*/
class StringUtils {
/**
- * Perform an operation equivalent to
+ * Perform an operation equivalent to
*
* preg_replace( "!$startDelim(.*?)$endDelim!", $replace, $subject );
*
* except that it's worst-case O(N) instead of O(N^2)
*
* Compared to delimiterReplace(), this implementation is fast but memory-
- * hungry and inflexible. The memory requirements are such that I don't
+ * hungry and inflexible. The memory requirements are such that I don't
* recommend using it on anything but guaranteed small chunks of text.
*/
static function hungryDelimiterReplace( $startDelim, $endDelim, $replace, $subject ) {
}
return $output;
}
-
+
/**
- * Perform an operation equivalent to
+ * Perform an operation equivalent to
*
* preg_replace_callback( "!$startDelim(.*)$endDelim!s$flags", $callback, $subject )
*
*/
# If the start delimiter ends with an initial substring of the end delimiter,
# e.g. in the case of C-style comments, the behaviour differs from the model
- # regex. In this implementation, the end must share no characters with the
- # start, so e.g. /*/ is not considered to be both the start and end of a
- # comment. /*/xy/*/ is considered to be a single comment with contents /xy/.
+ # regex. In this implementation, the end must share no characters with the
+ # start, so e.g. /*/ is not considered to be both the start and end of a
+ # comment. /*/xy/*/ is considered to be a single comment with contents /xy/.
static function delimiterReplaceCallback( $startDelim, $endDelim, $callback, $subject, $flags = '' ) {
$inputPos = 0;
$outputPos = 0;
$strcmp = strpos( $flags, 'i' ) === false ? 'strcmp' : 'strcasecmp';
$endLength = strlen( $endDelim );
$m = array();
-
- while ( $inputPos < strlen( $subject ) &&
- preg_match( "!($encStart)|($encEnd)!S$flags", $subject, $m, PREG_OFFSET_CAPTURE, $inputPos ) )
+
+ while ( $inputPos < strlen( $subject ) &&
+ preg_match( "!($encStart)|($encEnd)!S$flags", $subject, $m, PREG_OFFSET_CAPTURE, $inputPos ) )
{
$tokenOffset = $m[0][1];
if ( $m[1][0] != '' ) {
- if ( $foundStart &&
+ if ( $foundStart &&
$strcmp( $endDelim, substr( $subject, $tokenOffset, $endLength ) ) == 0 )
{
# An end match is present at the same location
}
if ( $tokenType == 'start' ) {
- $inputPos = $tokenOffset + $tokenLength;
# Only move the start position if we haven't already found a start
# This means that START START END matches outer pair
if ( !$foundStart ) {
# Found start
+ $inputPos = $tokenOffset + $tokenLength;
# Write out the non-matching section
$output .= substr( $subject, $outputPos, $tokenOffset - $outputPos );
$outputPos = $tokenOffset;
$contentPos = $inputPos;
$foundStart = true;
+ } else {
+ # Move the input position past the *first character* of START,
+ # to protect against missing END when it overlaps with START
+ $inputPos = $tokenOffset + 1;
}
} elseif ( $tokenType == 'end' ) {
if ( $foundStart ) {
}
/*
- * Perform an operation equivalent to
+ * Perform an operation equivalent to
*
* preg_replace( "!$startDelim(.*)$endDelim!$flags", $replace, $subject )
*
* @param string $startDelim Start delimiter regular expression
* @param string $endDelim End delimiter regular expression
- * @param string $replace Replacement string. May contain $1, which will be
+ * @param string $replace Replacement string. May contain $1, which will be
* replaced by the text between the delimiters
* @param string $subject String to search
* @return string The string with the matches replaced
*/
static function explodeMarkup( $separator, $text ) {
$placeholder = "\x00";
-
+
// Remove placeholder instances
$text = str_replace( $placeholder, '', $text );
-
+
// Replace instances of the separator inside HTML-like tags with the placeholder
$replacer = new DoubleReplacer( $separator, $placeholder );
$cleaned = StringUtils::delimiterReplaceCallback( '<', '>', $replacer->cb(), $text );
foreach( $items as $i => $str ) {
$items[$i] = str_replace( $placeholder, $separator, $str );
}
-
+
return $items;
}
$string = str_replace( '$', '\\$', $string );
return $string;
}
+
+ /**
+ * Workalike for explode() with limited memory usage.
