Try to pick the best language [bug 639818]

Given Accept-Language: en-gb, en;q=0.8, fr-fr;q=0.6, es;q=0.2, don't pick 'es'.

apps/sumo/tests/test_locale_middleware.py

@@ -1,4 +1,7 @@
+from nose.tools import eq_
+
 from sumo.tests import TestCase
+from sumo.urlresolvers import get_best_language
 
 
 class TestLocaleMiddleware(TestCase):
@@ -49,3 +52,32 @@ class TestLocaleMiddleware(TestCase):
         response = self.client.get('/search', follow=True,
                                    HTTP_ACCEPT_LANGUAGE='en-US,fr;q=0.3')
         self.assertRedirects(response, '/en-US/search', status_code=302)
+
+
+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('xx-YY, xx;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)

apps/sumo/urlresolvers.py

@@ -61,6 +61,25 @@ def find_supported(test):
             x.split('-', 1)[0] == test.lower().split('-', 1)[0]]
 
 
+def get_best_language(accept_lang):
+    """Given an Accept-Language header, return the best-matching language."""
+
+    LUM = settings.LANGUAGE_URL_MAP
+    langs = dict(LUM)
+    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]
+    # Could not find an acceptable language.
+    return False
+
+
 def split_path(path):
     """
     Split the requested path into (locale, path).
@@ -103,22 +122,10 @@ class Prefixer(object):
                 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].lower() 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