| 1 |
|
<?php |
| 2 |
|
/** |
| 3 |
|
* Xyster Framework |
| 4 |
|
* |
| 5 |
|
* This source file is subject to the new BSD license that is bundled |
| 6 |
|
* with this package in the file LICENSE.txt. |
| 7 |
|
* It is also available through the world-wide-web at this URL: |
| 8 |
|
* http://www.opensource.org/licenses/bsd-license.php |
| 9 |
|
* |
| 10 |
|
* @category Xyster |
| 11 |
|
* @package Xyster |
| 12 |
|
* @copyright Copyright (c) 2007-2008 Irrational Logic (http://irrationallogic.net) |
| 13 |
|
* @license http://www.opensource.org/licenses/bsd-license.php New BSD License |
| 14 |
|
* @version $Id: String.php 199 2008-01-19 21:54:40Z doublecompile $ |
| 15 |
|
*/ |
| 16 |
|
/** |
| 17 |
|
* Helper class for string manipulation |
| 18 |
|
* |
| 19 |
|
* @category Xyster |
| 20 |
|
* @package Xyster |
| 21 |
|
* @copyright Copyright (c) 2007-2008 Irrational Logic (http://irrationallogic.net) |
| 22 |
|
* @license http://www.opensource.org/licenses/bsd-license.php New BSD License |
| 23 |
|
*/ |
| 24 |
|
class Xyster_String |
| 25 |
|
{ |
| 26 |
|
/** |
| 27 |
|
* Used for matching parenthesis groups not inside quotes |
| 28 |
|
* |
| 29 |
|
* It works, but this still matches parentheses inside quotes that aren't in |
| 30 |
|
* parentheses. |
| 31 |
|
*/ |
| 32 |
|
const PARENTH_QUOTE_REGEX = '/(?<![\w\d])(\(((?:"[^"\\\\]*(?:\\\\.[^"\\\\]*)*"|[^()])|(?1))+\))/m'; |
| 33 |
|
|
| 34 |
|
/** |
| 35 |
|
* Match nested parentheses groups not inside of double-quotes |
| 36 |
|
* |
| 37 |
|
* This method will find all balanced parentheses not inside of double |
| 38 |
|
* quotes. It also respects escaped double quotes (i.e. with a preceding |
| 39 |
|
* backslash). |
| 40 |
|
* |
| 41 |
|
* For example, if you ran this method on a string containing JavaScript |
| 42 |
|
* code: |
| 43 |
|
* <code> |
| 44 |
|
* $string = <<<HEREDOC |
| 45 |
|
* var n = navigator; |
| 46 |
|
* return ( n.appName == "Netscape" ) || n.appVersion != "5.0 (Windows; en-US)"; |
| 47 |
|
* HEREDOC; |
| 48 |
|
* </code> |
| 49 |
|
* |
| 50 |
|
* You would receive back an array containing: |
| 51 |
|
* <pre> |
| 52 |
|
* array ( |
| 53 |
|
* [0]=>( n.appName == "Netscape" ) |
| 54 |
|
* ); |
| 55 |
|
* </pre> |
| 56 |
|
* |
| 57 |
|
* The second matched parentheses are inside of quotation marks and thus |
| 58 |
|
* should be treated as part of that string literal. |
| 59 |
|
* |
| 60 |
|
* @param string $string |
| 61 |
|
* @return array |
| 62 |
|
* @todo make this work with top-level strings with parenths in them |
| 63 |
|
*/ |
| 64 |
|
static public function matchGroups( $string ) |
| 65 |
|
{ |
| 66 |
55 |
if ( strpos($string, '(') === false ) { |
| 67 |
54 |
return array(); |
| 68 |
0 |
} |
| 69 |
|
|
| 70 |
38 |
$matches = array(); |
| 71 |
38 |
preg_match_all(self::PARENTH_QUOTE_REGEX, $string, $matches, PREG_SET_ORDER); |
| 72 |
|
|
| 73 |
38 |
$groups = array(); |
| 74 |
38 |
foreach( $matches as $group ) { |
| 75 |
36 |
$groups[] = $group[0]; |
| 76 |
36 |
} |
| 77 |
38 |
return $groups; |
| 78 |
|
} |
| 79 |
|
|
| 80 |
|
/** |
| 81 |
|
* Like explode(), but won't split inside of parentheses or double-quotes |
| 82 |
|
* |
| 83 |
|
* This method will split a string into an array using a string as a |
| 84 |
|
* seperator. If the seperator is contained within a pair of double-quotes |
| 85 |
|
* or between matching parentheses, it will be ignored. If the seperator is |
| 86 |
|
* not found, the method will return an array with one element containing |
| 87 |
|
* the entire string. |
| 88 |
|
* |
| 89 |
|
* Example: |
| 90 |
|
* <code> |
| 91 |
|
* $haystack = 'A (great (test)) of "this method"'; |
| 92 |
|
* print_r(Xyster_String::smartSplit(' ', $haystack)); |
| 93 |
|
* </code> |
| 94 |
|
* |
| 95 |
|
* Would print: |
| 96 |
|
* <pre> |
| 97 |
|
* array ( |
| 98 |
|
* [0]=>A |
| 99 |
|
* [1]=>(great (test)) |
| 100 |
|
* [2]=>of |
| 101 |
|
* [3]=>"this method" |
| 102 |
|
* ) |
| 103 |
|
* </pre> |
| 104 |
|
* |
| 105 |
|
* @param string $needle String used as seperator |
| 106 |
|
* @param string $haystack String to split |
| 107 |
|
* @param bool $caseInsensitive Whether to ignore case |
| 108 |
|
* @return array Split parts of $haystack |
| 109 |
|
*/ |
| 110 |
|
static public function smartSplit( $needle, $haystack, $caseInsensitive=true ) |
| 111 |
|
{ |
| 112 |
82 |
$split = array(); |
| 113 |
82 |
$buff = ""; |
| 114 |
82 |
$inPar = 0; |
| 115 |
82 |
$inStr = false; |
| 116 |
82 |
for( $i=0; $i<strlen($haystack); $i++ ) { |
| 117 |
82 |
$currNeed = substr( $haystack, $i, strlen($needle) ); |
| 118 |
82 |
$curr = $currNeed[0]; |
| 119 |
82 |
$last = ( $i ) ? $haystack[$i-1] : ""; |
| 120 |
82 |
if ( $curr == '"' && ( ( $inStr && $last != "\\") || !$inStr ) ) { |
| 121 |
23 |
$inStr = !$inStr; |
| 122 |
23 |
} |
| 123 |
82 |
if ( !$inStr ) { |
| 124 |
82 |
if ( $curr == "(" ) { |
| 125 |
24 |
$inPar++; |
| 126 |
82 |
} else if ( $curr == ")" ) { |
| 127 |
24 |
$inPar--; |
| 128 |
24 |
} |
| 129 |
82 |
} |
| 130 |
82 |
if ( !$inPar && !$inStr && |
| 131 |
82 |
( $currNeed == $needle || |
| 132 |
82 |
( $caseInsensitive && strcasecmp($currNeed,$needle) == 0 ) ) ) { |
| 133 |
63 |
$split[] = $buff; |
| 134 |
63 |
$buff = ""; |
| 135 |
63 |
$i = $i+(strlen($needle)-1); |
| 136 |
63 |
} else { |
| 137 |
82 |
$buff .= $curr; |
| 138 |
|
} |
| 139 |
82 |
} |
| 140 |
82 |
$split[] = $buff; |
| 141 |
82 |
return $split; |
| 142 |
|
} |
| 143 |
|
} |