Rev 1308 | Details | Compare with Previous | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
597 | daniel-mar | 1 | <?php |
1042 | daniel-mar | 2 | |
597 | daniel-mar | 3 | /** |
1042 | daniel-mar | 4 | * JavaScript minifier. |
597 | daniel-mar | 5 | * |
6 | * Please report bugs on https://github.com/matthiasmullie/minify/issues |
||
7 | * |
||
8 | * @author Matthias Mullie <minify@mullie.eu> |
||
9 | * @copyright Copyright (c) 2012, Matthias Mullie. All rights reserved |
||
10 | * @license MIT License |
||
11 | */ |
||
1042 | daniel-mar | 12 | |
597 | daniel-mar | 13 | namespace MatthiasMullie\Minify; |
14 | |||
15 | /** |
||
1042 | daniel-mar | 16 | * JavaScript Minifier Class. |
597 | daniel-mar | 17 | * |
18 | * Please report bugs on https://github.com/matthiasmullie/minify/issues |
||
19 | * |
||
20 | * @author Matthias Mullie <minify@mullie.eu> |
||
21 | * @author Tijs Verkoyen <minify@verkoyen.eu> |
||
22 | * @copyright Copyright (c) 2012, Matthias Mullie. All rights reserved |
||
23 | * @license MIT License |
||
24 | */ |
||
25 | class JS extends Minify |
||
26 | { |
||
27 | /** |
||
28 | * Var-matching regex based on http://stackoverflow.com/a/9337047/802993. |
||
29 | * |
||
30 | * Note that regular expressions using that bit must have the PCRE_UTF8 |
||
31 | * pattern modifier (/u) set. |
||
32 | * |
||
1042 | daniel-mar | 33 | * @internal |
34 | * |
||
597 | daniel-mar | 35 | * @var string |
36 | */ |
||
37 | const REGEX_VARIABLE = '\b[$A-Z\_a-z\xaa\xb5\xba\xc0-\xd6\xd8-\xf6\xf8-\x{02c1}\x{02c6}-\x{02d1}\x{02e0}-\x{02e4}\x{02ec}\x{02ee}\x{0370}-\x{0374}\x{0376}\x{0377}\x{037a}-\x{037d}\x{0386}\x{0388}-\x{038a}\x{038c}\x{038e}-\x{03a1}\x{03a3}-\x{03f5}\x{03f7}-\x{0481}\x{048a}-\x{0527}\x{0531}-\x{0556}\x{0559}\x{0561}-\x{0587}\x{05d0}-\x{05ea}\x{05f0}-\x{05f2}\x{0620}-\x{064a}\x{066e}\x{066f}\x{0671}-\x{06d3}\x{06d5}\x{06e5}\x{06e6}\x{06ee}\x{06ef}\x{06fa}-\x{06fc}\x{06ff}\x{0710}\x{0712}-\x{072f}\x{074d}-\x{07a5}\x{07b1}\x{07ca}-\x{07ea}\x{07f4}\x{07f5}\x{07fa}\x{0800}-\x{0815}\x{081a}\x{0824}\x{0828}\x{0840}-\x{0858}\x{08a0}\x{08a2}-\x{08ac}\x{0904}-\x{0939}\x{093d}\x{0950}\x{0958}-\x{0961}\x{0971}-\x{0977}\x{0979}-\x{097f}\x{0985}-\x{098c}\x{098f}\x{0990}\x{0993}-\x{09a8}\x{09aa}-\x{09b0}\x{09b2}\x{09b6}-\x{09b9}\x{09bd}\x{09ce}\x{09dc}\x{09dd}\x{09df}-\x{09e1}\x{09f0}\x{09f1}\x{0a05}-\x{0a0a}\x{0a0f}\x{0a10}\x{0a13}-\x{0a28}\x{0a2a}-\x{0a30}\x{0a32}\x{0a33}\x{0a35}\x{0a36}\x{0a38}\x{0a39}\x{0a59}-\x{0a5c}\x{0a5e}\x{0a72}-\x{0a74}\x{0a85}-\x{0a8d}\x{0a8f}-\x{0a91}\x{0a93}-\x{0aa8}\x{0aaa}-\x{0ab0}\x{0ab2}\x{0ab3}\x{0ab5}-\x{0ab9}\x{0abd}\x{0ad0}\x{0ae0}\x{0ae1}\x{0b05}-\x{0b0c}\x{0b0f}\x{0b10}\x{0b13}-\x{0b28}\x{0b2a}-\x{0b30}\x{0b32}\x{0b33}\x{0b35}-\x{0b39}\x{0b3d}\x{0b5c}\x{0b5d}\x{0b5f}-\x{0b61}\x{0b71}\x{0b83}\x{0b85}-\x{0b8a}\x{0b8e}-\x{0b90}\x{0b92}-\x{0b95}\x{0b99}\x{0b9a}\x{0b9c}\x{0b9e}\x{0b9f}\x{0ba3}\x{0ba4}\x{0ba8}-\x{0baa}\x{0bae}-\x{0bb9}\x{0bd0}\x{0c05}-\x{0c0c}\x{0c0e}-\x{0c10}\x{0c12}-\x{0c28}\x{0c2a}-\x{0c33}\x{0c35}-\x{0c39}\x{0c3d}\x{0c58}\x{0c59}\x{0c60}\x{0c61}\x{0c85}-\x{0c8c}\x{0c8e}-\x{0c90}\x{0c92}-\x{0ca8}\x{0caa}-\x{0cb3}\x{0cb5}-\x{0cb9}\x{0cbd}\x{0cde}\x{0ce0}\x{0ce1}\x{0cf1}\x{0cf2}\x{0d05}-\x{0d0c}\x{0d0e}-\x{0d10}\x{0d12}-\x{0d3a}\x{0d3d}\x{0d4e}\x{0d60}\x{0d61}\x{0d7a}-\x{0d7f}\x{0d85}-\x{0d96}\x{0d9a}-\x{0db1}\x{0db3}-\x{0dbb}\x{0dbd}\x{0dc0}-\x{0dc6}\x{0e01}-\x{0e30}\x{0e32}\x{0e33}\x{0e40}-\x{0e46}\x{0e81}\x{0e82}\x{0e84}\x{0e87}\x{0e88}\x{0e8a}\x{0e8d}\x{0e94}-\x{0e97}\x{0e99}-\x{0e9f}\x{0ea1}-\x{0ea3}\x{0ea5}\x{0ea7}\x{0eaa}\x{0eab}\x{0ead}-\x{0eb0}\x{0eb2}\x{0eb3}\x{0ebd}\x{0ec0}-\x{0ec4}\x{0ec6}\x{0edc}-\x{0edf}\x{0f00}\x{0f40}-\x{0f47}\x{0f49}-\x{0f6c}\x{0f88}-\x{0f8c}\x{1000}-\x{102a}\x{103f}\x{1050}-\x{1055}\x{105a}-\x{105d}\x{1061}\x{1065}\x{1066}\x{106e}-\x{1070}\x{1075}-\x{1081}\x{108e}\x{10a0}-\x{10c5}\x{10c7}\x{10cd}\x{10d0}-\x{10fa}\x{10fc}-\x{1248}\x{124a}-\x{124d}\x{1250}-\x{1256}\x{1258}\x{125a}-\x{125d}\x{1260}-\x{1288}\x{128a}-\x{128d}\x{1290}-\x{12b0}\x{12b2}-\x{12b5}\x{12b8}-\x{12be}\x{12c0}\x{12c2}-\x{12c5}\x{12c8}-\x{12d6}\x{12d8}-\x{1310}\x{1312}-\x{1315}\x{1318}-\x{135a}\x{1380}-\x{138f}\x{13a0}-\x{13f4}\x{1401}-\x{166c}\x{166f}-\x{167f}\x{1681}-\x{169a}\x{16a0}-\x{16ea}\x{16ee}-\x{16f0}\x{1700}-\x{170c}\x{170e}-\x{1711}\x{1720}-\x{1731}\x{1740}-\x{1751}\x{1760}-\x{176c}\x{176e}-\x{1770}\x{1780}-\x{17b3}\x{17d7}\x{17dc}\x{1820}-\x{1877}\x{1880}-\x{18a8}\x{18aa}\x{18b0}-\x{18f5}\x{1900}-\x{191c}\x{1950}-\x{196d}\x{1970}-\x{1974}\x{1980}-\x{19ab}\x{19c1}-\x{19c7}\x{1a00}-\x{1a16}\x{1a20}-\x{1a54}\x{1aa7}\x{1b05}-\x{1b33}\x{1b45}-\x{1b4b}\x{1b83}-\x{1ba0}\x{1bae}\x{1baf}\x{1bba}-\x{1be5}\x{1c00}-\x{1c23}\x{1c4d}-\x{1c4f}\x{1c5a}-\x{1c7d}\x{1ce9}-\x{1cec}\x{1cee}-\x{1cf1}\x{1cf5}\x{1cf6}\x{1d00}-\x{1dbf}\x{1e00}-\x{1f15}\x{1f18}-\x{1f1d}\x{1f20}-\x{1f45}\x{1f48}-\x{1f4d}\x{1f50}-\x{1f57}\x{1f59}\x{1f5b}\x{1f5d}\x{1f5f}-\x{1f7d}\x{1f80}-\x{1fb4}\x{1fb6}-\x{1fbc}\x{1fbe}\x{1fc2}-\x{1fc4}\x{1fc6}-\x{1fcc}\x{1fd0}-\x{1fd3}\x{1fd6}-\x{1fdb}\x{1fe0}-\x{1fec}\x{1ff2}-\x{1ff4}\x{1ff6}-\x{1ffc}\x{2071}\x{207f}\x{2090}-\x{209c}\x{2102}\x{2107}\x{210a}-\x{2113}\x{2115}\x{2119}-\x{211d}\x{2124}\x{2126}\x{2128}\x{212a}-\x{212d}\x{212f}-\x{2139}\x{213c}-\x{213f}\x{2145}-\x{2149}\x{214e}\x{2160}-\x{2188}\x{2c00}-\x{2c2e}\x{2c30}-\x{2c5e}\x{2c60}-\x{2ce4}\x{2ceb}-\x{2cee}\x{2cf2}\x{2cf3}\x{2d00}-\x{2d25}\x{2d27}\x{2d2d}\x{2d30}-\x{2d67}\x{2d6f}\x{2d80}-\x{2d96}\x{2da0}-\x{2da6}\x{2da8}-\x{2dae