| Index: trunk/phase3/includes/Tokenizer.php |
| — | — | @@ -8,8 +8,11 @@ |
| 9 | 9 | |
| 10 | 10 | /* private */ function Tokenizer() |
| 11 | 11 | { |
| | 12 | + global $wgLang; |
| | 13 | + |
| 12 | 14 | $this->mPos=0; |
| 13 | 15 | $this->mTokenQueue=array(); |
| | 16 | + $this->linkPrefixExtension = $wgLang->linkPrefixExtension(); |
| 14 | 17 | } |
| 15 | 18 | |
| 16 | 19 | # factory function |
| — | — | @@ -54,8 +57,7 @@ |
| 55 | 58 | // proceeds character by character through the text, looking for characters needing |
| 56 | 59 | // special attention. Those are currently: I, R, ', [, ], newline |
| 57 | 60 | // |
| 58 | | - // TODO: prefixed links for Arabic wikipedia not implemented yet |
| 59 | | - // handling of French blanks not yet implemented |
| | 61 | + // TODO: handling of French blanks not yet implemented |
| 60 | 62 | function nextToken() |
| 61 | 63 | { |
| 62 | 64 | $fname = "Tokenizer::nextToken"; |
| — | — | @@ -64,8 +66,8 @@ |
| 65 | 67 | if ( count( $this->mQueuedToken ) != 0 ) { |
| 66 | 68 | // still one token from the last round around. Return that one first. |
| 67 | 69 | $token = array_shift( $this->mQueuedToken ); |
| 68 | | - } else if ( $this->mPos > $this->mTextLength ) |
| 69 | | - { // If no text is left, return "false". |
| | 70 | + } else if ( $this->mPos > $this->mTextLength ) { |
| | 71 | + // If no text is left, return "false". |
| 70 | 72 | $token = false; |
| 71 | 73 | } else { |
| 72 | 74 | |
| — | — | @@ -98,8 +100,21 @@ |
| 99 | 101 | $this->mPos += 3; |
| 100 | 102 | break 2; // switch + while |
| 101 | 103 | } else if ( $this->continues("[") ) { |
| 102 | | - $queueToken["type"] = "[["; |
| | 104 | + $queueToken["type"] = "[["; |
| 103 | 105 | $queueToken["text"] = ""; |
| | 106 | + // Check for a "prefixed link", e.g. Al[[Khazar]] |
| | 107 | + // Mostly for arabic wikipedia |
| | 108 | + if ( $this->linkPrefixExtension ) { |
| | 109 | + while ( $this->linkPrefixExtension |
| | 110 | + && ($len = strlen( $token["text"] ) ) > 0 |
| | 111 | + && !ctype_space( $token["text"][$len-1] ) ) |
| | 112 | + { |
| | 113 | + //prepend the character to the link's open tag |
| | 114 | + $queueToken["text"] = $token["text"][$len-1] . $queueToken["text"]; |
| | 115 | + //remove character from the end of the text token |
| | 116 | + $token["text"] = substr( $token["text"], 0, -1); |
| | 117 | + } |
| | 118 | + } |
| 104 | 119 | $this->mQueuedToken[] = $queueToken; |
| 105 | 120 | $this->mPos += 2; |
| 106 | 121 | break 2; // switch + while |
| — | — | @@ -158,7 +173,7 @@ |
| 159 | 174 | |
| 160 | 175 | // function continues |
| 161 | 176 | // checks whether the mText continues with $cont from mPos+1 |
| 162 | | - function continues( $cont ) |
| | 177 | + /* private */ function continues( $cont ) |
| 163 | 178 | { |
| 164 | 179 | // If string is not long enough to contain $cont, return false |
| 165 | 180 | if ( $this->mTextLength < $this->mPos + strlen( $cont ) ) |