aviau | 039001d | 2016-09-29 16:39:05 -0400 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2016 SoapBox Innovations Inc. |
| 3 | * |
| 4 | * Permission is hereby granted, free of charge, to any person obtaining a copy |
| 5 | * of this software and associated documentation files (the "Software"), to deal |
| 6 | * in the Software without restriction, including without limitation the rights |
| 7 | * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell |
| 8 | * copies of the Software, and to permit persons to whom the Software is |
| 9 | * furnished to do so, subject to the following conditions: |
| 10 | * |
| 11 | * The above copyright notice and this permission notice shall be included in |
| 12 | * all copies or substantial portions of the Software. |
| 13 | * |
| 14 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 15 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 16 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 17 | * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 18 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
| 19 | * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN |
| 20 | * THE SOFTWARE. |
| 21 | */ |
| 22 | |
| 23 | 'use strict'; |
| 24 | |
| 25 | ;(function (window, linkify) { |
| 26 | var linkifyHtml = function (linkify) { |
| 27 | 'use strict'; |
| 28 | |
| 29 | var HTML5NamedCharRefs = {}; |
| 30 | |
| 31 | function EntityParser(named) { |
| 32 | this.named = named; |
| 33 | } |
| 34 | |
| 35 | var HEXCHARCODE = /^#[xX]([A-Fa-f0-9]+)$/; |
| 36 | var CHARCODE = /^#([0-9]+)$/; |
| 37 | var NAMED = /^([A-Za-z0-9]+)$/; |
| 38 | |
| 39 | EntityParser.prototype.parse = function (entity) { |
| 40 | if (!entity) { |
| 41 | return; |
| 42 | } |
| 43 | var matches = entity.match(HEXCHARCODE); |
| 44 | if (matches) { |
| 45 | return '&#x' + matches[1] + ';'; |
| 46 | } |
| 47 | matches = entity.match(CHARCODE); |
| 48 | if (matches) { |
| 49 | return '&#' + matches[1] + ';'; |
| 50 | } |
| 51 | matches = entity.match(NAMED); |
| 52 | if (matches) { |
| 53 | return '&' + matches[1] + ';'; |
| 54 | } |
| 55 | }; |
| 56 | |
| 57 | var WSP = /[\t\n\f ]/; |
| 58 | var ALPHA = /[A-Za-z]/; |
| 59 | var CRLF = /\r\n?/g; |
| 60 | |
| 61 | function isSpace(char) { |
| 62 | return WSP.test(char); |
| 63 | } |
| 64 | |
| 65 | function isAlpha(char) { |
| 66 | return ALPHA.test(char); |
| 67 | } |
| 68 | |
| 69 | function preprocessInput(input) { |
| 70 | return input.replace(CRLF, "\n"); |
| 71 | } |
| 72 | |
| 73 | function EventedTokenizer(delegate, entityParser) { |
| 74 | this.delegate = delegate; |
| 75 | this.entityParser = entityParser; |
| 76 | |
| 77 | this.state = null; |
| 78 | this.input = null; |
| 79 | |
| 80 | this.index = -1; |
| 81 | this.line = -1; |
| 82 | this.column = -1; |
| 83 | this.tagLine = -1; |
| 84 | this.tagColumn = -1; |
| 85 | |
| 86 | this.reset(); |
| 87 | } |
| 88 | |
| 89 | EventedTokenizer.prototype = { |
| 90 | reset: function reset() { |
| 91 | this.state = 'beforeData'; |
| 92 | this.input = ''; |
| 93 | |
| 94 | this.index = 0; |
| 95 | this.line = 1; |
| 96 | this.column = 0; |
| 97 | |
| 98 | this.tagLine = -1; |
| 99 | this.tagColumn = -1; |
| 100 | |
| 101 | this.delegate.reset(); |
| 102 | }, |
| 103 | |