}\x{2db0}-\x{2db6}\x{2db8}-\x{2dbe}\x{2dc0}-\x{2dc6}\x{2dc8}-\x{2dce}\x{2dd0}-\x{2dd6}\x{2dd8}-\x{2dde}\x{2e2f}\x{3005}-\x{3007}\x{3021}-\x{3029}\x{3031}-\x{3035}\x{3038}-\x{303c}\x{3041}-\x{3096}\x{309d}-\x{309f}\x{30a1}-\x{30fa}\x{30fc}-\x{30ff}\x{3105}-\x{312d}\x{3131}-\x{318e}\x{31a0}-\x{31ba}\x{31f0}-\x{31ff}\x{3400}-\x{4db5}\x{4e00}-\x{9fcc}\x{a000}-\x{a48c}\x{a4d0}-\x{a4fd}\x{a500}-\x{a60c}\x{a610}-\x{a61f}\x{a62a}\x{a62b}\x{a640}-\x{a66e}\x{a67f}-\x{a697}\x{a6a0}-\x{a6ef}\x{a717}-\x{a71f}\x{a722}-\x{a788}\x{a78b}-\x{a78e}\x{a790}-\x{a793}\x{a7a0}-\x{a7aa}\x{a7f8}-\x{a801}\x{a803}-\x{a805}\x{a807}-\x{a80a}\x{a80c}-\x{a822}\x{a840}-\x{a873}\x{a882}-\x{a8b3}\x{a8f2}-\x{a8f7}\x{a8fb}\x{a90a}-\x{a925}\x{a930}-\x{a946}\x{a960}-\x{a97c}\x{a984}-\x{a9b2}\x{a9cf}\x{aa00}-\x{aa28}\x{aa40}-\x{aa42}\x{aa44}-\x{aa4b}\x{aa60}-\x{aa76}\x{aa7a}\x{aa80}-\x{aaaf}\x{aab1}\x{aab5}\x{aab6}\x{aab9}-\x{aabd}\x{aac0}\x{aac2}\x{aadb}-\x{aadd}\x{aae0}-\x{aaea}\x{aaf2}-\x{aaf4}\x{ab01}-\x{ab06}\x{ab09}-\x{ab0e}\x{ab11}-\x{ab16}\x{ab20}-\x{ab26}\x{ab28}-\x{ab2e}\x{abc0}-\x{abe2}\x{ac00}-\x{d7a3}\x{d7b0}-\x{d7c6}\x{d7cb}-\x{d7fb}\x{f900}-\x{fa6d}\x{fa70}-\x{fad9}\x{fb00}-\x{fb06}\x{fb13}-\x{fb17}\x{fb1d}\x{fb1f}-\x{fb28}\x{fb2a}-\x{fb36}\x{fb38}-\x{fb3c}\x{fb3e}\x{fb40}\x{fb41}\x{fb43}\x{fb44}\x{fb46}-\x{fbb1}\x{fbd3}-\x{fd3d}\x{fd50}-\x{fd8f}\x{fd92}-\x{fdc7}\x{fdf0}-\x{fdfb}\x{fe70}-\x{fe74}\x{fe76}-\x{fefc}\x{ff21}-\x{ff3a}\x{ff41}-\x{ff5a}\x{ff66}-\x{ffbe}\x{ffc2}-\x{ffc7}\x{ffca}-\x{ffcf}\x{ffd2}-\x{ffd7}\x{ffda}-\x{ffdc}][$A-Z\_a-z\xaa\xb5\xba\xc0-\xd6\xd8-\xf6\xf8-\x{02c1}\x{02c6}-\x{02d1}\x{02e0}-\x{02e4}\x{02ec}\x{02ee}\x{0370}-\x{0374}\x{0376}\x{0377}\x{037a}-\x{037d}\x{0386}\x{0388}-\x{038a}\x{038c}\x{038e}-\x{03a1}\x{03a3}-\x{03f5}\x{03f7}-\x{0481}\x{048a}-\x{0527}\x{0531}-\x{0556}\x{0559}\x{0561}-\x{0587}\x{05d0}-\x{05ea}\x{05f0}-\x{05f2}\x{0620}-\x{064a}\x{066e}\x{066f}\x{0671}-\x{06d3}\x{06d5}\x{06e5}\x{06e6}\x{06ee}\x{06ef}\x{06fa}-\x{06fc}\x{06ff}\x{0710}\x{0712}-\x{072f}\x{074d}-\x{07a5}\x{07b1}\x{07ca}-\x{07ea}\x{07f4}\x{07f5}\x{07fa}\x{0800}-\x{0815}\x{081a}\x{0824}\x{0828}\x{0840}-\x{0858}\x{08a0}\x{08a2}-\x{08ac}\x{0904}-\x{0939}\x{093d}\x{0950}\x{0958}-\x{0961}\x{0971}-\x{0977}\x{0979}-\x{097f}\x{0985}-\x{098c}\x{098f}\x{0990}\x{0993}-\x{09a8}\x{09aa}-\x{09b0}\x{09b2}\x{09b6}-\x{09b9}\x{09bd}\x{09ce}\x{09dc}\x{09dd}\x{09df}-\x{09e1}\x{09f0}\x{09f1}\x{0a05}-\x{0a0a}\x{0a0f}\x{0a10}\x{0a13}-\x{0a28}\x{0a2a}-\x{0a30}\x{0a32}\x{0a33}\x{0a35}\x{0a36}\x{0a38}\x{0a39}\x{0a59}-\x{0a5c}\x{0a5e}\x{0a72}-\x{0a74}\x{0a85}-\x{0a8d}\x{0a8f}-\x{0a91}\x{0a93}-\x{0aa8}\x{0aaa}-\x{0ab0}\x{0ab2}\x{0ab3}\x{0ab5}-\x{0ab9}\x{0abd}\x{0ad0}\x{0ae0}\x{0ae1}\x{0b05}-\x{0b0c}\x{0b0f}\x{0b10}\x{0b13}-\x{0b28}\x{0b2a}-\x{0b30}\x{0b32}\x{0b33}\x{0b35}-\x{0b39}\x{0b3d}\x{0b5c}\x{0b5d}\x{0b5f}-\x{0b61}\x{0b71}\x{0b83}\x{0b85}-\x{0b8a}\x{0b8e}-\x{0b90}\x{0b92}-\x{0b95}\x{0b99}\x{0b9a}\x{0b9c}\x{0b9e}\x{0b9f}\x{0ba3}\x{0ba4}\x{0ba8}-\x{0baa}\x{0bae}-\x{0bb9}\x{0bd0}\x{0c05}-\x{0c0c}\x{0c0e}-\x{0c10}\x{0c12}-\x{0c28}\x{0c2a}-\x{0c33}\x{0c35}-\x{0c39}\x{0c3d}\x{0c58}\x{0c59}\x{0c60}\x{0c61}\x{0c85}-\x{0c8c}\x{0c8e}-\x{0c90}\x{0c92}-\x{0ca8}\x{0caa}-\x{0cb3}\x{0cb5}-\x{0cb9}\x{0cbd}\x{0cde}\x{0ce0}\x{0ce1}\x{0cf1}\x{0cf2}\x{0d05}-\x{0d0c}\x{0d0e}-\x{0d10}\x{0d12}-\x{0d3a}\x{0d3d}\x{0d4e}\x{0d60}\x{0d61}\x{0d7a}-\x{0d7f}\x{0d85}-\x{0d96}\x{0d9a}-\x{0db1}\x{0db3}-\x{0dbb}\x{0dbd}\x{0dc0}-\x{0dc6}\x{0e01}-\x{0e30}\x{0e32}\x{0e33}\x{0e40}-\x{0e46}\x{0e81}\x{0e82}\x{0e84}\x{0e87}\x{0e88}\x{0e8a}\x{0e8d}\x{0e94}-\x{0e97}\x{0e99}-\x{0e9f}\x{0ea1}-\x{0ea3}\x{0ea5}\x{0ea7}\x{0eaa}\x{0eab}\x{0ead}-\x{0eb0}\x{0eb2}\x{0eb3}\x{0ebd}\x{0ec0}-\x{0ec4}\x{0ec6}\x{0edc}-\x{0edf}\x{0f00}\x{0f40}-\x{0f47}\x{0f49}-\x{0f6c}\x{0f88}-\x{0f8c}\x{1000}-\x{102a}\x{103f}\x{1050}-\x{1055}\x{105a}-\x{105d}\x{1061}\x{1065}\x{1066}\x{106e}-\x{1070}\x{1075}-\x{1081}\x{108e}\x{10a0}-\x{10c5}\x{10c7}\x{10cd}\x{10d0}-\x{10fa}\x{10fc}-\x{1248}\x{124a}-\x{124d}\x{1250}-\x{1256}\x{1258}\x{125a}-\x{125d}\x{1260}-\x{1288}\x{128a}-\x{128d}\x{1290}-\x{12b0}\x{12b2}-\x{12b5}\x{12b8}-\x{12be}\x{12c0}\x{12c2}-\x{12c5}\x{12c8}-\x{12d6}\x{12d8}-\x{1310}\x{1312}-\x{1315}\x{1318}-\x{135a}\x{1380}-\x{138f}\x{13a0}-\x{13f4}\x{1401}-\x{166c}\x{166f}-\x{167f}\x{1681}-\x{169a}\x{16a0}-\x{16ea}\x{16ee}-\x{16f0}\x{1700}-\x{170c}\x{170e}-\x{1711}\x{1720}-\x{1731}\x{1740}-\x{1751}\x{1760}-\x{176c}\x{176e}-\x{1770}\x{1780}-\x{17b3}\x{17d7}\x{17dc}\x{1820}-\x{1877}\x{1880}-\x{18a8}\x{18aa}\x{18b0}-\x{18f5}\x{1900}-\x{191c}\x{1950}-\x{196d}\x{1970}-\x{1974}\x{1980}-\x{19ab}\x{19c1}-\x{19c7}\x{1a00}-\x{1a16}\x{1a20}-\x{1a54}\x{1aa7}\x{1b05}-\x{1b33}\x{1b45}-\x{1b4b}\x{1b83}-\x{1ba0}\x{1bae}\x{1baf}\x{1bba}-\x{1be5}\x{1c00}-\x{1c23}\x{1c4d}-\x{1c4f}\x{1c5a}-\x{1c7d}\x{1ce9}-\x{1cec}\x{1cee}-\x{1cf1}\x{1cf5}\x{1cf6}\x{1d00}-\x{1dbf}\x{1e00}-\x{1f15}\x{1f18}-\x{1f1d}\x{1f20}-\x{1f45}\x{1f48}-\x{1f4d}\x{1f50}-\x{1f57}\x{1f59}\x{1f5b}\x{1f5d}\x{1f5f}-\x{1f7d}\x{1f80}-\x{1fb4}\x{1fb6}-\x{1fbc}\x{1fbe}\x{1fc2}-\x{1fc4}\x{1fc6}-\x{1fcc}\x{1fd0}-\x{1fd3}\x{1fd6}-\x{1fdb}\x{1fe0}-\x{1fec}\x{1ff2}-\x{1ff4}\x{1ff6}-\x{1ffc}\x{2071}\x{207f}\x{2090}-\x{209c}\x{2102}\x{2107}\x{210a}-\x{2113}\x{2115}\x{2119}-\x{211d}\x{2124}\x{2126}\x{2128}\x{212a}-\x{212d}\x{212f}-\x{2139}\x{213c}-\x{213f}\x{2145}-\x{2149}\x{214e}\x{2160}-\x{2188}\x{2c00}-\x{2c2e}\x{2c30}-\x{2c5e}\x{2c60}-\x{2ce4}\x{2ceb}-\x{2cee}\x{2cf2}\x{2cf3}\x{2d00}-\x{2d25}\x{2d27}\x{2d2d}\x{2d30}-\x{2d67}\