1/** 2 * @license 3 * Copyright (C) 2006 Google Inc. 4 * 5 * Licensed under the Apache License, Version 2.0 (the "License"); 6 * you may not use this file except in compliance with the License. 7 * You may obtain a copy of the License at 8 * 9 * http://www.apache.org/licenses/LICENSE-2.0 10 * 11 * Unless required by applicable law or agreed to in writing, software 12 * distributed under the License is distributed on an "AS IS" BASIS, 13 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 * See the License for the specific language governing permissions and 15 * limitations under the License. 16 */ 17 18/** 19 * @fileoverview 20 * some functions for browser-side pretty printing of code contained in html. 21 * 22 * <p> 23 * For a fairly comprehensive set of languages see the 24 * <a href="https://github.com/google/code-prettify#for-which-languages-does-it-work">README</a> 25 * file that came with this source. At a minimum, the lexer should work on a 26 * number of languages including C and friends, Java, Python, Bash, SQL, HTML, 27 * XML, CSS, Javascript, and Makefiles. It works passably on Ruby, PHP and Awk 28 * and a subset of Perl, but, because of commenting conventions, doesn't work on 29 * Smalltalk, Lisp-like, or CAML-like languages without an explicit lang class. 30 * <p> 31 * Usage: <ol> 32 * <li> include this source file in an html page via 33 * {@code <script type="text/javascript" src="/path/to/prettify.js"></script>} 34 * <li> define style rules. See the example page for examples. 35 * <li> mark the {@code <pre>} and {@code <code>} tags in your source with 36 * {@code class=prettyprint.} 37 * You can also use the (html deprecated) {@code <xmp>} tag, but the pretty 38 * printer needs to do more substantial DOM manipulations to support that, so 39 * some css styles may not be preserved. 40 * </ol> 41 * That's it. I wanted to keep the API as simple as possible, so there's no 42 * need to specify which language the code is in, but if you wish, you can add 43 * another class to the {@code <pre>} or {@code <code>} element to specify the 44 * language, as in {@code <pre class="prettyprint lang-java">}. Any class that 45 * starts with "lang-" followed by a file extension, specifies the file type. 46 * See the "lang-*.js" files in this directory for code that implements 47 * per-language file handlers. 48 * <p> 49 * Change log:<br> 50 * cbeust, 2006/08/22 51 * <blockquote> 52 * Java annotations (start with "@") are now captured as literals ("lit") 53 * </blockquote> 54 * @requires console 55 */ 56 57// JSLint declarations 58/*global console, document, navigator, setTimeout, window, define */ 59 60/** 61 * @typedef {!Array.<number|string>} 62 * Alternating indices and the decorations that should be inserted there. 63 * The indices are monotonically increasing. 64 */ 65var DecorationsT; 66 67/** 68 * @typedef {!{ 69 * sourceNode: !Element, 70 * pre: !(number|boolean), 71 * langExtension: ?string, 72 * numberLines: ?(number|boolean), 73 * sourceCode: ?string, 74 * spans: ?(Array.<number|Node>), 75 * basePos: ?number, 76 * decorations: ?DecorationsT 77 * }} 78 * <dl> 79 * <dt>sourceNode<dd>the element containing the source 80 * <dt>sourceCode<dd>source as plain text 81 * <dt>pre<dd>truthy if white-space in text nodes 82 * should be considered significant. 83 * <dt>spans<dd> alternating span start indices into source 84 * and the text node or element (e.g. {@code <BR>}) corresponding to that 85 * span. 86 * <dt>decorations<dd>an array of style classes preceded 87 * by the position at which they start in job.sourceCode in order 88 * <dt>basePos<dd>integer position of this.sourceCode in the larger chunk of 89 * source. 90 * </dl> 91 */ 92var JobT; 93 94/** 95 * @typedef {!{ 96 * sourceCode: string, 97 * spans: !(Array.<number|Node>) 98 * }} 99 * <dl> 100 * <dt>sourceCode<dd>source as plain text 101 * <dt>spans<dd> alternating span start indices into source 102 * and the text node or element (e.g. {@code <BR>}) corresponding to that 103 * span. 104 * </dl> 105 */ 106var SourceSpansT; 107 108/** @define {boolean} */ 109var IN_GLOBAL_SCOPE = true; 110 111 112/** 113 * {@type !{ 114 * 'createSimpleLexer': function (Array, Array): (function (JobT)), 115 * 'registerLangHandler': function (function (JobT), Array.<string>), 116 * 'PR_ATTRIB_NAME': string, 117 * 'PR_ATTRIB_NAME': string, 118 * 'PR_ATTRIB_VALUE': string, 119 * 'PR_COMMENT': string, 120 * 'PR_DECLARATION': string, 121 * 'PR_KEYWORD': string, 122 * 'PR_LITERAL': string, 123 * 'PR_NOCODE': string, 124 * 'PR_PLAIN': string, 125 * 'PR_PUNCTUATION': string, 126 * 'PR_SOURCE': string, 127 * 'PR_STRING': string, 128 * 'PR_TAG': string, 129 * 'PR_TYPE': string, 130 * 'prettyPrintOne': function (string, string, number|boolean), 131 * 'prettyPrint': function (?function, ?(HTMLElement|HTMLDocument)) 132 * }} 133 * @const 134 */ 135var PR; 136 137/** 138 * Split {@code prettyPrint} into multiple timeouts so as not to interfere with 139 * UI events. 140 * If set to {@code false}, {@code prettyPrint()} is synchronous. 141 */ 142var PR_SHOULD_USE_CONTINUATION = true 143if (typeof window !== 'undefined') { 144 window['PR_SHOULD_USE_CONTINUATION'] = PR_SHOULD_USE_CONTINUATION; 145} 146 147/** 148 * Pretty print a chunk of code. 149 * @param {string} sourceCodeHtml The HTML to pretty print. 150 * @param {string} opt_langExtension The language name to use. 151 * Typically, a filename extension like 'cpp' or 'java'. 152 * @param {number|boolean} opt_numberLines True to number lines, 153 * or the 1-indexed number of the first line in sourceCodeHtml. 154 * @return {string} code as html, but prettier 155 */ 156var prettyPrintOne; 157/** 158 * Find all the {@code <pre>} and {@code <code>} tags in the DOM with 159 * {@code class=prettyprint} and prettify them. 160 * 161 * @param {Function} opt_whenDone called when prettifying is done. 162 * @param {HTMLElement|HTMLDocument} opt_root an element or document 163 * containing all the elements to pretty print. 164 * Defaults to {@code document.body}. 165 */ 166var prettyPrint; 167 168 169(function () { 170 var win = (typeof window !== 'undefined') ? window : {}; 171 // Keyword lists for various languages. 172 // We use things that coerce to strings to make them compact when minified 173 // and to defeat aggressive optimizers that fold large string constants. 174 var FLOW_CONTROL_KEYWORDS = ["break,continue,do,else,for,if,return,while"]; 175 var C_KEYWORDS = [FLOW_CONTROL_KEYWORDS,"auto,case,char,const,default," + 176 "double,enum,extern,float,goto,inline,int,long,register,restrict,short,signed," + 177 "sizeof,static,struct,switch,typedef,union,unsigned,void,volatile"]; 178 var COMMON_KEYWORDS = [C_KEYWORDS,"catch,class,delete,false,import," + 179 "new,operator,private,protected,public,this,throw,true,try,typeof"]; 180 var CPP_KEYWORDS = [COMMON_KEYWORDS,"alignas,alignof,align_union,asm,axiom,bool," + 181 "concept,concept_map,const_cast,constexpr,decltype,delegate," + 182 "dynamic_cast,explicit,export,friend,generic,late_check," + 183 "mutable,namespace,noexcept,noreturn,nullptr,property,reinterpret_cast,static_assert," + 184 "static_cast,template,typeid,typename,using,virtual,where"]; 185 var JAVA_KEYWORDS = [COMMON_KEYWORDS, 186 "abstract,assert,boolean,byte,extends,finally,final,implements,import," + 187 "instanceof,interface,null,native,package,strictfp,super,synchronized," + 188 "throws,transient"]; 189 var CSHARP_KEYWORDS = [COMMON_KEYWORDS, 190 "abstract,add,alias,as,ascending,async,await,base,bool,by,byte,checked,decimal,delegate,descending," + 191 "dynamic,event,finally,fixed,foreach,from,get,global,group,implicit,in,interface," + 192 "internal,into,is,join,let,lock,null,object,out,override,orderby,params," + 193 "partial,readonly,ref,remove,sbyte,sealed,select,set,stackalloc,string,select,uint,ulong," + 194 "unchecked,unsafe,ushort,value,var,virtual,where,yield"]; 195 var COFFEE_KEYWORDS = "all,and,by,catch,class,else,extends,false,finally," + 196 "for,if,in,is,isnt,loop,new,no,not,null,of,off,on,or,return,super,then," + 197 "throw,true,try,unless,until,when,while,yes"; 198 var JSCRIPT_KEYWORDS = [COMMON_KEYWORDS, 199 "abstract,async,await,constructor,debugger,enum,eval,export,from,function," + 200 "get,import,implements,instanceof,interface,let,null,of,set,undefined," + 201 "var,with,yield,Infinity,NaN"]; 202 var PERL_KEYWORDS = "caller,delete,die,do,dump,elsif,eval,exit,foreach,for," + 203 "goto,if,import,last,local,my,next,no,our,print,package,redo,require," + 204 "sub,undef,unless,until,use,wantarray,while,BEGIN,END"; 205 var PYTHON_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "and,as,assert,class,def,del," + 206 "elif,except,exec,finally,from,global,import,in,is,lambda," + 207 "nonlocal,not,or,pass,print,raise,try,with,yield," + 208 "False,True,None"]; 209 var RUBY_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "alias,and,begin,case,class," + 210 "def,defined,elsif,end,ensure,false,in,module,next,nil,not,or,redo," + 211 "rescue,retry,self,super,then,true,undef,unless,until,when,yield," + 212 "BEGIN,END"]; 213 var SH_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "case,done,elif,esac,eval,fi," + 214 "function,in,local,set,then,until"]; 215 var ALL_KEYWORDS = [ 216 CPP_KEYWORDS, CSHARP_KEYWORDS, JAVA_KEYWORDS, JSCRIPT_KEYWORDS, 217 PERL_KEYWORDS, PYTHON_KEYWORDS, RUBY_KEYWORDS, SH_KEYWORDS]; 218 var C_TYPES = /^(DIR|FILE|array|vector|(de|priority_)?queue|(forward_)?list|stack|(const_)?(reverse_)?iterator|(unordered_)?(multi)?(set|map)|bitset|u?(int|float)\d*)\b/; 219 220 // token style names. correspond to css classes 221 /** 222 * token style for a string literal 223 * @const 224 */ 225 var PR_STRING = 'str'; 226 /** 227 * token style for a keyword 228 * @const 229 */ 230 var PR_KEYWORD = 'kwd'; 231 /** 232 * token style for a comment 233 * @const 234 */ 235 var PR_COMMENT = 'com'; 236 /** 237 * token style for a type 238 * @const 239 */ 240 var PR_TYPE = 'typ'; 241 /** 242 * token style for a literal value. e.g. 1, null, true. 243 * @const 244 */ 245 var PR_LITERAL = 'lit'; 246 /** 247 * token style for a punctuation string. 248 * @const 249 */ 250 var PR_PUNCTUATION = 'pun'; 251 /** 252 * token style for plain text. 253 * @const 254 */ 255 var PR_PLAIN = 'pln'; 256 257 /** 258 * token style for an sgml tag. 259 * @const 260 */ 261 var PR_TAG = 'tag'; 262 /** 263 * token style for a markup declaration such as a DOCTYPE. 264 * @const 265 */ 266 var PR_DECLARATION = 'dec'; 267 /** 268 * token style for embedded source. 269 * @const 270 */ 271 var PR_SOURCE = 'src'; 272 /** 273 * token style for an sgml attribute name. 274 * @const 275 */ 276 var PR_ATTRIB_NAME = 'atn'; 277 /** 278 * token style for an sgml attribute value. 279 * @const 280 */ 281 var PR_ATTRIB_VALUE = 'atv'; 282 283 /** 284 * A class that indicates a section of markup that is not code, e.g. to allow 285 * embedding of line numbers within code listings. 286 * @const 287 */ 288 var PR_NOCODE = 'nocode'; 289 290 291 // Regex pattern below is automatically generated by regexpPrecederPatterns.pl 292 // Do not modify, your changes will be erased. 293 294 // CAVEAT: this does not properly handle the case where a regular 295 // expression immediately follows another since a regular expression may 296 // have flags for case-sensitivity and the like. Having regexp tokens 297 // adjacent is not valid in any language I'm aware of, so I'm punting. 298 // TODO: maybe style special characters inside a regexp as punctuation. 299 300 /** 301 * A set of tokens that can precede a regular expression literal in 302 * javascript 303 * http://web.archive.org/web/20070717142515/http://www.mozilla.org/js/language/js20/rationale/syntax.html 304 * has the full list, but I've removed ones that might be problematic when 305 * seen in languages that don't support regular expression literals. 306 * 307 * Specifically, I've removed any keywords that can't precede a regexp 308 * literal in a syntactically legal javascript program, and I've removed the 309 * "in" keyword since it's not a keyword in many languages, and might be used 310 * as a count of inches. 311 * 312 * The link above does not accurately describe EcmaScript rules since 313 * it fails to distinguish between (a=++/b/i) and (a++/b/i) but it works 314 * very well in practice. 315 * 316 * @private 317 * @const 318 */ 319 var REGEXP_PRECEDER_PATTERN = '(?:^^\\.?|[+-]|[!=]=?=?|\\#|%=?|&&?=?|\\(|\\*=?|[+\\-]=|->|\\/=?|::?|<<?=?|>>?>?=?|,|;|\\?|@|\\[|~|{|\\^\\^?=?|\\|\\|?=?|break|case|continue|delete|do|else|finally|instanceof|return|throw|try|typeof)\\s*'; 320 321 322 /** 323 * Given a group of {@link RegExp}s, returns a {@code RegExp} that globally 324 * matches the union of the sets of strings matched by the input RegExp. 325 * Since it matches globally, if the input strings have a start-of-input 326 * anchor (/^.../), it is ignored for the purposes of unioning. 327 * @param {Array.<RegExp>} regexs non multiline, non-global regexs. 328 * @return {RegExp} a global regex. 329 */ 330 function combinePrefixPatterns(regexs) { 331 var capturedGroupIndex = 0; 332 333 var needToFoldCase = false; 334 var ignoreCase = false; 335 for (var i = 0, n = regexs.length; i < n; ++i) { 336 var regex = regexs[i]; 337 if (regex.ignoreCase) { 338 ignoreCase = true; 339 } else if (/[a-z]/i.test(regex.source.replace( 340 /\\u[0-9a-f]{4}|\\x[0-9a-f]{2}|\\[^ux]/gi, ''))) { 341 needToFoldCase = true; 342 ignoreCase = false; 343 break; 344 } 345 } 346 347 var escapeCharToCodeUnit = { 348 'b': 8, 349 't': 9, 350 'n': 0xa, 351 'v': 0xb, 352 'f': 0xc, 353 'r': 0xd 354 }; 355 356 function decodeEscape(charsetPart) { 357 var cc0 = charsetPart.charCodeAt(0); 358 if (cc0 !== 92 /* \\ */) { 359 return cc0; 360 } 361 var c1 = charsetPart.charAt(1); 362 cc0 = escapeCharToCodeUnit[c1]; 363 if (cc0) { 364 return cc0; 365 } else if ('0' <= c1 && c1 <= '7') { 366 return parseInt(charsetPart.substring(1), 8); 367 } else if (c1 === 'u' || c1 === 'x') { 368 return parseInt(charsetPart.substring(2), 16); 369 } else { 370 return charsetPart.charCodeAt(1); 371 } 372 } 373 374 function encodeEscape(charCode) { 375 if (charCode < 0x20) { 376 return (charCode < 0x10 ? '\\x0' : '\\x') + charCode.toString(16); 377 } 378 var ch = String.fromCharCode(charCode); 379 return (ch === '\\' || ch === '-' || ch === ']' || ch === '^') 380 ? "\\" + ch : ch; 381 } 382 383 function caseFoldCharset(charSet) { 384 var charsetParts = charSet.substring(1, charSet.length - 1).match( 385 new RegExp( 386 '\\\\u[0-9A-Fa-f]{4}' 387 + '|\\\\x[0-9A-Fa-f]{2}' 388 + '|\\\\[0-3][0-7]{0,2}' 389 + '|\\\\[0-7]{1,2}' 390 + '|\\\\[\\s\\S]' 391 + '|-' 392 + '|[^-\\\\]', 393 'g')); 394 var ranges = []; 395 var inverse = charsetParts[0] === '^'; 396 397 var out = ['[']; 398 if (inverse) { out.push('^'); } 399 400 for (var i = inverse ? 1 : 0, n = charsetParts.length; i < n; ++i) { 401 var p = charsetParts[i]; 402 if (/\\[bdsw]/i.test(p)) { // Don't muck with named groups. 403 out.push(p); 404 } else { 405 var start = decodeEscape(p); 406 var end; 407 if (i + 2 < n && '-' === charsetParts[i + 1]) { 408 end = decodeEscape(charsetParts[i + 2]); 409 i += 2; 410 } else { 411 end = start; 412 } 413 ranges.push([start, end]); 414 // If the range might intersect letters, then expand it. 415 // This case handling is too simplistic. 416 // It does not deal with non-latin case folding. 417 // It works for latin source code identifiers though. 418 if (!(end < 65 || start > 122)) { 419 if (!