| 104 | tokenize: function tokenize(input) { |
| 105 | this.reset(); |
| 106 | this.tokenizePart(input); |
| 107 | this.tokenizeEOF(); |
| 108 | }, |
| 109 | |
| 110 | tokenizePart: function tokenizePart(input) { |
| 111 | this.input += preprocessInput(input); |
| 112 | |
| 113 | while (this.index < this.input.length) { |
| 114 | this.states[this.state].call(this); |
| 115 | } |
| 116 | }, |
| 117 | |
| 118 | tokenizeEOF: function tokenizeEOF() { |
| 119 | this.flushData(); |
| 120 | }, |
| 121 | |
| 122 | flushData: function flushData() { |
| 123 | if (this.state === 'data') { |
| 124 | this.delegate.finishData(); |
| 125 | this.state = 'beforeData'; |
| 126 | } |
| 127 | }, |
| 128 | |
| 129 | peek: function peek() { |
| 130 | return this.input.charAt(this.index); |
| 131 | }, |
| 132 | |
| 133 | consume: function consume() { |
| 134 | var char = this.peek(); |
| 135 | |
| 136 | this.index++; |
| 137 | |
| 138 | if (char === "\n") { |
| 139 | this.line++; |
| 140 | this.column = 0; |
| 141 | } else { |
| 142 | this.column++; |
| 143 | } |
| 144 | |
| 145 | return char; |
| 146 | }, |
| 147 | |
| 148 | consumeCharRef: function consumeCharRef() { |
| 149 | var endIndex = this.input.indexOf(';', this.index); |
| 150 | if (endIndex === -1) { |
| 151 | return; |
| 152 | } |
| 153 | var entity = this.input.slice(this.index, endIndex); |
| 154 | var chars = this.entityParser.parse(entity); |
| 155 | if (chars) { |
| 156 | this.index = endIndex + 1; |
| 157 | return chars; |
| 158 | } |
| 159 | }, |
| 160 | |
| 161 | markTagStart: function markTagStart() { |
| 162 | this.tagLine = this.line; |
| 163 | this.tagColumn = this.column; |
| 164 | }, |
| 165 | |
| 166 | states: { |
| 167 | beforeData: function beforeData() { |
| 168 | var char = this.peek(); |
| 169 | |
| 170 | if (char === "<") { |
| 171 | this.state = 'tagOpen'; |
| 172 | this.markTagStart(); |
| 173 | this.consume(); |
| 174 | } else { |
| 175 | this.state = 'data'; |
| 176 | this.delegate.beginData(); |
| 177 | } |
| 178 | }, |
| 179 | |
| 180 | data: function data() { |
| 181 | var char = this.peek(); |
| 182 | |
| 183 | if (char === "<") { |
| 184 | this.delegate.finishData(); |
| 185 | this.state = 'tagOpen'; |
| 186 | this.markTagStart(); |
| 187 | this.consume(); |
| 188 | } else if (char === "&") { |
| 189 | this.consume(); |
| 190 | this.delegate.appendToData(this.consumeCharRef() || "&"); |
| 191 | } else { |
| 192 | this.consume(); |
| 193 | this.delegate.appendToData(char); |
| 194 | } |
| 195 | }, |
| 196 | |
| 197 | tagOpen: function tagOpen() { |
| 198 | var char = this.consume(); |
| 199 | |
| 200 | if (char === "!") { |
| 201 | this.state = 'markupDeclaration'; |
| 202 | } else if (char === "/") { |
| 203 | this.state = 'endTagOpen'; |
| 204 | } else if (isAlpha(char)) { |
| 205 | this.state = 'tagName'; |
| 206 | this.delegate.beginStartTag(); |
| 207 | this.delegate.appendToTagName(char.toLowerCase()); |
| 208 | } |
| 209 | }, |
| 210 | |
| 211 | markupDeclaration: function markupDeclaration() { |
| 212 | var char = this.consume(); |
| 213 | |
| 214 | if (char === "-" && this.input.charAt(this.index) === "-") { |
| 215 | this.index++; |
| 216 | this.state = 'commentStart'; |
| 217 | this.delegate.beginComment(); |
| 218 | } |
