def check_blocks(self, old_file, new_file):
h = open(old_file, "rb")
old = list(bgzf.BgzfBlocks(h))
h.close()
h = open(new_file, "rb")
new = list(bgzf.BgzfBlocks(h))
h.close()
self.assertEqual(len(old), len(new))
self.assertEqual(old, new)
def check_blocks(self, old_file, new_file):
with open(old_file, "rb") as h:
old = list(bgzf.BgzfBlocks(h))
with open(new_file, "rb") as h:
new = list(bgzf.BgzfBlocks(h))
self.assertEqual(len(old), len(new))
self.assertEqual(old, new)
def populate_blocks():
with open(self.bgzipped_fasta_filename,
'rb') as fasta_file_for_blocks:
# Store uncompressed data start, uncompressed data end, compressed block start as a tuple
blocks = [(b[2], b[2] + b[3], b[0])
for b in bgzf.BgzfBlocks(fasta_file_for_blocks)]
verbose_print('\t\tFound {} blocks in {}'.format(
len(blocks),
datetime.datetime.now() - start_time))
return blocks
def check_random(self, filename):
"""Check BGZF random access by reading blocks in forward & reverse order"""
h = gzip.open(filename, "rb")
old = h.read()
h.close()
h = open(filename, "rb")
blocks = list(bgzf.BgzfBlocks(h))
h.close()
#Forward
new = _empty_bytes_string
h = bgzf.BgzfReader(filename, "rb")
self.assertTrue(h.seekable())
self.assertFalse(h.isatty())
self.assertEqual(h.fileno(), h._handle.fileno())
for start, raw_len, data_start, data_len in blocks:
#print start, raw_len, data_start, data_len
h.seek(bgzf.make_virtual_offset(start,0))
data = h.read(data_len)
self.assertEqual(len(data), data_len)
#self.assertEqual(start + raw_len, h._handle.tell())
self.assertEqual(len(new), data_start)
new += data
h.close()
self.assertEqual(len(old), len(new))
self.assertEqual(old, new)
#Reverse
new = _empty_bytes_string
h = bgzf.BgzfReader(filename, "rb")
for start, raw_len, data_start, data_len in blocks[::-1]:
#print start, raw_len, data_start, data_len
h.seek(bgzf.make_virtual_offset(start,0))
data = h.read(data_len)
self.assertEqual(len(data), data_len)
#self.assertEqual(start + raw_len, h._handle.tell())
new = data + new
h.close()
self.assertEqual(len(old), len(new))
self.assertEqual(old, new)
#Jump back - non-sequential seeking
if len(blocks) >= 3:
h = bgzf.BgzfReader(filename, "rb", max_cache = 1)
#Seek to a late block in the file,
#half way into the third last block
start, raw_len, data_start, data_len = blocks[-3]
voffset = bgzf.make_virtual_offset(start, data_len // 2)
h.seek(voffset)
self.assertEqual(voffset, h.tell())
data = h.read(1000)
self.assertTrue(data in old)
self.assertEqual(old.find(data), data_start + data_len // 2)
#Now seek to an early block in the file,
#half way into the second block
start, raw_len, data_start, data_len = blocks[1]
h.seek(bgzf.make_virtual_offset(start, data_len // 2))
voffset = bgzf.make_virtual_offset(start, data_len // 2)
h.seek(voffset)
self.assertEqual(voffset, h.tell())
#Now read all rest of this block and start of next block
data = h.read(data_len + 1000)
self.assertTrue(data in old)
self.assertEqual(old.find(data), data_start + data_len // 2)
h.close()
#Check seek/tell at block boundaries
v_offsets = []
for start, raw_len, data_start, data_len in blocks:
for within_offset in [0, 1, data_len // 2, data_len - 1]:
if within_offset < 0 or data_len <= within_offset:
continue
voffset = bgzf.make_virtual_offset(start, within_offset)
real_offset = data_start + within_offset
v_offsets.append((voffset, real_offset))
shuffle(v_offsets)
h = bgzf.BgzfReader(filename, "rb", max_cache = 1)
for voffset, real_offset in v_offsets:
h.seek(0)
assert voffset >= 0 and real_offset >= 0
self.assertEqual(h.read(real_offset), old[:real_offset])
self.assertEqual(h.tell(), voffset)
for voffset, real_offset in v_offsets:
h.seek(voffset)
self.assertEqual(h.tell(), voffset)
h.close()
def test_BgzfBlocks_TypeError(self):
"""Check get expected TypeError from BgzfBlocks."""
for mode in ("r", "rb"):
decompressed = bgzf.open("GenBank/cor6_6.gb.bgz", mode)
with self.assertRaises(TypeError):
list(bgzf.BgzfBlocks(decompressed))
