RAPPOR implementation

components/rappor/rappor_service.cc

Index: components/rappor/rappor_service.cc
diff --git a/components/rappor/rappor_service.cc b/components/rappor/rappor_service.cc
new file mode 100644
index 0000000000000000000000000000000000000000..e5334e0cd108e6c65693bc39c7380a923168ba84
--- /dev/null
+++ b/components/rappor/rappor_service.cc
@@ -0,0 +1,196 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "components/rappor/rappor_service.h"
+
+#include <cstdlib>
+
+#include "base/base64.h"
+#include "base/hash.h"
+#include "base/metrics/field_trial.h"
+#include "base/prefs/pref_registry_simple.h"
+#include "base/prefs/pref_service.h"
+#include "base/rand_util.h"
+#include "base/stl_util.h"
+#include "components/rappor/proto/rappor_metric.pb.h"
+#include "components/rappor/rappor_pref_names.h"
+#include "components/variations/variations_associated_data.h"
+#include "url/gurl.h"
+
+namespace {
+
+// The number of cohorts we divide clients into.
+const int kNumCohorts = 8;
+
+// Seconds before the initial log is generated.
+const int kInitialLogIntervalSeconds = 15;
+// Interval between ongoing logs.
+const int kLogIntervalSeconds = 30 * 60;
+
+const char kMimeType[] = "application/vnd.chrome.rappor";
+
+// Constants for the RAPPOR rollout field trial.
+const char kRapporRolloutFieldTrialName[] = "RapporRollout";
+
+// Constant for the finch parameter name for the server URL
+const char kRapporRolloutServerUrlParam[] = "ServerUrl";
+
+GURL GetServerUrl() {
+  return GURL(chrome_variations::GetVariationParamValue(
+      kRapporRolloutFieldTrialName,
+      kRapporRolloutServerUrlParam));
+}
+
+}  // namespace
+
+namespace rappor {
+
+RapporService::RapporService() : cohort_(-1) {}
+
+RapporService::~RapporService() {}
+
+void RapporService::Start(PrefService* pref_service,
+                          net::URLRequestContextGetter* request_context) {
+  GURL server_url = GetServerUrl();
+  if (!server_url.is_valid())
+    return;
+  DCHECK(!uploader_);
+  LoadSecret(pref_service);
+  LoadCohort(pref_service);
+  uploader_.reset(new LogUploader(server_url, kMimeType, request_context));
+  log_rotation_timer_.Start(
+      FROM_HERE,
+      base::TimeDelta::FromSeconds(kInitialLogIntervalSeconds),
+      this,
+      &RapporService::OnLogInterval);
+}
+
+void RapporService::OnLogInterval() {
+  DCHECK(uploader_);
+  RapporReports reports;
+  bool metrics_logged = LogMetrics(&reports);
+  if (metrics_logged) {
+    std::string log_text;
+    bool success = reports.SerializeToString(&log_text);
+    DCHECK(success);
+    uploader_->QueueLog(log_text);
+  }
+  log_rotation_timer_.Start(FROM_HERE,
+                            base::TimeDelta::FromSeconds(kLogIntervalSeconds),
+                            this,
+                            &RapporService::OnLogInterval);
+}
+
+// static
+void RapporService::RegisterPrefs(PrefRegistrySimple* registry) {
+  registry->RegisterStringPref(prefs::kRapporSecret, std::string());
+  registry->RegisterIntegerPref(prefs::kRapporCohort, -1);
+}
+
+void RapporService::LoadCohort(PrefService* pref_service) {

[Alexei Svitkine (slow), 2014/01/24 22:09:03]

Add a unit test for this and LoadSecret().

