state = State.COMMENT;
1942
+ this._stateComment(cp);
1943
+ }
1944
+ }
1945
+ }
1946
+ // Comment state
1947
+ //------------------------------------------------------------------
1948
+ _stateComment(cp) {
1949
+ const token = this.currentToken;
1950
+ switch (cp) {
1951
+ case $.HYPHEN_MINUS: {
1952
+ this.state = State.COMMENT_END_DASH;
1953
+ break;
1954
+ }
1955
+ case $.LESS_THAN_SIGN: {
1956
+ token.data += '<';
1957
+ this.state = State.COMMENT_LESS_THAN_SIGN;
1958
+ break;
1959
+ }
1960
+ case $.NULL: {
1961
+ this._err(ERR.unexpectedNullCharacter);
1962
+ token.data += REPLACEMENT_CHARACTER;
1963
+ break;
1964
+ }
1965
+ case $.EOF: {
1966
+ this._err(ERR.eofInComment);
1967
+ this.emitCurrentComment(token);
1968
+ this._emitEOFToken();
1969
+ break;
1970
+ }
1971
+ default: {
1972
+ token.data += String.fromCodePoint(cp);
1973
+ }
1974
+ }
1975
+ }
1976
+ // Comment less-than sign state
1977
+ //------------------------------------------------------------------
1978
+ _stateCommentLessThanSign(cp) {
1979
+ const token = this.currentToken;
1980
+ switch (cp) {
1981
+ case $.EXCLAMATION_MARK: {
1982
+ token.data += '!';
1983
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG;
1984
+ break;
1985
+ }
1986
+ case $.LESS_THAN_SIGN: {
1987
+ token.data += '<';
1988
+ break;
1989
+ }
1990
+ default: {
1991
+ this.state = State.COMMENT;
1992
+ this._stateComment(cp);
1993
+ }
1994
+ }
1995
+ }
1996
+ // Comment less-than sign bang state
1997
+ //------------------------------------------------------------------
1998
+ _stateCommentLessThanSignBang(cp) {
1999
+ if (cp === $.HYPHEN_MINUS) {
2000
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH;
2001
+ }
2002
+ else {
2003
+ this.state = State.COMMENT;
2004
+ this._stateComment(cp);
2005
+ }
2006
+ }
2007
+ // Comment less-than sign bang dash state
2008
+ //------------------------------------------------------------------
2009
+ _stateCommentLessThanSignBangDash(cp) {
2010
+ if (cp === $.HYPHEN_MINUS) {
2011
+ this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH;
2012
+ }
2013
+ else {
2014
+ this.state = State.COMMENT_END_DASH;
2015
+ this._stateCommentEndDash(cp);
2016
+ }
2017
+ }
2018
+ // Comment less-than sign bang dash dash state
2019
+ //------------------------------------------------------------------
2020
+ _stateCommentLessThanSignBangDashDash(cp) {
2021
+ if (cp !== $.GREATER_THAN_SIGN && cp !== $.EOF) {
2022
+ this._err(ERR.nestedComment);
2023
+ }
2024
+ this.state = State.COMMENT_END;
2025
+ this._stateCommentEnd(cp);
2026
+ }
2027
+ // Comment end dash state
2028
+ //------------------------------------------------------------------
2029
+ _stateCommentEndDash(cp) {
2030
+ const token = this.currentToken;
2031
+ switch (cp) {
2032
+ case $.HYPHEN_MINUS: {
2033
+ this.state = State.COMMENT_END;
2034
+ break;
2035
+ }
2036
+ case $.EOF: {
2037
+ this._err(ERR.eofInComment);
2038
+ this.emitCurrentComment(token);
2039
+ this._emitEOFToken();
2040
+ break;
2041
+ }
2042
+ default: {
2043
+ token.data += '-';
2044
+ this.state = State.COMMENT;
2045
+ this._stateComment(cp);
2046
+ }
2047
+ }
2048
+ }
2049
+ // Comment end state
2050
+ //------------------------------------------------------------------
2051
+ _stateCommentEnd(cp) {
2052
+ const token = this.currentToken;
2053
+ switch (cp) {
2054
+ case $.GREATER_THAN_SIGN: {
2055
+ this.state = State.DATA;
2056
+ this.emitCurrentComment(token);
2057
+ break;
2058
+ }
2059
+ case $.EXCLAMATION_MARK: {
2060
+ this.state = State.COMMENT_END_BANG;
2061
+ break;
2062
