maintenance/generate-entity-file.php

   1 #!/usr/bin/php
   2 <?php
   3
   4 /**
   5  * Parses *.ent files into an entity lookup table, and then serializes and
   6  * writes the whole kaboodle to a file. The resulting file should be versioned.
   7  */
   8
   9 chdir( dirname(__FILE__) );
  10
  11 // here's where the entity files are located, assuming working directory
  12 // is the same as the location of this PHP file. Needs trailing slash.
  13 $entity_dir = '../docs/entities/';
  14
  15 // defines the output file for the serialized content.
  16 $output_file = '../library/HTMLPurifier/EntityLookup/entities.ser';
  17
  18 // courtesy of a PHP manual comment
  19 function unichr($dec) {
  20     if ($dec < 128) {
  21         $utf  = chr($dec);
  22     } else if ($dec < 2048) {
  23         $utf  = chr(192 + (($dec - ($dec % 64)) / 64));
  24         $utf .= chr(128 + ($dec % 64));
  25     } else {
  26         $utf  = chr(224 + (($dec - ($dec % 4096)) / 4096));
  27         $utf .= chr(128 + ((($dec % 4096) - ($dec % 64)) / 64));
  28         $utf .= chr(128 + ($dec % 64));
  29     }
  30     return $utf;
  31 }
  32
  33 if ( !is_dir($entity_dir) ) exit("Fatal Error: Can't find entity directory.\n");
  34 if ( file_exists($output_file) ) exit("Fatal Error: entity-lookup.txt already exists.\n");
  35
  36 $dh = @opendir($entity_dir);
  37 if ( !$dh ) exit("Fatal Error: Cannot read entity directory.\n");
  38
  39 $entity_files = array();
  40 while (($file = readdir($dh)) !== false) {
  41     if (@$file[0] === '.') continue;
  42     if (substr(strrchr($file, "."), 1) !== 'ent') continue;
  43     $entity_files[] = $file;
  44 }
  45 closedir($dh);
  46
  47 if ( !$entity_files ) exit("Fatal Error: No entity files to parse.\n");
  48
  49 $entity_table = array();
  50 $regexp = '/<!ENTITY\s+([A-Za-z]+)\s+"&#(?:38;#)?([0-9]+);">/';
  51
  52 foreach ( $entity_files as $file ) {
  53     $contents = file_get_contents($entity_dir . $file);
  54     $matches = array();
  55     preg_match_all($regexp, $contents, $matches, PREG_SET_ORDER);
  56     foreach ($matches as $match) {
  57         $entity_table[$match[1]] = unichr($match[2]);
  58     }
  59 }
  60
  61 $output = serialize($entity_table);
  62
  63 $fh = fopen($output_file, 'w');
  64 fwrite($fh, $output);
  65 fclose($fh);
  66
  67 echo "Completed successfully.";
  68
  69 ?>