def test_fasta__from_file(func, tmp_path):
expected = [
FASTQ("This_is_FASTA!", None, "ACGTN", "12345"),
FASTQ("This_is_ALSO_FASTA!", None, "CGTNA", "56789"),
]
with func(tmp_path / "file", "wt") as handle:
for item in expected:
item.write(handle)
assert list(FASTQ.from_file(tmp_path / "file")) == expected
def main(argv):
args = parse_args(argv)
seq_retained_nts = 0
seq_retained_reads = 0
for filename in args.files:
qualities = FASTQualities()
for record in FASTQ.from_file(filename):
qualities.update(record)
seq_retained_reads += 1
seq_retained_nts += len(record.sequence)
offsets = qualities.offsets()
if offsets == FASTQualities.BOTH:
print(
"FASTQ file(s) contains quality scores with both quality offsets (33 "
"and 64); file may be unexpected format or corrupt. Please ensure that "
"this file contains valid FASTQ reads from a single source.",
file=sys.stderr,
)
return 1
elif offsets == FASTQualities.MISSING:
if args.no_empty:
print("FASTQ file is empty.", file=sys.stderr)
return 1
elif offsets not in (FASTQualities.AMBIGIOUS, args.offset):
print(
"FASTQ file contains quality scores with wrong quality score offset "
"(%i); expected reads with quality score offset %i. Ensure that the "
"'QualityOffset' specified in the makefile corresponds to the input."
% (offsets, args.offset),
file=sys.stderr,
)
return 1
print(
json.dumps(
{
"filenames": args.files,
"seq_retained_reads": seq_retained_reads,
"seq_retained_nts": seq_retained_nts,
"seq_collapsed": seq_retained_reads if args.collapsed else 0,
},
indent=2,
sort_keys=True,
)
)
return 0