+ }
2063
+ case $.HYPHEN_MINUS: {
2064
+ token.data += '-';
2065
+ break;
2066
+ }
2067
+ case $.EOF: {
2068
+ this._err(ERR.eofInComment);
2069
+ this.emitCurrentComment(token);
2070
+ this._emitEOFToken();
2071
+ break;
2072
+ }
2073
+ default: {
2074
+ token.data += '--';
2075
+ this.state = State.COMMENT;
2076
+ this._stateComment(cp);
2077
+ }
2078
+ }
2079
+ }
2080
+ // Comment end bang state
2081
+ //------------------------------------------------------------------
2082
+ _stateCommentEndBang(cp) {
2083
+ const token = this.currentToken;
2084
+ switch (cp) {
2085
+ case $.HYPHEN_MINUS: {
2086
+ token.data += '--!';
2087
+ this.state = State.COMMENT_END_DASH;
2088
+ break;
2089
+ }
2090
+ case $.GREATER_THAN_SIGN: {
2091
+ this._err(ERR.incorrectlyClosedComment);
2092
+ this.state = State.DATA;
2093
+ this.emitCurrentComment(token);
2094
+ break;
2095
+ }
2096
+ case $.EOF: {
2097
+ this._err(ERR.eofInComment);
2098
+ this.emitCurrentComment(token);
2099
+ this._emitEOFToken();
2100
+ break;
2101
+ }
2102
+ default: {
2103
+ token.data += '--!';
2104
+ this.state = State.COMMENT;
2105
+ this._stateComment(cp);
2106
+ }
2107
+ }
2108
+ }
2109
+ // DOCTYPE state
2110
+ //------------------------------------------------------------------
2111
+ _stateDoctype(cp) {
2112
+ switch (cp) {
2113
+ case $.SPACE:
2114
+ case $.LINE_FEED:
2115
+ case $.TABULATION:
2116
+ case $.FORM_FEED: {
2117
+ this.state = State.BEFORE_DOCTYPE_NAME;
2118
+ break;
2119
+ }
2120
+ case $.GREATER_THAN_SIGN: {
2121
+ this.state = State.BEFORE_DOCTYPE_NAME;
2122
+ this._stateBeforeDoctypeName(cp);
2123
+ break;
2124
+ }
2125
+ case $.EOF: {
2126
+ this._err(ERR.eofInDoctype);
2127
+ this._createDoctypeToken(null);
2128
+ const token = this.currentToken;
2129
+ token.forceQuirks = true;
2130
+ this.emitCurrentDoctype(token);
2131
+ this._emitEOFToken();
2132
+ break;
2133
+ }
2134
+ default: {
2135
+ this._err(ERR.missingWhitespaceBeforeDoctypeName);
2136
+ this.state = State.BEFORE_DOCTYPE_NAME;
2137
+ this._stateBeforeDoctypeName(cp);
2138
+ }
2139
+ }
2140
+ }
2141
+ // Before DOCTYPE name state
2142
+ //------------------------------------------------------------------
2143
+ _stateBeforeDoctypeName(cp) {
2144
+ if (isAsciiUpper(cp)) {
2145
+ this._createDoctypeToken(String.fromCharCode(toAsciiLower(cp)));
2146
+ this.state = State.DOCTYPE_NAME;
2147
+ }
2148
+ else
2149
+ switch (cp) {
2150
+ case $.SPACE:
2151
+ case $.LINE_FEED:
2152
+ case $.TABULATION:
2153
+ case $.FORM_FEED: {
2154
+ // Ignore whitespace
2155
+ break;
2156
+ }
2157
+ case $.NULL: {
2158
+ this._err(ERR.unexpectedNullCharacter);
2159
+ this._createDoctypeToken(REPLACEMENT_CHARACTER);
2160
+ this.state = State.DOCTYPE_NAME;
2161
+ break;
2162
+ }
2163
+ case $.GREATER_THAN_SIGN: {
2164
+ this._err(ERR.missingDoctypeName);
2165
+ this._createDoctypeToken(null);
2166
+ const token = this.currentToken;
2167
+ token.forceQuirks = true;
2168
+ this.emitCurrentDoctype(token);
2169
+ this.state = State.DATA;
2170
+ break;
2171
+ }
2172
+ case $.EOF: {
2173
+ this._err(ERR.eofInDoctype);
2174
+ this._createDoctypeToken(null);
2175
+ const token = this.currentToken;
2176
+ token.forceQuirks = true;
2177
+ this.emitCurrentDoctype(token);
2178
+ this._emitEOFToken();
2179
+ break;
2180
+ }
2181
+ default: {
2182
+ this._createDoctypeToken(String.fromCodePoint(cp));
2183
+ this.state = State.DOCTYPE_NAME;
2184
+ }
2185
+ }
2186
+ }
2187
+ // DOCTYPE name state
2188
+ //------------------------------------------------------------------
2189
+ _stateDoctypeName(cp) {
