Differences

This shows you the differences between two versions of the page.

--- rfc:is_literal [2021/02/12 18:42] – Add notes about the implementation from Danack, GC Flags, and Future Scope craigfrancis
+++ rfc:is_literal [2021/05/02 16:38] – Add performance stats, and notes from Dan craigfrancis
@@ Line 1: / Line 1: @@
 ====== PHP RFC: Is Literal Check ======
-  * Version: 0.2
+  * Version: 0.6
   * Date: 2020-03-21
-  * Updated: 2020-12-22
+  * Updated: 2021-04-30
   * Author: Craig Francis, craig#at#craigfrancis.co.uk
   * Status: Draft
@@ Line 11: / Line 11: @@
 ===== Introduction =====
-Add an //is_literal()// function, so developers/frameworks can check if a given variable is **safe**.
+A new function, //is_literal(string $string)//, to identify variables that have been created from a programmer defined string.
-As in, at runtime, being able to check if a variable has been created by literals, defined within a PHP script, by a trusted developer.
+This takes the concept of "taint checking" and makes it simpler and stricter.
-This simple check can be used to warn or completely block SQL Injection, Command Line Injection, and many cases of HTML Injection (aka XSS).
+It does not allow a variable to be marked as untainted, and it does not allowing escaping (important).
-===== The Problem =====
+For example, a database library that supports parametrised queries at the driver level, today a programmer could use either of these:
-Escaping strings for SQL, HTML, Commands, etc is **very** error prone.
-The vast majority of programmers should never do this (mistakes will be made).
-Unsafe values (often user supplied) **must** be kept separate (e.g. parameterised SQL), or be processed by something that understands the context (e.g. a HTML Templating Engine).
-This is primarily for security reasons, but it can also cause data to be damaged (e.g. ASCII/UTF-8 issues).
-Take these mistakes, where the value has come from the user:
 <code php>
-echo "<img src=" . $url . " alt='' />";
+$db->query('SELECT * FROM users WHERE id = ' . $_GET['id']); // INSECURE
-</code>
-Flawed, and unfortunately very common, a classic XSS vulnerability.
+$db->query('SELECT * FROM users WHERE id = ?', [$_GET['id']]);
-<code php>
-echo "<img src=" . htmlentities($url) . " alt='' />";
 </code>
-Flawed because the attribute value is not quoted, e.g. //$url = '/ onerror=alert(1)'//
+By rejecting the SQL that was not written as a literal (first example), and only accepting a literal string (written by the programmer), the library can provide an "inherently safe API".
-<code php>
+This definition of an "inherently safe API" comes from Christoph Kern, who did a talk in 2016 about [[https://www.youtube.com/watch?v=ccfEu-Jj0as|Preventing Security Bugs through Software Design]] (also at [[https://www.usenix.org/conference/usenixsecurity15/symposium-program/presentation/kern|USENIX Security 2015]]), which covers how this is used at Google. The idea is that we "Don't Blame the Developer, Blame the API"; where we need to put the burden on libraries (written once, used by many) to ensure that it's impossible for the developer to make these mistakes.
-echo "<img src='" . htmlentities($url) . "' alt='' />";
-</code>
-Flawed because //htmlentities()// does not encode single quotes by default, e.g. //$url = "/' onerror='alert(1)"//
+By adding a way for libraries to check if the strings they receive came from the developer (from trusted PHP source code), it allows the library to check they are being used in a safe way.
-<code php>
+===== Why =====
-echo '<a href="' . htmlentities($url) . '">Link</a>';
-</code>
-Flawed because a link can include JavaScript, e.g. //$url = 'javascript:alert(1)'//
+The [[https://owasp.org/www-project-top-ten/|OWASP Top 10]] lists common vulnerabilities sorted by prevalence, exploitability, detectability, and impact.
-<code html>
+**Injection vulnerabilities** remain at the top of the list (common prevalence, easy for attackers to detect/exploit, severe impact); and **XSS** comes in at 7 (widespread prevalence, easy for attackers to detect/exploit, moderate impact).
-<script>
-  var url = "<?= addslashes($url) ?>";
-</script>
-</code>
-Flawed because //addslashes()// is not HTML context aware, e.g. //$url = '</script><script>alert(1)</script>'//
+This is because injection mistakes are very easy to make, and hard to identify - is_literal() addresses this.
