Merge pull request #566 from pyinat/root-taxon-filter

Add root_id filter to taxon.make_tree()

HISTORY.md

@@ -7,6 +7,7 @@
 
 ### Models
 * Add `User.annotated_observations_count` field
+* Add `root_id` filter to `taxon.make_tree()` to explicitly set the root taxon instead of determining it automatically
 
 ### Rate limits, timeouts, and error handling
 * Increase default request timeout from 10 to 20 seconds

pyinaturalist/models/taxon.py

@@ -1,5 +1,6 @@
 import re
 from itertools import chain, groupby
+from logging import getLogger
 from typing import Any, Callable, Dict, Iterable, List, Optional
 
 from pyinaturalist.constants import (
@@ -33,6 +34,8 @@
     field,
 )
 
+logger = getLogger(__name__)
+
 
 @define_model
 class Taxon(BaseModel):
@@ -408,19 +411,23 @@ def make_tree(
     taxa: Iterable[Taxon],
     include_ranks: Optional[List[str]] = None,
     sort_key: Optional[TaxonSortKey] = None,
+    root_id: Optional[int] = None,
 ) -> Taxon:
     """Organize a list of taxa into a taxonomic tree. Expects exactly one root taxon.
 
     Args:
+        taxa: Taxon objects to organize
         sort_key: Key function for sorting children; defaults to rank and name
         include_ranks: If provided, only include taxa with these ranks; otherwise, include all ranks
+        root_id: ID of the root taxon; if provided, only that taxon and its descendants will
+            be included. Otherwise, the root taxon is determined automatically.
 
     Returns:
         Root taxon of the tree
     """
     include_ranks = [r.lower() for r in include_ranks or []]
-    root = _find_root(taxa, include_ranks)
     sort_key = sort_key if sort_key is not None else _sort_rank_name
+    root = _find_root(taxa, include_ranks, root_id)
 
     # Group taxa by parent ID, including any ungrafted children added directly to root
     taxa_by_parent: Dict[int, List[Taxon]] = _sort_groupby(taxa, key=lambda x: x.parent_id or -1)
@@ -444,16 +451,33 @@ def add_descendants(taxon, ancestors=None) -> Taxon:
     return add_descendants(root)
 
 
-def _find_root(taxa: Iterable[Taxon], include_ranks: Optional[List[str]] = None) -> Taxon:
+def _find_root(
+    taxa: Iterable[Taxon],
+    include_ranks: Optional[List[str]] = None,
+    root_id: Optional[int] = None,
+) -> Taxon:
     """Find the root taxon of a list of taxa, optionally filtering by rank.
     Handles ungrafted and multiple root taxa by adding under a new root node.
     """
+    # If a specific root taxon is requested, use that if possible
+    if root_id:
+        root = next((t for t in taxa if t.id == root_id), None)
+        if root and (not include_ranks or root.rank in include_ranks):
+            return root
+        else:
+            logger.warning(f'Root taxon {root_id} not found or filtered out; finding default root')
+
     # Typical case: exactly one root taxon ("Life")
     taxa_by_id = {t.id: t for t in taxa}
     if ROOT_TAXON_ID in taxa_by_id and (not include_ranks or 'stateofmatter' in include_ranks):
         return taxa_by_id[ROOT_TAXON_ID]
+    # Otherwise, find the highest-ranked taxa and graft them under a new root
+    else:
+        return _find_and_graft_root(taxa, include_ranks)
+
 
-    # Otherwise, find the taxa with the highest rank
+def _find_and_graft_root(taxa: Iterable[Taxon], include_ranks: Optional[List[str]] = None) -> Taxon:
+    taxa_by_id = {t.id: t for t in taxa}
     max_rank = max(t.rank_level for t in taxa if not include_ranks or t.rank in include_ranks)
     root_taxa = [t for t in taxa if t.rank_level == max_rank]
 

test/test_models.py

@@ -1240,6 +1240,33 @@ def test_make_tree__ungrafted():
     assert root.children[1].name == 'Monocots'
 
 
+def test_make_tree__explicit_root():
+    """If a root taxon is provided, it should be used as the root node if possible"""
+    taxa = Taxon.from_json_list(j_life_list_2)
+    root = make_tree(taxa, root_id=1)
+    assert root.id == 1
+    assert len(root.children) == 1
+    assert root.children[0].name == 'Arthropoda'
+
+
+def test_make_tree__explicit_root_not_found():
+    """If a root taxon is provided but not included in the list, fall back to default behavior"""
+    taxa = Taxon.from_json_list(j_life_list_2)
+    root = make_tree(taxa, root_id=12345)
+    assert root.id == ROOT_TAXON_ID
+    assert len(root.children) == 1
+
+
+def test_make_tree__explicit_root_filtered_out():
+    """If a root taxon is provided but filtered out, find the next root matching the filter"""
+    taxa = Taxon.from_json_list(j_life_list_2)
+    root = make_tree(taxa, root_id=1, include_ranks=['family', 'genus', 'species'])
+    assert root.id == 47221
+    assert root.name == 'Apidae'
+    assert root.rank == 'family'
+    assert len(root.children) == 1
+
+
 # Users
 # --------------------