2190
+ const token = this.currentToken;
2191
+ switch (cp) {
2192
+ case $.SPACE:
2193
+ case $.LINE_FEED:
2194
+ case $.TABULATION:
2195
+ case $.FORM_FEED: {
2196
+ this.state = State.AFTER_DOCTYPE_NAME;
2197
+ break;
2198
+ }
2199
+ case $.GREATER_THAN_SIGN: {
2200
+ this.state = State.DATA;
2201
+ this.emitCurrentDoctype(token);
2202
+ break;
2203
+ }
2204
+ case $.NULL: {
2205
+ this._err(ERR.unexpectedNullCharacter);
2206
+ token.name += REPLACEMENT_CHARACTER;
2207
+ break;
2208
+ }
2209
+ case $.EOF: {
2210
+ this._err(ERR.eofInDoctype);
2211
+ token.forceQuirks = true;
2212
+ this.emitCurrentDoctype(token);
2213
+ this._emitEOFToken();
2214
+ break;
2215
+ }
2216
+ default: {
2217
+ token.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
2218
+ }
2219
+ }
2220
+ }
2221
+ // After DOCTYPE name state
2222
+ //------------------------------------------------------------------
2223
+ _stateAfterDoctypeName(cp) {
2224
+ const token = this.currentToken;
2225
+ switch (cp) {
2226
+ case $.SPACE:
2227
+ case $.LINE_FEED:
2228
+ case $.TABULATION:
2229
+ case $.FORM_FEED: {
2230
+ // Ignore whitespace
2231
+ break;
2232
+ }
2233
+ case $.GREATER_THAN_SIGN: {
2234
+ this.state = State.DATA;
2235
+ this.emitCurrentDoctype(token);
2236
+ break;
2237
+ }
2238
+ case $.EOF: {
2239
+ this._err(ERR.eofInDoctype);
2240
+ token.forceQuirks = true;
2241
+ this.emitCurrentDoctype(token);
2242
+ this._emitEOFToken();
2243
+ break;
2244
+ }
2245
+ default:
2246
+ if (this._consumeSequenceIfMatch($$.PUBLIC, false)) {
2247
+ this.state = State.AFTER_DOCTYPE_PUBLIC_KEYWORD;
2248
+ }
2249
+ else if (this._consumeSequenceIfMatch($$.SYSTEM, false)) {
2250
+ this.state = State.AFTER_DOCTYPE_SYSTEM_KEYWORD;
2251
+ }
2252
+ //NOTE: sequence lookup can be abrupted by hibernation. In that case lookup
2253
+ //results are no longer valid and we will need to start over.
2254
+ else if (!this._ensureHibernation()) {
2255
+ this._err(ERR.invalidCharacterSequenceAfterDoctypeName);
2256
+ token.forceQuirks = true;
2257
+ this.state = State.BOGUS_DOCTYPE;
2258
+ this._stateBogusDoctype(cp);
2259
+ }
2260
+ }
2261
+ }
2262
+ // After DOCTYPE public keyword state
2263
+ //------------------------------------------------------------------
2264
+ _stateAfterDoctypePublicKeyword(cp) {
2265
+ const token = this.currentToken;
2266
+ switch (cp) {
2267
+ case $.SPACE:
2268
+ case $.LINE_FEED:
2269
+ case $.TABULATION:
2270
+ case $.FORM_FEED: {
2271
+ this.state = State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER;
2272
+ break;
2273
+ }
2274
+ case $.QUOTATION_MARK: {
2275
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2276
+ token.publicId = '';
2277
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2278
+ break;
2279
+ }
2280
+ case $.APOSTROPHE: {
2281
+ this._err(ERR.missingWhitespaceAfterDoctypePublicKeyword);
2282
+ token.publicId = '';
2283
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2284
+ break;
2285
+ }
2286
+ case $.GREATER_THAN_SIGN: {
2287
+ this._err(ERR.missingDoctypePublicIdentifier);
2288
+ token.forceQuirks = true;
2289
+ this.state = State.DATA;
2290
+ this.emitCurrentDoctype(token);
2291
+ break;
2292
+ }
2293
+ case $.EOF: {
2294
+ this._err(ERR.eofInDoctype);
2295
+ token.forceQuirks = true;
2296
+ this.emitCurrentDoctype(token);
2297
+ this._emitEOFToken();
2298
+ break;
2299
+ }
2300
+ default: {
2301
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2302
+ token.forceQuirks = true;
2303
+ this.state = State.BOGUS_DOCTYPE;
2304
+ this._stateBogusDoctype(cp);
2305
