From 445ba883a222938cecad9030687360a14390ba7e Mon Sep 17 00:00:00 2001
From: Sam Smith <phuedx@wikimedia.org>
Date: Wed, 30 Jun 2021 13:51:04 +0100
Subject: [PATCH] search: Disable query highlight for some languages

Following on from I551414b1, disable query highlighting for the list of
languages provided by @TJones in T281797.

The user's interface language can be different from the wiki's content
language and so the former is not available at configuration time. Thus,
we fetch the user's interface language at request time. Fortunately,
@TJones' list of languages is small so there should be little
perceivable performance impact from the perpective of the user.

Additional changes:

- Remove the config.VectorWvuiSearchOptions.value.highlightQuery
  property from skin.json

Bug: T281797
Change-Id: Ib39736a93fa64e82253f88551d125413e672558b
---
 includes/Hooks.php                    | 14 ++++++---
 includes/Services/LanguageService.php | 45 +++++++++++++++++++++++++++
 includes/VectorServices.php           | 10 ++++++
 skin.json                             |  3 +-
 4 files changed, 66 insertions(+), 6 deletions(-)
 create mode 100644 includes/Services/LanguageService.php

diff --git a/includes/Hooks.php b/includes/Hooks.php
index 332c6dff2..2fe4ca64b 100644
--- a/includes/Hooks.php
+++ b/includes/Hooks.php
@@ -53,16 +53,22 @@ class Hooks {
 	}
 
 	/**
-	 * Passes config variables to skins.vector.search ResourceLoader module.
+	 * Generates config variables for skins.vector.search Resource Loader module (defined in
+	 * skin.json).
+	 *
 	 * @param ResourceLoaderContext $context
 	 * @param Config $config
-	 * @return array
+	 * @return array<string,mixed>
 	 */
 	public static function getVectorWvuiSearchResourceLoaderConfig(
 		ResourceLoaderContext $context,
 		Config $config
-	) {
-		return $config->get( 'VectorWvuiSearchOptions' );
+	): array {
+		$result = $config->get( 'VectorWvuiSearchOptions' );
+		$result['highlightQuery'] =
+			VectorServices::getLanguageService()->canWordsBeSplitSafely( $context->getLanguage() );
+
+		return $result;
 	}
 
 	/**
diff --git a/includes/Services/LanguageService.php b/includes/Services/LanguageService.php
new file mode 100644
index 000000000..c750977ee
--- /dev/null
+++ b/includes/Services/LanguageService.php
@@ -0,0 +1,45 @@
+<?php
+
+namespace Vector\Services;
+
+class LanguageService {
+	/**
+	 * The list of language codes for those languages that the search autocomplete widget cannot
+	 * split a word on a Unicode code point followed by one or many combining marks (also code
+	 * points).
+	 *
+	 * This list was compiled by [@TJones](https://phabricator.wikimedia.org/p/TJones/) as part
+	 * of [T281797](https://phabricator.wikimedia.org/T281797).
+	 *
+	 * @var string[]
+	 */
+	private $splittableLanguages;
+
+	public function __construct() {
+		$this->splittableLanguages = [
+			'ar', 'ary', 'arz', 'ckb', 'fa', 'glk', 'ks', 'mzn', 'pnb', 'ps', 'sd', 'skr', 'ug', 'ur',
+			'as', 'bn', 'bpy',
+			'awa', 'bh', 'dty', 'gom', 'hi', 'ks', 'mai', 'mr', 'ne', 'new', 'pi', 'sa',
+			'gu',
+			'pa',
+			'kn', 'tcy',
+			'km',
+			'ml',
+			'or',
+			'si',
+			'ta',
+			'te',
+		];
+	}
+
+	/**
+	 * Gets whether or not we can split words arbitrarily, for example when highlighting the user's query in the search
+	 * autocomplete widget.
+	 *
+	 * @param string $code
+	 * @return bool
+	 */
+	public function canWordsBeSplitSafely( string $code ): bool {
+		return !in_array( $code, $this->splittableLanguages );
+	}
+}
diff --git a/includes/VectorServices.php b/includes/VectorServices.php
index e093d3272..9add2a48e 100644
--- a/includes/VectorServices.php
+++ b/includes/VectorServices.php
@@ -4,6 +4,7 @@ namespace Vector;
 
 use MediaWiki\MediaWikiServices;
 use Vector\FeatureManagement\FeatureManager;
+use Vector\Services\LanguageService;
 
 /**
  * A service locator for services specific to Vector.
@@ -24,4 +25,13 @@ final class VectorServices {
 	public static function getFeatureManager(): FeatureManager {
 		return MediaWikiServices::getInstance()->getService( Constants::SERVICE_FEATURE_MANAGER );
 	}
+
+	/**
+	 * Gets the language service.
+	 *
+	 * @return LanguageService
+	 */
+	public static function getLanguageService(): LanguageService {
+		return new LanguageService();
+	}
 }
diff --git a/skin.json b/skin.json
index 88e7f6f5c..80ba08d2e 100644
--- a/skin.json
+++ b/skin.json
@@ -277,8 +277,7 @@
 		"VectorWvuiSearchOptions": {
 			"value": {
 				"showThumbnail": true,
-				"showDescription": true,
-				"highlightQuery": true
+				"showDescription": true
 			}
 		},
 		"VectorDefaultSidebarVisibleForAuthorisedUser": {