def FeatureValueGivenReport(suspect):
"""Compute ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
Returns:
The ``FeatureValue`` of this feature.
"""
grouped_frame_infos = dep_to_grouped_frame_infos.get(suspect.dep_path, {})
matches = crash_util.MatchSuspectWithFrameInfos(suspect,
grouped_frame_infos,
self.Match)
if not matches:
return FeatureValue(name=self.name,
value=0.0,
reason=None,
changed_files=None)
return FeatureValue(
name=self.name,
value=1.0,
reason='\n'.join([str(match) for match in matches.itervalues()]),
changed_files=None)
def FeatureValueGivenReport(suspect, matches): # pylint: disable=W0613
"""Compute ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
matches(dict): Dict mapping crashed group(CrashedFile, CrashedDirectory)
to a list of ``Match``s representing all frames and all touched files
matched in the same crashed group(same crashed file or crashed
directory).
Returns:
The ``FeatureValue`` of this feature.
"""
if not matches:
return FeatureValue(name=self.name,
value=0.0,
reason=None,
changed_files=None)
return FeatureValue(name=self.name,
value=1.0,
reason='Touched files in stacktrace - %s' %
', '.join([
os.path.basename(touched_file.new_path)
for match in matches.itervalues()
for touched_file in match.touched_files
]),
changed_files=None)
def FeatureValueGivenReport(suspect):
"""Compute ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
Returns:
The ``FeatureValue`` of this feature.
"""
grouped_frame_infos = dep_to_grouped_frame_infos.get(
suspect.dep_path, {})
matches = crash_util.MatchSuspectWithFrameInfos(
suspect, grouped_frame_infos, self.Match)
if not matches:
return FeatureValue(name=self.name,
value=0.0,
reason=None,
changed_files=None)
crashed_directories = [directory.value for directory in matches]
plural = len(crashed_directories) > 1
reason = [
'Suspected changelist touched file(s) in the %s %s, which '
'appear%s in the stack trace.' %
('directories' if plural else 'directory',
', '.join(crashed_directories), '' if plural else 's')
]
return FeatureValue(name=self.name,
value=1.0,
reason=reason,
changed_files=None)
def __call__(self, x):
return lambda y: FeatureValue(self.name, y == (x > 5), 'reason0',
[ChangedFile(name='a.cc',
blame_url=None,
reasons=['file_reason0']),
ChangedFile(name='b.cc',
blame_url=None,
reasons=['file_reason0'])])
def FeatureValueGivenReport(suspect, matches): # pylint: disable=W0613
"""Computes ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
matches(dict): Dict mapping crashed group(CrashedFile, CrashedDirectory)
to a list of ``CrashMatch``s representing all frames and all touched
files matched in the same crashed group(same crashed file or crashed
directory).
Returns:
The ``FeatureValue`` of this feature.
"""
if not matches:
return FeatureValue(name=self.name,
value=0,
reason=None,
changed_files=None)
def GetMaxIdfInMatches(matches):
"""Gets the maximum idf and the file path with the maximum idf."""
file_paths = [
os.path.join(frame_info.frame.dep_path,
frame_info.frame.file_path)
for match in matches.itervalues()
for frame_info in match.frame_infos
]
max_idf, max_idf_file_path = max([(ComputeIdf(
file_path, self._inverted_index_table), file_path)
for file_path in file_paths],
key=lambda item: item[0])
return max_idf, max_idf_file_path
logging.info('Computing crash idf feature')
max_idf, max_idf_file_path = GetMaxIdfInMatches(matches)
value = LogRegressNomalize(max_idf)
reason = [
'Suspected changelist modified the file %s, which is not '
'commonly changed.' % os.path.basename(max_idf_file_path)
]
return FeatureValue(self.name, value, reason, None)
def FeatureValueGivenReport(suspect, matches): # pylint: disable=W0613
"""Computes ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
matches(dict): Dict mapping crashed group(CrashedFile, CrashedDirectory)
to a list of ``Match``s representing all frames and all touched files
matched in the same crashed group(same crashed file or crashed
directory).
Returns:
The ``FeatureValue`` of this feature.
"""
if not matches:
return FeatureValue(name=self.name,
value=0,
reason=None,
changed_files=None)
def TopFrameInMatches(matches):
frames = [
frame_info.frame for match in matches.itervalues()
for frame_info in match.frame_infos
]
frames.sort(key=lambda frame: frame.index)
return frames[0]
top_frame = TopFrameInMatches(matches)
value = LinearlyScaled(float(top_frame.index),
float(self.max_frame_index))
if value <= _MINIMUM_FEATURE_VALUE:
reason = None
else:
reason = 'Top touched frame is #%d %s(in %s)' % (
top_frame.index, re.sub('\(.*\)', '', top_frame.function),
os.path.basename(top_frame.file_path))
return FeatureValue(self.name, value, reason, None)
def FeatureValueGivenReport(suspect, matches): # pylint: disable=W0613
"""Computes ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
matches(dict): Dict mapping crashed group(CrashedFile, CrashedDirectory)
to a list of ``Match``s representing all frames and all touched files
matched in the same crashed group(same crashed file or crashed
directory).
