joomla
/
minify
29
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
minify/src/ef6438c4-76db-4aa1-bb15-bc9.../code.php

862 lines
44 KiB
PHP
Raw Normal View History

first commit
2023-04-15 14:58:37 +00:00
<?php
/**
* JavaScript minifier.
*
* Please report bugs on https://github.com/matthiasmullie/minify/issues
*
* @author Matthias Mullie <minify@mullie.eu>
* @copyright Copyright (c) 2012, Matthias Mullie. All rights reserved
* @license MIT License
*/
namespace VDM\Minify;
use VDM\Minify\Abstraction\Minify;
/**
* JavaScript Minifier Class.
*
* Please report bugs on https://github.com/matthiasmullie/minify/issues
*
* @author Matthias Mullie
* @author Tijs Verkoyen
* @copyright Copyright (c) 2012, Matthias Mullie. All rights reserved
* @license MIT License
*/
class JavaScript extends Minify
{
/**
* Var-matching regex based on http://stackoverflow.com/a/9337047/802993.
*
* Note that regular expressions using that bit must have the PCRE_UTF8
* pattern modifier (/u) set.
*
* @internal
*
* @var string
*/
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}
/**
* Full list of JavaScript reserved words.
* Will be loaded from /data/js/keywords_reserved.txt.
*
* @see https://mathiasbynens.be/notes/reserved-keywords
*
* @var string[]
*/
protected $keywordsReserved = [
'do',
'if',
'in',
'for',
'let',
'new',
'try',
'var',
'case',
'else',
'enum',
'eval',
'null',
'this',
'true',
'void',
'with',
'break',
'catch',
'class',
'const',
'false',
'super',
'throw',
'while',
'yield',
'delete',
'export',
'import',
'public',
'return',
'static',
'switch',
'typeof',
'default',
'extends',
'finally',
'package',
'private',
'continue',
'debugger',
'function',
'arguments',
'interface',
'protected',
'implements',
'instanceof',
'abstract',
'boolean',
'byte',
'char',
'double',
'final',
'float',
'goto',
'int',
'long',
'native',
'short',
'synchronized',
'throws',
'transient',
'volatile'
];
/**
* List of JavaScript reserved words that accept a <variable, value, ...>
* after them. Some end of lines are not the end of a statement, like with
* these keywords.
*
* E.g.: we shouldn't insert a ; after this else
* else
* console.log('this is quite fine')
*
* Will be loaded from /data/js/keywords_before.txt
*
* @var string[]
*/
protected $keywordsBefore = [
'do',
'in',
'let',
'new',
'var',
'case',
'else',
'enum',
'void',
'with',
'class',
'const',
'yield',
'delete',
'export',
'import',
'public',
'static',
'typeof',
'extends',
'package',
'private',
'function',
'protected',
'implements',
'instanceof'
];
/**
* List of JavaScript reserved words that accept a <variable, value, ...>
* before them. Some end of lines are not the end of a statement, like when
* continued by one of these keywords on the newline.
*
* E.g.: we shouldn't insert a ; before this instanceof
* variable
* instanceof String
*
* Will be loaded from /data/js/keywords_after.txt
*
* @var string[]
*/
protected $keywordsAfter = [
'in',
'public',
'extends',
'private',
'protected',
'implements',
'instanceof'
];
/**
* List of all JavaScript operators.
*
* Will be loaded from /data/js/operators.txt
*
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators
*
* @var string[]
*/
protected $operators = [
'+',
'-',
'*',
'/',
'%',
'=',
'+=',
'-=',
'*=',
'/=',
'%=',
'<<=',
'>>=',
'>>>=',
'&=',
'^=',
'|=',
'&',
'|',
'^',
'~',
'<<',
'>>',
'>>>',
'==',
'===',
'!=',
'!==',
'>',
'<',
'>=',
'<=',
'&&',
'||',
'!',
'.',
'[',
']',
'?',
':',
',',
';',
'(',
')',
'{',
'}'
];
/**
* List of JavaScript operators that accept a <variable, value, ...> after
* them. Some end of lines are not the end of a statement, like with these
* operators.
*
* Note: Most operators are fine, we've only removed ++ and --.
