Add LOCALE_NORMALIZATION_MAP for locale equivalences

2 files changed, 44 insertions, 7 deletions

diff --git a/python/phonenumbers/geocoder.py b/python/phonenumbers/geocoder.py
index 521ba5df..4297de92 100644
--- a/python/phonenumbers/geocoder.py
+++ b/python/phonenumbers/geocoder.py
@@ -64,6 +64,9 @@ except ImportError:  # pragma no cover
         raise
 
 
+_LOCALE_NORMALIZATION_MAP = {"zh_TW": "zh_Hant", "zh_HK": "zh_Hant", "zh_MO": "zh_Hant"}
+
+
 def _may_fall_back_to_english(lang):
     # Don't fall back to English if the requested language is among the following:
     # - Chinese
@@ -72,16 +75,38 @@ def _may_fall_back_to_english(lang):
     return lang != "zh" and lang != "ja" and lang != "ko"
 
 
+def _full_locale(lang, script, region):
+    if script is not None:
+        if region is not None:
+            return "%s_%s_%s" % (lang, script, region)
+        else:
+            return "%s_%s" % (lang, script)
+    elif region is not None:
+        return "%s_%s" % (lang, region)
+    else:
+        return lang
+
+
 def _find_lang(langdict, lang, script, region):
     """Return the entry in the dictionary for the given language information."""
-    # First look for lang, script as a combination
-    lang_script = "%s_%s" % (lang, script)
-    if lang_script in langdict:
-        return langdict[lang_script]
+    # Check if we should map this to a different locale.
+    full_locale = _full_locale(lang, script, region)
+    if (full_locale in _LOCALE_NORMALIZATION_MAP and
+        _LOCALE_NORMALIZATION_MAP[full_locale] in langdict):
+        return langdict[_LOCALE_NORMALIZATION_MAP[full_locale]]
+    # First look for the full locale
+    if full_locale in langdict:
+        return langdict[full_locale]
+    # Then look for lang, script as a combination
+    if script is not None:
+        lang_script = "%s_%s" % (lang, script)
+        if lang_script in langdict:
+            return langdict[lang_script]
     # Next look for lang, region as a combination
-    lang_region = "%s_%s" % (lang, region)
-    if lang_region in langdict:
-        return langdict[lang_region]
+    if region is not None:
+        lang_region = "%s_%s" % (lang, region)
+        if lang_region in langdict:
+            return langdict[lang_region]
     # Fall back to bare language code lookup
     if lang in langdict:
         return langdict[lang]
diff --git a/python/tests/geocodertest.py b/python/tests/geocodertest.py
index 4a9ecfe1..9cd136c3 100644
--- a/python/tests/geocodertest.py
+++ b/python/tests/geocodertest.py
@@ -169,6 +169,7 @@ class PhoneNumberGeocoderTest(unittest.TestCase):
         TEST_GEOCODE_DATA['1650960'] = {'en': u'Mountain View, CA',
                                         "en_GB": u'Mountain View California',
                                         "en_US": u'Mountain View, Sunny California',
+                                        "en_Xyzz_US": u'MTV - xyzz',
                                         "en_Latn": u'MountainView'}
         # The following test might one day return "Mountain View California"
         self.assertEqual("United States",
@@ -179,7 +180,18 @@ class PhoneNumberGeocoderTest(unittest.TestCase):
                          description_for_number(US_NUMBER2, _ENGLISH, script="Latn"))
         self.assertEqual("United States",
                          description_for_number(US_NUMBER2, _ENGLISH, script="Latn", region="GB"))
+        self.assertEqual("MTV - xyzz",
+                         description_for_number(US_NUMBER2, _ENGLISH, script="Xyzz", region="US"))
+        self.assertEqual("Mountain View, Sunny California",
+                         description_for_number(US_NUMBER2, _ENGLISH, script="Zazz", region="US"))
         # Get a different result when there is a script-specific variant
         self.assertEqual("MountainView",
                          description_for_number(US_NUMBER2, _ENGLISH, script="Latn", region="US"))
         TEST_GEOCODE_DATA['1650960'] = {'en': u'Mountain View, CA'}
+
+        # Test the locale mapping
+        TEST_GEOCODE_DATA['8868'] = {'zh': u'Chinese', 'zh_Hant': u'Hant-specific'}
+        tw_number = FrozenPhoneNumber(country_code=886, national_number=810080123)
+        self.assertEqual("Hant-specific",
+                         description_for_number(tw_number, "zh", region="TW"))
+        del TEST_GEOCODE_DATA['8868']