def get_tsv_errors(tsv_path, schema_name, optional_fields=[], offline=None, encoding=None):
'''
Validate the TSV.
'''
logging.info(f'Validating {schema_name} TSV...')
if not Path(tsv_path).exists():
return 'File does not exist'
try:
rows = dict_reader_wrapper(tsv_path, encoding=encoding)
except IsADirectoryError:
return 'Expected a TSV, but found a directory'
except UnicodeDecodeError as e:
return get_context_of_decode_error(e)
if not rows:
return 'File has no data rows.'
version = rows[0]['version'] if 'version' in rows[0] else '0'
try:
others = [
p.stem.split('-v')[0] for p in
(Path(__file__).parent / 'table-schemas/others').iterdir()
]
if schema_name in others:
schema = get_other_schema(schema_name, version, offline=offline)
else:
schema = get_table_schema(schema_name, version, offline=offline,
optional_fields=optional_fields)
except OSError as e:
return {e.strerror: Path(e.filename).name}
return get_table_errors(tsv_path, schema)
def main():
parser = argparse.ArgumentParser()
parser.add_argument('type', help='What type to generate')
parser.add_argument('target',
type=dir_path,
help='Directory to write output to')
args = parser.parse_args()
schema_versions = dict_schema_versions()
versions = sorted(schema_versions[args.type])
assert versions, f'No versions for {args.type}'
max_version = max(versions)
is_assay = get_is_assay(args.type)
if is_assay:
table_schemas = {v: get_table_schema(args.type, v) for v in versions}
directory_schema = get_directory_schema(args.type)
else:
table_schemas = {v: get_other_schema(args.type, v) for v in versions}
directory_schema = {}
# README.md:
with open(Path(args.target) / 'README.md', 'w') as f:
url = f'https://hubmapconsortium.github.io/ingest-validation-tools/{args.type}/'
f.write(f'Moved to [github pages]({url}).')
# index.md:
with open(Path(args.target) / 'index.md', 'w') as f:
f.write(
generate_readme_md(table_schemas,
directory_schema,
args.type,
is_assay=is_assay))
# YAML:
for v in versions:
schema = table_schemas[v]
first_field = schema['fields'][0]
if first_field['name'] == 'version':
assert first_field['constraints']['enum'] == [v], \
f'Wrong version constraint in {args.type}-v{v}.yaml'
assert schema['fields'][0]
with open(Path(args.target) / f'v{v}.yaml', 'w') as f:
f.write('# Generated YAML: PRs should not start here!\n' +
dump_yaml(schema))
# Data entry templates:
with open(
Path(args.target) / get_tsv_name(args.type, is_assay=is_assay),
'w') as f:
max_schema = table_schemas[max_version]
f.write(generate_template_tsv(max_schema))
create_xlsx(max_schema,
Path(args.target) /
get_xlsx_name(args.type, is_assay=is_assay),
idempotent=True,
sheet_name='Export as TSV')
def main():
parser = argparse.ArgumentParser()
parser.add_argument('type', help='What type to generate')
parser.add_argument('target',
type=dir_path,
help='Directory to write output to')
args = parser.parse_args()
schema_versions = dict_schema_versions()
versions = sorted(schema_versions[args.type])
max_version = max(versions)
is_assay = get_is_assay(args.type)
if is_assay:
table_schemas = {v: get_table_schema(args.type, v) for v in versions}
directory_schema = get_directory_schema(args.type)
else:
table_schemas = {v: get_other_schema(args.type, v) for v in versions}
directory_schema = {}
# README:
with open(Path(args.target) / 'README.md', 'w') as f:
f.write(
generate_readme_md(table_schemas,
directory_schema,
args.type,
is_assay=is_assay))
# YAML:
for v in versions:
with open(Path(args.target) / f'v{v}.yaml', 'w') as f:
f.write('# Generated YAML: PRs should not start here!\n' +
dump_yaml(table_schemas[v]))
# Data entry templates:
with open(
Path(args.target) / get_tsv_name(args.type, is_assay=is_assay),
'w') as f:
max_schema = table_schemas[max_version]
f.write(generate_template_tsv(max_schema))
create_xlsx(max_schema,
Path(args.target) /
get_xlsx_name(args.type, is_assay=is_assay),
idempotent=True,
sheet_name='Export as TSV')