def test_write_lines(self):
with TempDirectory() as output_file:
writer = FileWriter(os.path.join(output_file.path, "A.tmp"))
writer.open()
writer.write("1\n2\n")
writer.write("3")
writer.close()
actual_file = open(os.path.join(output_file.path, "A.tmp"))
actual_output = actual_file.readlines()
actual_file.close()
self.assertEquals(["1\n", "2\n", "3"], actual_output)
def execute(args, execution_context):
validate_args(args)
output_dir = os.path.abspath(args.output)
unclaimed_readers, trans_vcf_readers = _claim_readers(args)
_log_unclaimed_readers(unclaimed_readers)
logger.info("Processing [{}] VCF file(s) from [{}]",
len(trans_vcf_readers), args.input)
new_tags = [
_ExcludeMalformedRef(),
_ExcludeMalformedAlt(),
_ExcludeMissingAlt()
]
for i, trans_vcf_reader in enumerate(trans_vcf_readers):
logger.info("Translating file {}/{} [{}]", i + 1,
len(trans_vcf_readers), trans_vcf_reader.file_name)
new_filename = _mangle_output_filename(trans_vcf_reader.file_name)
output_filepath = os.path.join(output_dir, new_filename)
file_writer = FileWriter(output_filepath)
_translate_files(
trans_vcf_reader,
new_tags,
execution_context,
file_writer,
)
logger.info("Wrote [{}] VCF file(s)", len(trans_vcf_readers))
def test_write_lines(self):
with TempDirectory() as output_file:
writer = FileWriter(os.path.join(output_file.path, "A.tmp"))
writer.open()
writer.write("1\n2\n")
writer.write("3")
writer.close()
actual_file = open(os.path.join(output_file.path, "A.tmp"))
actual_output = actual_file.readlines()
actual_file.close()
self.assertEquals(["1\n", "2\n", "3"], actual_output)
def test_hashable(self):
s = set([FileWriter("foo")])
s.add(FileWriter("foo"))
self.assertEquals(1, len(s))
def test_file_name(self):
writer = FileWriter("foo/bar/baz.tmp")
self.assertEquals("baz.tmp", writer.file_name)
def test_equality(self):
self.assertEquals(FileWriter("foo"), FileWriter("foo"))
self.assertNotEquals(FileWriter("foo"), FileWriter("bar"))
self.assertNotEquals(FileWriter("foo"), 1)
def test_write(self):
with TempDirectory() as output_file:
file_path = os.path.join(output_file.path, "test.tmp")
writer = FileWriter(file_path)
writer.open()
writer.write("A")
writer.write("B\n")
writer.write("CD\n")
writer.close()
actual_output = output_file.read('test.tmp', encoding='utf8')
expected_output = "AB|CD|".replace('|', os.linesep)
self.assertEquals(expected_output, actual_output)
def test_write(self):
with TempDirectory() as output_file:
file_path = os.path.join(output_file.path, "test.tmp")
writer = FileWriter(file_path)
writer.open()
writer.write("A")
writer.write("B\n")
writer.write("CD\n")
writer.close()
actual_output = output_file.read("test.tmp", encoding="utf8")
expected_output = "AB|CD|".replace("|", os.linesep)
self.assertEquals(expected_output, actual_output)
def _sort_vcf(reader, sorted_dir):
vcf_records = []
reader.open()
for vcf_record in reader.vcf_records():
vcf_records.append(vcf_record)
reader.close()
vcf_records.sort()
writer = FileWriter(os.path.join(sorted_dir,
reader.file_name))
writer.open()
writer.write("\n".join(reader.metaheaders) + "\n")
writer.write(reader.column_header + "\n")
for vcf_record in vcf_records:
writer.write(vcf_record.text())
writer.close()
reader = MergeVcfReader(vcf.FileReader(writer.output_filepath))
return reader
def _sort_vcf(reader, sorted_dir):
vcf_records = []
reader.open()
for vcf_record in reader.vcf_records():
vcf_records.append(vcf_record)
reader.close()
vcf_records.sort()
writer = FileWriter(os.path.join(sorted_dir, reader.file_name))
writer.open()
writer.write("\n".join(reader.metaheaders) + "\n")
writer.write(reader.column_header + "\n")
for vcf_record in vcf_records:
writer.write(vcf_record.text())
writer.close()
reader = MergeVcfReader(vcf.FileReader(writer.output_filepath))
return reader
