Seculative resource prefetching for URLs CL.

chrome/browser/predictors/resource_prefetch_predictor_tables.cc

Index: chrome/browser/predictors/resource_prefetch_predictor_tables.cc
diff --git a/chrome/browser/predictors/resource_prefetch_predictor_tables.cc b/chrome/browser/predictors/resource_prefetch_predictor_tables.cc
new file mode 100644
index 0000000000000000000000000000000000000000..a8ed76431d24a83dd80aa4d5de4db3f5c3a06cad
--- /dev/null
+++ b/chrome/browser/predictors/resource_prefetch_predictor_tables.cc
@@ -0,0 +1,218 @@
+// Copyright (c) 2012 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 "chrome/browser/predictors/resource_prefetch_predictor_tables.h"
+
+#include "base/logging.h"
+#include "base/metrics/histogram.h"
+#include "base/stringprintf.h"
+#include "content/public/browser/browser_thread.h"
+#include "sql/statement.h"
+
+using content::BrowserThread;
+
+namespace {
+
+const char kResourcePredictorUrlTableName[] = "resource_prefetch_predictor_url";
+
+// The maximum length allowed for strings in the database.
+const size_t kMaxURLLength = 2048;
+
+void BindUrlTableRowToStatement(
+    const predictors::ResourcePrefetchPredictorTables::UrlTableRow& row,
+    sql::Statement* statement) {
+  statement->BindString(0, row.main_frame_url_.spec().substr(0, kMaxURLLength));
+  statement->BindString(1, row.resource_url_.spec().substr(0, kMaxURLLength));
+  statement->BindInt(2, static_cast<int>(row.resource_type_));
+  statement->BindInt(3, row.number_of_hits_);
+  statement->BindInt(4, row.number_of_misses_);
+  statement->BindInt(5, row.consecutive_misses_);
+  statement->BindDouble(6, row.average_position_);
+}
+
+bool StepAndInitializeUrlTableRow(
+    sql::Statement* statement,
+    predictors::ResourcePrefetchPredictorTables::UrlTableRow* row) {
+  if (!statement->Step())
+    return false;
+
+  row->main_frame_url_ = GURL(statement->ColumnString(0));
+  row->resource_url_ = GURL(statement->ColumnString(1));
+  row->resource_type_ = ResourceType::FromInt(statement->ColumnInt(2));
+  row->number_of_hits_ = statement->ColumnInt(3);
+  row->number_of_misses_ = statement->ColumnInt(4);
+  row->consecutive_misses_ = statement->ColumnInt(5);
+  row->average_position_ = statement->ColumnDouble(6);
+  return true;
+}
+
+}  // namespace
+
+namespace predictors {
+
+ResourcePrefetchPredictorTables::UrlTableRow::UrlTableRow()
+    : resource_type_(ResourceType::LAST_TYPE),
+      number_of_hits_(0),
+      number_of_misses_(0),
+      consecutive_misses_(0),
+      average_position_(0),

[dominich, 2012/05/21 16:16:53]

warning: initializing a double with an int.

[Shishir, 2012/05/23 01:46:46]

Done.

+      score_(0.0) {
+}
+
+ResourcePrefetchPredictorTables::UrlTableRow::UrlTableRow(
+    const UrlTableRow& other)
+        : main_frame_url_(other.main_frame_url_),
+          resource_url_(other.resource_url_),
+          resource_type_(other.resource_type_),
+          number_of_hits_(other.number_of_hits_),
+          number_of_misses_(other.number_of_misses_),
+          consecutive_misses_(other.consecutive_misses_),
+          average_position_(other.average_position_),
+          score_(other.score_) {
+}
+
+void ResourcePrefetchPredictorTables::UrlTableRow::UpdateScore() {

[dominich, 2012/05/21 16:16:53]

Comment explaining the heuristic here. Also, might this heuristic change or
might there be alternative field trials?

[Shishir, 2012/05/23 01:46:46]

Added comment. We might have additional field trials, but we want to start with
this simple heuristic.

+  static const int kMaxResourcesPerType = 100;
+  switch (resource_type_) {
+    case ResourceType::STYLESHEET:
+    case ResourceType::SCRIPT:
+      score_ = (2 * kMaxResourcesPerType) - average_position_;
+      break;
+
+    case ResourceType::IMAGE:
+      score_ = kMaxResourcesPerType - average_position_;
+      break;
+
+    default:

[dominich, 2012/05/21 16:16:53]

What other resource types should be handled? Maybe you can NOTREACHED in the
default case.

[Shishir, 2012/05/23 01:46:46]

Sometimes we do not know the type of resource, eg if a resource is loaded by
webkit inmemory cache or obtained through a 304. We still want to prefetch it,
since its cacheable.

