maintenance/merge-library.php

   1 #!/usr/bin/php
   2 <?php
   3
   4 require_once 'common.php';
   5 assertCli();
   6
   7 /**
   8  * Compiles all of HTML Purifier's library files into one big file
   9  * named HTMLPurifier.standalone.php.
  10  */
  11
  12 /**
  13  * Global hash that tracks already loaded includes
  14  */
  15 $GLOBALS['loaded'] = array('HTMLPurifier.php' => true);
  16
  17 /**
  18  * Custom FSTools for this script that overloads some behavior
  19  * @warning The overloading of copy() is not necessarily global for
  20  *          this script. Watch out!
  21  */
  22 class MergeLibraryFSTools extends FSTools
  23 {
  24     function copyable($entry) {
  25         // Skip hidden files
  26         if ($entry[0] == '.') {
  27             return false;
  28         }
  29         return true;
  30     }
  31     function copy($source, $dest) {
  32         copy_and_remove_includes($source, $dest);
  33     }
  34 }
  35 $FS = new MergeLibraryFSTools();
  36
  37 /**
  38  * Replaces the includes inside PHP source code with the corresponding
  39  * source.
  40  * @param string $text PHP source code to replace includes from
  41  */
  42 function replace_includes($text) {
  43     return preg_replace_callback(
  44         "/require_once ['\"]([^'\"]+)['\"];/",
  45         'replace_includes_callback',
  46         $text
  47     );
  48 }
  49
  50 /**
  51  * Removes leading PHP tags from included files. Assumes that there is
  52  * no trailing tag.
  53  * @note This is safe for files that have internal <?php
  54  * @param string $text Text to have leading PHP tag from
  55  */
  56 function remove_php_tags($text) {
  57     return substr($text, 5);
  58 }
  59
  60 /**
  61  * Creates an appropriate blank file, recursively generating directories
  62  * if necessary
  63  * @param string $file Filename to create blank for
  64  */
  65 function create_blank($file) {
  66     global $FS;
  67     $dir = dirname($file);
  68     $base = realpath('../tests/blanks/') . DIRECTORY_SEPARATOR ;
  69     if ($dir != '.') {
  70         $FS->mkdir($base . $dir);
  71     }
  72     file_put_contents($base . $file, '');
  73 }
  74
  75 /**
  76  * Copies the contents of a directory to the standalone directory
  77  * @param string $dir Directory to copy
  78  */
  79 function make_dir_standalone($dir) {
  80     global $FS;
  81     return $FS->copyr($dir, 'standalone/' . $dir);
  82 }
  83
  84 /**
  85  * Copies the contents of a file to the standalone directory
  86  * @param string $file File to copy
  87  */
  88 function make_file_standalone($file) {
  89     global $FS;
  90     $FS->mkdir('standalone/' . dirname($file));
  91     copy_and_remove_includes($file, 'standalone/' . $file);
  92     return true;
  93 }
  94
  95 /**
  96  * Copies a file to another location recursively, if it is a PHP file
  97  * remove includes
  98  * @param string $file Original file
  99  * @param string $sfile New location of file
 100  */
 101 function copy_and_remove_includes($file, $sfile) {
 102     $contents = file_get_contents($file);
 103     if (strrchr($file, '.') === '.php') $contents = replace_includes($contents);
 104     return file_put_contents($sfile, $contents);
 105 }
 106
 107 /**
 108  * @param $matches preg_replace_callback matches array, where index 1
 109  *        is the filename to include
 110  */
 111 function replace_includes_callback($matches) {
 112     $file = $matches[1];
 113     $preserve = array(
 114       // PHP 5 only
 115       'HTMLPurifier/Lexer/DOMLex.php' => 1,
 116       'HTMLPurifier/Printer.php' => 1,
 117       // PEAR (external)
 118       'XML/HTMLSax3.php' => 1
 119     );
 120     if (isset($preserve[$file])) {
 121         return $matches[0];
 122     }
 123     if (isset($GLOBALS['loaded'][$file])) return '';
 124     $GLOBALS['loaded'][$file] = true;
 125     create_blank($file);
 126     return replace_includes(remove_php_tags(file_get_contents($file)));
 127 }
 128
 129 chdir(dirname(__FILE__) . '/../library/');
 130 create_blank('HTMLPurifier.php');
 131
 132 echo 'Creating full file...';
 133 $contents = replace_includes(file_get_contents('HTMLPurifier.php'));
 134 $contents = str_replace(
 135     "define('HTMLPURIFIER_PREFIX', dirname(__FILE__));",
 136     "define('HTMLPURIFIER_PREFIX', dirname(__FILE__) . '/standalone');
 137 set_include_path(HTMLPURIFIER_PREFIX . PATH_SEPARATOR . get_include_path());",
 138     $contents
 139 );
 140 file_put_contents('HTMLPurifier.standalone.php', $contents);
 141 echo ' done!' . PHP_EOL;
 142
 143 echo 'Creating standalone directory...';
 144 $FS->rmdirr('standalone'); // ensure a clean copy
 145
 146 // data files
 147 $FS->mkdir('standalone/HTMLPurifier/DefinitionCache/Serializer');
 148 make_dir_standalone('HTMLPurifier/EntityLookup');
 149
 150 // non-standard inclusion setup
 151 make_dir_standalone('HTMLPurifier/Language');
 152
 153 // optional components
 154 make_file_standalone('HTMLPurifier/Printer.php');
 155 make_dir_standalone('HTMLPurifier/Printer');
 156 make_dir_standalone('HTMLPurifier/Filter');
 157 make_file_standalone('HTMLPurifier/Lexer/PEARSax3.php');
 158
 159 // PHP 5 only files
 160 make_file_standalone('HTMLPurifier/Lexer/DOMLex.php');
 161 make_file_standalone('HTMLPurifier/Lexer/PH5P.php');
 162 echo ' done!' . PHP_EOL;