xref: /PHP-8.1/scripts/dev/check_parameters.php (revision ceb6fa6d)
1#!/usr/bin/env php
2<?php
3/*
4  +----------------------------------------------------------------------+
5  | Copyright (c) The PHP Group                                          |
6  +----------------------------------------------------------------------+
7  | This source file is subject to version 3.01 of the PHP license,      |
8  | that is bundled with this package in the file LICENSE, and is        |
9  | available through the world-wide-web at the following url:           |
10  | https://www.php.net/license/3_01.txt                                 |
11  | If you did not receive a copy of the PHP license and are unable to   |
12  | obtain it through the world-wide-web, please send a note to          |
13  | license@php.net so we can mail you a copy immediately.               |
14  +----------------------------------------------------------------------+
15  | Author: Nuno Lopes <nlopess@php.net>                                 |
16  +----------------------------------------------------------------------+
17*/
18
19define('REPORT_LEVEL', 1); // 0 reports less false-positives. up to level 5.
20define('VERSION', '7.0');  // minimum is 7.0
21define('PHPDIR', realpath(dirname(__FILE__) . '/../..'));
22
23
24// be sure you have enough memory and stack for PHP. pcre will push the limits!
25ini_set('pcre.backtrack_limit', 10000000);
26
27
28// ------------------------ end of config ----------------------------
29
30
31$API_params = array(
32    'a' => array('zval**'), // array
33    'A' => array('zval**'), // array or object
34    'b' => array('bool*'), // boolean
35    'd' => array('double*'), // double
36    'f' => array('zend_fcall_info*', 'zend_fcall_info_cache*'), // function
37    'h' => array('HashTable**'), // array as an HashTable*
38    'H' => array('HashTable**'), // array or HASH_OF(object)
39    'l' => array('zend_long*'), // long
40    //TODO 'L' => array('zend_long*, '), // long
41    'o' => array('zval**'), //object
42    'O' => array('zval**', 'zend_class_entry*'), // object of given type
43    'P' => array('zend_string**'), // valid path
44    'r' => array('zval**'), // resource
45    'S' => array('zend_string**'), // string
46    'z' => array('zval**'), // zval*
47    'Z' => array('zval***') // zval**
48    // 's', 'p', 'C' handled separately
49);
50
51/** reports an error, according to its level */
52function error($str, $level = 0)
53{
54    global $current_file, $current_function, $line;
55
56    if ($level <= REPORT_LEVEL) {
57        if (strpos($current_file,PHPDIR) === 0) {
58            $filename = substr($current_file, strlen(PHPDIR)+1);
59        } else {
60            $filename = $current_file;
61        }
62        echo $filename , " [$line] $current_function : $str\n";
63    }
64}
65
66
67/** this updates the global var $line (for error reporting) */
68function update_lineno($offset)
69{
70    global $lines_offset, $line;
71
72    $left  = 0;
73    $right = $count = count($lines_offset)-1;
74
75    // a nice binary search :)
76    do {
77        $mid = intval(($left + $right)/2);
78        $val = $lines_offset[$mid];
79
80        if ($val < $offset) {
81            if (++$mid > $count || $lines_offset[$mid] > $offset) {
82                $line = $mid;
83                return;
84            } else {
85                $left = $mid;
86            }
87        } else if ($val > $offset) {
88            if ($lines_offset[--$mid] < $offset) {
89                $line = $mid+1;
90                return;
91            } else {
92                $right = $mid;
93            }
94        } else {
95            $line = $mid+1;
96            return;
97        }
98    } while (true);
99}
100
101
102/** parses the sources and fetches its vars name, type and if they are initialized or not */
103function get_vars($txt)
104{
105    $ret =  array();
106    preg_match_all('/((?:(?:unsigned|struct)\s+)?\w+)(?:\s*(\*+)\s+|\s+(\**))(\w+(?:\[\s*\w*\s*\])?)\s*(?:(=)[^,;]+)?((?:\s*,\s*\**\s*\w+(?:\[\s*\w*\s*\])?\s*(?:=[^,;]+)?)*)\s*;/S', $txt, $m, PREG_SET_ORDER);
107
108    foreach ($m as $x) {
109        // the first parameter is special
110        if (!in_array($x[1], array('else', 'endif', 'return'))) // hack to skip reserved words
111            $ret[$x[4]] = array($x[1] . $x[2] . $x[3], $x[5]);
112
113        // are there more vars?