| 219 | }, |
| 220 | |
| 221 | commentStart: function commentStart() { |
| 222 | var char = this.consume(); |
| 223 | |
| 224 | if (char === "-") { |
| 225 | this.state = 'commentStartDash'; |
| 226 | } else if (char === ">") { |
| 227 | this.delegate.finishComment(); |
| 228 | this.state = 'beforeData'; |
| 229 | } else { |
| 230 | this.delegate.appendToCommentData(char); |
| 231 | this.state = 'comment'; |
| 232 | } |
| 233 | }, |
| 234 | |
| 235 | commentStartDash: function commentStartDash() { |
| 236 | var char = this.consume(); |
| 237 | |
| 238 | if (char === "-") { |
| 239 | this.state = 'commentEnd'; |
| 240 | } else if (char === ">") { |
| 241 | this.delegate.finishComment(); |
| 242 | this.state = 'beforeData'; |
| 243 | } else { |
| 244 | this.delegate.appendToCommentData("-"); |
| 245 | this.state = 'comment'; |
| 246 | } |
| 247 | }, |
| 248 | |
| 249 | comment: function comment() { |
| 250 | var char = this.consume(); |
| 251 | |
| 252 | if (char === "-") { |
| 253 | this.state = 'commentEndDash'; |
| 254 | } else { |
| 255 | this.delegate.appendToCommentData(char); |
| 256 | } |
| 257 | }, |
| 258 | |
| 259 | commentEndDash: function commentEndDash() { |
| 260 | var char = this.consume(); |
| 261 | |
| 262 | if (char === "-") { |
| 263 | this.state = 'commentEnd'; |
| 264 | } else { |
| 265 | this.delegate.appendToCommentData("-" + char); |
| 266 | this.state = 'comment'; |
| 267 | } |
| 268 | }, |
| 269 | |
| 270 | commentEnd: function commentEnd() { |
| 271 | var char = this.consume(); |
| 272 | |
| 273 | if (char === ">") { |
| 274 | this.delegate.finishComment(); |
| 275 | this.state = 'beforeData'; |
| 276 | } else { |
| 277 | this.delegate.appendToCommentData("--" + char); |
| 278 | this.state = 'comment'; |
| 279 | } |
| 280 | }, |
| 281 | |
| 282 | tagName: function tagName() { |
| 283 | var char = this.consume(); |
| 284 | |
| 285 | if (isSpace(char)) { |
| 286 | this.state = 'beforeAttributeName'; |
| 287 | } else if (char === "/") { |
| 288 | this.state = 'selfClosingStartTag'; |
| 289 | } else if (char === ">") { |
| 290 | this.delegate.finishTag(); |
| 291 | this.state = 'beforeData'; |
| 292 | } else { |
| 293 | this.delegate.appendToTagName(char); |
| 294 | } |
| 295 | }, |
| 296 | |
| 297 | beforeAttributeName: function beforeAttributeName() { |
| 298 | var char = this.consume(); |
| 299 | |
| 300 | if (isSpace(char)) { |
| 301 | return; |
| 302 | } else if (char === "/") { |
| 303 | this.state = 'selfClosingStartTag'; |
| 304 | } else if (char === ">") { |
| 305 | this.delegate.finishTag(); |
| 306 | this.state = 'beforeData'; |
| 307 | } else { |
| 308 | this.state = 'attributeName'; |
| 309 | this.delegate.beginAttribute(); |
| 310 | this.delegate.appendToAttributeName(char); |
| 311 | } |
| 312 | }, |
| 313 | |
| 314 | attributeName: function attributeName() { |
| 315 | var char = this.consume(); |
| 316 | |
| 317 | if (isSpace(char)) { |
| 318 | this.state = 'afterAttributeName'; |
| 319 | } else if (char === "/") { |
| 320 | this.delegate.beginAttributeValue(false); |
| 321 | this.delegate.finishAttributeValue(); |
| 322 | this.state = 'selfClosingStartTag'; |
| 323 | } else if (char === "=") { |
| 324 | this.state = 'beforeAttributeValue'; |