-<code php>
+===== Examples =====
-echo '<a href="/path/?name=' . htmlentities($name) . '">Link</a>';
-</code>
-Flawed because //urlencode()// has not been used, e.g. //$name = 'A&B'//
+The [[https://www.doctrine-project.org/projects/doctrine-orm/en/current/reference/query-builder.html#high-level-api-methods|Doctrine Query Builder]] allows a custom WHERE clause to be provided as a string. This is intended for use with literals and placeholders, but does not protect against this simple mistake:
-<code html>
-<p><?= htmlentities($url) ?></p>
-</code>
-Flawed because the encoding is not guaranteed to be UTF-8 (or ISO-8859-1 before PHP 5.4), so the value could be corrupted.
-Also flawed because some browsers (e.g. IE 11), if the charset isn't defined (header or meta tag), could guess the output as UTF-7, e.g. //$url = '+ADw-script+AD4-alert(1)+ADw-+AC8-script+AD4-'//
 <code php>
-example.html:
+// INSECURE
-    <img src={{ url }} alt='' />
+$qb->select('u')
+   ->from('User', 'u')
-$loader = new \Twig\Loader\FilesystemLoader('./templates/');
+   ->where('u.id = ' . $_GET['id'])
-$twig = new \Twig\Environment($loader, ['autoescape' => 'name']);
-echo $twig->render('example.html', ['url' => $url]);
 </code>
-Flawed because Twig is not context aware (in this case, an unquoted HTML attribute), e.g. //$url = '/ onerror=alert(1)'//
+The definition of the //where()// method could check with //is_literal()// and throw an exception, advising the programmer to replace it with a safer use of placeholders:
 <code php>
-$sql = 'SELECT 1 FROM user WHERE id=' . $mysqli->escape_string($id);
+$qb->select('u')
+   ->from('User', 'u')
+   ->where('u.id = :identifier')
+   ->setParameter('identifier', $_GET['id']);
 </code>
-Flawed because the value has not been quoted, e.g. //$id = 'id', or '1 OR 1=1'//
+Similarly, Twig allows [[https://twig.symfony.com/doc/2.x/recipes.html#loading-a-template-from-a-string|loading a template from a string]], which could allow accidentally skipping the default escaping functionality:
 <code php>
-$sql = 'SELECT 1 FROM user WHERE id="' . $mysqli->escape_string($id) . '"';
+// INSECURE
+echo $twig->createTemplate('<p>Hi ' . $_GET['name'] . '</p>')->render();
 </code>
-Flawed if 'sql_mode' includes //NO_BACKSLASH_ESCAPES//, e.g. //$id = '2" or "1"="1'//
+If //createTemplate()// checked with //is_literal()//, the programmer could be advised to write this instead:
 <code php>
-$sql = 'INSERT INTO user (name) VALUES ("' . $mysqli->escape_string($name) . '")';
+echo $twig->createTemplate('<p>Hi {{ name }}</p>')->render(['name' => $_GET['name']]);
 </code>
-Flawed if 'SET NAMES latin1' has been used, and escape_string() uses 'utf8'.
+===== Failed Solutions =====
-<code php>
+==== Education ====
-$parameters = "-f$email";
-// $parameters = '-f' . escapeshellarg($email);
+Developer training has not worked, it simply does not scale (people start programming every day), and learning about every single issue is difficult.
-mail('a@example.com', 'Subject', 'Message', NULL, $parameters);
+Keeping in mind that programmers will frequently do just enough to complete their task (busy), where they often copy/paste a solution to their problem they find online (risky), modify it for their needs (risky), then move on.
-</code>
-Flawed because it's not possible to safely escape values in //$additional_parameters// for //mail()//, e.g. //$email = 'b@example.com -X/www/example.php'//
+We cannot keep saying they 'need to be careful', and relying on them to never make a mistake.
-===== Previous Solutions =====
+==== Escaping ====
-[[https://github.com/laruence/taint|Taint extension]] by Xinchen Hui, but this approach explicitly allows escaping, which doesn't address the issues listed above.
+Escaping is hard, and error prone.
-[[https://wiki.php.net/rfc/sql_injection_protection|Automatic SQL Injection Protection]] by Matt Tait, where it was noted:
+We have a list of common [[https://github.com/craigfrancis/php-is-literal-rfc/blob/main/justification.md#common-mistakes|escaping mistakes]].
-  * "unfiltered input can affect way more than only SQL" ([[https://news-web.php.net/php.internals/87355|Pierre Joye]]);
+Developers should use parameterised queries (e.g. SQL), or a well tested library that knows how to escape values based on their context (e.g. HTML).
