def test_dataset_card(dataset_name):
card_path = repo_path / "datasets" / dataset_name / "README.md"
assert card_path.exists()
error_messages = []
try:
readme = ReadMe.from_readme(card_path)
except Exception as readme_parsing_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nREADME Parsing:\n{readme_parsing_error}"
)
try:
readme = ReadMe.from_readme(card_path, suppress_parsing_errors=True)
readme.validate()
except Exception as readme_validation_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nREADME Validation:\n{readme_validation_error}"
)
try:
metadata = DatasetMetadata.from_readme(card_path)
metadata.validate()
except Exception as metadata_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nYAML tags:\n{metadata_error}"
)
if error_messages:
raise ValueError("\n".join(error_messages))
def test_readme_from_readme_suppress_parsing_errors(readme_md):
with tempfile.TemporaryDirectory() as tmp_dir:
path = Path(tmp_dir) / "README.md"
with open(path, "w+") as readme_file:
readme_file.write(readme_md)
ReadMe.from_readme(path,
example_yaml_structure,
suppress_parsing_errors=True)
def test_readme_from_readme_parsing_errors(readme_md, expected_error):
with tempfile.TemporaryDirectory() as tmp_dir:
path = Path(tmp_dir) / "README.md"
with open(path, "w+") as readme_file:
readme_file.write(readme_md)
expected_error = expected_error.format(path=path)
with pytest.raises(ValueError, match=re.escape(expected_error)):
ReadMe.from_readme(path, example_yaml_structure)
def test_dataset_card(dataset_name):
card_path = repo_path / "datasets" / dataset_name / "README.md"
assert card_path.exists()
error_messages = []
try:
ReadMe.from_readme(card_path)
except Exception as readme_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nREADME:\n{readme_error}"
)
try:
DatasetMetadata.from_readme(card_path)
except Exception as metadata_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nYAML tags:\n{metadata_error}"
)
if error_messages:
raise ValueError("\n".join(error_messages))
def test_readme_from_readme_correct(readme_md, expected_dict):
with tempfile.TemporaryDirectory() as tmp_dir:
path = Path(tmp_dir) / "README.md"
with open(path, "w+") as readme_file:
readme_file.write(readme_md)
out = ReadMe.from_readme(path, example_yaml_structure).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
if args.check_all:
readmes = [
dd / "README.md" for dd in (repo_path / "datasets").iterdir()
]
else:
changed_files = get_changed_files(repo_path)
readmes = [
f for f in changed_files
if f.exists() and f.name.lower() == "readme.md"
and f.parent.parent.name == "datasets"
]
failed: List[Path] = []
for readme in sorted(readmes):
try:
ReadMe.from_readme(readme)
logging.debug(f"✅️ Validated '{readme.relative_to(repo_path)}'")
except ValueError as e:
failed.append(readme)
logging.warning(
f"❌ Validation failed for '{readme.relative_to(repo_path)}':\n{e}"
)
except Exception as e:
failed.append(readme)
logging.warning(
f"⁉️ Something unexpected happened on '{readme.relative_to(repo_path)}':\n{e}"
)
if len(failed) > 0:
logging.info(f"❌ Failed on {len(failed)} files.")
exit(1)
def test_readme_from_string_suppress_parsing_errors(readme_md):
ReadMe.from_string(readme_md,
example_yaml_structure,
suppress_parsing_errors=True)
def test_readme_from_string_parsing_errors(readme_md, expected_error):
with pytest.raises(ValueError,
match=re.escape(expected_error.format(path="root"))):
ReadMe.from_string(readme_md, example_yaml_structure)
def test_readme_from_string_correct(readme_md, expected_dict):
assert ReadMe.from_string(
readme_md, example_yaml_structure).to_dict() == expected_dict
