22f5fdf810
https://github.com/tedivm/JShrink Signed-off-by: Anders Kaseorg <andersk@mit.edu>
587 lines
18 KiB
PHP
587 lines
18 KiB
PHP
<?php
|
|
/*
|
|
* This file is part of the JShrink package.
|
|
*
|
|
* (c) Robert Hafner <tedivm@tedivm.com>
|
|
*
|
|
* For the full copyright and license information, please view the LICENSE
|
|
* file that was distributed with this source code.
|
|
*/
|
|
|
|
/**
|
|
* JShrink
|
|
*
|
|
*
|
|
* @package JShrink
|
|
* @author Robert Hafner <tedivm@tedivm.com>
|
|
*/
|
|
|
|
namespace JShrink;
|
|
|
|
/**
|
|
* Minifier
|
|
*
|
|
* Usage - Minifier::minify($js);
|
|
* Usage - Minifier::minify($js, $options);
|
|
* Usage - Minifier::minify($js, array('flaggedComments' => false));
|
|
*
|
|
* @package JShrink
|
|
* @author Robert Hafner <tedivm@tedivm.com>
|
|
* @license http://www.opensource.org/licenses/bsd-license.php BSD License
|
|
*/
|
|
class Minifier
|
|
{
|
|
/**
|
|
* The input javascript to be minified.
|
|
*
|
|
* @var string
|
|
*/
|
|
protected $input;
|
|
|
|
/**
|
|
* The location of the character (in the input string) that is next to be
|
|
* processed.
|
|
*
|
|
* @var int
|
|
*/
|
|
protected $index = 0;
|
|
|
|
/**
|
|
* The first of the characters currently being looked at.
|
|
*
|
|
* @var string
|
|
*/
|
|
protected $a = '';
|
|
|
|
/**
|
|
* The next character being looked at (after a);
|
|
*
|
|
* @var string
|
|
*/
|
|
protected $b = '';
|
|
|
|
/**
|
|
* This character is only active when certain look ahead actions take place.
|
|
*
|
|
* @var string
|
|
*/
|
|
protected $c;
|
|
|
|
/**
|
|
* Contains the options for the current minification process.
|
|
*
|
|
* @var array
|
|
*/
|
|
protected $options;
|
|
|
|
/**
|
|
* Contains the default options for minification. This array is merged with
|
|
* the one passed in by the user to create the request specific set of
|
|
* options (stored in the $options attribute).
|
|
*
|
|
* @var array
|
|
*/
|
|
protected static $defaultOptions = array('flaggedComments' => true);
|
|
|
|
/**
|
|
* Contains lock ids which are used to replace certain code patterns and
|
|
* prevent them from being minified
|
|
*
|
|
* @var array
|
|
*/
|
|
protected $locks = array();
|
|
|
|
/**
|
|
* Takes a string containing javascript and removes unneeded characters in
|
|
* order to shrink the code without altering it's functionality.
|
|
*
|
|
* @param string $js The raw javascript to be minified
|
|
* @param array $options Various runtime options in an associative array
|
|
* @throws \Exception
|
|
* @return bool|string
|
|
*/
|
|
public static function minify($js, $options = array())
|
|
{
|
|
try {
|
|
ob_start();
|
|
|
|
$jshrink = new Minifier();
|
|
$js = $jshrink->lock($js);
|
|
$jshrink->minifyDirectToOutput($js, $options);
|
|
|
|
// Sometimes there's a leading new line, so we trim that out here.
|
|
$js = ltrim(ob_get_clean());
|
|
$js = $jshrink->unlock($js);
|
|
unset($jshrink);
|
|
|
|
return $js;
|
|
|
|
} catch (\Exception $e) {
|
|
|
|
if (isset($jshrink)) {
|
|
// Since the breakdownScript function probably wasn't finished
|
|
// we clean it out before discarding it.
|
|
$jshrink->clean();
|
|
unset($jshrink);
|
|
}
|
|
|
|
// without this call things get weird, with partially outputted js.
|
|
ob_end_clean();
|
|
throw $e;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Processes a javascript string and outputs only the required characters,
|
|
* stripping out all unneeded characters.
