def download_files(args):
"""Main function for the `taxadb download` sub-command.
This function can download taxump.tar.gz and the content of the
accession2taxid directory from the ncbi ftp.
Arguments:
args (object): The arguments from argparse
"""
logger = logging.getLogger(__name__)
# files to download
nucl_est = 'nucl_est.accession2taxid.gz'
nucl_gb = 'nucl_gb.accession2taxid.gz'
nucl_gss = 'nucl_gss.accession2taxid.gz'
nucl_wgs = 'nucl_wgs.accession2taxid.gz'
prot = 'prot.accession2taxid.gz'
taxdump = 'taxdump.tar.gz'
args.type = [x for y in args.type for x in y]
acc_dl_list = [taxdump]
for div in args.type:
if div in ['full', 'nucl', 'est']:
acc_dl_list.append(nucl_est)
if div in ['full', 'nucl', 'gb']:
acc_dl_list.append(nucl_gb)
if div in ['full', 'nucl', 'gss']:
acc_dl_list.append(nucl_gss)
if div in ['full', 'nucl', 'wgs']:
acc_dl_list.append(nucl_wgs)
if div in ['full', 'prot']:
acc_dl_list.append(prot)
try:
out = args.outdir
os.makedirs(os.path.abspath(out), exist_ok=args.force)
os.chdir(os.path.abspath(out))
except FileExistsError as e:
logger.error('%s exists. Consider using -f if you want to overwrite' %
out)
sys.exit(1)
for file in acc_dl_list:
if file != taxdump:
download.ncbi('pub/taxonomy/accession2taxid/', file)
download.ncbi('pub/taxonomy/accession2taxid/', file + '.md5')
util.md5_check(file)
else:
download.ncbi('pub/taxonomy/', taxdump)
download.ncbi('pub/taxonomy/', taxdump + '.md5')
util.md5_check(taxdump)
download.unpack(taxdump)
def download_files(args):
"""Main function for the `taxadb download` sub-command.
This function can download taxump.tar.gz and the content of the
accession2taxid directory from the ncbi ftp.
Arguments:
args (object): The arguments from argparse
"""
logger = logging.getLogger(__name__)
# files to download
# nucl_est = 'nucl_est.accession2taxid.gz' # deprecated
nucl_gb = 'nucl_gb.accession2taxid.gz'
# nucl_gss = 'nucl_gss.accession2taxid.gz' # deprecated
nucl_wgs = 'nucl_wgs.accession2taxid.gz'
prot = 'prot.accession2taxid.gz'
taxdump = 'taxdump.tar.gz'
args.type = [x for y in args.type for x in y]
acc_dl_list = [taxdump]
for div in args.type:
if div in ['full', 'nucl', 'gb']:
acc_dl_list.append(nucl_gb)
if div in ['full', 'nucl', 'wgs']:
acc_dl_list.append(nucl_wgs)
if div in ['full', 'prot']:
acc_dl_list.append(prot)
try:
out = args.outdir
os.makedirs(os.path.abspath(out), exist_ok=args.force)
os.chdir(os.path.abspath(out))
except FileExistsError as e:
logger.error('%s exists. Consider using -f if you want to overwrite'
% out)
sys.exit(1)
for file in acc_dl_list:
if file != taxdump:
download.ncbi('pub/taxonomy/accession2taxid/', file)
download.ncbi('pub/taxonomy/accession2taxid/', file + '.md5')
util.md5_check(file)
else:
download.ncbi('pub/taxonomy/', taxdump)
download.ncbi('pub/taxonomy/', taxdump + '.md5')
util.md5_check(taxdump)
download.unpack(taxdump)
def download(args):
"""Main function for the 'taxadb download' sub-command.
This function downloads taxump.tar.gz and the content of the
accession2taxid directory from the ncbi ftp.
Arguments:
args.output (:obj:`str`): output directory
"""
ncbi_ftp = 'ftp.ncbi.nlm.nih.gov'
# files to download in accession2taxid
nucl_est = 'nucl_est.accession2taxid.gz'
nucl_gb = 'nucl_gb.accession2taxid.gz'
nucl_gss = 'nucl_gss.accession2taxid.gz'
nucl_wgs = 'nucl_wgs.accession2taxid.gz'
prot = 'prot.accession2taxid.gz'
acc_dl_list = [nucl_est, nucl_gb, nucl_gss, nucl_wgs, prot]
taxdump = 'taxdump.tar.gz'
out = args.outdir
os.makedirs(os.path.abspath(out), exist_ok=True)
os.chdir(os.path.abspath(out))
for file in acc_dl_list:
print('Started Downloading %s' % file)
with ftputil.FTPHost(ncbi_ftp, 'anonymous', 'password') as ncbi:
ncbi.chdir('pub/taxonomy/accession2taxid/')
ncbi.download_if_newer(file, file)
ncbi.download_if_newer(file + '.md5', file + '.md5')
util.md5_check(file)
print('Started Downloading %s' % taxdump)
with ftputil.FTPHost(ncbi_ftp, 'anonymous', 'password') as ncbi:
ncbi.chdir('pub/taxonomy/')
ncbi.download_if_newer(taxdump, taxdump)
ncbi.download_if_newer(taxdump + '.md5', taxdump + '.md5')
util.md5_check(taxdump)
print('Unpacking %s' % taxdump)
with tarfile.open(taxdump, "r:gz") as tar:
tar.extractall()
tar.close()
def test_md5check_fails(self):
"""Check md5 fails"""
badfile = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'wrong.txt')
with self.assertRaises(AssertionError):
md5_check(badfile)
def test_md5check_success(self):
"""Check md5 is ok"""
okfile = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'good.txt')
self.assertIsNone(md5_check(okfile))
def test_md5check_fails(self):
"""Check md5 fails"""
badfile = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'wrong.txt')
with self.assertRaises(SystemExit):
md5_check(badfile)
def test_md5check_success(self):
"""Check md5 is ok"""
okfile = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'good.txt')
self.assertIsNone(md5_check(okfile))