Improvements for taxonomy trees and formatting

[JWCook]

Updates #499

Formatting:

• Omit rank in Taxon.full_name for species and below (to be more consistent with iNat's formatting)
• Fix common name title casing with apostrophes, parentheses, and dashes
• Add separate Taxon.rich_full_name property to use with terminal output
  • Italicize only for genus and below
  • Use for table output with pprint()

Ranks:

• Normalize Taxon.rank (casing, abbreviations, and other variations)
• Set Taxon.rank_level if rank name exists but level doesn't (for consistent rank sorting/comparison)
• Default Taxon.rank_level to 90 (unranked) if neither rank name or level is available
• Update constants.COMMON_RANKS to only include the main 7

Trees:

• Update Taxon.ancestors in make_tree()
• Add optional rank filtering to make_tree()
• Handle tree roots other than "Life" in make_tree()
• Handle ungrafted taxa and multiple roots in make_tree() by inserting "Life" as root node
• Remove LifeList.tree() in favor of make_tree()
  • This can be used with LifeList or a list of Taxon objects
• Add hide_root option to Taxon.flatten() to exclude the root from the list and from indentation level

[codecov[bot]]

Codecov Report

Merging #502 (e8d3bf2) into main (aeebf5a) will increase coverage by 0.19%. The diff coverage is 97.46%.

@@            Coverage Diff             @@
##             main     #502      +/-   ##
==========================================
+ Coverage   95.81%   96.01%   +0.19%     
==========================================
  Files          58       58              
  Lines        3179     3212      +33     
  Branches      609      622      +13     
==========================================
+ Hits         3046     3084      +38     
+ Misses         94       91       -3     
+ Partials       39       37       -2     

Impacted Files	Coverage Δ
pyinaturalist/models/identification.py	100.00% <ø> (ø)
pyinaturalist/models/observation.py	98.41% <ø> (ø)
pyinaturalist/models/taxon.py	98.54% <97.36%> (+3.11%)	:arrow_up:
pyinaturalist/constants.py	100.00% <100.00%> (ø)
pyinaturalist/formatters.py	91.00% <100.00%> (ø)

[synrg]

See the snippet from dronefly-core below which works when ranks_to_count is COMMON_RANKS, RANK_LEVELS.keys(), "leaf", or "species". However, I have a few unresolved issues with my code so far that could be helped by additional changes in this PR.

1. Maybe support filtering other than just by a list of included ranks? In particular, I had to do a special case for leaves. Users might have other particular criteria as well.

2. I had to add my own filter for the "species" case even though I passed it to make_tree() as one of the ranks to include because of the auto-inclusion of the Life root. I don't want Life included in my output, and in any case, the generated node is created as a Taxon not a TaxonCount, making it unusable in my output without further adjustments. Because it isn't a TaxonCount, it lacks descendant_obs_count - that would throw an exception if I did include it in my display. I'm not exactly sure what to do here - maybe this is a case where having the filter in the flatten as well makes sense?

3. Also, maybe the default sort order could be improved? With the following, the per species and per any cases produce nicely sorted output. However, the per main and per leaf cases produce out-of-order lists of species due to filtered ranks collapsing higher rank level groupings.

def taxa_per_rank(life_list: LifeList, ranks_to_count: Union(list[str], str)):
    options = {}
    if isinstance(ranks_to_count, list):
        options["include_ranks"] = ranks_to_count
        include = lambda _t: True
    else:
        if ranks_to_count == "leaf":
            include = lambda t: t.count == t.descendant_obs_count
        else:
            options["include_ranks"] = [ranks_to_count]
            include = lambda t: t.rank == ranks_to_count
    for taxon_count in make_tree(life_list.data, **options).flatten():
        if include(taxon_count):
            yield taxon_count

[synrg]

Good. Apart from this slight glitch in how my TaxonFormatter deals with the autocreated TaxonCount for life, everything looks fine now:

According to this, is_active should be True for Life:

[JWCook]

Thanks for testing this out!

I don't want Life included in my output

In the case where this needs to be added due to multiple root nodes, maybe I could add a flag indicating it can be ignored elsewhere (including in flatten())?

the generated node is created as a Taxon not a TaxonCount

Fixed.

out-of-order lists of species due to filtered ranks collapsing higher rank level groupings

Ah, I thought I accounted for that by applying sorting before skipping any ranks, but it must not be working as intended. To help with debugging, do you have an example that shows this?

[JWCook]

is_active should be True for Life:

Fixed

[synrg]

Thanks. I haven't fully caught up to all the changes (e.g. not yet making use of ancestors you've built) but at least our branches are now compatible. If the flattened tree has Life at the root I just skip over that one now and decrease the indent level at the same time. I imagine the code I just added will be even simpler once I fully take advantage of everything you've done. This is coming together very nicely!

[JWCook]

Okay, if this is working for you now, I'll go ahead and merge what I've got so far.

I also added a hide_root option to Taxon.flatten(), but we could replace that with a better solution later on.