def test_read_invalid_input(self, L):
# cm: context_manager
# ref: http://stackoverflow.com/questions/15672151/is-it-possible-for-a-unit-test-to-assert-that-a-method-calls-sys-exit
# with LogCapture() as L:
with self.assertRaises(SystemExit) as cm:
utils.read(self.invalid_input)
self.assertEqual(cm.exception.code, 1)
L.check(
('rsempipeline.preprocess.utils', 'ERROR',
'Please correct the invalid entries in {0}'.format(self.invalid_input)),
('rsempipeline.preprocess.utils', 'ERROR',
'If unsure of the correct format, check {0}'.format(os.path.join(SHARE_DIR, 'GSE_GSM.example.csv'))),
)
def test_read_invalid_input(self, L):
# cm: context_manager
# ref: http://stackoverflow.com/questions/15672151/is-it-possible-for-a-unit-test-to-assert-that-a-method-calls-sys-exit
# with LogCapture() as L:
with self.assertRaises(SystemExit) as cm:
utils.read(self.invalid_input)
self.assertEqual(cm.exception.code, 1)
L.check(
('rsempipeline.preprocess.utils', 'ERROR',
'Please correct the invalid entries in {0}'.format(
self.invalid_input)),
('rsempipeline.preprocess.utils', 'ERROR',
'If unsure of the correct format, check {0}'.format(
os.path.join(SHARE_DIR, 'GSE_GSM.example.csv'))),
)
def download_soft(input_csv, soft_outdir): # pragma: no cover
ftp = SOFTDownloader()
current_gse = None
for gse, gsm in read(input_csv):
if current_gse is None or gse != current_gse:
ftp.gen_soft(gse, soft_outdir)
current_gse = gse
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_read_valid_input_with_commented_line(self):
self.assertIsInstance(utils.read(self.valid_input3),
types.GeneratorType)
def test_read_valid_input_with_appending_semicolon(self):
self.assertIsInstance(utils.read(self.valid_input2),
types.GeneratorType)
def test_read_valid_input(self):
self.assertIsInstance(utils.read(self.valid_input1),
types.GeneratorType)
def test_yield_gse_gsm_from_valid_input_with_commented_line(self):
for __ in utils.read(self.valid_input3):
self.check_yield_result(__)
def test_yield_gse_gsm_from_valid_input_with_appending_semicolon(self):
for __ in utils.read(self.valid_input2):
self.check_yield_result(__)
def test_read_valid_input_with_commented_line(self):
self.assertIsInstance(utils.read(self.valid_input3), types.GeneratorType)
def test_read_valid_input_with_appending_semicolon(self):
self.assertIsInstance(utils.read(self.valid_input2), types.GeneratorType)
def test_read_valid_input(self):
self.assertIsInstance(utils.read(self.valid_input1), types.GeneratorType)
def test_yield_gse_gsm_from_valid_input_with_commented_line(self):
for __ in utils.read(self.valid_input3):
self.check_yield_result(__)
def test_yield_gse_gsm_from_valid_input_with_appending_semicolon(self):
for __ in utils.read(self.valid_input2):
self.check_yield_result(__)
