Return comment stuffs
[lhc/web/wiklou.git] / includes / StringUtils.php
1 <?php
2 /**
3 * A collection of static methods to play with strings.
4 */
5 class StringUtils {
6 /**
7 * Perform an operation equivalent to
8 *
9 * preg_replace( "!$startDelim(.*?)$endDelim!", $replace, $subject );
10 *
11 * except that it's worst-case O(N) instead of O(N^2)
12 *
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.
16 */
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;
24 } else {
25 $output .= $replace . substr( $s, $endDelimPos + strlen( $endDelim ) );
26 }
27 }
28 return $output;
29 }
30
31 /**
32 * Perform an operation equivalent to
33 *
34 * preg_replace_callback( "!$startDelim(.*)$endDelim!s$flags", $callback, $subject )
35 *
36 * This implementation is slower than hungryDelimiterReplace but uses far less
37 * memory. The delimiters are literal strings, not regular expressions.
38 *
39 * @param $startDelim String: start delimiter
40 * @param $endDelim String: end delimiter
41 * @param $callback Callback: function to call on each match
42 * @param $subject String
43 * @param $flags String: regular expression flags
44 */
45 # If the start delimiter ends with an initial substring of the end delimiter,
46 # e.g. in the case of C-style comments, the behaviour differs from the model
47 # regex. In this implementation, the end must share no characters with the
48 # start, so e.g. /*/ is not considered to be both the start and end of a
49 # comment. /*/xy/*/ is considered to be a single comment with contents /xy/.
50 static function delimiterReplaceCallback( $startDelim, $endDelim, $callback, $subject, $flags = '' ) {
51 $inputPos = 0;
52 $outputPos = 0;
53 $output = '';
54 $foundStart = false;
55 $encStart = preg_quote( $startDelim, '!' );
56 $encEnd = preg_quote( $endDelim, '!' );
57 $strcmp = strpos( $flags, 'i' ) === false ? 'strcmp' : 'strcasecmp';
58 $endLength = strlen( $endDelim );
59 $m = array();
60
61 while ( $inputPos < strlen( $subject ) &&
62 preg_match( "!($encStart)|($encEnd)!S$flags", $subject, $m, PREG_OFFSET_CAPTURE, $inputPos ) )
63 {
64 $tokenOffset = $m[0][1];
65 if ( $m[1][0] != '' ) {
66 if ( $foundStart &&
67 $strcmp( $endDelim, substr( $subject, $tokenOffset, $endLength ) ) == 0 )
68 {
69 # An end match is present at the same location
70 $tokenType = 'end';
71 $tokenLength = $endLength;
72 } else {
73 $tokenType = 'start';
74 $tokenLength = strlen( $m[0][0] );
75 }
76 } elseif ( $m[2][0] != '' ) {
77 $tokenType = 'end';
78 $tokenLength = strlen( $m[0][0] );
79 } else {
80 throw new MWException( 'Invalid delimiter given to ' . __METHOD__ );
81 }
82
83 if ( $tokenType == 'start' ) {
84 # Only move the start position if we haven't already found a start
85 # This means that START START END matches outer pair
86 if ( !$foundStart ) {
87 # Found start
88 $inputPos = $tokenOffset + $tokenLength;
89 # Write out the non-matching section
90 $output .= substr( $subject, $outputPos, $tokenOffset - $outputPos );
91 $outputPos = $tokenOffset;
92 $contentPos = $inputPos;
93 $foundStart = true;
94 } else {
95 # Move the input position past the *first character* of START,
96 # to protect against missing END when it overlaps with START
97 $inputPos = $tokenOffset + 1;
98 }
99 } elseif ( $tokenType == 'end' ) {
100 if ( $foundStart ) {
101 # Found match
102 $output .= call_user_func( $callback, array(
103 substr( $subject, $outputPos, $tokenOffset + $tokenLength - $outputPos ),
104 substr( $subject, $contentPos, $tokenOffset - $contentPos )
105 ));
106 $foundStart = false;
107 } else {
108 # Non-matching end, write it out
109 $output .= substr( $subject, $inputPos, $tokenOffset + $tokenLength - $outputPos );
110 }
111 $inputPos = $outputPos = $tokenOffset + $tokenLength;
112 } else {
113 throw new MWException( 'Invalid delimiter given to ' . __METHOD__ );
114 }
115 }
116 if ( $outputPos < strlen( $subject ) ) {
117 $output .= substr( $subject, $outputPos );