* ++ & -- have to be joined with the value they're in-/decrementing.
*
* Will be loaded from /data/js/operators_before.txt
*
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators
*
* @var string[]
*/
protected $operatorsBefore = [
'+',
'-',
'*',
'/',
'%',
'=',
'+=',
'-=',
'*=',
'/=',
'%=',
'<<=',
'>>=',
'>>>=',
'&=',
'^=',
'|=',
'&',
'|',
'^',
'~',
'<<',
'>>',
'>>>',
'==',
'===',
'!=',
'!==',
'>',
'<',
'>=',
'<=',
'&&',
'||',
'!',
'.',
'[',
'?',
':',
',',
';',
'(',
'{',
];
/**
* List of JavaScript operators that accept a <variable, value, ...> before
* them. Some end of lines are not the end of a statement, like when
* continued by one of these operators on the newline.
*
* Note: Most operators are fine, we've only removed ), ], ++, --, ! and ~.
* There can't be a newline separating ! or ~ and whatever it is negating.
* ++ & -- have to be joined with the value they're in-/decrementing.
* ) & ] are "special" in that they have lots or usecases. () for example
* is used for function calls, for grouping, in if () and for (), ...
*
* Will be loaded from /data/js/operators_after.txt
*
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Expressions_and_Operators
*
* @var string[]
*/
protected $operatorsAfter = [
'+',
'-',
'*',
'/',
'%',
'=',
'+=',
'-=',
'*=',
'/=',
'%=',
'<<=',
'>>=',
'>>>=',
'&=',
'^=',
'|=',
'&',
'|',
'^',
'<<',
'>>',
'>>>',
'==',
'===',
'!=',
'!==',
'>',
'<',
'>=',
'<=',
'&&',
'||',
'.',
'[',
']',
'?',
':',
',',
';',
'(',
')',
'}',
];
/**
* Minify the data.
* Perform JS optimizations.
*
* @param string[optional] $path Path to write the data to
*
* @return string The minified data
*/
public function execute($path = null)
{
$content = '';
/*
* Let's first take out strings, comments and regular expressions.
* All of these can contain JS code-like characters, and we should make
* sure any further magic ignores anything inside of these.
*
* Consider this example, where we should not strip any whitespace:
* var str = "a test";
*
* Comments will be removed altogether, strings and regular expressions
* will be replaced by placeholder text, which we'll restore later.
*/
$this->extractStrings('\'"`');
$this->stripComments();
$this->extractRegex();
// loop files
foreach ($this->data as $source => $js) {
// take out strings, comments & regex (for which we've registered
// the regexes just a few lines earlier)
$js = $this->replace($js);
$js = $this->propertyNotation($js);
$js = $this->shortenBools($js);
$js = $this->stripWhitespace($js);
// combine js: separating the scripts by a ;
$content .= $js . ';';
}
// clean up leftover `;`s from the combination of multiple scripts
$content = ltrim($content, ';');
$content = (string) substr($content, 0, -1);
/*
* Earlier, we extracted strings & regular expressions and replaced them
* with placeholder text. This will restore them.
*/
$content = $this->restoreExtractedData($content);
return $content;
}
/**
* Strip comments from source code.
*/
protected function stripComments()
{
// PHP only supports $this inside anonymous functions since 5.4
$minifier = $this;
$callback = function ($match) use ($minifier) {
if (
substr((string) $match[2], 0, 1) === '!' ||
strpos((string) $match[2], '@license') !== false ||
strpos((string) $match[2], '@preserve') !== false
) {
// preserve multi-line comments that start with /*!
// or contain @license or @preserve annotations
$count = count($minifier->extracted);
$placeholder = '/*' . $count . '*/';
$minifier->extracted[$placeholder] = $match[0];
return $match[1] . $placeholder . $match[3];
}
return $match[1] . $match[3];
};
// multi-line comments
$this->registerPattern('/(\n?)\/\*(.*?)\*\/(\n?)/s', $callback);
// single-line comments
$this->registerPattern('/\/\/.*$/m', '');
}
/**
* JS can have /-delimited regular expressions, like: /ab+c/.match(string).
