xref: /PHP-7.2/scripts/dev/check_parameters.php (revision fbfdd1e1) 
1<?php
2/*
3  +----------------------------------------------------------------------+
4  | PHP Version 7                                                        |
5  +----------------------------------------------------------------------+
6  | Copyright (c) 1997-2018 The PHP Group                                |
7  +----------------------------------------------------------------------+
8  | This source file is subject to version 3.01 of the PHP license,      |
9  | that is bundled with this package in the file LICENSE, and is        |
10  | available through the world-wide-web at the following url:           |
11  | http://www.php.net/license/3_01.txt                                  |
12  | If you did not receive a copy of the PHP license and are unable to   |
13  | obtain it through the world-wide-web, please send a note to          |
14  | license@php.net so we can mail you a copy immediately.               |
15  +----------------------------------------------------------------------+
16  | Author: Nuno Lopes <nlopess@php.net>                                 |
17  +----------------------------------------------------------------------+
18*/
19
20/* $Id$ */
21
22
23define('REPORT_LEVEL', 1); // 0 reports less false-positives. up to level 5.
24define('VERSION', '7.0');  // minimum is 7.0
25define('PHPDIR', realpath(dirname(__FILE__) . '/../..'));
26
27
28// be sure you have enough memory and stack for PHP. pcre will push the limits!
29ini_set('pcre.backtrack_limit', 10000000);
30
31
32// ------------------------ end of config ----------------------------
33
34
35$API_params = array(
36	'a' => array('zval**'), // array
37	'A' => array('zval**'), // array or object
38	'b' => array('zend_bool*'), // boolean
39	'd' => array('double*'), // double
40	'f' => array('zend_fcall_info*', 'zend_fcall_info_cache*'), // function
41	'h' => array('HashTable**'), // array as an HashTable*
42	'H' => array('HashTable**'), // array or HASH_OF(object)
43	'l' => array('zend_long*'), // long
44	//TODO 'L' => array('zend_long*, '), // long
45	'o' => array('zval**'), //object
46	'O' => array('zval**', 'zend_class_entry*'), // object of given type
47	'P' => array('zend_string**'), // valid path
48	'r' => array('zval**'), // resource
49	'S' => array('zend_string**'), // string
50	'z' => array('zval**'), // zval*
51	'Z' => array('zval***') // zval**
52	// 's', 'p', 'C' handled separately
53);
54
55/** reports an error, according to its level */
56function error($str, $level = 0)
57{
58	global $current_file, $current_function, $line;
59
60	if ($level <= REPORT_LEVEL) {
61		if (strpos($current_file,PHPDIR) === 0) {
62			$filename = substr($current_file, strlen(PHPDIR)+1);
63		} else {
64			$filename = $current_file;
65		}
66		echo $filename , " [$line] $current_function : $str\n";
67	}
68}
69
70
71/** this updates the global var $line (for error reporting) */
72function update_lineno($offset)
73{
74	global $lines_offset, $line;
75
76	$left  = 0;
77	$right = $count = count($lines_offset)-1;
78
79	// a nice binary search :)
80	do {
81		$mid = intval(($left + $right)/2);
82		$val = $lines_offset[$mid];
83
84		if ($val < $offset) {
85			if (++$mid > $count || $lines_offset[$mid] > $offset) {
86				$line = $mid;
87				return;
88			} else {
89				$left = $mid;
90			}
91		} else if ($val > $offset) {
92			if ($lines_offset[--$mid] < $offset) {
93				$line = $mid+1;
94				return;
95			} else {
96				$right = $mid;
97			}
98		} else {
99			$line = $mid+1;
100			return;
101		}
102	} while (true);
103}
104
105
106/** parses the sources and fetches its vars name, type and if they are initialized or not */
107function get_vars($txt)
108{
109	$ret =  array();
110	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, $m, PREG_SET_ORDER);
111
112	foreach ($m as $x) {
113		// the first parameter is special
114		if (!in_array($x[1], array('else', 'endif', 'return'))) // hack to skip reserved words
115			$ret[$x[4]] = array($x[1] . $x[2] . $x[3], $x[5]);
116
117		// are there more vars?
