def test_cache_all_data(self):
from ir_log import IRLog
from ir_config import IRConfig
from ir_tfidf import IRTFIDF
IRLog.get_instance().start_log()
IRConfig.get_instance().load('../data/test/bug_test.cfg')
IRTFIDF.cache_all_data()
IRLog.get_instance().stop_log()
def server_cache(msg, res):
from ir_log import IRLog
from ir_text import IRText
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
IRLog.get_instance().println('Server is caching data')
IRText.cache_all_data()
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRLog.get_instance().println('Server cached data')
return SIGNAL_CONTINUE
def server_cache(msg, res):
from ir_log import IRLog
from ir_text import IRText
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
IRLog.get_instance().println('Server is caching data')
IRText.cache_all_data()
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRLog.get_instance().println('Server cached data')
return SIGNAL_CONTINUE
def start_shell(cls):
"""Start a shell that do recommending interactively"""
from ir_log import IRLog
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
from ir_report import IRReport
IRLog.get_instance().println("Starting Intereport...")
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRLog.get_instance().println("Intereport Started. Waiting for input")
new_report = None
while 1:
cmd = raw_input("Input command:").strip()
if cmd == 'exit':
IRLog.get_instance().println('Exiting')
break
elif cmd == 'new':
IRLog.get_instance().println('Creating New Report')
import time
cur_time = -1
while cur_time < 0:
try:
cur_time = int(time.mktime(time.strptime(
raw_input("Input Time (e.g., 2011-05-05): "),
'%Y-%m-%d')))
except:
cur_time = -1
product = raw_input("Input Product: ")
summary = raw_input("Summary: ")
raw_description = raw_input("Description:\n")
new_report = IRReport.from_string(IRReport.separator.join([
str(cur_time), product.lower(), summary, raw_description,
'', '']))
cls.__print_report(new_report)
elif cmd == 'do':
IRLog.get_instance().println('Do Recommending')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
cls.do_recommend(new_report)
elif cmd == 'ls':
IRLog.get_instance().println('Show Current Report')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
cls.__print_report(new_report)
elif cmd == 'ad':
IRLog.get_instance().println('Appending Description')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
append_description = raw_input("Append Description:\n")
description =' '.join([new_report.get_description_text(),
append_description])
dummy_report = IRReport(new_report.get_summary_text(),
description)
dummy_report.set_stacktrace(new_report.get_stacktrace())
dummy_report.set_basic_info(new_report.get_create_ts(),
new_report.get_product())
dummy_report.set_penalty_terms(new_report.get_penalty_terms())
dummy_report.set_dummy_bug_id(new_report.get_dummy_bug_id())
new_report = dummy_report
IRLog.get_instance().println('Description: %s' % description)
elif cmd == 'ap':
IRLog.get_instance().println('Appending Penalties')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
raw = []
while raw.__len__() < 1:
raw = raw_input('Input Penalties (split by \',\'):').split(',')
from ir_term_count import IRTermCount
penalty = new_report.get_penalty_terms()
if penalty is None:
penalty = []
penalty += IRTermCount.do_stemming(raw)
new_report.set_penalty_terms(penalty)
print len(penalty), penalty
IRLog.get_instance().println('Penalties: %s' % \
(', '.join(penalty)))
elif cmd == 'sd':
IRLog.get_instance().println('Set Dummy Bug ID')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
bug_id = -1
while bug_id <= 0:
try:
bug_id = int(raw_input('Dummy Bug ID: '))
except:
bug_id = -1
new_report.set_dummy_bug_id(bug_id)
IRLog.get_instance().println('Dummy Bug ID: %d' % bug_id)
elif cmd == 'help':
cls.__show_help()
else:
IRLog.get_instance().println('Error! Unkown command: %s' \
% cmd)
cls.__show_help()
# end of while 1
IRLog.get_instance().println("Bye")
if mode == 'file':
test_file = sys.argv[3]
bug_id = int(sys.argv[4])
from ir_sim_bug_evaluator import IRSimBugEvaluator
new_report = IRSimBugEvaluator.get_report_from_test_file(test_file, bug_id)
if new_report is None:
IRLog.get_instance().println('Error! Cannot find report %d in %s' % \
(bug_id, test_file))
else:
if sys.argv.__len__() > 5:
from ir_term_count import IRTermCount
penalty_terms_raw = sys.argv[4].split(',')
penalty_terms = set(IRTermCount.do_stemming(penalty_terms_raw))
IRLog.get_instance().println('%d penalty terms: %s:' \
% (penalty_terms.__len__(), ','.join(penalty_terms)))
new_report.set_penalty_terms(penalty_terms)
elif mode == 'text':
text = sys.argv[3]
new_report = IRReport.from_string(text)
elif mode == 'inte':
IRRecommender.start_shell()
exit()
else:
IRLog.get_instance().println('Error! Known mode %s' % mode)
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRRecommender.do_recommend(new_report)
IRLog.get_instance().stop_log()
def start_shell(cls):
"""Start a shell that do recommending interactively"""
from ir_log import IRLog
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
from ir_report import IRReport
IRLog.get_instance().println("Starting Intereport...")