*
* The content inside the regex can contain characters that may be confused
* for JS code: e.g. it could contain whitespace it needs to match & we
* don't want to strip whitespace in there.
*
* The regex can be pretty simple: we don't have to care about comments,
* (which also use slashes) because stripComments() will have stripped those
* already.
*
* This method will replace all string content with simple REGEX#
* placeholder text, so we've rid all regular expressions from characters
* that may be misinterpreted. Original regex content will be saved in
* $this->extracted and after doing all other minifying, we can restore the
* original content via restoreRegex()
*/
protected function extractRegex()
{
// PHP only supports $this inside anonymous functions since 5.4
$minifier = $this;
$callback = function ($match) use ($minifier) {
$count = count($minifier->extracted);
$placeholder = '"' . $count . '"';
$minifier->extracted[$placeholder] = $match[0];
return $placeholder;
};
// match all chars except `/` and `\`
// `\` is allowed though, along with whatever char follows (which is the
// one being escaped)
// this should allow all chars, except for an unescaped `/` (= the one
// closing the regex)
// then also ignore bare `/` inside `[]`, where they don't need to be
// escaped: anything inside `[]` can be ignored safely
$pattern = '\\/(?!\*)(?:[^\\[\\/\\\\\n\r]++|(?:\\\\.)++|(?:\\[(?:[^\\]\\\\\n\r]++|(?:\\\\.)++)++\\])++)++\\/[gimuy]*';
// a regular expression can only be followed by a few operators or some
// of the RegExp methods (a `\` followed by a variable or value is
// likely part of a division, not a regex)
$keywords = array('do', 'in', 'new', 'else', 'throw', 'yield', 'delete', 'return', 'typeof');
$before = '(^|[=:,;\+\-\*\?\/\}\(\{\[&\|!]|' . implode('|', $keywords) . ')\s*';
$propertiesAndMethods = array(
// https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp#Properties_2
'constructor',
'flags',
'global',
'ignoreCase',
'multiline',
'source',
'sticky',
'unicode',
// https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp#Methods_2
'compile(',
'exec(',
'test(',
'toSource(',
'toString(',
);
$delimiters = array_fill(0, count($propertiesAndMethods), '/');
$propertiesAndMethods = array_map('preg_quote', $propertiesAndMethods, $delimiters);
$after = '(?=\s*([\.,;:\)\}&\|+]|\/\/|$|\.(' . implode('|', $propertiesAndMethods) . ')))';
$this->registerPattern('/' . $before . '\K' . $pattern . $after . '/', $callback);
// regular expressions following a `)` are rather annoying to detect...
// quite often, `/` after `)` is a division operator & if it happens to
// be followed by another one (or a comment), it is likely to be
// confused for a regular expression
// however, it's perfectly possible for a regex to follow a `)`: after
// a single-line `if()`, `while()`, ... statement, for example
// since, when they occur like that, they're always the start of a
// statement, there's only a limited amount of ways they can be useful:
// by calling the regex methods directly
// if a regex following `)` is not followed by `.<property or method>`,
// it's quite likely not a regex
$before = '\)\s*';
$after = '(?=\s*\.(' . implode('|', $propertiesAndMethods) . '))';
$this->registerPattern('/' . $before . '\K' . $pattern . $after . '/', $callback);
// 1 more edge case: a regex can be followed by a lot more operators or
// keywords if there's a newline (ASI) in between, where the operator
// actually starts a new statement
// (https://github.com/matthiasmullie/minify/issues/56)
$operators = $this->getOperatorsForRegex($this->operatorsBefore, '/');
$operators += $this->getOperatorsForRegex($this->keywordsReserved, '/');
$after = '(?=\s*\n\s*(' . implode('|', $operators) . '))';
$this->registerPattern('/' . $pattern . $after . '/', $callback);
}
/**
* Strip whitespace.
*
* We won't strip *all* whitespace, but as much as possible. The thing that
* we'll preserve are newlines we're unsure about.
* JavaScript doesn't require statements to be terminated with a semicolon.
* It will automatically fix missing semicolons with ASI (automatic semi-
* colon insertion) at the end of line causing errors (without semicolon.)