+  DCHECK_EQ(cohort_, -1);
+  cohort_ = pref_service->GetInteger(prefs::kRapporCohort);
+  if (cohort_ > 0 && cohort_ < kNumCohorts)
+    return;
+
+  cohort_ = base::RandGenerator(kNumCohorts);
+  pref_service->SetInteger(prefs::kRapporCohort, cohort_);
+}
+
+void RapporService::LoadSecret(PrefService* pref_service) {
+  DCHECK(secret_.empty());
+  std::string secret_base64 =
+      pref_service->GetString(prefs::kRapporSecret);
+  if (!secret_base64.empty()) {
+    bool decoded = base::Base64Decode(secret_base64, &secret_);
+    if (decoded)
+      return;
+    // If the preference fails to decode, it must be corrupt, so continue as
+    // though it didn't exist yet and generate a new one.
+  }
+
+  secret_ = base::RandBytesAsString(128);
+  base::Base64Encode(secret_, &secret_base64);
+  pref_service->SetString(prefs::kRapporSecret, secret_base64);
+}
+
+void RapporService::LogMetric(const RapporMetric& metric,
+                              RapporReports::Report* report) {
+  report->set_name_hash(base::Hash(metric.parameters()->rappor_name));
+  ByteVector bytes = metric.GetReport(secret_);
+  std::string byte_string(bytes.begin(), bytes.end());
+  report->set_bits(byte_string);
+}
+
+bool RapporService::LogMetrics(RapporReports* reports) {
+  base::AutoLock auto_lock(lock_);
+  if (metrics_map_.empty())
+    return false;
+
+  DCHECK_GE(cohort_, 0);
+  reports->set_cohort(cohort_);
+
+  for (std::map<std::string, RapporMetric*>::iterator it = metrics_map_.begin();
+       metrics_map_.end() != it;
+       ++it) {
+    const RapporMetric* metric = it->second;
+    DCHECK_EQ(it->first, metric->parameters()->rappor_name);
+    RapporReports::Report* report = reports->add_report();
+    LogMetric(*metric, report);
+  }
+  STLDeleteContainerPairSecondPointers(
+      metrics_map_.begin(), metrics_map_.end());
+  metrics_map_.clear();
+  return true;
+}
+
+void RapporService::RecordSamples(const RapporParameters& parameters,
+                                  const std::vector<std::string>& samples) {
+  // Ignore the sample if the service hasn't started yet.
+  if (!cohort_ < 0)

[Alexei Svitkine (slow), 2014/01/24 22:09:03]

Make a helper function for this, rather than duplicating this logic in several
places.

+    return;
+  RecordSamplesInternal(parameters, samples);
+}
+
+void RapporService::RecordUrl(const RapporParameters& parameters,
+                              const GURL& url) {
+  // Ignore the sample if the service hasn't started yet.
+  if (cohort_ < 0)
+    return;
+  std::vector<std::string> samples;
+  samples.push_back(url.spec());
+  samples.push_back(url.host());
+  // TODO(holte): break up host more
+  samples.push_back(url.query());
+  samples.push_back(url.path());
+  RecordSamplesInternal(parameters, samples);
+}
+
+void RapporService::RecordSamplesInternal(
+    const RapporParameters& parameters,
+    const std::vector<std::string>& samples) {
+  DCHECK_GE(cohort_, 0);
+  base::AutoLock auto_lock(lock_);
+
+  RapporMetric* metric = LookupMetric(parameters);
+  metric->AddSamples(samples);
+}
+
+RapporMetric* RapporService::LookupMetric(
+    const RapporParameters& parameters) {
+  DCHECK_GE(cohort_, 0);
+  std::map<std::string, RapporMetric*>::iterator it =
+      metrics_map_.find(parameters.rappor_name);
+  if (metrics_map_.end() != it) {
+    RapporMetric* metric = it->second;
+    DCHECK_EQ(parameters.ToString(), metric->parameters()->ToString());
+    return metric;
+  }
+
+  RapporMetric* new_metric = new RapporMetric(parameters, cohort_);
+  metrics_map_[parameters.rappor_name] = new_metric;
+  return new_metric;
+}
+
+}  // namespace rappor