def _resolve_url(base_url, path):
"""
If path is a URL or an absolute path return URL
If path is a relative path return base_url joined with path
>>> _resolve_url('file:///foo/abc', 'bar')
'file:///foo/bar'
>>> _resolve_url('file:///foo/abc', 'file:///bar')
'file:///bar'
>>> _resolve_url('file:///foo/abc', None)
'file:///foo/abc'
>>> _resolve_url('file:///foo/abc', '/bar')
'file:///bar'
>>> _resolve_url('http://foo.com/abc/odc-metadata.yaml', 'band-5.tif')
'http://foo.com/abc/band-5.tif'
>>> _resolve_url('s3://foo.com/abc/odc-metadata.yaml', 'band-5.tif')
's3://foo.com/abc/band-5.tif'
>>> _resolve_url('s3://foo.com/abc/odc-metadata.yaml?something', 'band-5.tif')
's3://foo.com/abc/band-5.tif'
"""
if path:
if is_url(path):
url_str = path
elif Path(path).is_absolute():
url_str = Path(path).as_uri()
else:
url_str = urljoin(base_url, path)
else:
url_str = base_url
return url_str
def get_metadata_path(possible_path: Union[str, Path]) -> str:
"""
Find a metadata path for a given input/dataset path.
Needs to handle local files as well as remote URLs
"""
# We require exact URLs, lets skip any sort of fancy investigation and mapping
if isinstance(possible_path, str) and is_url(possible_path):
return possible_path
dataset_path = Path(possible_path)
# They may have given us a metadata file directly.
if dataset_path.is_file() and is_supported_document_type(dataset_path):
return str(dataset_path)
# Otherwise there may be a sibling file with appended suffix '.agdc-md.yaml'.
expected_name = dataset_path.parent.joinpath('{}.agdc-md'.format(
dataset_path.name))
found = _find_any_metadata_suffix(expected_name)
if found:
return str(found)
# Otherwise if it's a directory, there may be an 'agdc-metadata.yaml' file describing all contained datasets.
if dataset_path.is_dir():
expected_name = dataset_path.joinpath('agdc-metadata')
found = _find_any_metadata_suffix(expected_name)
if found:
return str(found)
raise ValueError('No metadata found for input %r' % dataset_path)
def expand_paths_as_uris(
input_paths: Iterable[str], ) -> Generator[Tuple[Path, bool], None, None]:
"""
For any paths that are directories, find inner documents that are known.
Returns Tuples: path as a URL, and whether it was specified explicitly by user.
"""
for input_ in input_paths:
if is_url(input_):
yield input_, True
else:
path = Path(input_).resolve()
if path.is_dir():
for found_path in path.rglob("*"):
if _readable_doc_extension(
found_path.as_uri()) is not None:
yield found_path.as_uri(), False
else:
yield path.as_uri(), True
def _web_reference(ref: str):
"""
A reference to a schema via a URL
eg http://geojson.org/schemas/Features.json'
"""
if not is_url(ref):
raise ValueError(f"Expected URL? Got {ref!r}")
(scheme, netloc, offset, params, query, fragment) = urllib.parse.urlparse(ref)
# We used `wget -r` to download the remote schemas locally.
# It puts into hostname/path folders by default. Eg. 'geojson.org/schema/Feature.json'
path = _SCHEMA_BASE / f"{netloc}{offset}"
if not path.exists():
if ALLOW_INTERNET:
path.parent.mkdir(parents=True, exist_ok=True)
path.write_bytes(urlopen(ref).read())
else:
raise ValueError(
f"No local copy exists of schema {ref!r}.\n"
f"\tPerhaps we need to add it to ./update.sh in the tests folder?\n"
f"\t(looked in {path})"
)
return read_document(path)
def _resolve_url(base_url, path):
"""
If path is a URL or an absolute path return URL
If path is a relative path return base_url joined with path
>>> _resolve_url('file:///foo/abc', 'bar')
'file:///foo/bar'
>>> _resolve_url('file:///foo/abc', 'file:///bar')
'file:///bar'
>>> _resolve_url('file:///foo/abc', None)
'file:///foo/abc'
>>> _resolve_url('file:///foo/abc', '/bar')
'file:///bar'
"""
if path:
if is_url(path):
url_str = path
elif Path(path).is_absolute():
url_str = Path(path).as_uri()
else:
url_str = urljoin(base_url, path)
else:
url_str = base_url
return url_str
def get_extension(url: str) -> jsonschema.Draft7Validator:
if not is_url(url):
raise ValueError(
f"stac extensions are now expected to be URLs in 1.0.0. " f"Got {url!r}"
)
return load_schema_doc(_web_reference(url), location=url)