-  * this amount of work isn't ideal for "just for one use case" ([[https://news-web.php.net/php.internals/87647|Julien Pauli]]);
-  * It would have effected every SQL function, such as //mysqli_query()//, //$pdo->query()//, //odbc_exec()//, etc (concerns raised by [[https://news-web.php.net/php.internals/87436|Lester Caine]] and [[https://news-web.php.net/php.internals/87650|Anthony Ferrara]]);
-  * Each of those functions would need a bypass for cases where unsafe SQL was intentionally being used (e.g. phpMyAdmin taking SQL from POST data) because some applications intentionally "pass raw, user submitted, SQL" (Ronald Chmara [[https://news-web.php.net/php.internals/87406|1]]/[[https://news-web.php.net/php.internals/87446|2]]).
-I also agree that "SQL injection is almost a solved problem [by using] prepared statements" ([[https://news-web.php.net/php.internals/87400|Scott Arciszewski]]), but we still need something to identify mistakes.
+==== Taint Checking ====
-===== Related Go Implementation =====
+Some languages implement a "taint flag" which tracks whether values are considered "safe".
-As discussed by [[https://www.usenix.org/conference/usenixsecurity15/symposium-program/presentation/kern|Christoph Kern (Google) in 2015]] and in [[https://www.youtube.com/watch?v=ccfEu-Jj0as|2016]], this approach works.
+There is a [[https://github.com/laruence/taint|Taint extension for PHP]] by Xinchen Hui, and [[https://wiki.php.net/rfc/taint|a previous RFC proposing it be added to the language]].
-The Go language can do this by checking for "compile time constants", which isn't as good as a run time solution (e.g. the //WHERE IN// issue), but it does work.
+These solutions rely on the assumption that the output of an escaping function is safe for a particular context. This sounds reasonable in theory, but the operation of escaping functions, and the context for which their output is safe, are very hard to define. This leads to a feature that is both complex and unreliable.
-Google currently avoids most issues by using [[https://blogtitle.github.io/go-safe-html/|go-safe-html]] and [[https://github.com/google/go-safeweb/tree/master/safesql|safesql]].
+This proposal avoids the complexity by addressing a different part of the problem: separating inputs supplied by the programmer, from inputs supplied by the user.
-===== Related JavaScript Implementation =====
+==== Static Analysis ====
-This RFC is taking some ideas from TC39, where a similar idea is being discussed for JavaScript, to support the introduction of Trusted Types.
+While I agree with [[https://news-web.php.net/php.internals/109192|Tyson Andre]], it is highly recommended to use Static Analysis.
-https://github.com/tc39/proposal-array-is-template-object\\
+But they nearly always focus on other issues (type checking, basic logic flaws, code formatting, etc).
-https://github.com/mikewest/tc39-proposal-literals
-They are looking at "Distinguishing strings from a trusted developer, from strings that may be attacker controlled".
+Those that attempt to address injection vulnerabilities, do so via Taint Checking (see above), and are [[https://github.com/vimeo/psalm/commit/2122e4a1756dac68a83ec3f5abfbc60331630781|often incomplete]].
-===== Solution =====
+For a quick example, psalm, even in its strictest errorLevel (1), and/or running //--taint-analysis//, will not notice the missing quote marks in this SQL, and will incorrectly assume this is perfectly safe:
-Literals are safe values, defined within the PHP scripts, for example:
 <code php>
-$a = 'Example';
+$db = new mysqli('...');
-is_literal($a); // true
-$a = 'Example ' . $a . ', ' . 5;
-is_literal($a); // true
-$a = 'Example ' . $_GET['id'];
-is_literal($a); // false
-$a = 'Example ' . time();
-is_literal($a); // false
-$a = sprintf('LIMIT %d', 3);
-is_literal($a); // false
-$c = count($ids);
+$id = (string) ($_GET['id'] ?? 'id'); // Keep the type checker happy.
-$a = 'WHERE id IN (' . implode(',', array_fill(0, $c, '?')) . ')';
-is_literal($a); // true, the odd one that involves functions.
-$limit = 10;
+$db->prepare('SELECT * FROM users WHERE id = ' . $db->real_escape_string($id));
-$a = 'LIMIT ' . ($limit + 1);
-is_literal($a); // false, but might need some discussion.
 </code>
-This uses a similar definition of [[https://wiki.php.net/rfc/sql_injection_protection#safeconst|SafeConst]] from Matt Tait's RFC, but it doesn't need to accept Integer or FloatingPoint variables as safe (unless it makes the implementation easier), nor should this proposal effect any existing functions.
+When psalm comes to taint checking the usage of a library (like Doctrine), it assumes all methods are safe, because none of them note the sinks (and even if they did, you're back to escaping being an issue).
-Thanks to [[https://chat.stackoverflow.com/transcript/message/51565346#51565346|NikiC]], it looks like we can reuse the GC_PROTECTED flag for strings.
