[Findit] Enable project classifier and component classifier

appengine/findit/crash/project_classifier.py

Index: appengine/findit/crash/project_classifier.py
diff --git a/appengine/findit/crash/project_classifier.py b/appengine/findit/crash/project_classifier.py
new file mode 100644
index 0000000000000000000000000000000000000000..1633a45e6ddfbe1d51b9c5facd08e857fe46521b
--- /dev/null
+++ b/appengine/findit/crash/project_classifier.py
@@ -0,0 +1,108 @@
+# Copyright 2016 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 copy
+
+from crash import classifier
+from crash.type_enums import CallStackLanguageType
+from model.crash.crash_config import CrashConfig
+
+
+class ProjectClassifier(classifier.Classifier):
+  """Determines the project of a crash - (project_name, project_path).
+
+  For example: ('chromium', 'src/'), ('skia', 'src/skia/'), ...etc.
+  """
+
+  def __init__(self):
+    super(ProjectClassifier, self).__init__()
+    self.project_classifier_config = CrashConfig.Get().project_classifier
+    self.project_classifier_config['host_directories'].sort(
+        cmp=self._CompareHosts)

[stgao, 2016/05/24 22:28:48]

This could be simplified by just comparing the length of the string. Longer one
at beginning of the list.

BTW, if we want to sort in place, it's better to do a deepcopy from the config.

[Sharu Jiang, 2016/05/24 23:30:24]

I think only when a host is contained by another, the longer one is smaller. for
others, we should still use alphabetic order.
For example: 'src/testing < src/third_party/WebKit/'

We won't do put() to crash config, even we do somehow(probably will not), it
won't affect the CrashConfig, maybe it's even better we sorted them :) So I
think no need to deepcopy here?

[stgao, 2016/05/24 23:35:36]

For those that are not contained by another, does order really matter for them
like the example you gave?
Even though it won't have side effect at the moment, I guess it is not a good
practice.
Anyway, your call :)

[Sharu Jiang, 2016/05/25 00:08:07]

Done.

+
+  def _CompareHosts(self, host1, host2):
+    """Compares 2 host directories.
+
+    If a host is contained in another host, the shorter the length the smaller,
+    otherwise, use the alphabetic order.
+    For example, 'src/tools' < 'src', 'src/chrome/ < src/media'.
+    """
+    if host1.startswith(host2) or host2.startswith(host1):
+      return -cmp(len(host1), len(host2))
+
+    return cmp(host1, host2)
+
+  def _GetProjectFromDepPath(self, dep_path):
+    """Returns the project name from a dep path."""
+    if not dep_path:
+      return ''
+
+    if dep_path == 'src/':
+      return 'chromium'
+
+    for host_directory in self.project_classifier_config['host_directories']:
+      if dep_path.startswith(host_directory):
+        path = dep_path[len(host_directory):]
+        return 'chromium-%s' % path.split('/')[0].lower()
+
+    # Unknown path, return the whole path as project name.
+    return 'chromium-%s' % '_'.join(dep_path.split('/'))
+
+  def GetClassFromStackFrame(self, frame):
+    """Returns a tuple (project_name, project_path) of a StackFrame."""
+    for marker, name in self.project_classifier_config[
+        'function_marker_to_project_name'].iteritems():
+      if frame.function.startswith(marker):
+        return name
+
+    for marker, name in self.project_classifier_config[
+        'file_path_marker_to_project_name'].iteritems():
+      if marker in frame.file_path or marker in frame.raw_file_path:
+        return name
+
+    return self._GetProjectFromDepPath(frame.dep_path)
+
+  def GetClassFromResult(self, result):
+    """Returns (project_name, project_path) of a Result."""
+    if result.file_to_stack_infos:
+      # A file in culprit result should always have its stack_info, namely a
+      # list of (frame, callstack_priority) pairs.
+      frame, _ = result.file_to_stack_infos.values()[0][0]
+      return self.GetClassFromStackFrame(frame)
+
+    return ''
+
+  def Classify(self, results, crash_stack):
+    """Classify project of a crash.
+
+    Args:
+      results (list of Result): culprit results.
+      crash_stack (CallStack): the callstack that caused the crash.
+
+    Returns:
+      A tuple, project of the crash - (project_name, project_path).
+    """
+    def _GetRankFunction(language_type):
+      if language_type == CallStackLanguageType.JAVA:
+        def _RankFunctionForJava(class_occurrences_info):
+          project_name = class_occurrences_info.name
+          return (len(class_occurrences_info.occurrences),
+                  0 if 'chromium' in project_name else
+                  self.project_classifier_config[
+                      'non_chromium_project_rank_priority'][project_name])
+
+        return _RankFunctionForJava
+
+      return classifier.DefaultRankFunction
+
+    # Set the max_classes to 1, so the returned projects only has one element.
+    projects = self._Classify(
+        results, crash_stack,
+        self.project_classifier_config['top_n'], 1,
+        rank_function=_GetRankFunction(crash_stack.language_type))
+
+    if projects:
+      return projects[0]
+
+    return ''