+ }
2306
+ }
2307
+ }
2308
+ // Before DOCTYPE public identifier state
2309
+ //------------------------------------------------------------------
2310
+ _stateBeforeDoctypePublicIdentifier(cp) {
2311
+ const token = this.currentToken;
2312
+ switch (cp) {
2313
+ case $.SPACE:
2314
+ case $.LINE_FEED:
2315
+ case $.TABULATION:
2316
+ case $.FORM_FEED: {
2317
+ // Ignore whitespace
2318
+ break;
2319
+ }
2320
+ case $.QUOTATION_MARK: {
2321
+ token.publicId = '';
2322
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
2323
+ break;
2324
+ }
2325
+ case $.APOSTROPHE: {
2326
+ token.publicId = '';
2327
+ this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
2328
+ break;
2329
+ }
2330
+ case $.GREATER_THAN_SIGN: {
2331
+ this._err(ERR.missingDoctypePublicIdentifier);
2332
+ token.forceQuirks = true;
2333
+ this.state = State.DATA;
2334
+ this.emitCurrentDoctype(token);
2335
+ break;
2336
+ }
2337
+ case $.EOF: {
2338
+ this._err(ERR.eofInDoctype);
2339
+ token.forceQuirks = true;
2340
+ this.emitCurrentDoctype(token);
2341
+ this._emitEOFToken();
2342
+ break;
2343
+ }
2344
+ default: {
2345
+ this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
2346
+ token.forceQuirks = true;
2347
+ this.state = State.BOGUS_DOCTYPE;
2348
+ this._stateBogusDoctype(cp);
2349
+ }
2350
+ }
2351
+ }
2352
+ // DOCTYPE public identifier (double-quoted) state
2353
+ //------------------------------------------------------------------
2354
+ _stateDoctypePublicIdentifierDoubleQuoted(cp) {
2355
+ const token = this.currentToken;
2356
+ switch (cp) {
2357
+ case $.QUOTATION_MARK: {
2358
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2359
+ break;
2360
+ }
2361
+ case $.NULL: {
2362
+ this._err(ERR.unexpectedNullCharacter);
2363
+ token.publicId += REPLACEMENT_CHARACTER;
2364
+ break;
2365
+ }
2366
+ case $.GREATER_THAN_SIGN: {
2367
+ this._err(ERR.abruptDoctypePublicIdentifier);
2368
+ token.forceQuirks = true;
2369
+ this.emitCurrentDoctype(token);
2370
+ this.state = State.DATA;
2371
+ break;
2372
+ }
2373
+ case $.EOF: {
2374
+ this._err(ERR.eofInDoctype);
2375
+ token.forceQuirks = true;
2376
+ this.emitCurrentDoctype(token);
2377
+ this._emitEOFToken();
2378
+ break;
2379
+ }
2380
+ default: {
2381
+ token.publicId += String.fromCodePoint(cp);
2382
+ }
2383
+ }
2384
+ }
2385
+ // DOCTYPE public identifier (single-quoted) state
2386
+ //------------------------------------------------------------------
2387
+ _stateDoctypePublicIdentifierSingleQuoted(cp) {
2388
+ const token = this.currentToken;
2389
+ switch (cp) {
2390
+ case $.APOSTROPHE: {
2391
+ this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
2392
+ break;
2393
+ }
2394
+ case $.NULL: {
2395
+ this._err(ERR.unexpectedNullCharacter);
2396
+ token.publicId += REPLACEMENT_CHARACTER;
2397
+ break;
2398
+ }
2399
+ case $.GREATER_THAN_SIGN: {
2400
+ this._err(ERR.abruptDoctypePublicIdentifier);
2401
+ token.forceQuirks = true;
2402
+ this.emitCurrentDoctype(token);
2403
+ this.state = State.DATA;
2404
+ break;
2405
+ }
2406
+ case $.EOF: {
2407
+ this._err(ERR.eofInDoctype);
2408
+ token.forceQuirks = true;
2409
+ this.emitCurrentDoctype(token);
2410
+ this._emitEOFToken();
2411
+ break;
2412
+ }
2413
+ default: {
2414
+ token.publicId += String.fromCodePoint(cp);
2415
+ }
2416
+ }
2417
+ }
2418
+ // After DOCTYPE public identifier state
2419
+ //------------------------------------------------------------------
2420
+ _stateAfterDoctypePublicIdentifier(cp) {
2421
+ const token = this.currentToken;
2422
+ switch (cp) {
2423
+ case $.SPACE:
