Autolink.php
5 years ago
Configuration.php
5 years ago
EmojiRegex.php
5 years ago
Extractor.php
5 years ago
HitHighlighter.php
5 years ago
ParseResults.php
5 years ago
Parser.php
5 years ago
Regex.php
5 years ago
StringUtils.php
5 years ago
TldLists.php
5 years ago
Validator.php
5 years ago
Regex.php
839 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * @author Mike Cochrane <mikec@mikenz.geek.nz> |
| 5 | * @author Nick Pope <nick@nickpope.me.uk> |
| 6 | * @copyright Copyright © 2010, Mike Cochrane, Nick Pope |
| 7 | * @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0 |
| 8 | * @package Twitter.Text |
| 9 | */ |
| 10 | |
| 11 | namespace Twitter\Text; |
| 12 | |
| 13 | /** |
| 14 | * Twitter Regex Abstract Class |
| 15 | * |
| 16 | * Used by subclasses that need to parse tweets. |
| 17 | * |
| 18 | * Originally written by {@link http://github.com/mikenz Mike Cochrane}, this |
| 19 | * is based on code by {@link http://github.com/mzsanford Matt Sanford} and |
| 20 | * heavily modified by {@link http://github.com/ngnpope Nick Pope}. |
| 21 | * |
| 22 | * @author Mike Cochrane <mikec@mikenz.geek.nz> |
| 23 | * @author Nick Pope <nick@nickpope.me.uk> |
| 24 | * @copyright Copyright © 2010, Mike Cochrane, Nick Pope |
| 25 | * @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0 |
| 26 | * @package Twitter |
| 27 | */ |
| 28 | class Regex |
| 29 | { |
| 30 | /** |
| 31 | * Expression to match whitespace characters. |
| 32 | * |
| 33 | * 0x0009-0x000D Cc # <control-0009>..<control-000D> |
| 34 | * 0x0020 Zs # SPACE |
| 35 | * 0x0085 Cc # <control-0085> |
| 36 | * 0x00A0 Zs # NO-BREAK SPACE |
| 37 | * 0x1680 Zs # OGHAM SPACE MARK |
| 38 | * 0x180E Zs # MONGOLIAN VOWEL SEPARATOR |
| 39 | * 0x2000-0x200A Zs # EN QUAD..HAIR SPACE |
| 40 | * 0x2028 Zl # LINE SEPARATOR |
| 41 | * 0x2029 Zp # PARAGRAPH SEPARATOR |
| 42 | * 0x202F Zs # NARROW NO-BREAK SPACE |
| 43 | * 0x205F Zs # MEDIUM MATHEMATICAL SPACE |
| 44 | * 0x3000 Zs # IDEOGRAPHIC SPACE |
| 45 | * |
| 46 | * @var string |
| 47 | */ |
| 48 | // @codingStandardsIgnoreStart |
| 49 | private static $spaces = '\x{0009}-\x{000D}\x{0020}\x{0085}\x{00a0}\x{1680}\x{180E}\x{2000}-\x{200a}\x{2028}\x{2029}\x{202f}\x{205f}\x{3000}'; // @codingStandardsIgnoreEnd |
| 50 | |
| 51 | /** |
| 52 | * Expression to match latin accented characters. |
| 53 | * |
| 54 | * 0x00C0-0x00D6 |
| 55 | * 0x00D8-0x00F6 |
| 56 | * 0x00F8-0x00FF |
| 57 | * 0x0100-0x024f |
| 58 | * 0x0253-0x0254 |
| 59 | * 0x0256-0x0257 |
| 60 | * 0x0259 |
| 61 | * 0x025b |
| 62 | * 0x0263 |
| 63 | * 0x0268 |
| 64 | * 0x026f |
| 65 | * 0x0272 |
| 66 | * 0x0289 |
| 67 | * 0x028b |
| 68 | * 0x02bb |
| 69 | * 0x0300-0x036f |
| 70 | * 0x1e00-0x1eff |
| 71 | * |
| 72 | * Excludes 0x00D7 - multiplication sign (confusable with 'x'). |
| 73 | * Excludes 0x00F7 - division sign. |
| 74 | * |
| 75 | * @var string |
| 76 | */ |
| 77 | // @codingStandardsIgnoreStart |
| 78 | private static $latinAccents = '\x{00c0}-\x{00d6}\x{00d8}-\x{00f6}\x{00f8}-\x{00ff}\x{0100}-\x{024f}\x{0253}-\x{0254}\x{0256}-\x{0257}\x{0259}\x{025b}\x{0263}\x{0268}\x{026f}\x{0272}\x{0289}\x{028b}\x{02bb}\x{0300}-\x{036f}\x{1e00}-\x{1eff}'; // @codingStandardsIgnoreEnd |
| 79 | |
| 80 | /** |
| 81 | * Invalid Characters |
| 82 | * |
| 83 | * 0xFFFE,0xFEFF # BOM |
| 84 | * 0xFFFF # Special |
| 85 | * 0x202A-0x202E # Directional change |
| 86 | */ |
| 87 | private static $invalidCharacters = '\x{202a}-\x{202e}\x{feff}\x{fffe}\x{ffff}'; |
| 88 | |
| 89 | /** |
| 90 | * Directional Characters |
| 91 | * |
