maintenance/mediawiki.Title/generatePhpCharToUpperMappings.php

   1 #!/usr/bin/env php
   2 <?php
   3 /**
   4  * Utility to generate mapping file used in phpCharToUpper.js
   5  *
   6  * Compares output of String.toUpperCase in JavaScript with
   7  * mb_strtoupper in PHP, and outputs a list of lower:upper
   8  * mappings where they differ. This is then used by Title.js
   9  * to provide the same normalization in the client as on
  10  * the server.
  11  */
  12
  13 $data = [];
  14
  15 // phpcs:disable MediaWiki.Usage.ForbiddenFunctions.exec
  16 $jsUpperChars = json_decode( exec( 'node generateJsToUpperCaseList.js' ) );
  17 // phpcs:enable MediaWiki.Usage.ForbiddenFunctions.exec
  18
  19 for ( $i = 0; $i < 65536; $i++ ) {
  20         if ( $i >= 0xd800 && $i <= 0xdfff ) {
  21                 // Skip surrogate pairs
  22                 continue;
  23         }
  24         $char = mb_convert_encoding( '&#' . $i . ';', 'UTF-8', 'HTML-ENTITIES' );
  25         $phpUpper = mb_strtoupper( $char );
  26         $jsUpper = $jsUpperChars[$i];
  27         if ( $jsUpper !== $phpUpper ) {
  28                 $data[$char] = $phpUpper;
  29         }
  30 }
  31
  32 echo 'var toUpperMapping = ' . str_replace( '"', "'",
  33         str_replace( '    ', "\t",
  34                 json_encode( $data, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE )
  35         )
  36 ) . ";\n";