fix bug 647796 - better accept-language logic

apps/devmo/tests/test_middleware.py

@@ -0,0 +1,37 @@
+from nose.tools import eq_
+from test_utils import TestCase
+
+from devmo.urlresolvers import get_best_language
+
+
+class BestLanguageTests(TestCase):
+    def test_english_only(self):
+        best = get_best_language('en-us, en;q=0.8')
+        eq_('en-US', best)
+
+    def test_en_GB(self):
+        """Stick with English if you can."""
+        best = get_best_language('en-gb, fr;q=0.8')
+        eq_('en-US', best)
+
+    def test_not_worst_choice(self):
+        """Try not to fall back to 'es' here."""
+        best = get_best_language('en-gb, en;q=0.8, fr-fr;q=0.6, es;q=0.2')
+        eq_('en-US', best)
+
+    def test_fr_FR(self):
+        best = get_best_language('fr-FR, es;q=0.8')
+        eq_('fr', best)
+
+    def test_non_existent(self):
+        best = get_best_language('xy-YY, xy;q=0.8')
+        eq_(False, best)
+
+    def test_prefix_matching(self):
+        """en-US is a better match for en-gb, es;q=0.2 than es."""
+        best = get_best_language('en-gb, es;q=0.2')
+        eq_('en-US', best)
+
+    def test_serbian(self):
+        """sr -> sr-CYRL, not sr-LATN."""
+        eq_('sr-CYRL', get_best_language('sr'))

apps/devmo/urlresolvers.py

@@ -9,6 +9,30 @@
 _prefixes = {}
 
 
+def get_best_language(accept_lang):
+    """Given an Accept-Language header, return the best-matching language."""
+
+    LUM = settings.LANGUAGE_URL_MAP
+    NSL = settings.NON_SUPPORTED_LOCALES
+    LC = settings.LANGUAGE_CODE
+    langs = dict(LUM)
+    # Add in non-supported first to allow overriding prefix behavior.
+    langs.update((k.lower(), v if v else LC) for k, v in NSL.items() if
+                 k.lower() not in langs)
+    langs.update((k.split('-')[0], v) for k, v in LUM.items() if
+                 k.split('-')[0] not in langs)
+    ranked = parse_accept_lang_header(accept_lang)
+    for lang, _ in ranked:
+        lang = lang.lower()
+        if lang in langs:
+            return langs[lang]
+        pre = lang.split('-')[0]
+        if pre in langs:
+            return langs[pre]
+    # Couldn't find any acceptable locale.
+    return False
+
+
 def set_url_prefix(prefix):
     """Set the ``prefix`` for the current thread."""
     _prefixes[currentThread()] = prefix
@@ -83,22 +107,10 @@ def get_language(self):
                 return settings.LANGUAGE_URL_MAP[lang]
 
         if self.request.META.get('HTTP_ACCEPT_LANGUAGE'):
-            ranked_languages = parse_accept_lang_header(
+            best = get_best_language(
                 self.request.META['HTTP_ACCEPT_LANGUAGE'])
-
-            # Do we support or remap their locale?
-            supported = [lang[0] for lang in ranked_languages if lang[0]
-                        in settings.LANGUAGE_URL_MAP]
-
-            # Do we support a less specific locale? (xx-YY -> xx)
-            if not len(supported):
-                for lang in ranked_languages:
-                    supported = find_supported(lang[0])
-                    if supported:
-                        break
-
-            if len(supported):
-                return settings.LANGUAGE_URL_MAP[supported[0].lower()]
+            if best:
+                return best
 
         return settings.LANGUAGE_CODE
 

apps/sumo/urlresolvers.py

@@ -10,6 +10,30 @@
 _locals = threading.local()
 
 
+def get_best_language(accept_lang):
+    """Given an Accept-Language header, return the best-matching language."""
+
+    LUM = settings.LANGUAGE_URL_MAP
+    NSL = settings.NON_SUPPORTED_LOCALES
+    LC = settings.LANGUAGE_CODE
+    langs = dict(LUM)
+    # Add in non-supported first to allow overriding prefix behavior.
+    langs.update((k.lower(), v if v else LC) for k, v in NSL.items() if
+                 k.lower() not in langs)
+    langs.update((k.split('-')[0], v) for k, v in LUM.items() if
+                 k.split('-')[0] not in langs)
+    ranked = parse_accept_lang_header(accept_lang)
+    for lang, _ in ranked:
+        lang = lang.lower()
+        if lang in langs:
+            return langs[lang]
+        pre = lang.split('-')[0]
+        if pre in langs:
+            return langs[pre]
+    # Couldn't find any acceptable locale.
+    return False
+
+
 def set_url_prefixer(prefixer):
     """Set the Prefixer for the current thread."""
     _locals.prefixer = prefixer
@@ -103,22 +127,10 @@ def get_language(self):
                 return settings.LANGUAGE_URL_MAP[lang]
 
         if self.request.META.get('HTTP_ACCEPT_LANGUAGE'):
-            ranked_languages = parse_accept_lang_header(
+            best = get_best_language(
                 self.request.META['HTTP_ACCEPT_LANGUAGE'])
-
-            # Do we support or remap their locale?
-            supported = [lang[0] for lang in ranked_languages if lang[0]
-                        in settings.LANGUAGE_URL_MAP]
-
-            # Do we support a less specific locale? (xx-YY -> xx)
-            if not len(supported):
-                for lang in ranked_languages:
-                    supported = find_supported(lang[0])
-                    if supported:
-                        break
-
-            if len(supported):
-                return settings.LANGUAGE_URL_MAP[supported[0].lower()]
+            if best:
+                return best
 
         return settings.LANGUAGE_CODE
 

settings.py

@@ -118,6 +118,21 @@
     'xx-testing': 'x-testing',
 }
 
+# Locales that are known but unsupported. Keys are the locale, values
+# are an optional fallback locale, or None, to use the LANGUAGE_CODE.
+NON_SUPPORTED_LOCALES = {
+    'af': None,
+    'an': 'es',
+    'br': 'fr',
+    'csb': 'pl',
+    'lij': 'it',
+    'nb-NO': 'no',
+    'nn-NO': 'no',
+    'oc': 'fr',
+    'sr': 'sr-CYRL',  # Override the tendency to go sr->sr-LATN.
+    'sv-SE': 'sv',
+}
+
 try:
     DEV_LANGUAGES = [
         loc.replace('_','-') for loc in os.listdir(path('locale'))
@@ -1073,4 +1088,4 @@ def read_only_mode(env):
 GRAPHITE_PREFIX = 'devmo'
 GRAPHITE_TIMEOUT = 1
 
-ES_DISABLED = True
+ES_DISABLED = True