+But the biggest problem is that Static Analysis is simply not used by most developers, especially those who are new to programming (usage tends to be higher by those writing well tested libraries).
-As an aside, [[https://news-web.php.net/php.internals/87396|Xinchen Hui]] found the Taint extension was complex in PHP5, but "with PHP7's new zend_string, and string flags, the implementation will become easier". Also, [[https://chat.stackoverflow.com/transcript/message/48927813#48927813|MarkR]] suggested that it might be possible to use the fact that "interned strings in PHP have a flag", which is there because these "can't be freed".
+===== Proposal =====
-Commands can be checked to ensure they are a "programmer supplied constant/static/validated string", and all other unsafe variables are provided separately (as noted by [[https://news-web.php.net/php.internals/87725|Yasuo Ohgaki]]).
+This RFC proposes adding three functions:
-This approach allows all systems/frameworks to decide if they want to **block**, **educate** (via a notice), or **ignore** these issues (to avoid the "don't nanny" concern raised by [[https://news-web.php.net/php.internals/87383|Lester Caine]]).
+  * //is_literal(string $string): bool// to check if a variable represents a value written into the source code or not.
+  * //literal_combine(string $piece, string $pieces): string// to allow concatenating literal strings.
+  * //literal_implode(string $glue, array $pieces): string// to implode an array of literals, with a literal.
-Unlike the Taint extension, there must **not** be an equivalent //untaint()// function, or support any kind of escaping.
+A literal is defined as a value (string) which has been written by the programmer. The value may be passed between functions, as long as it is not modified in any way.
-==== Solution: SQL Injection ====
+<code php>
+is_literal('Example'); // true
-Database abstractions (e.g. ORMs) will be able to ensure they are provided with strings that are safe.
+$a = 'Hello';
+$b = 'World';
-[[https://www.doctrine-project.org/projects/doctrine-orm/en/2.7/reference/query-builder.html#high-level-api-methods|Doctrine]] could use this to ensure //->where($predicates)// is a literal:
+is_literal($a); // true
+is_literal($a . $b); // false, no concat at run time (details below).
-<code php>
+is_literal(literal_combine($a, $b)); // true
-$users = $queryBuilder
-  ->select('u')
-  ->from('User', 'u')
-  ->where('u.id = ' . $_GET['id'])
-  ->getQuery()
-  ->getResult();
-// example.php?id=u.id
+is_literal($_GET['id']); // false
+is_literal('WHERE id = ' . intval($_GET['id'])); // false
+is_literal(rand(0, 10)); // false
+is_literal(sprintf('LIMIT %d', 3)); // false, should use parameters
 </code>
-This mistake can be easily identified by:
+There is no way to manually mark a string as a literal (i.e. no equivalent to //untaint()//); as soon as the value has been manipulated in any way, it is no longer marked as a literal.
-<code php>
+*Technical detail: Strings that are concatenated in place at compile time are treated as a literal.*
-public function where($predicates)
-{
-    if (function_exists('is_literal') && !is_literal($predicates)) {
-        throw new Exception('->where() can only accept a literal');
-    }
-    ...
-}
-</code>
-[[https://redbeanphp.com/index.php?p=/finding|RedBean]] could check //$sql// is a literal:
+===== Previous Work =====
-<code php>
+Google uses "compile time constants" in Go, which isn't as good as a run time solution (e.g. the //WHERE IN// issue), but it works, and is used by [[https://blogtitle.github.io/go-safe-html/|go-safe-html]] and [[https://github.com/google/go-safeweb/tree/master/safesql|go-safesql]].
-$users = R::find('user', 'id = ' . $_GET['id']);
-</code>
-[[http://propelorm.org/Propel/reference/model-criteria.html#relational-api|PropelORM]] could check //$clause// is a literal:
+Google also uses [[https://errorprone.info/|Error Prone]] in Java to augment the compiler's type analysis, where [[https://errorprone.info/bugpattern/CompileTimeConstant|@CompileTimeConstant]] ensures method parameters can only use "compile-time constant expressions" (this isn't a complete solution either).
-<code php>
+Perl has a [[https://perldoc.perl.org/perlsec#Taint-mode|Taint Mode]], via the -T flag, where all input is marked as "tainted", and cannot be used by some methods (like commands that modify files), unless you use a regular expression to match and return known-good values (where regular expressions are easy to get wrong).
-$users = UserQuery::create()->where('id = ' . $_GET['id'])->find();
-</code>
-The //is_literal()// function could also be used internally by ORM developers, so they can be sure they have created their SQL strings out of literals. This would avoid mistakes such as the ORDER BY issues in the Zend framework [[https://framework.zend.com/security/advisory/ZF2014-04|1]]/[[https://framework.zend.com/security/advisory/ZF2016-03|2]].
