From 66b7be09656798b745f95ff0ce7669db014b3b38 Mon Sep 17 00:00:00 2001
From: Marc Abonce Seguin <marc-abonce@mailbox.org>
Date: Mon, 4 Oct 2021 22:44:58 -0700
Subject: [PATCH] [fix] fix match_language issue to make zh-TW match to
 zh-Hant-TW

pybabel separates locales with underscores but we use hyphens
everywhere babel doesn't directly touch
---
 searx/utils.py            | 26 ++++++++++++++++++++++----
 tests/unit/test_utils.py  |  8 ++++++++
 tests/unit/test_webapp.py |  4 ++--
 3 files changed, 32 insertions(+), 6 deletions(-)

diff --git a/searx/utils.py b/searx/utils.py
index 4c2c9e429..163892e93 100644
--- a/searx/utils.py
+++ b/searx/utils.py
@@ -369,6 +369,16 @@ def _get_lang_to_lc_dict(lang_list):
     return value
 
 
+# babel's get_global contains all sorts of miscellaneous locale and territory related data
+# see get_global in: https://github.com/python-babel/babel/blob/master/babel/core.py
+def _get_from_babel(lang_code, key):
+    match = get_global(key).get(lang_code.replace('-', '_'))
+    # for some keys, such as territory_aliases, match may be a list
+    if isinstance(match, str):
+        return match.replace('_', '-')
+    return match
+
+
 def _match_language(lang_code, lang_list=[], custom_aliases={}):  # pylint: disable=W0102
     """auxiliary function to match lang_code in lang_list"""
     # replace language code with a custom alias if necessary
@@ -379,9 +389,11 @@ def _match_language(lang_code, lang_list=[], custom_aliases={}):  # pylint: disa
         return lang_code
 
     # try to get the most likely country for this language
-    subtags = get_global('likely_subtags').get(lang_code)
+    subtags = _get_from_babel(lang_code, 'likely_subtags')
     if subtags:
-        subtag_parts = subtags.split('_')
+        if subtags in lang_list:
+            return subtags
+        subtag_parts = subtags.split('-')
         new_code = subtag_parts[0] + '-' + subtag_parts[-1]
         if new_code in custom_aliases:
             new_code = custom_aliases[new_code]
@@ -402,16 +414,22 @@ def match_language(locale_code, lang_list=[], custom_aliases={}, fallback='en-US
     locale_parts = locale_code.split('-')
     lang_code = locale_parts[0]
 
+    # if locale_code has script, try matching without it
+    if len(locale_parts) > 2:
+        language = _match_language(lang_code + '-' + locale_parts[-1], lang_list, custom_aliases)
+        if language:
+            return language
+
     # try to get language using an equivalent country code
     if len(locale_parts) > 1:
-        country_alias = get_global('territory_aliases').get(locale_parts[-1])
+        country_alias = _get_from_babel(locale_parts[-1], 'territory_aliases')
         if country_alias:
             language = _match_language(lang_code + '-' + country_alias[0], lang_list, custom_aliases)
             if language:
                 return language
 
     # try to get language using an equivalent language code
-    alias = get_global('language_aliases').get(lang_code)
+    alias = _get_from_babel(lang_code, 'language_aliases')
     if alias:
         language = _match_language(alias, lang_list, custom_aliases)
         if language:
diff --git a/tests/unit/test_utils.py b/tests/unit/test_utils.py
index e9c247382..bea28c0cc 100644
--- a/tests/unit/test_utils.py
+++ b/tests/unit/test_utils.py
@@ -92,6 +92,14 @@ class TestUtils(SearxTestCase):
         self.assertEqual(utils.match_language('es', [], fallback='fallback'), 'fallback')
         self.assertEqual(utils.match_language('ja', ['jp'], {'ja': 'jp'}), 'jp')
 
+        # handle script tags
+        self.assertEqual(utils.match_language('zh-CN', ['zh-Hans-CN', 'zh-Hant-TW']), 'zh-Hans-CN')
+        self.assertEqual(utils.match_language('zh-TW', ['zh-Hans-CN', 'zh-Hant-TW']), 'zh-Hant-TW')
+        self.assertEqual(utils.match_language('zh-Hans-CN', ['zh-CN', 'zh-TW']), 'zh-CN')
+        self.assertEqual(utils.match_language('zh-Hant-TW', ['zh-CN', 'zh-TW']), 'zh-TW')
+        self.assertEqual(utils.match_language('zh-Hans', ['zh-CN', 'zh-TW', 'zh-HK']), 'zh-CN')
+        self.assertEqual(utils.match_language('zh-Hant', ['zh-CN', 'zh-TW', 'zh-HK']), 'zh-TW')
+
         aliases = {'en-GB': 'en-UK', 'he': 'iw'}
 
         # guess country
diff --git a/tests/unit/test_webapp.py b/tests/unit/test_webapp.py
index c2d57f80c..30b2839a8 100644
--- a/tests/unit/test_webapp.py
+++ b/tests/unit/test_webapp.py
@@ -211,12 +211,12 @@ class ViewsTestCase(SearxTestCase):
         result = self.app.get('/preferences', headers={'Accept-Language': 'zh-tw;q=0.8'})
         self.assertEqual(result.status_code, 200)
         self.assertIn(
-            b'<option value="zh_TW" selected="selected">',
+            b'<option value="zh_Hant_TW" selected="selected">',
             result.data,
             'Interface locale ignored browser preference.'
         )
         self.assertIn(
-            b'<option value="zh-TW" selected="selected">',
+            b'<option value="zh_Hant_TW" selected="selected">',
             result.data,
             'Search language ignored browser preference.'
         )