def test_get_format():
assert f.get_format(f.ContentType.APP_JSON) == Format(
f.ContentType.APP_JSON) # basic
assert f.get_format(f.ContentType.APP_JSON + "; charset=UTF-8") == Format(
f.ContentType.APP_JSON)
assert f.get_format(f.ContentType.APP_GEOJSON) == Format(
f.ContentType.APP_GEOJSON) # pywps vendor MIME-type
assert f.get_format(f.ContentType.APP_NETCDF
).encoding == "base64" # extra encoding data available
def test_get_cwl_file_format_synonym():
"""
Test handling of special non-official MIME-type that have a synonym redirection to an official one.
"""
res = f.get_cwl_file_format(f.ContentType.APP_TAR_GZ,
make_reference=False,
must_exist=True,
allow_synonym=False)
assert res == (
None, None
), "Non-official MIME-type without allowed synonym should resolve as not-found"
res = f.get_cwl_file_format(f.ContentType.APP_TAR_GZ,
make_reference=False,
must_exist=True,
allow_synonym=True)
assert isinstance(res, tuple)
assert res != (
None,
None), "Synonym type should have been mapped to its base reference"
assert res[1].split(
":"
)[1] == f.ContentType.APP_GZIP, "Synonym type should have been mapped to its base reference"
assert f.get_extension(
f.ContentType.APP_TAR_GZ
) == ".tar.gz", "Original extension resolution needed, not synonym"
fmt = f.get_format(f.ContentType.APP_TAR_GZ)
assert fmt.extension == ".tar.gz"
assert fmt.mime_type == f.ContentType.APP_TAR_GZ
# above tests validated that synonym is defined and works, so following must not use that synonym
res = f.get_cwl_file_format(f.ContentType.APP_TAR_GZ,
make_reference=True,
must_exist=False,
allow_synonym=True)
assert res.endswith(f.ContentType.APP_TAR_GZ), \
"Literal MIME-type expected instead of its existing synonym since non-official is allowed (must_exist=False)"
def format_inputs(self, workflow_inputs):
# type: (CWL_RuntimeInputList) -> OWS_InputDataValues
"""
Convert submitted :term:`CWL` workflow inputs into corresponding :mod:`OWSLib.wps` representation for execution.
:param workflow_inputs: mapping of input IDs and values submitted to the workflow.
:returns: converted OWS inputs ready for submission to remote WPS process.
"""
# prepare inputs
complex_inputs = []
for process_input in self.wps_process.dataInputs:
if WPS_COMPLEX_DATA in process_input.dataType:
complex_inputs.append(process_input.identifier)
wps_inputs = []
for input_item in workflow_inputs:
input_key = get_any_id(input_item)
input_val = get_any_value(input_item)
# ignore optional inputs resolved as omitted
if input_val is None:
continue
# in case of array inputs, must repeat (id,value)
# in case of complex input (File), obtain location, otherwise get data value
if not isinstance(input_val, list):
input_val = [input_val]
input_values = []
for val in input_val:
mime_type = None
encoding = None
if isinstance(val, dict):
fmt = val.get("format") # format as namespace:link
val = val["location"]
if fmt:
fmt = get_format(
fmt,
default=DEFAULT_FORMAT) # format as content-type
mime_type = fmt.mime_type or None
encoding = fmt.encoding or None # avoid empty string
# owslib only accepts strings, not numbers directly
if isinstance(val, (int, float)):
val = str(val)
input_values.append((val, mime_type, encoding))
# need to use ComplexDataInput structure for complex input
# TODO: BoundingBox not supported
for input_value, mime_type, encoding in input_values:
if input_key in complex_inputs:
input_value = ComplexDataInput(input_value,
mimeType=mime_type,
encoding=encoding)
wps_inputs.append((input_key, input_value))
return wps_inputs
def test_get_format_media_type_no_extension():
for ctype in [
f.ContentType.APP_OCTET_STREAM,
f.ContentType.APP_FORM,
f.ContentType.MULTI_PART_FORM,
]:
fmt = f.get_format(ctype)
assert fmt == Format(ctype, extension=None)
assert fmt.extension == ""
def test_get_format_default_no_extension():
for val in ["", None]:
for ctype in [
f.ContentType.APP_OCTET_STREAM,
f.ContentType.APP_FORM,
f.ContentType.MULTI_PART_FORM,
]:
fmt = f.get_format(val, default=ctype)
assert fmt == Format(ctype, extension=None)
assert fmt.extension == ""
def get_results(job, container, value_key=None, ogc_api=False):
# type: (Job, AnySettingsContainer, Optional[str], bool) -> Union[List[JSON], JSON]
"""
Obtains the job results with extended full WPS output URL as applicable and according to configuration settings.
:param job: job from which to retrieve results.
:param container: any container giving access to instance settings (to resolve reference output location).
:param value_key:
If not specified, the returned values will have the appropriate ``data``/``href`` key according to the content.
Otherwise, all values will have the specified key.
:param ogc_api:
If ``True``, formats the results using the ``OGC API - Processes`` format.
:returns: list of all outputs each with minimally an ID and value under the requested key.
