Merge pull request #1514 from GabiThume/bug916905

Fixes Bug 916905 - removing unique uuids and dates

webapp-django/crashstats/crashstats/models.py

@@ -12,6 +12,7 @@
 import stat
 import time
 import urllib
+import re
 
 from django.conf import settings
 from django.core.cache import cache
@@ -176,12 +177,23 @@ def _process_response(self, method, url, status_code):
         path = urlparse.urlparse(url).path
         path_info = urllib.quote(path.encode('utf-8'))
 
+        # Removes uuids from path_info
+        if "uuid/" in url:
+            uuid = path_info.rsplit("/uuid/")
+            if len(uuid) == 2:
+                path_info = uuid[0] + '/uuid' + uuid[1][uuid[1].find('/'):]
+
+        # Replaces dates for XXXX-XX-XX
+        replaces = re.findall(r'(\d{4}-\d{2}-\d{2})', path_info)
+        for date in replaces:
+            date = path_info[path_info.find(date):].rsplit("/")[0]
+            path_info = path_info.replace(date, "XXXX-XX-XX")
+
         metric = u"middleware.{0}.{1}.{2}".format(
             method.upper(),
             path_info.lstrip('/').replace('.', '-'),
             status_code
         )
-
         metric = metric.encode('utf-8')
         statsd.incr(metric)
 

webapp-django/crashstats/crashstats/tests/test_models.py

@@ -1085,7 +1085,69 @@ def mocked_get(url, **options):
         rget.side_effect = mocked_get
         api.get(crash_id='crash-id')
         assert incr.called
-        metric = 'middleware.GET.crash_data/uuid/crash-id/datatype/meta/.200'
+        metric = 'middleware.GET.crash_data/uuid/datatype/meta/.200'
+        incr.assert_called_with(metric)
+
+        # Test if replaces dates for XXXX-XX-XX
+        model = models.TCBS
+        api = model()
+
+        def mocked_get(**options):
+            assert 'crashes/signatures' in options['url']
+            return Response("""
+                {
+                    "start_date": "2012-05-10",
+                    "end_date": "2012-05-24"
+                }
+            """)
+
+        rget.side_effect = mocked_get
+        today = datetime.datetime.utcnow()
+        api.get(
+            product='Thunderbird',
+            version='12.0',
+            end_date=today,
+        )
+        assert incr.called
+        metric = "middleware.GET.crashes/signatures/product/Thunderbird/" \
+            "version/12-0/end_date/XXXX-XX-XX/limit/300/.200"
+        incr.assert_called_with(metric)
+
+        # Test if removes unique uuids
+        model = models.ProcessedCrash
+        api = model()
+
+        def mocked_get(url, **options):
+            assert '/crash_data/' in url
+            ok_('/datatype/processed/' in url)
+            return Response("""
+            {
+              "product": "WaterWolf",
+              "uuid": "7c44ade2-fdeb-4d6c-830a-07d302120525",
+              "version": "13.0",
+              "build": "20120501201020",
+              "ReleaseChannel": "beta",
+              "os_name": "Windows NT",
+              "date_processed": "2012-05-25 11:35:57",
+              "success": true,
+              "signature": "CLocalEndpointEnumerator::OnMediaNotific",
+              "addons": [
+                [
+                  "testpilot@labs.mozilla.com",
+                  "1.2.1"
+                ],
+                [
+                  "{972ce4c6-7e08-4474-a285-3208198ce6fd}",
+                  "13.0"
+                ]
+              ]
+            }
+            """)
+
+        rget.side_effect = mocked_get
+        api.get(crash_id='7c44ade2-fdeb-4d6c-830a-07d302120525')
+        assert incr.called
+        metric = "middleware.GET.crash_data/datatype/processed/uuid/.200"
         incr.assert_called_with(metric)