x{2d6f}\x{2d80}-\x{2d96}\x{2da0}-\x{2da6}\x{2da8}-\x{2dae}\x{2db0}-\x{2db6}\x{2db8}-\x{2dbe}\x{2dc0}-\x{2dc6}\x{2dc8}-\x{2dce}\x{2dd0}-\x{2dd6}\x{2dd8}-\x{2dde}\x{2e2f}\x{3005}-\x{3007}\x{3021}-\x{3029}\x{3031}-\x{3035}\x{3038}-\x{303c}\x{3041}-\x{3096}\x{309d}-\x{309f}\x{30a1}-\x{30fa}\x{30fc}-\x{30ff}\x{3105}-\x{312d}\x{3131}-\x{318e}\x{31a0}-\x{31ba}\x{31f0}-\x{31ff}\x{3400}-\x{4db5}\x{4e00}-\x{9fcc}\x{a000}-\x{a48c}\x{a4d0}-\x{a4fd}\x{a500}-\x{a60c}\x{a610}-\x{a61f}\x{a62a}\x{a62b}\x{a640}-\x{a66e}\x{a67f}-\x{a697}\x{a6a0}-\x{a6ef}\x{a717}-\x{a71f}\x{a722}-\x{a788}\x{a78b}-\x{a78e}\x{a790}-\x{a793}\x{a7a0}-\x{a7aa}\x{a7f8}-\x{a801}\x{a803}-\x{a805}\x{a807}-\x{a80a}\x{a80c}-\x{a822}\x{a840}-\x{a873}\x{a882}-\x{a8b3}\x{a8f2}-\x{a8f7}\x{a8fb}\x{a90a}-\x{a925}\x{a930}-\x{a946}\x{a960}-\x{a97c}\x{a984}-\x{a9b2}\x{a9cf}\x{aa00}-\x{aa28}\x{aa40}-\x{aa42}\x{aa44}-\x{aa4b}\x{aa60}-\x{aa76}\x{aa7a}\x{aa80}-\x{aaaf}\x{aab1}\x{aab5}\x{aab6}\x{aab9}-\x{aabd}\x{aac0}\x{aac2}\x{aadb}-\x{aadd}\x{aae0}-\x{aaea}\x{aaf2}-\x{aaf4}\x{ab01}-\x{ab06}\x{ab09}-\x{ab0e}\x{ab11}-\x{ab16}\x{ab20}-\x{ab26}\x{ab28}-\x{ab2e}\x{abc0}-\x{abe2}\x{ac00}-\x{d7a3}\x{d7b0}-\x{d7c6}\x{d7cb}-\x{d7fb}\x{f900}-\x{fa6d}\x{fa70}-\x{fad9}\x{fb00}-\x{fb06}\x{fb13}-\x{fb17}\x{fb1d}\x{fb1f}-\x{fb28}\x{fb2a}-\x{fb36}\x{fb38}-\x{fb3c}\x{fb3e}\x{fb40}\x{fb41}\x{fb43}\x{fb44}\x{fb46}-\x{fbb1}\x{fbd3}-\x{fd3d}\x{fd50}-\x{fd8f}\x{fd92}-\x{fdc7}\x{fdf0}-\x{fdfb}\x{fe70}-\x{fe74}\x{fe76}-\x{fefc}\x{ff21}-\x{ff3a}\x{ff41}-\x{ff5a}\x{ff66}-\x{ffbe}\x{ffc2}-\x{ffc7}\x{ffca}-\x{ffcf}\x{ffd2}-\x{ffd7}\x{ffda}-\x{ffdc}0-9\x{0300}-\x{036f}\x{0483}-\x{0487}\x{0591}-\x{05bd}\x{05bf}\x{05c1}\x{05c2}\x{05c4}\x{05c5}\x{05c7}\x{0610}-\x{061a}\x{064b}-\x{0669}\x{0670}\x{06d6}-\x{06dc}\x{06df}-\x{06e4}\x{06e7}\x{06e8}\x{06ea}-\x{06ed}\x{06f0}-\x{06f9}\x{0711}\x{0730}-\x{074a}\x{07a6}-\x{07b0}\x{07c0}-\x{07c9}\x{07eb}-\x{07f3}\x{0816}-\x{0819}\x{081b}-\x{0823}\x{0825}-\x{0827}\x{0829}-\x{082d}\x{0859}-\x{085b}\x{08e4}-\x{08fe}\x{0900}-\x{0903}\x{093a}-\x{093c}\x{093e}-\x{094f}\x{0951}-\x{0957}\x{0962}\x{0963}\x{0966}-\x{096f}\x{0981}-\x{0983}\x{09bc}\x{09be}-\x{09c4}\x{09c7}\x{09c8}\x{09cb}-\x{09cd}\x{09d7}\x{09e2}\x{09e3}\x{09e6}-\x{09ef}\x{0a01}-\x{0a03}\x{0a3c}\x{0a3e}-\x{0a42}\x{0a47}\x{0a48}\x{0a4b}-\x{0a4d}\x{0a51}\x{0a66}-\x{0a71}\x{0a75}\x{0a81}-\x{0a83}\x{0abc}\x{0abe}-\x{0ac5}\x{0ac7}-\x{0ac9}\x{0acb}-\x{0acd}\x{0ae2}\x{0ae3}\x{0ae6}-\x{0aef}\x{0b01}-\x{0b03}\x{0b3c}\x{0b3e}-\x{0b44}\x{0b47}\x{0b48}\x{0b4b}-\x{0b4d}\x{0b56}\x{0b57}\x{0b62}\x{0b63}\x{0b66}-\x{0b6f}\x{0b82}\x{0bbe}-\x{0bc2}\x{0bc6}-\x{0bc8}\x{0bca}-\x{0bcd}\x{0bd7}\x{0be6}-\x{0bef}\x{0c01}-\x{0c03}\x{0c3e}-\x{0c44}\x{0c46}-\x{0c48}\x{0c4a}-\x{0c4d}\x{0c55}\x{0c56}\x{0c62}\x{0c63}\x{0c66}-\x{0c6f}\x{0c82}\x{0c83}\x{0cbc}\x{0cbe}-\x{0cc4}\x{0cc6}-\x{0cc8}\x{0cca}-\x{0ccd}\x{0cd5}\x{0cd6}\x{0ce2}\x{0ce3}\x{0ce6}-\x{0cef}\x{0d02}\x{0d03}\x{0d3e}-\x{0d44}\x{0d46}-\x{0d48}\x{0d4a}-\x{0d4d}\x{0d57}\x{0d62}\x{0d63}\x{0d66}-\x{0d6f}\x{0d82}\x{0d83}\x{0dca}\x{0dcf}-\x{0dd4}\x{0dd6}\x{0dd8}-\x{0ddf}\x{0df2}\x{0df3}\x{0e31}\x{0e34}-\x{0e3a}\x{0e47}-\x{0e4e}\x{0e50}-\x{0e59}\x{0eb1}\x{0eb4}-\x{0eb9}\x{0ebb}\x{0ebc}\x{0ec8}-\x{0ecd}\x{0ed0}-\x{0ed9}\x{0f18}\x{0f19}\x{0f20}-\x{0f29}\x{0f35}\x{0f37}\x{0f39}\x{0f3e}\x{0f3f}\x{0f71}-\x{0f84}\x{0f86}\x{0f87}\x{0f8d}-\x{0f97}\x{0f99}-\x{0fbc}\x{0fc6}\x{102b}-\x{103e}\x{1040}-\x{1049}\x{1056}-\x{1059}\x{105e}-\x{1060}\x{1062}-\x{1064}\x{1067}-\x{106d}\x{1071}-\x{1074}\x{1082}-\x{108d}\x{108f}-\x{109d}\x{135d}-\x{135f}\x{1712}-\x{1714}\x{1732}-\x{1734}\x{1752}\x{1753}\x{1772}\x{1773}\x{17b4}-\x{17d3}\x{17dd}\x{17e0}-\x{17e9}\x{180b}-\x{180d}\x{1810}-\x{1819}\x{18a9}\x{1920}-\x{192b}\x{1930}-\x{193b}\x{1946}-\x{194f}\x{19b0}-\x{19c0}\x{19c8}\x{19c9}\x{19d0}-\x{19d9}\x{1a17}-\x{1a1b}\x{1a55}-\x{1a5e}\x{1a60}-\x{1a7c}\x{1a7f}-\x{1a89}\x{1a90}-\x{1a99}\x{1b00}-\x{1b04}\x{1b34}-\x{1b44}\x{1b50}-\x{1b59}\x{1b6b}-\x{1b73}\x{1b80}-\x{1b82}\x{1ba1}-\x{1bad}\x{1bb0}-\x{1bb9}\x{1be6}-\x{1bf3}\x{1c24}-\x{1c37}\x{1c40}-\x{1c49}\x{1c50}-\x{1c59}\x{1cd0}-\x{1cd2}\x{1cd4}-\x{1ce8}\x{1ced}\x{1cf2}-\x{1cf4}\x{1dc0}-\x{1de6}\x{1dfc}-\x{1dff}\x{200c}\x{200d}\x{203f}\x{2040}\x{2054}\x{20d0}-\x{20dc}\x{20e1}\x{20e5}-\x{20f0}\x{2cef}-\x{2cf1}\x{2d7f}\x{2de0}-\x{2dff}\x{302a}-\x{302f}\x{3099}\x{309a}\x{a620}-\x{a629}\x{a66f}\x{a674}-\x{a67d}\x{a69f}\x{a6f0}\x{a6f1}\x{a802}\x{a806}\x{a80b}\x{a823}-\x{a827}\x{a880}\x{a881}\x{a8b4}-\x{a8c4}\x{a8d0}-\x{a8d9}\x{a8e0}-\x{a8f1}\x{a900}-\x{a909}\x{a926}-\x{a92d}\x{a947}-\x{a953}\x{a980}-\x{a983}\x{a9b3}-\x{a9c0}\x{a9d0}-\x{a9d9}\x{aa29}-\x{aa36}\x{aa43}\x{aa4c}\x{aa4d}\x{aa50}-\x{aa59}\x{aa7b}\x{aab0}\x{aab2}-\x{aab4}\x{aab7}\x{aab8}\x{aabe}\x{aabf}\x{aac1}\x{aaeb}-\x{aaef}\x{aaf5}\x{aaf6}\x{abe3}-\x{abea}\x{abec}\x{abed}\x{abf0}-\x{abf9}\x{fb1e}\x{fe00}-\x{fe0f}\x{fe20}-\x{fe26}\x{fe33}\x{fe34}\x{fe4d}-\x{fe4f}\x{ff10}-\x{ff19}\x{ff3f}]*\b'; |
||
38 | |||
39 | /** |
||
40 | * Full list of JavaScript reserved words. |
||
41 | * Will be loaded from /data/js/keywords_reserved.txt. |
||
42 | * |
||
43 | * @see https://mathiasbynens.be/notes/reserved-keywords |
||
44 | * |
||
45 | * @var string[] |
||
46 | */ |
||
47 | protected $keywordsReserved = array(); |
||
48 | |||
49 | /** |
||
50 | * List of JavaScript reserved words that accept a <variable, value, ...> |
||
51 | * after them. Some end of lines are not the end of a statement, like with |
||
52 | * these keywords. |
||
53 | * |
||
