maintenance/config-scanner.php

   1 #!/usr/bin/php
   2 <?php
   3
   4 chdir(dirname(__FILE__));
   5 require_once 'common.php';
   6 require_once '../library/HTMLPurifier.auto.php';
   7 assertCli();
   8
   9 if (version_compare(PHP_VERSION, '5.2.2', '<')) {
  10     echo "This script requires PHP 5.2.2 or later, for tokenizer line numbers.";
  11     exit(1);
  12 }
  13
  14 /**
  15  * @file
  16  * Scans HTML Purifier source code for $config tokens and records the
  17  * directive being used; configdoc can use this info later.
  18  *
  19  * Currently, this just dumps all the info onto the console. Eventually, it
  20  * will create an XML file that our XSLT transform can use.
  21  */
  22
  23 $FS = new FSTools();
  24 chdir(dirname(__FILE__) . '/../library/');
  25 $raw_files = $FS->globr('.', '*.php');
  26 $files = array();
  27 foreach ($raw_files as $file) {
  28     $file = substr($file, 2); // rm leading './'
  29     if (strncmp('standalone/', $file, 11) === 0) continue; // rm generated files
  30     if (substr_count($file, '.') > 1) continue; // rm meta files
  31     $files[] = $file;
  32 }
  33
  34 /**
  35  * Moves the $i cursor to the next non-whitespace token
  36  */
  37 function consumeWhitespace($tokens, &$i) {
  38     do {$i++;} while (is_array($tokens[$i]) && $tokens[$i][0] === T_WHITESPACE);
  39 }
  40
  41 /**
  42  * Tests whether or not a token is a particular type. There are three run-cases:
  43  *      - ($token, $expect_token): tests if the token is $expect_token type;
  44  *      - ($token, $expect_value): tests if the token is the string $expect_value;
  45  *      - ($token, $expect_token, $expect_value): tests if token is $expect_token type, and
  46  *        its string representation is $expect_value
  47  */
  48 function testToken($token, $value_or_token, $value = null) {
  49     if (is_null($value)) {
  50         if (is_int($value_or_token)) return is_array($token) && $token[0] === $value_or_token;
  51         else return $token === $value_or_token;
  52     } else {
  53         return is_array($token) && $token[0] === $value_or_token && $token[1] === $value;
  54     }
  55 }
  56
  57 $counter = 0;
  58 $full_counter = 0;
  59 $tracker = array();
  60
  61 foreach ($files as $file) {
  62     $tokens = token_get_all(file_get_contents($file));
  63     $file = str_replace('\\', '/', $file);
  64     for ($i = 0, $c = count($tokens); $i < $c; $i++) {
  65         $ok = false;
  66         // Match $config
  67         if (!$ok && testToken($tokens[$i], T_VARIABLE, '$config')) $ok = true;
  68         // Match $this->config
  69         while (!$ok && testToken($tokens[$i], T_VARIABLE, '$this')) {
  70             consumeWhitespace($tokens, $i);
  71             if (!testToken($tokens[$i], T_OBJECT_OPERATOR)) break;
  72             consumeWhitespace($tokens, $i);
  73             if (testToken($tokens[$i], T_STRING, 'config')) $ok = true;
  74             break;
  75         }
  76         if (!$ok) continue;
  77
  78         $ok = false;
  79         for($i++; $i < $c; $i++) {
  80             if ($tokens[$i] === ',' || $tokens[$i] === ')' || $tokens[$i] === ';') {
  81                 break;
  82             }
  83             if (is_string($tokens[$i])) continue;
  84             if ($tokens[$i][0] === T_OBJECT_OPERATOR) {
  85                 $ok = true;
  86                 break;
  87             }
  88         }
  89         if (!$ok) continue;
  90
  91         $line = $tokens[$i][2];
  92
  93         consumeWhitespace($tokens, $i);
  94         if (!testToken($tokens[$i], T_STRING, 'get')) continue;
  95
  96         consumeWhitespace($tokens, $i);
  97         if (!testToken($tokens[$i], '(')) continue;
  98
  99         $full_counter++;
 100
 101         $matched = false;
 102         do {
 103
 104             // What we currently don't match are batch retrievals, and
 105             // wildcard retrievals. This data might be useful in the future,
 106             // which is why we have a do {} while loop that doesn't actually
 107             // do anything.
 108
 109             consumeWhitespace($tokens, $i);
 110             if (!testToken($tokens[$i], T_CONSTANT_ENCAPSED_STRING)) continue;
 111             $namespace = substr($tokens[$i][1], 1, -1);
 112
 113             consumeWhitespace($tokens, $i);
 114             if (!testToken($tokens[$i], ',')) continue;
 115
 116             consumeWhitespace($tokens, $i);
 117             if (!testToken($tokens[$i], T_CONSTANT_ENCAPSED_STRING)) continue;
 118             $directive = substr($tokens[$i][1], 1, -1);
 119
 120             $counter++;
 121             $matched = true;
 122
 123             $id = "$namespace.$directive";
 124             if (!isset($tracker[$id])) $tracker[$id] = array();
 125             if (!isset($tracker[$id][$file])) $tracker[$id][$file] = array();
 126             $tracker[$id][$file][] = $line;
 127
 128         } while (0);
 129
 130         //echo "$file:$line uses $namespace.$directive\n";
 131     }
 132 }
 133
 134 echo "\n$counter/$full_counter instances of \$config or \$this->config found in source code.\n";
 135
 136 echo "Generating XML... ";
 137
 138 $xw = new XMLWriter();
 139 $xw->openURI('../configdoc/usage.xml');
 140 $xw->setIndent(true);
 141 $xw->startDocument('1.0', 'UTF-8');
 142 $xw->startElement('usage');
 143 foreach ($tracker as $id => $files) {
 144     $xw->startElement('directive');
 145     $xw->writeAttribute('id', $id);
 146     foreach ($files as $file => $lines) {
 147         $xw->startElement('file');
 148         $xw->writeAttribute('name', $file);
 149         foreach ($lines as $line) {
 150             $xw->writeElement('line', $line);
 151         }
 152         $xw->endElement();
 153     }
 154     $xw->endElement();
 155 }
 156 $xw->endElement();
 157 $xw->flush();
 158
 159 echo "done!\n";
 160
 161 // vim: et sw=4 sts=4