Add support for tracking websocket connections in extension telemetry…

… service.

Enhance RemoteHostContacted signal and signal processor classes to support websocket connections. These changes are non-functional until the actual signal interception logic in web_request_api is landed.

Bug: 1362637
Change-Id: Ia8e6c09a03d57b446f3b7024eec2980bd373cf83
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3921947
Reviewed-by: Xinghui Lu <xinghuilu@chromium.org>
Commit-Queue: Adam Psarouthakis <psarouthakis@google.com>
Cr-Commit-Position: refs/heads/main@{#1054295}

chrome/browser/safe_browsing/extension_telemetry/extension_telemetry_service_browsertest.cc

@@ -155,6 +155,8 @@ IN_PROC_BROWSER_TEST_F(ExtensionTelemetryServiceBrowserTest,
         remote_host_contacted_info.remote_host(0);
     EXPECT_EQ(remote_host_info.contact_count(), static_cast<uint32_t>(1));
     EXPECT_EQ(remote_host_info.url(), kExtensionContactedHost);
+    EXPECT_EQ(remote_host_info.connection_protocol(),
+              RemoteHostInfo::HTTP_HTTPS);
   }
 }
 

chrome/browser/safe_browsing/extension_telemetry/potential_password_theft_signal_processor.cc

@@ -33,8 +33,8 @@ void PotentialPasswordTheftSignalProcessor::ProcessSignal(
   if (signal.GetType() == ExtensionSignalType::kRemoteHostContacted) {
     const auto& rhc_signal =
         static_cast<const RemoteHostContactedSignal&>(signal);
-    // Extract only the host portion of the contacted URLs.
-    std::string host_url = rhc_signal.contacted_host_url().host();
+    // Extract only the host portion of the remote host URLs.
+    std::string host_url = rhc_signal.remote_host_url().host();
     extension_id = rhc_signal.extension_id();
     (remote_host_url_queue_[extension_id])
         .emplace_back(std::make_pair(host_url, signal_creation_time));

chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal.cc

@@ -3,13 +3,20 @@
 // found in the LICENSE file.
 
 #include "chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal.h"
+#include "components/safe_browsing/core/common/proto/csd.pb.h"
 
 namespace safe_browsing {
 
+using RemoteHostInfo = ExtensionTelemetryReportRequest::SignalInfo::
+    RemoteHostContactedInfo::RemoteHostInfo;
+
 RemoteHostContactedSignal::RemoteHostContactedSignal(
     const extensions::ExtensionId& extension_id,
-    const GURL& host_url)
-    : ExtensionSignal(extension_id), contacted_host_url_(host_url) {}
+    const GURL& host_url,
+    RemoteHostInfo::ProtocolType protocol)
+    : ExtensionSignal(extension_id),
+      remote_host_url_(host_url),
+      protocol_(protocol) {}
 
 RemoteHostContactedSignal::~RemoteHostContactedSignal() = default;
 

chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal.h

@@ -6,25 +6,38 @@
 #define CHROME_BROWSER_SAFE_BROWSING_EXTENSION_TELEMETRY_REMOTE_HOST_CONTACTED_SIGNAL_H_
 
 #include "chrome/browser/safe_browsing/extension_telemetry/extension_signal.h"
+#include "components/safe_browsing/core/common/proto/csd.pb.h"
 #include "url/gurl.h"
 
 namespace safe_browsing {
 
+using RemoteHostInfo = ExtensionTelemetryReportRequest::SignalInfo::
+    RemoteHostContactedInfo::RemoteHostInfo;
+
 // A signal that is created when an extension initiates a web request.
 class RemoteHostContactedSignal : public ExtensionSignal {
  public:
-  RemoteHostContactedSignal(const extensions::ExtensionId& extension_id,
-                            const GURL& host_url);
+  RemoteHostContactedSignal(
+      const extensions::ExtensionId& extension_id,
+      const GURL& host_url,
+      RemoteHostInfo::ProtocolType protocol = RemoteHostInfo::HTTP_HTTPS);
   ~RemoteHostContactedSignal() override;
 
   // ExtensionSignal:
   ExtensionSignalType GetType() const override;
 
