[fix] add missing territory (country) name

Related-to: https://github.com/searxng/searxng/pull/1029#issuecomment-1086824911
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>

3 files changed, 41 insertions, 27 deletions

diff --git a/searx/languages.py b/searx/languages.py
index 177a43bfe..e9f167103 100644
--- a/searx/languages.py
+++ b/searx/languages.py
@@ -2,11 +2,11 @@
 # list of language codes
 # this file is generated automatically by utils/fetch_languages.py
 language_codes = (
-    ('af-ZA', 'Afrikaans', '', 'Afrikaans', '\U0001f1ff\U0001f1e6'),
-    ('ar-EG', 'العربية', '', 'Arabic', '\U0001f1ea\U0001f1ec'),
-    ('be-BY', 'Беларуская', '', 'Belarusian', '\U0001f1e7\U0001f1fe'),
+    ('af-ZA', 'Afrikaans', 'Suid-Afrika', 'Afrikaans', '\U0001f1ff\U0001f1e6'),
+    ('ar-EG', 'العربية', 'مصر', 'Arabic', '\U0001f1ea\U0001f1ec'),
+    ('be-BY', 'Беларуская', 'Беларусь', 'Belarusian', '\U0001f1e7\U0001f1fe'),
     ('bg-BG', 'Български', 'България', 'Bulgarian', '\U0001f1e7\U0001f1ec'),
-    ('ca-ES', 'Català', '', 'Catalan', '\U0001f1ea\U0001f1f8'),
+    ('ca-ES', 'Català', 'Espanya', 'Catalan', '\U0001f1ea\U0001f1f8'),
     ('cs-CZ', 'Čeština', 'Česko', 'Czech', '\U0001f1e8\U0001f1ff'),
     ('da-DK', 'Dansk', 'Danmark', 'Danish', '\U0001f1e9\U0001f1f0'),
     ('de', 'Deutsch', '', 'German', '\U0001f310'),
@@ -28,26 +28,30 @@ language_codes = (
     ('es-ES', 'Español', 'España', 'Spanish', '\U0001f1ea\U0001f1f8'),
     ('es-MX', 'Español', 'México', 'Spanish', '\U0001f1f2\U0001f1fd'),
     ('et-EE', 'Eesti', 'Eesti', 'Estonian', '\U0001f1ea\U0001f1ea'),
-    ('fa-IR', 'فارسی', '', 'Persian', '\U0001f1ee\U0001f1f7'),
+    ('fa-IR', 'فارسی', 'ایران', 'Persian', '\U0001f1ee\U0001f1f7'),
     ('fi-FI', 'Suomi', 'Suomi', 'Finnish', '\U0001f1eb\U0001f1ee'),
-    ('fil-PH', 'Filipino', '', 'Filipino', '\U0001f1f5\U0001f1ed'),
+    ('fil-PH', 'Filipino', 'Pilipinas', 'Filipino', '\U0001f1f5\U0001f1ed'),
     ('fr', 'Français', '', 'French', '\U0001f310'),
     ('fr-BE', 'Français', 'Belgique', 'French', '\U0001f1e7\U0001f1ea'),
     ('fr-CA', 'Français', 'Canada', 'French', '\U0001f1e8\U0001f1e6'),
     ('fr-CH', 'Français', 'Suisse', 'French', '\U0001f1e8\U0001f1ed'),
     ('fr-FR', 'Français', 'France', 'French', '\U0001f1eb\U0001f1f7'),
-    ('he-IL', 'עברית', '', 'Hebrew', '\U0001f1ee\U0001f1f1'),
-    ('hi-IN', 'हिन्दी', '', 'Hindi', '\U0001f1ee\U0001f1f3'),
-    ('hr-HR', 'Hrvatski', '', 'Croatian', '\U0001f1ed\U0001f1f7'),
+    ('he-IL', 'עברית', 'ישראל', 'Hebrew', '\U0001f1ee\U0001f1f1'),
+    ('hi-IN', 'हिन्दी', 'भारत', 'Hindi', '\U0001f1ee\U0001f1f3'),
+    ('hr-HR', 'Hrvatski', 'Hrvatska', 'Croatian', '\U0001f1ed\U0001f1f7'),
     ('hu-HU', 'Magyar', 'Magyarország', 'Hungarian', '\U0001f1ed\U0001f1fa'),
-    ('id-ID', 'Indonesia', '', 'Indonesian', '\U0001f1ee\U0001f1e9'),
-    ('is-IS', 'Íslenska', '', 'Icelandic', '\U0001f1ee\U0001f1f8'),
+    ('id-ID', 'Indonesia', 'Indonesia', 'Indonesian', '\U0001f1ee\U0001f1e9'),
+    ('is-IS', 'Íslenska', 'Ísland', 'Icelandic', '\U0001f1ee\U0001f1f8'),
     ('it-IT', 'Italiano', 'Italia', 'Italian', '\U0001f1ee\U0001f1f9'),
-    ('ja-JP', '日本語', '', 'Japanese', '\U0001f1ef\U0001f1f5'),
-    ('ko-KR', '한국어', '', 'Korean', '\U0001f1f0\U0001f1f7'),
-    ('lt-LT', 'Lietuvių', '', 'Lithuanian', '\U0001f1f1\U0001f1f9'),
-    ('lv-LV', 'Latviešu', '', 'Latvian', '\U0001f1f1\U0001f1fb'),
-    ('nb-NO', 'Norsk Bokmål', '', 'Norwegian Bokmål', '\U0001f1f3\U0001f1f4'),
+    ('ja-JP', '日本語', '日本', 'Japanese', '\U0001f1ef\U0001f1f5'),
+    ('ko-KR', '한국어', '대한민국', 'Korean', '\U0001f1f0\U0001f1f7'),
+    ('lt-LT', 'Lietuvių', 'Lietuva', 'Lithuanian', '\U0001f1f1\U0001f1f9'),
+    ('lv-LV', 'Latviešu', 'Latvija', 'Latvian', '\U0001f1f1\U0001f1fb'),
+    (   'nb-NO',
+        'Norsk Bokmål',
+        'Norge',
+        'Norwegian Bokmål',
+        '\U0001f1f3\U0001f1f4'),
     ('nl', 'Nederlands', '', 'Dutch', '\U0001f310'),
     ('nl-BE', 'Nederlands', 'België', 'Dutch', '\U0001f1e7\U0001f1ea'),
     ('nl-NL', 'Nederlands', 'Nederland', 'Dutch', '\U0001f1f3\U0001f1f1'),
@@ -56,16 +60,16 @@ language_codes = (
     ('pt-BR', 'Português', 'Brasil', 'Portuguese', '\U0001f1e7\U0001f1f7'),
     ('pt-PT', 'Português', 'Portugal', 'Portuguese', '\U0001f1f5\U0001f1f9'),
     ('ro-RO', 'Română', 'România', 'Romanian', '\U0001f1f7\U0001f1f4'),
-    ('ru-RU', 'Русский', '', 'Russian', '\U0001f1f7\U0001f1fa'),
-    ('sk-SK', 'Slovenčina', '', 'Slovak', '\U0001f1f8\U0001f1f0'),
-    ('sl-SI', 'Slovenščina', '', 'Slovenian', '\U0001f1f8\U0001f1ee'),
-    ('sr-RS', 'Српски', '', 'Serbian', '\U0001f1f7\U0001f1f8'),
+    ('ru-RU', 'Русский', 'Россия', 'Russian', '\U0001f1f7\U0001f1fa'),
+    ('sk-SK', 'Slovenčina', 'Slovensko', 'Slovak', '\U0001f1f8\U0001f1f0'),
+    ('sl-SI', 'Slovenščina', 'Slovenija', 'Slovenian', '\U0001f1f8\U0001f1ee'),
+    ('sr-RS', 'Српски', 'Србија', 'Serbian', '\U0001f1f7\U0001f1f8'),
     ('sv-SE', 'Svenska', 'Sverige', 'Swedish', '\U0001f1f8\U0001f1ea'),
-    ('sw-TZ', 'Kiswahili', '', 'Swahili', '\U0001f1f9\U0001f1ff'),
-    ('th-TH', 'ไทย', '', 'Thai', '\U0001f1f9\U0001f1ed'),
-    ('tr-TR', 'Türkçe', '', 'Turkish', '\U0001f1f9\U0001f1f7'),
-    ('uk-UA', 'Українська', '', 'Ukrainian', '\U0001f1fa\U0001f1e6'),
-    ('vi-VN', 'Tiếng Việt', '', 'Vietnamese', '\U0001f1fb\U0001f1f3'),
+    ('sw-TZ', 'Kiswahili', 'Tanzania', 'Swahili', '\U0001f1f9\U0001f1ff'),
+    ('th-TH', 'ไทย', 'ไทย', 'Thai', '\U0001f1f9\U0001f1ed'),
+    ('tr-TR', 'Türkçe', 'Türkiye', 'Turkish', '\U0001f1f9\U0001f1f7'),
+    ('uk-UA', 'Українська', 'Україна', 'Ukrainian', '\U0001f1fa\U0001f1e6'),
+    ('vi-VN', 'Tiếng Việt', 'Việt Nam', 'Vietnamese', '\U0001f1fb\U0001f1f3'),
     ('zh', '中文', '', 'Chinese', '\U0001f310'),
     ('zh-CN', '中文', '中国', 'Chinese', '\U0001f1e8\U0001f1f3'),
     ('zh-HK', '中文', '中國香港特別行政區', 'Chinese', '\U0001f1ed\U0001f1f0'),
diff --git a/searxng_extra/update/update_languages.py b/searxng_extra/update/update_languages.py
index 92083f39f..95ee5bd3e 100755
--- a/searxng_extra/update/update_languages.py
+++ b/searxng_extra/update/update_languages.py
@@ -113,6 +113,16 @@ def get_unicode_flag(lang_code):
     return c1 + c2
 