(end < 65 || start > 90)) { 420 ranges.push([Math.max(65, start) | 32, Math.min(end, 90) | 32]); 421 } 422 if (!(end < 97 || start > 122)) { 423 ranges.push([Math.max(97, start) & ~32, Math.min(end, 122) & ~32]); 424 } 425 } 426 } 427 } 428 429 // [[1, 10], [3, 4], [8, 12], [14, 14], [16, 16], [17, 17]] 430 // -> [[1, 12], [14, 14], [16, 17]] 431 ranges.sort(function (a, b) { return (a[0] - b[0]) || (b[1] - a[1]); }); 432 var consolidatedRanges = []; 433 var lastRange = []; 434 for (var i = 0; i < ranges.length; ++i) { 435 var range = ranges[i]; 436 if (range[0] <= lastRange[1] + 1) { 437 lastRange[1] = Math.max(lastRange[1], range[1]); 438 } else { 439 consolidatedRanges.push(lastRange = range); 440 } 441 } 442 443 for (var i = 0; i < consolidatedRanges.length; ++i) { 444 var range = consolidatedRanges[i]; 445 out.push(encodeEscape(range[0])); 446 if (range[1] > range[0]) { 447 if (range[1] + 1 > range[0]) { out.push('-'); } 448 out.push(encodeEscape(range[1])); 449 } 450 } 451 out.push(']'); 452 return out.join(''); 453 } 454 455 function allowAnywhereFoldCaseAndRenumberGroups(regex) { 456 // Split into character sets, escape sequences, punctuation strings 457 // like ('(', '(?:', ')', '^'), and runs of characters that do not 458 // include any of the above. 459 var parts = regex.source.match( 460 new RegExp( 461 '(?:' 462 + '\\[(?:[^\\x5C\\x5D]|\\\\[\\s\\S])*\\]' // a character set 463 + '|\\\\u[A-Fa-f0-9]{4}' // a unicode escape 464 + '|\\\\x[A-Fa-f0-9]{2}' // a hex escape 465 + '|\\\\[0-9]+' // a back-reference or octal escape 466 + '|\\\\[^ux0-9]' // other escape sequence 467 + '|\\(\\?[:!=]' // start of a non-capturing group 468 + '|[\\(\\)\\^]' // start/end of a group, or line start 469 + '|[^\\x5B\\x5C\\(\\)\\^]+' // run of other characters 470 + ')', 471 'g')); 472 var n = parts.length; 473 474 // Maps captured group numbers to the number they will occupy in 475 // the output or to -1 if that has not been determined, or to 476 // undefined if they need not be capturing in the output. 477 var capturedGroups = []; 478 479 // Walk over and identify back references to build the capturedGroups 480 // mapping. 481 for (var i = 0, groupIndex = 0; i < n; ++i) { 482 var p = parts[i]; 483 if (p === '(') { 484 // groups are 1-indexed, so max group index is count of '(' 485 ++groupIndex; 486 } else if ('\\' === p.charAt(0)) { 487 var decimalValue = +p.substring(1); 488 if (decimalValue) { 489 if (decimalValue <= groupIndex) { 490 capturedGroups[decimalValue] = -1; 491 } else { 492 // Replace with an unambiguous escape sequence so that 493 // an octal escape sequence does not turn into a backreference 494 // to a capturing group from an earlier regex. 495 parts[i] = encodeEscape(decimalValue); 496 } 497 } 498 } 499 } 500 501 // Renumber groups and reduce capturing groups to non-capturing groups 502 // where possible. 503 for (var i = 1; i < capturedGroups.length; ++i) { 504 if (-1 === capturedGroups[i]) { 505 capturedGroups[i] = ++capturedGroupIndex; 506 } 507 } 508 for (var i = 0, groupIndex = 0; i < n; ++i) { 509 var p = parts[i]; 510 if (p === '(') { 511 ++groupIndex; 512 if (!capturedGroups[groupIndex]) { 513 parts[i] = '(?:'; 514 } 515 } else if ('\\' === p.charAt(0)) { 516 var decimalValue = +p.substring(1); 517 if (decimalValue && decimalValue <= groupIndex) { 518 parts[i] = '\\' + capturedGroups[decimalValue]; 519 } 520 } 521 } 522 523 // Remove any prefix anchors so that the output will match anywhere. 524 // ^^ really does mean an anchored match though. 525 for (var i = 0; i < n; ++i) { 526 if ('^' === parts[i] && '^' !== parts[i + 1]) { parts[i] = ''; } 527 } 528 529 // Expand letters to groups to handle mixing of case-sensitive and 530 // case-insensitive patterns if necessary. 531 if (regex.ignoreCase && needToFoldCase) { 532 for (var i = 0; i < n; ++i) { 533 var p = parts[i]; 534 var ch0 = p.charAt(0); 535 if (p.length >= 2 && ch0 === '[') { 536 parts[i] = caseFoldCharset(p); 537 } else if (ch0 !== '\\') { 538 // TODO: handle letters in numeric escapes. 539 parts[i] = p.replace( 540 /[a-zA-Z]/g, 541 function (ch) { 542 var cc = ch.charCodeAt(0); 543 return '[' + String.fromCharCode(cc & ~32, cc | 32) + ']'; 544 }); 545 } 546 } 547 } 548 549 return parts.join(''); 550 } 551 552 var rewritten = []; 553 for (var i = 0, n = regexs.length; i < n; ++i) { 554 var regex = regexs[i]; 555 if (regex.global || regex.multiline) { throw new Error('' + regex); } 556 rewritten.push( 557 '(?:' + allowAnywhereFoldCaseAndRenumberGroups(regex) + ')'); 558 } 559 560 return new RegExp(rewritten.join('|'), ignoreCase ? 'gi' : 'g'); 561 } 562 563 564 /** 565 * Split markup into a string of source code and an array mapping ranges in 566 * that string to the text nodes in which they appear. 567 * 568 * <p> 569 * The HTML DOM structure:</p> 570 * <pre> 571 * (Element "p" 572 * (Element "b" 573 * (Text "print ")) ; #1 574 * (Text "'Hello '") ; #2 575 * (Element "br") ; #3 576 * (Text " + 'World';")) ; #4 577 * </pre> 578 * <p> 579 * corresponds to the HTML 580 * {@code <p><b>print </b>'Hello '<br> + 'World';</p>}.</p> 581 * 582 * <p> 583 * It will produce the output:</p> 584 * <pre> 585 * { 586 * sourceCode: "print 'Hello '\n + 'World';", 587 * // 1 2 588 * // 012345678901234 5678901234567 589 * spans: [0, #1, 6, #2, 14, #3, 15, #4] 590 * } 591 * </pre> 592 * <p> 593 * where #1 is a reference to the {@code "print "} text node above, and so 594 * on for the other text nodes. 595 * </p> 596 * 597 * <p> 598 * The {@code} spans array is an array of pairs. Even elements are the start 599 * indices of substrings, and odd elements are the text nodes (or BR elements) 600 * that contain the text for those substrings. 601 * Substrings continue until the next index or the end of the source. 602 * </p> 603 * 604 * @param {Node} node an HTML DOM subtree containing source-code. 605 * @param {boolean|number} isPreformatted truthy if white-space in 606 * text nodes should be considered significant. 607 * @return {SourceSpansT} source code and the nodes in which they occur. 608 */ 609 function extractSourceSpans(node, isPreformatted) { 610 var nocode = /(?:^|\s)nocode(?:\s|$)/; 611 612 var chunks = []; 613 var length = 0; 614 var spans = []; 615 var k = 0; 616 617 function walk(node) { 618 var type = node.nodeType; 619 if (type == 1) { // Element 620 if (nocode.test(node.className)) { return; } 621 for (var child = node.firstChild; child; child = child.nextSibling) { 622 walk(child); 623 } 624 var nodeName = node.nodeName.toLowerCase(); 625 if ('br' === nodeName || 'li' === nodeName) { 626 chunks[k] = '\n'; 627 spans[k << 1] = length++; 628 spans[(k++ << 1) | 1] = node; 629 } 630 } else if (type == 3 || type == 4) { // Text 631 var text = node.nodeValue; 632 if (text.length) { 633 if (!isPreformatted) { 634 text = text.replace(/[ \t\r\n]+/g, ' '); 635 } else { 636 text = text.replace(/\r\n?/g, '\n'); // Normalize newlines. 637 } 638 // TODO: handle tabs here? 639 chunks[k] = text; 640 spans[k << 1] = length; 641 length += text.length; 642 spans[(k++ << 1) | 1] = node; 643 } 644 } 645 } 646 647 walk(node); 648 649 return { 650 sourceCode: chunks.join('').replace(/\n$/, ''), 651 spans: spans 652 }; 653 } 654 655 656 /** 657 * Apply the given language handler to sourceCode and add the resulting 658 * decorations to out. 659 * @param {!Element} sourceNode 660 * @param {number} basePos the index of sourceCode within the chunk of source 661 * whose decorations are already present on out. 662 * @param {string} sourceCode 663 * @param {function(JobT)} langHandler 664 * @param {DecorationsT} out 665 */ 666 function appendDecorations( 667 sourceNode, basePos, sourceCode, langHandler, out) { 668 if (!sourceCode) { return; } 669 /** @type {JobT} */ 670 var job = { 671 sourceNode: sourceNode, 672 pre: 1, 673 langExtension: null, 674 numberLines: null, 675 sourceCode: sourceCode, 676 spans: null, 677 basePos: basePos, 678 decorations: null 679 }; 680 langHandler(job); 681 out.push.apply(out, job.decorations); 682 } 683 684 var notWs = /\S/; 685 686 /** 687 * Given an element, if it contains only one child element and any text nodes 688 * it contains contain only space characters, return the sole child element. 