Total axes

[zschira]

This PR changes the extractor behavior so missing Axes will be treated as Totals.

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: +11.46% 🎉

Comparison is base (0572a99) 77.25% compared to head (19ae97c) 88.72%.

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #118       +/-   ##
===========================================
+ Coverage   77.25%   88.72%   +11.46%     
===========================================
  Files           8        8               
  Lines         554      541       -13     
===========================================
+ Hits          428      480       +52     
+ Misses        126       61       -65     

Files Changed	Coverage Δ
src/ferc_xbrl_extractor/arelle_interface.py	100.00% <100.00%> (ø)
src/ferc_xbrl_extractor/cli.py	77.27% <100.00%> (ø)
src/ferc_xbrl_extractor/datapackage.py	98.52% <100.00%> (+10.41%)	⬆️
src/ferc_xbrl_extractor/instance.py	97.43% <100.00%> (+40.52%)	⬆️
src/ferc_xbrl_extractor/taxonomy.py	86.20% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jdangerx]

The only actually blocking thing is the question about whether dropna() is doing what we want it to do. But I would feel warm and fuzzy inside if you engaged with my various suggestions, too!

Broader questions I wanted to reiterate at the top level:

• do we need to filter by context dimensions in get_facts? it might make sense to just grab all the facts. then we can filter contexts by whether or not each context fits with the schema.primary_key, and avoid passing extra state around.
• should we consider not smearing the concept names into their own columns? I'm sure there was some sort of decision made about this in the ancient times, but what was the rationale there?
• should we have the output dataframes be indexed by schema.primary_key?

Finally, when I was testing the code I'm suggesting, I found a couple weird things that I needed to handle:

• empty files, obviously
• filings with duplicated ("c_id", "name") values in facts... that seemed surprising.