add Tables.jl interface

[rafaqz]

This PR adds a standard Tables.jl interface and removes Requires.jl and DataFrames code, as it's no longer needed.

Now a CSV can be saved directly from a GBIFRecords object without DataFrames.jl being involved.

A DataFrame will now include all fields as-is form the GBIFRecord (which may be up for debate, but e.g. basisOfRecord can be good to have in a table...), as well as the taxon fields previously included.

This is done by adding methods to propertynames and getproperty, which means a GBIFRecord also has the same properties accessible with the record.property syntax. To get the actual taxon object, you use getfield(record, :taxon). Or we can add a wrapper method.

[rafaqz]

The error seems to be an issue with DataFrames and CategoricalArrays? wierd

[tpoisot]

Oh it's what is making the tests fail in #42 - I like the idea (@gabrieldansereau might like it even more as he's our resident "export thing as CSV" evangelist).

If the issue really is 1.5, we might make this a major version and set 1.6 as the lower bound.

[rafaqz]

The docs on all these PRs also have a problem with QT not finding the xcb library to run visually without X on linux.

[codecov[bot]]

Codecov Report

Merging #41 (1e04917) into main (23b6921) will decrease coverage by 1.03%. The diff coverage is 81.81%.

@@            Coverage Diff             @@
##             main      #41      +/-   ##
==========================================
- Coverage   85.95%   84.91%   -1.04%     
==========================================
  Files          10        9       -1     
  Lines         178      179       +1     
==========================================
- Hits          153      152       -1     
- Misses         25       27       +2     

Flag	Coverage Δ
unittests	84.91% <81.81%> (-1.04%)	:arrow_down:

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/tables.jl	66.66% <66.66%> (ø)
src/types/GBIFRecords.jl	90.62% <87.50%> (-1.05%)	:arrow_down:

---

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 23b6921...1e04917. Read the comment docs.

[rafaqz]

The problem was just old dependencies in the tests. I'm not sure why Pkg chose them, but adding fixed versions like 1 for DataFrames seems to have fixed the issue. This is good to go.

[rafaqz]

I would really like to merge this... but it needs review some.

Specifically it may break everything that built off dataframes output before because it includes most of the fields now, and some appear to have previously been named differently to GBIF, and to the struct field names? Now there is no distinction.

[tpoisot]

Can you rebase on the main branch? The type piracy issue is gone

[tpoisot]

I really want a review from @gabrieldansereau on this - if it's going to break things in your workflows, then maybe we shouldn't merge it yet.

[tpoisot]

I also won't merge unless properly documented - I'll start my review when all the elements are here.

[rafaqz]

I also won't merge unless properly documented - I'll start my review when all the elements are here.

Before I start, can I clarify what you want documented? Nothing in the current docs appears to be changed by this PR.

We could however add a listing of available table columns, which are not currently documented.

[tpoisot]

We could however add a listing of available table columns, which are not currently documented.

Yep

[rafaqz]

This is documented and ready to go, besides possible changes to GBIFRecord that would make the code a lot simpler.