def test_backup_file_with_previously_backuped_file(self, mock_exists,
mock_rename):
mock_exists.side_effect = [True, True, False]
with LogCapture() as L:
misc.backup_file('path/file')
L.check(('rsempipeline.utils.misc', 'INFO',
'Backing up path/file to path/#file.2#'), )
mock_rename.called_once_with('path/file', 'path/#file.2#')
def test_backup_file(self, mock_exists, mock_rename):
# Need to mock multiple return values for mock_exists because the file
# to backup has to exist, but the backuped file with new name cannnot
# exist
# http://stackoverflow.com/questions/24897145/python-mock-multiple-return-values
mock_exists.side_effect = [True, False]
with LogCapture() as L:
misc.backup_file('path/file')
L.check(('rsempipeline.utils.misc', 'INFO', 'Backing up path/file to path/#file.1#'),)
mock_rename.called_once_with('path/file', 'path/#file.1#')
def test_backup_file(self, mock_exists, mock_rename):
# Need to mock multiple return values for mock_exists because the file
# to backup has to exist, but the backuped file with new name cannnot
# exist
# http://stackoverflow.com/questions/24897145/python-mock-multiple-return-values
mock_exists.side_effect = [True, False]
with LogCapture() as L:
misc.backup_file('path/file')
L.check(('rsempipeline.utils.misc', 'INFO',
'Backing up path/file to path/#file.1#'), )
mock_rename.called_once_with('path/file', 'path/#file.1#')
def generate_csv(input_csv, outdir, num_threads):
# Sometimes GSM data could be private, so no species information will be
# extracted. e.g. GSE49366 GSM1198168
res, res_no_species = [], []
# execute in parallel
queue = Queue.Queue()
def worker():
while True:
GSE, GSM = queue.get()
species = find_species(GSE, GSM, outdir)
row = [GSE, species, GSM]
if species:
res.append(row)
else:
res_no_species.append(row)
queue.task_done()
for i in range(num_threads):
thrd = threading.Thread(target=worker)
thrd.daemon = True
thrd.start()
for gse, gsm in read(input_csv):
queue.put([gse, gsm])
queue.join()
# write output
out_csv = os.path.join(outdir, SPECIES_CSV_BASENAME)
no_species_csv = os.path.join(outdir, NO_SPECIES_CSV_BASENAME)
backup_file(out_csv)
write_csv(res, out_csv)
if res_no_species:
backup_file(no_species_csv)
write_csv(res_no_species, no_species_csv)
def generate_csv(input_csv, outdir, num_threads):
# Sometimes GSM data could be private, so no species information will be
# extracted. e.g. GSE49366 GSM1198168
res, res_no_species = [], []
# execute in parallel
queue = Queue.Queue()
def worker():
while True:
GSE, GSM = queue.get()
species = find_species(GSE, GSM, outdir)
row = [GSE, species, GSM]
if species:
res.append(row)
else:
res_no_species.append(row)
queue.task_done()
for i in range(num_threads):
thrd = threading.Thread(target=worker)
thrd.daemon = True
thrd.start()
for gse, gsm in read(input_csv):
queue.put([gse, gsm])
queue.join()
# write output
out_csv = os.path.join(outdir, SPECIES_CSV_BASENAME)
no_species_csv = os.path.join(outdir, NO_SPECIES_CSV_BASENAME)
backup_file(out_csv)
write_csv(res, out_csv)
if res_no_species:
backup_file(no_species_csv)
write_csv(res_no_species, no_species_csv)
def test_backup_file_with_nonexistent_file(self, mock_os):
mock_os.path.exists.return_value = False
with LogCapture() as L:
misc.backup_file('some_nonexistent_file')
L.check(('rsempipeline.utils.misc', 'WARNING',
'some_nonexistent_file doesn\'t exist'), )
def test_backup_file_with_nonexistent_file(self, mock_os):
mock_os.path.exists.return_value = False
with LogCapture() as L:
misc.backup_file('some_nonexistent_file')
L.check(('rsempipeline.utils.misc', 'WARNING', 'some_nonexistent_file doesn\'t exist'),)
def test_backup_file_with_previously_backuped_file(self, mock_exists, mock_rename):
mock_exists.side_effect = [True, True, False]
with LogCapture() as L:
misc.backup_file('path/file')
L.check(('rsempipeline.utils.misc', 'INFO', 'Backing up path/file to path/#file.2#'),)
mock_rename.called_once_with('path/file', 'path/#file.2#')