689 * Otherwise returns undefined. 690 * <p> 691 * This is meant to return the CODE element in {@code <pre><code ...>} when 692 * there is a single child element that contains all the non-space textual 693 * content, but not to return anything where there are multiple child elements 694 * as in {@code <pre><code>...</code><code>...</code></pre>} or when there 695 * is textual content. 696 */ 697 function childContentWrapper(element) { 698 var wrapper = undefined; 699 for (var c = element.firstChild; c; c = c.nextSibling) { 700 var type = c.nodeType; 701 wrapper = (type === 1) // Element Node 702 ? (wrapper ? element : c) 703 : (type === 3) // Text Node 704 ? (notWs.test(c.nodeValue) ? element : wrapper) 705 : wrapper; 706 } 707 return wrapper === element ? undefined : wrapper; 708 } 709 710 /** Given triples of [style, pattern, context] returns a lexing function, 711 * The lexing function interprets the patterns to find token boundaries and 712 * returns a decoration list of the form 713 * [index_0, style_0, index_1, style_1, ..., index_n, style_n] 714 * where index_n is an index into the sourceCode, and style_n is a style 715 * constant like PR_PLAIN. index_n-1 <= index_n, and style_n-1 applies to 716 * all characters in sourceCode[index_n-1:index_n]. 717 * 718 * The stylePatterns is a list whose elements have the form 719 * [style : string, pattern : RegExp, DEPRECATED, shortcut : string]. 720 * 721 * Style is a style constant like PR_PLAIN, or can be a string of the 722 * form 'lang-FOO', where FOO is a language extension describing the 723 * language of the portion of the token in $1 after pattern executes. 724 * E.g., if style is 'lang-lisp', and group 1 contains the text 725 * '(hello (world))', then that portion of the token will be passed to the 726 * registered lisp handler for formatting. 727 * The text before and after group 1 will be restyled using this decorator 728 * so decorators should take care that this doesn't result in infinite 729 * recursion. For example, the HTML lexer rule for SCRIPT elements looks 730 * something like ['lang-js', /<[s]cript>(.+?)<\/script>/]. This may match 731 * '<script>foo()<\/script>', which would cause the current decorator to 732 * be called with '<script>' which would not match the same rule since 733 * group 1 must not be empty, so it would be instead styled as PR_TAG by 734 * the generic tag rule. The handler registered for the 'js' extension would 735 * then be called with 'foo()', and finally, the current decorator would 736 * be called with '<\/script>' which would not match the original rule and 737 * so the generic tag rule would identify it as a tag. 738 * 739 * Pattern must only match prefixes, and if it matches a prefix, then that 740 * match is considered a token with the same style. 741 * 742 * Context is applied to the last non-whitespace, non-comment token 743 * recognized. 744 * 745 * Shortcut is an optional string of characters, any of which, if the first 746 * character, gurantee that this pattern and only this pattern matches. 747 * 748 * @param {Array} shortcutStylePatterns patterns that always start with 749 * a known character. Must have a shortcut string. 750 * @param {Array} fallthroughStylePatterns patterns that will be tried in 751 * order if the shortcut ones fail. May have shortcuts. 752 * 753 * @return {function (JobT)} a function that takes an undecorated job and 754 * attaches a list of decorations. 755 */ 756 function createSimpleLexer(shortcutStylePatterns, fallthroughStylePatterns) { 757 var shortcuts = {}; 758 var tokenizer; 759 (function () { 760 var allPatterns = shortcutStylePatterns.concat(fallthroughStylePatterns); 761 var allRegexs = []; 762 var regexKeys = {}; 763 for (var i = 0, n = allPatterns.length; i < n; ++i) { 764 var patternParts = allPatterns[i]; 765 var shortcutChars = patternParts[3]; 766 if (shortcutChars) { 767 for (var c = shortcutChars.length; --c >= 0;) { 768 shortcuts[shortcutChars.charAt(c)] = patternParts; 769 } 770 } 771 var regex = patternParts[1]; 772 var k = '' + regex; 773 if (!regexKeys.hasOwnProperty(k)) { 774 allRegexs.push(regex); 775 regexKeys[k] = null; 776 } 777 } 778 allRegexs.push(/[\0-\uffff]/); 779 tokenizer = combinePrefixPatterns(allRegexs); 780 })(); 781 782 var nPatterns = fallthroughStylePatterns.length; 783 784 /** 785 * Lexes job.sourceCode and attaches an output array job.decorations of 786 * style classes preceded by the position at which they start in 787 * job.sourceCode in order. 788 * 789 * @type{function (JobT)} 790 */ 791 var decorate = function (job) { 792 var sourceCode = job.sourceCode, basePos = job.basePos; 793 var sourceNode = job.sourceNode; 794 /** Even entries are positions in source in ascending order. Odd enties 795 * are style markers (e.g., PR_COMMENT) that run from that position until 796 * the end. 797 * @type {DecorationsT} 798 */ 799 var decorations = [basePos, PR_PLAIN]; 800 var pos = 0; // index into sourceCode 801 var tokens = sourceCode.match(tokenizer) || []; 802 var styleCache = {}; 803 804 for (var ti = 0, nTokens = tokens.length; ti < nTokens; ++ti) { 805 var token = tokens[ti]; 806 var style = styleCache[token]; 807 var match = void 0; 808 809 var isEmbedded; 810 if (typeof style === 'string') { 811 isEmbedded = false; 812 } else { 813 var patternParts = shortcuts[token.charAt(0)]; 814 if (patternParts) { 815 match = token.match(patternParts[1]); 816 style = patternParts[0]; 817 } else { 818 for (var i = 0; i < nPatterns; ++i) { 819 patternParts = fallthroughStylePatterns[i]; 820 match = token.match(patternParts[1]); 821 if (match) { 822 style = patternParts[0]; 823 break; 824 } 825 } 826 827 if (!match) { // make sure that we make progress 828 style = PR_PLAIN; 829 } 830 } 831 832 isEmbedded = style.length >= 5 && 'lang-' === style.substring(0, 5); 833 if (isEmbedded && !(match && typeof match[1] === 'string')) { 834 isEmbedded = false; 835 style = PR_SOURCE; 836 } 837 838 if (!isEmbedded) { styleCache[token] = style; } 839 } 840 841 var tokenStart = pos; 842 pos += token.length; 843 844 if (!isEmbedded) { 845 decorations.push(basePos + tokenStart, style); 846 } else { // Treat group 1 as an embedded block of source code. 847 var embeddedSource = match[1]; 848 var embeddedSourceStart = token.indexOf(embeddedSource); 849 var embeddedSourceEnd = embeddedSourceStart + embeddedSource.length; 850 if (match[2]) { 851 // If embeddedSource can be blank, then it would match at the 852 // beginning which would cause us to infinitely recurse on the 853 // entire token, so we catch the right context in match[2]. 854 embeddedSourceEnd = token.length - match[2].length; 855 embeddedSourceStart = embeddedSourceEnd - embeddedSource.length; 856 } 857 var lang = style.substring(5); 858 // Decorate the left of the embedded source 859 appendDecorations( 860 sourceNode, 861 basePos + tokenStart, 862 token.substring(0, embeddedSourceStart), 863 decorate, decorations); 864 // Decorate the embedded source 865 appendDecorations( 866 sourceNode, 867 basePos + tokenStart + embeddedSourceStart, 868 embeddedSource, 869 langHandlerForExtension(lang, embeddedSource), 870 decorations); 871 // Decorate the right of the embedded section 872 appendDecorations( 873 sourceNode, 874 basePos + tokenStart + embeddedSourceEnd, 875 token.substring(embeddedSourceEnd), 876 decorate, decorations); 877 } 878 } 879 job.decorations = decorations; 880 }; 881 return decorate; 882 } 883 884 /** returns a function that produces a list of decorations from source text. 885 * 886 * This code treats ", ', and ` as string delimiters, and \ as a string 887 * escape. It does not recognize perl's qq() style strings. 