-  const GURL& contacted_host_url() const { return contacted_host_url_; }
+  const GURL& remote_host_url() const { return remote_host_url_; }
+
+  const RemoteHostInfo::ProtocolType& protocol_type() const {
+    return protocol_;
+  }
 
  protected:
   // Url of the remote contacted host.
-  GURL contacted_host_url_;
+  GURL remote_host_url_;
+
+  // The protocol used to contact the remote host.
+  RemoteHostInfo::ProtocolType protocol_;
 };
 
 }  // namespace safe_browsing

chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal_processor.cc

@@ -6,7 +6,6 @@
 
 #include "base/check_op.h"
 #include "chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal.h"
-#include "components/safe_browsing/core/common/proto/csd.pb.h"
 
 namespace safe_browsing {
 
@@ -21,15 +20,16 @@ void RemoteHostContactedSignalProcessor::ProcessSignal(
   const auto& rhc_signal =
       static_cast<const RemoteHostContactedSignal&>(signal);
   // Extract only the host portion of the urls.
-  const std::string host_url = rhc_signal.contacted_host_url().host();
-  ++((remote_host_url_store_[rhc_signal.extension_id()])[host_url]);
+  std::string host_url = rhc_signal.remote_host_url().host();
+  ++(((remote_host_info_store_[rhc_signal.extension_id()])
+          [rhc_signal.protocol_type()])[host_url]);
 }
 
 std::unique_ptr<ExtensionTelemetryReportRequest_SignalInfo>
 RemoteHostContactedSignalProcessor::GetSignalInfoForReport(
     const extensions::ExtensionId& extension_id) {
-  auto remote_host_url_store_it = remote_host_url_store_.find(extension_id);
-  if (remote_host_url_store_it == remote_host_url_store_.end())
+  auto remote_host_info_store_it = remote_host_info_store_.find(extension_id);
+  if (remote_host_info_store_it == remote_host_info_store_.end())
     return nullptr;
 
   // Create the signal info protobuf.
@@ -40,21 +40,26 @@ RemoteHostContactedSignalProcessor::GetSignalInfoForReport(
       remote_host_contacted_info =
           signal_info->mutable_remote_host_contacted_info();
 
-  for (auto& remote_host_contacted_urls_it : remote_host_url_store_it->second) {
-    ExtensionTelemetryReportRequest_SignalInfo_RemoteHostContactedInfo_RemoteHostInfo*
-        remote_host_pb = remote_host_contacted_info->add_remote_host();
-    remote_host_pb->set_url(std::move(remote_host_contacted_urls_it.first));
-    remote_host_pb->set_contact_count(remote_host_contacted_urls_it.second);
+  for (auto& remote_host_urls_by_protocol_it :
+       remote_host_info_store_it->second) {
+    for (auto& remote_host_urls_it : remote_host_urls_by_protocol_it.second) {
+      ExtensionTelemetryReportRequest_SignalInfo_RemoteHostContactedInfo_RemoteHostInfo*
+          remote_host_pb = remote_host_contacted_info->add_remote_host();
+      remote_host_pb->set_url(std::move(remote_host_urls_it.first));
+      remote_host_pb->set_contact_count(remote_host_urls_it.second);
+      remote_host_pb->set_connection_protocol(
+          remote_host_urls_by_protocol_it.first);
+    }
   }
 
   // Clear the data in the remote host urls store.
-  remote_host_url_store_.erase(remote_host_url_store_it);
+  remote_host_info_store_.erase(remote_host_info_store_it);
 
   return signal_info;
 }
 
 bool RemoteHostContactedSignalProcessor::HasDataToReportForTest() const {
-  return !remote_host_url_store_.empty();
+  return !remote_host_info_store_.empty();
 }
 