118 }
119 return $output;
120 }
121
122 /**
123 * Perform an operation equivalent to
124 *
125 * preg_replace( "!$startDelim(.*)$endDelim!$flags", $replace, $subject )
126 *
127 * @param $startDelim String: start delimiter regular expression
128 * @param $endDelim String: end delimiter regular expression
129 * @param $replace String: replacement string. May contain $1, which will be
130 * replaced by the text between the delimiters
131 * @param $subject String to search
132 * @param $flags String: regular expression flags
133 * @return String: The string with the matches replaced
134 */
135 static function delimiterReplace( $startDelim, $endDelim, $replace, $subject, $flags = '' ) {
136 $replacer = new RegexlikeReplacer( $replace );
137 return self::delimiterReplaceCallback( $startDelim, $endDelim,
138 $replacer->cb(), $subject, $flags );
139 }
140
141 /**
142 * More or less "markup-safe" explode()
143 * Ignores any instances of the separator inside <...>
144 * @param $separator String
145 * @param $text String
146 * @return array
147 */
148 static function explodeMarkup( $separator, $text ) {
149 $placeholder = "\x00";
150
151 // Remove placeholder instances
152 $text = str_replace( $placeholder, '', $text );
153
154 // Replace instances of the separator inside HTML-like tags with the placeholder
155 $replacer = new DoubleReplacer( $separator, $placeholder );
156 $cleaned = StringUtils::delimiterReplaceCallback( '<', '>', $replacer->cb(), $text );
157
158 // Explode, then put the replaced separators back in
159 $items = explode( $separator, $cleaned );
160 foreach( $items as $i => $str ) {
161 $items[$i] = str_replace( $placeholder, $separator, $str );
162 }
163
164 return $items;
165 }
166
167 /**
168 * Escape a string to make it suitable for inclusion in a preg_replace()
169 * replacement parameter.
170 *
171 * @param $string String
172 * @return String
173 */
174 static function escapeRegexReplacement( $string ) {
175 $string = str_replace( '\\', '\\\\', $string );
176 $string = str_replace( '$', '\\$', $string );
177 return $string;
178 }
179
180 /**
181 * Workalike for explode() with limited memory usage.
182 * Returns an Iterator
183 */
184 static function explode( $separator, $subject ) {
185 if ( substr_count( $subject, $separator ) > 1000 ) {
186 return new ExplodeIterator( $separator, $subject );
187 } else {
188 return new ArrayIterator( explode( $separator, $subject ) );
189 }
190 }
191 }
192
193 /**
194 * Base class for "replacers", objects used in preg_replace_callback() and
195 * StringUtils::delimiterReplaceCallback()
196 */
197 class Replacer {
198 function cb() {
199 return array( &$this, 'replace' );
200 }
201 }
202
203 /**
204 * Class to replace regex matches with a string similar to that used in preg_replace()
205 */
206 class RegexlikeReplacer extends Replacer {
207 var $r;
208 function __construct( $r ) {
209 $this->r = $r;
210 }
211
212 function replace( $matches ) {
213 $pairs = array();
214 foreach ( $matches as $i => $match ) {
215 $pairs["\$$i"] = $match;
216 }
217 return strtr( $this->r, $pairs );
218 }
219
220 }
221
222 /**
223 * Class to perform secondary replacement within each replacement string
224 */
225 class DoubleReplacer extends Replacer {
226 function __construct( $from, $to, $index = 0 ) {
227 $this->from = $from;
228 $this->to = $to;
229 $this->index = $index;
230 }
231
232 function replace( $matches ) {
233 return str_replace( $this->from, $this->to, $matches[$this->index] );
234 }
235 }
236
237 /**
238 * Class to perform replacement based on a simple hashtable lookup
239 */
240 class HashtableReplacer extends Replacer {
241 var $table, $index;
242
243 function __construct( $table, $index = 0 ) {
244 $this->table = $table;
245 $this->index = $index;
246 }
247
248 function replace( $matches ) {
249 return $this->table[$matches[$this->index]];
250 }
251 }
252
253 /**
