| 1 |
<?php |
| 2 |
|
| 3 |
namespace WPCoder\Optimization; |
| 4 |
|
| 5 |
defined( 'ABSPATH' ) || exit; |
| 6 |
|
| 7 |
class ParseMaster { |
| 8 |
public $ignoreCase = false; |
| 9 |
public $escapeChar = ''; |
| 10 |
|
| 11 |
// constants |
| 12 |
const EXPRESSION = 0; |
| 13 |
const REPLACEMENT = 1; |
| 14 |
const LENGTH = 2; |
| 15 |
|
| 16 |
// used to determine nesting levels |
| 17 |
private $GROUPS = '/\\(/'; //g |
| 18 |
private $SUB_REPLACE = '/\\$\\d/'; |
| 19 |
private $INDEXED = '/^\\$\\d+$/'; |
| 20 |
private $TRIM = '/([\'"])\\1\\.(.*)\\.\\1\\1$/'; |
| 21 |
private $ESCAPE = '/\\\./'; //g |
| 22 |
private $QUOTE = '/\'/'; |
| 23 |
private $DELETED = '/\\x01[^\\x01]*\\x01/'; //g |
| 24 |
|
| 25 |
public function add( $expression, $replacement = '' ) { |
| 26 |
// count the number of sub-expressions |
| 27 |
// - add one because each pattern is itself a sub-expression |
| 28 |
$length = 1 + preg_match_all( $this->GROUPS, $this->_internalEscape( (string) $expression ), $out ); |
| 29 |
|
| 30 |
// treat only strings $replacement |
| 31 |
if ( is_string( $replacement ) ) { |
| 32 |
// does the pattern deal with sub-expressions? |
| 33 |
if ( preg_match( $this->SUB_REPLACE, $replacement ) ) { |
| 34 |
// a simple lookup? (e.g. "$2") |
| 35 |
if ( preg_match( $this->INDEXED, $replacement ) ) { |
| 36 |
// store the index (used for fast retrieval of matched strings) |
| 37 |
$replacement = (int) ( substr( $replacement, 1 ) ) - 1; |
| 38 |
} else { // a complicated lookup (e.g. "Hello $2 $1") |
| 39 |
// build a function to do the lookup |
| 40 |
$quote = preg_match( $this->QUOTE, $this->_internalEscape( $replacement ) ) |
| 41 |
? '"' : "'"; |
| 42 |
$replacement = array( |
| 43 |
'fn' => '_backReferences', |
| 44 |
'data' => array( |
| 45 |
'replacement' => $replacement, |
| 46 |
'length' => $length, |
| 47 |
'quote' => $quote |
| 48 |
) |
| 49 |
); |
| 50 |
} |
| 51 |
} |
| 52 |
} |
| 53 |
// pass the modified arguments |
| 54 |
if ( ! empty( $expression ) ) { |
| 55 |
$this->_add( $expression, $replacement, $length ); |
| 56 |
} else { |
| 57 |
$this->_add( '/^$/', $replacement, $length ); |
| 58 |
} |
| 59 |
} |
| 60 |
|
| 61 |
public function exec( $string ) { |
| 62 |
// execute the global replacement |
| 63 |
$this->_escaped = array(); |
| 64 |
|
| 65 |
// simulate the _patterns.toSTring of Dean |
| 66 |
$regexp = '/'; |
| 67 |
foreach ( $this->_patterns as $reg ) { |
| 68 |
$regexp .= '(' . substr( $reg[ self::EXPRESSION ], 1, - 1 ) . ')|'; |
| 69 |
} |
| 70 |
$regexp = substr( $regexp, 0, - 1 ) . '/'; |
| 71 |
$regexp .= ( $this->ignoreCase ) ? 'i' : ''; |
| 72 |
|
| 73 |
$string = $this->_escape( $string, $this->escapeChar ); |
| 74 |
$string = preg_replace_callback( |
| 75 |
$regexp, |
| 76 |
array( |
| 77 |
&$this, |
| 78 |
'_replacement' |
| 79 |
), |
| 80 |
$string |
| 81 |
); |
| 82 |
$string = $this->_unescape( $string, $this->escapeChar ); |
| 83 |
|
| 84 |
return preg_replace( $this->DELETED, '', $string ); |
| 85 |
} |
| 86 |
|
| 87 |
public function reset() { |
| 88 |
// clear the patterns collection so that this object may be re-used |
| 89 |
$this->_patterns = array(); |
| 90 |
} |
| 91 |
|
| 92 |
// private |
| 93 |
private $_escaped = array(); // escaped characters |
| 94 |
private $_patterns = array(); // patterns stored by index |
| 95 |
|
| 96 |
// create and add a new pattern to the patterns collection |
| 97 |
private function _add() { |
| 98 |
$arguments = func_get_args(); |
| 99 |
$this->_patterns[] = $arguments; |
| 100 |
} |
| 101 |
|
| 102 |
// this is the global replace function (it's quite complicated) |
| 103 |
private function _replacement( $arguments ) { |
| 104 |
if ( empty( $arguments ) ) { |
| 105 |
return ''; |
| 106 |
} |
| 107 |
|
| 108 |
$i = 1; |
| 109 |
$j = 0; |
| 110 |