54 | * E.g.: we shouldn't insert a ; after this else |
||
55 | * else |
||
56 | * console.log('this is quite fine') |
||
57 | * |
||
58 | * Will be loaded from /data/js/keywords_before.txt |
||
59 | * |
||
60 | * @var string[] |
||
61 | */ |
||
62 | protected $keywordsBefore = array(); |
||
63 | |||
64 | /** |
||
65 | * List of JavaScript reserved words that accept a <variable, value, ...> |
||
66 | * before them. Some end of lines are not the end of a statement, like when |
||
67 | * continued by one of these keywords on the newline. |
||
68 | * |
||
69 | * E.g.: we shouldn't insert a ; before this instanceof |
||
70 | * variable |
||
71 | * instanceof String |
||
72 | * |
||
73 | * Will be loaded from /data/js/keywords_after.txt |
||
74 | * |
||
75 | * @var string[] |
||
76 | */ |
||
77 | protected $keywordsAfter = array(); |
||
78 | |||
79 | /** |
||
80 | * List of all JavaScript operators. |
||
81 | * |
||
82 | * Will be loaded from /data/js/operators.txt |
||
83 | * |
||
84 | * @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators |
||
85 | * |
||
86 | * @var string[] |
||
87 | */ |
||
88 | protected $operators = array(); |
||
89 | |||
90 | /** |
||
91 | * List of JavaScript operators that accept a <variable, value, ...> after |
||
92 | * them. Some end of lines are not the end of a statement, like with these |
||
93 | * operators. |
||
94 | * |
||
95 | * Note: Most operators are fine, we've only removed ++ and --. |
||
96 | * ++ & -- have to be joined with the value they're in-/decrementing. |
||
97 | * |
||
98 | * Will be loaded from /data/js/operators_before.txt |
||
99 | * |
||
100 | * @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators |
||
101 | * |
||
102 | * @var string[] |
||
103 | */ |
||
104 | protected $operatorsBefore = array(); |
||
105 | |||
106 | /** |
||
107 | * List of JavaScript operators that accept a <variable, value, ...> before |
||
108 | * them. Some end of lines are not the end of a statement, like when |
||
109 | * continued by one of these operators on the newline. |
||
110 | * |
||
111 | * Note: Most operators are fine, we've only removed ), ], ++, --, ! and ~. |
||
112 | * There can't be a newline separating ! or ~ and whatever it is negating. |
||
113 | * ++ & -- have to be joined with the value they're in-/decrementing. |
||
114 | * ) & ] are "special" in that they have lots or usecases. () for example |
||
115 | * is used for function calls, for grouping, in if () and for (), ... |
||
116 | * |
||
117 | * Will be loaded from /data/js/operators_after.txt |
||
118 | * |
||
119 | * @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators |
||
120 | * |
||
121 | * @var string[] |
||
122 | */ |
||
123 | protected $operatorsAfter = array(); |
||
124 | |||
125 | public function __construct() |
||
126 | { |
||
1042 | daniel-mar | 127 | call_user_func_array(array('\\MatthiasMullie\Minify\\Minify', '__construct'), func_get_args()); |
597 | daniel-mar | 128 | |
1042 | daniel-mar | 129 | $dataDir = __DIR__ . '/../data/js/'; |
597 | daniel-mar | 130 | $options = FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES; |
1042 | daniel-mar | 131 | $this->keywordsReserved = file($dataDir . 'keywords_reserved.txt', $options); |
132 | $this->keywordsBefore = file($dataDir . 'keywords_before.txt', $options); |
||
133 | $this->keywordsAfter = file($dataDir . 'keywords_after.txt', $options); |
||
134 | $this->operators = file($dataDir . 'operators.txt', $options); |
||
135 | $this->operatorsBefore = file($dataDir . 'operators_before.txt', $options); |
||
136 | $this->operatorsAfter = file($dataDir . 'operators_after.txt', $options); |
||
597 | daniel-mar | 137 | } |
138 | |||
139 | /** |
||
140 | * Minify the data. |
||
141 | * Perform JS optimizations. |
||
142 | * |
||
143 | * @param string[optional] $path Path to write the data to |
||
144 | * |
||
145 | * @return string The minified data |
||
146 | */ |
||
147 | public function execute($path = null) |
||
148 | { |
||
149 | $content = ''; |
||
150 | |||
151 | /* |
||
152 | * Let's first take out strings, comments and regular expressions. |
||
153 | * All of these can contain JS code-like characters, and we should make |
||
154 | * sure any further magic ignores anything inside of these. |
||
155 | * |
||
156 | * Consider this example, where we should not strip any whitespace: |
||
157 | * var str = "a test"; |
||
158 | * |
||
159 | * Comments will be removed altogether, strings and regular expressions |
||
160 | * will be replaced by placeholder text, which we'll restore later. |
||
161 | */ |
||
162 | $this->extractStrings('\'"`'); |
||
163 | $this->stripComments(); |
||
164 | $this->extractRegex(); |
||
165 | |||
166 | // loop files |
||
167 | foreach ($this->data as $source => $js) { |
||
168 | // take out strings, comments & regex (for which we've registered |
||
169 | // the regexes just a few lines earlier) |
||
170 | $js = $this->replace($js); |
||
171 | |||
172 | $js = $this->propertyNotation($js); |
||
173 | $js = $this->shortenBools($js); |
||
174 | $js = $this->stripWhitespace($js); |
||
175 | |||
176 | // combine js: separating the scripts by a ; |
||
1042 | daniel-mar | 177 | $content .= $js . ';'; |
597 | daniel-mar | 178 | } |
179 | |||
180 | // clean up leftover `;`s from the combination of multiple scripts |
||
181 | $content = ltrim($content, ';'); |
||
182 | $content = (string) substr($content, 0, -1); |
||
183 | |||
184 | /* |
||
