feat: improve the lookup & usage of Taxon names (#696)

ami/main/admin.py

@@ -348,6 +348,10 @@ class TaxonAdmin(admin.ModelAdmin[Taxon]):
     )
     list_filter = ("lists", "rank", TaxonParentFilter)
     search_fields = ("name",)
+    autocomplete_fields = (
+        "parent",
+        "synonym_of",
+    )
 
     # annotate queryset with occurrence counts and allow sorting
     # https://docs.djangoproject.com/en/3.2/ref/contrib/admin/#django.contrib.admin.ModelAdmin.list_display

ami/main/api/views.py

@@ -1080,6 +1080,7 @@ def suggest(self, request):
                 taxa = (
                     Taxon.objects.select_related("parent", "parent__parent")
                     .annotate(similarity=TrigramSimilarity("name", query))
+                    .filter(active=True)
                     .order_by("-similarity")[:limit]
                 )
                 return Response(TaxonNestedSerializer(taxa, many=True, context={"request": request}).data)
@@ -1088,6 +1089,7 @@ def suggest(self, request):
                     Taxon.objects.filter(name__icontains=query)
                     .annotate(similarity=TrigramSimilarity("name", query))
                     .order_by("-similarity")[:default_results_limit]
+                    .filter(active=True)
                     .values("id", "name", "rank")[:limit]
                 )
                 return Response(TaxonSearchResultSerializer(taxa, many=True, context={"request": request}).data)

ami/main/models.py

@@ -2573,6 +2573,7 @@ class Taxon(BaseModel):
     gbif_taxon_key = models.BigIntegerField("GBIF taxon key", blank=True, null=True)
     bold_taxon_bin = models.CharField("BOLD taxon BIN", max_length=255, blank=True, null=True)
     inat_taxon_id = models.BigIntegerField("iNaturalist taxon ID", blank=True, null=True)
+    # lepsai_id = models.BigIntegerField("LepsAI / Fieldguide ID", blank=True, null=True)
 
     notes = models.TextField(blank=True)
 

ami/ml/models/algorithm.py

@@ -78,7 +78,12 @@ def with_taxa(self, category_field="label", only_indexes: list[int] | None = Non
             labels_data = self.data
             labels_label = self.labels
 
-        taxa = Taxon.objects.filter(models.Q(name__in=labels_label) | models.Q(search_names__overlap=labels_label))
+        # @TODO standardize species search / lookup.
+        # See similar query in ml.models.pipeline.get_or_create_taxon_for_classification()
+        taxa = Taxon.objects.filter(
+            models.Q(name__in=labels_label) | models.Q(search_names__overlap=labels_label),
+            active=True,
+        )
         taxon_map = {taxon.name: taxon for taxon in taxa}
 
         for category in labels_data:

ami/ml/models/pipeline.py

@@ -482,25 +482,28 @@ def get_or_create_taxon_for_classification(
 
     :return: The Taxon object
     """
-    taxa_list, _created = TaxaList.objects.get_or_create(
+    taxa_list, created = TaxaList.objects.get_or_create(
         name=f"Taxa returned by {algorithm.name}",
     )
-    if _created:
+    if created:
         logger.info(f"Created new taxa list {taxa_list}")
     else:
         logger.debug(f"Using existing taxa list {taxa_list}")
 
     # Get top label from classification scores
     assert algorithm.category_map, f"No category map found for algorithm {algorithm}"
     label_data: dict = algorithm.category_map.data[classification_resp.scores.index(max(classification_resp.scores))]
-    taxon, _created = Taxon.objects.get_or_create(
-        name=classification_resp.classification,
-        defaults={
-            "name": classification_resp.classification,
-            "rank": label_data.get("taxon_rank", TaxonRank.UNKNOWN),
-        },
-    )
-    if _created:
+    returned_taxon_name = classification_resp.classification
+    # @TOOD standardize the Taxon search / lookup. See similar query in ml.models.algorithm.AlgorithmCategoryMap
+    taxon = Taxon.objects.filter(
+        models.Q(name=returned_taxon_name) | models.Q(search_names__overlap=[returned_taxon_name]),
+        active=True,
+    ).first()
+    if not taxon:
+        taxon = Taxon.objects.create(
+            name=returned_taxon_name,
+            rank=label_data.get("taxon_rank", TaxonRank.UNKNOWN),
+        )
         logger.info(f"Registered new taxon {taxon}")
 
     taxa_list.taxa.add(taxon)