Improves heuristics and adds testing

canonicalwebteam/http/heuristics.py

@@ -1,26 +1,35 @@
-import calendar
+import locale
+from datetime import datetime, timedelta
+
 from cachecontrol.heuristics import BaseHeuristic
-from datetime import timedelta, datetime
-from email.utils import formatdate
 
 
 def expire_after(delta, date=None):
     date = date or datetime.utcnow()
     return date + delta
 
 
-def datetime_to_header(dt):
-    return formatdate(calendar.timegm(dt.timetuple()))
+def datetime_to_HTTP_date(date_and_time):
+    """
+    Returns a HTTP-date as defined in rfc7234 section 5.3
+    for a datetime object"""
+    locale.setlocale(locale.LC_ALL, "en_GB.utf8")
+
+    return datetime.astimezone(date_and_time).strftime(
+        "%a, %d %b %Y %H:%M:%S %Z"
+    )
 
 
-def cache_control_in_response_headers(headers):
+def cache_directives_in_headers(headers):
     """
-    Checks if CacheControl is set in response headers
+    Checks if cache controls are set in response headers
     """
     cache_control = "cache-control" in headers
     pragma = "pragma" in headers and "no-cache" in headers["pragma"]
     expires = "expires" in headers
 
+    return expires or pragma or cache_control
+
 
 class ExpiresAfterIfNoCacheControl(BaseHeuristic):
     """
@@ -31,21 +40,30 @@ def __init__(self, **kw):
         self.delta = timedelta(**kw)
 
     def update_headers(self, response):
-        if cache_directives_in_headers(reponse.headers):
+        """
+        If no caching controls are present, a default expires header will be set
+        """
+        if cache_directives_in_headers(response.headers):
             return
 
         expires = expire_after(self.delta)
 
         return {
-            "expires": datetime_to_header(expires),
+            "expires": datetime_to_HTTP_date(expires),
             "cache-control": "public",
         }
 
     def warning(self):
+        """
+        Adds a warning to the response as defined in rfc7234 section 5.5
+        """
         template = "110 - Automatically cached for %s. Response might be stale"
         return template % self.delta
 
     def apply(self, response):
+        """
+        Applies the heuristic to the response
+        """
         updated_headers = self.update_headers(response)
 
         if updated_headers:

tests/test_heuristics.py

@@ -0,0 +1,42 @@
+import locale
+import unittest
+from datetime import datetime, timedelta
+
+from canonicalwebteam.http import heuristics
+
+
+class TestHeuristics(unittest.TestCase):
+    def test_custom_heuristic(self):
+        today = datetime.utcnow()
+        one_day_delta = timedelta(days=1)
+        tomorrow = today + one_day_delta
+
+        self.assertEqual(
+            tomorrow, heuristics.expire_after(one_day_delta, today)
+        )
+
+    def test_datetime_to_header_string(self):
+        locale.setlocale(locale.LC_ALL, "en_GB.utf8")
+        date_string = "Thu, 01 Dec 1994 16:00:00 GMT"
+        date = datetime.strptime(date_string, "%a, %d %b %Y %H:%M:%S %Z")
+
+        self.assertEqual(date_string, heuristics.datetime_to_HTTP_date(date))
+
+    def test_cache_directives_in_headers(self):
+        headers = {}
+
+        self.assertEqual(
+            heuristics.cache_directives_in_headers(headers), False
+        )
+
+        headers = {"expires": "1"}
+
+        self.assertEqual(heuristics.cache_directives_in_headers(headers), True)
+
+        headers = {"pragma": "no-cache"}
+
+        self.assertEqual(heuristics.cache_directives_in_headers(headers), True)
+
+        headers = {"cache-control": "yeah"}
+
+        self.assertEqual(heuristics.cache_directives_in_headers(headers), True)