+ * Returns an Iterator
+ */
+ static function explode( $separator, $subject ) {
+ if ( substr_count( $subject, $separator ) > 1000 ) {
+ return new ExplodeIterator( $separator, $subject );
+ } else {
+ return new ArrayIterator( explode( $separator, $subject ) );
+ }
+ }
}
/**
- * Base class for "replacers", objects used in preg_replace_callback() and
+ * Base class for "replacers", objects used in preg_replace_callback() and
* StringUtils::delimiterReplaceCallback()
*/
class Replacer {
$this->to = $to;
$this->index = $index;
}
-
+
function replace( $matches ) {
return str_replace( $this->from, $this->to, $matches[$this->index] );
}
$this->fss = false;
}
+ function removePair( $from ) {
+ unset($this->data[$from]);
+ $this->fss = false;
+ }
+
+ function removeArray( $data ) {
+ foreach( $data as $from => $to )
+ $this->removePair( $from );
+ $this->fss = false;
+ }
+
function replace( $subject ) {
if ( function_exists( 'fss_prep_replace' ) ) {
wfProfileIn( __METHOD__.'-fss' );
}
}
+/**
+ * An iterator which works exactly like:
+ *
+ * foreach ( explode( $delim, $s ) as $element ) {
+ * ...
+ * }
+ *
+ * Except it doesn't use 193 byte per element
+ */
+class ExplodeIterator implements Iterator {
+ // The subject string
+ var $subject, $subjectLength;
+
+ // The delimiter
+ var $delim, $delimLength;
+
+ // The position of the start of the line
+ var $curPos;
+
+ // The position after the end of the next delimiter
+ var $endPos;
+
+ // The current token
+ var $current;
+
+ /**
+ * Construct a DelimIterator
+ */
+ function __construct( $delim, $s ) {
+ $this->subject = $s;
+ $this->delim = $delim;
+
+ // Micro-optimisation (theoretical)
+ $this->subjectLength = strlen( $s );
+ $this->delimLength = strlen( $delim );
+
+ $this->rewind();
+ }
+
+ function rewind() {
+ $this->curPos = 0;
+ $this->endPos = strpos( $this->subject, $this->delim );
+ $this->refreshCurrent();
+ }
+
+
+ function refreshCurrent() {
+ if ( $this->curPos === false ) {
+ $this->current = false;
+ } elseif ( $this->curPos >= $this->subjectLength ) {
+ $this->current = '';
+ } elseif ( $this->endPos === false ) {
+ $this->current = substr( $this->subject, $this->curPos );
+ } else {
+ $this->current = substr( $this->subject, $this->curPos, $this->endPos - $this->curPos );
+ }
+ }
+
+ function current() {
+ return $this->current;
+ }
+
+ function key() {
+ return $this->curPos;
+ }
+
+ function next() {
+ if ( $this->endPos === false ) {
+ $this->curPos = false;
+ } else {
+ $this->curPos = $this->endPos + $this->delimLength;
+ if ( $this->curPos >= $this->subjectLength ) {
+ $this->endPos = false;
+ } else {
+ $this->endPos = strpos( $this->subject, $this->delim, $this->curPos );
+ }
+ }
+ $this->refreshCurrent();
+ return $this->current;
+ }
+
+ function valid() {
+ return $this->curPos !== false;
+ }
+}
scripts.mit.edu / autoinstallsdev / mediawiki.git / blobdiff