| 92 | * 0x061C ARABIC LETTER MARK (ALM) |
| 93 | * 0x200E LEFT-TO-RIGHT MARK (LRM) |
| 94 | * 0x200F RIGHT-TO-LEFT MARK (RLM) |
| 95 | * 0x202A LEFT-TO-RIGHT EMBEDDING (LRE) |
| 96 | * 0x202B RIGHT-TO-LEFT EMBEDDING (RLE) |
| 97 | * 0x202C POP DIRECTIONAL FORMATTING (PDF) |
| 98 | * 0x202D LEFT-TO-RIGHT OVERRIDE (LRO) |
| 99 | * 0x202E RIGHT-TO-LEFT OVERRIDE (RLO) |
| 100 | * 0x2066 LEFT-TO-RIGHT ISOLATE (LRI) |
| 101 | * 0x2067 RIGHT-TO-LEFT ISOLATE (RLI) |
| 102 | * 0x2068 FIRST STRONG ISOLATE (FSI) |
| 103 | * 0x2069 POP DIRECTIONAL ISOLATE (PDI) |
| 104 | */ |
| 105 | private static $directionalCharacters = '\x{061c}\x{200e}\x{200f}\x{202a}\x{202e}\x{2066}\x{2069}'; |
| 106 | |
| 107 | /** |
| 108 | * Expression to match RTL characters. |
| 109 | * |
| 110 | * 0x0600-0x06FF Arabic |
| 111 | * 0x0750-0x077F Arabic Supplement |
| 112 | * 0x08A0-0x08FF Arabic Extended-A |
| 113 | * 0x0590-0x05FF Hebrew |
| 114 | * 0xFB50-0xFDFF Arabic Presentation Forms-A |
| 115 | * 0xFE70-0xFEFF Arabic Presentation Forms-B |
| 116 | * |
| 117 | * @var string |
| 118 | */ |
| 119 | // @codingStandardsIgnoreStart |
| 120 | private static $rtlChars = '\x{0600}-\x{06ff}\x{0750}-\x{077f}\x{08a0}-\x{08ff}\x{0590}-\x{05ff}\x{fb50}-\x{fdff}\x{fe70}-\x{feff}'; // @codingStandardsIgnoreEnd |
| 121 | |
| 122 | # Expression to match at and hash sign characters: |
| 123 | private static $atSigns = '@＠'; |
| 124 | |
| 125 | private static $hashSigns = '#＃'; |
| 126 | |
| 127 | # cash tags |
| 128 | private static $cashSigns = '\$'; |
| 129 | |
| 130 | private static $cashtag = '[a-z]{1,6}(?:[._][a-z]{1,2})?'; |
| 131 | |
| 132 | # These URL validation pattern strings are based on the ABNF from RFC 3986 |
| 133 | private static $validateUrlUnreserved = '[a-z\p{Cyrillic}0-9\-._~]'; |
| 134 | |
| 135 | private static $validateUrlPctEncoded = '(?:%[0-9a-f]{2})'; |
| 136 | |
| 137 | private static $validateUrlSubDelims = '[!$&\'()*+,;=]'; |
| 138 | |
| 139 | private static $validUrlQueryChars = '[a-z0-9!?\*\'\(\);:&=\+\$\/%#\[\]\-_\.,~|@]'; |
| 140 | |
| 141 | private static $validUrlQueryEndingChars = '[a-z0-9_&=#\/\-]'; |
| 142 | |
| 143 | // @codingStandardsIgnoreStart |
| 144 | private static $validateUrlIpv4 = '(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])(?:\.(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])){3})'; // @codingStandardsIgnoreEnd |
| 145 | |
| 146 | private static $validateUrlIpv6 = '(?:\[[a-f0-9:\.]+\])'; |
| 147 | |
| 148 | private static $validateUrlPort = '[0-9]{1,5}'; |
| 149 | |
| 150 | # URL related hash regex collection |
| 151 | private static $validSpecialCcTLD = '(?:(?:co|tv)(?=[^0-9a-z@]|$))'; |
| 152 | |
| 153 | private static $validPunycode = '(?:xn--[0-9a-z]+)'; |
| 154 | |
| 155 | /** |
| 156 | * Get invalid characters matcher |
| 157 | * |
| 158 | * @staticvar string $regexp |
| 159 | * @return string |
| 160 | */ |
| 161 | public static function getInvalidCharactersMatcher() |
| 162 | { |
| 163 | static $regexp = null; |
| 164 | |
| 165 | if ($regexp === null) { |
| 166 | $regexp = '/[' . static::$invalidCharacters . ']/u'; |
| 167 | } |
| 168 | |
| 169 | return $regexp; |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * Get RTL characters matcher |
| 174 | * |
| 175 | * @staticvar string $regexp |
| 176 | * @return string |
| 177 | */ |
| 178 | public static function getRtlCharsMatcher() |
| 179 | { |
| 180 | static $regexp = null; |
| 181 | |
| 182 | if ($regexp === null) { |
| 183 | $regexp = '/[' . static::$rtlChars . ']/iu'; |
| 184 | } |
| 185 | |
