From e874f3fb17c4d589864db72c7f81a5a6c8c37eb3 Mon Sep 17 00:00:00 2001
From: Thiemo Kreuz <thiemo.kreuz@wikimedia.de>
Date: Thu, 14 May 2020 19:44:30 +0200
Subject: [PATCH] Make WebRequest::getAcceptLang() return floats instead of
 strings

This is a follow-up for Ifcf0bc4.

I checked the callers of this method. The majority of them does not
care about these numeric values, only about the language codes. Most
callers even do an array_keys() immediatelly. The only effect I could
find will be visible in the ApiQueryUserInfo API module. Example:
https://en.wikipedia.org/wiki/Special:ApiSandbox#action=query&meta=userinfo&formatversion=2&uiprop=acceptlang
In this response, the numbers are currently reported as strings, but
will be reported as floats. I would argue this counts as a bugfix and
doesn't need anouncement.

This patch also splits the responsible regular expression pattern for
improved readability, utilizing the /x extended mode. No change is
made to the pattern.

Change-Id: I42de46f56d7544f9781b02a2f0ed49ef85ee44a1
---
 includes/WebRequest.php                   | 16 ++++++++++++---
 tests/phpunit/includes/WebRequestTest.php | 24 +++++++++++------------
 2 files changed, 25 insertions(+), 15 deletions(-)

diff --git a/includes/WebRequest.php b/includes/WebRequest.php
index 4337a6528cc..ac55308af17 100644
--- a/includes/WebRequest.php
+++ b/includes/WebRequest.php
@@ -1218,7 +1218,17 @@ class WebRequest {
 
 		// Break up string into pieces (languages and q factors)
 		if ( !preg_match_all(
-			'/([a-z]{1,8}(?:-[a-z]{1,8})*|\*)\s*(?:;\s*q\s*=\s*(1(?:\.0{0,3})?|0(?:\.[0-9]{0,3})?)?)?/',
+			'/
+				# a language code or a star is required
+				([a-z]{1,8}(?:-[a-z]{1,8})*|\*)
+				# from here everything is optional
+				\s*
+				(?:
+					# this accepts only numbers in the range ;q=0.000 to ;q=1.000
+					;\s*q\s*=\s*
+					(1(?:\.0{0,3})?|0(?:\.\d{0,3})?)?
+				)?
+			/x',
 			$acceptLang,
 			$matches,
 			PREG_SET_ORDER
@@ -1231,8 +1241,8 @@ class WebRequest {
 		foreach ( $matches as $match ) {
 			$languageCode = $match[1];
 			// When not present, the default value is 1
-			$qValue = $match[2] ?? 1;
-			if ( $qValue > 0 ) {
+			$qValue = (float)( $match[2] ?? 1.0 );
+			if ( $qValue ) {
 				$langs[$languageCode] = $qValue;
 			}
 		}
diff --git a/tests/phpunit/includes/WebRequestTest.php b/tests/phpunit/includes/WebRequestTest.php
index e996ef832b2..7242e58dbea 100644
--- a/tests/phpunit/includes/WebRequestTest.php
+++ b/tests/phpunit/includes/WebRequestTest.php
@@ -604,33 +604,33 @@ class WebRequestTest extends MediaWikiIntegrationTestCase {
 	public static function provideLanguageData() {
 		return [
 			[ '', [], 'Empty Accept-Language header' ],
-			[ 'en', [ 'en' => 1 ], 'One language' ],
-			[ 'en;q=', [ 'en' => 1 ], 'Empty q= defaults to 1' ],
+			[ 'en', [ 'en' => 1.0 ], 'One language' ],
+			[ 'en;q=', [ 'en' => 1.0 ], 'Empty q= defaults to 1' ],
 			[ 'en;q=0, de;q=0. pt;q=0.0 it;q=0.0000', [], 'Zeros to be skipped' ],
-			[ 'EN;Q=1.0009', [ 'en' => '1.000' ], 'Limited to max. 3 decimal places' ],
-			[ 'en, ar', [ 'en' => 1, 'ar' => 1 ], 'Two languages listed in appearance order.' ],
+			[ 'EN;Q=1.0009', [ 'en' => 1.000 ], 'Limited to max. 3 decimal places' ],
+			[ 'en, ar', [ 'en' => 1.0, 'ar' => 1.0 ], 'Two languages listed in appearance order.' ],
 			[
 				'zh-cn,zh-tw',
-				[ 'zh-cn' => 1, 'zh-tw' => 1 ],
+				[ 'zh-cn' => 1.0, 'zh-tw' => 1.0 ],
 				'Two equally prefered languages, listed in appearance order per rfc3282. Checks c9119'
 			],
 			[
 				'es, en; q=0.5',
-				[ 'es' => 1, 'en' => '0.5' ],
+				[ 'es' => 1.0, 'en' => 0.5 ],
 				'Spanish as first language and English and second'
 			],
-			[ 'en; q=0.5, es', [ 'es' => 1, 'en' => '0.5' ], 'Less prefered language first' ],
-			[ 'fr, en; q=0.5, es', [ 'fr' => 1, 'es' => 1, 'en' => '0.5' ], 'Three languages' ],
-			[ 'en; q=0.5, es', [ 'es' => 1, 'en' => '0.5' ], 'Two languages' ],
-			[ 'en, zh;q=0', [ 'en' => 1 ], "It's Chinese to me" ],
+			[ 'en; q=0.5, es', [ 'es' => 1.0, 'en' => 0.5 ], 'Less prefered language first' ],
+			[ 'fr, en; q=0.5, es', [ 'fr' => 1.0, 'es' => 1.0, 'en' => 0.5 ], 'Three languages' ],
+			[ 'en; q=0.5, es', [ 'es' => 1.0, 'en' => 0.5 ], 'Two languages' ],
+			[ 'en, zh;q=0', [ 'en' => 1.0 ], "It's Chinese to me" ],
 			[
 				'es; q=1, pt;q=0.7, it; q=0.6, de; q=0.1, ru;q=0',
-				[ 'es' => '1', 'pt' => '0.7', 'it' => '0.6', 'de' => '0.1' ],
+				[ 'es' => 1.0, 'pt' => 0.7, 'it' => 0.6, 'de' => 0.1 ],
 				'Preference for Romance languages'
 			],
 			[
 				'en-gb, en-us; q=1',
-				[ 'en-gb' => 1, 'en-us' => '1' ],
+				[ 'en-gb' => 1.0, 'en-us' => 1.0 ],
 				'Two equally prefered English variants'
 			],
 			[ '_', [], 'Invalid input' ],