Taxon updates (#181)

* Add conservation status, establishment means, and wikipedia
description + url to DbTaxon model
* Don't update a full taxon record with a partial one

pyinaturalist_convert/_models.py

@@ -175,18 +175,22 @@ class DbTaxon:
     id: int = sa_field(Integer, primary_key=True)
     ancestor_ids: str = sa_field(String, default=None)
     child_ids: str = sa_field(String, default=None)
+    conservation_status: str = sa_field(String, default=None)
+    establishment_means: str = sa_field(String, default=None)
     iconic_taxon_id: int = sa_field(Integer, default=0)
     is_active: bool = sa_field(Boolean, default=None)
     leaf_taxa_count: int = sa_field(Integer, default=0)
     observations_count: int = sa_field(Integer, default=0)
     observations_count_rg: int = sa_field(Integer, default=0)
     name: str = sa_field(String, default=None, index=True)
     parent_id: int = sa_field(ForeignKey('taxon.id'), default=None, index=True)
-    partial: int = sa_field(Boolean, default=False)
+    partial: bool = sa_field(Boolean, default=False)
     photo_urls: str = sa_field(String, default=None)
     preferred_common_name: str = sa_field(String, default=None)
     rank: str = sa_field(String, default=None)
     reference_url: str = sa_field(String, default=None)
+    wikipedia_summary: str = sa_field(String, default=None)
+    wikipedia_url: str = sa_field(String, default=None)
 
     @classmethod
     def from_model(cls, taxon: Taxon) -> 'DbTaxon':
@@ -195,6 +199,8 @@ def from_model(cls, taxon: Taxon) -> 'DbTaxon':
             id=taxon.id,
             ancestor_ids=_join_list(taxon.ancestor_ids),
             child_ids=_join_list(taxon.child_ids),
+            conservation_status=taxon.conservation_status,
+            establishment_means=taxon.establishment_means,
             iconic_taxon_id=taxon.iconic_taxon_id,
             is_active=taxon.is_active,
             leaf_taxa_count=taxon.complete_species_count,
@@ -206,6 +212,8 @@ def from_model(cls, taxon: Taxon) -> 'DbTaxon':
             rank=taxon.rank,
             reference_url=taxon.reference_url,
             photo_urls=photo_urls,
+            wikipedia_summary=taxon.wikipedia_summary,
+            wikipedia_url=taxon.wikipedia_url,
         )
 
     def to_model(self) -> Taxon:
@@ -214,6 +222,8 @@ def to_model(self) -> Taxon:
             id=self.id,
             ancestors=_get_taxa(self.ancestor_ids),
             children=_get_taxa(self.child_ids),
+            conservation_status=self.conservation_status,
+            establishment_means=self.establishment_means,
             default_photo=photos[0] if photos else None,
             iconic_taxon_id=self.iconic_taxon_id,
             is_active=self.is_active,
@@ -226,17 +236,23 @@ def to_model(self) -> Taxon:
             rank=self.rank,
             reference_url=self.reference_url,
             taxon_photos=photos,
+            wikipedia_summary=self.wikipedia_summary,
+            wikipedia_url=self.wikipedia_url,
         )
 
     def update(self, taxon: Taxon):
-        """Update an existing record"""
+        """Merge new values into an existing record"""
+        # Don't update a full taxon record with a partial one
+        if self.partial is False and taxon._partial is True:
+            return
+
         new_taxon = self.from_model(taxon)
         for col in [c.name for c in inspect(self).mapper.columns]:
-            if not (new_val := getattr(new_taxon, col)):
+            if (new_val := getattr(new_taxon, col)) is None:
                 continue
             if col not in PRECOMPUTED_COLUMNS:
                 setattr(self, col, new_val)
-            # Precomputed columns: Don't overwrite non-null values
+            # Precomputed columns: Only overwrite null values
             elif getattr(self, col) in [None, 0]:
                 setattr(self, col, new_val)
 

test/test_db.py

@@ -173,3 +173,28 @@ def test_save_taxa__preserve_precomputed_cols(tmp_path):
     assert results[0].observations_count == 18017625
     assert results[1].complete_species_count == 416
     assert results[1].observations_count == 218279
+
+
+def test_save_taxa__partial(tmp_path):
+    db_path = tmp_path / 'observations.db'
+    taxon_1 = Taxon(
+        id=3,
+        name='Aves',
+        rank='class',
+        preferred_common_name='Birds',
+        complete_species_count=10672,
+        observations_count=18017625,
+    )
+
+    create_tables(db_path)
+    save_taxa([taxon_1], db_path=db_path)
+
+    # Save with updated values for precomputed columns
+    taxon_1.preferred_common_name = 'updated!'
+    taxon_1._partial = True
+    save_taxa([taxon_1], db_path=db_path)
+
+    # Only previously null values (taxon_2.observations_count) in DB should be updated
+    results = list(get_db_taxa(db_path))
+    assert results[0].preferred_common_name == 'Birds'
+    assert results[0]._partial is False