def test_longer_locus_line(self):
"""Check that we can read and write files with longer locus lines."""
# Create example file from existing file
with open(path.join("GenBank", "DS830848.gb"), 'r') as inhandle:
data = inhandle.readlines()
data[0] = "LOCUS AZZZAA021234567891234 2147483647 bp DNA linear PRI 15-OCT-2018\n"
# Create memory file from modified genbank file
in_tmp = StringIO()
in_tmp.writelines(data)
in_tmp.seek(0)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
in_tmp.seek(0)
record = SeqIO.read(in_tmp, 'genbank')
# Create temporary output memory file
out_tmp = StringIO()
SeqIO.write(record, out_tmp, 'genbank')
# Check that the written file can be read back in
out_tmp.seek(0)
record_in = SeqIO.read(out_tmp, 'genbank')
self.assertEqual(record_in.id, "DS830848.1")
self.assertEqual(record_in.name, "AZZZAA021234567891234")
self.assertEqual(len(record_in.seq), 2147483647)
def read_longer_than_maxsize():
with open(path.join("GenBank", "DS830848.gb"), 'r') as inhandle:
data2 = inhandle.readlines()
data2[0] = "LOCUS AZZZAA02123456789 " + str(sys.maxsize + 1) + " bp DNA linear PRI 15-OCT-2018\n"
long_in_tmp = StringIO()
long_in_tmp.writelines(data2)
long_in_tmp.seek(0)
record = SeqIO.read(long_in_tmp, 'genbank')
def test_extremely_long_sequence(self):
"""Tests if extremely long sequences can be read.
This is only run if sys.maxsize > 2147483647.
"""
# Create example file from existing file
with open(path.join("GenBank", "DS830848.gb"), 'r') as inhandle:
data = inhandle.readlines()
data[0] = "LOCUS AZZZAA02123456789 10000000000 bp DNA linear PRI 15-OCT-2018\n"
# Create memory file from modified genbank file
in_tmp = StringIO()
in_tmp.writelines(data)
in_tmp.seek(0)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
in_tmp.seek(0)
record = SeqIO.read(in_tmp, 'genbank')
# Create temporary output memory file
out_tmp = StringIO()
SeqIO.write(record, out_tmp, 'genbank')
# Check that the written file can be read back in
out_tmp.seek(0)
record_in = SeqIO.read(out_tmp, 'genbank')
self.assertEqual(record_in.id, "DS830848.1")
self.assertEqual(record_in.name, "AZZZAA02123456789")
self.assertEqual(len(record_in.seq), 10000000000)
def read_longer_than_maxsize():
with open(path.join("GenBank", "DS830848.gb"), 'r') as inhandle:
data2 = inhandle.readlines()
data2[0] = "LOCUS AZZZAA02123456789 " + str(sys.maxsize + 1) + " bp DNA linear PRI 15-OCT-2018\n"
long_in_tmp = StringIO()
long_in_tmp.writelines(data2)
long_in_tmp.seek(0)
record = SeqIO.read(long_in_tmp, 'genbank')
self.assertRaises(ValueError, read_longer_than_maxsize)