def test_empty_encoding_writes_nothing(self):
encoding = []
ostream = BytesIO()
PrgEncoder.write(encoding, ostream)
ostream.seek(0)
self.assertEqual(ostream.read(), b"")
def write_prg(output_prefix: str, prg_string: str):
"""
Writes the prg to outfile.
Writes it as a human readable string, and also as an integer vector
"""
prg_filename = Path(output_prefix + ".prg")
with prg_filename.open("w") as prg:
regex = re.compile(
r"^(?P<sample>.+)\.max_nest(?P<max_nest>\d+)\.min_match(?P<min_match>\d+)"
)
match = regex.search(prg_filename.stem)
try:
sample = match.group("sample")
except IndexError:
logging.warning(
"A sample name couldn't be parsed from the prefix. "
"Using 'sample' as sample name."
)
sample = "sample"
max_nest = int(match.group("max_nest"))
min_match = int(match.group("min_match"))
header = f"{sample} max_nest={max_nest} min_match={min_match}"
print(f">{header}\n{prg_string}", file=prg)
prg_ints_fpath = Path(output_prefix + ".bin")
prg_encoder = PrgEncoder()
prg_ints: PRG_Ints = prg_encoder.encode(prg_string)
with prg_ints_fpath.open("wb") as ostream:
prg_encoder.write(prg_ints, ostream)
def test_write_single_int(self):
prg_ints = [1]
write_to = BytesIO()
PrgEncoder.write(prg_ints, write_to)
write_to.seek(0)
actual = write_to.read()
expected = to_bytes(1)
self.assertEqual(actual, expected)
def test_write_two_ints(self):
encoding = [1, 4]
write_to = BytesIO()
PrgEncoder.write(encoding, write_to)
write_to.seek(0)
actual = write_to.read()
expected = to_bytes(1) + to_bytes(4)
self.assertEqual(actual, expected)
def build_from_msas(report, action, build_paths, args):
"""
For use by gramtools
"""
log.info(f"Building prg from prgs in {args.prgs_bed}")
built_intervals, rescaled_prg_ints = standalone_build_from_msas(
args.prgs_bed,
args.reference,
build_paths.coords_file,
build_paths.built_prg_dirname,
args.max_threads,
)
built_intervals.saveas(build_paths.built_prg_bed)
with open(build_paths.prg, "wb") as fhandle_out:
PrgEncoder.write(rescaled_prg_ints, fhandle_out)
def test_write_multiple_ints_function_calls(self):
num_elems = 100
prg_ints = [random.randint for _ in range(num_elems)]
ostream = BytesIO()
ostream.write = MagicMock(spec=True)
# Check call to write triggers the production of bytes
with patch("make_prg.prg_encoder.to_bytes",
return_value=b"") as mocked_to_bytes:
PrgEncoder.write(prg_ints, ostream)
expected_bytes_calls = [call(i) for i in prg_ints]
self.assertEqual(expected_bytes_calls,
mocked_to_bytes.call_args_list)
# Check ostream write operation only called once
ostream.write.assert_called_once()
def write_prg(prg_fname: Path, prg_string: str, options: ArgumentParser):
"""
Writes th prg to `output_file`.
Writes it as a human readable string, and also as an integer vector
"""
seqid = options.seqid or options.prg_name
if options.output_type.prg:
with prg_fname.open("w") as prg:
header = f">{seqid} max_nest={options.max_nesting} min_match={options.min_match_length}"
print(f"{header}\n{prg_string}", file=prg)
if options.output_type.binary:
prg_ints_fpath = prg_fname.with_suffix(".bin")
prg_encoder = PrgEncoder()
prg_ints: PRG_Ints = prg_encoder.encode(prg_string)
with prg_ints_fpath.open("wb") as ostream:
prg_encoder.write(prg_ints, ostream)
def encode_and_write_prg(self, prg_string: str):
prg_encoder = PrgEncoder()
prg_ints = prg_encoder.encode(prg_string)
with open(self.out_fname, "wb") as fhandle_out:
prg_encoder.write(prg_ints, fhandle_out)
