Cannot load CWL definition directly with Python objects using Factory class

[fmigneault]

Expected Behavior

from cwltool.factory import Factory
import json 
with open("/tmp/base.json", mode="r", encoding="utf-8") as f:
    cwl = json.load(f)
cwl
{'cwlVersion': 'v1.2', 'class': 'CommandLineTool', 'baseCommand': 'echo', 'inputs': {'input': {'type': 'int'}}, 'outputs': {'output': 'stdout'}}

#----------
tool = Factory().make("/tmp/base.json")
Resolved '/tmp/base.json' to 'file:///tmp/base.json'

tool(input=1)
[job base.json] /tmp/2ntkyyif$ echo > /tmp/2ntkyyif/054dc6206b72fb355c36f9ab1c37be5233d88d70
[job base.json] completed success
{'output': {'location': 'file:///tmp/2ntkyyif/054dc6206b72fb355c36f9ab1c37be5233d88d70', 'basename': '054dc6206b72fb355c36f9ab1c37be5233d88d70', 'nameroot': '054dc6206b72fb355c36f9ab1c37be5233d88d70', 'nameext': '', 'class': 'File', 'checksum': 'sha1$adc83b19e793491b1c6ea0fd8b46cd9f32e592fc', 'size': 1, 'http://commonwl.org/cwltool#generation': 0}}

#----------
tool = Factory().make(cwl)
Traceback (most recent call last):
  File "/opt/pycharm-2023.2/plugins/python/helpers/pydev/pydevconsole.py", line 364, in runcode
    coro = func()
  File "<input>", line 1, in <module>
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/factory.py", line 67, in make
    load = load_tool.load_tool(cwl, self.loading_context)
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/load_tool.py", line 605, in load_tool
    loadingContext, uri = resolve_and_validate_document(
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/load_tool.py", line 519, in resolve_and_validate_document
    uri = processobj["id"]
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/ruamel/yaml/comments.py", line 927, in __getitem__
    return ordereddict.__getitem__(self, key)
KeyError: 'id'

#-------- 
cwl["id"] = "file:///tmp/base.json"
tool = Factory().make(cwl)

tool(input=1)
[job base.json_2] /tmp/vn0i5ukp$ echo > /tmp/vn0i5ukp/054dc6206b72fb355c36f9ab1c37be5233d88d70
[job base.json_2] completed success
{'output': {'location': 'file:///tmp/vn0i5ukp/054dc6206b72fb355c36f9ab1c37be5233d88d70', 'basename': '054dc6206b72fb355c36f9ab1c37be5233d88d70', 'nameroot': '054dc6206b72fb355c36f9ab1c37be5233d88d70', 'nameext': '', 'class': 'File', 'checksum': 'sha1$adc83b19e793491b1c6ea0fd8b46cd9f32e592fc', 'size': 1, 'http://commonwl.org/cwltool#generation': 0}}

Actual Behavior

When the id field is missing and the CWL is provided directly as Python objects, the following error is raised:

KeyError: processobj["id"]

However, the code seems to attempt adding a temporary id (in the form _:<UUID> beforehand) to work around the "missing" document reference when the contents are provided directly. Therefore, the above processobj reference seems invalid.

Workflow Code

see above

Full Traceback

Traceback (most recent call last):
  File "/opt/pycharm-2023.2/plugins/python/helpers/pydev/pydevconsole.py", line 364, in runcode
    coro = func()
  File "<input>", line 1, in <module>
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/factory.py", line 67, in make
    load = load_tool.load_tool(cwl, self.loading_context)
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/load_tool.py", line 605, in load_tool
    loadingContext, uri = resolve_and_validate_document(
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/cwltool/load_tool.py", line 519, in resolve_and_validate_document
    uri = processobj["id"]
  File "/home/francis/dev/conda/envs/weaver/lib/python3.10/site-packages/ruamel/yaml/comments.py", line 927, in __getitem__
    return ordereddict.__getitem__(self, key)
KeyError: 'id'

Your Environment

• cwltool version: 3.1.20230906142556