| 186 | return $regexp; |
| 187 | } |
| 188 | |
| 189 | // ================================================================================================================= |
| 190 | |
| 191 | /** |
| 192 | * Get valid ascii domain matcher |
| 193 | * |
| 194 | * @staticvar string $regexp |
| 195 | * @return string |
| 196 | */ |
| 197 | public static function getValidAsciiDomainMatcher() |
| 198 | { |
| 199 | static $regexp = null; |
| 200 | |
| 201 | if ($regexp === null) { |
| 202 | $regexp = '/' . static::getValidSubdomain() . '*(' . static::getValidDomainName() |
| 203 | . ')(?:' . TldLists::getValidGTLD() . '|' . TldLists::getValidCcTLD() |
| 204 | . '|' . static::$validPunycode . ')/iu'; |
| 205 | } |
| 206 | |
| 207 | return $regexp; |
| 208 | } |
| 209 | |
| 210 | /** |
| 211 | * Get valid tco url matcher |
| 212 | * |
| 213 | * Used by the extractor for stricter t.co URL extraction |
| 214 | * |
| 215 | * @staticvar string $regexp |
| 216 | * @return string |
| 217 | */ |
| 218 | public static function getValidTcoUrlMatcher() |
| 219 | { |
| 220 | static $regexp = null; |
| 221 | |
| 222 | if ($regexp === null) { |
| 223 | $regexp = '/^https?:\/\/t\.co\/([a-z0-9]+)' |
| 224 | . '(?:\?' . static::$validUrlQueryChars . '*' . static::$validUrlQueryEndingChars . ')?/iu'; |
| 225 | } |
| 226 | |
| 227 | return $regexp; |
| 228 | } |
| 229 | |
| 230 | /** |
| 231 | * Get invalid short domain matcher |
| 232 | * |
| 233 | * @staticvar string $regexp |
| 234 | * @return string |
| 235 | */ |
| 236 | public static function getInvalidShortDomainMatcher() |
| 237 | { |
| 238 | static $regexp = null; |
| 239 | |
| 240 | if ($regexp === null) { |
| 241 | $regexp = '/\A' . static::getValidDomainName() . TldLists::getValidCcTLD() . '\Z/iu'; |
| 242 | } |
| 243 | |
| 244 | return $regexp; |
| 245 | } |
| 246 | |
| 247 | /** |
| 248 | * Get valid special short domain matcher |
| 249 | * |
| 250 | * @staticvar string $regexp |
| 251 | * @return string |
| 252 | */ |
| 253 | public static function getValidSpecialShortDomainMatcher() |
| 254 | { |
| 255 | static $regexp = null; |
| 256 | |
| 257 | if ($regexp === null) { |
| 258 | $regexp = '/\A' . static::getValidDomainName() . static::$validSpecialCcTLD . '\Z/iu'; |
| 259 | } |
| 260 | |
| 261 | return $regexp; |
| 262 | } |
| 263 | |
| 264 | /** |
| 265 | * Get invalid url without protocol preceding chars matcher |
| 266 | * |
| 267 | * @staticvar string $regexp |
| 268 | * @return string |
| 269 | */ |
| 270 | public static function getInvalidUrlWithoutProtocolPrecedingCharsMatcher() |
| 271 | { |
| 272 | static $regexp = null; |
| 273 | |
| 274 | if ($regexp === null) { |
| 275 | $regexp = '/[\-_.\/]\z/iu'; |
| 276 | } |
| 277 | |
| 278 | return $regexp; |
| 279 | } |
| 280 | |
| 281 | /** |
| 282 | * Get valid url |
| 283 | * |
| 284 | * @staticvar string $regexp |
| 285 | * @return string |
| 286 | */ |
| 287 | public static function getValidUrlMatcher() |
| 288 | { |
| 289 | static $regexp = null; |
| 290 | |
| 291 | if ($regexp === null) { |
| 292 | $validUrlPrecedingChars = '(?:[^a-z0-9_@＠\$#＃' . static::$invalidCharacters . ']|[' . static::$directionalCharacters . ']|^)'; |
| 293 | $validPortNumber = '[0-9]+'; |
| 294 | |
| 295 | $regexp = '/(?:' # $1 Complete match (preg_match() already matches everything.) |
| 296 | . '(' . $validUrlPrecedingChars . ')' # $2 Preceding characters |
| 297 | . '(' # $3 Complete URL |
| 298 | . '(https?:\/\/)?' # $4 Protocol (optional) |
| 299 | . '(' . static::getValidDomain() . ')' # $5 Domain(s) |