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRLog.get_instance().println("Intereport Started. Waiting for input")
new_report = None
while 1:
cmd = raw_input("Input command:").strip()
if cmd == 'exit':
IRLog.get_instance().println('Exiting')
break
elif cmd == 'new':
IRLog.get_instance().println('Creating New Report')
import time
cur_time = -1
while cur_time < 0:
try:
cur_time = int(
time.mktime(
time.strptime(
raw_input(
"Input Time (e.g., 2011-05-05): "),
'%Y-%m-%d')))
except:
cur_time = -1
product = raw_input("Input Product: ")
summary = raw_input("Summary: ")
raw_description = raw_input("Description:\n")
new_report = IRReport.from_string(
IRReport.separator.join([
str(cur_time),
product.lower(), summary, raw_description, '', ''
]))
cls.__print_report(new_report)
elif cmd == 'do':
IRLog.get_instance().println('Do Recommending')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
cls.do_recommend(new_report)
elif cmd == 'ls':
IRLog.get_instance().println('Show Current Report')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
cls.__print_report(new_report)
elif cmd == 'ad':
IRLog.get_instance().println('Appending Description')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
append_description = raw_input("Append Description:\n")
description = ' '.join([
new_report.get_description_text(), append_description
])
dummy_report = IRReport(new_report.get_summary_text(),
description)
dummy_report.set_stacktrace(new_report.get_stacktrace())
dummy_report.set_basic_info(new_report.get_create_ts(),
new_report.get_product())
dummy_report.set_penalty_terms(
new_report.get_penalty_terms())
dummy_report.set_dummy_bug_id(
new_report.get_dummy_bug_id())
new_report = dummy_report
IRLog.get_instance().println('Description: %s' %
description)
elif cmd == 'ap':
IRLog.get_instance().println('Appending Penalties')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
raw = []
while raw.__len__() < 1:
raw = raw_input(
'Input Penalties (split by \',\'):').split(',')
from ir_term_count import IRTermCount
penalty = new_report.get_penalty_terms()
if penalty is None:
penalty = []
penalty += IRTermCount.do_stemming(raw)
new_report.set_penalty_terms(penalty)
print len(penalty), penalty
IRLog.get_instance().println('Penalties: %s' % \
(', '.join(penalty)))
elif cmd == 'sd':
IRLog.get_instance().println('Set Dummy Bug ID')
if new_report is None:
IRLog.get_instance().println('Error! Please create '
'report first.')