*
* Because it's sometimes hard to tell if a newline is part of a statement
* that should be terminated or not, we'll just leave some of them alone.
*
* @param string $content The content to strip the whitespace for
*
* @return string
*/
protected function stripWhitespace($content)
{
// uniform line endings, make them all line feed
$content = str_replace(array("\r\n", "\r"), "\n", $content);
// collapse all non-line feed whitespace into a single space
$content = preg_replace('/[^\S\n]+/', ' ', $content);
// strip leading & trailing whitespace
$content = str_replace(array(" \n", "\n "), "\n", $content);
// collapse consecutive line feeds into just 1
$content = preg_replace('/\n+/', "\n", $content);
$operatorsBefore = $this->getOperatorsForRegex($this->operatorsBefore, '/');
$operatorsAfter = $this->getOperatorsForRegex($this->operatorsAfter, '/');
$operators = $this->getOperatorsForRegex($this->operators, '/');
$keywordsBefore = $this->getKeywordsForRegex($this->keywordsBefore, '/');
$keywordsAfter = $this->getKeywordsForRegex($this->keywordsAfter, '/');
// strip whitespace that ends in (or next line begin with) an operator
// that allows statements to be broken up over multiple lines
unset($operatorsBefore['+'], $operatorsBefore['-'], $operatorsAfter['+'], $operatorsAfter['-']);
$content = preg_replace(
array(
'/(' . implode('|', $operatorsBefore) . ')\s+/',
'/\s+(' . implode('|', $operatorsAfter) . ')/',
),
'\\1',
$content
);
// make sure + and - can't be mistaken for, or joined into ++ and --
$content = preg_replace(
array(
'/(?<![\+\-])\s*([\+\-])(?![\+\-])/',
'/(?<![\+\-])([\+\-])\s*(?![\+\-])/',
),
'\\1',
$content
);
// collapse whitespace around reserved words into single space
$content = preg_replace('/(^|[;\}\s])\K(' . implode('|', $keywordsBefore) . ')\s+/', '\\2 ', $content);
$content = preg_replace('/\s+(' . implode('|', $keywordsAfter) . ')(?=([;\{\s]|$))/', ' \\1', $content);
/*
* We didn't strip whitespace after a couple of operators because they
* could be used in different contexts and we can't be sure it's ok to
* strip the newlines. However, we can safely strip any non-line feed
* whitespace that follows them.
*/
$operatorsDiffBefore = array_diff($operators, $operatorsBefore);
$operatorsDiffAfter = array_diff($operators, $operatorsAfter);
$content = preg_replace('/(' . implode('|', $operatorsDiffBefore) . ')[^\S\n]+/', '\\1', $content);
$content = preg_replace('/[^\S\n]+(' . implode('|', $operatorsDiffAfter) . ')/', '\\1', $content);
/*
* Whitespace after `return` can be omitted in a few occasions
* (such as when followed by a string or regex)
* Same for whitespace in between `)` and `{`, or between `{` and some
* keywords.
*/
$content = preg_replace('/\breturn\s+(["\'\/\+\-])/', 'return$1', $content);
$content = preg_replace('/\)\s+\{/', '){', $content);
$content = preg_replace('/}\n(else|catch|finally)\b/', '}$1', $content);
/*
* Get rid of double semicolons, except where they can be used like:
* "for(v=1,_=b;;)", "for(v=1;;v++)" or "for(;;ja||(ja=true))".
* I'll safeguard these double semicolons inside for-loops by
* temporarily replacing them with an invalid condition: they won't have
* a double semicolon and will be easy to spot to restore afterwards.
*/
$content = preg_replace('/\bfor\(([^;]*);;([^;]*)\)/', 'for(\\1;-;\\2)', $content);
$content = preg_replace('/;+/', ';', $content);
$content = preg_replace('/\bfor\(([^;]*);-;([^;]*)\)/', 'for(\\1;;\\2)', $content);
/*
* Next, we'll be removing all semicolons where ASI kicks in.
* for-loops however, can have an empty body (ending in only a
* semicolon), like: `for(i=1;i<3;i++);`, of `for(i in list);`
* Here, nothing happens during the loop; it's just used to keep
* increasing `i`. With that ; omitted, the next line would be expected
* to be the for-loop's body... Same goes for while loops.