185 | * Earlier, we extracted strings & regular expressions and replaced them |
||
186 | * with placeholder text. This will restore them. |
||
187 | */ |
||
188 | $content = $this->restoreExtractedData($content); |
||
189 | |||
190 | return $content; |
||
191 | } |
||
192 | |||
193 | /** |
||
194 | * Strip comments from source code. |
||
195 | */ |
||
196 | protected function stripComments() |
||
197 | { |
||
1308 | daniel-mar | 198 | $this->stripMultilineComments(); |
597 | daniel-mar | 199 | |
200 | // single-line comments |
||
201 | $this->registerPattern('/\/\/.*$/m', ''); |
||
202 | } |
||
203 | |||
204 | /** |
||
205 | * JS can have /-delimited regular expressions, like: /ab+c/.match(string). |
||
206 | * |
||
207 | * The content inside the regex can contain characters that may be confused |
||
208 | * for JS code: e.g. it could contain whitespace it needs to match & we |
||
209 | * don't want to strip whitespace in there. |
||
210 | * |
||
211 | * The regex can be pretty simple: we don't have to care about comments, |
||
212 | * (which also use slashes) because stripComments() will have stripped those |
||
213 | * already. |
||
214 | * |
||
215 | * This method will replace all string content with simple REGEX# |
||
216 | * placeholder text, so we've rid all regular expressions from characters |
||
217 | * that may be misinterpreted. Original regex content will be saved in |
||
218 | * $this->extracted and after doing all other minifying, we can restore the |
||
219 | * original content via restoreRegex() |
||
220 | */ |
||
221 | protected function extractRegex() |
||
222 | { |
||
223 | // PHP only supports $this inside anonymous functions since 5.4 |
||
224 | $minifier = $this; |
||
225 | $callback = function ($match) use ($minifier) { |
||
226 | $count = count($minifier->extracted); |
||
1042 | daniel-mar | 227 | $placeholder = '"' . $count . '"'; |
597 | daniel-mar | 228 | $minifier->extracted[$placeholder] = $match[0]; |
229 | |||
230 | return $placeholder; |
||
231 | }; |
||
232 | |||
233 | // match all chars except `/` and `\` |
||
234 | // `\` is allowed though, along with whatever char follows (which is the |
||
235 | // one being escaped) |
||
236 | // this should allow all chars, except for an unescaped `/` (= the one |
||
237 | // closing the regex) |
||
238 | // then also ignore bare `/` inside `[]`, where they don't need to be |
||
239 | // escaped: anything inside `[]` can be ignored safely |
||
240 | $pattern = '\\/(?!\*)(?:[^\\[\\/\\\\\n\r]++|(?:\\\\.)++|(?:\\[(?:[^\\]\\\\\n\r]++|(?:\\\\.)++)++\\])++)++\\/[gimuy]*'; |
||
241 | |||
242 | // a regular expression can only be followed by a few operators or some |
||
243 | // of the RegExp methods (a `\` followed by a variable or value is |
||
244 | // likely part of a division, not a regex) |
||
245 | $keywords = array('do', 'in', 'new', 'else', 'throw', 'yield', 'delete', 'return', 'typeof'); |
||
1042 | daniel-mar | 246 | $before = '(^|[=:,;\+\-\*\?\/\}\(\{\[&\|!]|' . implode('|', $keywords) . ')\s*'; |
597 | daniel-mar | 247 | $propertiesAndMethods = array( |
248 | // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp#Properties_2 |
||
249 | 'constructor', |
||
250 | 'flags', |
||
251 | 'global', |
||
252 | 'ignoreCase', |
||
253 | 'multiline', |
||
254 | 'source', |
||
255 | 'sticky', |
||
256 | 'unicode', |
||
257 | // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp#Methods_2 |
||
258 | 'compile(', |
||
259 | 'exec(', |
||
260 | 'test(', |
||
261 | 'toSource(', |
||
262 | 'toString(', |
||
263 | ); |
||
264 | $delimiters = array_fill(0, count($propertiesAndMethods), '/'); |
||
265 | $propertiesAndMethods = array_map('preg_quote', $propertiesAndMethods, $delimiters); |
||
1042 | daniel-mar | 266 | $after = '(?=\s*([\.,;:\)\}&\|+]|\/\/|$|\.(' . implode('|', $propertiesAndMethods) . ')))'; |
267 | $this->registerPattern('/' . $before . '\K' . $pattern . $after . '/', $callback); |
||
597 | daniel-mar | 268 | |
269 | // regular expressions following a `)` are rather annoying to detect... |
||
270 | // quite often, `/` after `)` is a division operator & if it happens to |
||
271 | // be followed by another one (or a comment), it is likely to be |
||
272 | // confused for a regular expression |
||
273 | // however, it's perfectly possible for a regex to follow a `)`: after |
||
274 | // a single-line `if()`, `while()`, ... statement, for example |
||
275 | // since, when they occur like that, they're always the start of a |
||
276 | // statement, there's only a limited amount of ways they can be useful: |
||
277 | // by calling the regex methods directly |
||
278 | // if a regex following `)` is not followed by `.<property or method>`, |
||
279 | // it's quite likely not a regex |
||
280 | $before = '\)\s*'; |
||
1042 | daniel-mar | 281 | $after = '(?=\s*\.(' . implode('|', $propertiesAndMethods) . '))'; |
282 | $this->registerPattern('/' . $before . '\K' . $pattern . $after . '/', $callback); |
||
597 | daniel-mar | 283 | |
284 | // 1 more edge case: a regex can be followed by a lot more operators or |
||
285 | // keywords if there's a newline (ASI) in between, where the operator |
||
286 | // actually starts a new statement |
||
287 | // (https://github.com/matthiasmullie/minify/issues/56) |
||
288 | $operators = $this->getOperatorsForRegex($this->operatorsBefore, '/'); |
||
289 | $operators += $this->getOperatorsForRegex($this->keywordsReserved, '/'); |
||
1042 | daniel-mar | 290 | $after = '(?=\s*\n\s*(' . implode('|', $operators) . '))'; |