+[[https://github.com/tc39/proposal-array-is-template-object|JavaScript might get isTemplateObject]] to "Distinguishing strings from a trusted developer from strings that may be attacker controlled" (intended to be [[https://github.com/mikewest/tc39-proposal-literals|used with Trusted Types]]).
-==== Solution: SQL Injection, Basic ====
+As noted above, there is the [[https://github.com/laruence/taint|Taint extension for PHP]] by Xinchen Hui.
-A simple example:
+And there is the [[https://wiki.php.net/rfc/sql_injection_protection|Automatic SQL Injection Protection]] RFC by Matt Tait, where this RFC uses a similar concept of the [[https://wiki.php.net/rfc/sql_injection_protection#safeconst|SafeConst]]. When Matt's RFC was being discussed, it was noted:
-<code php>
+  * "unfiltered input can affect way more than only SQL" ([[https://news-web.php.net/php.internals/87355|Pierre Joye]]);
-$sql = 'SELECT * FROM table WHERE id = ?';
+  * this amount of work isn't ideal for "just for one use case" ([[https://news-web.php.net/php.internals/87647|Julien Pauli]]);
+  * It would have effected every SQL function, such as //mysqli_query()//, //$pdo->query()//, //odbc_exec()//, etc (concerns raised by [[https://news-web.php.net/php.internals/87436|Lester Caine]] and [[https://news-web.php.net/php.internals/87650|Anthony Ferrara]]);
+  * Each of those functions would need a bypass for cases where unsafe SQL was intentionally being used (e.g. phpMyAdmin taking SQL from POST data) because some applications intentionally "pass raw, user submitted, SQL" (Ronald Chmara [[https://news-web.php.net/php.internals/87406|1]]/[[https://news-web.php.net/php.internals/87446|2]]).
-$result = $db->exec($sql, [$id]);
+I also agree that "SQL injection is almost a solved problem [by using] prepared statements" ([[https://news-web.php.net/php.internals/87400|Scott Arciszewski]]), and this is where //is_literal()// can be used to check that no mistakes are made when using prepared statements.
-</code>
-Checked in the framework by:
+===== Usage =====
+By libraries:
 <code php>
 class db {
+  protected $level = 2; // Probably should default to 1 at first.
-  public function exec($sql, $parameters = []) {
+  function literal_check($var) {
+    if (function_exists('is_literal') && !is_literal($var)) {
-    if (!is_literal($sql)) {
+      if ($this->level === 0) {
-      throw new Exception('SQL must be a literal.');
+        // Programmer aware, and is choosing to bypass this check.
+      } else if ($this->level === 1) {
+        trigger_error('Non-literal detected!', E_USER_WARNING);
+      } else {
+        throw new Exception('Non-literal detected!');
+      }
     }
-    $statement = $this->pdo->prepare($sql);
-    $statement->execute($parameters);
-    return $statement->fetchAll();
   }
+  function unsafe_disable_injection_protection() {
+    $this->level = 0;
+  }
+  function where($sql, $parameters = []) {
+    $this->literal_check($sql);
+    // ...
+  }
 }
-</code>
-This also works with string concatenation:
+$db->where('id = ?'); // OK
+$db->where('id = ' . $_GET['id']); // Exception thrown
-<code php>
-define('TABLE', 'example');
-$sql = 'SELECT * FROM ' . TABLE . ' WHERE id = ?';
-  is_literal($sql); // Returns true
-$sql .= ' AND id = ' . $mysqli->escape_string($_GET['id']);
-  is_literal($sql); // Returns false
 </code>
-==== Solution: SQL Injection, ORDER BY ====
+Table and Fields in SQL, which cannot use parameters; for example //ORDER BY//:
-To ensure //ORDER BY// can be set via the user, but only use acceptable values:
 <code php>
@@ Line 286: / Line 216: @@
 $order_id = array_search(($_GET['sort'] ?? NULL), $order_fields);
-$sql = ' ORDER BY ' . $order_fields[$order_id];
+$sql = literal_combine(' ORDER BY ', $order_fields[$order_id]);
 </code>
-==== Solution: SQL Injection, WHERE IN ====
+Undefined number of parameters; for example //WHERE IN//:
-Most SQL strings can be a simple concatenations of literal values, but //WHERE x IN (?,?,?)// needs to use a variable number of literal placeholders.