* I'm going to double that semicolon (if any) so after the next line,
* which strips semicolons here & there, we're still left with this one.
* Note the special recursive construct in the three inner parts of the for:
* (\{([^\{\}]*(?-2))*[^\{\}]*\})? - it is intended to match inline
* functions bodies, e.g.: i<arr.map(function(e){return e}).length.
* Also note that the construct is applied only once and multiplied
* for each part of the for, otherwise it risks a catastrophic backtracking.
* The limitation is that it will not allow closures in more than one
* of the three parts for a specific for() case.
* REGEX throwing catastrophic backtracking: $content = preg_replace('/(for\([^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*;[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*;[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*\));(\}|$)/s', '\\1;;\\8', $content);
*/
$content = preg_replace('/(for\((?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*);[^;\{]*;[^;\{]*\));(\}|$)/s', '\\1;;\\4', $content);
$content = preg_replace('/(for\([^;\{]*;(?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*);[^;\{]*\));(\}|$)/s', '\\1;;\\4', $content);
$content = preg_replace('/(for\([^;\{]*;[^;\{]*;(?:[^;\{]*|[^;\{]*function[^;\{]*(\{([^\{\}]*(?-2))*[^\{\}]*\})?[^;\{]*)\));(\}|$)/s', '\\1;;\\4', $content);
$content = preg_replace('/(for\([^;\{]+\s+in\s+[^;\{]+\));(\}|$)/s', '\\1;;\\2', $content);
/*
* Do the same for the if's that don't have a body but are followed by ;}
*/
$content = preg_replace('/(\bif\s*\([^{;]*\));\}/s', '\\1;;}', $content);
/*
* Below will also keep `;` after a `do{}while();` along with `while();`
* While these could be stripped after do-while, detecting this
* distinction is cumbersome, so I'll play it safe and make sure `;`
* after any kind of `while` is kept.
*/
$content = preg_replace('/(while\([^;\{]+\));(\}|$)/s', '\\1;;\\2', $content);
/*
* We also can't strip empty else-statements. Even though they're
* useless and probably shouldn't be in the code in the first place, we
* shouldn't be stripping the `;` that follows it as it breaks the code.
* We can just remove those useless else-statements completely.
*
* @see https://github.com/matthiasmullie/minify/issues/91
*/
$content = preg_replace('/else;/s', '', $content);
/*
* We also don't really want to terminate statements followed by closing
* curly braces (which we've ignored completely up until now) or end-of-
* script: ASI will kick in here & we're all about minifying.
* Semicolons at beginning of the file don't make any sense either.
*/
$content = preg_replace('/;(\}|$)/s', '\\1', $content);
$content = ltrim($content, ';');
// get rid of remaining whitespace af beginning/end
return trim($content);
}
/**
* We'll strip whitespace around certain operators with regular expressions.
* This will prepare the given array by escaping all characters.
*
* @param string[] $operators
* @param string $delimiter
*
* @return string[]
*/
protected function getOperatorsForRegex(array $operators, $delimiter = '/')
{
// escape operators for use in regex
$delimiters = array_fill(0, count($operators), $delimiter);
$escaped = array_map('preg_quote', $operators, $delimiters);
$operators = array_combine($operators, $escaped);
// ignore + & - for now, they'll get special treatment
unset($operators['+'], $operators['-']);
// dot can not just immediately follow a number; it can be confused for
// decimal point, or calling a method on it, e.g. 42 .toString()
$operators['.'] = '(?<![0-9]\s)\.';
// don't confuse = with other assignment shortcuts (e.g. +=)
$chars = preg_quote('+-*\=<>%&|', $delimiter);
$operators['='] = '(?<![' . $chars . '])\=';
return $operators;
}
/**
* We'll strip whitespace around certain keywords with regular expressions.
* This will prepare the given array by escaping all characters.