|
|
*
|
|
* @param string $js The raw javascript to be minified
|
|
* @param array $options Various runtime options in an associative array
|
|
*/
|
|
protected function minifyDirectToOutput($js, $options)
|
|
{
|
|
$this->initialize($js, $options);
|
|
$this->loop();
|
|
$this->clean();
|
|
}
|
|
|
|
/**
|
|
* Initializes internal variables, normalizes new lines,
|
|
*
|
|
* @param string $js The raw javascript to be minified
|
|
* @param array $options Various runtime options in an associative array
|
|
*/
|
|
protected function initialize($js, $options)
|
|
{
|
|
$this->options = array_merge(static::$defaultOptions, $options);
|
|
$js = str_replace("\r\n", "\n", $js);
|
|
$js = str_replace('/**/', '', $js);
|
|
$this->input = str_replace("\r", "\n", $js);
|
|
|
|
// We add a newline to the end of the script to make it easier to deal
|
|
// with comments at the bottom of the script- this prevents the unclosed
|
|
// comment error that can otherwise occur.
|
|
$this->input .= PHP_EOL;
|
|
|
|
// Populate "a" with a new line, "b" with the first character, before
|
|
// entering the loop
|
|
$this->a = "\n";
|
|
$this->b = $this->getReal();
|
|
}
|
|
|
|
/**
|
|
* The primary action occurs here. This function loops through the input string,
|
|
* outputting anything that's relevant and discarding anything that is not.
|
|
*/
|
|
protected function loop()
|
|
{
|
|
while ($this->a !== false && !is_null($this->a) && $this->a !== '') {
|
|
|
|
switch ($this->a) {
|
|
// new lines
|
|
case "\n":
|
|
// if the next line is something that can't stand alone preserve the newline
|
|
if (strpos('(-+{[@', $this->b) !== false) {
|
|
echo $this->a;
|
|
$this->saveString();
|
|
break;
|
|
}
|
|
|
|
// if B is a space we skip the rest of the switch block and go down to the
|
|
// string/regex check below, resetting $this->b with getReal
|
|
if($this->b === ' ')
|
|
break;
|
|
|
|
// otherwise we treat the newline like a space
|
|
|
|
case ' ':
|
|
if(static::isAlphaNumeric($this->b))
|
|
echo $this->a;
|
|
|
|
$this->saveString();
|
|
break;
|
|
|
|
default:
|
|
switch ($this->b) {
|
|
case "\n":
|
|
if (strpos('}])+-"\'', $this->a) !== false) {
|
|
echo $this->a;
|
|
$this->saveString();
|
|
break;
|
|
} else {
|
|
if (static::isAlphaNumeric($this->a)) {
|
|
echo $this->a;
|
|
$this->saveString();
|
|
}
|
|
}
|
|
break;
|
|
|
|
case ' ':
|
|
if(!static::isAlphaNumeric($this->a))
|
|
break;
|
|
|
|
default:
|
|
// check for some regex that breaks stuff
|
|
if ($this->a === '/' && ($this->b === '\'' || $this->b === '"')) {
|
|
$this->saveRegex();
|
|
continue;
|
|
}
|
|
|
|
echo $this->a;
|
|
$this->saveString();
|
|
break;
|
|
}
|
|
}
|
|
|
|
// do reg check of doom
|
|
$this->b = $this->getReal();
|
|
|
|
if(($this->b == '/' && strpos('(,=:[!&|?', $this->a) !== false))
|
|
$this->saveRegex();
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Resets attributes that do not need to be stored between requests so that
|
|
* the next request is ready to go. Another reason for this is to make sure
|
|
* the variables are cleared and are not taking up memory.
|
|
*/
|
|
protected function clean()
|
|
{
|
|
unset($this->input);
|
|
$this->index = 0;
|
|
$this->a = $this->b = '';
|
|
unset($this->c);
|
|
unset($this->options);
|
|
}
|
|
|
|
/**
|
|
* Returns the next string for processing based off of the current index.
|
|
*
|
|
* @return string
|
|
*/
|
|
protected function getChar()
|
|
{
|
|
// Check to see if we had anything in the look ahead buffer and use that.
|
|
if (isset($this->c)) {
|
|
$char = $this->c;
|
|
unset($this->c);
|
|
|
|
// Otherwise we start pulling from the input.
|
|
} else {
|
|
$char = substr($this->input, $this->index, 1);
|
|
|
|
// If the next character doesn't exist return false.
|
|
if (isset($char) && $char === false) {
|
|
return false;
|
|
}
|
|
|
|
// Otherwise increment the pointer and use this char.
|
|
$this->index++;
|
|
}
|
|
|
|
// Normalize all whitespace except for the newline character into a
|
|
// standard space.
|
|
if($char !== "\n" && ord($char) < 32)
|
|
|
|
return ' ';
|
|
|
|
return $char;
|
|
}
|
|
|
|
/**
|
|
* This function gets the next "real" character. It is essentially a wrapper
|
|
* around the getChar function that skips comments. This has significant
|
|
* performance benefits as the skipping is done using native functions (ie,
|
|
* c code) rather than in script php.