2424
+ case $.LINE_FEED:
2425
+ case $.TABULATION:
2426
+ case $.FORM_FEED: {
2427
+ this.state = State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS;
2428
+ break;
2429
+ }
2430
+ case $.GREATER_THAN_SIGN: {
2431
+ this.state = State.DATA;
2432
+ this.emitCurrentDoctype(token);
2433
+ break;
2434
+ }
2435
+ case $.QUOTATION_MARK: {
2436
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2437
+ token.systemId = '';
2438
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2439
+ break;
2440
+ }
2441
+ case $.APOSTROPHE: {
2442
+ this._err(ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
2443
+ token.systemId = '';
2444
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2445
+ break;
2446
+ }
2447
+ case $.EOF: {
2448
+ this._err(ERR.eofInDoctype);
2449
+ token.forceQuirks = true;
2450
+ this.emitCurrentDoctype(token);
2451
+ this._emitEOFToken();
2452
+ break;
2453
+ }
2454
+ default: {
2455
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2456
+ token.forceQuirks = true;
2457
+ this.state = State.BOGUS_DOCTYPE;
2458
+ this._stateBogusDoctype(cp);
2459
+ }
2460
+ }
2461
+ }
2462
+ // Between DOCTYPE public and system identifiers state
2463
+ //------------------------------------------------------------------
2464
+ _stateBetweenDoctypePublicAndSystemIdentifiers(cp) {
2465
+ const token = this.currentToken;
2466
+ switch (cp) {
2467
+ case $.SPACE:
2468
+ case $.LINE_FEED:
2469
+ case $.TABULATION:
2470
+ case $.FORM_FEED: {
2471
+ // Ignore whitespace
2472
+ break;
2473
+ }
2474
+ case $.GREATER_THAN_SIGN: {
2475
+ this.emitCurrentDoctype(token);
2476
+ this.state = State.DATA;
2477
+ break;
2478
+ }
2479
+ case $.QUOTATION_MARK: {
2480
+ token.systemId = '';
2481
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2482
+ break;
2483
+ }
2484
+ case $.APOSTROPHE: {
2485
+ token.systemId = '';
2486
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2487
+ break;
2488
+ }
2489
+ case $.EOF: {
2490
+ this._err(ERR.eofInDoctype);
2491
+ token.forceQuirks = true;
2492
+ this.emitCurrentDoctype(token);
2493
+ this._emitEOFToken();
2494
+ break;
2495
+ }
2496
+ default: {
2497
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2498
+ token.forceQuirks = true;
2499
+ this.state = State.BOGUS_DOCTYPE;
2500
+ this._stateBogusDoctype(cp);
2501
+ }
2502
+ }
2503
+ }
2504
+ // After DOCTYPE system keyword state
2505
+ //------------------------------------------------------------------
2506
+ _stateAfterDoctypeSystemKeyword(cp) {
2507
+ const token = this.currentToken;
2508
+ switch (cp) {
2509
+ case $.SPACE:
2510
+ case $.LINE_FEED:
2511
+ case $.TABULATION:
2512
+ case $.FORM_FEED: {
2513
+ this.state = State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER;
2514
+ break;
2515
+ }
2516
+ case $.QUOTATION_MARK: {
2517
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2518
+ token.systemId = '';
2519
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2520
+ break;
2521
+ }
2522
+ case $.APOSTROPHE: {
2523
+ this._err(ERR.missingWhitespaceAfterDoctypeSystemKeyword);
2524
+ token.systemId = '';
2525
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2526
+ break;
2527
+ }
2528
+ case $.GREATER_THAN_SIGN: {
2529
+ this._err(ERR.missingDoctypeSystemIdentifier);
2530
+ token.forceQuirks = true;
2531
+ this.state = State.DATA;
2532
+ this.emitCurrentDoctype(token);
2533
+ break;
2534
+ }
2535
+ case $.EOF: {
2536
+ this._err(ERR.eofInDoctype);
2537
+ token.forceQuirks = true;
2538
+ this.emitCurrentDoctype(token);
2539
+ this._emitEOFToken();
2540
+ break;
2541
+ }
2542
