3 * A collection of static methods to play with strings.
7 * Perform an operation equivalent to
9 * preg_replace( "!$startDelim(.*?)$endDelim!", $replace, $subject );
11 * except that it's worst-case O(N) instead of O(N^2)
13 * Compared to delimiterReplace(), this implementation is fast but memory-
14 * hungry and inflexible. The memory requirements are such that I don't
15 * recommend using it on anything but guaranteed small chunks of text.
17 static function hungryDelimiterReplace( $startDelim, $endDelim, $replace, $subject ) {
18 $segments = explode( $startDelim, $subject );
19 $output = array_shift( $segments );
20 foreach ( $segments as $s ) {
21 $endDelimPos = strpos( $s, $endDelim );
22 if ( $endDelimPos === false ) {
23 $output .= $startDelim . $s;
25 $output .= $replace . substr( $s, $endDelimPos +
strlen( $endDelim ) );
32 * Perform an operation equivalent to
34 * preg_replace_callback( "!$startDelim(.*)$endDelim!s$flags", $callback, $subject )
36 * This implementation is slower than hungryDelimiterReplace but uses far less
37 * memory. The delimiters are literal strings, not regular expressions.
39 * @param string $flags Regular expression flags
41 # If the start delimiter ends with an initial substring of the end delimiter,
42 # e.g. in the case of C-style comments, the behaviour differs from the model
43 # regex. In this implementation, the end must share no characters with the
44 # start, so e.g. /*/ is not considered to be both the start and end of a
45 # comment. /*/xy/*/ is considered to be a single comment with contents /xy/.
46 static function delimiterReplaceCallback( $startDelim, $endDelim, $callback, $subject, $flags = '' ) {
51 $encStart = preg_quote( $startDelim, '!' );
52 $encEnd = preg_quote( $endDelim, '!' );
53 $strcmp = strpos( $flags, 'i' ) === false ?
'strcmp' : 'strcasecmp';
54 $endLength = strlen( $endDelim );
57 while ( $inputPos < strlen( $subject ) &&
58 preg_match( "!($encStart)|($encEnd)!S$flags", $subject, $m, PREG_OFFSET_CAPTURE
, $inputPos ) )
60 $tokenOffset = $m[0][1];
61 if ( $m[1][0] != '' ) {
63 $strcmp( $endDelim, substr( $subject, $tokenOffset, $endLength ) ) == 0 )
65 # An end match is present at the same location
67 $tokenLength = $endLength;
70 $tokenLength = strlen( $m[0][0] );
72 } elseif ( $m[2][0] != '' ) {
74 $tokenLength = strlen( $m[0][0] );
76 throw new MWException( 'Invalid delimiter given to ' . __METHOD__
);
79 if ( $tokenType == 'start' ) {
80 $inputPos = $tokenOffset +
$tokenLength;
81 # Only move the start position if we haven't already found a start
82 # This means that START START END matches outer pair
85 # Write out the non-matching section
86 $output .= substr( $subject, $outputPos, $tokenOffset - $outputPos );
87 $outputPos = $tokenOffset;
88 $contentPos = $inputPos;
91 } elseif ( $tokenType == 'end' ) {
94 $output .= call_user_func( $callback, array(
95 substr( $subject, $outputPos, $tokenOffset +
$tokenLength - $outputPos ),
96 substr( $subject, $contentPos, $tokenOffset - $contentPos )
100 # Non-matching end, write it out
101 $output .= substr( $subject, $inputPos, $tokenOffset +
$tokenLength - $outputPos );
103 $inputPos = $outputPos = $tokenOffset +
$tokenLength;
105 throw new MWException( 'Invalid delimiter given to ' . __METHOD__
);
108 if ( $outputPos < strlen( $subject ) ) {
109 $output .= substr( $subject, $outputPos );
115 * Perform an operation equivalent to
117 * preg_replace( "!$startDelim(.*)$endDelim!$flags", $replace, $subject )
119 * @param string $startDelim Start delimiter regular expression
120 * @param string $endDelim End delimiter regular expression
121 * @param string $replace Replacement string. May contain $1, which will be
122 * replaced by the text between the delimiters
123 * @param string $subject String to search
124 * @return string The string with the matches replaced
126 static function delimiterReplace( $startDelim, $endDelim, $replace, $subject, $flags = '' ) {
127 $replacer = new RegexlikeReplacer( $replace );
128 return self
::delimiterReplaceCallback( $startDelim, $endDelim,
129 $replacer->cb(), $subject, $flags );
133 * More or less "markup-safe" explode()
134 * Ignores any instances of the separator inside <...>
135 * @param string $separator
136 * @param string $text
139 static function explodeMarkup( $separator, $text ) {
140 $placeholder = "\x00";
142 // Remove placeholder instances
143 $text = str_replace( $placeholder, '', $text );
145 // Replace instances of the separator inside HTML-like tags with the placeholder
146 $replacer = new DoubleReplacer( $separator, $placeholder );
147 $cleaned = StringUtils
::delimiterReplaceCallback( '<', '>', $replacer->cb(), $text );
149 // Explode, then put the replaced separators back in
150 $items = explode( $separator, $cleaned );
151 foreach( $items as $i => $str ) {
152 $items[$i] = str_replace( $placeholder, $separator, $str );
159 * Escape a string to make it suitable for inclusion in a preg_replace()
160 * replacement parameter.