// loop through the patterns |
| 111 |
while ( isset( $this->_patterns[ $j ] ) ) { |
| 112 |
$pattern = $this->_patterns[ $j ++ ]; |
| 113 |
// do we have a result? |
| 114 |
if ( isset( $arguments[ $i ] ) && ( $arguments[ $i ] != '' ) ) { |
| 115 |
$replacement = $pattern[ self::REPLACEMENT ]; |
| 116 |
|
| 117 |
if ( is_array( $replacement ) && isset( $replacement['fn'] ) ) { |
| 118 |
|
| 119 |
if ( isset( $replacement['data'] ) ) { |
| 120 |
$this->buffer = $replacement['data']; |
| 121 |
} |
| 122 |
|
| 123 |
return call_user_func( array( &$this, $replacement['fn'] ), $arguments, $i ); |
| 124 |
|
| 125 |
} elseif ( is_int( $replacement ) ) { |
| 126 |
return $arguments[ $replacement + $i ]; |
| 127 |
|
| 128 |
} |
| 129 |
$delete = ( $this->escapeChar == '' |
| 130 |
|| strpos( $arguments[ $i ], $this->escapeChar ) === false ) |
| 131 |
? '' : "\x01" . $arguments[ $i ] . "\x01"; |
| 132 |
|
| 133 |
return $delete . $replacement; |
| 134 |
|
| 135 |
// skip over references to sub-expressions |
| 136 |
} else { |
| 137 |
$i += $pattern[ self::LENGTH ]; |
| 138 |
} |
| 139 |
} |
| 140 |
} |
| 141 |
|
| 142 |
private function _backReferences( $match, $offset ) { |
| 143 |
$replacement = $this->buffer['replacement']; |
| 144 |
$quote = $this->buffer['quote']; |
| 145 |
$i = $this->buffer['length']; |
| 146 |
while ( $i ) { |
| 147 |
$replacement = str_replace( '$' . $i --, $match[ $offset + $i ], $replacement ); |
| 148 |
} |
| 149 |
|
| 150 |
return $replacement; |
| 151 |
} |
| 152 |
|
| 153 |
private function _replace_name( $match, $offset ) { |
| 154 |
$length = strlen( $match[ $offset + 2 ] ); |
| 155 |
$start = $length - max( $length - strlen( $match[ $offset + 3 ] ), 0 ); |
| 156 |
|
| 157 |
return substr( $match[ $offset + 1 ], $start, $length ) . $match[ $offset + 4 ]; |
| 158 |
} |
| 159 |
|
| 160 |
private function _replace_encoded( $match, $offset ) { |
| 161 |
return $this->buffer[ $match[ $offset ] ]; |
| 162 |
} |
| 163 |
|
| 164 |
|
| 165 |
// php : we cannot pass additional data to preg_replace_callback, |
| 166 |
// and we cannot use &$this in create_function, so let's go to lower level |
| 167 |
private $buffer; |
| 168 |
|
| 169 |
// encode escaped characters |
| 170 |
private function _escape( $string, $escapeChar ) { |
| 171 |
if ( $escapeChar ) { |
| 172 |
$this->buffer = $escapeChar; |
| 173 |
|
| 174 |
return preg_replace_callback( |
| 175 |
'/\\' . $escapeChar . '(.)' . '/', |
| 176 |
array( &$this, '_escapeBis' ), |
| 177 |
$string |
| 178 |
); |
| 179 |
|
| 180 |
} else { |
| 181 |
return $string; |
| 182 |
} |
| 183 |
} |
| 184 |
|
| 185 |
private function _escapeBis( $match ) { |
| 186 |
$this->_escaped[] = $match[1]; |
| 187 |
|
| 188 |
return $this->buffer; |
| 189 |
} |
| 190 |
|
| 191 |
// decode escaped characters |
| 192 |
private function _unescape( $string, $escapeChar ) { |
| 193 |
if ( $escapeChar ) { |
| 194 |
$regexp = '/' . '\\' . $escapeChar . '/'; |
| 195 |
$this->buffer = array( 'escapeChar' => $escapeChar, 'i' => 0 ); |
| 196 |
|
| 197 |
return preg_replace_callback |
| 198 |
( |
| 199 |
$regexp, |
| 200 |
array( &$this, '_unescapeBis' ), |
| 201 |
$string |
| 202 |
); |
| 203 |
|
| 204 |
} else { |
| 205 |
return $string; |
| 206 |
} |
| 207 |
} |
| 208 |
|
| 209 |
private function _unescapeBis() { |
| 210 |
if ( isset( $this->_escaped[ $this->buffer['i'] ] ) |
| 211 |
&& $this->_escaped[ $this->buffer['i'] ] != '' |
| 212 |
) { |
| 213 |
$temp = $this->_escaped[ $this->buffer['i'] ]; |
| 214 |
} else { |
| 215 |
$temp = ''; |
| 216 |
} |
| 217 |
$this->buffer['i'] ++; |
| 218 |
|
| 219 |
return $this->buffer['escapeChar'] . $temp; |
| 220 |
} |
| 221 |
|
| 222 |
private function _internalEscape( $string ) { |
| 223 |
return preg_replace( $this->ESCAPE, '', $string ); |
| 224 |
} |
| 225 |
} |
