def similarity_over_all(self):
"""Calculate similarity between bug (summary, description) over
all.
Returns:
dict, {bug_id -> [score, summary_score, description_score, stacktrace_score]}
"""
from ir_log import IRLog
from ir_config import IRConfig
from ir_mongodb_helper import IRCollection
from ir_text import IRText
from ir_tfidf import IRTFIDF
logger = IRLog.get_instance()
search_time_span = 2 * 3600 * 24 * 365
bug_id_name = IRConfig.get_instance().get('bug_id_name')
create_ts_name = IRConfig.get_instance().get('bug_create_ts_name')
product_name = IRConfig.get_instance().get('bug_product_name')
basic_collection = IRCollection('bug_db_name',
'bug_basic_collection_name', 'r')
reports2scan = basic_collection.find({
product_name: self.get_product(),
create_ts_name: {
'$gt': self.get_create_ts() - search_time_span
},
bug_id_name: {
'$nin': self.__exclude_report_ids
}
})
result = {}
IRLog.get_instance().println('Comparing with %d reports.' \
% (reports2scan.count()) )
print self.__summary_text
print self.__description_text
for report in reports2scan:
bug_id = report[bug_id_name]
if bug_id == self.get_dummy_bug_id():
continue
# because we don't want to load stacktrace in case of self.__stacktrace
# being none, we create and fill the info of report manually
other_report = IRReport("", "")
other_report.__summary_tfidf, other_report.__description_tfidf = \
IRTFIDF.get_tfidf_of_bug(bug_id)
# if self.__stacktrace is empty, we don't need to do this
if self.get_stacktrace() is not None and \
self.get_stacktrace().__len__() > 0:
other_report.__stacktrace = IRText.get_stacktrace_of_bug(
bug_id)
if other_report.__stacktrace is None:
other_report.__stacktrace = []
result[bug_id] = self.similarity_with(other_report)
return result
def __generate_single_bug(self, bug_id, drop_rate):
"""Generate an incomplete bug report text.
Args:
bug_id: int, original bug id.
drop_rate: float, 0.0 for not drop, 1.0 for totally drop.
Returns:
IRReport
"""
from ir_text import IRText
from ir_term_count import IRTermCount
from ir_report import IRReport
# get description and summary
summary, description = IRText.get_summary_and_description_of_bug(bug_id)
create_ts, product = IRText.get_basic_info_of_bug(bug_id)
if drop_rate > 0.001:
summary, description = \
IRTermCount.create_incomplete_report(summary, description, drop_rate)
print description
new_report = IRReport(summary, description)
new_report.set_stacktrace(IRText.get_stacktrace_of_bug(bug_id))
new_report.set_dummy_bug_id(bug_id)
new_report.set_basic_info(create_ts, product)
return new_report
def __generate_single_bug(self, bug_id, drop_rate):
"""Generate an incomplete bug report text.
Args:
bug_id: int, original bug id.
drop_rate: float, 0.0 for not drop, 1.0 for totally drop.
Returns:
IRReport
"""
from ir_text import IRText
from ir_term_count import IRTermCount
from ir_report import IRReport
# get description and summary
summary, description = IRText.get_summary_and_description_of_bug(
bug_id)
create_ts, product = IRText.get_basic_info_of_bug(bug_id)
if drop_rate > 0.001:
summary, description = \
IRTermCount.create_incomplete_report(summary, description, drop_rate)
print description
new_report = IRReport(summary, description)
new_report.set_stacktrace(IRText.get_stacktrace_of_bug(bug_id))
new_report.set_dummy_bug_id(bug_id)
new_report.set_basic_info(create_ts, product)
return new_report
def get_stacktrace(self):
if self.__bug_id is None:
return self.__stacktrace
else:
if self.__allow_cache and self.__stacktrace is not None:
return self.__stacktrace
from ir_text import IRText
stack = IRText.get_stacktrace_of_bug(self.__bug_id)
if self.__allow_cache:
self.__stacktrace = stack
return stack
def similarity_over_all(self):
"""Calculate similarity between bug (summary, description) over
all.
Returns:
dict, {bug_id -> [score, summary_score, description_score, stacktrace_score]}
"""
from ir_log import IRLog
from ir_config import IRConfig
from ir_mongodb_helper import IRCollection
from ir_text import IRText
from ir_tfidf import IRTFIDF
logger = IRLog.get_instance()
search_time_span = 2 * 3600 * 24 * 365
bug_id_name = IRConfig.get_instance().get('bug_id_name')
create_ts_name = IRConfig.get_instance().get('bug_create_ts_name')
product_name = IRConfig.get_instance().get('bug_product_name')
basic_collection = IRCollection(
'bug_db_name', 'bug_basic_collection_name', 'r')
reports2scan = basic_collection.find({
product_name : self.get_product(),
create_ts_name : {'$gt' : self.get_create_ts() - search_time_span},
bug_id_name : {'$nin' : self.__exclude_report_ids} })
result = {}
IRLog.get_instance().println('Comparing with %d reports.' \
% (reports2scan.count()) )
print self.__summary_text
print self.__description_text
for report in reports2scan:
bug_id = report[bug_id_name]
if bug_id == self.get_dummy_bug_id():
continue
# because we don't want to load stacktrace in case of self.__stacktrace
# being none, we create and fill the info of report manually
other_report = IRReport("", "")
other_report.__summary_tfidf, other_report.__description_tfidf = \
IRTFIDF.get_tfidf_of_bug(bug_id)
# if self.__stacktrace is empty, we don't need to do this
if self.get_stacktrace() is not None and \
self.get_stacktrace().__len__() > 0:
other_report.__stacktrace = IRText.get_stacktrace_of_bug(bug_id)
if other_report.__stacktrace is None:
other_report.__stacktrace = []
result[bug_id] = self.similarity_with(other_report)
return result
def get_stacktrace(self):
if self.__bug_id is None:
return self.__stacktrace
else:
if self.__allow_cache and self.__stacktrace is not None:
return self.__stacktrace
from ir_text import IRText
stack = IRText.get_stacktrace_of_bug(self.__bug_id)
if self.__allow_cache:
self.__stacktrace = stack
return stack