118		if ($x[6]) {
119			preg_match_all('/(\**)\s*(\w+(?:\[\s*\w*\s*\])?)\s*(=?)/S', $x[6], $y, PREG_SET_ORDER);
120			foreach ($y as $z) {
121				$ret[$z[2]] = array($x[1] . $z[1], $z[3]);
122			}
123		}
124	}
125
126//	if ($GLOBALS['current_function'] == 'for_debugging') { print_r($m);print_r($ret); }
127	return $ret;
128}
129
130
131/** run diagnostic checks against one var. */
132function check_param($db, $idx, $exp, $optional, $allow_uninit = false)
133{
134	global $error_few_vars_given;
135
136	if ($idx >= count($db)) {
137		if (!$error_few_vars_given) {
138			error("too few variables passed to function");
139			$error_few_vars_given = true;
140		}
141		return;
142	} elseif ($db[$idx][0] === '**dummy**') {
143		return;
144	}
145
146	if ($db[$idx][1] != $exp) {
147		error("{$db[$idx][0]}: expected '$exp' but got '{$db[$idx][1]}' [".($idx+1).']');
148	}
149
150	if (!$optional && $db[$idx][2]) {
151		error("not optional var is initialized: {$db[$idx][0]} [".($idx+1).']', 2);
152	}
153	if (!$allow_uninit && $optional && !$db[$idx][2]) {
154		error("optional var not initialized: {$db[$idx][0]} [".($idx+1).']', 1);
155	}
156}
157
158/** fetch params passed to zend_parse_params*() */
159function get_params($vars, $str)
160{
161	$ret = array();
162	preg_match_all('/(?:\([^)]+\))?(&?)([\w>.()-]+(?:\[\w+\])?)\s*,?((?:\)*\s*=)?)/S', $str, $m, PREG_SET_ORDER);
163
164	foreach ($m as $x) {
165		$name = $x[2];
166
167		// little hack for last parameter
168		if (strpos($name, '(') === false) {
169			$name = rtrim($name, ')');
170		}
171
172		if (empty($vars[$name][0])) {
173			error("variable not found: '$name'", 3);
174			$ret[][] = '**dummy**';
175
176		} else {
177			$ret[] = array($name, $vars[$name][0] . ($x[1] ? '*' : ''), $vars[$name][1]);
178		}
179
180		// the end (yes, this is a little hack :P)
181		if ($x[3]) {
182			break;
183		}
184	}
185
186//	if ($GLOBALS['current_function'] == 'for_debugging') { var_dump($m); var_dump($ret); }
187	return $ret;
188}
189
190
191/** run tests on a function. the code is passed in $txt */
192function check_function($name, $txt, $offset)
193{
194	global $API_params;
195
196	$regex = '/
197		(?: zend_parse_parameters(?:_throw)?               \s*\([^,]+
198		|   zend_parse_(?:parameters_ex|method_parameters) \s*\([^,]+,[^,]+
199		|   zend_parse_method_parameters_ex                \s*\([^,]+,[^,]+,[^,+]
200		)
201		,\s*"([^"]*)"\s*
202		,\s*([^{;]*)
203	/Sx';
204	if (preg_match_all($regex, $txt, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
205
206		$GLOBALS['current_function'] = $name;
207
208		foreach ($matches as $m) {
209			$GLOBALS['error_few_vars_given'] = false;
210			update_lineno($offset + $m[2][1]);
211
212			$vars = get_vars(substr($txt, 0, $m[0][1])); // limit var search to current location
213			$params = get_params($vars, $m[2][0]);
214			$optional = $varargs = false;
215			$last_char = '';
216			$j = -1;
217
218			$spec = $m[1][0];
219			$len = strlen($spec);
220			for ($i = 0; $i < $len; ++$i) {
221				$char = $spec[$i];
222				switch ($char = $spec[$i]) {
223					// separator for optional parameters
224					case '|':
225						if ($optional) {
226							error("more than one optional separator at char #$i");
227						} else {
228							$optional = true;
229							if ($i == $len-1) {
230								error("unnecessary optional separator");
231							}
232						}
233					break;
234
235					// separate_zval_if_not_ref
236					case '/':
237						if (in_array($last_char, array('l', 'L', 'd', 'b'))) {
238							error("the '/' specifier should not be applied to '$last_char'");
239						}
240					break;
241
242					// nullable arguments
243					case '!':
244						if (in_array($last_char, array('l', 'L', 'd', 'b'))) {
245							check_param($params, ++$j, 'zend_bool*', $optional);
246						}
247					break;
248
249					// variadic arguments
250					case '+':
251					case '*':
252						if ($varargs) {
253							error("A varargs specifier can only be used once. repeated char at column $i");
254						} else {
255							check_param($params, ++$j, 'zval**', $optional);
256							check_param($params, ++$j, 'int*', $optional);
257							$varargs = true;
258						}
259					break;
260
261					case 's':
262					case 'p':
263						check_param($params, ++$j, 'char**', $optional, $allow_uninit=true);
264						check_param($params, ++$j, 'size_t*', $optional, $allow_uninit=true);
265						if ($optional && !$params[$j-1][2] && !$params[$j][2]
266								&& $params[$j-1][0] !== '**dummy**' && $params[$j][0] !== '**dummy**') {
267							error("one of optional vars {$params[$j-1][0]} or {$params[$j][0]} must be initialized", 1);
268						}
269					break;
270
271					case 'C':
272						// C must always be initialized, independently of whether it's optional
273						check_param($params, ++$j, 'zend_class_entry**', false);
274					break;
275
276					default:
277						if (!isset($API_params[$char])) {
278							error("unknown char ('$char') at column $i");
279						}
280
281						// If an is_null flag is in use, only that flag is required to be
282						// initialized
283						$allow_uninit = $i+1 < $len && $spec[$i+1] === '!'