else:
bug_id = -1
while bug_id <= 0:
try:
bug_id = int(raw_input('Dummy Bug ID: '))
except:
bug_id = -1
new_report.set_dummy_bug_id(bug_id)
IRLog.get_instance().println('Dummy Bug ID: %d' % bug_id)
elif cmd == 'help':
cls.__show_help()
else:
IRLog.get_instance().println('Error! Unkown command: %s' \
% cmd)
cls.__show_help()
# end of while 1
IRLog.get_instance().println("Bye")
test_file = sys.argv[3]
bug_id = int(sys.argv[4])
from ir_sim_bug_evaluator import IRSimBugEvaluator
new_report = IRSimBugEvaluator.get_report_from_test_file(
test_file, bug_id)
if new_report is None:
IRLog.get_instance().println('Error! Cannot find report %d in %s' % \
(bug_id, test_file))
else:
if sys.argv.__len__() > 5:
from ir_term_count import IRTermCount
penalty_terms_raw = sys.argv[4].split(',')
penalty_terms = set(IRTermCount.do_stemming(penalty_terms_raw))
IRLog.get_instance().println('%d penalty terms: %s:' \
% (penalty_terms.__len__(), ','.join(penalty_terms)))
new_report.set_penalty_terms(penalty_terms)
elif mode == 'text':
text = sys.argv[3]
new_report = IRReport.from_string(text)
elif mode == 'inte':
IRRecommender.start_shell()
exit()
else:
IRLog.get_instance().println('Error! Known mode %s' % mode)
from ir_tfidf import IRTFIDF
from ir_document_count import IRDocumentCount
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
IRRecommender.do_recommend(new_report)
IRLog.get_instance().stop_log()
def do_test_over_file(self, filename):
"""Do test over the file.
Args:
filename: str, the input file which generated by
generate_incomplete_test_file.
"""
from ir_log import IRLog
from ir_config import IRConfig
from ir_duplicate_group import IRDuplicateGroup
from ir_text import IRText
from ir_term_count import IRTermCount
from ir_tfidf import IRTFIDF
from ir_report import IRReport
from ir_document_count import IRDocumentCount
IRText.cache_all_data()
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
remove_self_bug_id = IRConfig.get_instance().get_bool('remove_self_bug_id', True)
sim_tot_precision = 0.0
sim_tot_recall = 0.0
sim_bi_tot_recall = 0.0
sim_tot_size = 0
dup_tot_precision = 0.0
dup_tot_recall = 0.0
dup_bi_toto_recall = 0.0
dup_num = 0
test_num = 0
infile = open(filename, 'r')
for line in infile:
IRLog.get_instance().println('----test----')
test_num += 1
line.strip()
new_report = IRReport.from_string(line)
ori_report = IRReport(new_report.get_dummy_bug_id())
#IRLog.get_instance().println('Summary')
#IRTermCount.show_dict_compare(ori_report.get_summary_termcount(),
# new_report.get_summary_termcount())
#IRLog.get_instance().println('Description')
#IRTermCount.show_dict_compare(ori_report.get_description_termcount(),
# new_report.get_description_termcount())
# do test for single
similarities, duplicates = new_report.similarities_and_duplicates()
sim_ids = [sim[0] for sim in similarities]
dup_ids = [dup[0] for dup in duplicates]
IRLog.get_instance().println('Sim ids: %s' % str(sim_ids))
IRLog.get_instance().println('Dup ids: %s' % str(dup_ids))
# evaluate sim
sim_hit, sim_nothit, real_duplicates = \
IRDuplicateGroup.is_in_same_duplicate_group(
new_report.get_dummy_bug_id(), sim_ids, remove_self_bug_id)
# some group contain only one
if real_duplicates.__len__() == 0:
test_num -= 1
continue
precision, recall = self.__report_result(
new_report.get_dummy_bug_id(), sim_hit, sim_nothit, real_duplicates)
sim_tot_precision += precision
sim_tot_recall += recall
sim_tot_size += sim_ids.__len__()
sim_bi_tot_recall += 1 if recall > 0.0 else 0
if dup_ids.__len__() > 0:
dup_num += 1
dup_hit, dup_nothit, real_duplicates = \
IRDuplicateGroup.is_in_same_duplicate_group(
new_report.get_dummy_bug_id(), dup_ids, remove_self_bug_id)
precision, recall = self.__report_result(
new_report.get_dummy_bug_id(), dup_hit, dup_nothit, real_duplicates)
dup_tot_precision += precision
dup_tot_recall += recall
dup_bi_toto_recall += 1 if recall > 0.0 else 0
# general conclusion
if dup_num == 0:
dup_num = 1.0
IRLog.get_instance().println(','.join(['#cases', 'sim pre', 'sim rec', 'sim birec', 'sim size',\
'#dup', 'dup pre', 'dup rec', 'dup birec']))
IRLog.get_instance().println(','.join([str(test_num), \
str(sim_tot_precision/test_num), str(sim_tot_recall/test_num), str(sim_bi_tot_recall/test_num), str(float(sim_tot_size)/test_num), \
str(dup_num), \
str(dup_tot_precision/dup_num), str(dup_tot_recall/dup_num), str(dup_bi_toto_recall/dup_num)]))
infile.close()
def do_test_over_file(self, filename):
"""Do test over the file.