| 300 | . '(?::(' . $validPortNumber . '))?' # $6 Port number (optional) |
| 301 | . '(\/' . static::getValidUrlPath() . '*)?' # $7 URL Path |
| 302 | . '(\?' . static::$validUrlQueryChars . '*' . static::$validUrlQueryEndingChars . ')?' # $8 Query String |
| 303 | . ')' |
| 304 | . ')/iux'; |
| 305 | } |
| 306 | |
| 307 | return $regexp; |
| 308 | } |
| 309 | |
| 310 | /** |
| 311 | * Get valid domain chars |
| 312 | * |
| 313 | * @return string |
| 314 | */ |
| 315 | private static function getValidDomainChars() |
| 316 | { |
| 317 | return '0-9a-z' . static::$latinAccents; |
| 318 | } |
| 319 | |
| 320 | /** |
| 321 | * Get valid subdomain |
| 322 | * |
| 323 | * @return string |
| 324 | */ |
| 325 | private static function getValidSubdomain() |
| 326 | { |
| 327 | $domainValidChars = static::getValidDomainChars(); |
| 328 | |
| 329 | return '(?>(?:[' . $domainValidChars . '][' . $domainValidChars . '\-_]*)?[' . $domainValidChars . ']\.)'; |
| 330 | } |
| 331 | |
| 332 | /** |
| 333 | * Get valid domain name |
| 334 | * |
| 335 | * @return string |
| 336 | */ |
| 337 | private static function getValidDomainName() |
| 338 | { |
| 339 | $domainValidChars = static::getValidDomainChars(); |
| 340 | |
| 341 | return '(?:(?:[' . $domainValidChars . '][' . $domainValidChars . '\-]*)?[' . $domainValidChars . ']\.)'; |
| 342 | } |
| 343 | |
| 344 | /** |
| 345 | * Get valid unicode domain chars |
| 346 | * |
| 347 | * @return string |
| 348 | */ |
| 349 | private static function getValidUnicodeDomainChars() |
| 350 | { |
| 351 | return '[^\p{P}\p{Z}\p{C}' . static::$invalidCharacters . static::$spaces . ']'; |
| 352 | } |
| 353 | |
| 354 | /** |
| 355 | * Get valid unicode domain name |
| 356 | * |
| 357 | * @return string |
| 358 | */ |
| 359 | private static function getValidUnicodeDomainName() |
| 360 | { |
| 361 | $domainValidChars = static::getValidUnicodeDomainChars(); |
| 362 | |
| 363 | return '(?:(?:' . $domainValidChars . '(?:' . $domainValidChars . '|[\-])*)?' . $domainValidChars . '\.)'; |
| 364 | } |
| 365 | |
| 366 | /** |
| 367 | * Get valid domain |
| 368 | * |
| 369 | * @return string |
| 370 | */ |
| 371 | private static function getValidDomain() |
| 372 | { |
| 373 | $validSubdomain = static::getValidSubdomain(); |
| 374 | $validDomainName = static::getValidDomainName(); |
| 375 | $validUnicodeDomainName = static::getValidUnicodeDomainName(); |
| 376 | $validGTLD = TldLists::getValidGTLD(); |
| 377 | $validCcTLD = TldLists::getValidCcTLD(); |
| 378 | |
| 379 | return '' |
| 380 | // optional sub-domain + domain + TLD |
| 381 | // e.g. twitter.com, foo.co.jp, bar.co.uk |
| 382 | . '(?:' . $validSubdomain . '*' . $validDomainName |
| 383 | . '(?:' . $validGTLD . '|' . $validCcTLD . '|' . static::$validPunycode . '))' |
| 384 | // domain + gTLD | protocol + unicode domain + gTLD |
| 385 | . '|(?:' |
| 386 | . '(?:' . $validSubdomain . '+' . $validDomainName |
| 387 | . '|' . $validDomainName |
| 388 | . '|(?:(?<=http:\/\/|https:\/\/)' . $validUnicodeDomainName . ')' |
| 389 | . ')' |
| 390 | . $validGTLD |
| 391 | . ')' |
| 392 | // protocol + (domain | unicode domain) + ccTLD |
| 393 | . '|(?:(?<=http:\/\/|https:\/\/)' |
| 394 | . '(?:' . $validDomainName . '|' . $validUnicodeDomainName . ')' |
| 395 | . $validCcTLD . ')' |
| 396 | // domain + ccTLD + '/' |
| 397 | // e.g. t.co/ |
| 398 | . '|(?:' . $validDomainName . $validCcTLD . '(?=\/))'; |
| 399 | } |
| 400 | |
| 401 | /** |
| 402 | * Get valid url path |
| 403 | * |
| 404 | * @return string |
| 405 | */ |
