From 9d5802cb9f0897e98c018420f4d2db11547ec43e Mon Sep 17 00:00:00 2001
From: epriestley <git@epriestley.com>
Date: Wed, 17 Feb 2021 12:42:09 -0800
Subject: [PATCH] Provide some "preg_*" wrappers which raise exceptions on
 failure

Summary: Ref T13608. Ref T13100. Ref T13586. Properly checking "preg_match()" and similar calls for failure and raising useful exceptions is complicated and error-prone. Provide wrapper functions with an API that's more consistent with the rest of the codebase: matches are returned; and errors raise detailed exceptions.

Test Plan: See next change.

Maniphest Tasks: T13608, T13586, T13100

Differential Revision: https://secure.phabricator.com/D21561
---
 src/__phutil_library_map__.php         |  5 ++
 src/exception/PhutilRegexException.php |  3 +
 src/utils/utils.php                    | 86 ++++++++++++++++++++++++++
 3 files changed, 94 insertions(+)
 create mode 100644 src/exception/PhutilRegexException.php

diff --git a/src/__phutil_library_map__.php b/src/__phutil_library_map__.php
index 0bd7549a..8c46addd 100644
--- a/src/__phutil_library_map__.php
+++ b/src/__phutil_library_map__.php
@@ -849,6 +849,7 @@ phutil_register_library_map(array(
     'PhutilRawEnglishLocale' => 'internationalization/locales/PhutilRawEnglishLocale.php',
     'PhutilReadableSerializer' => 'readableserializer/PhutilReadableSerializer.php',
     'PhutilReadableSerializerTestCase' => 'readableserializer/__tests__/PhutilReadableSerializerTestCase.php',
+    'PhutilRegexException' => 'exception/PhutilRegexException.php',
     'PhutilRope' => 'utils/PhutilRope.php',
     'PhutilRopeTestCase' => 'utils/__tests__/PhutilRopeTestCase.php',
     'PhutilServiceProfiler' => 'serviceprofiler/PhutilServiceProfiler.php',
@@ -1009,6 +1010,9 @@ phutil_register_library_map(array(
     'phutil_partition' => 'utils/utils.php',
     'phutil_passthru' => 'future/exec/execx.php',
     'phutil_person' => 'internationalization/pht.php',
+    'phutil_preg_match' => 'utils/utils.php',
+    'phutil_preg_match_all' => 'utils/utils.php',
+    'phutil_raise_preg_exception' => 'utils/utils.php',
     'phutil_register_library' => 'init/lib/core.php',
     'phutil_register_library_map' => 'init/lib/core.php',
     'phutil_set_system_locale' => 'utils/utf8.php',
@@ -1925,6 +1929,7 @@ phutil_register_library_map(array(
     'PhutilRawEnglishLocale' => 'PhutilLocale',
     'PhutilReadableSerializer' => 'Phobject',
     'PhutilReadableSerializerTestCase' => 'PhutilTestCase',
+    'PhutilRegexException' => 'Exception',
     'PhutilRope' => 'Phobject',
     'PhutilRopeTestCase' => 'PhutilTestCase',
     'PhutilServiceProfiler' => 'Phobject',
diff --git a/src/exception/PhutilRegexException.php b/src/exception/PhutilRegexException.php
new file mode 100644
index 00000000..90b1b3bf
--- /dev/null
+++ b/src/exception/PhutilRegexException.php
@@ -0,0 +1,3 @@
+<?php
+
+final class PhutilRegexException extends Exception {}
diff --git a/src/utils/utils.php b/src/utils/utils.php
index 3ae606d7..cbc4d5fb 100644
--- a/src/utils/utils.php
+++ b/src/utils/utils.php
@@ -2008,3 +2008,89 @@ function phutil_partition(array $map) {
 
   return $partitions;
 }
+
+function phutil_preg_match(
+  $pattern,
+  $subject,
+  $flags = 0,
+  $offset = 0) {
+
+  $matches = null;
+  $result = @preg_match($pattern, $subject, $matches, $flags, $offset);
+  if ($result === false || $result === null) {
+    phutil_raise_preg_exception(
+      'preg_match',
+      array(
+        $pattern,
+        $subject,
+        $matches,
+        $flags,
+        $offset,
+      ));
+  }
+
+  return $matches;
+}
+
+function phutil_preg_match_all(
+  $pattern,
+  $subject,
+  $flags = 0,
+  $offset = 0) {
+
+  $matches = null;
+  $result = @preg_match_all($pattern, $subject, $matches, $flags, $offset);
+  if ($result === false || $result === null) {
+    phutil_raise_preg_exception(
+      'preg_match_all',
+      array(
+        $pattern,
+        $subject,
+        $matches,
+        $flags,
+        $offset,
+      ));
+  }
+
+  return $matches;
+}
+
+function phutil_raise_preg_exception($function, array $argv) {
+  $trap = new PhutilErrorTrap();
+
+  // NOTE: This ugly construction to avoid issues with reference behavior when
+  // passing values through "call_user_func_array()".
+
+  switch ($function) {
+    case 'preg_match':
+      @preg_match($argv[0], $argv[1], $argv[2], $argv[3], $argv[4]);
+      break;
+    case 'preg_match_all':
+      @preg_match_all($argv[0], $argv[1], $argv[2], $argv[3], $argv[4]);
+      break;
+  }
+  $error_message = $trap->getErrorsAsString();
+
+  $trap->destroy();
+
+  $pattern = $argv[0];
+  $pattern_display = sprintf(
+    '"%s"',
+    addcslashes($pattern, '\\\"'));
+
+  $message = array();
+  $message[] = pht(
+    'Call to %s(%s, ...) failed.',
+    $function,
+    $pattern_display);
+
+  if (strlen($error_message)) {
+    $message[] = pht(
+      'Regular expression engine emitted message: %s',
+      $error_message);
+  }
+
+  $message = implode("\n\n", $message);
+
+  throw new PhutilRegexException($message);
+}