888 * It has no special handling for double delimiter escapes as in basic, or 889 * the tripled delimiters used in python, but should work on those regardless 890 * although in those cases a single string literal may be broken up into 891 * multiple adjacent string literals. 892 * 893 * It recognizes C, C++, and shell style comments. 894 * 895 * @param {Object} options a set of optional parameters. 896 * @return {function (JobT)} a function that examines the source code 897 * in the input job and builds a decoration list which it attaches to 898 * the job. 899 */ 900 function sourceDecorator(options) { 901 var shortcutStylePatterns = [], fallthroughStylePatterns = []; 902 if (options['tripleQuotedStrings']) { 903 // '''multi-line-string''', 'single-line-string', and double-quoted 904 shortcutStylePatterns.push( 905 [PR_STRING, /^(?:\'\'\'(?:[^\'\\]|\\[\s\S]|\'{1,2}(?=[^\']))*(?:\'\'\'|$)|\"\"\"(?:[^\"\\]|\\[\s\S]|\"{1,2}(?=[^\"]))*(?:\"\"\"|$)|\'(?:[^\\\']|\\[\s\S])*(?:\'|$)|\"(?:[^\\\"]|\\[\s\S])*(?:\"|$))/, 906 null, '\'"']); 907 } else if (options['multiLineStrings']) { 908 // 'multi-line-string', "multi-line-string" 909 shortcutStylePatterns.push( 910 [PR_STRING, /^(?:\'(?:[^\\\']|\\[\s\S])*(?:\'|$)|\"(?:[^\\\"]|\\[\s\S])*(?:\"|$)|\`(?:[^\\\`]|\\[\s\S])*(?:\`|$))/, 911 null, '\'"`']); 912 } else { 913 // 'single-line-string', "single-line-string" 914 shortcutStylePatterns.push( 915 [PR_STRING, 916 /^(?:\'(?:[^\\\'\r\n]|\\.)*(?:\'|$)|\"(?:[^\\\"\r\n]|\\.)*(?:\"|$))/, 917 null, '"\'']); 918 } 919 if (options['verbatimStrings']) { 920 // verbatim-string-literal production from the C# grammar. See issue 93. 921 fallthroughStylePatterns.push( 922 [PR_STRING, /^@\"(?:[^\"]|\"\")*(?:\"|$)/, null]); 923 } 924 var hc = options['hashComments']; 925 if (hc) { 926 if (options['cStyleComments']) { 927 if (hc > 1) { // multiline hash comments 928 shortcutStylePatterns.push( 929 [PR_COMMENT, /^#(?:##(?:[^#]|#(?!##))*(?:###|$)|.*)/, null, '#']); 930 } else { 931 // Stop C preprocessor declarations at an unclosed open comment 932 shortcutStylePatterns.push( 933 [PR_COMMENT, /^#(?:(?:define|e(?:l|nd)if|else|error|ifn?def|include|line|pragma|undef|warning)\b|[^\r\n]*)/, 934 null, '#']); 935 } 936 // #include <stdio.h> 937 fallthroughStylePatterns.push( 938 [PR_STRING, 939 /^<(?:(?:(?:\.\.\/)*|\/?)(?:[\w-]+(?:\/[\w-]+)+)?[\w-]+\.h(?:h|pp|\+\+)?|[a-z]\w*)>/, 940 null]); 941 } else { 942 shortcutStylePatterns.push([PR_COMMENT, /^#[^\r\n]*/, null, '#']); 943 } 944 } 945 if (options['cStyleComments']) { 946 fallthroughStylePatterns.push([PR_COMMENT, /^\/\/[^\r\n]*/, null]); 947 fallthroughStylePatterns.push( 948 [PR_COMMENT, /^\/\*[\s\S]*?(?:\*\/|$)/, null]); 949 } 950 var regexLiterals = options['regexLiterals']; 951 if (regexLiterals) { 952 /** 953 * @const 954 */ 955 var regexExcls = regexLiterals > 1 956 ? '' // Multiline regex literals 957 : '\n\r'; 958 /** 959 * @const 960 */ 961 var regexAny = regexExcls ? '.' : '[\\S\\s]'; 962 /** 963 * @const 964 */ 965 var REGEX_LITERAL = ( 966 // A regular expression literal starts with a slash that is 967 // not followed by * or / so that it is not confused with 968 // comments. 969 '/(?=[^/*' + regexExcls + '])' 970 // and then contains any number of raw characters, 971 + '(?:[^/\\x5B\\x5C' + regexExcls + ']' 972 // escape sequences (\x5C), 973 + '|\\x5C' + regexAny 974 // or non-nesting character sets (\x5B\x5D); 975 + '|\\x5B(?:[^\\x5C\\x5D' + regexExcls + ']' 976 + '|\\x5C' + regexAny + ')*(?:\\x5D|$))+' 977 // finally closed by a /. 978 + '/'); 979 fallthroughStylePatterns.push( 980 ['lang-regex', 981 RegExp('^' + REGEXP_PRECEDER_PATTERN + '(' + REGEX_LITERAL + ')') 982 ]); 983 } 984 985 var types = options['types']; 986 if (types) { 987 fallthroughStylePatterns.push([PR_TYPE, types]); 988 } 989 990 var keywords = ("" + options['keywords']).replace(/^ | $/g, ''); 991 if (keywords.length) { 992 fallthroughStylePatterns.push( 993 [PR_KEYWORD, 994 new RegExp('^(?:' + keywords.replace(/[\s,]+/g, '|') + ')\\b'), 995 null]); 996 } 997 998 shortcutStylePatterns.push([PR_PLAIN, /^\s+/, null, ' \r\n\t\xA0']); 999 1000 var punctuation = 1001 // The Bash man page says 1002 1003 // A word is a sequence of characters considered as a single 1004 // unit by GRUB. Words are separated by metacharacters, 1005 // which are the following plus space, tab, and newline: { } 1006 // | & $ ; < > 1007 // ... 1008 1009 // A word beginning with # causes that word and all remaining 1010 // characters on that line to be ignored. 1011 1012 // which means that only a '#' after /(?:^|[{}|&$;<>\s])/ starts a 1013 // comment but empirically 1014 // $ echo {#} 1015 // {#} 1016 // $ echo \$# 1017 // $# 1018 // $ echo }# 1019 // }# 1020 1021 // so /(?:^|[|&;<>\s])/ is more appropriate. 1022 1023 // http://gcc.gnu.org/onlinedocs/gcc-2.95.3/cpp_1.html#SEC3 1024 // suggests that this definition is compatible with a 1025 // default mode that tries to use a single token definition 1026 // to recognize both bash/python style comments and C 1027 // preprocessor directives. 1028 1029 // This definition of punctuation does not include # in the list of 1030 // follow-on exclusions, so # will not be broken before if preceeded 1031 // by a punctuation character. We could try to exclude # after 1032 // [|&;<>] but that doesn't seem to cause many major problems. 1033 // If that does turn out to be a problem, we should change the below 1034 // when hc is truthy to include # in the run of punctuation characters 1035 // only when not followint [|&;<>]. 1036 '^.[^\\s\\w.$@\'"`/\\\\]*'; 1037 if (options['regexLiterals']) { 1038 punctuation += '(?!\s*\/)'; 1039 } 1040 1041 fallthroughStylePatterns.push( 1042 // TODO(mikesamuel): recognize non-latin letters and numerals in idents 1043 [PR_LITERAL, /^@[a-z_$][a-z_$@0-9]*/i, null], 1044 [PR_TYPE, /^(?:[@_]?[A-Z]+[a-z][A-Za-z_$@0-9]*|\w+_t\b)/, null], 1045 [PR_PLAIN, /^[a-z_$][a-z_$@0-9]*/i, null], 1046 [PR_LITERAL, 1047 new RegExp( 1048 '^(?:' 1049 // A hex number 1050 + '0x[a-f0-9]+' 1051 // or an octal or decimal number, 1052 + '|(?:\\d(?:_\\d+)*\\d*(?:\\.\\d*)?|\\.\\d\\+)' 1053 // possibly in scientific notation 1054 + '(?:e[+\\-]?\\d+)?' 1055 + ')' 1056 // with an optional modifier like UL for unsigned long 1057 + '[a-z]*', 'i'), 1058 null, '0123456789'], 1059 // Don't treat escaped quotes in bash as starting strings. 1060 // See issue 144. 1061 [PR_PLAIN, /^\\[\s\S]?/, null], 1062 [PR_PUNCTUATION, new RegExp(punctuation), null]); 1063 1064 return createSimpleLexer(shortcutStylePatterns, fallthroughStylePatterns); 1065 } 1066 1067 var decorateSource = sourceDecorator({ 1068 'keywords': ALL_KEYWORDS, 1069 'hashComments': true, 1070 'cStyleComments': true, 1071 'multiLineStrings': true, 1072 'regexLiterals': true 1073 }); 1074 1075 /** 1076 * Given a DOM subtree, wraps it in a list, and puts each line into its own 1077 * list item. 1078 * 1079 * @param {Node} node modified in place. Its content is pulled into an 1080 * HTMLOListElement, and each line is moved into a separate list item. 1081 * This requires cloning elements, so the input might not have unique 1082 * IDs after numbering. 1083 * @param {number|null|boolean} startLineNum 1084 * If truthy, coerced to an integer which is the 1-indexed line number 1085 * of the first line of code. The number of the first line will be 1086 * attached to the list. 1087 * @param {boolean} isPreformatted true iff white-space in text nodes should 1088 * be treated as significant. 