Improved ExtractionTest Validation of Attachment and Extracted Children Expected vs Found Count

[sambish5]

This is in reference to NationalSecurityAgency/burrito-grande#884.
Improved the ExtractionTest to further checks on Attachment and Extracted Children counts. Includes new tests in TestExtractionTest to verify count check works.

[dev-mlb]

lgtm - just need to fix the import statements

[drivenflywheel]

This seems way more burdensome than the original ask, which was "don't allow the XML to have more child attachment or extractedRecord elements than specified in the numAttachments / extractionCount elements".

As currently implemented, we can no longer assert the numAttachments or extractionCount without creating a child element for every one of those children. That could cause an enormous burden for downstream integration. and it will actively hinder extraction tests for some packaged file processing. If nothing else, the ExtractionTest class should provide the ability to disable this check on a class-by-class or test-by-test basis so downstream projects can integrate this version of emissary without requiring a huge integration PR with thousands of lines of new XML.

[sambish5]

@drivenflywheel @dev-mlb some major updates within this. Instead of having it fail if <numAttachments> and <extractCount> do not match <att#> and <extract#>, it will now just log the non-matching counts. I currently have it written so even if numAttachments/extractCount are not specified, the check still occurs (this can easily be moved to within the if > -1 parts if we only want it to check when those are specified, but I figured we'd want it to always be specified if there are att or extract in the xml).

If you guys would prefer that method however, I can move it inside of those ifs, and in the future if we want to build back out the functionality to fail if the counts do not match, we can.

[drivenflywheel]

Last changes? 🤞

[sambish5]

@drivenflywheel made the suggested changes, had to make some slight changes to the assertion things to make it work like how it did when it was just logging

[drivenflywheel]

Looks good. Sorry for the high number cycles.

[sambish5]

rebased and made some minor updates to just include the test name when validation finds expected vs found count not equal on assertion, hopefully this can close this PR 🤞

[sambish5]

Changes made by request of @jpdahlke
Overview of changes:

• no longer fails if <numAttachments> & <extractCount> DNE <att#> or <extract#> in answer file, now just logs differences.
• Has a new priority system to make sure that counts in answer file match counts in payload:
  - Checks <numAttachments>/<extractCount> if present in answer file against count from payload
  - If those are not present, check against <att#>/<extract#> if present
  - If neither are present, and there are atts or extracts in the payload, FAIL the test, saying <numAttachments>/<extractCount> matching payload count needs to be in answer file

[jpdahlke]

I think this update makes sense given our integration experiences.