resources/lib/CLDRPluralRuleParser/CLDRPluralRuleParser.js

   1 /**
   2  * cldrpluralparser.js
   3  * A parser engine for CLDR plural rules.
   4  *
   5  * Copyright 2012-2014 Santhosh Thottingal and other contributors
   6  * Released under the MIT license
   7  * http://opensource.org/licenses/MIT
   8  *
   9  * @version 0.1.0
  10  * @source https://github.com/santhoshtr/CLDRPluralRuleParser
  11  * @author Santhosh Thottingal <santhosh.thottingal@gmail.com>
  12  * @author Timo Tijhof
  13  * @author Amir Aharoni
  14  */
  15
  16 /**
  17  * Evaluates a plural rule in CLDR syntax for a number
  18  * @param {string} rule
  19  * @param {integer} number
  20  * @return {boolean} true if evaluation passed, false if evaluation failed.
  21  */
  22
  23 // UMD returnExports https://github.com/umdjs/umd/blob/master/returnExports.js
  24 (function(root, factory) {
  25         if (typeof define === 'function' && define.amd) {
  26                 // AMD. Register as an anonymous module.
  27                 define(factory);
  28         } else if (typeof exports === 'object') {
  29                 // Node. Does not work with strict CommonJS, but
  30                 // only CommonJS-like environments that support module.exports,
  31                 // like Node.
  32                 module.exports = factory();
  33         } else {
  34                 // Browser globals (root is window)
  35                 root.pluralRuleParser = factory();
  36         }
  37 }(this, function() {
  38
  39 function pluralRuleParser(rule, number) {
  40         'use strict';
  41
  42         /*
  43         Syntax: see http://unicode.org/reports/tr35/#Language_Plural_Rules
  44         -----------------------------------------------------------------
  45         condition     = and_condition ('or' and_condition)*
  46                 ('@integer' samples)?
  47                 ('@decimal' samples)?
  48         and_condition = relation ('and' relation)*
  49         relation      = is_relation | in_relation | within_relation
  50         is_relation   = expr 'is' ('not')? value
  51         in_relation   = expr (('not')? 'in' | '=' | '!=') range_list
  52         within_relation = expr ('not')? 'within' range_list
  53         expr          = operand (('mod' | '%') value)?
  54         operand       = 'n' | 'i' | 'f' | 't' | 'v' | 'w'
  55         range_list    = (range | value) (',' range_list)*
  56         value         = digit+
  57         digit         = 0|1|2|3|4|5|6|7|8|9
  58         range         = value'..'value
  59         samples       = sampleRange (',' sampleRange)* (',' ('…'|'...'))?
  60         sampleRange   = decimalValue '~' decimalValue
  61         decimalValue  = value ('.' value)?
  62         */
  63
  64         // We don't evaluate the samples section of the rule. Ignore it.
  65         rule = rule.split('@')[0].replace(/^\s*/, '').replace(/\s*$/, '');
  66
  67         if (!rule.length) {
  68                 // Empty rule or 'other' rule.
  69                 return true;
  70         }
  71
  72         // Indicates the current position in the rule as we parse through it.
  73         // Shared among all parsing functions below.
  74         var pos = 0,
  75                 operand,
  76                 expression,
  77                 relation,
  78                 result,
  79                 whitespace = makeRegexParser(/^\s+/),
  80                 value = makeRegexParser(/^\d+/),
  81                 _n_ = makeStringParser('n'),
  82                 _i_ = makeStringParser('i'),
  83                 _f_ = makeStringParser('f'),
  84                 _t_ = makeStringParser('t'),
  85                 _v_ = makeStringParser('v'),
  86                 _w_ = makeStringParser('w'),
  87                 _is_ = makeStringParser('is'),
  88                 _isnot_ = makeStringParser('is not'),
  89                 _isnot_sign_ = makeStringParser('!='),
  90                 _equal_ = makeStringParser('='),
  91                 _mod_ = makeStringParser('mod'),
  92                 _percent_ = makeStringParser('%'),
  93                 _not_ = makeStringParser('not'),
  94                 _in_ = makeStringParser('in'),
  95                 _within_ = makeStringParser('within'),
  96                 _range_ = makeStringParser('..'),
  97                 _comma_ = makeStringParser(','),
  98                 _or_ = makeStringParser('or'),
  99                 _and_ = makeStringParser('and');
 100
 101         function debug() {
 102                 // console.log.apply(console, arguments);
 103         }
 104
 105         debug('pluralRuleParser', rule, number);
 106
 107         // Try parsers until one works, if none work return null
 108         function choice(parserSyntax) {
 109                 return function() {
 110                         var i, result;
 111
 112                         for (i = 0; i < parserSyntax.length; i++) {
 113                                 result = parserSyntax[i]();
 114
 115                                 if (result !== null) {
 116                                         return result;
 117                                 }
 118                         }
 119
 120                         return null;
 121                 };
 122         }
 123
 124         // Try several parserSyntax-es in a row.