Args:
filename: str, the input file which generated by
generate_incomplete_test_file.
"""
from ir_log import IRLog
from ir_config import IRConfig
from ir_duplicate_group import IRDuplicateGroup
from ir_text import IRText
from ir_term_count import IRTermCount
from ir_tfidf import IRTFIDF
from ir_report import IRReport
from ir_document_count import IRDocumentCount
IRText.cache_all_data()
IRTFIDF.cache_all_data()
IRDocumentCount.cache_all_data()
remove_self_bug_id = IRConfig.get_instance().get_bool(
'remove_self_bug_id', True)
sim_tot_precision = 0.0
sim_tot_recall = 0.0
sim_bi_tot_recall = 0.0
sim_tot_size = 0
dup_tot_precision = 0.0
dup_tot_recall = 0.0
dup_bi_toto_recall = 0.0
dup_num = 0
test_num = 0
infile = open(filename, 'r')
for line in infile:
IRLog.get_instance().println('----test----')
test_num += 1
line.strip()
new_report = IRReport.from_string(line)
ori_report = IRReport(new_report.get_dummy_bug_id())
#IRLog.get_instance().println('Summary')
#IRTermCount.show_dict_compare(ori_report.get_summary_termcount(),
# new_report.get_summary_termcount())
#IRLog.get_instance().println('Description')
#IRTermCount.show_dict_compare(ori_report.get_description_termcount(),
# new_report.get_description_termcount())
# do test for single
similarities, duplicates = new_report.similarities_and_duplicates()
sim_ids = [sim[0] for sim in similarities]
dup_ids = [dup[0] for dup in duplicates]
IRLog.get_instance().println('Sim ids: %s' % str(sim_ids))
IRLog.get_instance().println('Dup ids: %s' % str(dup_ids))
# evaluate sim
sim_hit, sim_nothit, real_duplicates = \
IRDuplicateGroup.is_in_same_duplicate_group(
new_report.get_dummy_bug_id(), sim_ids, remove_self_bug_id)
# some group contain only one
if real_duplicates.__len__() == 0:
test_num -= 1
continue
precision, recall = self.__report_result(
new_report.get_dummy_bug_id(), sim_hit, sim_nothit,
real_duplicates)
sim_tot_precision += precision
sim_tot_recall += recall
sim_tot_size += sim_ids.__len__()
sim_bi_tot_recall += 1 if recall > 0.0 else 0
if dup_ids.__len__() > 0:
dup_num += 1
dup_hit, dup_nothit, real_duplicates = \
IRDuplicateGroup.is_in_same_duplicate_group(
new_report.get_dummy_bug_id(), dup_ids, remove_self_bug_id)
precision, recall = self.__report_result(
new_report.get_dummy_bug_id(), dup_hit, dup_nothit,
real_duplicates)
dup_tot_precision += precision
dup_tot_recall += recall
dup_bi_toto_recall += 1 if recall > 0.0 else 0
# general conclusion
if dup_num == 0:
dup_num = 1.0
IRLog.get_instance().println(','.join(['#cases', 'sim pre', 'sim rec', 'sim birec', 'sim size',\
'#dup', 'dup pre', 'dup rec', 'dup birec']))
IRLog.get_instance().println(','.join([str(test_num), \
str(sim_tot_precision/test_num), str(sim_tot_recall/test_num), str(sim_bi_tot_recall/test_num), str(float(sim_tot_size)/test_num), \
str(dup_num), \
str(dup_tot_precision/dup_num), str(dup_tot_recall/dup_num), str(dup_bi_toto_recall/dup_num)]))
infile.close()