114        if ($x[6]) {
115            preg_match_all('/(\**)\s*(\w+(?:\[\s*\w*\s*\])?)\s*(=?)/S', $x[6], $y, PREG_SET_ORDER);
116            foreach ($y as $z) {
117                $ret[$z[2]] = array($x[1] . $z[1], $z[3]);
118            }
119        }
120    }
121
122//	if ($GLOBALS['current_function'] == 'for_debugging') { print_r($m);print_r($ret); }
123    return $ret;
124}
125
126
127/** run diagnostic checks against one var. */
128function check_param($db, $idx, $exp, $optional, $allow_uninit = false)
129{
130    global $error_few_vars_given;
131
132    if ($idx >= count($db)) {
133        if (!$error_few_vars_given) {
134            error("too few variables passed to function");
135            $error_few_vars_given = true;
136        }
137        return;
138    } elseif ($db[$idx][0] === '**dummy**') {
139        return;
140    }
141
142    if ($db[$idx][1] != $exp) {
143        error("{$db[$idx][0]}: expected '$exp' but got '{$db[$idx][1]}' [".($idx+1).']');
144    }
145
146    if (!$optional && $db[$idx][2]) {
147        error("not optional var is initialized: {$db[$idx][0]} [".($idx+1).']', 2);
148    }
149    if (!$allow_uninit && $optional && !$db[$idx][2]) {
150        error("optional var not initialized: {$db[$idx][0]} [".($idx+1).']', 1);
151    }
152}
153
154/** fetch params passed to zend_parse_params*() */
155function get_params($vars, $str)
156{
157    $ret = array();
158    preg_match_all('/(?:\([^)]+\))?(&?)([\w>.()-]+(?:\[\w+\])?)\s*,?((?:\)*\s*=)?)/S', $str, $m, PREG_SET_ORDER);
159
160    foreach ($m as $x) {
161        $name = $x[2];
162
163        // little hack for last parameter
164        if (strpos($name, '(') === false) {
165            $name = rtrim($name, ')');
166        }
167
168        if (empty($vars[$name][0])) {
169            error("variable not found: '$name'", 3);
170            $ret[][] = '**dummy**';
171
172        } else {
173            $ret[] = array($name, $vars[$name][0] . ($x[1] ? '*' : ''), $vars[$name][1]);
174        }
175
176        // the end (yes, this is a little hack :P)
177        if ($x[3]) {
178            break;
179        }
180    }
181
182//	if ($GLOBALS['current_function'] == 'for_debugging') { var_dump($m); var_dump($ret); }
183    return $ret;
184}
185
186
187/** run tests on a function. the code is passed in $txt */
188function check_function($name, $txt, $offset)
189{
190    global $API_params;
191
192    $regex = '/
193        (?: zend_parse_parameters(?:_throw)?               \s*\([^,]+
194        |   zend_parse_(?:parameters_ex|method_parameters) \s*\([^,]+,[^,]+
195        |   zend_parse_method_parameters_ex                \s*\([^,]+,[^,]+,[^,+]
196        )
197        ,\s*"([^"]*)"\s*
198        ,\s*([^{;]*)
199    /Sx';
200    if (preg_match_all($regex, $txt, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
201
202        $GLOBALS['current_function'] = $name;
203
204        foreach ($matches as $m) {
205            $GLOBALS['error_few_vars_given'] = false;
206            update_lineno($offset + $m[2][1]);
207
208            $vars = get_vars(substr($txt, 0, $m[0][1])); // limit var search to current location
209            $params = get_params($vars, $m[2][0]);
210            $optional = $varargs = false;
211            $last_char = '';
212            $j = -1;
213
214            $spec = $m[1][0];
215            $len = strlen($spec);
216            for ($i = 0; $i < $len; ++$i) {
217                $char = $spec[$i];
218                switch ($char = $spec[$i]) {
219                    // separator for optional parameters
220                    case '|':
221                        if ($optional) {
222                            error("more than one optional separator at char #$i");
223                        } else {
224                            $optional = true;
225                            if ($i == $len-1) {
226                                error("unnecessary optional separator");
227                            }
228                        }
229                    break;
230
231                    // separate_zval_if_not_ref
232                    case '/':
233                        if (in_array($last_char, array('l', 'L', 'd', 'b'))) {
234                            error("the '/' specifier should not be applied to '$last_char'");
235                        }
236                    break;
237
238                    // nullable arguments
239                    case '!':
240                        if (in_array($last_char, array('l', 'L', 'd', 'b'))) {
241                            check_param($params, ++$j, 'bool*', $optional);
242                        }
243                    break;
244
245                    // variadic arguments
246                    case '+':
247                    case '*':
248                        if ($varargs) {
249                            error("A varargs specifier can only be used once. repeated char at column $i");
250                        } else {
251                            check_param($params, ++$j, 'zval**', $optional);
252                            check_param($params, ++$j, 'int*', $optional);
253                            $varargs = true;
254                        }
255                    break;
256
257                    case 's':
258                    case 'p':
259                        check_param($params, ++$j, 'char**', $optional, $allow_uninit=true);
260                        check_param($params, ++$j, 'size_t*', $optional, $allow_uninit=true);
261                        if ($optional && !$params[$j-1][2] && !$params[$j][2]
262                                && $params[$j-1][0] !== '**dummy**' && $params[$j][0] !== '**dummy**') {
263                            error("one of optional vars {$params[$j-1][0]} or {$params[$j][0]} must be initialized", 1);
264                        }
265                    break;
266
267                    case 'C':
268                        // C must always be initialized, independently of whether it's optional
269                        check_param($params, ++$j, 'zend_class_entry**', false);
270                    break;
271
272                    default:
273                        if (!isset($API_params[$char])) {
274                            error("unknown char ('$char') at column $i");
275                        }
276
277                        // If an is_null flag is in use, only that flag is required to be
278                        // initialized
279                        $allow_uninit = $i+1 < $len && $spec[$i+1] === '!'
