def get_bugs(self, date="today", bug_ids=[]):
self.query_url = ""
# Ignore already analyzed commits.
for commit in repository.get_commits():
pass
rev_start = f"children({commit['node']})"
commits = repository.download_commits(self.repo_dir,
rev_start,
ret=True)
commits = [
commit for commit in commits if not commit["ever_backedout"]
]
probs = self.model.classify(commits, True)
indexes = probs.argmax(axis=-1)
result = {}
for commit, prob, index in zip(commits, probs, indexes):
result[commit["node"]] = {
"id": commit["node"],
"summary": commit["desc"].split("\n", 1)[0],
"result": "Risky" if prob[1] > 0.5 else "Not risky",
"confidence": nice_round(prob[index]),
}
return result
def get_bugs(self, date="today", bug_ids=[]):
# Retrieve the bugs with the fields defined in get_bz_params
raw_bugs = super().get_bugs(date=date,
bug_ids=bug_ids,
chunk_size=7000)
if len(raw_bugs) == 0:
return {}
# Extract the bug ids
bug_ids = list(raw_bugs.keys())
# Classify those bugs
bugs = get_bug_ids_classification("regression", bug_ids)
results = {}
for bug_id in sorted(bugs.keys()):
bug_data = bugs[bug_id]
if not bug_data.get("available", True):
# The bug was not available, it was either removed or is a
# security bug
continue
if not {"prob"}.issubset(bug_data.keys()):
raise Exception(f"Invalid bug response {bug_id}: {bug_data!r}")
bug = raw_bugs[bug_id]
prob = bug_data["prob"]
if prob[1] < 0.5:
continue
bug_id = str(bug_id)
results[bug_id] = {
"id": bug_id,
"summary": bug["summary"],
"confidence": nice_round(prob[1]),
"autofixed": False,
}
# Only autofix results for which we are sure enough.
if prob[1] >= self.get_config("confidence_threshold"):
results[bug_id]["autofixed"] = True
self.autofix_regression.append((bug_id, prob[1]))
return results
def get_bugs(self, date="today", bug_ids=[]):
# Retrieve the bugs with the fields defined in get_bz_params
raw_bugs = super().get_bugs(date=date,
bug_ids=bug_ids,
chunk_size=7000)
if len(raw_bugs) == 0:
return {}
# Extract the bug ids
bug_ids = list(raw_bugs.keys())
# Classify those bugs
bugs = get_bug_ids_classification("stepstoreproduce", bug_ids)
results = {}
for bug_id in sorted(bugs.keys()):
bug_data = bugs[bug_id]
if not bug_data.get("available", True):
# The bug was not available, it was either removed or is a
# security bug
continue
if not {"prob", "index"}.issubset(bug_data.keys()):
raise Exception(f"Invalid bug response {bug_id}: {bug_data!r}")
bug = raw_bugs[bug_id]
prob = bug_data["prob"]
index = bug_data["index"]
results[bug_id] = {
"id": bug_id,
"summary": bug["summary"],
"has_str": "yes" if prob[1] > 0.5 else "no",
"confidence": nice_round(prob[index]),
"autofixed": False,
}
return results
def get_bugs(self, date="today", bug_ids=[]):
# Retrieve the bugs with the fields defined in get_bz_params
raw_bugs = super().get_bugs(date=date,
bug_ids=bug_ids,
chunk_size=7000)
if len(raw_bugs) == 0:
return {}
# Extract the bug ids
bug_ids = list(raw_bugs.keys())
# Classify those bugs
bugs = get_bug_ids_classification("spambug", bug_ids)
for bug_id in sorted(bugs.keys()):
bug_data = bugs[bug_id]
if not bug_data.get("available", True):
# The bug was not available, it was either removed or is a
# security bug
continue
if not {"prob", "index"}.issubset(bug_data.keys()):
raise Exception(f"Invalid bug response {bug_id}: {bug_data!r}")
bug = raw_bugs[bug_id]
prob = bug_data["prob"]
if prob[1] < self.get_config("confidence_threshold"):
continue
self.autofix_bugs[bug_id] = {
"id": bug_id,
"summary": bug["summary"],
"confidence": nice_round(prob[1]),
}
return self.autofix_bugs
def get_bugs(self, date="today", bug_ids=[]):
# Retrieve the bugs with the fields defined in get_bz_params
raw_bugs = super().get_bugs(date=date,
bug_ids=bug_ids,
chunk_size=7000)
if len(raw_bugs) == 0:
return {}
# Extract the bug ids
bug_ids = list(raw_bugs.keys())
# Classify those bugs
bugs = get_bug_ids_classification("defectenhancementtask", bug_ids)