Returns:
The ``FeatureValue`` of this feature.
"""
if not matches:
return FeatureValue(name=self.name,
value=0,
reason=None,
changed_files=None)
def TopFrameInMatches(matches):
frames = [
frame_info.frame for match in matches.itervalues()
for frame_info in match.frame_infos
]
frames.sort(key=lambda frame: frame.index)
return frames[0]
top_frame = TopFrameInMatches(matches)
value = LinearlyScaled(float(top_frame.index),
float(self.max_frame_index))
# This feature reason is redundant to MinDistance or TouchedCrashedFile
# features, which also mentioned which files the suspected cl touched with
# more useful information. So the reason for this feature is not very
# helpful, we are not providing the reason for this feature.
return FeatureValue(self.name, value, None, None)
def FeatureValueGivenReport(suspect):
"""Compute ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
Returns:
The ``FeatureValue`` of this feature.
"""
grouped_frame_infos = dep_to_grouped_frame_infos.get(
suspect.dep_path, {})
matches = crash_util.MatchSuspectWithFrameInfos(
suspect, grouped_frame_infos,
self.GetMatchFunction(suspect.dep_path))
if not matches:
return FeatureValue(name=self.name,
value=0.0,
reason=None,
changed_files=None)
crashed_components = [component.value for component in matches]
plural = len(crashed_components) > 1
reason = [
'Suspected changelist touched file(s) associated with the '
'component%s %s, which we believe is related to this testcase based '
'on information in OWNERS files.' %
('s' if plural else '', ', '.join(crashed_components))
]
return FeatureValue(name=self.name,
value=1.0,
reason=reason,
changed_files=None)
def FeatureValueGivenReport(suspect): # pylint: disable=W0613
"""Compute ``FeatureValue`` for a suspect.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
Returns:
The ``FeatureValue`` of this feature.
"""
number_of_files = len(suspect.changelog.touched_files)
return FeatureValue(name=self.name,
value=LinearlyScaled(number_of_files,
self._maximum_files,
self._minimum_files,
offset=-1.0),
reason=None,
changed_files=None)
def FeatureValueGivenReport(suspect, matches):
"""Function mapping suspect related data to MinDistance FeatureValue.
Args:
suspect (Suspect): The suspected changelog and some meta information
about it.
matches(dict): Dict mapping crashed group(CrashedFile, CrashedDirectory)
to a list of ``Match``s representing all frames and all touched files
matched in the same crashed group(same crashed file or crashed
directory).
Returns:
The ``FeatureValue`` of this feature.
"""
if not matches:
FeatureValue(name=self.name,
value=0.0,
reason=None,
changed_files=None)
distance = Distance(float('inf'), None)
touched_file_to_distance = {}
for match in matches.itervalues():
if len(match.touched_files) != 1:
logging.warning('There should be only one touched file per crashed '
'file group.')
continue
touched_file = match.touched_files[0]
# Records the closest frame (the frame has minimum distance between
# crashed lines and touched lines) for each touched file of the suspect.
distance_per_file = self.DistanceBetweenTouchedFileAndFrameInfos(
suspect.changelog.revision, touched_file,
match.frame_infos, report.dependencies[suspect.dep_path])
# Failed to get blame information of a file.
if not distance_per_file:
logging.warning('suspect\'s change cannot be blamed due to lack of'
'blame information for crashed file %s' %
touched_file.new_path)
continue
# It is possible that a changelog doesn't show in the blame of a file,
# in this case, treat the changelog as if it didn't change the file.
if distance_per_file.IsInfinity():
continue
touched_file_to_distance[touched_file] = distance_per_file
distance.Update(distance_per_file.distance,
distance_per_file.frame)
value = LinearlyScaled(float(distance.distance), float(self._maximum))
if distance.frame is not None:
current_distance = int(distance.distance)
file_name = os.path.basename(distance.frame.file_path)
if current_distance == 0:
reason = [
'Suspected changelist touched the crashing line in ' + file_name]
else:
plural = current_distance > 1
reason = [
'Suspected changelist touched line(s) near the crashing line in '
'%s (%d line%s away)' % (file_name, current_distance,
's' if plural else '')]
else:
reason = None
if value <= _MINIMUM_FEATURE_VALUE:
changed_files = None
else:
changed_files = MinDistanceFeature.ChangedFiles(
suspect, touched_file_to_distance,
report.crashed_version)
return FeatureValue(name=self.name,
value=value,
reason=reason,
changed_files=changed_files)
def __call__(self, x): return lambda y: FeatureValue(self.name, y == (x < 9), None, None)
def __call__(self, x): return lambda y: FeatureValue(self.name, y == (x <= 7), 'reason2', None)
def __call__(self, x):
return lambda y: FeatureValue(self.name, y == ((x % 2) == 1), 'reason1',
[ChangedFile(name='b.cc',
blame_url=None,
reasons=['file_reason1'])])