| 325 | } else if (char === ">") { |
| 326 | this.delegate.beginAttributeValue(false); |
| 327 | this.delegate.finishAttributeValue(); |
| 328 | this.delegate.finishTag(); |
| 329 | this.state = 'beforeData'; |
| 330 | } else { |
| 331 | this.delegate.appendToAttributeName(char); |
| 332 | } |
| 333 | }, |
| 334 | |
| 335 | afterAttributeName: function afterAttributeName() { |
| 336 | var char = this.consume(); |
| 337 | |
| 338 | if (isSpace(char)) { |
| 339 | return; |
| 340 | } else if (char === "/") { |
| 341 | this.delegate.beginAttributeValue(false); |
| 342 | this.delegate.finishAttributeValue(); |
| 343 | this.state = 'selfClosingStartTag'; |
| 344 | } else if (char === "=") { |
| 345 | this.state = 'beforeAttributeValue'; |
| 346 | } else if (char === ">") { |
| 347 | this.delegate.beginAttributeValue(false); |
| 348 | this.delegate.finishAttributeValue(); |
| 349 | this.delegate.finishTag(); |
| 350 | this.state = 'beforeData'; |
| 351 | } else { |
| 352 | this.delegate.beginAttributeValue(false); |
| 353 | this.delegate.finishAttributeValue(); |
| 354 | this.state = 'attributeName'; |
| 355 | this.delegate.beginAttribute(); |
| 356 | this.delegate.appendToAttributeName(char); |
| 357 | } |
| 358 | }, |
| 359 | |
| 360 | beforeAttributeValue: function beforeAttributeValue() { |
| 361 | var char = this.consume(); |
| 362 | |
| 363 | if (isSpace(char)) {} else if (char === '"') { |
| 364 | this.state = 'attributeValueDoubleQuoted'; |
| 365 | this.delegate.beginAttributeValue(true); |
| 366 | } else if (char === "'") { |
| 367 | this.state = 'attributeValueSingleQuoted'; |
| 368 | this.delegate.beginAttributeValue(true); |
| 369 | } else if (char === ">") { |
| 370 | this.delegate.beginAttributeValue(false); |
| 371 | this.delegate.finishAttributeValue(); |
| 372 | this.delegate.finishTag(); |
| 373 | this.state = 'beforeData'; |
| 374 | } else { |
| 375 | this.state = 'attributeValueUnquoted'; |
| 376 | this.delegate.beginAttributeValue(false); |
| 377 | this.delegate.appendToAttributeValue(char); |
| 378 | } |
| 379 | }, |
| 380 | |
| 381 | attributeValueDoubleQuoted: function attributeValueDoubleQuoted() { |
| 382 | var char = this.consume(); |
| 383 | |
| 384 | if (char === '"') { |
| 385 | this.delegate.finishAttributeValue(); |
| 386 | this.state = 'afterAttributeValueQuoted'; |
| 387 | } else if (char === "&") { |
| 388 | this.delegate.appendToAttributeValue(this.consumeCharRef('"') || "&"); |
| 389 | } else { |
| 390 | this.delegate.appendToAttributeValue(char); |
| 391 | } |
| 392 | }, |
| 393 | |
| 394 | attributeValueSingleQuoted: function attributeValueSingleQuoted() { |
| 395 | var char = this.consume(); |
| 396 | |
| 397 | if (char === "'") { |
| 398 | this.delegate.finishAttributeValue(); |
| 399 | this.state = 'afterAttributeValueQuoted'; |
| 400 | } else if (char === "&") { |
| 401 | this.delegate.appendToAttributeValue(this.consumeCharRef("'") || "&"); |
| 402 | } else { |
| 403 | this.delegate.appendToAttributeValue(char); |
| 404 | } |
| 405 | }, |
| 406 | |
| 407 | attributeValueUnquoted: function attributeValueUnquoted() { |
| 408 | var char = this.consume(); |
| 409 | |
| 410 | if (isSpace(char)) { |
| 411 | this.delegate.finishAttributeValue(); |
| 412 | this.state = 'beforeAttributeName'; |