 }  // namespace safe_browsing

chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal_processor.h

@@ -10,6 +10,7 @@
 
 #include "base/containers/flat_map.h"
 #include "chrome/browser/safe_browsing/extension_telemetry/extension_signal_processor.h"
+#include "components/safe_browsing/core/common/proto/csd.pb.h"
 #include "url/gurl.h"
 
 namespace safe_browsing {
@@ -36,12 +37,18 @@ class RemoteHostContactedSignalProcessor : public ExtensionSignalProcessor {
   bool HasDataToReportForTest() const override;
 
  protected:
+  using RemoteHostInfo = ExtensionTelemetryReportRequest::SignalInfo::
+      RemoteHostContactedInfo::RemoteHostInfo;
   // Maps remote hosts url to contact count.
   using RemoteHostURLs = base::flat_map<std::string, uint32_t>;
-  // Maps extension id to (remote host url, contact count).
-  using RemoteHostURLsPerExtension =
-      base::flat_map<extensions::ExtensionId, RemoteHostURLs>;
-  RemoteHostURLsPerExtension remote_host_url_store_;
+  // Maps connection protocols to remote hosts contacted.
+  using RemoteHostURLsByConnectionProtocol =
+      base::flat_map<RemoteHostInfo::ProtocolType, RemoteHostURLs>;
+  // Maps extension id to remote hosts contacted.
+  using RemoteHostInfoPerExtension =
+      base::flat_map<extensions::ExtensionId,
+                     RemoteHostURLsByConnectionProtocol>;
+  RemoteHostInfoPerExtension remote_host_info_store_;
 };
 
 }  // namespace safe_browsing

chrome/browser/safe_browsing/extension_telemetry/remote_host_contacted_signal_processor_unittest.cc

@@ -21,7 +21,8 @@ using RemoteHostInfo =
 
 constexpr const char* kExtensionId[] = {"crx-0", "crx-1"};
 const char* host_urls[] = {"http://www.google.com", "http://www.youtube.com"};
-
+RemoteHostInfo::ProtocolType protocolType[] = {RemoteHostInfo::HTTP_HTTPS,
+                                               RemoteHostInfo::WEBSOCKET};
 class RemoteHostContactedSignalProcessorTest : public ::testing::Test {
  protected:
   RemoteHostContactedSignalProcessorTest() = default;
@@ -36,7 +37,8 @@ TEST_F(RemoteHostContactedSignalProcessorTest,
 
 TEST_F(RemoteHostContactedSignalProcessorTest,
        StoresDataAfterProcessingSignal) {
-  auto signal = RemoteHostContactedSignal(kExtensionId[0], GURL(host_urls[0]));
+  auto signal = RemoteHostContactedSignal(kExtensionId[0], GURL(host_urls[0]),
+                                          protocolType[0]);
   processor_.ProcessSignal(signal);
 
   // Verify that processor now has some data to report.
@@ -47,28 +49,28 @@ TEST_F(RemoteHostContactedSignalProcessorTest,
   EXPECT_FALSE(processor_.GetSignalInfoForReport(kExtensionId[1]));
 }
 
-// Test contacted_count and mapping functionalities.
+// Test contacted_count, protocol type, and mapping functionalities.
 TEST_F(RemoteHostContactedSignalProcessorTest, ReportsSignalInfoCorrectly) {
   // Process 3 signals for the first extension, each corresponding to the
   // web request sent to the first test url.
   for (int i = 0; i < 3; i++) {
-    auto signal =
-        RemoteHostContactedSignal(kExtensionId[0], GURL(host_urls[0]));
-    processor_.ProcessSignal(signal);
+    auto signal = RemoteHostContactedSignal(kExtensionId[0], GURL(host_urls[0]),
+                                            protocolType[0]);
+    processor_.ProcessSignal(std::move(signal));
   }
 
   // Process 3 signals for the second extension. Two signal corresponds to the
   // web request sent to the first url, the third to the web request
   // sent to the second url.
   for (int i = 0; i < 2; i++) {
-    auto signal =
-        RemoteHostContactedSignal(kExtensionId[1], GURL(host_urls[0]));
-    processor_.ProcessSignal(signal);
+    auto signal = RemoteHostContactedSignal(kExtensionId[1], GURL(host_urls[0]),
+                                            protocolType[0]);
+    processor_.ProcessSignal(std::move(signal));
   }
   {
-    auto signal =
-        RemoteHostContactedSignal(kExtensionId[1], GURL(host_urls[1]));
-    processor_.ProcessSignal(signal);
+    auto signal = RemoteHostContactedSignal(kExtensionId[1], GURL(host_urls[1]),
+                                            protocolType[1]);
+    processor_.ProcessSignal(std::move(signal));
   }
 
