use locale setting for regexes, but revert after preg call

1 files changed, 14 insertions, 26 deletions

diff --git a/it.class b/it.class
index 2fe2682..26b2055 100644
--- a/it.class
+++ b/it.class
@@ -165,7 +165,7 @@ function toascii($text)
 
 
 /**
- * Convert regex for preg (redefines \w,\W,\b,\B, adds and escapes delimiter, adds modifiers)
+ * Convert regex for preg (adds and escapes delimiter, adds modifiers)
  * @param $pattern Regex to convert
  * @param named parameter casesensitive Regex is case sensitive (omit modifier i)
  * @param named parameter multiline add modifier m
@@ -176,31 +176,8 @@ function toascii($text)
  */
 function convertregex( $pattern, $p = array() )
 {
-	$wordchar = 'a-zA-Z0-9_\xa0-\xff';
-	$nonwordchar = '\x00-\x2f\x3a-\x40\x5B-\x60\x7b-\x9f';
-	
-	#matches only even number of backslashes (double escaped for php and preg)
-	$nonesc = '(?<!\\\\)(?:\\\\\\\\)*';
-	
-	#matches unescaped '[' until unescaped and not starting ']', starting ] as in []gna] or [^]gna]
-	$incharclass = "$nonesc\\[\\^?\\]?(?:[^]]|$nonesc\\\\])*";
-	
 	$pattern = preg_replace( "|/|", '\/', $pattern ); 
-	$replaces = array( 
-		# \w in character class
-		"/($incharclass$nonesc)\\\\w/" => "\$1$wordchar",
-		# \W in character class
-		"/($incharclass$nonesc)\\\\W/" => "\$1$nonwordchar",
-		# normal \w
-		"/($nonesc)\\\\w/" => "\$1[$wordchar]",
-		# normal \W
-		"/($nonesc)\\\\W/" => "\$1[$nonwordchar]",
-		# \b (use negative assertions to match at end of string)
-		"/($nonesc)\\\\b/" => "\$1(?:(?<![$wordchar])(?![$nonwordchar])|(?<![$nonwordchar])(?![$wordchar]))",
-		# \B (use positive assertions to not match at end of string)
-		"/($nonesc)\\\\B/" => "\$1(?:(?<=[$wordchar])(?=[$wordchar])|(?<=[$nonwordchar])(?=[$nonwordchar]))",
-		);
-	$pattern = preg_replace( array_keys( $replaces ), array_values( $replaces ), $pattern );
+	$modifiers = '';
 	if( ! $p['casesensitive'] )
 		$modifiers .= 'i';
 	foreach( array(
@@ -227,10 +204,15 @@ function match( $pattern, $string, $p = array() )
 	$flags = 0;
 	if( $p['offset_capture'] )
 		$flags |= PREG_OFFSET_CAPTURE;
+	$oldlocale = setlocale( LC_CTYPE, 0 );
+	if( $oldlocale != 'de_CH' )
+		setlocale( LC_CTYPE, 'de_CH' );
 	if( $p['all'] )
 		$r = preg_match_all( it::convertregex( $pattern, $p ), $string, $m, $flags | PREG_PATTERN_ORDER, $p['offset'] );
 	else
 		$r = preg_match( it::convertregex( $pattern, $p ), $string, $m, $flags, $p['offset'] );
+	if( $oldlocale != 'de_CH' )
+		setlocale( LC_CTYPE, $oldlocale );
 	# no match
 	if( !$r )
 		return $p['all'] ? array() : false;
@@ -259,7 +241,13 @@ function replace( $replacements, $string, $p = array() )
 	$patterns = array();
 	foreach( array_keys( $replacements ) as $pat ) 
 		$patterns[] = it::convertregex( $pat, $p );
-	return preg_replace( $patterns, array_values( $replacements ), $string, isset( $p['limit'] ) ? $p['limit'] : -1 );
+	$oldlocale = setlocale( LC_CTYPE, 0 );
+	if( $oldlocale != 'de_CH' )
+		setlocale( LC_CTYPE, 'de_CH' );
+	$r = preg_replace( $patterns, array_values( $replacements ), $string, isset( $p['limit'] ) ? $p['limit'] : -1 );
+	if( $oldlocale != 'de_CH' )
+		setlocale( LC_CTYPE, $oldlocale );
+	return $r;
 }
 
 }