+ default: {
2543
+ this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
2544
+ token.forceQuirks = true;
2545
+ this.state = State.BOGUS_DOCTYPE;
2546
+ this._stateBogusDoctype(cp);
2547
+ }
2548
+ }
2549
+ }
2550
+ // Before DOCTYPE system identifier state
2551
+ //------------------------------------------------------------------
2552
+ _stateBeforeDoctypeSystemIdentifier(cp) {
2553
+ const token = this.currentToken;
2554
+ switch (cp) {
2555
+ case $.SPACE:
2556
+ case $.LINE_FEED:
2557
+ case $.TABULATION:
2558
+ case $.FORM_FEED: {
2559
+ // Ignore whitespace
2560
+ break;
2561
+ }
2562
+ case $.QUOTATION_MARK: {
2563
+ token.systemId = '';
2564
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
2565
+ break;
2566
+ }
2567
+ case $.APOSTROPHE: {
2568
+ token.systemId = '';
2569
+ this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
2570
+ break;
2571
+ }
2572
+ case $.GREATER_THAN_SIGN: {
2573
+ this._err(ERR.missingDoctypeSystemIdentifier);
2574
+ token.forceQuirks = true;
2575
+ this.state = State.DATA;
2576
+ this.emitCurrentDoctype(token);
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.missingQuoteBeforeDoctypeSystemIdentifier);
2588
+ token.forceQuirks = true;
2589
+ this.state = State.BOGUS_DOCTYPE;
2590
+ this._stateBogusDoctype(cp);
2591
+ }
2592
+ }
2593
+ }
2594
+ // DOCTYPE system identifier (double-quoted) state
2595
+ //------------------------------------------------------------------
2596
+ _stateDoctypeSystemIdentifierDoubleQuoted(cp) {
2597
+ const token = this.currentToken;
2598
+ switch (cp) {
2599
+ case $.QUOTATION_MARK: {
2600
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2601
+ break;
2602
+ }
2603
+ case $.NULL: {
2604
+ this._err(ERR.unexpectedNullCharacter);
2605
+ token.systemId += REPLACEMENT_CHARACTER;
2606
+ break;
2607
+ }
2608
+ case $.GREATER_THAN_SIGN: {
2609
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2610
+ token.forceQuirks = true;
2611
+ this.emitCurrentDoctype(token);
2612
+ this.state = State.DATA;
2613
+ break;
2614
+ }
2615
+ case $.EOF: {
2616
+ this._err(ERR.eofInDoctype);
2617
+ token.forceQuirks = true;
2618
+ this.emitCurrentDoctype(token);
2619
+ this._emitEOFToken();
2620
+ break;
2621
+ }
2622
+ default: {
2623
+ token.systemId += String.fromCodePoint(cp);
2624
+ }
2625
+ }
2626
+ }
2627
+ // DOCTYPE system identifier (single-quoted) state
2628
+ //------------------------------------------------------------------
2629
+ _stateDoctypeSystemIdentifierSingleQuoted(cp) {
2630
+ const token = this.currentToken;
2631
+ switch (cp) {
2632
+ case $.APOSTROPHE: {
2633
+ this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
2634
+ break;
2635
+ }
2636
+ case $.NULL: {
2637
+ this._err(ERR.unexpectedNullCharacter);
2638
+ token.systemId += REPLACEMENT_CHARACTER;
2639
+ break;
2640
+ }
2641
+ case $.GREATER_THAN_SIGN: {
2642
+ this._err(ERR.abruptDoctypeSystemIdentifier);
2643
+ token.forceQuirks = true;
2644
+ this.emitCurrentDoctype(token);
2645
+ this.state = State.DATA;
2646
+ break;
2647
+ }
2648
+ case $.EOF: {
2649
+ this._err(ERR.eofInDoctype);
2650
+ token.forceQuirks = true;
2651
+ this.emitCurrentDoctype(token);
2652
+ this._emitEOFToken();
2653
+ break;
2654
+ }
2655
+ default: {
2656
+ token.systemId += String.fromCodePoint(cp);
2657
+ }
2658
+ }
2659
+ }
2660
+ // After DOCTYPE system identifier state
2661
+ //------------------------------------------------------------------
2662
+ _stateAfterDoctypeSystemIdentifier(cp) {
2663
+ const token = this.currentToken;
2664
+ switch (cp) {
2665
+ case $.SPACE:
