def _filter_column_list(column_spec_list, potential_col_list,
column_spec_filename):
actual_column_list = []
glossary_fields = OrderedDict()
for i, column_regex in enumerate(column_spec_list):
no_columns_found = True
for (column_name, vcf_tag) in potential_col_list.items():
column_exists = column_name in actual_column_list
regex_matches = re.match("^" + column_regex + "$", column_name)
if regex_matches and not column_exists:
glossary_fields[vcf_tag] = 1
actual_column_list.append(column_name)
no_columns_found = False
if no_columns_found:
logger.warning(UNUSED_REGEX_WARNING_FORMAT, column_regex,
column_spec_filename, i + 1)
if not actual_column_list:
raise utils.JQException(
"The selected_columns_file [{}] would "
"exclude all input columns. Review "
"inputs/usage and try again.", column_spec_filename)
glossary_fields = [x for x in glossary_fields.keys() if x]
return actual_column_list, glossary_fields
def _filter_column_list(column_spec_list,
potential_col_list,
column_spec_filename):
actual_column_list = []
glossary_fields = OrderedDict()
for i, column_regex in enumerate(column_spec_list):
no_columns_found = True
for (column_name, vcf_tag) in potential_col_list.items():
column_exists = column_name in actual_column_list
regex_matches = re.match("^" + column_regex + "$", column_name)
if regex_matches and not column_exists:
glossary_fields[vcf_tag] = 1
actual_column_list.append(column_name)
no_columns_found = False
if no_columns_found:
logger.warning(UNUSED_REGEX_WARNING_FORMAT,
column_regex,
column_spec_filename,
i + 1)
if not actual_column_list:
raise utils.JQException("The selected_columns_file [{}] would "
"exclude all input columns. Review "
"inputs/usage and try again.",
column_spec_filename)
glossary_fields = [x for x in glossary_fields.keys() if x]
return actual_column_list, glossary_fields
def test_warning(self):
tool = "foo"
args = Namespace(subparser_name=tool,
log_file=None,
verbose=None)
logger.initialize_logger(args)
self.assertFalse(logger.WARNING_OCCURRED)
logger.warning("bar")
self.assertTrue(logger.WARNING_OCCURRED)
root_logger = logger.logging.getLogger()
current_time = datetime.now().strftime('%Y-%m-%d')
output_lines = self.output.getvalue().rstrip().split("\n")
self.assertEquals(["root: WARNING: bar"], root_logger.handlers[0].buffer)
self.assertRegexpMatches(output_lines[0], ""+current_time+r".*\|WARNING\|foo\|bar")
def test_warning(self):
tool = "foo"
args = Namespace(subparser_name=tool, log_file=None, verbose=None)
logger.initialize_logger(args)
self.assertFalse(logger.WARNING_OCCURRED)
logger.warning("bar")
self.assertTrue(logger.WARNING_OCCURRED)
root_logger = logger.logging.getLogger()
current_time = datetime.now().strftime('%Y-%m-%d')
output_lines = self.output.getvalue().rstrip().split("\n")
self.assertEquals(["root: WARNING: bar"],
root_logger.handlers[0].buffer)
self.assertRegexpMatches(output_lines[0],
"" + current_time + r".*\|WARNING\|foo\|bar")
def _validate_consistent_samples(file_readers):
readers_per_patient = _get_readers_per_patient(file_readers)
all_callers = set()
for callers in readers_per_patient.values():
all_callers.update(callers)
warning = 0
for patient, callers in readers_per_patient.items():
missing_callers = set(all_callers).difference(set(callers))
if missing_callers:
warning = 1
msg = "Sample [{}] is missing VCF(s): {}"
logger.warning(msg, patient, sorted(list(missing_callers)))
if warning:
msg = "Some samples appear to be missing VCF(s)"
logger.warning(msg)
def _validate_consistent_samples(file_readers):
readers_per_patient = _get_readers_per_patient(file_readers)
all_callers = set()
for callers in readers_per_patient.values():
all_callers.update(callers)
warning = 0
for patient, callers in readers_per_patient.items():
missing_callers = set(all_callers).difference(set(callers))
if missing_callers:
warning = 1
msg = "Sample [{}] is missing VCF(s): {}"
logger.warning(msg,
patient,
sorted(list(missing_callers)))
if warning:
msg = "Some samples appear to be missing VCF(s)"
logger.warning(msg)
def _build_format_tags(format_tag_regex, vcf_readers):
retained_tags = set()
regexes_used = set()
for vcf_reader in vcf_readers:
for tag_regex in format_tag_regex:
for original_tag, new_tag in list(vcf_reader.format_tags.items()):
if re.match(tag_regex + "$", original_tag):
retained_tags.add(new_tag)
regexes_used.add(tag_regex)
if len(retained_tags) == 0:
msg = ("The specified format tag regex [{}] would exclude all format "
"tags. Review inputs/usage and try again")
raise utils.JQException(msg, format_tag_regex)
unused_regexes = set(format_tag_regex).difference(regexes_used)
if unused_regexes:
for unused_regex in unused_regexes:
msg = ("In the specified list of regexes {}, the regex [{}] does "
"not match any format tags; this expression may be "
"irrelevant.")
logger.warning(msg, format_tag_regex, unused_regex)
return sorted(list(retained_tags))
def _build_format_tags(format_tag_regex, vcf_readers):
retained_tags = set()
regexes_used = set()
for vcf_reader in vcf_readers:
for tag_regex in format_tag_regex:
for original_tag, new_tag in list(vcf_reader.format_tags.items()):
if re.match(tag_regex + "$", original_tag):
retained_tags.add(new_tag)
regexes_used.add(tag_regex)
if len(retained_tags) == 0:
msg = ("The specified format tag regex [{}] would exclude all format "
"tags. Review inputs/usage and try again")
raise utils.JQException(msg, format_tag_regex)
unused_regexes = set(format_tag_regex).difference(regexes_used)
if unused_regexes:
for unused_regex in unused_regexes:
msg = ("In the specified list of regexes {}, the regex [{}] does "
"not match any format tags; this expression may be "
"irrelevant.")
logger.warning(msg, format_tag_regex, unused_regex)
return sorted(list(retained_tags))
def _log_unclaimed_readers(unclaimed_readers):
unclaimed_log_message = "The input file [{}] will not be translated"
for reader in sorted(unclaimed_readers):
msg = unclaimed_log_message.format(reader.file_name)
logger.warning(msg)
def _log_unclaimed_readers(unclaimed_readers):
unclaimed_log_message = "The input file [{}] will not be translated"
for reader in sorted(unclaimed_readers):
msg = unclaimed_log_message.format(reader.file_name)
logger.warning(msg)
