Fix deceptively-simple pattern in pure-Lua ustring

The pure-Lua ustring pattern matching functions short-circuit to the
much faster string library when the pattern would match the same against
the raw bytes.

A pattern like "[^a-z]" can match a partial UTF-8 character when applied
bytewise, and so must be detected as unsafe.

Let's also directly test the pure-Lua module, instead of me having to
comment out lines in Scribunto_LuaUstringLibrary::register() whenever I
want to test them.

Change-Id: I91ed3374aadfea379b9db2e13b4248ab20df509e

engines/LuaCommon/lualib/ustring/ustring.lua

@@ -735,6 +735,7 @@ end
 --  * If it contains any bytes over 0x7f. We could skip these if they're not
 --    inside brackets and aren't followed by quantifiers and aren't part of a
 --    '%b', but that's too complicated to check.
+--  * If it contains a negated character set.
 --  * If it contains "%a" or any of the other %-prefixed character sets except
 --    %z or %Z.
 --  * If it contains a '.' not followed by '*', '+', or '-'. A bare '.' or '.?'
@@ -747,6 +748,7 @@ end
 local function patternIsSimple( pattern )
 	return not (
 		S.find( pattern, '[\128-\255]' ) or
+		S.find( pattern, '%[%^' ) or
 		S.find( pattern, '%%[acdlpsuwxACDLPSUWX]' ) or
 		S.find( pattern, '%.[^*+-]' ) or
 		S.find( pattern, '()', 1, true )

tests/engines/LuaCommon/UstringLibraryPureLuaTest.php

@@ -0,0 +1,20 @@
+<?php
+
+require_once( __DIR__ . '/UstringLibraryTest.php' );
+
+class Scribunto_LuaUstringLibraryPureLuaTests extends Scribunto_LuaUstringLibraryTests {
+	function setUp() {
+		parent::setUp();
+
+		// Override mw.ustring with the pure-Lua version
+		$interpreter = $this->getEngine()->getInterpreter();
+		$interpreter->callFunction(
+			$interpreter->loadString( '
+				local ustring = require( "ustring" )
+				ustring.maxStringLength = mw.ustring.maxStringLength
+				ustring.maxPatternLength = mw.ustring.maxPatternLength
+				mw.ustring = ustring
+			', 'fortest' )
+		);
+	}
+}

tests/engines/LuaCommon/UstringLibraryTests.lua

@@ -425,6 +425,10 @@ return testframework.getTestProvider( {
 	  args = { "foo foofóó foófoó bar", '(f%a+)%1' },
 	  expect = { 12, 17, 'foó' }
 	},
+	{ name = 'find: deceptively-simple pattern', func = mw.ustring.find,
+	  args = { "fóó", '([^a-z])' },
+	  expect = { 2, 2, 'ó' }
+	},
 
 	{ name = 'match: (1)', func = mw.ustring.match,
 	  args = { "bar fóo bar", 'f%a+' },