results = {}
for bug_id in sorted(bugs.keys()):
bug_data = bugs[bug_id]
if not bug_data.get("available", True):
# The bug was not available, it was either removed or is a
# security bug
continue
if not {"prob", "index", "class", "extra_data"}.issubset(
bug_data.keys()):
raise Exception(f"Invalid bug response {bug_id}: {bug_data!r}")
bug = raw_bugs[bug_id]
prob = bug_data["prob"]
index = bug_data["index"]
suggestion = bug_data["class"]
labels_map = bug_data["extra_data"]["labels_map"]
assert suggestion in {
"defect",
"enhancement",
"task",
}, f"Suggestion {suggestion} is invalid"
if bug["type"] == suggestion:
continue
defect_prob = prob[labels_map["defect"]]
enhancement_prob = prob[labels_map["enhancement"]]
task_prob = prob[labels_map["task"]]
results[bug_id] = {
"id": bug_id,
"summary": bug["summary"],
"type": bug["type"],
"bugbug_type": suggestion,
"confidence": nice_round(prob[index]),
"confidences":
f"defect {nice_round(defect_prob)}, enhancement {nice_round(enhancement_prob)}, task {nice_round(task_prob)}",
"autofixed": False,
}
# Only autofix results for which we are sure enough.
# And only autofix defect -> task/enhancement for now, unless we're 100% sure.
"""if prob[index] == 1.0 or (
bug["type"] == "defect"
and (enhancement_prob + task_prob)
>= self.get_config("confidence_threshold")
):"""
if prob[index] == 1.0:
results[bug_id]["autofixed"] = True
self.autofix_type[bug["id"]] = suggestion
return results
def get_bugs(self, date="today", bug_ids=[]):
# Retrieve the bugs with the fields defined in get_bz_params
raw_bugs = super().get_bugs(date=date,
bug_ids=bug_ids,
chunk_size=7000)
if len(raw_bugs) == 0:
return {}
# Extract the bug ids
bug_ids = list(raw_bugs.keys())
# Classify those bugs
bugs = get_bug_ids_classification("component", bug_ids)
results = {}
for bug_id in sorted(bugs.keys()):
bug_data = bugs[bug_id]
if not bug_data.get("available", True):
# The bug was not available, it was either removed or is a
# security bug
continue
if not {"prob", "index", "class", "extra_data"}.issubset(
bug_data.keys()):
raise Exception(f"Invalid bug response {bug_id}: {bug_data!r}")
bug = raw_bugs[bug_id]
prob = bug_data["prob"]
index = bug_data["index"]
suggestion = bug_data["class"]
conflated_components_mapping = bug_data["extra_data"][
"conflated_components_mapping"]
# Skip product-only suggestions that are not useful.
if "::" not in suggestion and bug["product"] == suggestion:
continue
suggestion = conflated_components_mapping.get(
suggestion, suggestion)
if "::" not in suggestion:
logger.error(
f"There is something wrong with this component suggestion! {suggestion}"
)
continue
i = suggestion.index("::")
suggested_product = suggestion[:i]
suggested_component = suggestion[i + 2:]
# When moving bugs out of the 'General' component, we don't want to change the product (unless it is Firefox).
if bug["component"] == "General" and bug["product"] not in {
suggested_product,
"Firefox",
}:
continue
# Don't move bugs from Firefox::General to Core::Internationalization.
if (bug["product"] == "Firefox" and bug["component"] == "General"
and suggested_product == "Core"
and suggested_component == "Internationalization"):
continue
bug_id = str(bug["id"])
result = {
"id": bug_id,
"summary": bug["summary"],
"component": suggestion,
"confidence": nice_round(prob[index]),
"autofixed": False,
}
# In daily mode, we send an email with all results.
if self.frequency == "daily":
results[bug_id] = result
confidence_threshold_conf = ("confidence_threshold"
if bug["component"] != "General" else
"general_confidence_threshold")
if prob[index] >= self.get_config(confidence_threshold_conf):
self.autofix_component[bug_id] = {
"product": suggested_product,
"component": suggested_component,
}
result["autofixed"] = True
# In hourly mode, we send an email with only the bugs we acted upon.
if self.frequency == "hourly":
results[bug_id] = result
return results