254 * Replacement array for FSS with fallback to strtr()
255 * Supports lazy initialisation of FSS resource
256 */
257 class ReplacementArray {
258 /*mostly private*/ var $data = false;
259 /*mostly private*/ var $fss = false;
260
261 /**
262 * Create an object with the specified replacement array
263 * The array should have the same form as the replacement array for strtr()
264 */
265 function __construct( $data = array() ) {
266 $this->data = $data;
267 }
268
269 function __sleep() {
270 return array( 'data' );
271 }
272
273 function __wakeup() {
274 $this->fss = false;
275 }
276
277 /**
278 * Set the whole replacement array at once
279 */
280 function setArray( $data ) {
281 $this->data = $data;
282 $this->fss = false;
283 }
284
285 function getArray() {
286 return $this->data;
287 }
288
289 /**
290 * Set an element of the replacement array
291 */
292 function setPair( $from, $to ) {
293 $this->data[$from] = $to;
294 $this->fss = false;
295 }
296
297 function mergeArray( $data ) {
298 $this->data = array_merge( $this->data, $data );
299 $this->fss = false;
300 }
301
302 function merge( $other ) {
303 $this->data = array_merge( $this->data, $other->data );
304 $this->fss = false;
305 }
306
307 function removePair( $from ) {
308 unset($this->data[$from]);
309 $this->fss = false;
310 }
311
312 function removeArray( $data ) {
313 foreach( $data as $from => $to )
314 $this->removePair( $from );
315 $this->fss = false;
316 }
317
318 function replace( $subject ) {
319 if ( function_exists( 'fss_prep_replace' ) ) {
320 wfProfileIn( __METHOD__.'-fss' );
321 if ( $this->fss === false ) {
322 $this->fss = fss_prep_replace( $this->data );
323 }
324 $result = fss_exec_replace( $this->fss, $subject );
325 wfProfileOut( __METHOD__.'-fss' );
326 } else {
327 wfProfileIn( __METHOD__.'-strtr' );
328 $result = strtr( $subject, $this->data );
329 wfProfileOut( __METHOD__.'-strtr' );
330 }
331 return $result;
332 }
333 }
334
335 /**
336 * An iterator which works exactly like:
337 *
338 * foreach ( explode( $delim, $s ) as $element ) {
339 * ...
340 * }
341 *
342 * Except it doesn't use 193 byte per element
343 */
344 class ExplodeIterator implements Iterator {
345 // The subject string
346 var $subject, $subjectLength;
347
348 // The delimiter
349 var $delim, $delimLength;
350
351 // The position of the start of the line
352 var $curPos;
353
354 // The position after the end of the next delimiter
355 var $endPos;
356
357 // The current token
358 var $current;
359
360 /**
361 * Construct a DelimIterator
362 */
363 function __construct( $delim, $s ) {
364 $this->subject = $s;
365 $this->delim = $delim;
366
367 // Micro-optimisation (theoretical)
368 $this->subjectLength = strlen( $s );
369 $this->delimLength = strlen( $delim );
370
371 $this->rewind();
372 }
373
374 function rewind() {
375 $this->curPos = 0;
376 $this->endPos = strpos( $this->subject, $this->delim );
377 $this->refreshCurrent();
378 }
379
380
381 function refreshCurrent() {
382 if ( $this->curPos === false ) {
383 $this->current = false;
384 } elseif ( $this->curPos >= $this->subjectLength ) {
385 $this->current = '';
386 } elseif ( $this->endPos === false ) {
387 $this->current = substr( $this->subject, $this->curPos );
388 } else {
389 $this->current = substr( $this->subject, $this->curPos, $this->endPos - $this->curPos );
390 }
391 }
392
393 function current() {
394 return $this->current;
395 }
396
397 function key() {
398 return $this->curPos;
399 }
400
401 function next() {
402 if ( $this->endPos === false ) {
403 $this->curPos = false;
404 } else {
405 $this->curPos = $this->endPos + $this->delimLength;
406 if ( $this->curPos >= $this->subjectLength ) {
407 $this->endPos = false;
408 } else {
409 $this->endPos = strpos( $this->subject, $this->delim, $this->curPos );
410 }
411 }
412 $this->refreshCurrent();
413 return $this->current;
414 }
415
416 function valid() {
417 return $this->curPos !== false;
418 }
419 }
420