1089 */ 1090 function numberLines(node, startLineNum, isPreformatted) { 1091 var nocode = /(?:^|\s)nocode(?:\s|$)/; 1092 var lineBreak = /\r\n?|\n/; 1093 1094 var document = node.ownerDocument; 1095 1096 var li = document.createElement('li'); 1097 while (node.firstChild) { 1098 li.appendChild(node.firstChild); 1099 } 1100 // An array of lines. We split below, so this is initialized to one 1101 // un-split line. 1102 var listItems = [li]; 1103 1104 function walk(node) { 1105 var type = node.nodeType; 1106 if (type == 1 && !nocode.test(node.className)) { // Element 1107 if ('br' === node.nodeName.toLowerCase()) { 1108 breakAfter(node); 1109 // Discard the <BR> since it is now flush against a </LI>. 1110 if (node.parentNode) { 1111 node.parentNode.removeChild(node); 1112 } 1113 } else { 1114 for (var child = node.firstChild; child; child = child.nextSibling) { 1115 walk(child); 1116 } 1117 } 1118 } else if ((type == 3 || type == 4) && isPreformatted) { // Text 1119 var text = node.nodeValue; 1120 var match = text.match(lineBreak); 1121 if (match) { 1122 var firstLine = text.substring(0, match.index); 1123 node.nodeValue = firstLine; 1124 var tail = text.substring(match.index + match[0].length); 1125 if (tail) { 1126 var parent = node.parentNode; 1127 parent.insertBefore( 1128 document.createTextNode(tail), node.nextSibling); 1129 } 1130 breakAfter(node); 1131 if (!firstLine) { 1132 // Don't leave blank text nodes in the DOM. 1133 node.parentNode.removeChild(node); 1134 } 1135 } 1136 } 1137 } 1138 1139 // Split a line after the given node. 1140 function breakAfter(lineEndNode) { 1141 // If there's nothing to the right, then we can skip ending the line 1142 // here, and move root-wards since splitting just before an end-tag 1143 // would require us to create a bunch of empty copies. 1144 while (!lineEndNode.nextSibling) { 1145 lineEndNode = lineEndNode.parentNode; 1146 if (!lineEndNode) { return; } 1147 } 1148 1149 function breakLeftOf(limit, copy) { 1150 // Clone shallowly if this node needs to be on both sides of the break. 1151 var rightSide = copy ? limit.cloneNode(false) : limit; 1152 var parent = limit.parentNode; 1153 if (parent) { 1154 // We clone the parent chain. 1155 // This helps us resurrect important styling elements that cross lines. 1156 // E.g. in <i>Foo<br>Bar</i> 1157 // should be rewritten to <li><i>Foo</i></li><li><i>Bar</i></li>. 1158 var parentClone = breakLeftOf(parent, 1); 1159 // Move the clone and everything to the right of the original 1160 // onto the cloned parent. 1161 var next = limit.nextSibling; 1162 parentClone.appendChild(rightSide); 1163 for (var sibling = next; sibling; sibling = next) { 1164 next = sibling.nextSibling; 1165 parentClone.appendChild(sibling); 1166 } 1167 } 1168 return rightSide; 1169 } 1170 1171 var copiedListItem = breakLeftOf(lineEndNode.nextSibling, 0); 1172 1173 // Walk the parent chain until we reach an unattached LI. 1174 for (var parent; 1175 // Check nodeType since IE invents document fragments. 1176 (parent = copiedListItem.parentNode) && parent.nodeType === 1;) { 1177 copiedListItem = parent; 1178 } 1179 // Put it on the list of lines for later processing. 1180 listItems.push(copiedListItem); 1181 } 1182 1183 // Split lines while there are lines left to split. 1184 for (var i = 0; // Number of lines that have been split so far. 1185 i < listItems.length; // length updated by breakAfter calls. 1186 ++i) { 1187 walk(listItems[i]); 1188 } 1189 1190 // Make sure numeric indices show correctly. 1191 if (startLineNum === (startLineNum|0)) { 1192 listItems[0].setAttribute('value', startLineNum); 1193 } 1194 1195 var ol = document.createElement('ol'); 1196 ol.className = 'linenums'; 1197 var offset = Math.max(0, ((startLineNum - 1 /* zero index */)) | 0) || 0; 1198 for (var i = 0, n = listItems.length; i < n; ++i) { 1199 li = listItems[i]; 1200 // Stick a class on the LIs so that stylesheets can 1201 // color odd/even rows, or any other row pattern that 1202 // is co-prime with 10. 1203 li.className = 'L' + ((i + offset) % 10); 1204 if (!li.firstChild) { 1205 li.appendChild(document.createTextNode('\xA0')); 1206 } 1207 ol.appendChild(li); 1208 } 1209 1210 node.appendChild(ol); 1211 } 1212 1213 1214 /** 1215 * Breaks {@code job.sourceCode} around style boundaries in 1216 * {@code job.decorations} and modifies {@code job.sourceNode} in place. 1217 * @param {JobT} job 1218 * @private 1219 */ 1220 function recombineTagsAndDecorations(job) { 1221 var isIE8OrEarlier = /\bMSIE\s(\d+)/.exec(navigator.userAgent); 1222 isIE8OrEarlier = isIE8OrEarlier && +isIE8OrEarlier[1] <= 8; 1223 var newlineRe = /\n/g; 1224 1225 var source = job.sourceCode; 1226 var sourceLength = source.length; 1227 // Index into source after the last code-unit recombined. 1228 var sourceIndex = 0; 1229 1230 var spans = job.spans; 1231 var nSpans = spans.length; 1232 // Index into spans after the last span which ends at or before sourceIndex. 1233 var spanIndex = 0; 1234 1235 var decorations = job.decorations; 1236 var nDecorations = decorations.length; 1237 // Index into decorations after the last decoration which ends at or before 1238 // sourceIndex. 1239 var decorationIndex = 0; 1240 1241 // Remove all zero-length decorations. 1242 decorations[nDecorations] = sourceLength; 1243 var decPos, i; 1244 for (i = decPos = 0; i < nDecorations;) { 1245 if (decorations[i] !== decorations[i + 2]) { 1246 decorations[decPos++] = decorations[i++]; 1247 decorations[decPos++] = decorations[i++]; 1248 } else { 1249 i += 2; 1250 } 1251 } 1252 nDecorations = decPos; 1253 1254 // Simplify decorations. 1255 for (i = decPos = 0; i < nDecorations;) { 1256 var startPos = decorations[i]; 1257 // Conflate all adjacent decorations that use the same style. 1258 var startDec = decorations[i + 1]; 1259 var end = i + 2; 1260 while (end + 2 <= nDecorations && decorations[end + 1] === startDec) { 1261 end += 2; 1262 } 1263 decorations[decPos++] = startPos; 1264 decorations[decPos++] = startDec; 1265 i = end; 1266 } 1267 1268 nDecorations = decorations.length = decPos; 1269 1270 var sourceNode = job.sourceNode; 1271 var oldDisplay = ""; 1272 if (sourceNode) { 1273 oldDisplay = sourceNode.style.display; 1274 sourceNode.style.display = 'none'; 1275 } 1276 try { 1277 var decoration = null; 1278 while (spanIndex < nSpans) { 1279 var spanStart = spans[spanIndex]; 1280 var spanEnd = /** @type{number} */ (spans[spanIndex + 2]) 1281 || sourceLength; 1282 1283 var decEnd = decorations[decorationIndex + 2] || sourceLength; 1284 1285 var end = Math.min(spanEnd, decEnd); 1286 1287 var textNode = /** @type{Node} */ (spans[spanIndex + 1]); 1288 var styledText; 1289 if (textNode.nodeType !== 1 // Don't muck with <BR>s or <LI>s 1290 // Don't introduce spans around empty text nodes. 1291 && (styledText = source.substring(sourceIndex, end))) { 1292 // This may seem bizarre, and it is. Emitting LF on IE causes the 1293 // code to display with spaces instead of line breaks. 1294 // Emitting Windows standard issue linebreaks (CRLF) causes a blank 1295 // space to appear at the beginning of every line but the first. 1296 // Emitting an old Mac OS 9 line separator makes everything spiffy. 1297 if (isIE8OrEarlier) { 1298 styledText = styledText.replace(newlineRe, '\r'); 1299 } 1300 textNode.nodeValue = styledText; 1301 var document = textNode.ownerDocument; 1302 var span = document.createElement('span'); 1303 span.className = decorations[decorationIndex + 1]; 1304 var parentNode = textNode.parentNode; 1305 parentNode.replaceChild(span, textNode); 1306 span.appendChild(textNode); 1307 if (sourceIndex < spanEnd) { // Split off a text node. 1308 spans[spanIndex + 1] = textNode 1309 // TODO: Possibly optimize by using '' if there's no flicker. 