Precache tracking database

components/precache/core/precache_statistics_table.cc

Index: components/precache/core/precache_statistics_table.cc
diff --git a/components/precache/core/precache_statistics_table.cc b/components/precache/core/precache_statistics_table.cc
new file mode 100644
index 0000000000000000000000000000000000000000..aaf98ad6ed18ebe592bfadf81dc5e7c842c06347
--- /dev/null
+++ b/components/precache/core/precache_statistics_table.cc
@@ -0,0 +1,168 @@
+// Copyright 2013 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/precache/core/precache_statistics_table.h"
+
+#include "base/logging.h"
+#include "base/strings/stringprintf.h"
+#include "sql/connection.h"
+#include "sql/statement.h"
+
+using sql::Statement;
+
+namespace {
+
+const char kStatisticsTableName[] = "precache_statistics";
+
+// Rounds |time_on_day| down to midnight of the start of the day of
+// |time_on_day|, and returns the internal value.
+int64 GetKey(const base::Time& time_on_day) {
+  return time_on_day.LocalMidnight().ToInternalValue();

[davidben, 2013/11/06 23:50:53]

What happens if the user changes time zones? Seems you could end up with some
strange results if you key dates by their local time midnight time. Not sure
what would be a better mechanism. Could you store hourly in UTC and then
aggregate after the statistics have been uploaded? I guess that wouldn't give
you grouping by local days either.

It does seem a little odd to me how you have data aggregated by day, but then
UMA itself also can aggregate histograms and such. Particularly with the old
days' data thing mmenke mentioned. But I'm not actually very familiar with our
metrics.

[sclittle, 2013/11/09 01:07:58]

Good point about time zones. If we switch time zones, the keys for the day would
change, and weird stuff would happen. I've changed it to not care about local
times at all, and just track 24 hour intervals.

The reason that we track daily stats, as opposed to just reporting UMA per
request, is because we want to be able to measure the usefulness and overhead of
precaching for the typical user. Reporting UMA per-request would show us the
usefulness and overhead for the typical request, which we don't care about.
Using the overall sums of these per-request stats wouldn't tell us anything
about typical users either, since it would be biased toward outliers who made a
lot of big requests.

[davidben, 2013/11/11 19:06:26]

Ah, so if you rely on UMA's aggregation, you'll weight a user's contributions by
the number of requests they make. Makes sense. (I'm not very familiar with you
can/can't do with UMA.)

