From ba19a82c069ad22b9455803834dbbb4d87c7a811 Mon Sep 17 00:00:00 2001
From: Brad Jorsch <bjorsch@wikimedia.org>
Date: Thu, 24 Mar 2016 10:08:29 -0400
Subject: [PATCH] Add handling for PCRE errors in ustringGsub

Bug: T130823
Change-Id: I6fab71c82ddab92daf6b369cb9857d9892f2d246
---
 engines/LuaCommon/UstringLibrary.php          | 46 +++++++++++++++++++
 .../LuaCommon/UstringLibraryPureLuaTest.php   | 14 ++++++
 .../engines/LuaCommon/UstringLibraryTest.php  | 37 +++++++++++++++
 3 files changed, 97 insertions(+)

diff --git a/engines/LuaCommon/UstringLibrary.php b/engines/LuaCommon/UstringLibrary.php
index 6b8ec2c5..7511a1b6 100644
--- a/engines/LuaCommon/UstringLibrary.php
+++ b/engines/LuaCommon/UstringLibrary.php
@@ -695,6 +695,52 @@ class Scribunto_LuaUstringLibrary extends Scribunto_LuaLibraryBase {
 
 		$count = 0;
 		$s2 = preg_replace_callback( $re, $cb, $s, $n, $count );
+		if ( $s2 === null ) {
+			self::handlePCREError( preg_last_error(), $pattern );
+		}
 		return array( $s2, $count );
 	}
+
+	/**
+	 * Handle a PCRE error
+	 * @param int $error From preg_last_error()
+	 * @param string $pattern Pattern being matched
+	 * @throws Scribunto_LuaError
+	 */
+	private function handlePCREError( $error, $pattern ) {
+		$PREG_JIT_STACKLIMIT_ERROR = defined( 'PREG_JIT_STACKLIMIT_ERROR' )
+			? PREG_JIT_STACKLIMIT_ERROR
+			: 'PREG_JIT_STACKLIMIT_ERROR';
+
+		$error = preg_last_error();
+		switch ( $error ) {
+			case PREG_NO_ERROR:
+				// Huh?
+				break;
+			case PREG_INTERNAL_ERROR:
+				throw new Scribunto_LuaError( "PCRE internal error" );
+			case PREG_BACKTRACK_LIMIT_ERROR:
+				throw new Scribunto_LuaError(
+					"PCRE backtrack limit reached while matching pattern '$pattern'"
+				);
+			case PREG_RECURSION_LIMIT_ERROR:
+				throw new Scribunto_LuaError(
+					"PCRE recursion limit reached while matching pattern '$pattern'"
+				);
+			case PREG_BAD_UTF8_ERROR:
+				// Should have alreay been caught, but just in case
+				throw new Scribunto_LuaError( "PCRE bad UTF-8 error" );
+			case PREG_BAD_UTF8_OFFSET_ERROR:
+				// Shouldn't happen, but just in case
+				throw new Scribunto_LuaError( "PCRE bad UTF-8 offset error" );
+			case $PREG_JIT_STACKLIMIT_ERROR:
+				throw new Scribunto_LuaError(
+					"PCRE JIT stack limit reached while matching pattern '$pattern'"
+				);
+			default:
+				throw new Scribunto_LuaError(
+					"PCRE error code $error while matching pattern '$pattern'"
+				);
+		}
+	}
 }
diff --git a/tests/engines/LuaCommon/UstringLibraryPureLuaTest.php b/tests/engines/LuaCommon/UstringLibraryPureLuaTest.php
index ad55265f..89e3be68 100644
--- a/tests/engines/LuaCommon/UstringLibraryPureLuaTest.php
+++ b/tests/engines/LuaCommon/UstringLibraryPureLuaTest.php
@@ -18,4 +18,18 @@ class Scribunto_LuaUstringLibraryPureLuaTests extends Scribunto_LuaUstringLibrar
 			', 'fortest' )
 		);
 	}
+
+	/**
+	 * @dataProvider providePCREErrors
+	 */
+	public function testPCREErrors( $ini, $args, $error ) {
+		// Not applicable
+		$this->assertTrue( true );
+	}
+
+	public static function providePCREErrors() {
+		return array(
+			array( array(), array(), null ),
+		);
+	}
 }
diff --git a/tests/engines/LuaCommon/UstringLibraryTest.php b/tests/engines/LuaCommon/UstringLibraryTest.php
index 4c8ef81e..918c3846 100644
--- a/tests/engines/LuaCommon/UstringLibraryTest.php
+++ b/tests/engines/LuaCommon/UstringLibraryTest.php
@@ -60,6 +60,43 @@ class Scribunto_LuaUstringLibraryTests extends Scribunto_LuaEngineTestBase {
 		$this->assertSame( $expected, $actual );
 		$this->luaTestName = null;
 	}
+
+	/**
+	 * @dataProvider providePCREErrors
+	 */
+	public function testPCREErrors( $ini, $args, $error ) {
+		$reset = array();
+		foreach ( $ini as $key => $value ) {
+			$old = ini_set( $key, $value );
+			if ( $old === false ) {
+				$this->markTestSkipped( "Failed to set ini setting $key = $value" );
+			}
+			$reset[] = new ScopedCallback( 'ini_set', array( $key, $old ) );
+		}
+
+		$interpreter = $this->getEngine()->getInterpreter();
+		$func = $interpreter->loadString( 'return mw.ustring.gsub( ... )', 'fortest' );
+		try {
+			call_user_func_array(
+				array( $interpreter, 'callFunction' ),
+				array_merge( array( $func ), $args )
+			);
+			$this->fail( 'Expected exception not thrown' );
+		} catch ( Scribunto_LuaError $e ) {
+			$this->assertSame( $error, $e->getMessage() );
+		}
+	}
+
+	public static function providePCREErrors() {
+		return array(
+			array(
+				array( 'pcre.backtrack_limit' => 10 ),
+				array( 'zzzzzzzzzzzzzzzzzzzz', '^(.-)[abc]*$', '%1' ),
+				'Lua error: PCRE backtrack limit reached while matching pattern \'^(.-)[abc]*$\'.'
+			),
+			// @TODO: Figure out patterns that hit other PCRE limits
+		);
+	}
 }
 
 class UstringLibraryNormalizationTestProvider extends Scribunto_LuaDataProvider {