2666
+ case $.LINE_FEED:
2667
+ case $.TABULATION:
2668
+ case $.FORM_FEED: {
2669
+ // Ignore whitespace
2670
+ break;
2671
+ }
2672
+ case $.GREATER_THAN_SIGN: {
2673
+ this.emitCurrentDoctype(token);
2674
+ this.state = State.DATA;
2675
+ break;
2676
+ }
2677
+ case $.EOF: {
2678
+ this._err(ERR.eofInDoctype);
2679
+ token.forceQuirks = true;
2680
+ this.emitCurrentDoctype(token);
2681
+ this._emitEOFToken();
2682
+ break;
2683
+ }
2684
+ default: {
2685
+ this._err(ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
2686
+ this.state = State.BOGUS_DOCTYPE;
2687
+ this._stateBogusDoctype(cp);
2688
+ }
2689
+ }
2690
+ }
2691
+ // Bogus DOCTYPE state
2692
+ //------------------------------------------------------------------
2693
+ _stateBogusDoctype(cp) {
2694
+ const token = this.currentToken;
2695
+ switch (cp) {
2696
+ case $.GREATER_THAN_SIGN: {
2697
+ this.emitCurrentDoctype(token);
2698
+ this.state = State.DATA;
2699
+ break;
2700
+ }
2701
+ case $.NULL: {
2702
+ this._err(ERR.unexpectedNullCharacter);
2703
+ break;
2704
+ }
2705
+ case $.EOF: {
2706
+ this.emitCurrentDoctype(token);
2707
+ this._emitEOFToken();
2708
+ break;
2709
+ }
2710
+ default:
2711
+ // Do nothing
2712
+ }
2713
+ }
2714
+ // CDATA section state
2715
+ //------------------------------------------------------------------
2716
+ _stateCdataSection(cp) {
2717
+ switch (cp) {
2718
+ case $.RIGHT_SQUARE_BRACKET: {
2719
+ this.state = State.CDATA_SECTION_BRACKET;
2720
+ break;
2721
+ }
2722
+ case $.EOF: {
2723
+ this._err(ERR.eofInCdata);
2724
+ this._emitEOFToken();
2725
+ break;
2726
+ }
2727
+ default: {
2728
+ this._emitCodePoint(cp);
2729
+ }
2730
+ }
2731
+ }
2732
+ // CDATA section bracket state
2733
+ //------------------------------------------------------------------
2734
+ _stateCdataSectionBracket(cp) {
2735
+ if (cp === $.RIGHT_SQUARE_BRACKET) {
2736
+ this.state = State.CDATA_SECTION_END;
2737
+ }
2738
+ else {
2739
+ this._emitChars(']');
2740
+ this.state = State.CDATA_SECTION;
2741
+ this._stateCdataSection(cp);
2742
+ }
2743
+ }
2744
+ // CDATA section end state
2745
+ //------------------------------------------------------------------
2746
+ _stateCdataSectionEnd(cp) {
2747
+ switch (cp) {
2748
+ case $.GREATER_THAN_SIGN: {
2749
+ this.state = State.DATA;
2750
+ break;
2751
+ }
2752
+ case $.RIGHT_SQUARE_BRACKET: {
2753
+ this._emitChars(']');
2754
+ break;
2755
+ }
2756
+ default: {
2757
+ this._emitChars(']]');
2758
+ this.state = State.CDATA_SECTION;
2759
+ this._stateCdataSection(cp);
2760
+ }
2761
+ }
2762
+ }
2763
+ // Character reference state
2764
+ //------------------------------------------------------------------
2765
+ _stateCharacterReference(cp) {
2766
+ if (cp === $.NUMBER_SIGN) {
2767
+ this.state = State.NUMERIC_CHARACTER_REFERENCE;
2768
+ }
2769
+ else if (isAsciiAlphaNumeric(cp)) {
2770
+ this.state = State.NAMED_CHARACTER_REFERENCE;
2771
+ this._stateNamedCharacterReference(cp);
2772
+ }
2773
+ else {
2774
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2775
+ this._reconsumeInState(this.returnState);
2776
+ }
2777
+ }
2778
+ // Named character reference state
2779
+ //------------------------------------------------------------------
2780
+ _stateNamedCharacterReference(cp) {
2781
+ const matchResult = this._matchNamedCharacterReference(cp);
2782
+ //NOTE: Matching can be abrupted by hibernation. In that case, match
2783
+ //results are no longer valid and we will need to start over.
2784
+ if (this._ensureHibernation()) {
2785
+ // Stay in the state, try again.