+      score_ = kMaxResourcesPerType - average_position_;
+      break;
+  }
+}
+
+bool ResourcePrefetchPredictorTables::UrlTableRowSorter::operator()(
+    const UrlTableRow& x,
+    const UrlTableRow& y) const {
+  return x.score_ > y.score_;
+}
+
+ResourcePrefetchPredictorTables::ResourcePrefetchPredictorTables()
+    : PredictorTableBase() {
+}
+
+ResourcePrefetchPredictorTables::~ResourcePrefetchPredictorTables() {
+}
+
+void ResourcePrefetchPredictorTables::CreateTableIfNonExistent() {
+  CHECK(BrowserThread::CurrentlyOn(BrowserThread::DB));
+  if (CantAccessDatabase())
+    return;
+
+  std::string url_table_creation_statement = base::StringPrintf(
+      "CREATE TABLE %s ( "
+      "main_page_url TEXT, "
+      "resource_url TEXT, "
+      "resource_type INTEGER, "
+      "number_of_hits INTEGER, "
+      "number_of_misses INTEGER, "
+      "consecutive_misses INTEGER, "
+      "average_position DOUBLE, "
+      "PRIMARY KEY(main_page_url, resource_url))",
+      kResourcePredictorUrlTableName);
+
+  if (!DB()->DoesTableExist(kResourcePredictorUrlTableName))

[dominich, 2012/05/21 16:16:53]

add braces as the inner if has them.

[Shishir, 2012/05/23 01:46:46]

Done.

+    if (!DB()->Execute(url_table_creation_statement.c_str())) {
+      ResetDB();
+      return;
+    }
+}
+
+void ResourcePrefetchPredictorTables::LogDatabaseStats()  {

[dominich, 2012/05/21 16:16:53]

Consider tracking the raw file size.

[Shishir, 2012/05/23 01:46:46]

That happens in the PredictorDatabaseInternal.

+  CHECK(BrowserThread::CurrentlyOn(BrowserThread::DB));
+  if (CantAccessDatabase())
+    return;
+
+  sql::Statement url_statement(DB()->GetUniqueStatement(
+      base::StringPrintf("SELECT count(*) FROM %s",
+                         kResourcePredictorUrlTableName).c_str()));
+  if (url_statement.Step())
+    UMA_HISTOGRAM_COUNTS("ResourcePrefetchPredictor.UrlTableRowCount",
+                         url_statement.ColumnInt(0));
+}
+
+void ResourcePrefetchPredictorTables::GetAllRows(
+    std::vector<UrlTableRow>* url_row_buffer) {
+  CHECK(BrowserThread::CurrentlyOn(BrowserThread::DB));
+  if (CantAccessDatabase())
+    return;
+
+  CHECK(url_row_buffer);
+  url_row_buffer->clear();
+  sql::Statement url_statement(DB()->GetCachedStatement(SQL_FROM_HERE,
+      base::StringPrintf("SELECT * FROM %s",
+                         kResourcePredictorUrlTableName).c_str()));
+
+  UrlTableRow url_row;
+  while (StepAndInitializeUrlTableRow(&url_statement, &url_row)) {
+    url_row.UpdateScore();

[dominich, 2012/05/21 16:16:53]

why don't you store the score?

[Shishir, 2012/05/23 01:46:46]

Not storing it right now because we want to run experiments and keep changing
it. May store it in the future if its not a simple function of the other fields.

+    url_row_buffer->push_back(url_row);
+  }
+}
+
+void ResourcePrefetchPredictorTables::UpdateRowsForUrl(
+    const GURL& main_page_url,
+    const std::vector<UrlTableRow>& row_buffer) {
+  CHECK(BrowserThread::CurrentlyOn(BrowserThread::DB));
+  if (CantAccessDatabase())
+    return;
+
+  DB()->BeginTransaction();
+
+  sql::Statement delete_statement(DB()->GetUniqueStatement(

[dominich, 2012/05/21 16:16:53]

would it be more optimal to SELECT and then UPDATE/INSERT as necessary?

[Shishir, 2012/05/23 01:46:46]

May not be, since we will end up updating all the rows for the URL.

+      base::StringPrintf("DELETE FROM %s WHERE main_page_url='%s'",
+                         kResourcePredictorUrlTableName,
+                         main_page_url.spec().c_str()).c_str()));
+  if (!delete_statement.Run()) {
+    DB()->RollbackTransaction();
+    return;
+  }
+
+  for (int i = 0; i < static_cast<int>(row_buffer.size()); ++i) {
+    const UrlTableRow& row = row_buffer.at(i);
+
+    sql::Statement add_statement(
+        DB()->GetCachedStatement(
+            SQL_FROM_HERE,
+            base::StringPrintf(
+                "INSERT INTO %s "
+                "(main_page_url, resource_url, resource_type, number_of_hits, "
+                "number_of_misses, consecutive_misses, average_position) "
+                "VALUES (?,?,?,?,?,?,?)",
+                kResourcePredictorUrlTableName).c_str()));
+    BindUrlTableRowToStatement(row, &add_statement);
+    if (!add_statement.Run()) {
+      DB()->RollbackTransaction();
+      return;
+    }
+  }
+
+  DB()->CommitTransaction();
+}
+
+void ResourcePrefetchPredictorTables::DeleteUrlRows(
+    const std::vector<GURL>& urls) {
+  CHECK(BrowserThread::CurrentlyOn(BrowserThread::DB));
+  if (CantAccessDatabase())
+    return;
+
+  // These do not need to be a transaction.
+  for (int i = 0; i < static_cast<int>(urls.size()); ++i) {
+    sql::Statement delete_statement(DB()->GetCachedStatement(
+        SQL_FROM_HERE,
+        base::StringPrintf("DELETE FROM %s WHERE main_page_url='%s'",
+                           kResourcePredictorUrlTableName,
+                           urls[i].spec().c_str()).c_str()));
+  }
+}
+
+}  // namespace predictors