maintenance/generate-entity-file.php

   1 #!/usr/bin/php
   2 <?php
   3
   4 chdir(dirname(__FILE__));
   5 require_once 'common.php';
   6 assertCli();
   7
   8 /**
   9  * @file
  10  * Parses *.ent files into an entity lookup table, and then serializes and
  11  * writes the whole kaboodle to a file. The resulting file should be versioned.
  12  */
  13
  14 // here's where the entity files are located, assuming working directory
  15 // is the same as the location of this PHP file. Needs trailing slash.
  16 $entity_dir = '../docs/entities/';
  17
  18 // defines the output file for the serialized content.
  19 $output_file = '../library/HTMLPurifier/EntityLookup/entities.ser';
  20
  21 // courtesy of a PHP manual comment
  22 function unichr($dec) {
  23     if ($dec < 128) {
  24         $utf  = chr($dec);
  25     } else if ($dec < 2048) {
  26         $utf  = chr(192 + (($dec - ($dec % 64)) / 64));
  27         $utf .= chr(128 + ($dec % 64));
  28     } else {
  29         $utf  = chr(224 + (($dec - ($dec % 4096)) / 4096));
  30         $utf .= chr(128 + ((($dec % 4096) - ($dec % 64)) / 64));
  31         $utf .= chr(128 + ($dec % 64));
  32     }
  33     return $utf;
  34 }
  35
  36 if ( !is_dir($entity_dir) ) exit("Fatal Error: Can't find entity directory.\n");
  37 if ( file_exists($output_file) ) exit("Fatal Error: entity-lookup.txt already exists.\n");
  38
  39 $dh = @opendir($entity_dir);
  40 if ( !$dh ) exit("Fatal Error: Cannot read entity directory.\n");
  41
  42 $entity_files = array();
  43 while (($file = readdir($dh)) !== false) {
  44     if (@$file[0] === '.') continue;
  45     if (substr(strrchr($file, "."), 1) !== 'ent') continue;
  46     $entity_files[] = $file;
  47 }
  48 closedir($dh);
  49
  50 if ( !$entity_files ) exit("Fatal Error: No entity files to parse.\n");
  51
  52 $entity_table = array();
  53 $regexp = '/<!ENTITY\s+([A-Za-z]+)\s+"&#(?:38;#)?([0-9]+);">/';
  54
  55 foreach ( $entity_files as $file ) {
  56     $contents = file_get_contents($entity_dir . $file);
  57     $matches = array();
  58     preg_match_all($regexp, $contents, $matches, PREG_SET_ORDER);
  59     foreach ($matches as $match) {
  60         $entity_table[$match[1]] = unichr($match[2]);
  61     }
  62 }
  63
  64 $output = serialize($entity_table);
  65
  66 $fh = fopen($output_file, 'w');
  67 fwrite($fh, $output);
  68 fclose($fh);
  69
  70 echo "Completed successfully.";
  71