2786
+ }
2787
+ else if (matchResult) {
2788
+ for (let i = 0; i < matchResult.length; i++) {
2789
+ this._flushCodePointConsumedAsCharacterReference(matchResult[i]);
2790
+ }
2791
+ this.state = this.returnState;
2792
+ }
2793
+ else {
2794
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2795
+ this.state = State.AMBIGUOUS_AMPERSAND;
2796
+ }
2797
+ }
2798
+ // Ambiguos ampersand state
2799
+ //------------------------------------------------------------------
2800
+ _stateAmbiguousAmpersand(cp) {
2801
+ if (isAsciiAlphaNumeric(cp)) {
2802
+ this._flushCodePointConsumedAsCharacterReference(cp);
2803
+ }
2804
+ else {
2805
+ if (cp === $.SEMICOLON) {
2806
+ this._err(ERR.unknownNamedCharacterReference);
2807
+ }
2808
+ this._reconsumeInState(this.returnState);
2809
+ }
2810
+ }
2811
+ // Numeric character reference state
2812
+ //------------------------------------------------------------------
2813
+ _stateNumericCharacterReference(cp) {
2814
+ this.charRefCode = 0;
2815
+ if (cp === $.LATIN_SMALL_X || cp === $.LATIN_CAPITAL_X) {
2816
+ this.state = State.HEXADEMICAL_CHARACTER_REFERENCE_START;
2817
+ }
2818
+ else {
2819
+ this.state = State.DECIMAL_CHARACTER_REFERENCE_START;
2820
+ this._stateDecimalCharacterReferenceStart(cp);
2821
+ }
2822
+ }
2823
+ // Hexademical character reference start state
2824
+ //------------------------------------------------------------------
2825
+ _stateHexademicalCharacterReferenceStart(cp) {
2826
+ if (isAsciiHexDigit(cp)) {
2827
+ this.state = State.HEXADEMICAL_CHARACTER_REFERENCE;
2828
+ this._stateHexademicalCharacterReference(cp);
2829
+ }
2830
+ else {
2831
+ this._err(ERR.absenceOfDigitsInNumericCharacterReference);
2832
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2833
+ this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
2834
+ this._unconsume(2);
2835
+ this.state = this.returnState;
2836
+ }
2837
+ }
2838
+ // Decimal character reference start state
2839
+ //------------------------------------------------------------------
2840
+ _stateDecimalCharacterReferenceStart(cp) {
2841
+ if (isAsciiDigit(cp)) {
2842
+ this.state = State.DECIMAL_CHARACTER_REFERENCE;
2843
+ this._stateDecimalCharacterReference(cp);
2844
+ }
2845
+ else {
2846
+ this._err(ERR.absenceOfDigitsInNumericCharacterReference);
2847
+ this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
2848
+ this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
2849
+ this._reconsumeInState(this.returnState);
2850
+ }
2851
+ }
2852
+ // Hexademical character reference state
2853
+ //------------------------------------------------------------------
2854
+ _stateHexademicalCharacterReference(cp) {
2855
+ if (isAsciiUpperHexDigit(cp)) {
2856
+ this.charRefCode = this.charRefCode * 16 + cp - 0x37;
2857
+ }
2858
+ else if (isAsciiLowerHexDigit(cp)) {
2859
+ this.charRefCode = this.charRefCode * 16 + cp - 0x57;
2860
+ }
2861
+ else if (isAsciiDigit(cp)) {
2862
+ this.charRefCode = this.charRefCode * 16 + cp - 0x30;
2863
+ }
2864
+ else if (cp === $.SEMICOLON) {
2865
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2866
+ }
2867
+ else {
2868
+ this._err(ERR.missingSemicolonAfterCharacterReference);
2869
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2870
+ this._stateNumericCharacterReferenceEnd();
2871
+ }
2872
+ }
2873
+ // Decimal character reference state
2874
+ //------------------------------------------------------------------
2875
+ _stateDecimalCharacterReference(cp) {
2876
+ if (isAsciiDigit(cp)) {
2877
+ this.charRefCode = this.charRefCode * 10 + cp - 0x30;
2878
+ }
2879
+ else if (cp === $.SEMICOLON) {
2880
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2881
+ }
2882
+ else {
2883
+ this._err(ERR.missingSemicolonAfterCharacterReference);
2884
+ this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
2885
+ this._stateNumericCharacterReferenceEnd();
2886
+ }
2887
+ }
2888
+ // Numeric character reference end state
2889
+ //------------------------------------------------------------------
2890
+ _stateNumericCharacterReferenceEnd() {
2891
+ if (this.charRefCode === $.NULL) {
2892
+ this._err(ERR.nullCharacterReference);
2893
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2894
+ }
2895
+ else if (this.charRefCode > 1114111) {
2896
+ this._err(ERR.characterReferenceOutsideUnicodeRange);
2897
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2898
+ }
2899
+ else if (isSurrogate(this.charRefCode)) {
2900
+ this._err(ERR.surrogateCharacterReference);
2901
+ this.charRefCode = $.REPLACEMENT_CHARACTER;
2902
+ }
2903
+ else if (isUndefinedCodePoint(this.charRefCode)) {
2904
+ this._err(ERR.noncharacterCharacterReference);
2905
+ }
2906
+ else if (isControlCodePoint(this.charRefCode) || this.charRefCode === $.CARRIAGE_RETURN) {
2907
+ this._err(ERR.controlCharacterReference);
2908
+ const replacement = C1_CONTROLS_REFERENCE_REPLACEMENTS.get(this.charRefCode);
2909
+ if (replacement !== undefined) {
2910
+ this.charRefCode = replacement;
2911
+ }
2912
+ }
2913
+ this._flushCodePointConsumedAsCharacterReference(this.charRefCode);
2914
+ this._reconsumeInState(this.returnState);
2915
+ }
2916
+ }
2917
+ //# sourceMappingURL=index.js.map