| 406 | private static function getValidUrlPath() |
| 407 | { |
| 408 | $validGeneralUrlPathChars = '[a-z0-9' . preg_quote("!*';:=+,.$/%#[]–\x{2013}_~", '/') |
| 409 | . '|&@' . static::$latinAccents . '\p{Cyrillic}]'; |
| 410 | |
| 411 | # Allow URL paths to contain up to two nested levels of balanced parentheses: |
| 412 | # 1. Used in Wikipedia URLs, e.g. /Primer_(film) |
| 413 | # 2. Used in IIS sessions, e.g. /S(dfd346)/ |
| 414 | # 3. Used in Rdio URLs like /track/We_Up_(Album_Version_(Edited))/ |
| 415 | $validUrlBalancedParens = '(?:\(' |
| 416 | . '(?:' . $validGeneralUrlPathChars . '+' |
| 417 | . '|' |
| 418 | // allow one nested level of balanced parentheses |
| 419 | . '(?:' |
| 420 | . $validGeneralUrlPathChars . '*' |
| 421 | . '\(' . $validGeneralUrlPathChars . '+' . '\)' |
| 422 | . $validGeneralUrlPathChars . '*' |
| 423 | . ')' |
| 424 | . ')' |
| 425 | . '\))'; |
| 426 | # Valid end-of-path characters (so /foo. does not gobble the period). |
| 427 | # 1. Allow =&# for empty URL parameters and other URL-join artifacts. |
| 428 | $validUrlPathEndingChars = '[a-z0-9=_#\/\+\-' . static::$latinAccents . '\p{Cyrillic}]' |
| 429 | . '|(?:' . $validUrlBalancedParens . ')'; |
| 430 | |
| 431 | return '(?:(?:' |
| 432 | . $validGeneralUrlPathChars . '*(?:' |
| 433 | . $validUrlBalancedParens . ' ' |
| 434 | . $validGeneralUrlPathChars . '*)*' |
| 435 | . $validUrlPathEndingChars . ')|(?:@' |
| 436 | . $validGeneralUrlPathChars . '+\/))'; |
| 437 | } |
| 438 | |
| 439 | // ================================================================================================================= |
| 440 | |
| 441 | # NOTE: PHP doesn't have Ruby's $' (dollar apostrophe) so we have to capture |
| 442 | # $after in the following regular expression. Note that we only use a |
| 443 | # look-ahead capture here and don't append $after when we return. |
| 444 | |
| 445 | /** |
| 446 | * Get valid mentions or lists matcher |
| 447 | * |
| 448 | * @staticvar string $regexp |
| 449 | * @return string |
| 450 | */ |
| 451 | public static function getValidMentionsOrListsMatcher() |
| 452 | { |
| 453 | static $regexp = null; |
| 454 | |
| 455 | if ($regexp === null) { |
| 456 | $mentionPrecedingChars = '([^a-z0-9_!#\$%&*@＠\/]|^|(?:^|[^a-z0-9_+~.-])RT:?)'; |
| 457 | $regexp = '/' . $mentionPrecedingChars |
| 458 | . '([' . static::$atSigns . '])([a-z0-9_]{1,20})(\/[a-z][a-z0-9_\-]{0,24})?(?=(.*|$))/iu'; |
| 459 | } |
| 460 | |
| 461 | return $regexp; |
| 462 | } |
| 463 | |
| 464 | /** |
| 465 | * Get valid hashtag matcher |
| 466 | * |
| 467 | * @staticvar string $regexp |
| 468 | * @return string |
| 469 | */ |
| 470 | public static function getValidReplyMatcher() |
| 471 | { |
| 472 | static $regexp = null; |
| 473 | |
| 474 | if ($regexp === null) { |
| 475 | $regexp = '/^(?:[' . static::$spaces . static::$directionalCharacters . '])*[' . static::$atSigns . ']([a-z0-9_]{1,20})(?=(.*|$))/iu'; |
| 476 | } |
| 477 | |
| 478 | return $regexp; |
| 479 | } |
| 480 | |
| 481 | /** |
| 482 | * Get end of hashtag matcher |
| 483 | * |
| 484 | * @staticvar string $regexp |
| 485 | * @return string |
| 486 | */ |
| 487 | public static function getEndMentionMatcher() |
| 488 | { |
| 489 | static $regexp = null; |
| 490 | |
| 491 | if ($regexp === null) { |
| 492 | $regexp = '/\A(?:[' . static::$atSigns . ']|[' . static::$latinAccents . ']|:\/\/)/iu'; |
| 493 | } |
| 494 | |
| 495 | return $regexp; |
| 496 | } |
| 497 | |
| 498 | // ================================================================================================================= |
