2022-12 Change filter wastewater structures

[sjib]

• filtering on fk_wastewater_structures instead of selected wastewater_nodes:
  • replaced query = query.filter(QGEP.wastewater_networkelement.objid.in(subset_ids)) with query = query.filter(QGEP.wastewater_networkelement.fk_wastewaterstructure.in(wws_ids))
• to be able to do that qgep_export needs new argument selection2 with the obj_id's of the selected elements from vw_gep_wastewater_structure
• this needs a new function def selected_wws(self) in gui_export.py
• adapting the qgep_export call in init.py: qgep_export(selection=export_dialog.selected_ids, selection2=export_dialog.selected_wws, labels_file=labels_file_path)

[sjib]

Should fix https://github.com/QGEP/qgepqwat2ili/issues/93

[sjib]

Second wastewater_node now ok, but reach point of overflow pipe at wrong place:

and wrong relation to ak instead of overflow ak

[ponceta]

@olivierdalang will review this.

[olivierdalang]

@sjib Could you add a regression test (a test that would fail without your fix, and that works with your fix) ? You can copy test_case_e_export_selection in test_qgep.py to TestRegressions, adapt the selected ids as well as the assertions in the output (the tests use the demo data, so you can just use whatever you used when you manually reproduced the issue).

If you're not sure how this can be done, can you give me an example of an object id to select that creates an export with missing objects, also stating the ids of the missing objects in the export ?

Once that's in place, I will suggest some refactoring directly in the code:

• mainly reuse the same selection argument, as I think it's easier to just keep one list of selection ids, the fact that they are of different type of objects shouldn't matter
• remove comments from the code (since we're using git, comments with the date at which one line was added are better left out, this information is already present in the git history)

(it's better to do the regression test first, so we can ensure your fix will still work after refactoring)

[sjib]

I cannot find an example in the testdataset Arbon at the moment. I have to check back on the data with the issue. The integration of additional datamodels in Export / Import has higher priority.

[olivierdalang]

(converting this to draft so we don't merge by accident until comment are sorted)

[sjib]

Should solve: [Additional wastwater_nodes not exported with Limit to selection

93](https://github.com/QGEP/qgepqwat2ili/issues/93)

[ponceta]

@sjib do we want this for 1.6.1 or do we save it for later?

What has to be achieved to finish this work?

[sjib]

@sjib do we want this for 1.6.1 or do we save it for later?

What has to be achieved to finish this work?

The main point that Olivier suggested was to write a regression test. I tried, but I was not sucessful to find a suitable example in our demodata. So if this is the only missing part for 1.6.1 I suggest not to merge it for now, but still include this point in testing, if all wastewater_nodes will get exported (see original issue: "Additional wastwater_nodes not exported with Limit to selection"

93).

We then would have a reconfirmation of the need during testing (here especially include @DflGruBoe ) and could still integrate it in the final version.