-There needs to be a special case for //array_fill()//+//implode()//, so the //is_literal// state can be preserved, allowing us to create the safe literal string '?,?,?':
 <code php>
-$in_sql = implode(',', array_fill(0, count($ids), '?'));
+function where_in_sql($count) { // Should check for 0
+  $sql = [];
-$sql = 'SELECT * FROM table WHERE id IN (' . $in_sql . ')';
+  for ($k = 0; $k < $count; $k++) {
+    $sql[] = '?';
+  }
+  return literal_implode(',', $sql);
+}
+$sql = literal_combine('WHERE id IN (', where_in_sql(count($ids)), ')');
 </code>
-==== Solution: CLI Injection ====
+===== Considerations =====
-Rather than using functions such as:
+==== Naming ====
-  * //exec()//
+Literal string is the standard name for strings in source code. See https://www.google.com/search?q=what+is+literal+string+in+php
-  * //shell_exec()//
-  * //system()//
-  * //passthru()//
-Frameworks (or PHP) could introduce something similar to //pcntl_exec()//, where arguments are provided separately.
+> A string literal is the notation for representing a string value within the text of a computer program. In PHP, strings can be created with single quotes, double quotes or using the heredoc or the nowdoc syntax. ... The heredoc preserves the line breaks and other whitespace (including indentation) in the text.
-Or, take a safe literal for the command, and use parameters for the arguments (like SQL does):
+Alternatives suggestions have included //is_from_literal()// from [[https://news-web.php.net/php.internals/109197|Jakob Givoni]]. I think //is_safe_string()// might be asking for trouble. Other terms have included "compile time constants" and "code string".
-<code php>
+==== Supporting Int/Float/Boolean values. ====
-$output = parameterised_exec('grep ? /path/to/file | wc -l', [
-    'example',
-  ]);
-</code>
-Rough implementation:
+When converting to string, they aren't guaranteed (and often don't) have the exact same value they have in source code.
-<code php>
+For example, //TRUE// and //true// when cast to string give "1".
-function parameterised_exec($cmd, $args = []) {
-  if (!is_literal($cmd)) {
+It's also a very low value feature, where there might not be space for a flag to be added.
-    throw new Exception('The first argument must be a literal');
-  }
-  $offset = 0;
+==== Supporting Concatenation ====
-  $k = 0;
-  while (($pos = strpos($cmd, '?', $offset)) !== false) {
-    if (!isset($args[$k])) {
-      throw new Exception('Missing parameter "' . ($k + 1) . '"');
-      exit();
-    }
-    $arg = escapeshellarg($args[$k]);
-    $cmd = substr($cmd, 0, $pos) . $arg . substr($cmd, ($pos + 1));
-    $offset = ($pos + strlen($arg));
-    $k++;
-  }
-  if (isset($args[$k])) {
-    throw new Exception('Unused parameter "' . ($k + 1) . '"');
-    exit();
-  }
-  return exec($cmd);
+Unfortunately early testing suggests there will be too much of a performance impact, and is why //literal_combine()// or //literal_implode()// exists.
-}
+It isn't needed for most libraries, like an ORM or Query Builder, where their methods nearly always take a small literal string.
-</code>
-==== Solution: HTML Injection ====
+It was considered because it would have made it easier for existing projects currently using string concatenation to adopt.
-Template engines should receive variables separately from the raw HTML.
+Joe Watkins has created a version that does support string concatenation, which we might re-consider at a later date (Joe's implementation also sets the literal flag, and was used as the basis for this implementation).
-Often the engine will get the HTML from static files (safe):
+Máté Kocsis did the [[https://github.com/craigfrancis/php-is-literal-rfc/blob/main/tests/results/with-concat/kocsismate.pdf|primary testing on the string concat version]], and found a 0.124% performance hit for the Laravel Demo app, 0.161% for Symfony, and a more severe -3.719% when running this [[https://github.com/kocsismate/php-version-benchmarks/blob/main/app/zend/concat.php#L25|concat test]].
-<code php>
+In my own [[https://github.com/craigfrancis/php-is-literal-rfc/tree/main/tests|simplistic testing]], the [[https://github.com/craigfrancis/php-is-literal-rfc/blob/main/tests/results/with-concat/local.pdf|results]] found a 0.42% performance hit for the Laravel Demo app, 0.40% for Symfony, and 1.81% when running my [[https://github.com/craigfrancis/php-is-literal-rfc/blob/main/tests/001.phpt|concat test]] via //./cli/php//.
-$html = file_get_contents('/path/to/template.html');
-</code>
-But small snippets of HTML are often easier to define as a literal within the PHP script:
+Dan Ackroyd also notes that the use of //literal_combine()// or //literal_implode()// will make it easier to identify exactly where mistakes are made, rather than it being picked up at the end of a potentially long script, after multiple string concatenations, e.g.