*
* @param string[] $keywords
* @param string $delimiter
*
* @return string[]
*/
protected function getKeywordsForRegex(array $keywords, $delimiter = '/')
{
// escape keywords for use in regex
$delimiter = array_fill(0, count($keywords), $delimiter);
$escaped = array_map('preg_quote', $keywords, $delimiter);
// add word boundaries
array_walk($keywords, fn($value) => '\b' . $value . '\b');
$keywords = array_combine($keywords, $escaped);
return $keywords;
}
/**
* Replaces all occurrences of array['key'] by array.key.
*
* @param string $content
*
* @return string
*/
protected function propertyNotation($content)
{
// PHP only supports $this inside anonymous functions since 5.4
$minifier = $this;
$keywords = $this->keywordsReserved;
$callback = function ($match) use ($minifier, $keywords) {
$property = trim($minifier->extracted[$match[1]], '\'"');
/*
* Check if the property is a reserved keyword. In this context (as
* property of an object literal/array) it shouldn't matter, but IE8
* freaks out with "Expected identifier".
*/
if (in_array($property, $keywords)) {
return $match[0];
}
/*
* See if the property is in a variable-like format (e.g.
* array['key-here'] can't be replaced by array.key-here since '-'
* is not a valid character there.
*/
if (!preg_match('/^' . $minifier::REGEX_VARIABLE . '$/u', $property)) {
return $match[0];
}
return '.' . $property;
};
/*
* Figure out if previous character is a variable name (of the array
* we want to use property notation on) - this is to make sure
* standalone ['value'] arrays aren't confused for keys-of-an-array.
* We can (and only have to) check the last character, because PHP's
* regex implementation doesn't allow unfixed-length look-behind
* assertions.
*/
preg_match('/(\[[^\]]+\])[^\]]*$/', static::REGEX_VARIABLE, $previousChar);
$previousChar = $previousChar[1];
/*
* Make sure word preceding the ['value'] is not a keyword, e.g.
* return['x']. Because -again- PHP's regex implementation doesn't allow
* unfixed-length look-behind assertions, I'm just going to do a lot of
* separate look-behind assertions, one for each keyword.
*/
$keywords = $this->getKeywordsForRegex($keywords);
$keywords = '(?<!' . implode(')(?<!', $keywords) . ')';
return preg_replace_callback('/(?<=' . $previousChar . '|\])' . $keywords . '\[\s*(([\'"])[0-9]+\\2)\s*\]/u', $callback, $content);
}
/**
* Replaces true & false by !0 and !1.
*
* @param string $content
*
* @return string
*/
protected function shortenBools($content)
{
/*
* 'true' or 'false' could be used as property names (which may be
* followed by whitespace) - we must not replace those!
* Since PHP doesn't allow variable-length (to account for the
* whitespace) lookbehind assertions, I need to capture the leading
* character and check if it's a `.`
*/
$callback = function ($match) {
if (trim((string) $match[1]) === '.') {
return $match[0];
}
return $match[1] . ($match[2] === 'true' ? '!0' : '!1');
};
$content = preg_replace_callback('/(^|.\s*)\b(true|false)\b(?!:)/', $callback, $content);
// for(;;) is exactly the same as while(true), but shorter :)
$content = preg_replace('/\bwhile\(!0\){/', 'for(;;){', $content);
// now make sure we didn't turn any do ... while(true) into do ... for(;;)
preg_match_all('/\bdo\b/', $content, $dos, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
// go backward to make sure positional offsets aren't altered when $content changes
$dos = array_reverse($dos);
foreach ($dos as $do) {
$offsetDo = $do[0][1];
// find all `while` (now `for`) following `do`: one of those must be
// associated with the `do` and be turned back into `while`
preg_match_all('/\bfor\(;;\)/', $content, $whiles, PREG_OFFSET_CAPTURE | PREG_SET_ORDER, $offsetDo);
foreach ($whiles as $while) {
$offsetWhile = $while[0][1];
$open = substr_count($content, '{', $offsetDo, $offsetWhile - $offsetDo);
$close = substr_count($content, '}', $offsetDo, $offsetWhile - $offsetDo);
if ($open === $close) {
// only restore `while` if amount of `{` and `}` are the same;
// otherwise, that `for` isn't associated with this `do`
$content = substr_replace($content, 'while(!0)', $offsetWhile, strlen('for(;;)'));
break;
}
}
}
return $content;
}
}