split read/write schema validation

[marscher]

Changes

WeldxFile now supports two custom_schema sub-arguments. The first argument will be used upon reading, the second upon writing. That way we can use separate schema for both operations. It turned out, that this is useful during schema development.

Related Issues

Closes #690

Checks

• updated CHANGELOG.rst
• updated tests
• updated doc/
• update example/tutorial notebooks
• update manifest file

[github-actions]

Unit Test Results

       1 files  ±0         1 suites  ±0   1m 40s ⏱️ +18s
2 108 tests +7  2 108 ✔️ +7  0 💤 ±0  0 ❌ ±0 

Results for commit 96abb03. ± Comparison against base commit ae2dff5.

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

Merging #697 (96abb03) into master (ae2dff5) will decrease coverage by 0.01%.
The diff coverage is 93.75%.

@@            Coverage Diff             @@
##           master     #697      +/-   ##
==========================================
- Coverage   96.08%   96.06%   -0.02%     
==========================================
  Files          92       92              
  Lines        6330     6350      +20     
==========================================
+ Hits         6082     6100      +18     
- Misses        248      250       +2     

Impacted Files	Coverage Δ
weldx/asdf/file.py	96.59% <93.75%> (-0.49%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ae2dff5...96abb03. Read the comment docs.

[marscher]

The sphinx warnings are due to the intersphinx mapping to Scipy is currently broken.

[CagtayFabry]

Thank you for adding a testcase for this !

I think there is a way with asdf to provide a schema without saving an actual file but I am not sure if it works with our WeldxFile wrapper

Here is an example:
https://github.com/asdf-format/asdf/blob/66abc5e4599db381850dddcb293a90c4a9afd999/asdf/tests/test_schema.py#L196-L204

I will look into if we can apply this somehow in our test cases, it should make a lot of things much easier

but this solution is good

[marscher]

Thanks for providing the file-free version of schema validation! This looks very useful. Since WeldxFile respects outer Config contexts, I think it would be possible to use this pattern. I will try it out quickly.

[CagtayFabry]

I think what might be an issue is that we only look for schemas as files in our current implementation

weldx/weldx/asdf/file.py

Lines 221 to 230 in ae2dff5

	if custom_schema is not None:
	_custom_schema_path = pathlib.Path(custom_schema)
	if not _custom_schema_path.exists():
	try:
	custom_schema = get_schema_path(custom_schema)
	except ValueError:
	raise ValueError(
	f"provided custom_schema {custom_schema} " "does not exist."
	)
	asdffile_kwargs["custom_schema"] = custom_schema

what we might need is another 'fallback' into asdf.schema.load_schema() in the place where we 'resolve' the schema

[marscher]

Although it is a nice pattern, the custom_schema argument routing (e.g. tuple unpacking) is only possible within the wrapper. Also AsdfFile expects user schemas to be a file or URI: https://github.com/asdf-format/asdf/blob/66abc5e4599db381850dddcb293a90c4a9afd999/asdf/tests/test_schema.py#L899

[CagtayFabry]

So we cant use it here?

[CagtayFabry]

I like this solution 👍