1310 = document.createTextNode(source.substring(end, spanEnd)); 1311 parentNode.insertBefore(textNode, span.nextSibling); 1312 } 1313 } 1314 1315 sourceIndex = end; 1316 1317 if (sourceIndex >= spanEnd) { 1318 spanIndex += 2; 1319 } 1320 if (sourceIndex >= decEnd) { 1321 decorationIndex += 2; 1322 } 1323 } 1324 } finally { 1325 if (sourceNode) { 1326 sourceNode.style.display = oldDisplay; 1327 } 1328 } 1329 } 1330 1331 1332 /** Maps language-specific file extensions to handlers. */ 1333 var langHandlerRegistry = {}; 1334 /** Register a language handler for the given file extensions. 1335 * @param {function (JobT)} handler a function from source code to a list 1336 * of decorations. Takes a single argument job which describes the 1337 * state of the computation and attaches the decorations to it. 1338 * @param {Array.<string>} fileExtensions 1339 */ 1340 function registerLangHandler(handler, fileExtensions) { 1341 for (var i = fileExtensions.length; --i >= 0;) { 1342 var ext = fileExtensions[i]; 1343 if (!langHandlerRegistry.hasOwnProperty(ext)) { 1344 langHandlerRegistry[ext] = handler; 1345 } else if (win['console']) { 1346 console['warn']('cannot override language handler %s', ext); 1347 } 1348 } 1349 } 1350 function langHandlerForExtension(extension, source) { 1351 if (!(extension && langHandlerRegistry.hasOwnProperty(extension))) { 1352 // Treat it as markup if the first non whitespace character is a < and 1353 // the last non-whitespace character is a >. 1354 extension = /^\s*</.test(source) 1355 ? 'default-markup' 1356 : 'default-code'; 1357 } 1358 return langHandlerRegistry[extension]; 1359 } 1360 registerLangHandler(decorateSource, ['default-code']); 1361 registerLangHandler( 1362 createSimpleLexer( 1363 [], 1364 [ 1365 [PR_PLAIN, /^[^<?]+/], 1366 [PR_DECLARATION, /^<!\w[^>]*(?:>|$)/], 1367 [PR_COMMENT, /^<\!--[\s\S]*?(?:-\->|$)/], 1368 // Unescaped content in an unknown language 1369 ['lang-', /^<\?([\s\S]+?)(?:\?>|$)/], 1370 ['lang-', /^<%([\s\S]+?)(?:%>|$)/], 1371 [PR_PUNCTUATION, /^(?:<[%?]|[%?]>)/], 1372 ['lang-', /^<xmp\b[^>]*>([\s\S]+?)<\/xmp\b[^>]*>/i], 1373 // Unescaped content in javascript. (Or possibly vbscript). 1374 ['lang-js', /^<script\b[^>]*>([\s\S]*?)(<\/script\b[^>]*>)/i], 1375 // Contains unescaped stylesheet content 1376 ['lang-css', /^<style\b[^>]*>([\s\S]*?)(<\/style\b[^>]*>)/i], 1377 ['lang-in.tag', /^(<\/?[a-z][^<>]*>)/i] 1378 ]), 1379 ['default-markup', 'htm', 'html', 'mxml', 'xhtml', 'xml', 'xsl']); 1380 registerLangHandler( 1381 createSimpleLexer( 1382 [ 1383 [PR_PLAIN, /^[\s]+/, null, ' \t\r\n'], 1384 [PR_ATTRIB_VALUE, /^(?:\"[^\"]*\"?|\'[^\']*\'?)/, null, '\"\''] 1385 ], 1386 [ 1387 [PR_TAG, /^^<\/?[a-z](?:[\w.:-]*\w)?|\/?>$/i], 1388 [PR_ATTRIB_NAME, /^(?!style[\s=]|on)[a-z](?:[\w:-]*\w)?/i], 1389 ['lang-uq.val', /^=\s*([^>\'\"\s]*(?:[^>\'\"\s\/]|\/(?=\s)))/], 1390 [PR_PUNCTUATION, /^[=<>\/]+/], 1391 ['lang-js', /^on\w+\s*=\s*\"([^\"]+)\"/i], 1392 ['lang-js', /^on\w+\s*=\s*\'([^\']+)\'/i], 1393 ['lang-js', /^on\w+\s*=\s*([^\"\'>\s]+)/i], 1394 ['lang-css', /^style\s*=\s*\"([^\"]+)\"/i], 1395 ['lang-css', /^style\s*=\s*\'([^\']+)\'/i], 1396 ['lang-css', /^style\s*=\s*([^\"\'>\s]+)/i] 1397 ]), 1398 ['in.tag']); 1399 registerLangHandler( 1400 createSimpleLexer([], [[PR_ATTRIB_VALUE, /^[\s\S]+/]]), ['uq.val']); 1401 registerLangHandler(sourceDecorator({ 1402 'keywords': CPP_KEYWORDS, 1403 'hashComments': true, 1404 'cStyleComments': true, 1405 'types': C_TYPES 1406 }), ['c', 'cc', 'cpp', 'cxx', 'cyc', 'm']); 1407 registerLangHandler(sourceDecorator({ 1408 'keywords': 'null,true,false' 1409 }), ['json']); 1410 registerLangHandler(sourceDecorator({ 1411 'keywords': CSHARP_KEYWORDS, 1412 'hashComments': true, 1413 'cStyleComments': true, 1414 'verbatimStrings': true, 1415 'types': C_TYPES 1416 }), ['cs']); 1417 registerLangHandler(sourceDecorator({ 1418 'keywords': JAVA_KEYWORDS, 1419 'cStyleComments': true 1420 }), ['java']); 1421 registerLangHandler(sourceDecorator({ 1422 'keywords': SH_KEYWORDS, 1423 'hashComments': true, 1424 'multiLineStrings': true 1425 }), ['bash', 'bsh', 'csh', 'sh']); 1426 registerLangHandler(sourceDecorator({ 1427 'keywords': PYTHON_KEYWORDS, 1428 'hashComments': true, 1429 'multiLineStrings': true, 1430 'tripleQuotedStrings': true 1431 }), ['cv', 'py', 'python']); 1432 registerLangHandler(sourceDecorator({ 1433 'keywords': PERL_KEYWORDS, 1434 'hashComments': true, 1435 'multiLineStrings': true, 1436 'regexLiterals': 2 // multiline regex literals 1437 }), ['perl', 'pl', 'pm']); 1438 registerLangHandler(sourceDecorator({ 1439 'keywords': RUBY_KEYWORDS, 1440 'hashComments': true, 1441 'multiLineStrings': true, 1442 'regexLiterals': true 1443 }), ['rb', 'ruby']); 1444 registerLangHandler(sourceDecorator({ 1445 'keywords': JSCRIPT_KEYWORDS, 1446 'cStyleComments': true, 1447 'regexLiterals': true 1448 }), ['javascript', 'js', 'ts', 'typescript']); 1449 registerLangHandler(sourceDecorator({ 1450 'keywords': COFFEE_KEYWORDS, 1451 'hashComments': 3, // ### style block comments 1452 'cStyleComments': true, 1453 'multilineStrings': true, 1454 'tripleQuotedStrings': true, 1455 'regexLiterals': true 1456 }), ['coffee']); 1457 registerLangHandler( 1458 createSimpleLexer([], [[PR_STRING, /^[\s\S]+/]]), ['regex']); 1459 1460 /** @param {JobT} job */ 1461 function applyDecorator(job) { 1462 var opt_langExtension = job.langExtension; 1463 1464 try { 1465 // Extract tags, and convert the source code to plain text. 1466 var sourceAndSpans = extractSourceSpans(job.sourceNode, job.pre); 1467 /** Plain text. @type {string} */ 1468 var source = sourceAndSpans.sourceCode; 1469 job.sourceCode = source; 1470 job.spans = sourceAndSpans.spans; 1471 job.basePos = 0; 1472 1473 // Apply the appropriate language handler 1474 langHandlerForExtension(opt_langExtension, source)(job); 1475 1476 // Integrate the decorations and tags back into the source code, 1477 // modifying the sourceNode in place. 1478 recombineTagsAndDecorations(job); 1479 } catch (e) { 1480 if (win['console']) { 1481 console['log'](e && e['stack'] || e); 1482 } 1483 } 1484 } 1485 1486 /** 1487 * Pretty print a chunk of code. 1488 * @param sourceCodeHtml {string} The HTML to pretty print. 1489 * @param opt_langExtension {string} The language name to use. 1490 * Typically, a filename extension like 'cpp' or 'java'. 1491 * @param opt_numberLines {number|boolean} True to number lines, 1492 * or the 1-indexed number of the first line in sourceCodeHtml. 1493 */ 1494 function $prettyPrintOne(sourceCodeHtml, opt_langExtension, opt_numberLines) { 1495 /** @type{number|boolean} */ 1496 var nl = opt_numberLines || false; 1497 /** @type{string|null} */ 1498 var langExtension = opt_langExtension || null; 1499 /** @type{!Element} */ 1500 var container = document.createElement('div'); 1501 // This could cause images to load and onload listeners to fire. 1502 // E.g. <img onerror="alert(1337)" src="nosuchimage.png">. 1503 // We assume that the inner HTML is from a trusted source. 1504 // The pre-tag is required for IE8 which strips newlines from innerHTML 1505 // when it is injected into a <pre> tag. 1506 // http://stackoverflow.com/questions/451486/pre-tag-loses-line-breaks-when-setting-innerhtml-in-ie 1507 // http://stackoverflow.com/questions/195363/inserting-a-newline-into-a-pre-tag-ie-javascript 1508 container.innerHTML = '<pre>' + sourceCodeHtml + '</pre>'; 1509 container = /** @type{!Element} */(container.firstChild); 1510 if (nl) { 1511 numberLines(container, nl, true); 1512 } 1513 1514 /** @type{JobT} */ 1515 var job = { 1516 langExtension: langExtension, 1517 numberLines: nl, 1518 sourceNode: container, 1519 pre: 1, 1520 sourceCode: null, 1521 basePos: null, 1522 spans: null, 1523 decorations: null 1524 }; 1525 applyDecorator(job); 1526 return container.innerHTML; 1527 } 1528 1529 /** 1530 * Find all the {@code <pre>} and {@code <code>} tags in the DOM with 1531 * {@code class=prettyprint} and prettify them. 1532 * 1533 * @param {Function} opt_whenDone called when prettifying is done. 1534 * @param {HTMLElement|HTMLDocument} opt_root an element or document 1535 * containing all the elements to pretty print. 