 <code php>
-$template_html = '
+$sortOrder = 'ASC';
-  <p>Hello <span id="username"></span></p>
-  <p><a>Website</a></p>';
-</code>
-Where the variables are supplied separately, in this example I'm using XPath:
+// 300 lines of code, or multiple function calls
-<code php>
+$sql .= ' ORDER BY name ' . $sortOrder;
-$values = [
-    '//span[@id="username"]' => [
-        NULL      => 'Name', // The textContent
-        'class'   => 'admin',
-        'data-id' => '123',
-      ],
-    '//a' => [
-        'href' => 'https://example.com',
-      ],
-  ];
-echo template_parse($template_html, $values);
+// 300 lines of code, or multiple function calls
+$db->query($sql);
 </code>
-The templating engine can then accept and apply the supplied variables for the relevant context.
+If a developer changed the literal //'ASC'// to //$_GET['order']//, the error raised by //$db->query()// would not be clear where the mistake was made. Whereas using //literal_combine()// highlights exactly where the issue happened:
-As a simple example, this can be done with:
 <code php>
-function template_parse($html, $values) {
+$sql = literal_combine($sql, ' ORDER BY name ', $sortOrder);
+</code>
-  if (!is_literal($html)) {
+==== Performance ====
-    throw new Exception('Invalid Template HTML.');
-  }
-  $dom = new DomDocument();
+The proposed implementation has:
-  $dom->loadHTML('<' . '?xml encoding="UTF-8">' . $html);
-  $xpath = new DOMXPath($dom);
+    [TODO]
-  foreach ($values as $query => $attributes) {
+Also, see the section above, where string concatenation support would have introduced a ~1% performance hit.
-    if (!is_literal($query)) {
+==== Values from INI/JSON/YAML ====
-      throw new Exception('Invalid Template XPath.');
-    }
-    foreach ($xpath->query($query) as $element) {
+As noted by [[https://news-web.php.net/php.internals/87667|Dennis Birkholz]], Systems/Frameworks that define certain variables (e.g. table name prefixes) without the use of a literal (e.g. ini/json/yaml files), might need to make some changes to use this feature (depending on where they use the is_literal check).
-      foreach ($attributes as $attribute => $value) {
-        if (!is_literal($attribute)) {
+==== Existing String Functions ====
-          throw new Exception('Invalid Template Attribute.');
-        }
-        if ($attribute) {
+Trying to determine if the //is_literal// flag should be passed through functions like //str_repeat()//, or //substr()// etc is difficult. Having a security feature be difficult to reason about, gives a much higher chance of making a mistake.
-          $safe = false;
-          if ($attribute == 'href') {
-            if (preg_match('/^https?:\/\//', $value)) {
-              $safe = true; // Not "javascript:..."
-            }
-          } else if ($attribute == 'class') {
-            if (in_array($value, ['admin', 'important'])) {
-              $safe = true; // Only allow specific classes?
-            }
-          } else if (preg_match('/^data-[a-z]+$/', $attribute)) {
-            if (preg_match('/^[a-z0-9 ]+$/i', $value)) {
-              $safe = true;
-            }
-          }
-          if ($safe) {
-            $element->setAttribute($attribute, $value);
-          }
-        } else {
-          $element->textContent = $value;
-        }
-      }
+For any use-case where dynamic strings are required, it would be better to build those strings with an appropriate query builder, or by using //literal_combine()/////literal_implode()//.
-    }
-  }
-  $html = '';
-  $body = $dom->documentElement->firstChild;
-  if ($body->hasChildNodes()) {
-    foreach ($body->childNodes as $node) {
-      $html .= $dom->saveXML($node);
-    }
-  }
-  return $html;
-}
-</code>
 ===== Backward Incompatible Changes =====
-None
+No known BC breaks, except for code-bases that already contain userland functions //is_literal()//, //literal_implode()// or //literal_combine()//.
 ===== Proposed PHP Version(s) =====
-PHP 8.1?
+PHP 8.1
 ===== RFC Impact =====
@@ Line 467: / Line 314: @@
 ==== To SAPIs ====
-Not sure
+None known
 ==== To Existing Extensions ====
@@ Line 479: / Line 326: @@
 ===== Open Issues =====
-On [[https://github.com/craigfrancis/php-is-literal-rfc/issues|GitHub]]:
+None
-  - Would this cause performance issues? Presumably not as bad a type checking.
-  - Can //array_fill()//+//implode()// pass though the "is_literal" flag for the "WHERE IN" case?
-  - Should the function be named //is_from_literal()//? (suggestion from [[https://news-web.php.net/php.internals/109197|Jakob Givoni]])
-  - Systems/Frameworks that define certain variables (e.g. table name prefixes) without the use of a literal (e.g. ini/json/yaml files), so they might need to make some changes to use this check, as originally noted by [[https://news-web.php.net/php.internals/87667|Dennis Birkholz]].