284								&& in_array($char, array('l', 'L', 'd', 'b'));
285
286						foreach ($API_params[$char] as $exp) {
287							check_param($params, ++$j, $exp, $optional, $allow_uninit);
288						}
289				}
290
291				$last_char = $char;
292			}
293		}
294	}
295}
296
297
298/** the main recursion function. splits files in functions and calls the other functions */
299function recurse($path)
300{
301	foreach (scandir($path) as $file) {
302		if ($file == '.' || $file == '..' || $file == 'CVS') continue;
303
304		$file = "$path/$file";
305		if (is_dir($file)) {
306			recurse($file);
307			continue;
308		}
309
310		// parse only .c and .cpp files
311		if (substr_compare($file, '.c', -2) && substr_compare($file, '.cpp', -4)) continue;
312
313		$txt = file_get_contents($file);
314		// remove comments (but preserve the number of lines)
315		$txt = preg_replace('@//.*@S', '', $txt);
316		$txt = preg_replace_callback('@/\*.*\*/@SsU', function($matches) {
317			return preg_replace("/[^\r\n]+/S", "", $matches[0]);
318		}, $txt);
319
320		$split = preg_split('/PHP_(?:NAMED_)?(?:FUNCTION|METHOD)\s*\((\w+(?:,\s*\w+)?)\)/S', $txt, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_OFFSET_CAPTURE);
321
322		if (count($split) < 2) continue; // no functions defined on this file
323		array_shift($split); // the first part isn't relevant
324
325
326		// generate the line offsets array
327		$j = 0;
328		$lines = preg_split("/(\r\n?|\n)/S", $txt, -1, PREG_SPLIT_DELIM_CAPTURE);
329		$lines_offset = array();
330
331		for ($i = 0; $i < count($lines); ++$i) {
332			$j += strlen($lines[$i]) + strlen(@$lines[++$i]);
333			$lines_offset[] = $j;
334		}
335
336		$GLOBALS['lines_offset'] = $lines_offset;
337		$GLOBALS['current_file'] = $file;
338
339
340		for ($i = 0; $i < count($split); $i+=2) {
341			// if the /* }}} */ comment is found use it to reduce false positives
342			// TODO: check the other indexes
343			list($f) = preg_split('@/\*\s*}}}\s*\*/@S', $split[$i+1][0]);
344			check_function(preg_replace('/\s*,\s*/S', '::', $split[$i][0]), $f, $split[$i][1]);
345		}
346	}
347}
348
349$dirs = array();
350
351if (isset($argc) && $argc > 1) {
352	if ($argv[1] == '-h' || $argv[1] == '-help' || $argv[1] == '--help') {
353		echo <<<HELP
354Synopsis:
355    php check_parameters.php [directories]
356
357HELP;
358		exit(0);
359	}
360	for ($i = 1; $i < $argc; $i++) {
361		$dirs[] = $argv[$i];
362	}
363} else {
364	$dirs[] = PHPDIR;
365}
366
367foreach($dirs as $dir) {
368	if (is_dir($dir)) {
369		if (!is_readable($dir)) {
370			echo "ERROR: directory '", $dir ,"' is not readable\n";
371			exit(1);
372		}
373	} else {
374		echo "ERROR: bogus directory '", $dir ,"'\n";
375		exit(1);
376	}
377}
378
379foreach ($dirs as $dir) {
380	recurse(realpath($dir));
381}
382