 125         // All must succeed; otherwise, return null.
 126         // This is the only eager one.
 127         function sequence(parserSyntax) {
 128                 var i, parserRes,
 129                         originalPos = pos,
 130                         result = [];
 131
 132                 for (i = 0; i < parserSyntax.length; i++) {
 133                         parserRes = parserSyntax[i]();
 134
 135                         if (parserRes === null) {
 136                                 pos = originalPos;
 137
 138                                 return null;
 139                         }
 140
 141                         result.push(parserRes);
 142                 }
 143
 144                 return result;
 145         }
 146
 147         // Run the same parser over and over until it fails.
 148         // Must succeed a minimum of n times; otherwise, return null.
 149         function nOrMore(n, p) {
 150                 return function() {
 151                         var originalPos = pos,
 152                                 result = [],
 153                                 parsed = p();
 154
 155                         while (parsed !== null) {
 156                                 result.push(parsed);
 157                                 parsed = p();
 158                         }
 159
 160                         if (result.length < n) {
 161                                 pos = originalPos;
 162
 163                                 return null;
 164                         }
 165
 166                         return result;
 167                 };
 168         }
 169
 170         // Helpers - just make parserSyntax out of simpler JS builtin types
 171         function makeStringParser(s) {
 172                 var len = s.length;
 173
 174                 return function() {
 175                         var result = null;
 176
 177                         if (rule.substr(pos, len) === s) {
 178                                 result = s;
 179                                 pos += len;
 180                         }
 181
 182                         return result;
 183                 };
 184         }
 185
 186         function makeRegexParser(regex) {
 187                 return function() {
 188                         var matches = rule.substr(pos).match(regex);
 189
 190                         if (matches === null) {
 191                                 return null;
 192                         }
 193
 194                         pos += matches[0].length;
 195
 196                         return matches[0];
 197                 };
 198         }
 199
 200         /**
 201          * Integer digits of n.
 202          */
 203         function i() {
 204                 var result = _i_();
 205
 206                 if (result === null) {
 207                         debug(' -- failed i', parseInt(number, 10));
 208
 209                         return result;
 210                 }
 211
 212                 result = parseInt(number, 10);
 213                 debug(' -- passed i ', result);
 214
 215                 return result;
 216         }
 217
 218         /**
 219          * Absolute value of the source number (integer and decimals).
 220          */
 221         function n() {
 222                 var result = _n_();
 223
 224                 if (result === null) {
 225                         debug(' -- failed n ', number);
 226
 227                         return result;
 228                 }
 229
 230                 result = parseFloat(number, 10);
 231                 debug(' -- passed n ', result);
 232
 233                 return result;
 234         }
 235
 236         /**
 237          * Visible fractional digits in n, with trailing zeros.
 238          */
 239         function f() {
 240                 var result = _f_();
 241
 242                 if (result === null) {
 243                         debug(' -- failed f ', number);
 244
 245                         return result;
 246                 }
 247
 248                 result = (number + '.').split('.')[1] || 0;
 249                 debug(' -- passed f ', result);
 250
 251                 return result;
 252         }
 253
 254         /**
 255          * Visible fractional digits in n, without trailing zeros.
 256          */
 257         function t() {
 258                 var result = _t_();
 259
 260                 if (result === null) {
 261                         debug(' -- failed t ', number);
 262
 263                         return result;
 264                 }
 265
 266                 result = (number + '.').split('.')[1].replace(/0$/, '') || 0;
 267                 debug(' -- passed t ', result);
 268
 269                 return result;
 270         }
 271
 272         /**
 273          * Number of visible fraction digits in n, with trailing zeros.
 274          */
 275         function v() {
 276                 var result = _v_();
 277
 278                 if (result === null) {
 279                         debug(' -- failed v ', number);
 280
 281                         return result;
 282                 }
 283
 284                 result = (number + '.').split('.')[1].length || 0;
 285                 debug(' -- passed v ', result);
 286
 287                 return result;
 288         }
 289
 290         /**
 291          * Number of visible fraction digits in n, without trailing zeros.
