Telemetry: Create new profile creator large_profile_creator.

tools/perf/profile_creators/history_profile_extender.py

Index: tools/perf/profile_creators/history_profile_extender.py
diff --git a/tools/perf/profile_creators/history_profile_extender.py b/tools/perf/profile_creators/history_profile_extender.py
new file mode 100644
index 0000000000000000000000000000000000000000..fdd932db257a2f23d1d20cc0493013185546e3bf
--- /dev/null
+++ b/tools/perf/profile_creators/history_profile_extender.py
@@ -0,0 +1,78 @@
+# Copyright 2015 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.
+import multiprocessing
+import tempfile
+import os
+
+from profile_creators import fast_navigation_profile_extender
+
+
+class HistoryProfileExtender(
+    fast_navigation_profile_extender.FastNavigationProfileExtender):
+  """This extender navigates Chrome to a large number of URIs pointing to local
+  files. It continues running until the history DB becomes full."""
+  _HISTORY_DB_MAX_SIZE_IN_MB = 10
+
+  def __init__(self):
+    # The rate limiting factors are the speed of page navigation, and the speed
+    # of python bindings. The former is larger than the latter, so having a
+    # large batch size skews the amortized average time per page load towards
+    # the latter.
+    batch_size = multiprocessing.cpu_count() * 2
+    super(HistoryProfileExtender, self).__init__(batch_size)
+
+    # A list of paths of temporary files. The instance is responsible for
+    # making sure that the files are deleted before they are removed from this
+    # list.
+    self._file_paths = []
+
+  def GetUrlsToNavigate(self, maximum_batch_size):
+    """Superclass override.
+
+    This method makes temporary files. It returns URIs to these files, which
+    allows Chrome to populate its history database by navigating to these files
+    on the local file system.
+
+    This method fills the member |_file_paths| with the absolute paths of the
+    temporary files, which must be removed at a later point in time.
+    """
+    urls = []
+    for _ in range(maximum_batch_size):
+      suffix = "reallylongsuffixintendedtoartificiallyincreasethelengthoftheurl"

[nednguyen, 2015/02/19 00:33:32]

Can you add some comment about why we want a long suffix?

[erikchen, 2015/02/19 01:56:03]

Done.

+      handle, path = tempfile.mkstemp(suffix=suffix)
+      os.close(handle)
+      self._file_paths.append(path)
+
+      file_url = "file://" + path
+      urls.append(file_url)
+
+    return urls
+
+  def ShouldExitAfterBatchNavigation(self):
+    """Superclass override."""
+    return self._IsHistoryDBAtMaxSize()
+
+  def TearDown(self):
+    """Superclass override."""
+    super(HistoryProfileExtender, self).TearDown()
+    for path in self._file_paths:
+      os.remove(path)
+    self._file_paths = []
+
+  def CleanUpAfterBatchNavigation(self):
+    """Superclass override."""
+    for path in self._file_paths:
+      os.remove(path)

[nednguyen, 2015/02/19 00:33:32]

You can move TemporaryFile which is deleted on close() instead.

[erikchen, 2015/02/19 01:56:03]

For a full explanation, see my earlier response. In short:

TemporaryFile() does not guarantee the creation of a file-system visible file. 
NamedTemporaryFile() is no different from mkstemp(), since we still need to
close the file handle on Windows, and set the flag Delete=False.

+    self._file_paths = []
+
+  def _IsHistoryDBAtMaxSize(self):
+    """Whether the history DB has reached its maximum size."""
+    history_db_path = os.path.join(self.profile_path, "Default", "History")
+    stat_info = os.stat(history_db_path)
+    size = stat_info.st_size
+
+    max_size_threshold = 0.95
+    max_size = ((10**6 * HistoryProfileExtender._HISTORY_DB_MAX_SIZE_IN_MB) *

[nednguyen, 2015/02/19 00:33:32]

Technically, this should be 2 ** 20?

[erikchen, 2015/02/19 01:56:03]

Done.

+        max_size_threshold)
+    return size > max_size