'Island' doesn't register as 'Iceland' (IS) anymore if there is no la…

…nguage specified

CHANGELOG.md

@@ -6,6 +6,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 ## [5.2.3] - 2024-11-04
 
 - "Pacific Island" no longer matches Iceland.
+- "Island" no longer matches Iceland when language is not set.
 
 ## [5.2.2] - 2024-08-23
 

geoconvert/convert.py

@@ -9,6 +9,7 @@
     DE_POSTCODE_RANGE,
     NUTS_CODES_BY_COUNTRY,
     all_nuts_regex,
+    ambiguous_countries,
     br_postcode_regex,
     br_state_code_regex,
     br_state_name_regex,
@@ -413,8 +414,15 @@ def _full_name_to_country_code(text, lang, language_to_full_names):
         # If the language is unknown, just do not use any language.
         language_to_full_names = {}
 
-    for lang, full_names in language_to_full_names.items():
-        country_code = _full_name_to_country_code_for_lang(text, lang, full_names)
+    for language, full_names in language_to_full_names.items():
+        # When language is not set, we remove ambiguous countries names from the countries list
+        if not lang:
+            full_names = {
+                name: code
+                for name, code in full_names.items()
+                if name not in ambiguous_countries
+            }
+        country_code = _full_name_to_country_code_for_lang(text, language, full_names)
         if country_code:
             return country_code
 

geoconvert/data/__init__.py

@@ -1,5 +1,6 @@
 from .capitals import capitals_de, capitals_en, capitals_fr, language_to_capital_names
 from .countries import (
+    ambiguous_countries,
     countries_de,
     countries_en,
     countries_fr,

geoconvert/data/countries.py

@@ -55,6 +55,11 @@
     "pacific island": None,  # en
 }
 
+# We register here names matching common words in other languages to prevent
+# matching them when language is not defined
+ambiguous_countries = ["island"]
+
+
 countries_fr = {
     **special_countries,
     **{

tests/test_countries.py

@@ -67,9 +67,10 @@ class TestCountries:
             ("Rhode Island", {}, "US"),  # en
             ("Pacific island", {}, None),  # en
             ("Fiji/Pacific island", {}, "FJ"),  # en
+            ("island", {}, None),  # en
             # However, in cases where island is singular instead of plural,
             # there can be confusion.
-            ("Solomon Island Nationals", {}, "IS"),  # en
+            ("Solomon Island Nationals", {}, None),  # en
             # Any capitalization for lang works
             ("Germany", {"lang": "en"}, "DE"),
             ("Germany", {"lang": "En"}, "DE"),
@@ -166,6 +167,7 @@ def test_country_name_to_country_code_fr(self, input_data, expected):
             ("state of palestine", "PS"),
             ("palestine, state of", "PS"),
             ("Irak", None),
+            ("island", None),
         ],
     )
     def test_country_name_to_country_code_en(self, input_data, expected):
@@ -194,6 +196,7 @@ def test_country_name_to_country_code_en(self, input_data, expected):
             ("Dschibuti-Stadt", "DJ"),
             ("elfenbeinküste und ecuador    ", "CI"),
             ("Fiji/Pacific island", None),  # en
+            ("island", "IS"),
         ],
     )
     def test_country_name_to_country_code_de(self, input_data, expected):