 292          */
 293         function w() {
 294                 var result = _w_();
 295
 296                 if (result === null) {
 297                         debug(' -- failed w ', number);
 298
 299                         return result;
 300                 }
 301
 302                 result = (number + '.').split('.')[1].replace(/0$/, '').length || 0;
 303                 debug(' -- passed w ', result);
 304
 305                 return result;
 306         }
 307
 308         // operand       = 'n' | 'i' | 'f' | 't' | 'v' | 'w'
 309         operand = choice([n, i, f, t, v, w]);
 310
 311         // expr          = operand (('mod' | '%') value)?
 312         expression = choice([mod, operand]);
 313
 314         function mod() {
 315                 var result = sequence(
 316                         [operand, whitespace, choice([_mod_, _percent_]), whitespace, value]
 317                 );
 318
 319                 if (result === null) {
 320                         debug(' -- failed mod');
 321
 322                         return null;
 323                 }
 324
 325                 debug(' -- passed ' + parseInt(result[0], 10) + ' ' + result[2] + ' ' + parseInt(result[4], 10));
 326
 327                 return parseInt(result[0], 10) % parseInt(result[4], 10);
 328         }
 329
 330         function not() {
 331                 var result = sequence([whitespace, _not_]);
 332
 333                 if (result === null) {
 334                         debug(' -- failed not');
 335
 336                         return null;
 337                 }
 338
 339                 return result[1];
 340         }
 341
 342         // is_relation   = expr 'is' ('not')? value
 343         function is() {
 344                 var result = sequence([expression, whitespace, choice([_is_]), whitespace, value]);
 345
 346                 if (result !== null) {
 347                         debug(' -- passed is : ' + result[0] + ' == ' + parseInt(result[4], 10));
 348
 349                         return result[0] === parseInt(result[4], 10);
 350                 }
 351
 352                 debug(' -- failed is');
 353
 354                 return null;
 355         }
 356
 357         // is_relation   = expr 'is' ('not')? value
 358         function isnot() {
 359                 var result = sequence(
 360                         [expression, whitespace, choice([_isnot_, _isnot_sign_]), whitespace, value]
 361                 );
 362
 363                 if (result !== null) {
 364                         debug(' -- passed isnot: ' + result[0] + ' != ' + parseInt(result[4], 10));
 365
 366                         return result[0] !== parseInt(result[4], 10);
 367                 }
 368
 369                 debug(' -- failed isnot');
 370
 371                 return null;
 372         }
 373
 374         function not_in() {
 375                 var i, range_list,
 376                         result = sequence([expression, whitespace, _isnot_sign_, whitespace, rangeList]);
 377
 378                 if (result !== null) {
 379                         debug(' -- passed not_in: ' + result[0] + ' != ' + result[4]);
 380                         range_list = result[4];
 381
 382                         for (i = 0; i < range_list.length; i++) {
 383                                 if (parseInt(range_list[i], 10) === parseInt(result[0], 10)) {
 384                                         return false;
 385                                 }
 386                         }
 387
 388                         return true;
 389                 }
 390
 391                 debug(' -- failed not_in');
 392
 393                 return null;
 394         }
 395
 396         // range_list    = (range | value) (',' range_list)*
 397         function rangeList() {
 398                 var result = sequence([choice([range, value]), nOrMore(0, rangeTail)]),
 399                         resultList = [];
 400
 401                 if (result !== null) {
 402                         resultList = resultList.concat(result[0]);
 403
 404                         if (result[1][0]) {
 405                                 resultList = resultList.concat(result[1][0]);
 406                         }
 407
 408                         return resultList;
 409                 }
 410
 411                 debug(' -- failed rangeList');
 412
 413                 return null;
 414         }
 415
 416         function rangeTail() {
 417                 // ',' range_list
 418                 var result = sequence([_comma_, rangeList]);
 419
 420                 if (result !== null) {
 421                         return result[1];
 422                 }
 423
 424                 debug(' -- failed rangeTail');
 425
 426                 return null;
 427         }
 428
 429         // range         = value'..'value
 430         function range() {
 431                 var i, array, left, right,
 432                         result = sequence([value, _range_, value]);
 433
 434                 if (result !== null) {
 435                         debug(' -- passed range');
 436
 437                         array = [];
 438                         left = parseInt(result[0], 10);
 439                         right = parseInt(result[2], 10);
 440
 441                         for (i = left; i <= right; i++) {
 442                                 array.push(i);
 443                         }
 444
 445                         return array;
 446                 }
 447
 448                 debug(' -- failed range');
 449
 450                 return null;
 451         }
 452
 453         function _in() {
 454                 var result, range_list, i;
 455
 456                 // in_relation   = expr ('not')? 'in' range_list
 457                 result = sequence(
 458                         [expression, nOrMore(0, not), whitespace, choice([_in_, _equal_]), whitespace, rangeList]
 459                 );
 460
 461                 if (result !== null) {
 462                         debug(' -- passed _in:' + result);
 463
 464                         range_list = result[5];
 465
 466                         for (i = 0; i < range_list.length; i++) {
 467                                 if (parseInt(range_list[i], 10) === parseInt(result[0], 10)) {
 468                                         return (result[1][0] !== 'not');
 469                                 }
 470                         }
 471
 472                         return (result[1][0] === 'not');
 473                 }
 474
 475                 debug(' -- failed _in ');
 476
 477                 return null;
 478         }
 479
 480         /**
 481          * The difference between "in" and "within" is that
 482          * "in" only includes integers in the specified range,
 483          * while "within" includes all values.