162 * @param string $string
165 static function escapeRegexReplacement( $string ) {
166 $string = str_replace( '\\', '\\\\', $string );
167 $string = str_replace( '$', '\\$', $string );
172 * Workalike for explode() with limited memory usage.
173 * Returns an Iterator
175 static function explode( $separator, $subject ) {
176 if ( substr_count( $subject, $separator ) > 1000 ) {
177 return new ExplodeIterator( $separator, $subject );
179 return new ArrayIterator( explode( $separator, $subject ) );
184 * Clean characters that are invalid in the given character set
185 * from a given string.
187 * @param $string \type{$string} String to clean
188 * @param $charset \type{$string} Character set (if unspecified, assume $wgOutputEncoding)
189 * @return \type{$string} Cleaned string
191 public static function cleanForCharset( $string, $charset='' ) {
192 global $wgOutputEncoding;
193 switch ( $charset ?
$charset : $wgOutputEncoding ) {
194 # UTF-8 should be all we need to worry about. :)
196 return self
::cleanUtf8( $string );
203 * Clean invalid UTF-8 characters and sequences from a given string,
204 * replacing them with U+FFFD.
205 * Should be RFC 3629 compliant.
207 * @param $string \type{$string} String to clean
208 * @return \type{$string} Cleaned string
210 private static function cleanUtf8( $str ) {
212 # ABANDON ALL HOPE, ALL YE WHO ENTER THE BITWISE HELLFIRE.
214 $illegal = array( 0xD800, 0xDB7F, 0xDB80, 0xDBFF,
215 0xDC00, 0xDF80, 0xDFFF, 0xFFFE, 0xFFFF );
216 $len = strlen( $str );
218 for ( $i = 0; $i < $len; $i++
) {
219 $ch = ord( $str[$i] );
221 if ( !($ch & 0x80 ) )
223 $left = (( $ch & 0xFE ) == 0xFC ?
5 :
224 (( $ch & 0xFC ) == 0xF8 ?
4 :
225 (( $ch & 0xF8 ) == 0xF0 ?
3 :
226 (( $ch & 0xF0 ) == 0xE0 ?
2 :
227 (( $ch & 0xE0 ) == 0xC0 ?
1 :
231 $sum = $ch & ( 0xFF >> $bytes +
1 );
233 } else if ( $ch & 0x80 ) {
236 } else if ( ( $ch & 0xC0 ) == 0x80 ) {
239 if ( --$left ) continue;
240 if ( ( $bytes == 2 && $sum < 0x80 ) ||
241 ( $bytes == 3 && $sum < 0x800 ) ||
242 ( $bytes == 4 && $sum < 0x10000 ) ||
243 ( $bytes > 4 ||
$sum > 0x10FFFF ) ||
244 in_array( $sum, $illegal ) ) {
252 $str = ( substr( $str, 0, $i - $bytes +
1 ) .
