| Index: trunk/phase3/includes/WebRequest.php |
| — | — | @@ -220,13 +220,10 @@ |
| 221 | 221 | */ |
| 222 | 222 | function getGPCVal( $arr, $name, $default ) { |
| 223 | 223 | if( isset( $arr[$name] ) ) { |
| 224 | | - global $wgContLang; |
| 225 | 224 | $data = $arr[$name]; |
| 226 | 225 | if( isset( $_GET[$name] ) && !is_array( $data ) ) { |
| 227 | 226 | # Check for alternate/legacy character encoding. |
| 228 | | - if( isset( $wgContLang ) ) { |
| 229 | | - $data = $wgContLang->checkTitleEncoding( $data ); |
| 230 | | - } |
| | 227 | + $data = $this->checkTitleEncoding( $data ); |
| 231 | 228 | } |
| 232 | 229 | $data = $this->normalizeUnicode( $data ); |
| 233 | 230 | return $data; |
| — | — | @@ -234,7 +231,21 @@ |
| 235 | 232 | return $default; |
| 236 | 233 | } |
| 237 | 234 | } |
| | 235 | + |
| | 236 | + protected function checkTitleEncoding( $s ) { |
| | 237 | + global $wgContLang; |
| | 238 | + if( !isset($wgContLang) ) return $s; |
| | 239 | + # Check for non-UTF-8 URLs |
| | 240 | + $ishigh = preg_match( '/[\x80-\xff]/', $s); |
| | 241 | + if( !$ishigh ) return $s; |
| 238 | 242 | |
| | 243 | + $isutf8 = preg_match( '/^([\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' . |
| | 244 | + '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s ); |
| | 245 | + if( $isutf8 ) return $s; |
| | 246 | + # Do the heavy lifting by unstubbing $wgContLang |
| | 247 | + return $wgContLang->iconv( $wgContLang->fallback8bitEncoding(), "utf-8", $s ); |
| | 248 | + } |
| | 249 | + |
| 239 | 250 | /** |
| 240 | 251 | * Fetch a scalar from the input or return $default if it's not set. |
| 241 | 252 | * Returns a string. Arrays are discarded. Useful for |