|
|
*
|
|
*
|
|
* @return string Next 'real' character to be processed.
|
|
* @throws \RuntimeException
|
|
*/
|
|
protected function getReal()
|
|
{
|
|
$startIndex = $this->index;
|
|
$char = $this->getChar();
|
|
|
|
// Check to see if we're potentially in a comment
|
|
if ($char !== '/') {
|
|
return $char;
|
|
}
|
|
|
|
$this->c = $this->getChar();
|
|
|
|
if ($this->c === '/') {
|
|
return $this->processOneLineComments($startIndex);
|
|
|
|
} elseif ($this->c === '*') {
|
|
return $this->processMultiLineComments($startIndex);
|
|
}
|
|
|
|
return $char;
|
|
}
|
|
|
|
/**
|
|
* Removed one line comments, with the exception of some very specific types of
|
|
* conditional comments.
|
|
*
|
|
* @param int $startIndex The index point where "getReal" function started
|
|
* @return string
|
|
*/
|
|
protected function processOneLineComments($startIndex)
|
|
{
|
|
$thirdCommentString = substr($this->input, $this->index, 1);
|
|
|
|
// kill rest of line
|
|
$this->getNext("\n");
|
|
|
|
if ($thirdCommentString == '@') {
|
|
$endPoint = $this->index - $startIndex;
|
|
unset($this->c);
|
|
$char = "\n" . substr($this->input, $startIndex, $endPoint);
|
|
} else {
|
|
// first one is contents of $this->c
|
|
$this->getChar();
|
|
$char = $this->getChar();
|
|
}
|
|
|
|
return $char;
|
|
}
|
|
|
|
/**
|
|
* Skips multiline comments where appropriate, and includes them where needed.
|
|
* Conditional comments and "license" style blocks are preserved.
|
|
*
|
|
* @param int $startIndex The index point where "getReal" function started
|
|
* @return bool|string False if there's no character
|
|
* @throws \RuntimeException Unclosed comments will throw an error
|
|
*/
|
|
protected function processMultiLineComments($startIndex)
|
|
{
|
|
$this->getChar(); // current C
|
|
$thirdCommentString = $this->getChar();
|
|
|
|
// kill everything up to the next */ if it's there
|
|
if ($this->getNext('*/')) {
|
|
|
|
$this->getChar(); // get *
|
|
$this->getChar(); // get /
|
|
$char = $this->getChar(); // get next real character
|
|
|
|
// Now we reinsert conditional comments and YUI-style licensing comments
|
|
if (($this->options['flaggedComments'] && $thirdCommentString === '!')
|
|
|| ($thirdCommentString === '@') ) {
|
|
|
|
// If conditional comments or flagged comments are not the first thing in the script
|
|
// we need to echo a and fill it with a space before moving on.
|
|
if ($startIndex > 0) {
|
|
echo $this->a;
|
|
$this->a = " ";
|
|
|
|
// If the comment started on a new line we let it stay on the new line
|
|
if ($this->input[($startIndex - 1)] === "\n") {
|
|
echo "\n";
|
|
}
|
|
}
|
|
|
|
$endPoint = ($this->index - 1) - $startIndex;
|
|
echo substr($this->input, $startIndex, $endPoint);
|
|
|
|
return $char;
|
|
}
|
|
|
|
} else {
|
|
$char = false;
|
|
}
|
|
|
|
if($char === false)
|
|
throw new \RuntimeException('Unclosed multiline comment at position: ' . ($this->index - 2));
|
|
|
|
// if we're here c is part of the comment and therefore tossed
|
|
if(isset($this->c))
|
|
unset($this->c);
|
|
|
|
return $char;
|
|
}
|
|
|
|
/**
|
|
* Pushes the index ahead to the next instance of the supplied string. If it
|
|
* is found the first character of the string is returned and the index is set
|
|
* to it's position.
|
|
*
|
|
* @param string $string
|
|
* @return string|false Returns the first character of the string or false.
|
|
*/
|
|
protected function getNext($string)
|
|
{
|
|
// Find the next occurrence of "string" after the current position.
|
|
$pos = strpos($this->input, $string, $this->index);
|
|
|
|
// If it's not there return false.
|
|
if($pos === false)
|
|
|
|
return false;
|
|
|
|
// Adjust position of index to jump ahead to the asked for string
|
|
$this->index = $pos;
|
|
|
|
// Return the first character of that string.
|
|
return substr($this->input, $this->index, 1);
|
|
}
|
|
|
|
/**
|
|
* When a javascript string is detected this function crawls for the end of
|
|
* it and saves the whole string.