| 413 | } else if (char === "&") { |
| 414 | this.delegate.appendToAttributeValue(this.consumeCharRef(">") || "&"); |
| 415 | } else if (char === ">") { |
| 416 | this.delegate.finishAttributeValue(); |
| 417 | this.delegate.finishTag(); |
| 418 | this.state = 'beforeData'; |
| 419 | } else { |
| 420 | this.delegate.appendToAttributeValue(char); |
| 421 | } |
| 422 | }, |
| 423 | |
| 424 | afterAttributeValueQuoted: function afterAttributeValueQuoted() { |
| 425 | var char = this.peek(); |
| 426 | |
| 427 | if (isSpace(char)) { |
| 428 | this.consume(); |
| 429 | this.state = 'beforeAttributeName'; |
| 430 | } else if (char === "/") { |
| 431 | this.consume(); |
| 432 | this.state = 'selfClosingStartTag'; |
| 433 | } else if (char === ">") { |
| 434 | this.consume(); |
| 435 | this.delegate.finishTag(); |
| 436 | this.state = 'beforeData'; |
| 437 | } else { |
| 438 | this.state = 'beforeAttributeName'; |
| 439 | } |
| 440 | }, |
| 441 | |
| 442 | selfClosingStartTag: function selfClosingStartTag() { |
| 443 | var char = this.peek(); |
| 444 | |
| 445 | if (char === ">") { |
| 446 | this.consume(); |
| 447 | this.delegate.markTagAsSelfClosing(); |
| 448 | this.delegate.finishTag(); |
| 449 | this.state = 'beforeData'; |
| 450 | } else { |
| 451 | this.state = 'beforeAttributeName'; |
| 452 | } |
| 453 | }, |
| 454 | |
| 455 | endTagOpen: function endTagOpen() { |
| 456 | var char = this.consume(); |
| 457 | |
| 458 | if (isAlpha(char)) { |
| 459 | this.state = 'tagName'; |
| 460 | this.delegate.beginEndTag(); |
| 461 | this.delegate.appendToTagName(char.toLowerCase()); |
| 462 | } |
| 463 | } |
| 464 | } |
| 465 | }; |
| 466 | |
| 467 | function Tokenizer(entityParser, options) { |
| 468 | this.token = null; |
| 469 | this.startLine = 1; |
| 470 | this.startColumn = 0; |
| 471 | this.options = options || {}; |
| 472 | this.tokenizer = new EventedTokenizer(this, entityParser); |
| 473 | } |
| 474 | |
| 475 | Tokenizer.prototype = { |
| 476 | tokenize: function tokenize(input) { |
| 477 | this.tokens = []; |
| 478 | this.tokenizer.tokenize(input); |
| 479 | return this.tokens; |
| 480 | }, |
| 481 | |
| 482 | tokenizePart: function tokenizePart(input) { |
| 483 | this.tokens = []; |
| 484 | this.tokenizer.tokenizePart(input); |
| 485 | return this.tokens; |
| 486 | }, |
| 487 | |
| 488 | tokenizeEOF: function tokenizeEOF() { |
| 489 | this.tokens = []; |
| 490 | this.tokenizer.tokenizeEOF(); |
| 491 | return this.tokens[0]; |
| 492 | }, |
| 493 | |
| 494 | reset: function reset() { |
| 495 | this.token = null; |
| 496 | this.startLine = 1; |
| 497 | this.startColumn = 0; |
| 498 | }, |
| 499 | |
| 500 | addLocInfo: function addLocInfo() { |
| 501 | if (this.options.loc) { |
| 502 | this.token.loc = { |
| 503 | start: { |
| 504 | line: this.startLine, |
| 505 | column: this.startColumn |
| 506 | }, |
| 507 | end: { |
| 508 | line: this.tokenizer.line, |
| 509 | column: this.tokenizer.column |
| 510 | } |
| 511 | }; |
| 512 | } |
| 513 | this.startLine = this.tokenizer.line; |
| 514 | this.startColumn = this.tokenizer.column; |
| 515 | }, |
| 516 | |
| 517 | // Data |
| 518 | |
| 519 | beginData: function beginData() { |
| 520 | this.token = { |
| 521 | type: 'Chars', |
| 522 | chars: '' |