| 499 | |
| 500 | /** |
| 501 | * Get hashtag matcher |
| 502 | * |
| 503 | * @return string matcher |
| 504 | */ |
| 505 | private static function getHashtagPattern() |
| 506 | { |
| 507 | $hashtag_letters = '\p{L}\p{M}'; |
| 508 | $hashtag_numerals = '\p{Nd}'; |
| 509 | # Hashtag special chars |
| 510 | # |
| 511 | # _ underscore |
| 512 | # 0x200c ZERO WIDTH NON-JOINER (ZWNJ) |
| 513 | # 0x200d ZERO WIDTH JOINER (ZWJ) |
| 514 | # 0xa67e CYRILLIC KAVYKA |
| 515 | # 0x05be HEBREW PUNCTUATION MAQAF |
| 516 | # 0x05f3 HEBREW PUNCTUATION GERESH |
| 517 | # 0x05f4 HEBREW PUNCTUATION GERSHAYIM |
| 518 | # 0xff5e FULLWIDTH TILDE |
| 519 | # 0x301c WAVE DASH |
| 520 | # 0x309b KATAKANA-HIRAGANA VOICED SOUND MARK |
| 521 | # 0x309c KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK |
| 522 | # 0x30a0 KATAKANA-HIRAGANA DOUBLE HYPHEN |
| 523 | # 0x30fb KATAKANA MIDDLE DOT |
| 524 | # 0x3003 DITTO MARK |
| 525 | # 0x0f0b TIBETAN MARK INTERSYLLABIC TSHEG |
| 526 | # 0x0f0c TIBETAN MARK DELIMITER TSHEG BSTAR |
| 527 | # 0x00b7 MIDDLE DOT |
| 528 | $hashtag_special_chars = '_\x{200c}\x{200d}\x{a67e}\x{05be}\x{05f3}\x{05f4}' |
| 529 | . '\x{ff5e}\x{301c}\x{309b}\x{309c}\x{30a0}\x{30fb}\x{3003}\x{0f0b}\x{0f0c}\x{00b7}'; |
| 530 | $hashtag_letters_numerals_set = '[' . $hashtag_letters . $hashtag_numerals . $hashtag_special_chars . ']'; |
| 531 | $hashtag_letters_set = '[' . $hashtag_letters . ']'; |
| 532 | $hashtag_boundary = '(?:\A|\x{fe0e}|\x{fe0f}|[^&' |
| 533 | . $hashtag_letters . $hashtag_numerals . $hashtag_special_chars . '])'; |
| 534 | |
| 535 | return '(' . $hashtag_boundary . ')(#|\x{ff03})(?!\x{fe0f}|\x{20e3})(' |
| 536 | . $hashtag_letters_numerals_set . '*' . $hashtag_letters_set . $hashtag_letters_numerals_set . '*)'; |
| 537 | } |
| 538 | |
| 539 | /** |
| 540 | * Get valid hashtag matcher |
| 541 | * |
| 542 | * @staticvar string $regexp |
| 543 | * @return string |
| 544 | */ |
| 545 | public static function getValidHashtagMatcher() |
| 546 | { |
| 547 | static $regexp = null; |
| 548 | |
| 549 | if ($regexp === null) { |
| 550 | $regexp = '/' . static::getHashtagPattern() . '(?=(.*|$))/iu'; |
| 551 | } |
| 552 | |
| 553 | return $regexp; |
| 554 | } |
| 555 | |
| 556 | /** |
| 557 | * Get end of hashtag matcher |
| 558 | * |
| 559 | * @staticvar string $regexp |
| 560 | * @return string |
| 561 | */ |
| 562 | public static function getEndHashtagMatcher() |
| 563 | { |
| 564 | static $regexp = null; |
| 565 | |
| 566 | if ($regexp === null) { |
| 567 | $regexp = '/\A(?:[' . static::$hashSigns . ']|:\/\/)/u'; |
| 568 | } |
| 569 | |
| 570 | return $regexp; |
| 571 | } |
| 572 | |
| 573 | // ================================================================================================================= |
| 574 | |
| 575 | /** |
| 576 | * Get valid cachtag matcher |
| 577 | * |
| 578 | * @staticvar string $regexp |
| 579 | * @return string |
| 580 | */ |
| 581 | public static function getValidCashtagMatcher() |
| 582 | { |
| 583 | static $regexp = null; |
| 584 | |
| 585 | if ($regexp === null) { |
| 586 | $regexp = '/(^|[' . static::$spaces . static::$directionalCharacters . '])([' . static::$cashSigns . '])' |
| 587 | . '(' . static::$cashtag . ')(?=($|\s|[[:punct:]]))/iu'; |
| 588 | } |
| 589 | |
| 590 | return $regexp; |
| 591 | } |
| 592 | |
| 593 | /** |
| 594 | * Get end of cachtag matcher |
| 595 | * |
| 596 | * @staticvar string $regexp |
| 597 | * @return string |
| 598 | */ |
| 599 | public static function getEndCashtagMatcher() |