-===== Alternatives =====
-  - The current Taint Extension (notes above)
-  - Using static analysis (not at runtime), for example [[https://psalm.dev/|psalm]] (thanks [[https://news-web.php.net/php.internals/109192|Tyson Andre]]). But I can't find any which do these checks by default (if they even try), and we can't expect all programmers to use static analysis (especially those who have just stated).
 ===== Unaffected PHP Functionality =====
-Not sure
+None known
 ===== Future Scope =====
-As noted by [[https://chat.stackoverflow.com/transcript/message/51573226#51573226|MarkR]], the benefit will come when it can be used by PDO and similar functions (//mysqli_query//, //preg_match//, etc).
+As noted by MarkR, the biggest benefit will come when it can be used by PDO and similar functions (//mysqli_query//, //preg_match//, //exec//, etc). But the basic idea can be used immediately by frameworks and general abstraction libraries, and they can give feedback for future work.
-This check could be used to throw an exception, or generate an error/warning/notice, providing a way for PHP to teach new programmers, and/or completely block unsafe values in SQL, HTML, CLI, etc.
+**Phase 2** could introduce a way for programmers to specify certain PHP function/method arguments can only accept literals, and/or specific value-objects their project trusts (this idea comes from [[https://web.dev/trusted-types/|Trusted Types]] in JavaScript).
-PHP could also have a mode where output (e.g. //echo '<html>'//) is blocked, and this can be bypassed (maybe via //ini_set//) when the HTML Templating Engine has created the correctly encoded output.
+For example, a project could require the second argument for //pg_query()// only accept literals or their //query_builder// object (which provides a //__toString// method); and that any output (print, echo, readfile, etc) must use the //html_output// object that's returned by their trusted HTML Templating system (using //ob_start()// might be useful here).
-And maybe there could be a [[https://chat.stackoverflow.com/transcript/message/51573091#51573091|is_figurative()]] function, which [[https://chat.stackoverflow.com/transcript/message/48927770#48927770|MarkR]] seems to really want :-)
+**Phase 3** could set a default of 'only literals' for all of the relevant PHP function arguments, so developers are given a warning, and later prevented (via an exception), when they provide an unsafe value to those functions (they could still specify that unsafe values are allowed, e.g. phpMyAdmin).
+And, for a bit of silliness (Spaß ist verboten), MarkR would like a //is_figurative()// function (functionality to be confirmed).
 ===== Proposed Voting Choices =====
-N/A
+Accept the RFC. Yes/No
 ===== Patches and Tests =====
@@ Line 515: / Line 354: @@
 ===== Implementation =====
-[[https://github.com/Danack/|Danack]] has [[https://github.com/php/php-src/compare/master...Danack:is_literal_attempt_two|started an implementation]].
+Dan Ackroyd has [[https://github.com/php/php-src/compare/master...Danack:is_literal_attempt_two|started an implementation]], which uses functions like [[https://github.com/php/php-src/compare/master...Danack:is_literal_attempt_two#diff-2b0486443df74cd919c949f33f895eacf97c34b8490e7554e032e770ab11e4d8R2761|literal_combine()]] to avoid performance concerns.
+Joe Watkins has [[https://github.com/php/php-src/compare/master...krakjoe:literals|created an implementation]] which includes string concat. As noted above, the performance impact is probably too much for it to be accepted.
+===== References =====
+N/A
 ===== Rejected Features =====
 N/A
+===== Thanks =====
+  - **Dan Ackroyd**, DanAck, for starting the first implementation (which made this a reality), and followup on the version that uses functions instead of string concat.
+  - **Joe Watkins**, krakjoe, for finding how to set the literal flag (tricky), and creating the implementation that does support string concat.
+  - **Rowan Tommins**, IMSoP, for re-writing this RFC to focus on the key features, and putting it in context of how it can be used by libraries.
+  - **Nikita Popov**, NikiC, for suggesting where the literal flag could be stored. Initially this was going to be the "GC_PROTECTED flag for strings", which allowed Dan to start the first implementation.
+  - **Mark Randall**, MarkR, for alternative ideas, and noting that "interned strings in PHP have a flag", which started the conversation on how this could be implemented.
+  - **Xinchen Hui**, who created the Taint Extension, allowing me to test the idea; and noting how Taint in PHP5 was complex, but "with PHP7's new zend_string, and string flags, the implementation will become easier" [[https://news-web.php.net/php.internals/87396|source]].

Differences

Page Tools