"""
wps_url = get_wps_output_url(container)
if not wps_url.endswith("/"):
wps_url = wps_url + "/"
outputs = {} if ogc_api else []
fmt_key = "mediaType" if ogc_api else "mimeType"
for result in job.results:
rtype = "data" if any(k in result
for k in ["data", "value"]) else "href"
value = get_any_value(result)
out_id = get_any_id(result)
out_key = rtype
if rtype == "href":
# fix paths relative to instance endpoint, but leave explicit links as is (eg: S3 bucket, remote HTTP, etc.)
if value.startswith("/"):
value = str(value).lstrip("/")
if "://" not in value:
value = wps_url + value
elif ogc_api:
out_key = "value"
elif value_key:
out_key = value_key
output = {out_key: value}
if rtype == "href": # required for the rest to be there, other fields optional
if "mimeType" not in result:
result["mimeType"] = get_format(
value, default=CONTENT_TYPE_TEXT_PLAIN).mime_type
output["format"] = {fmt_key: result["mimeType"]}
for field in ["encoding", "schema"]:
if field in result:
output["format"][field] = result[field]
elif rtype != "href":
# literal data
# FIXME: BoundingBox not implemented (https://github.com/crim-ca/weaver/issues/51)
dtype = result.get(
"dataType",
any2wps_literal_datatype(value, is_value=True) or "string")
if ogc_api:
output["dataType"] = {"name": dtype}
else:
output["dataType"] = dtype
if ogc_api:
if out_id in outputs:
output_list = outputs[out_id]
if not isinstance(output_list, list):
output_list = [output_list]
output_list.append(output)
outputs[out_id] = output_list
else:
outputs[out_id] = output
else:
# if ordered insert supported by python version, insert ID first
output = dict([("id", out_id)] + list(output.items())) # noqa
outputs.append(output)
return outputs
def get_results(
job, # type: Job
container, # type: AnySettingsContainer
value_key=None, # type: Optional[str]
schema=JobInputsOutputsSchema.OLD, # type: JobInputsOutputsSchemaType
link_references=False, # type: bool
): # type: (...) -> Tuple[ExecutionResults, HeadersTupleType]
"""
Obtains the job results with extended full WPS output URL as applicable and according to configuration settings.
:param job: job from which to retrieve results.
:param container: any container giving access to instance settings (to resolve reference output location).
:param value_key:
If not specified, the returned values will have the appropriate ``data``/``href`` key according to the content.
Otherwise, all values will have the specified key.
:param schema:
Selects which schema to employ for representing the output results (listing or mapping).
:param link_references:
If enabled, an output that was requested by reference instead of value will be returned as ``Link`` reference.
:returns:
Tuple with:
- List or mapping of all outputs each with minimally an ID and value under the requested key.
- List of ``Link`` headers for reference outputs when requested. Empty otherwise.
"""
settings = get_settings(container)
wps_url = get_wps_output_url(settings)
if not wps_url.endswith("/"):
wps_url = wps_url + "/"
schema = JobInputsOutputsSchema.get(str(schema).lower(),
default=JobInputsOutputsSchema.OLD)
strict = schema.endswith("+strict")
schema = schema.split("+")[0]
ogc_api = schema == JobInputsOutputsSchema.OGC
outputs = {} if ogc_api else []
fmt_key = "mediaType" if ogc_api else "mimeType"
out_ref = convert_output_params_schema(
job.outputs, JobInputsOutputsSchema.OGC) if link_references else {}
references = {}
for result in job.results:
rtype = "data" if any(k in result
for k in ["data", "value"]) else "href"
value = get_any_value(result)
out_key = rtype
out_id = get_any_id(result)
out_mode = out_ref.get(out_id, {}).get("transmissionMode")
as_ref = link_references and out_mode == ExecuteTransmissionMode.REFERENCE
if rtype == "href":
# fix paths relative to instance endpoint, but leave explicit links as is (eg: S3 bucket, remote HTTP, etc.)
if value.startswith("/"):
value = str(value).lstrip("/")
if "://" not in value:
value = wps_url + value
elif ogc_api:
out_key = "value"
elif value_key:
out_key = value_key
output = {out_key: value}
if rtype == "href": # required for the rest to be there, other fields optional
if "mimeType" not in result:
result["mimeType"] = get_format(
value, default=ContentType.TEXT_PLAIN).mime_type
if ogc_api or not strict:
output["type"] = result["mimeType"]
if not ogc_api or not strict or as_ref:
output["format"] = {fmt_key: result["mimeType"]}
for field in ["encoding", "schema"]:
if field in result:
output["format"][field] = result[field]
elif rtype != "href":
# literal data
# FIXME: BoundingBox not implemented (https://github.com/crim-ca/weaver/issues/51)
dtype = result.get(
"dataType",
any2wps_literal_datatype(value, is_value=True) or "string")
if ogc_api:
output["dataType"] = {"name": dtype}
else:
output["dataType"] = dtype
if ogc_api or as_ref:
mapping = references if as_ref else outputs
if out_id in mapping:
output_list = mapping[out_id]
if not isinstance(output_list, list):
output_list = [output_list]
output_list.append(output)
mapping[out_id] = output_list
else:
mapping[out_id] = output
else:
# if ordered insert supported by python version, insert ID first
output = dict([("id", out_id)] + list(output.items())) # noqa
outputs.append(output)
# needed to collect and aggregate outputs of same ID first in case of array
# convert any requested link references using indices if needed
headers = []
for out_id, output in references.items():
res_links = make_result_link(out_id, output, job.id, settings)
headers.extend([("Link", link) for link in res_links])
return outputs, headers