maintenance/merge-library.php

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