From: Platonides <platonides@users.mediawiki.org>
Date: Tue, 26 Jan 2010 18:58:07 +0000 (+0000)
Subject: Step 4: Profit!!
X-Git-Tag: 1.31.0-rc.0~38098
X-Git-Url: http://git.cyclocoop.org/%24self?a=commitdiff_plain;h=11f8b8390cec0b0331f431b8a11721436a6035f8;p=lhc%2Fweb%2Fwiklou.git

Step 4: Profit!!
Add and use PregSplitIterator instead of a direct preg_split.
Slower, but with an upper bound on memory usage.
---

diff --git a/includes/StringUtils.php b/includes/StringUtils.php
index c437b3c19e..bab9be4bbe 100644
--- a/includes/StringUtils.php
+++ b/includes/StringUtils.php
@@ -179,6 +179,14 @@ class StringUtils {
 			return new ArrayIterator( explode( $separator, $subject ) );
 		}
 	}
+
+	/**
+	 * Workalike for preg_split() with limited memory usage.
+	 * Returns an Iterator
+	 */
+	static function preg_split( $pattern, $subject, $limit = -1, $flags = 0 ) {
+		return new PregSplitIterator( $pattern, $subject, $limit, $flags );
+	}
 }
 
 /**
@@ -409,3 +417,82 @@ class ExplodeIterator implements Iterator {
 	}
 }
 
+
+/**
+ * An iterator which works exactly like:
+ * 
+ * foreach ( preg_split( $pattern, $s, $limit, $flags ) as $element ) {
+ *    ...
+ * }
+ *
+ * Except it doesn't use huge amounts of memory when $limit is -1
+ *
+ * The flag PREG_SPLIT_OFFSET_CAPTURE isn't supported.
+ */
+class PregSplitIterator implements Iterator {
+	// The subject string
+	var $pattern, $subject, $originalLimit, $flags;
+
+	// The last extracted group of items.
+	var $smallArray;
+
+	// The position on the iterator.
+	var $curPos;
+
+	const MAX_LIMIT = 100;
+
+	/** 
+	 * Construct a PregSplitIterator
+	 */
+	function __construct( $pattern, $s, $limit, $flags) {
+		$this->pattern = $pattern;
+		$this->subject = $s;
+		$this->originalLimit = $limit;
+		$this->flags = $flags;
+
+		$this->rewind();
+	}
+
+	private function effectiveLimit() {
+		if ($this->originalLimit == -1) {
+			return self::MAX_LIMIT + 1;
+		} else if ($this->limit > self::MAX_LIMIT) {
+			$this->limit -= self::MAX_LIMIT;
+			return self::MAX_LIMIT + 1;
+		} else {
+			$old = $this->limit;
+			$this->limit = 0;
+			return $old;
+		}
+	}
+
+	function rewind() {
+		$this->curPos = 0;
+		$this->limit = 	$this->originalLimit;
+		if ($this->limit == -1) $this->limit = self::MAX_LIMIT;
+		$this->smallArray = preg_split( $this->pattern, $this->subject, $this->effectiveLimit(), $this->flags);
+	}
+
+	function current() {
+		return $this->smallArray[$this->curPos % self::MAX_LIMIT];
+	}
+
+	function key() {
+		return $this->curPos;
+	}
+
+	function next() {
+		$this->curPos++;
+		if ( $this->curPos % self::MAX_LIMIT == 0 ) {
+			# Last item contains the rest unsplitted.
+			if ($this->limit > 0) {
+				$this->smallArray = preg_split( $this->pattern, $this->smallArray[self::MAX_LIMIT], $this->effectiveLimit(), $this->flags);
+			}
+		}
+		return;
+	}
+
+	function valid() {
+		return $this->curPos % self::MAX_LIMIT < count($this->smallArray);
+	}
+}
diff --git a/includes/parser/Parser.php b/includes/parser/Parser.php
index e419a30df4..50e15dc4f8 100644
--- a/includes/parser/Parser.php
+++ b/includes/parser/Parser.php
@@ -1154,7 +1154,7 @@ class Parser
 			# be text, and the remaining three constitute mark-up for bold text.
 			# If there are more than 6 apostrophes in a row, assume they're all
 			# text except for the last 6.		
-			$arr = preg_split( "/('{2,3}(?:''')?)(?!')/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
+			$arr = Stringutils::preg_split( "/('{2,3}(?:''')?)(?!')/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 
 
 			# Now let's actually convert our apostrophic mush to HTML!
diff --git a/tests/preg_split_test.php b/tests/preg_split_test.php
new file mode 100644
index 0000000000..69c977f774
--- /dev/null
+++ b/tests/preg_split_test.php
@@ -0,0 +1,24 @@
+<?php
+include "../includes/StringUtils.php";
+
+$pattern = "/('')+/";
+$subject = str_repeat("'' ", 1024*1024 + 7);
+
+$m = memory_get_usage();
+
+$ps1 = preg_split($pattern, $subject);
+
+$r = "";
+foreach ($ps1 as $c) {
+	$r .= $c . "|";
+}
+echo "Original preg_split: " . md5($r) . "  " . (memory_get_usage()-$m) . "\n";
+
+unset($ps1);
+
+$r = "";
+$ps2 = StringUtils::preg_split($pattern, $subject);
+foreach ($ps2 as $c) {
+	$r .= $c . "|";
+}
+echo "StringUtils preg_split: " . md5($r) . "  " . (memory_get_usage()-$m) . "\n";