|
|
*
|
|
* @throws \RuntimeException Unclosed strings will throw an error
|
|
*/
|
|
protected function saveString()
|
|
{
|
|
$startpos = $this->index;
|
|
|
|
// saveString is always called after a gets cleared, so we push b into
|
|
// that spot.
|
|
$this->a = $this->b;
|
|
|
|
// If this isn't a string we don't need to do anything.
|
|
if ($this->a !== "'" && $this->a !== '"') {
|
|
return;
|
|
}
|
|
|
|
// String type is the quote used, " or '
|
|
$stringType = $this->a;
|
|
|
|
// Echo out that starting quote
|
|
echo $this->a;
|
|
|
|
// Loop until the string is done
|
|
while (true) {
|
|
|
|
// Grab the very next character and load it into a
|
|
$this->a = $this->getChar();
|
|
|
|
switch ($this->a) {
|
|
|
|
// If the string opener (single or double quote) is used
|
|
// output it and break out of the while loop-
|
|
// The string is finished!
|
|
case $stringType:
|
|
break 2;
|
|
|
|
// New lines in strings without line delimiters are bad- actual
|
|
// new lines will be represented by the string \n and not the actual
|
|
// character, so those will be treated just fine using the switch
|
|
// block below.
|
|
case "\n":
|
|
throw new \RuntimeException('Unclosed string at position: ' . $startpos );
|
|
break;
|
|
|
|
// Escaped characters get picked up here. If it's an escaped new line it's not really needed
|
|
case '\\':
|
|
|
|
// a is a slash. We want to keep it, and the next character,
|
|
// unless it's a new line. New lines as actual strings will be
|
|
// preserved, but escaped new lines should be reduced.
|
|
$this->b = $this->getChar();
|
|
|
|
// If b is a new line we discard a and b and restart the loop.
|
|
if ($this->b === "\n") {
|
|
break;
|
|
}
|
|
|
|
// echo out the escaped character and restart the loop.
|
|
echo $this->a . $this->b;
|
|
break;
|
|
|
|
|
|
// Since we're not dealing with any special cases we simply
|
|
// output the character and continue our loop.
|
|
default:
|
|
echo $this->a;
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* When a regular expression is detected this function crawls for the end of
|
|
* it and saves the whole regex.
|
|
*
|
|
* @throws \RuntimeException Unclosed regex will throw an error
|
|
*/
|
|
protected function saveRegex()
|
|
{
|
|
echo $this->a . $this->b;
|
|
|
|
while (($this->a = $this->getChar()) !== false) {
|
|
if($this->a === '/')
|
|
break;
|
|
|
|
if ($this->a === '\\') {
|
|
echo $this->a;
|
|
$this->a = $this->getChar();
|
|
}
|
|
|
|
if($this->a === "\n")
|
|
throw new \RuntimeException('Unclosed regex pattern at position: ' . $this->index);
|
|
|
|
echo $this->a;
|
|
}
|
|
$this->b = $this->getReal();
|
|
}
|
|
|
|
/**
|
|
* Checks to see if a character is alphanumeric.
|
|
*
|
|
* @param string $char Just one character
|
|
* @return bool
|
|
*/
|
|
protected static function isAlphaNumeric($char)
|
|
{
|
|
return preg_match('/^[\w\$\pL]$/', $char) === 1 || $char == '/';
|
|
}
|
|
|
|
/**
|
|
* Replace patterns in the given string and store the replacement
|
|
*
|
|
* @param string $js The string to lock
|
|
* @return bool
|
|
*/
|
|
protected function lock($js)
|
|
{
|
|
/* lock things like <code>"asd" + ++x;</code> */
|
|
$lock = '"LOCK---' . crc32(time()) . '"';
|
|
|
|
$matches = array();
|
|
preg_match('/([+-])(\s+)([+-])/S', $js, $matches);
|
|
if (empty($matches)) {
|
|
return $js;
|
|
}
|
|
|
|
$this->locks[$lock] = $matches[2];
|
|
|
|
$js = preg_replace('/([+-])\s+([+-])/S', "$1{$lock}$2", $js);
|
|
/* -- */
|
|
|
|
return $js;
|
|
}
|
|
|
|
/**
|
|
* Replace "locks" with the original characters
|
|
*
|
|
* @param string $js The string to unlock
|
|
* @return bool
|
|
*/
|
|
protected function unlock($js)
|
|
{
|
|
if (empty($this->locks)) {
|
|
return $js;
|
|
}
|
|
|
|
foreach ($this->locks as $lock => $replacement) {
|
|
$js = str_replace($lock, $replacement, $js);
|
|
}
|
|
|
|
return $js;
|
|
}
|
|
|
|
}
|