| Index: trunk/phase3/tests/preg_split_test.php |
| — | — | @@ -0,0 +1,24 @@ |
| | 2 | +<?php |
| | 3 | +include "../includes/StringUtils.php"; |
| | 4 | + |
| | 5 | +$pattern = "/('')+/"; |
| | 6 | +$subject = str_repeat("'' ", 1024*1024 + 7); |
| | 7 | + |
| | 8 | +$m = memory_get_usage(); |
| | 9 | + |
| | 10 | +$ps1 = preg_split($pattern, $subject); |
| | 11 | + |
| | 12 | +$r = ""; |
| | 13 | +foreach ($ps1 as $c) { |
| | 14 | + $r .= $c . "|"; |
| | 15 | +} |
| | 16 | +echo "Original preg_split: " . md5($r) . " " . (memory_get_usage()-$m) . "\n"; |
| | 17 | + |
| | 18 | +unset($ps1); |
| | 19 | + |
| | 20 | +$r = ""; |
| | 21 | +$ps2 = StringUtils::preg_split($pattern, $subject); |
| | 22 | +foreach ($ps2 as $c) { |
| | 23 | + $r .= $c . "|"; |
| | 24 | +} |
| | 25 | +echo "StringUtils preg_split: " . md5($r) . " " . (memory_get_usage()-$m) . "\n"; |
| Property changes on: trunk/phase3/tests/preg_split_test.php |
| ___________________________________________________________________ |
| Name: svn:eol-style |
| 1 | 26 | + native |
| Index: trunk/phase3/includes/parser/Parser.php |
| — | — | @@ -1154,7 +1154,7 @@ |
| 1155 | 1155 | # be text, and the remaining three constitute mark-up for bold text. |
| 1156 | 1156 | # If there are more than 6 apostrophes in a row, assume they're all |
| 1157 | 1157 | # text except for the last 6. |
| 1158 | | - $arr = preg_split( "/('{2,3}(?:''')?)(?!')/", $text, -1, PREG_SPLIT_DELIM_CAPTURE ); |
| | 1158 | + $arr = Stringutils::preg_split( "/('{2,3}(?:''')?)(?!')/", $text, -1, PREG_SPLIT_DELIM_CAPTURE ); |
| 1159 | 1159 | |
| 1160 | 1160 | |
| 1161 | 1161 | # Now let's actually convert our apostrophic mush to HTML! |
| Index: trunk/phase3/includes/StringUtils.php |
| — | — | @@ -179,6 +179,14 @@ |
| 180 | 180 | return new ArrayIterator( explode( $separator, $subject ) ); |
| 181 | 181 | } |
| 182 | 182 | } |
| | 183 | + |
| | 184 | + /** |
| | 185 | + * Workalike for preg_split() with limited memory usage. |
| | 186 | + * Returns an Iterator |
| | 187 | + */ |
| | 188 | + static function preg_split( $pattern, $subject, $limit = -1, $flags = 0 ) { |
| | 189 | + return new PregSplitIterator( $pattern, $subject, $limit, $flags ); |
| | 190 | + } |
| 183 | 191 | } |
| 184 | 192 | |
| 185 | 193 | /** |
| — | — | @@ -409,3 +417,82 @@ |
| 410 | 418 | } |
| 411 | 419 | } |
| 412 | 420 | |
| | 421 | + |
| | 422 | +/** |
| | 423 | + * An iterator which works exactly like: |
| | 424 | + * |
| | 425 | + * foreach ( preg_split( $pattern, $s, $limit, $flags ) as $element ) { |
| | 426 | + * ... |
| | 427 | + * } |
| | 428 | + * |
| | 429 | + * Except it doesn't use huge amounts of memory when $limit is -1 |
| | 430 | + * |
| | 431 | + * The flag PREG_SPLIT_OFFSET_CAPTURE isn't supported. |
| | 432 | + */ |
| | 433 | +class PregSplitIterator implements Iterator { |
| | 434 | + // The subject string |
| | 435 | + var $pattern, $subject, $originalLimit, $flags; |
| | 436 | + |
| | 437 | + // The last extracted group of items. |
| | 438 | + var $smallArray; |
| | 439 | + |
| | 440 | + // The position on the iterator. |
| | 441 | + var $curPos; |
| | 442 | + |
| | 443 | + const MAX_LIMIT = 100; |
| | 444 | + |
| | 445 | + /** |
| | 446 | + * Construct a PregSplitIterator |
| | 447 | + */ |
| | 448 | + function __construct( $pattern, $s, $limit, $flags) { |
| | 449 | + $this->pattern = $pattern; |
| | 450 | + $this->subject = $s; |
| | 451 | + $this->originalLimit = $limit; |
| | 452 | + $this->flags = $flags; |
| | 453 | + |
| | 454 | + $this->rewind(); |
| | 455 | + } |
| | 456 | + |
| | 457 | + private function effectiveLimit() { |
| | 458 | + if ($this->originalLimit == -1) { |
| | 459 | + return self::MAX_LIMIT + 1; |
| | 460 | + } else if ($this->limit > self::MAX_LIMIT) { |
| | 461 | + $this->limit -= self::MAX_LIMIT; |
| | 462 | + return self::MAX_LIMIT + 1; |
| | 463 | + } else { |
| | 464 | + $old = $this->limit; |
| | 465 | + $this->limit = 0; |
| | 466 | + return $old; |
| | 467 | + } |
| | 468 | + } |
| | 469 | + |
| | 470 | + function rewind() { |
| | 471 | + $this->curPos = 0; |
| | 472 | + $this->limit = $this->originalLimit; |
| | 473 | + if ($this->limit == -1) $this->limit = self::MAX_LIMIT; |
| | 474 | + $this->smallArray = preg_split( $this->pattern, $this->subject, $this->effectiveLimit(), $this->flags); |
| | 475 | + } |
| | 476 | + |
| | 477 | + function current() { |
| | 478 | + return $this->smallArray[$this->curPos % self::MAX_LIMIT]; |
| | 479 | + } |
| | 480 | + |
| | 481 | + function key() { |
| | 482 | + return $this->curPos; |
| | 483 | + } |
| | 484 | + |
| | 485 | + function next() { |
| | 486 | + $this->curPos++; |
| | 487 | + if ( $this->curPos % self::MAX_LIMIT == 0 ) { |
| | 488 | + # Last item contains the rest unsplitted. |
| | 489 | + if ($this->limit > 0) { |
| | 490 | + $this->smallArray = preg_split( $this->pattern, $this->smallArray[self::MAX_LIMIT], $this->effectiveLimit(), $this->flags); |
| | 491 | + } |
| | 492 | + } |
| | 493 | + return; |
| | 494 | + } |
| | 495 | + |
| | 496 | + function valid() { |
| | 497 | + return $this->curPos % self::MAX_LIMIT < count($this->smallArray); |
| | 498 | + } |
| | 499 | +} |