254 substr( $str, $i +
1 ) );
265 * Base class for "replacers", objects used in preg_replace_callback() and
266 * StringUtils::delimiterReplaceCallback()
270 return array( &$this, 'replace' );
275 * Class to replace regex matches with a string similar to that used in preg_replace()
277 class RegexlikeReplacer
extends Replacer
{
279 function __construct( $r ) {
283 function replace( $matches ) {
285 foreach ( $matches as $i => $match ) {
286 $pairs["\$$i"] = $match;
288 return strtr( $this->r
, $pairs );
294 * Class to perform secondary replacement within each replacement string
296 class DoubleReplacer
extends Replacer
{
297 function __construct( $from, $to, $index = 0 ) {
300 $this->index
= $index;
303 function replace( $matches ) {
304 return str_replace( $this->from
, $this->to
, $matches[$this->index
] );
309 * Class to perform replacement based on a simple hashtable lookup
311 class HashtableReplacer
extends Replacer
{
314 function __construct( $table, $index = 0 ) {
315 $this->table
= $table;
316 $this->index
= $index;
319 function replace( $matches ) {
320 return $this->table
[$matches[$this->index
]];
325 * Replacement array for FSS with fallback to strtr()
326 * Supports lazy initialisation of FSS resource
328 class ReplacementArray
{
329 /*mostly private*/ var $data = false;
330 /*mostly private*/ var $fss = false;
333 * Create an object with the specified replacement array
334 * The array should have the same form as the replacement array for strtr()
336 function __construct( $data = array() ) {
341 return array( 'data' );
344 function __wakeup() {
349 * Set the whole replacement array at once
351 function setArray( $data ) {
356 function getArray() {
361 * Set an element of the replacement array
363 function setPair( $from, $to ) {
364 $this->data
[$from] = $to;
368 function mergeArray( $data ) {
369 $this->data
= array_merge( $this->data
, $data );
373 function merge( $other ) {
374 $this->data
= array_merge( $this->data
, $other->data
);
378 function removePair( $from ) {
379 unset($this->data
[$from]);
383 function removeArray( $data ) {
384 foreach( $data as $from => $to )
385 $this->removePair( $from );
389 function replace( $subject ) {
390 if ( function_exists( 'fss_prep_replace' ) ) {
391 wfProfileIn( __METHOD__
.'-fss' );
392 if ( $this->fss
=== false ) {
393 $this->fss
= fss_prep_replace( $this->data
);
395 $result = fss_exec_replace( $this->fss
, $subject );
396 wfProfileOut( __METHOD__
.'-fss' );
398 wfProfileIn( __METHOD__
.'-strtr' );
399 $result = strtr( $subject, $this->data
);
400 wfProfileOut( __METHOD__
.'-strtr' );
407 * An iterator which works exactly like:
409 * foreach ( explode( $delim, $s ) as $element ) {
413 * Except it doesn't use 193 byte per element
415 class ExplodeIterator
implements Iterator
{
416 // The subject string
417 var $subject, $subjectLength;
420 var $delim, $delimLength;
422 // The position of the start of the line
425 // The position after the end of the next delimiter
432 * Construct a DelimIterator
434 function __construct( $delim, $s ) {
436 $this->delim
= $delim;
438 // Micro-optimisation (theoretical)
439 $this->subjectLength
= strlen( $s );
440 $this->delimLength
= strlen( $delim );
447 $this->endPos
= strpos( $this->subject
, $this->delim
);
448 $this->refreshCurrent();
452 function refreshCurrent() {
453 if ( $this->curPos
=== false ) {
454 $this->current
= false;
455 } elseif ( $this->curPos
>= $this->subjectLength
) {
457 } elseif ( $this->endPos
=== false ) {
458 $this->current
= substr( $this->subject
, $this->curPos
);
460 $this->current
= substr( $this->subject
, $this->curPos
, $this->endPos
- $this->curPos
);
465 return $this->current
;
469 return $this->curPos
;
473 if ( $this->endPos
=== false ) {
474 $this->curPos
= false;
476 $this->curPos
= $this->endPos +
$this->delimLength
;
477 if ( $this->curPos
>= $this->subjectLength
) {
478 $this->endPos
= false;
480 $this->endPos
= strpos( $this->subject
, $this->delim
, $this->curPos
);
483 $this->refreshCurrent();
484 return $this->current
;
488 return $this->curPos
!== false;