| 600 | { |
| 601 | static $regexp = null; |
| 602 | |
| 603 | if ($regexp === null) { |
| 604 | $regexp = '/\A(?:[' . static::$cashSigns . ']|:\/\/)/u'; |
| 605 | } |
| 606 | |
| 607 | return $regexp; |
| 608 | } |
| 609 | |
| 610 | // ================================================================================================================= |
| 611 | |
| 612 | /** |
| 613 | * Get url matcher |
| 614 | * |
| 615 | * @staticvar string $regexp |
| 616 | * @return string |
| 617 | */ |
| 618 | public static function getValidateUrlUnencodedMatcher() |
| 619 | { |
| 620 | static $regexp = null; |
| 621 | |
| 622 | if ($regexp === null) { |
| 623 | # Modified version of RFC 3986 Appendix B |
| 624 | $regexp = '/\A' # Full URL |
| 625 | . '(?:' |
| 626 | . '([^:\/?#]+):\/\/' # $1 Scheme |
| 627 | . ')?' |
| 628 | . '([^\/?#]*)' # $2 Authority |
| 629 | . '([^?#]*)' # $3 Path |
| 630 | . '(?:' |
| 631 | . '\?([^#]*)' # $4 Query |
| 632 | . ')?' |
| 633 | . '(?:' |
| 634 | . '\#(.*)' # $5 Fragment |
| 635 | . ')?\z/iux'; |
| 636 | } |
| 637 | |
| 638 | return $regexp; |
| 639 | } |
| 640 | |
| 641 | /** |
| 642 | * Get valid url ip |
| 643 | * |
| 644 | * @return string matcher |
| 645 | */ |
| 646 | private static function getValidateUrlIp() |
| 647 | { |
| 648 | return '(?:' . static::$validateUrlIpv4 . '|' . static::$validateUrlIpv6 . ')'; #/iox |
| 649 | } |
| 650 | |
| 651 | /** |
| 652 | * Get valid url domain |
| 653 | * |
| 654 | * @return string matcher |
| 655 | */ |
| 656 | private static function getValidateUrlDomain() |
| 657 | { |
| 658 | $subdomain = '(?:[a-z0-9](?:[a-z0-9_\-]*[a-z0-9])?)'; #/i |
| 659 | $domain = '(?:[a-z0-9](?:[a-z0-9\-]*[a-z0-9])?)'; #/i |
| 660 | $tld = '(?:[a-z](?:[a-z0-9\-]*[a-z0-9])?)'; #/i |
| 661 | |
| 662 | return '(?:(?:' . $subdomain . '\.)*(?:' . $domain . '\.)' . $tld . ')'; #/iox |
| 663 | } |
| 664 | |
| 665 | /** |
| 666 | * Get valid url host |
| 667 | * |
| 668 | * @return string matcher |
| 669 | */ |
| 670 | private static function getValidateUrlHost() |
| 671 | { |
| 672 | return '(?:' . static::getValidateUrlIp() . '|' . static::getValidateUrlDomain() . ')'; #/iox |
| 673 | } |
| 674 | |
| 675 | /** |
| 676 | * Get valid url unicode domain |
| 677 | * |
| 678 | * @return string matcher |
| 679 | */ |
| 680 | private static function getValidateUrlUnicodeDomain() |
| 681 | { |
| 682 | $subdomain = '(?:(?:[a-z0-9]|[^\x00-\x7f])(?:(?:[a-z0-9_\-]|[^\x00-\x7f])*(?:[a-z0-9]|[^\x00-\x7f]))?)'; #/ix |
| 683 | $domain = '(?:(?:[a-z0-9]|[^\x00-\x7f])(?:(?:[a-z0-9\-]|[^\x00-\x7f])*(?:[a-z0-9]|[^\x00-\x7f]))?)'; #/ix |
| 684 | $tld = '(?:(?:[a-z]|[^\x00-\x7f])(?:(?:[a-z0-9\-]|[^\x00-\x7f])*(?:[a-z0-9]|[^\x00-\x7f]))?)'; #/ix |
| 685 | |
| 686 | return '(?:(?:' . $subdomain . '\.)*(?:' . $domain . '\.)' . $tld . ')'; #/iox |
| 687 | } |
| 688 | |
| 689 | /** |
| 690 | * Get valid url unicode host |
| 691 | * |
| 692 | * @return string matcher |
| 693 | */ |
| 694 | private static function getValidateUrlUnicodeHost() |
| 695 | { |
| 696 | return '(?:' . static::getValidateUrlIp() . '|' . static::getValidateUrlUnicodeDomain() . ')'; #/iox |
| 697 | } |
| 698 | |
| 699 | /** |
| 700 | * Get valid url userinfo |
| 701 | * |
| 702 | * @return string matcher |
| 703 | */ |
| 704 | private static function getValidateUrlUserinfo() |
| 705 | { |
| 706 | return '(?:' . static::$validateUrlUnreserved |
| 707 | . '|' . static::$validateUrlPctEncoded |
| 708 | . '|' . static::$validateUrlSubDelims |
| 709 | . '|:)*'; #/iox |
| 710 | } |
