def should_report_invalid_json(
validation_results_factory_mock: MagicMock) -> None:
# Given
metadata_url = any_s3_url()
url_reader = MockJSONURLReader({metadata_url: StringIO(initial_value="{")})
validator = STACDatasetValidator(url_reader,
validation_results_factory_mock)
# When
with raises(JSONDecodeError):
validator.validate(metadata_url)
# Then
assert validation_results_factory_mock.mock_calls == [
call.save(
metadata_url,
Check.JSON_PARSE,
ValidationResult.FAILED,
details={
"message":
"Expecting property name enclosed in double quotes:"
" line 1 column 2 (char 1)"
},
),
]
def should_collect_assets_from_validated_collection_metadata_files(
subtests: SubTests) -> None:
# Given one asset in another directory and one relative link
base_url = any_s3_url()
metadata_url = f"{base_url}/{any_safe_filename()}"
stac_object = deepcopy(MINIMAL_VALID_STAC_COLLECTION_OBJECT)
first_asset_url = f"{base_url}/{any_safe_filename()}/{any_safe_filename()}"
first_asset_multihash = any_hex_multihash()
second_asset_filename = any_safe_filename()
second_asset_url = f"{base_url}/{second_asset_filename}"
second_asset_multihash = any_hex_multihash()
stac_object["assets"] = {
any_asset_name(): {
"href": first_asset_url,
"file:checksum": first_asset_multihash
},
any_asset_name(): {
"href": second_asset_filename,
"file:checksum": second_asset_multihash,
},
}
expected_assets = [
{
"multihash": first_asset_multihash,
"url": first_asset_url
},
{
"multihash": second_asset_multihash,
"url": second_asset_url
},
]
expected_metadata = [{"url": metadata_url}]
url_reader = MockJSONURLReader({metadata_url: stac_object})
with patch(
"backend.check_stac_metadata.utils.processing_assets_model_with_meta"
):
validator = STACDatasetValidator(url_reader,
MockValidationResultFactory())
# When
validator.validate(metadata_url)
# Then
with subtests.test():
assert _sort_assets(
validator.dataset_assets) == _sort_assets(expected_assets)
with subtests.test():
assert validator.dataset_metadata == expected_metadata
def should_collect_assets_from_validated_item_metadata_files(subtests: SubTests) -> None:
base_url = any_s3_url()
metadata_url = f"{base_url}/{any_safe_filename()}"
stac_object = deepcopy(MINIMAL_VALID_STAC_ITEM_OBJECT)
first_asset_url = f"{base_url}/{any_safe_filename()}"
first_asset_multihash = any_hex_multihash()
second_asset_filename = any_safe_filename()
second_asset_multihash = any_hex_multihash()
stac_object[STAC_ASSETS_KEY] = {
any_asset_name(): {
STAC_HREF_KEY: first_asset_url,
STAC_FILE_CHECKSUM_KEY: first_asset_multihash,
},
any_asset_name(): {
STAC_HREF_KEY: second_asset_filename,
STAC_FILE_CHECKSUM_KEY: second_asset_multihash,
},
}
expected_assets = [
{
PROCESSING_ASSET_MULTIHASH_KEY: first_asset_multihash,
PROCESSING_ASSET_URL_KEY: first_asset_url,
},
{
PROCESSING_ASSET_MULTIHASH_KEY: second_asset_multihash,
PROCESSING_ASSET_URL_KEY: f"{base_url}/{second_asset_filename}",
},
]
expected_metadata = [{PROCESSING_ASSET_URL_KEY: metadata_url}]
url_reader = MockJSONURLReader({metadata_url: stac_object})
with patch("backend.check_stac_metadata.utils.processing_assets_model_with_meta"):
validator = STACDatasetValidator(url_reader, MockValidationResultFactory())
validator.validate(metadata_url)
with subtests.test():
assert _sort_assets(validator.dataset_assets) == _sort_assets(expected_assets)
with subtests.test():
assert validator.dataset_metadata == expected_metadata