+}
+
+}  // namespace
+
+namespace precache {
+
+PrecacheStatisticsTable::PrecacheStatistics::PrecacheStatistics()
+    : downloaded_precache_motivated_bytes(0),
+      downloaded_non_precache_bytes(0),
+      downloaded_non_precache_bytes_cellular(0),
+      saved_bytes(0),
+      saved_bytes_cellular(0) {}
+
+PrecacheStatisticsTable::PrecacheStatistics::PrecacheStatistics(
+    int64 downloaded_precache_motivated_bytes,
+    int64 downloaded_non_precache_bytes,
+    int64 downloaded_non_precache_bytes_cellular, int64 saved_bytes,
+    int64 saved_bytes_cellular)
+    : downloaded_precache_motivated_bytes(downloaded_precache_motivated_bytes),
+      downloaded_non_precache_bytes(downloaded_non_precache_bytes),
+      downloaded_non_precache_bytes_cellular(
+          downloaded_non_precache_bytes_cellular),
+      saved_bytes(saved_bytes),
+      saved_bytes_cellular(saved_bytes_cellular) {}
+
+bool PrecacheStatisticsTable::PrecacheStatistics::operator==(
+    const PrecacheStatistics& other) const {
+  return downloaded_precache_motivated_bytes ==
+             other.downloaded_precache_motivated_bytes &&
+         downloaded_non_precache_bytes == other.downloaded_non_precache_bytes &&
+         downloaded_non_precache_bytes_cellular ==
+             other.downloaded_non_precache_bytes_cellular &&
+         saved_bytes == other.saved_bytes &&
+         saved_bytes_cellular == other.saved_bytes_cellular;
+}
+
+PrecacheStatisticsTable::PrecacheStatisticsTable() : db_(NULL) {}
+
+PrecacheStatisticsTable::~PrecacheStatisticsTable() {}
+
+void PrecacheStatisticsTable::Init(sql::Connection* db) {
+  DCHECK(!db_);  // Init must only be called once.
+  DCHECK(db);    // The database connection must be non-NULL.
+  db_ = db;
+  CreateTableIfNonExistent();
+}
+
+void PrecacheStatisticsTable::IncreaseDailyStats(
+    const base::Time& time_on_day, const PrecacheStatistics& delta) {
+  DCHECK(db_);
+
+  // Create a row for that day if there is no existing one.
+  Statement insert_statement(db_->GetCachedStatement(
+      SQL_FROM_HERE,
+      base::StringPrintf("INSERT OR IGNORE INTO %s (date) VALUES(?)",
+                         kStatisticsTableName).c_str()));
+
+  insert_statement.BindInt64(0, GetKey(time_on_day));
+  insert_statement.Run();
+
+  // Increase the recorded statistics for that day.
+  Statement update_statement(db_->GetCachedStatement(
+      SQL_FROM_HERE,
+      base::StringPrintf(
+          "UPDATE %s SET "
+              "downloaded_precache_motivated_bytes = "
+                  "downloaded_precache_motivated_bytes + ?, "
+              "downloaded_non_precache_bytes = "
+                  "downloaded_non_precache_bytes + ?, "
+              "downloaded_non_precache_bytes_cellular = "
+                  "downloaded_non_precache_bytes_cellular + ?, "
+              "saved_bytes = saved_bytes + ?, "
+              "saved_bytes_cellular = saved_bytes_cellular + ? "
+          "WHERE date = ?", kStatisticsTableName).c_str()));
+
+  update_statement.BindInt64(0, delta.downloaded_precache_motivated_bytes);
+  update_statement.BindInt64(1, delta.downloaded_non_precache_bytes);
+  update_statement.BindInt64(2, delta.downloaded_non_precache_bytes_cellular);
+  update_statement.BindInt64(3, delta.saved_bytes);
+  update_statement.BindInt64(4, delta.saved_bytes_cellular);
+  update_statement.BindInt64(5, GetKey(time_on_day));
+
+  update_statement.Run();
+}
+
+void PrecacheStatisticsTable::GetAllStatsUntil(
+    const base::Time& end_date, PrecacheStatisticsMap* stats_map) {
+  DCHECK(db_);
+  DCHECK(stats_map);  // |stats_map| must not be NULL.
+  stats_map->clear();
+
+  Statement statement(db_->GetCachedStatement(
+      SQL_FROM_HERE, base::StringPrintf(
+                         "SELECT date, "
+                         "downloaded_precache_motivated_bytes, "
+                         "downloaded_non_precache_bytes, "
+                         "downloaded_non_precache_bytes_cellular, "
+                         "saved_bytes, "
+                         "saved_bytes_cellular "
+                         "FROM %s WHERE date <= ?",
+                         kStatisticsTableName).c_str()));
+  statement.BindInt64(0, GetKey(end_date));
+
+  while (statement.Step()) {
+    base::Time date = base::Time::FromInternalValue(statement.ColumnInt64(0));
+    PrecacheStatistics stats;
+
+    stats.downloaded_precache_motivated_bytes = statement.ColumnInt64(1);
+    stats.downloaded_non_precache_bytes = statement.ColumnInt64(2);
+    stats.downloaded_non_precache_bytes_cellular = statement.ColumnInt64(3);
+    stats.saved_bytes = statement.ColumnInt64(4);
+    stats.saved_bytes_cellular = statement.ColumnInt64(5);
+
+    (*stats_map)[date] = stats;
+  }
+}
+
+void PrecacheStatisticsTable::DeleteAllStatsUntil(const base::Time& end_date) {
+  DCHECK(db_);
+
+  Statement statement(db_->GetCachedStatement(
+      SQL_FROM_HERE, base::StringPrintf("DELETE FROM %s WHERE date <= ?",
+                                        kStatisticsTableName).c_str()));
+
+  statement.BindInt64(0, GetKey(end_date));
+  statement.Run();
+}
+
+void PrecacheStatisticsTable::CreateTableIfNonExistent() {
+  DCHECK(db_);
+  if (db_->DoesTableExist(kStatisticsTableName))
+    return;
+
+  if (!db_->Execute(base::StringPrintf(
+           "CREATE TABLE %s (date INTEGER, "
+           "downloaded_precache_motivated_bytes INTEGER DEFAULT 0, "
+           "downloaded_non_precache_bytes INTEGER DEFAULT 0, "
+           "downloaded_non_precache_bytes_cellular INTEGER DEFAULT 0, "
+           "saved_bytes INTEGER DEFAULT 0, "
+           "saved_bytes_cellular INTEGER DEFAULT 0, "
+           "PRIMARY KEY(date))",
+           kStatisticsTableName).c_str())) {
+    DLOG(WARNING) << "Could not create precache statistics table in database.";
+  }
+}
+
+}  // namespace precache