 484          */
 485         function within() {
 486                 var range_list, result;
 487
 488                 // within_relation = expr ('not')? 'within' range_list
 489                 result = sequence(
 490                         [expression, nOrMore(0, not), whitespace, _within_, whitespace, rangeList]
 491                 );
 492
 493                 if (result !== null) {
 494                         debug(' -- passed within');
 495
 496                         range_list = result[5];
 497
 498                         if ((result[0] >= parseInt(range_list[0], 10)) &&
 499                                 (result[0] < parseInt(range_list[range_list.length - 1], 10))) {
 500
 501                                 return (result[1][0] !== 'not');
 502                         }
 503
 504                         return (result[1][0] === 'not');
 505                 }
 506
 507                 debug(' -- failed within ');
 508
 509                 return null;
 510         }
 511
 512         // relation      = is_relation | in_relation | within_relation
 513         relation = choice([is, not_in, isnot, _in, within]);
 514
 515         // and_condition = relation ('and' relation)*
 516         function and() {
 517                 var i,
 518                         result = sequence([relation, nOrMore(0, andTail)]);
 519
 520                 if (result) {
 521                         if (!result[0]) {
 522                                 return false;
 523                         }
 524
 525                         for (i = 0; i < result[1].length; i++) {
 526                                 if (!result[1][i]) {
 527                                         return false;
 528                                 }
 529                         }
 530
 531                         return true;
 532                 }
 533
 534                 debug(' -- failed and');
 535
 536                 return null;
 537         }
 538
 539         // ('and' relation)*
 540         function andTail() {
 541                 var result = sequence([whitespace, _and_, whitespace, relation]);
 542
 543                 if (result !== null) {
 544                         debug(' -- passed andTail' + result);
 545
 546                         return result[3];
 547                 }
 548
 549                 debug(' -- failed andTail');
 550
 551                 return null;
 552
 553         }
 554         //  ('or' and_condition)*
 555         function orTail() {
 556                 var result = sequence([whitespace, _or_, whitespace, and]);
 557
 558                 if (result !== null) {
 559                         debug(' -- passed orTail: ' + result[3]);
 560
 561                         return result[3];
 562                 }
 563
 564                 debug(' -- failed orTail');
 565
 566                 return null;
 567         }
 568
 569         // condition     = and_condition ('or' and_condition)*
 570         function condition() {
 571                 var i,
 572                         result = sequence([and, nOrMore(0, orTail)]);
 573
 574                 if (result) {
 575                         for (i = 0; i < result[1].length; i++) {
 576                                 if (result[1][i]) {
 577                                         return true;
 578                                 }
 579                         }
 580
 581                         return result[0];
 582                 }
 583
 584                 return false;
 585         }
 586
 587         result = condition();
 588
 589         /**
 590          * For success, the pos must have gotten to the end of the rule
 591          * and returned a non-null.
 592          * n.b. This is part of language infrastructure,
 593          * so we do not throw an internationalizable message.
 594          */
 595         if (result === null) {
 596                 throw new Error('Parse error at position ' + pos.toString() + ' for rule: ' + rule);
 597         }
 598
 599         if (pos !== rule.length) {
 600                 debug('Warning: Rule not parsed completely. Parser stopped at ' + rule.substr(0, pos) + ' for rule: ' + rule);
 601         }
 602
 603         return result;
 604 }
 605
 606 return pluralRuleParser;
 607
 608 }));