| 523 | }; |
| 524 | this.tokens.push(this.token); |
| 525 | }, |
| 526 | |
| 527 | appendToData: function appendToData(char) { |
| 528 | this.token.chars += char; |
| 529 | }, |
| 530 | |
| 531 | finishData: function finishData() { |
| 532 | this.addLocInfo(); |
| 533 | }, |
| 534 | |
| 535 | // Comment |
| 536 | |
| 537 | beginComment: function beginComment() { |
| 538 | this.token = { |
| 539 | type: 'Comment', |
| 540 | chars: '' |
| 541 | }; |
| 542 | this.tokens.push(this.token); |
| 543 | }, |
| 544 | |
| 545 | appendToCommentData: function appendToCommentData(char) { |
| 546 | this.token.chars += char; |
| 547 | }, |
| 548 | |
| 549 | finishComment: function finishComment() { |
| 550 | this.addLocInfo(); |
| 551 | }, |
| 552 | |
| 553 | // Tags - basic |
| 554 | |
| 555 | beginStartTag: function beginStartTag() { |
| 556 | this.token = { |
| 557 | type: 'StartTag', |
| 558 | tagName: '', |
| 559 | attributes: [], |
| 560 | selfClosing: false |
| 561 | }; |
| 562 | this.tokens.push(this.token); |
| 563 | }, |
| 564 | |
| 565 | beginEndTag: function beginEndTag() { |
| 566 | this.token = { |
| 567 | type: 'EndTag', |
| 568 | tagName: '' |
| 569 | }; |
| 570 | this.tokens.push(this.token); |
| 571 | }, |
| 572 | |
| 573 | finishTag: function finishTag() { |
| 574 | this.addLocInfo(); |
| 575 | }, |
| 576 | |
| 577 | markTagAsSelfClosing: function markTagAsSelfClosing() { |
| 578 | this.token.selfClosing = true; |
| 579 | }, |
| 580 | |
| 581 | // Tags - name |
| 582 | |
| 583 | appendToTagName: function appendToTagName(char) { |
| 584 | this.token.tagName += char; |
| 585 | }, |
| 586 | |
| 587 | // Tags - attributes |
| 588 | |
| 589 | beginAttribute: function beginAttribute() { |
| 590 | this._currentAttribute = ["", "", null]; |
| 591 | this.token.attributes.push(this._currentAttribute); |
| 592 | }, |
| 593 | |
| 594 | appendToAttributeName: function appendToAttributeName(char) { |
| 595 | this._currentAttribute[0] += char; |
| 596 | }, |
| 597 | |
| 598 | beginAttributeValue: function beginAttributeValue(isQuoted) { |
| 599 | this._currentAttribute[2] = isQuoted; |
| 600 | }, |
| 601 | |
| 602 | appendToAttributeValue: function appendToAttributeValue(char) { |
| 603 | this._currentAttribute[1] = this._currentAttribute[1] || ""; |
| 604 | this._currentAttribute[1] += char; |
| 605 | }, |
| 606 | |
| 607 | finishAttributeValue: function finishAttributeValue() {} |
| 608 | }; |
| 609 | |
| 610 | function tokenize(input, options) { |
| 611 | var tokenizer = new Tokenizer(new EntityParser(HTML5NamedCharRefs), options); |
| 612 | return tokenizer.tokenize(input); |
| 613 | } |
| 614 | |
| 615 | var HTML5Tokenizer = { |
| 616 | HTML5NamedCharRefs: HTML5NamedCharRefs, |
| 617 | EntityParser: EntityParser, |
| 618 | EventedTokenizer: EventedTokenizer, |
| 619 | Tokenizer: Tokenizer, |
| 620 | tokenize: tokenize |
| 621 | }; |
| 622 | |
| 623 | var options = linkify.options; |
| 624 | var Options = options.Options; |
| 625 | |
| 626 | |
| 627 | var StartTag = 'StartTag'; |
| 628 | var EndTag = 'EndTag'; |
| 629 | var Chars = 'Chars'; |
| 630 | var Comment = 'Comment'; |
| 631 | |
| 632 | /** |
| 633 | `tokens` and `token` in this section refer to tokens generated by the HTML |
| 634 | parser. |