 
+def get_territory_name(lang_code):
+    country_name = None
+    locale = get_locale(lang_code)
+    try:
+        country_name = locale.get_territory_name()
+    except FileNotFoundError as exc:
+        print("ERROR: %s --> %s" % (locale, exc))
+    return country_name
+
+
 # Join all language lists.
 def join_language_lists(engines_languages):
     language_list = {}
@@ -274,7 +284,7 @@ def write_languages_file(languages):
         item = (
             code,
             languages[code]['name'].split(' (')[0],
-            languages[code].get('country_name') or '',
+            get_territory_name(code) or '',
             languages[code].get('english_name') or '',
             UnicodeEscape(flag),
         )
diff --git a/tests/unit/test_query.py b/tests/unit/test_query.py
index 05fcafe30..0bbf8b4ba 100644
--- a/tests/unit/test_query.py
+++ b/tests/unit/test_query.py
@@ -130,7 +130,7 @@ class TestLanguageParser(SearxTestCase):
         self.assertEqual(query.autocomplete_list, [":hu-hu"])
 
         query = RawTextQuery(':v', [])
-        self.assertEqual(query.autocomplete_list, [":vi", ":tiếng việt"])
+        self.assertEqual(query.autocomplete_list, [':vi', ':tiếng việt', ':việt_nam'])
 
 
 class TestTimeoutParser(SearxTestCase):