   // Retrieve signal info for first extension.
@@ -93,33 +95,37 @@ TEST_F(RemoteHostContactedSignalProcessorTest, ReportsSignalInfoCorrectly) {
     const RemoteHostContactedInfo& remote_host_contacted_info =
         extension_0_signal_info->remote_host_contacted_info();
 
-    // Verify data stored : only 1 url (contacted 3 times).
+    // Verify data stored : only 1 url (contacted 3 times) and
+    // contacted via HTTP_HTTPS protocol.
     ASSERT_EQ(remote_host_contacted_info.remote_host_size(), 1);
     const RemoteHostInfo& remote_host_info =
         remote_host_contacted_info.remote_host(0);
     EXPECT_EQ(remote_host_info.contact_count(), static_cast<uint32_t>(3));
+    EXPECT_EQ(remote_host_info.connection_protocol(), protocolType[0]);
   }
 
   // Verify signal info contents for second extension.
   {
     const RemoteHostContactedInfo& remote_host_contacted_info =
         extension_1_signal_info->remote_host_contacted_info();
 
-    // Verify data stored : 2 URLs (2 contacted counts for 1st, 1 for the 2nd).
+    // Verify data stored : 2 URLs (2 contacted counts for 1st, 1 for the 2nd)
+    // and contacted via HTTP_HTTPS and WEBSOCKET protocols.
     ASSERT_EQ(remote_host_contacted_info.remote_host_size(), 2);
     {
       const RemoteHostInfo& remote_host_info =
           remote_host_contacted_info.remote_host(0);
       EXPECT_EQ(remote_host_info.contact_count(), static_cast<uint32_t>(2));
+      EXPECT_EQ(remote_host_info.connection_protocol(), protocolType[0]);
     }
     {
       const RemoteHostInfo& remote_host_info =
           remote_host_contacted_info.remote_host(1);
       EXPECT_EQ(remote_host_info.contact_count(), static_cast<uint32_t>(1));
+      EXPECT_EQ(remote_host_info.connection_protocol(), protocolType[1]);
     }
   }
 }
-
 }  // namespace
 
 }  // namespace safe_browsing

components/safe_browsing/core/common/features.cc

@@ -118,6 +118,10 @@ BASE_FEATURE(kExtensionTelemetryReportContactedHosts,
              "SafeBrowsingExtensionTelemetryReportContactedHosts",
              base::FEATURE_DISABLED_BY_DEFAULT);
 
+BASE_FEATURE(kExtensionTelemetryReportHostsContactedViaWebSocket,
+             "SafeBrowsingExtensionTelemetryReportHostsContactedViaWebsocket",
+             base::FEATURE_DISABLED_BY_DEFAULT);
+
 BASE_FEATURE(kExtensionTelemetryPotentialPasswordTheft,
              "SafeBrowsingExtensionTelemetryPotentialPasswordTheft",
              base::FEATURE_DISABLED_BY_DEFAULT);
@@ -234,6 +238,7 @@ constexpr struct {
     {&kEnhancedProtectionPhase2IOS, true},
     {&kExtensionTelemetry, true},
     {&kExtensionTelemetryReportContactedHosts, true},
+    {&kExtensionTelemetryReportHostsContactedViaWebSocket, true},
     {&kExtensionTelemetryCookiesGetAllSignal, true},
     {&kExtensionTelemetryPersistence, true},
     {&kExtensionTelemetryPotentialPasswordTheft, true},

components/safe_browsing/core/common/features.h

@@ -109,6 +109,9 @@ BASE_DECLARE_FEATURE(kExtensionTelemetryTabsExecuteScriptSignal);
 // Enables reporting of remote hosts contacted by extensions in telemetry.
 BASE_DECLARE_FEATURE(kExtensionTelemetryReportContactedHosts);
 
+// Enables reporting of remote hosts contacted by extensions via websockets;
+BASE_DECLARE_FEATURE(kExtensionTelemetryReportHostsContactedViaWebSocket);
+
 // Enables collection of potential password theft data and uploads
 // telemetry reports to SB servers.
 BASE_DECLARE_FEATURE(kExtensionTelemetryPotentialPasswordTheft);