| 635 | */ |
| 636 | function linkifyHtml(str) { |
| 637 | var opts = arguments.length <= 1 || arguments[1] === undefined ? {} : arguments[1]; |
| 638 | |
| 639 | var tokens = HTML5Tokenizer.tokenize(str); |
| 640 | var linkifiedTokens = []; |
| 641 | var linkified = []; |
| 642 | var i; |
| 643 | |
| 644 | opts = new Options(opts); |
| 645 | |
| 646 | // Linkify the tokens given by the parser |
| 647 | for (i = 0; i < tokens.length; i++) { |
| 648 | var token = tokens[i]; |
| 649 | |
| 650 | if (token.type === StartTag) { |
| 651 | linkifiedTokens.push(token); |
| 652 | |
| 653 | // Ignore all the contents of ignored tags |
| 654 | var tagName = token.tagName.toUpperCase(); |
| 655 | var isIgnored = tagName === 'A' || options.contains(opts.ignoreTags, tagName); |
| 656 | if (!isIgnored) { |
| 657 | continue; |
| 658 | } |
| 659 | |
| 660 | var preskipLen = linkifiedTokens.length; |
| 661 | skipTagTokens(tagName, tokens, ++i, linkifiedTokens); |
| 662 | i += linkifiedTokens.length - preskipLen - 1; |
| 663 | continue; |
| 664 | } else if (token.type !== Chars) { |
| 665 | // Skip this token, it's not important |
| 666 | linkifiedTokens.push(token); |
| 667 | continue; |
| 668 | } |
| 669 | |
| 670 | // Valid text token, linkify it! |
| 671 | var linkifedChars = linkifyChars(token.chars, opts); |
| 672 | linkifiedTokens.push.apply(linkifiedTokens, linkifedChars); |
| 673 | } |
| 674 | |
| 675 | // Convert the tokens back into a string |
| 676 | for (i = 0; i < linkifiedTokens.length; i++) { |
| 677 | var _token = linkifiedTokens[i]; |
| 678 | switch (_token.type) { |
| 679 | case StartTag: |
| 680 | var link = '<' + _token.tagName; |
| 681 | if (_token.attributes.length > 0) { |
| 682 | var attrs = attrsToStrings(_token.attributes); |
| 683 | link += ' ' + attrs.join(' '); |
| 684 | } |
| 685 | link += '>'; |
| 686 | linkified.push(link); |
| 687 | break; |
| 688 | case EndTag: |
| 689 | linkified.push('</' + _token.tagName + '>'); |
| 690 | break; |
| 691 | case Chars: |
| 692 | linkified.push(escapeText(_token.chars)); |
| 693 | break; |
| 694 | case Comment: |
| 695 | linkified.push('<!--' + escapeText(_token.chars) + '-->'); |
| 696 | break; |
| 697 | } |
| 698 | } |
| 699 | |
| 700 | return linkified.join(''); |
| 701 | } |
| 702 | |
| 703 | /** |
| 704 | `tokens` and `token` in this section referes to tokens returned by |
| 705 | `linkify.tokenize`. `linkified` will contain HTML Parser-style tokens |
| 706 | */ |
| 707 | function linkifyChars(str, opts) { |
| 708 | var tokens = linkify.tokenize(str); |
| 709 | var result = []; |
| 710 | |
| 711 | for (var i = 0; i < tokens.length; i++) { |
| 712 | var token = tokens[i]; |
| 713 | |
| 714 | if (token.type === 'nl' && opts.nl2br) { |
| 715 | result.push({ |
| 716 | type: StartTag, |
| 717 | tagName: 'br', |
| 718 | attributes: [], |
| 719 | selfClosing: true |
| 720 | }); |
| 721 | continue; |
| 722 | } else if (!token.isLink || !opts.check(token)) { |
| 723 | result.push({ type: Chars, chars: token.toString() }); |
| 724 | continue; |
| 725 | } |
| 726 | |
| 727 | var _opts$resolve = opts.resolve(token); |
| 728 | |
| 729 | var href = _opts$resolve.href; |
| 730 | var formatted = _opts$resolve.formatted; |
| 731 | var formattedHref = _opts$resolve.formattedHref; |