291 | $this->registerPattern('/' . $pattern . $after . '/', $callback); |
||
597 | daniel-mar | 292 | } |
293 | |||
294 | /** |
||
295 | * Strip whitespace. |
||
296 | * |
||
297 | * We won't strip *all* whitespace, but as much as possible. The thing that |
||
298 | * we'll preserve are newlines we're unsure about. |
||
299 | * JavaScript doesn't require statements to be terminated with a semicolon. |
||
300 | * It will automatically fix missing semicolons with ASI (automatic semi- |
||
301 | * colon insertion) at the end of line causing errors (without semicolon.) |
||
302 | * |
||
303 | * Because it's sometimes hard to tell if a newline is part of a statement |
||
304 | * that should be terminated or not, we'll just leave some of them alone. |
||
305 | * |
||
306 | * @param string $content The content to strip the whitespace for |
||
307 | * |
||
308 | * @return string |
||
309 | */ |
||
310 | protected function stripWhitespace($content) |
||
311 | { |
||
312 | // uniform line endings, make them all line feed |
||
313 | $content = str_replace(array("\r\n", "\r"), "\n", $content); |
||
314 | |||
315 | // collapse all non-line feed whitespace into a single space |
||
316 | $content = preg_replace('/[^\S\n]+/', ' ', $content); |
||
317 | |||
318 | // strip leading & trailing whitespace |
||
319 | $content = str_replace(array(" \n", "\n "), "\n", $content); |
||
320 | |||
321 | // collapse consecutive line feeds into just 1 |
||
322 | $content = preg_replace('/\n+/', "\n", $content); |
||
323 | |||
324 | $operatorsBefore = $this->getOperatorsForRegex($this->operatorsBefore, '/'); |
||
325 | $operatorsAfter = $this->getOperatorsForRegex($this->operatorsAfter, '/'); |
||
326 | $operators = $this->getOperatorsForRegex($this->operators, '/'); |
||
327 | $keywordsBefore = $this->getKeywordsForRegex($this->keywordsBefore, '/'); |
||
328 | $keywordsAfter = $this->getKeywordsForRegex($this->keywordsAfter, '/'); |
||
329 | |||
330 | // strip whitespace that ends in (or next line begin with) an operator |
||
331 | // that allows statements to be broken up over multiple lines |
||
332 | unset($operatorsBefore['+'], $operatorsBefore['-'], $operatorsAfter['+'], $operatorsAfter['-']); |
||
333 | $content = preg_replace( |
||
334 | array( |
||
1042 | daniel-mar | 335 | '/(' . implode('|', $operatorsBefore) . ')\s+/', |
336 | '/\s+(' . implode('|', $operatorsAfter) . ')/', |
||
597 | daniel-mar | 337 | ), |
338 | '\\1', |
||
339 | $content |
||
340 | ); |
||
341 | |||
342 | // make sure + and - can't be mistaken for, or joined into ++ and -- |
||
343 | $content = preg_replace( |
||
344 | array( |
||
345 | '/(?<![\+\-])\s*([\+\-])(?![\+\-])/', |
||
346 | '/(?<![\+\-])([\+\-])\s*(?![\+\-])/', |
||
347 | ), |
||
348 | '\\1', |
||
349 | $content |
||
350 | ); |
||
351 | |||
352 | // collapse whitespace around reserved words into single space |
||
1042 | daniel-mar | 353 | $content = preg_replace('/(^|[;\}\s])\K(' . implode('|', $keywordsBefore) . ')\s+/', '\\2 ', $content); |
354 | $content = preg_replace('/\s+(' . implode('|', $keywordsAfter) . ')(?=([;\{\s]|$))/', ' \\1', $content); |
||
597 | daniel-mar | 355 | |
356 | /* |
||
357 | * We didn't strip whitespace after a couple of operators because they |
||
358 | * could be used in different contexts and we can't be sure it's ok to |
||
359 | * strip the newlines. However, we can safely strip any non-line feed |
||
360 | * whitespace that follows them. |
||
361 | */ |
||
362 | $operatorsDiffBefore = array_diff($operators, $operatorsBefore); |
||
363 | $operatorsDiffAfter = array_diff($operators, $operatorsAfter); |
||
1042 | daniel-mar | 364 | $content = preg_replace('/(' . implode('|', $operatorsDiffBefore) . ')[^\S\n]+/', '\\1', $content); |
365 | $content = preg_replace('/[^\S\n]+(' . implode('|', $operatorsDiffAfter) . ')/', '\\1', $content); |
||
597 | daniel-mar | 366 | |
367 | /* |
||
368 | * Whitespace after `return` can be omitted in a few occasions |
||
369 | * (such as when followed by a string or regex) |
||
370 | * Same for whitespace in between `)` and `{`, or between `{` and some |
||
371 | * keywords. |
||
372 | */ |
||
373 | $content = preg_replace('/\breturn\s+(["\'\/\+\-])/', 'return$1', $content); |
||
374 | $content = preg_replace('/\)\s+\{/', '){', $content); |
||
375 | $content = preg_replace('/}\n(else|catch|finally)\b/', '}$1', $content); |
||
376 | |||
377 | /* |
||
378 | * Get rid of double semicolons, except where they can be used like: |
||
379 | * "for(v=1,_=b;;)", "for(v=1;;v++)" or "for(;;ja||(ja=true))". |
||
380 | * I'll safeguard these double semicolons inside for-loops by |
||
381 | * temporarily replacing them with an invalid condition: they won't have |
||
382 | * a double semicolon and will be easy to spot to restore afterwards. |
||
383 | */ |
||
384 | $content = preg_replace('/\bfor\(([^;]*);;([^;]*)\)/', 'for(\\1;-;\\2)', $content); |
||
385 | $content = preg_replace('/;+/', ';', $content); |
||
386 | $content = preg_replace('/\bfor\(([^;]*);-;([^;]*)\)/', 'for(\\1;;\\2)', $content); |
||
387 | |||
388 | /* |
||
389 | * Next, we'll be removing all semicolons where ASI kicks in. |
||
390 | * for-loops however, can have an empty body (ending in only a |
||
391 | * semicolon), like: `for(i=1;i<3;i++);`, of `for(i in list);` |
||
392 | * Here, nothing happens during the loop; it's just used to keep |
||
393 | * increasing `i`. With that ; omitted, the next line would be expected |