| 711 | |
| 712 | /** |
| 713 | * Get url unicode authority matcher |
| 714 | * |
| 715 | * Unencoded internationalized domains - this doesn't check for invalid UTF-8 sequences |
| 716 | * |
| 717 | * @staticvar string $regexp |
| 718 | * @return string |
| 719 | */ |
| 720 | public static function getValidateUrlUnicodeAuthorityMatcher() |
| 721 | { |
| 722 | static $regexp = null; |
| 723 | |
| 724 | if ($regexp === null) { |
| 725 | $regexp = '/' |
| 726 | . '(?:(' . static::getValidateUrlUserinfo() . ')@)?' # $1 userinfo |
| 727 | . '(' . static::getValidateUrlUnicodeHost() . ')' # $2 host |
| 728 | . '(?::(' . static::$validateUrlPort . '))?' # $3 port |
| 729 | . '/iux'; |
| 730 | } |
| 731 | |
| 732 | return $regexp; |
| 733 | } |
| 734 | |
| 735 | /** |
| 736 | * Get url authority matcher |
| 737 | * |
| 738 | * This is more strict than the rfc specifies |
| 739 | * |
| 740 | * @staticvar string $regexp |
| 741 | * @return string |
| 742 | */ |
| 743 | public static function getValidateUrlAuthorityMatcher() |
| 744 | { |
| 745 | static $regexp = null; |
| 746 | |
| 747 | if ($regexp === null) { |
| 748 | $regexp = '/' |
| 749 | . '(?:(' . static::getValidateUrlUserinfo() . ')@)?' # $1 userinfo |
| 750 | . '(' . static::getValidateUrlHost() . ')' # $2 host |
| 751 | . '(?::(' . static::$validateUrlPort . '))?' # $3 port |
| 752 | . '/ix'; |
| 753 | } |
| 754 | |
| 755 | return $regexp; |
| 756 | } |
| 757 | |
| 758 | /** |
| 759 | * Get url scheme matcher |
| 760 | * |
| 761 | * @staticvar string $regexp |
| 762 | * @return string |
| 763 | */ |
| 764 | public static function getValidateUrlSchemeMatcher() |
| 765 | { |
| 766 | static $regexp = null; |
| 767 | |
| 768 | if ($regexp === null) { |
| 769 | $regexp = '/(?:[a-z][a-z0-9+\-.]*)/i'; |
| 770 | } |
| 771 | |
| 772 | return $regexp; |
| 773 | } |
| 774 | |
| 775 | /** |
| 776 | * Get valid url charactors |
| 777 | * |
| 778 | * @return string matcher |
| 779 | */ |
| 780 | private static function getValidateUrlPchar() |
| 781 | { |
| 782 | return '(?:' . static::$validateUrlUnreserved |
| 783 | . '|' . static::$validateUrlPctEncoded |
| 784 | . '|' . static::$validateUrlSubDelims |
| 785 | . '|[:\|@])'; #/iox |
| 786 | } |
| 787 | |
| 788 | /** |
| 789 | * Get url path matcher |
| 790 | * |
| 791 | * @staticvar string $regexp |
| 792 | * @return string |
| 793 | */ |
| 794 | public static function getValidateUrlPathMatcher() |
| 795 | { |
| 796 | static $regexp = null; |
| 797 | |
| 798 | if ($regexp === null) { |
| 799 | $regexp = '/(\/' . static::getValidateUrlPchar() . '*)*/iu'; |
| 800 | } |
| 801 | |
| 802 | return $regexp; |
| 803 | } |
| 804 | |
| 805 | /** |
| 806 | * Get url query matcher |
| 807 | * |
| 808 | * @staticvar string $regexp |
| 809 | * @return string |
| 810 | */ |
| 811 | public static function getValidateUrlQueryMatcher() |
| 812 | { |
| 813 | static $regexp = null; |
| 814 | |
| 815 | if ($regexp === null) { |
| 816 | $regexp = '/(' . static::getValidateUrlPchar() . '|\/|\?)*/iu'; |
| 817 | } |
| 818 | |
| 819 | return $regexp; |
| 820 | } |
| 821 | |
| 822 | /** |
| 823 | * Get url flagment matcher |
| 824 | * |
| 825 | * @staticvar string $regexp |
| 826 | * @return string |
| 827 | */ |
| 828 | public static function getValidateUrlFragmentMatcher() |
| 829 | { |
| 830 | static $regexp = null; |
| 831 | |
| 832 | if ($regexp === null) { |
| 833 | $regexp = '/(' . static::getValidateUrlPchar() . '|\/|\?)*/iu'; |
| 834 | } |
| 835 | |
| 836 | return $regexp; |
| 837 | } |
| 838 | } |
| 839 |