| 732 | var tagName = _opts$resolve.tagName; |
| 733 | var className = _opts$resolve.className; |
| 734 | var target = _opts$resolve.target; |
| 735 | var attributes = _opts$resolve.attributes; |
| 736 | |
| 737 | // Build up attributes |
| 738 | |
| 739 | var attributeArray = [['href', formattedHref]]; |
| 740 | |
| 741 | if (className) { |
| 742 | attributeArray.push(['class', className]); |
| 743 | } |
| 744 | |
| 745 | if (target) { |
| 746 | attributeArray.push(['target', target]); |
| 747 | } |
| 748 | |
| 749 | for (var attr in attributes) { |
| 750 | attributeArray.push([attr, attributes[attr]]); |
| 751 | } |
| 752 | |
| 753 | // Add the required tokens |
| 754 | result.push({ |
| 755 | type: StartTag, |
| 756 | tagName: tagName, |
| 757 | attributes: attributeArray, |
| 758 | selfClosing: false |
| 759 | }); |
| 760 | result.push({ type: Chars, chars: formatted }); |
| 761 | result.push({ type: EndTag, tagName: tagName }); |
| 762 | } |
| 763 | |
| 764 | return result; |
| 765 | } |
| 766 | |
| 767 | /** |
| 768 | Returns a list of tokens skipped until the closing tag of tagName. |
| 769 | |
| 770 | * `tagName` is the closing tag which will prompt us to stop skipping |
| 771 | * `tokens` is the array of tokens generated by HTML5Tokenizer which |
| 772 | * `i` is the index immediately after the opening tag to skip |
| 773 | * `skippedTokens` is an array which skipped tokens are being pushed into |
| 774 | |
| 775 | Caveats |
| 776 | |
| 777 | * Assumes that i is the first token after the given opening tagName |
| 778 | * The closing tag will be skipped, but nothing after it |
| 779 | * Will track whether there is a nested tag of the same type |
| 780 | */ |
| 781 | function skipTagTokens(tagName, tokens, i, skippedTokens) { |
| 782 | |
| 783 | // number of tokens of this type on the [fictional] stack |
| 784 | var stackCount = 1; |
| 785 | |
| 786 | while (i < tokens.length && stackCount > 0) { |
| 787 | var token = tokens[i]; |
| 788 | if (token.type === StartTag && token.tagName.toUpperCase() === tagName) { |
| 789 | // Nested tag of the same type, "add to stack" |
| 790 | stackCount++; |
| 791 | } else if (token.type === EndTag && token.tagName.toUpperCase() === tagName) { |
| 792 | // Closing tag |
| 793 | stackCount--; |
| 794 | } |
| 795 | skippedTokens.push(token); |
| 796 | i++; |
| 797 | } |
| 798 | |
| 799 | // Note that if stackCount > 0 here, the HTML is probably invalid |
| 800 | return skippedTokens; |
| 801 | } |
| 802 | |
| 803 | function escapeText(text) { |
| 804 | // Not required, HTML tokenizer ensures this occurs properly |
| 805 | return text; |
| 806 | } |
| 807 | |
| 808 | function escapeAttr(attr) { |
| 809 | return attr.replace(/"/g, '"'); |
| 810 | } |
| 811 | |
| 812 | function attrsToStrings(attrs) { |
| 813 | var attrStrs = []; |
| 814 | for (var i = 0; i < attrs.length; i++) { |
| 815 | var _attrs$i = attrs[i]; |
| 816 | var name = _attrs$i[0]; |
| 817 | var value = _attrs$i[1]; |
| 818 | |
| 819 | attrStrs.push(name + '="' + escapeAttr(value) + '"'); |
| 820 | } |
| 821 | return attrStrs; |
| 822 | } |
| 823 | |
| 824 | return linkifyHtml; |
| 825 | }(linkify); |
| 826 | window.linkifyHtml = linkifyHtml; |
| 827 | })(window, linkify); |