280                                && in_array($char, array('l', 'L', 'd', 'b'));
281
282                        foreach ($API_params[$char] as $exp) {
283                            check_param($params, ++$j, $exp, $optional, $allow_uninit);
284                        }
285                }
286
287                $last_char = $char;
288            }
289        }
290    }
291}
292
293
294/** the main recursion function. splits files in functions and calls the other functions */
295function recurse($path)
296{
297    foreach (scandir($path) as $file) {
298        if ($file == '.' || $file == '..' || $file == 'CVS') continue;
299
300        $file = "$path/$file";
301        if (is_dir($file)) {
302            recurse($file);
303            continue;
304        }
305
306        // parse only .c and .cpp files
307        if (substr_compare($file, '.c', -2) && substr_compare($file, '.cpp', -4)) continue;
308
309        $txt = file_get_contents($file);
310        // remove comments (but preserve the number of lines)
311        $txt = preg_replace('@//.*@S', '', $txt);
312        $txt = preg_replace_callback('@/\*.*\*/@SsU', function($matches) {
313            return preg_replace("/[^\r\n]+/S", "", $matches[0]);
314        }, $txt);
315
316        $split = preg_split('/PHP_(?:NAMED_)?(?:FUNCTION|METHOD)\s*\((\w+(?:,\s*\w+)?)\)/S', $txt, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_OFFSET_CAPTURE);
317
318        if (count($split) < 2) continue; // no functions defined on this file
319        array_shift($split); // the first part isn't relevant
320
321
322        // generate the line offsets array
323        $j = 0;
324        $lines = preg_split("/(\r\n?|\n)/S", $txt, -1, PREG_SPLIT_DELIM_CAPTURE);
325        $lines_offset = array();
326
327        for ($i = 0; $i < count($lines); ++$i) {
328            $j += strlen($lines[$i]) + strlen(@$lines[++$i]);
329            $lines_offset[] = $j;
330        }
331
332        $GLOBALS['lines_offset'] = $lines_offset;
333        $GLOBALS['current_file'] = $file;
334
335
336        for ($i = 0; $i < count($split); $i+=2) {
337            // if the /* }}} */ comment is found use it to reduce false positives
338            // TODO: check the other indexes
339            list($f) = preg_split('@/\*\s*}}}\s*\*/@S', $split[$i+1][0]);
340            check_function(preg_replace('/\s*,\s*/S', '::', $split[$i][0]), $f, $split[$i][1]);
341        }
342    }
343}
344
345$dirs = array();
346
347if (isset($argc) && $argc > 1) {
348    if ($argv[1] == '-h' || $argv[1] == '-help' || $argv[1] == '--help') {
349        echo <<<HELP
350Synopsis:
351    php check_parameters.php [directories]
352
353HELP;
354        exit(0);
355    }
356    for ($i = 1; $i < $argc; $i++) {
357        $dirs[] = $argv[$i];
358    }
359} else {
360    $dirs[] = PHPDIR;
361}
362
363foreach($dirs as $dir) {
364    if (is_dir($dir)) {
365        if (!is_readable($dir)) {
366            echo "ERROR: directory '", $dir ,"' is not readable\n";
367            exit(1);
368        }
369    } else {
370        echo "ERROR: bogus directory '", $dir ,"'\n";
371        exit(1);
372    }
373}
374
375foreach ($dirs as $dir) {
376    recurse(realpath($dir));
377}
378