||
394 | * to be the for-loop's body... Same goes for while loops. |
||
395 | * I'm going to double that semicolon (if any) so after the next line, |
||
396 | * which strips semicolons here & there, we're still left with this one. |
||
842 | daniel-mar | 397 | * Note the special recursive construct in the three inner parts of the for: |
398 | * (\{([^\{\}]*(?-2))*[^\{\}]*\})? - it is intended to match inline |
||
399 | * functions bodies, e.g.: i<arr.map(function(e){return e}).length. |
||
400 | * Also note that the construct is applied only once and multiplied |
||
401 | * for each part of the for, otherwise it risks a catastrophic backtracking. |
||
402 | * The limitation is that it will not allow closures in more than one |
||
403 | * of the three parts for a specific for() case. |
||
404 | * REGEX throwing catastrophic backtracking: $content = preg_replace('/(for\([^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*;[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*;[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*\));(\}|$)/s', '\\1;;\\8', $content); |
||
597 | daniel-mar | 405 | */ |
846 | daniel-mar | 406 | $content = preg_replace('/(for\((?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*);[^;\{]*;[^;\{]*\));(\}|$)/s', '\\1;;\\4', $content); |
407 | $content = preg_replace('/(for\([^;\{]*;(?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*);[^;\{]*\));(\}|$)/s', '\\1;;\\4', $content); |
||
408 | $content = preg_replace('/(for\([^;\{]*;[^;\{]*;(?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*)\));(\}|$)/s', '\\1;;\\4', $content); |
||
842 | daniel-mar | 409 | |
597 | daniel-mar | 410 | $content = preg_replace('/(for\([^;\{]+\s+in\s+[^;\{]+\));(\}|$)/s', '\\1;;\\2', $content); |
846 | daniel-mar | 411 | |
597 | daniel-mar | 412 | /* |
846 | daniel-mar | 413 | * Do the same for the if's that don't have a body but are followed by ;} |
414 | */ |
||
415 | $content = preg_replace('/(\bif\s*\([^{;]*\));\}/s', '\\1;;}', $content); |
||
416 | |||
417 | /* |
||
597 | daniel-mar | 418 | * Below will also keep `;` after a `do{}while();` along with `while();` |
419 | * While these could be stripped after do-while, detecting this |
||
420 | * distinction is cumbersome, so I'll play it safe and make sure `;` |
||
421 | * after any kind of `while` is kept. |
||
422 | */ |
||
423 | $content = preg_replace('/(while\([^;\{]+\));(\}|$)/s', '\\1;;\\2', $content); |
||
424 | |||
425 | /* |
||
426 | * We also can't strip empty else-statements. Even though they're |
||
427 | * useless and probably shouldn't be in the code in the first place, we |
||
428 | * shouldn't be stripping the `;` that follows it as it breaks the code. |
||
429 | * We can just remove those useless else-statements completely. |
||
430 | * |
||
431 | * @see https://github.com/matthiasmullie/minify/issues/91 |
||
432 | */ |
||
433 | $content = preg_replace('/else;/s', '', $content); |
||
434 | |||
435 | /* |
||
436 | * We also don't really want to terminate statements followed by closing |
||
437 | * curly braces (which we've ignored completely up until now) or end-of- |
||
438 | * script: ASI will kick in here & we're all about minifying. |
||
439 | * Semicolons at beginning of the file don't make any sense either. |
||
440 | */ |
||
441 | $content = preg_replace('/;(\}|$)/s', '\\1', $content); |
||
442 | $content = ltrim($content, ';'); |
||
443 | |||
444 | // get rid of remaining whitespace af beginning/end |
||
445 | return trim($content); |
||
446 | } |
||
447 | |||
448 | /** |
||
449 | * We'll strip whitespace around certain operators with regular expressions. |
||
450 | * This will prepare the given array by escaping all characters. |
||
451 | * |
||
452 | * @param string[] $operators |
||
453 | * @param string $delimiter |
||
454 | * |
||
455 | * @return string[] |
||
456 | */ |
||
457 | protected function getOperatorsForRegex(array $operators, $delimiter = '/') |
||
458 | { |
||
459 | // escape operators for use in regex |
||
460 | $delimiters = array_fill(0, count($operators), $delimiter); |
||
461 | $escaped = array_map('preg_quote', $operators, $delimiters); |
||
462 | |||
463 | $operators = array_combine($operators, $escaped); |
||
464 | |||
465 | // ignore + & - for now, they'll get special treatment |
||
466 | unset($operators['+'], $operators['-']); |
||
467 | |||
468 | // dot can not just immediately follow a number; it can be confused for |
||
469 | // decimal point, or calling a method on it, e.g. 42 .toString() |
||
470 | $operators['.'] = '(?<![0-9]\s)\.'; |
||
471 | |||
472 | // don't confuse = with other assignment shortcuts (e.g. +=) |
||
473 | $chars = preg_quote('+-*\=<>%&|', $delimiter); |
||
1042 | daniel-mar | 474 | $operators['='] = '(?<![' . $chars . '])\='; |
597 | daniel-mar | 475 | |
476 | return $operators; |
||
477 | } |
||
478 | |||
479 | /** |
||
480 | * We'll strip whitespace around certain keywords with regular expressions. |
||
481 | * This will prepare the given array by escaping all characters. |
||
482 | * |
||
483 | * @param string[] $keywords |
||
484 | * @param string $delimiter |
||
485 | * |
||
486 | * @return string[] |
||
487 | */ |
||
488 | protected function getKeywordsForRegex(array $keywords, $delimiter = '/') |
||
489 | { |
||
490 | // escape keywords for use in regex |
||
491 | $delimiter = array_fill(0, count($keywords), $delimiter); |
||
492 | $escaped = array_map('preg_quote', $keywords, $delimiter); |
