Integration tests for desktop platforms and safebrowsing

tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py

Index: tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py
diff --git a/tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py b/tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py
index 8857a56e4cff95393360195f389fa1e3a96bc157..fde33b1a78fa45325c8db6f248ff33c176630ad2 100644
--- a/tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py
+++ b/tools/chrome_proxy/integration_tests/chrome_proxy_metrics.py
@@ -270,7 +270,7 @@ class ChromeProxyMetric(network_metrics.NetworkMetric):
       results.AddValue(scalar.ScalarValue(
           results.current_page, 'bypass', 'count', bypass_count))
 
-  def AddResultsForSafebrowsing(self, tab, results):
+  def AddResultsForSafebrowsingMobile(self, tab, results):
     count = 0
     safebrowsing_count = 0
 
@@ -310,8 +310,26 @@ class ChromeProxyMetric(network_metrics.NetworkMetric):
                   r.headers))
         via_fallback_count += 1
 
-    results.AddValue(scalar.ScalarValue(
-        results.current_page, 'via_fallback', 'count', via_fallback_count))
+  def AddResultsForSafebrowsingDesktop(self, tab, results):

[sclittle, 2015/01/22 23:15:14]

Move this method definition up above AddResultsForHTTPFallback.

[Not at Google. Contact bengr, 2015/01/23 17:11:34]

Done.

+    count = 0
+    safebrowsing_count = 0
+    for resp in self.IterResponses(tab):

[sclittle, 2015/01/22 23:15:14]

What do you actually expect to happen here? Should the DRP return a safebrowsing
response, and the client should ignore it, or should the real response come in
from the DRP, or should the DRP cause a bypass? Or should the client not send
any request at all and pop up an interstitial?

Please comment about what is expected to happen on desktop.

[Not at Google. Contact bengr, 2015/01/23 17:11:34]

Done.

+      count += 1
+      if resp.HasChromeProxyViaHeader():
+        safebrowsing_count += 1

[sclittle, 2015/01/22 23:15:14]

Why is this called safebrowsing_count if you're counting via headers?

[Not at Google. Contact bengr, 2015/01/23 17:11:34]

Changed to viaheader_count. Had originally kept it safebrowsing_count because
via header is the expected response for a site with malware in it for desktop.
Don't have a strong preference either way.

+      else:
+        r = resp.response
+        raise ChromeProxyMetricException, (
+            '%s: Safebrowsing feature should be off for desktop.\n'
+            'Reponse: status=(%d, %s)\nHeaders:\n %s' % (
+                r.url, r.status, r.status_text, r.headers))
+    if count == safebrowsing_count:
+      results.AddValue(scalar.ScalarValue(
+          results.current_page, 'safebrowsing', 'boolean', True))

[sclittle, 2015/01/22 23:15:14]

Instead of just a boolean, how about adding both count and safebrowsing_count as
results? You could update the regular safebrowsing test to report these as well.

[Not at Google. Contact bengr, 2015/01/23 17:11:34]

Done. Added count. safebrowsing_count is redundant since this only gets printed
when they are equal.

+    else:
+      raise ChromeProxyMetricException, (
+          'Safebrowsing failed (count=%d, safebrowsing_count=%d)\n' % (
+              count, safebrowsing_count))
 
   def AddResultsForHTTPToDirectFallback(self, tab, results):
     via_fallback_count = 0