1536 * Defaults to {@code document.body}. 1537 */ 1538 function $prettyPrint(opt_whenDone, opt_root) { 1539 var root = opt_root || document.body; 1540 var doc = root.ownerDocument || document; 1541 function byTagName(tn) { return root.getElementsByTagName(tn); } 1542 // fetch a list of nodes to rewrite 1543 var codeSegments = [byTagName('pre'), byTagName('code'), byTagName('xmp')]; 1544 var elements = []; 1545 for (var i = 0; i < codeSegments.length; ++i) { 1546 for (var j = 0, n = codeSegments[i].length; j < n; ++j) { 1547 elements.push(codeSegments[i][j]); 1548 } 1549 } 1550 codeSegments = null; 1551 1552 var clock = Date; 1553 if (!clock['now']) { 1554 clock = { 'now': function () { return +(new Date); } }; 1555 } 1556 1557 // The loop is broken into a series of continuations to make sure that we 1558 // don't make the browser unresponsive when rewriting a large page. 1559 var k = 0; 1560 1561 var langExtensionRe = /\blang(?:uage)?-([\w.]+)(?!\S)/; 1562 var prettyPrintRe = /\bprettyprint\b/; 1563 var prettyPrintedRe = /\bprettyprinted\b/; 1564 var preformattedTagNameRe = /pre|xmp/i; 1565 var codeRe = /^code$/i; 1566 var preCodeXmpRe = /^(?:pre|code|xmp)$/i; 1567 var EMPTY = {}; 1568 1569 function doWork() { 1570 var endTime = (win['PR_SHOULD_USE_CONTINUATION'] ? 1571 clock['now']() + 250 /* ms */ : 1572 Infinity); 1573 for (; k < elements.length && clock['now']() < endTime; k++) { 1574 var cs = elements[k]; 1575 1576 // Look for a preceding comment like 1577 // <?prettify lang="..." linenums="..."?> 1578 var attrs = EMPTY; 1579 { 1580 for (var preceder = cs; (preceder = preceder.previousSibling);) { 1581 var nt = preceder.nodeType; 1582 // <?foo?> is parsed by HTML 5 to a comment node (8) 1583 // like <!--?foo?-->, but in XML is a processing instruction 1584 var value = (nt === 7 || nt === 8) && preceder.nodeValue; 1585 if (value 1586 ? !/^\??prettify\b/.test(value) 1587 : (nt !== 3 || /\S/.test(preceder.nodeValue))) { 1588 // Skip over white-space text nodes but not others. 1589 break; 1590 } 1591 if (value) { 1592 attrs = {}; 1593 value.replace( 1594 /\b(\w+)=([\w:.%+-]+)/g, 1595 function (_, name, value) { attrs[name] = value; }); 1596 break; 1597 } 1598 } 1599 } 1600 1601 var className = cs.className; 1602 if ((attrs !== EMPTY || prettyPrintRe.test(className)) 1603 // Don't redo this if we've already done it. 1604 // This allows recalling pretty print to just prettyprint elements 1605 // that have been added to the page since last call. 1606 && !prettyPrintedRe.test(className)) { 1607 1608 // make sure this is not nested in an already prettified element 1609 var nested = false; 1610 for (var p = cs.parentNode; p; p = p.parentNode) { 1611 var tn = p.tagName; 1612 if (preCodeXmpRe.test(tn) 1613 && p.className && prettyPrintRe.test(p.className)) { 1614 nested = true; 1615 break; 1616 } 1617 } 1618 if (!nested) { 1619 // Mark done. If we fail to prettyprint for whatever reason, 1620 // we shouldn't try again. 1621 cs.className += ' prettyprinted'; 1622 1623 // If the classes includes a language extensions, use it. 1624 // Language extensions can be specified like 1625 // <pre class="prettyprint lang-cpp"> 1626 // the language extension "cpp" is used to find a language handler 1627 // as passed to PR.registerLangHandler. 1628 // HTML5 recommends that a language be specified using "language-" 1629 // as the prefix instead. Google Code Prettify supports both. 1630 // http://dev.w3.org/html5/spec-author-view/the-code-element.html 1631 var langExtension = attrs['lang']; 1632 if (!langExtension) { 1633 langExtension = className.match(langExtensionRe); 1634 // Support <pre class="prettyprint"><code class="language-c"> 1635 var wrapper; 1636 if (!langExtension && (wrapper = childContentWrapper(cs)) 1637 && codeRe.test(wrapper.tagName)) { 1638 langExtension = wrapper.className.match(langExtensionRe); 1639 } 1640 1641 if (langExtension) { langExtension = langExtension[1]; } 1642 } 1643 1644 var preformatted; 1645 if (preformattedTagNameRe.test(cs.tagName)) { 1646 preformatted = 1; 1647 } else { 1648 var currentStyle = cs['currentStyle']; 1649 var defaultView = doc.defaultView; 1650 var whitespace = ( 1651 currentStyle 1652 ? currentStyle['whiteSpace'] 1653 : (defaultView 1654 && defaultView.getComputedStyle) 1655 ? defaultView.getComputedStyle(cs, null) 1656 .getPropertyValue('white-space') 1657 : 0); 1658 preformatted = whitespace 1659 && 'pre' === whitespace.substring(0, 3); 1660 } 1661 1662 // Look for a class like linenums or linenums:<n> where <n> is the 1663 // 1-indexed number of the first line. 1664 var lineNums = attrs['linenums']; 1665 if (!(lineNums = lineNums === 'true' || +lineNums)) { 1666 lineNums = className.match(/\blinenums\b(?::(\d+))?/); 1667 lineNums = 1668 lineNums 1669 ? lineNums[1] && lineNums[1].length 1670 ? +lineNums[1] : true 1671 : false; 1672 } 1673 if (lineNums) { numberLines(cs, lineNums, preformatted); } 1674 1675 // do the pretty printing 1676 var prettyPrintingJob = { 1677 langExtension: langExtension, 1678 sourceNode: cs, 1679 numberLines: lineNums, 1680 pre: preformatted, 1681 sourceCode: null, 1682 basePos: null, 1683 spans: null, 1684 decorations: null 1685 }; 1686 applyDecorator(prettyPrintingJob); 1687 } 1688 } 1689 } 1690 if (k < elements.length) { 1691 // finish up in a continuation 1692 win.setTimeout(doWork, 250); 1693 } else if ('function' === typeof opt_whenDone) { 1694 opt_whenDone(); 1695 } 1696 } 1697 1698 doWork(); 1699 } 1700 1701 /** 1702 * Contains functions for creating and registering new language handlers. 1703 * @type {Object} 1704 */ 1705 var PR = win['PR'] = { 1706 'createSimpleLexer': createSimpleLexer, 1707 'registerLangHandler': registerLangHandler, 1708 'sourceDecorator': sourceDecorator, 1709 'PR_ATTRIB_NAME': PR_ATTRIB_NAME, 1710 'PR_ATTRIB_VALUE': PR_ATTRIB_VALUE, 1711 'PR_COMMENT': PR_COMMENT, 1712 'PR_DECLARATION': PR_DECLARATION, 1713 'PR_KEYWORD': PR_KEYWORD, 1714 'PR_LITERAL': PR_LITERAL, 1715 'PR_NOCODE': PR_NOCODE, 1716 'PR_PLAIN': PR_PLAIN, 1717 'PR_PUNCTUATION': PR_PUNCTUATION, 1718 'PR_SOURCE': PR_SOURCE, 1719 'PR_STRING': PR_STRING, 1720 'PR_TAG': PR_TAG, 1721 'PR_TYPE': PR_TYPE, 1722 'prettyPrintOne': 1723 IN_GLOBAL_SCOPE 1724 ? (win['prettyPrintOne'] = $prettyPrintOne) 1725 : (prettyPrintOne = $prettyPrintOne), 1726 'prettyPrint': 1727 IN_GLOBAL_SCOPE 1728 ? (win['prettyPrint'] = $prettyPrint) 1729 : (prettyPrint = $prettyPrint) 1730 }; 1731 1732 // Make PR available via the Asynchronous Module Definition (AMD) API. 1733 // Per https://github.com/amdjs/amdjs-api/wiki/AMD: 1734 // The Asynchronous Module Definition (AMD) API specifies a 1735 // mechanism for defining modules such that the module and its 1736 // dependencies can be asynchronously loaded. 1737 // ... 1738 // To allow a clear indicator that a global define function (as 1739 // needed for script src browser loading) conforms to the AMD API, 1740 // any global define function SHOULD have a property called "amd" 1741 // whose value is an object. This helps avoid conflict with any 1742 // other existing JavaScript code that could have defined a define() 1743 // function that does not conform to the AMD API. 1744 var define = win['define']; 1745 if (typeof define === "function" && define['amd']) { 1746 define("google-code-prettify", [], function () { 1747 return PR; 1748 }); 1749 } 1750})(); 1751