||
493 | |||
494 | // add word boundaries |
||
495 | array_walk($keywords, function ($value) { |
||
1042 | daniel-mar | 496 | return '\b' . $value . '\b'; |
597 | daniel-mar | 497 | }); |
498 | |||
499 | $keywords = array_combine($keywords, $escaped); |
||
500 | |||
501 | return $keywords; |
||
502 | } |
||
503 | |||
504 | /** |
||
505 | * Replaces all occurrences of array['key'] by array.key. |
||
506 | * |
||
507 | * @param string $content |
||
508 | * |
||
509 | * @return string |
||
510 | */ |
||
511 | protected function propertyNotation($content) |
||
512 | { |
||
513 | // PHP only supports $this inside anonymous functions since 5.4 |
||
514 | $minifier = $this; |
||
515 | $keywords = $this->keywordsReserved; |
||
516 | $callback = function ($match) use ($minifier, $keywords) { |
||
517 | $property = trim($minifier->extracted[$match[1]], '\'"'); |
||
518 | |||
519 | /* |
||
520 | * Check if the property is a reserved keyword. In this context (as |
||
521 | * property of an object literal/array) it shouldn't matter, but IE8 |
||
522 | * freaks out with "Expected identifier". |
||
523 | */ |
||
524 | if (in_array($property, $keywords)) { |
||
525 | return $match[0]; |
||
526 | } |
||
527 | |||
528 | /* |
||
529 | * See if the property is in a variable-like format (e.g. |
||
530 | * array['key-here'] can't be replaced by array.key-here since '-' |
||
531 | * is not a valid character there. |
||
532 | */ |
||
1042 | daniel-mar | 533 | if (!preg_match('/^' . $minifier::REGEX_VARIABLE . '$/u', $property)) { |
597 | daniel-mar | 534 | return $match[0]; |
535 | } |
||
536 | |||
1042 | daniel-mar | 537 | return '.' . $property; |
597 | daniel-mar | 538 | }; |
539 | |||
540 | /* |
||
541 | * Figure out if previous character is a variable name (of the array |
||
542 | * we want to use property notation on) - this is to make sure |
||
543 | * standalone ['value'] arrays aren't confused for keys-of-an-array. |
||
544 | * We can (and only have to) check the last character, because PHP's |
||
545 | * regex implementation doesn't allow unfixed-length look-behind |
||
546 | * assertions. |
||
547 | */ |
||
548 | preg_match('/(\[[^\]]+\])[^\]]*$/', static::REGEX_VARIABLE, $previousChar); |
||
549 | $previousChar = $previousChar[1]; |
||
550 | |||
551 | /* |
||
552 | * Make sure word preceding the ['value'] is not a keyword, e.g. |
||
553 | * return['x']. Because -again- PHP's regex implementation doesn't allow |
||
554 | * unfixed-length look-behind assertions, I'm just going to do a lot of |
||
555 | * separate look-behind assertions, one for each keyword. |
||
556 | */ |
||
557 | $keywords = $this->getKeywordsForRegex($keywords); |
||
1042 | daniel-mar | 558 | $keywords = '(?<!' . implode(')(?<!', $keywords) . ')'; |
597 | daniel-mar | 559 | |
1042 | daniel-mar | 560 | return preg_replace_callback('/(?<=' . $previousChar . '|\])' . $keywords . '\[\s*(([\'"])[0-9]+\\2)\s*\]/u', $callback, $content); |
597 | daniel-mar | 561 | } |
562 | |||
563 | /** |
||
564 | * Replaces true & false by !0 and !1. |
||
565 | * |
||
566 | * @param string $content |
||
567 | * |
||
568 | * @return string |
||
569 | */ |
||
570 | protected function shortenBools($content) |
||
571 | { |
||
572 | /* |
||
573 | * 'true' or 'false' could be used as property names (which may be |
||
574 | * followed by whitespace) - we must not replace those! |
||
575 | * Since PHP doesn't allow variable-length (to account for the |
||
576 | * whitespace) lookbehind assertions, I need to capture the leading |
||
577 | * character and check if it's a `.` |
||
578 | */ |
||
579 | $callback = function ($match) { |
||
580 | if (trim($match[1]) === '.') { |
||
581 | return $match[0]; |
||
582 | } |
||
583 | |||
1042 | daniel-mar | 584 | return $match[1] . ($match[2] === 'true' ? '!0' : '!1'); |
597 | daniel-mar | 585 | }; |
586 | $content = preg_replace_callback('/(^|.\s*)\b(true|false)\b(?!:)/', $callback, $content); |
||
587 | |||
588 | // for(;;) is exactly the same as while(true), but shorter :) |
||
589 | $content = preg_replace('/\bwhile\(!0\){/', 'for(;;){', $content); |
||
590 | |||
591 | // now make sure we didn't turn any do ... while(true) into do ... for(;;) |
||
592 | preg_match_all('/\bdo\b/', $content, $dos, PREG_OFFSET_CAPTURE | PREG_SET_ORDER); |
||
593 | |||
594 | // go backward to make sure positional offsets aren't altered when $content changes |
||
595 | $dos = array_reverse($dos); |
||
596 | foreach ($dos as $do) { |
||
597 | $offsetDo = $do[0][1]; |
||
598 | |||
599 | // find all `while` (now `for`) following `do`: one of those must be |
||
600 | // associated with the `do` and be turned back into `while` |
||
601 | preg_match_all('/\bfor\(;;\)/', $content, $whiles, PREG_OFFSET_CAPTURE | PREG_SET_ORDER, $offsetDo); |
||
602 | foreach ($whiles as $while) { |
||
603 | $offsetWhile = $while[0][1]; |
||
604 | |||
605 | $open = substr_count($content, '{', $offsetDo, $offsetWhile - $offsetDo); |
||
606 | $close = substr_count($content, '}', $offsetDo, $offsetWhile - $offsetDo); |
||
607 | if ($open === $close) { |
||
608 | // only restore `while` if amount of `{` and `}` are the same; |
||
609 | // otherwise, that `for` isn't associated with this `do` |
||
610 | $content = substr_replace($content, 'while(!0)', $offsetWhile, strlen('for(;;)')); |
||
611 | break; |
||
612 | } |
||
613 | } |
||
614 | } |
||
615 | |||
616 | return $content; |
||
617 | } |
||
618 | } |