publish-directory - Test repeat publish using --force mode#508
publish-directory - Test repeat publish using --force mode#508ce10-sanger wants to merge 6 commits intowtsi-npg:develfrom
--force mode#508Conversation
| for x in DataObject(dest / "c.txt").metadata() | ||
| if x.attribute == "dcterms:created" | ||
| ] | ||
| assert len(created_values) == 1 |
There was a problem hiding this comment.
Assertion on line 225 should be moved here. However, I am not sure what is being tested on line 225. Are we testing that the original file creation date does not change when we asked to update the content of the file? Fair enough. We should also test that the content of the file in iRODS has been updated and a new md5 has been assigned iRODS. So we should capture md5 of the original file on disk and md5 of the updated file on disk, make sure that they differ and check that we get correct md5 and optionally get the content of the file from iRODS and compare to the updated file on disk.
| # Arrange | ||
| src = Path("./tests/data/ultima/minimal").absolute() | ||
| src = tmp_path / "minimal" | ||
| shutil.copytree("./tests/data/ultima/minimal", src) |
There was a problem hiding this comment.
Really, test prep should be set up in a fixture (see conftest.py) so that the majority of the code in the tests is testing things, rather than building state.
| @m.context( | ||
| 'When run in place of npg_publish_tree.pl in "Uploading Ultima run to iRODS" SOP from rnd_platforms' | ||
| ) | ||
| @m.it("Should be compatible with npg_publish_tree.pl") |
There was a problem hiding this comment.
This is pretty broad - what aspects of the behaviour are being picked out? The following test is large and by the end it has overflowed my context window, so I'm not sure what's covered (and I'm familiar with both systems).
| assert [x.attribute for x in DataObject(dest / "b.txt").metadata()] == [ | ||
| "dcterms:created", | ||
| "dcterms:creator", | ||
| "md5", | ||
| "type", | ||
| ] | ||
|
|
||
| assert [x.attribute for x in DataObject(dest / "c.txt").metadata()] == [ | ||
| "dcterms:created", | ||
| "dcterms:creator", | ||
| "md5", | ||
| "md5", | ||
| "type", | ||
| ] | ||
| assert [ | ||
| x.value | ||
| for x in DataObject(dest / "c.txt").metadata() | ||
| if x.attribute == "md5" | ||
| ] == [ | ||
| c_txt_original_md5, | ||
| c_txt_modified_md5, | ||
| ], "New md5 attribute should be added, previous preserved to provide history" | ||
| assert ( | ||
| DataObject(dest / "c.txt").read() == "modified" | ||
| ), "Contents of data object in iRODS should have updated" | ||
| assert ( | ||
| DataObject(dest / "c.txt").checksum() == c_txt_modified_md5 | ||
| ), "Checksum should have updated" |
There was a problem hiding this comment.
Take care not to recapitulate tests that are already covered in partisan (which is responsible for what you're testing here).
There was a problem hiding this comment.
Cheers for the feedback Keith. Yea I've been wondering about the structure/approach here too - I'll tighten it up so we've got something tighter to discuss around, share context of what I'm trying to do here and then perhaps we can take a look at it together
3 participants
JIRA: NPG-3524
Fix docs. Default publish_directory(force) is False #488
Test repeat publish using
--forcemode.Currently
--fillmode fails in case of updated file. Captured addressing and switching to this mode in NPG-3758