dev:minimal WoRMS compatibility: handle remote deletion

Author: grololo06

py/API_operations/TaxoManager.py

@@ -567,25 +567,28 @@ def pull_updates(self) -> Dict[str, Any]:
         nbr_rows = len(updates)
         nbr_updates = nbr_inserts = 0
 
-        to_rename: Dict[ClassifIDT, ClassifIDT] = {}
+        to_delete: List[ClassifIDT] = []
 
         for a_json_taxon in updates:
             # Convert non-str fields
             json_taxon_id = int(a_json_taxon["id"])
             lastupdate_datetime = datetime.datetime.strptime(
                 a_json_taxon["lastupdate_datetime"], "%Y-%m-%d %H:%M:%S"
             )
-            # Store rename intentions
-            if a_json_taxon["rename_to"]:
-                to_rename[json_taxon_id] = int(a_json_taxon["rename_to"])
+            must_delete = a_json_taxon["taxostatus"] == "X"
             # Read taxon from DB
             taxon = self.session.query(Taxonomy).get(json_taxon_id)
             if taxon is not None:
+                if must_delete:
+                    to_delete.append(json_taxon_id)
+                    continue
                 # The taxon is already present
                 if taxon.lastupdate_datetime == lastupdate_datetime:
                     continue  # already up to date
                 nbr_updates += 1
             else:
+                if must_delete:
+                    continue # Should not happen if timestamps are OK
                 # The taxon is not present, create it
                 nbr_inserts += 1
                 taxon = Taxonomy()
@@ -595,10 +598,9 @@ def pull_updates(self) -> Dict[str, Any]:
             for a_col in self.UpdatableCols:
                 setattr(taxon, a_col, a_json_taxon[a_col])
             taxon.lastupdate_datetime = lastupdate_datetime
-            self.session.commit()
-        # Manage rename_to
-        if len(to_rename) > 0:
-            TaxonomyBO.do_renames(self.session, to_rename)
+        if len(to_delete) > 0:
+            TaxonomyBO.do_deletes(self.session, to_delete)
+        self.session.commit()
 
         # if gvp('updatestat') == 'Y':
         #     msg = DoSyncStatUpdate()

py/BO/Taxonomy.py

@@ -419,6 +419,44 @@ def do_renames(session: Session, to_rename: Dict[ClassifIDT, ClassifIDT]):
         #     !!!! ECOPART LINK !!!!
         #     appli.part.prj.GlobalTaxoCompute()
 
+    @classmethod
+    def do_deletes(cls, session: Session, to_delete: List[ClassifIDT]) -> None:
+        #### Direct Foreign Keys to `taxonomy.id`
+        # 1.  **`obj_head.classif_id`**: Points to `taxonomy.id`. This represents the current classification of an object.
+        # 2.  **`objectsclassifhisto.classif_id`**: Points to `taxonomy.id` (with `ondelete="CASCADE"`). This stores the history of classifications for an object.
+        # 3.  **`taxo_change_log.to_id`**: Points to `taxonomy.id` (with `ondelete="CASCADE"`). This logs the target taxon in a mass classification change.
+        # 4.  **`taxo_change_log.from_id`**: Points to `taxonomy.id` (with `ondelete="CASCADE"`). This logs the original taxon in a mass classification change.
+        # 5.  **`prediction.classif_id`**: Points to `taxonomy.id` (with `ondelete="CASCADE"`). This stores the predicted taxon for an object.
+        # 6.  **`prediction_histo.classif_id`**: Points to `taxonomy.id` (with `ondelete="CASCADE"`). This stores the history of predicted taxa.
+        #### Other Implicit Relations
+        # *   **`taxonomy.parent_id`**: Although not explicitly defined with a `ForeignKey` constraint in the SQLAlchemy model (it is a simple `INTEGER` column), it logically points to `taxonomy.id` to represent the taxonomic tree structure.
+        # *   **`taxonomy.rename_to`**: Similarly, this `INTEGER` column is used to store an "advised" target taxon for mass category changes, logically referring to another `taxonomy.id`.
+        # *   **`taxo_recast.transforms`**: This `JSONB` column stores mapping in the form `{from:to}`, where both values are taxonomic IDs, though they are not enforced by database-level foreign key constraints.
+
+        # We want to protect 1. and 2.
+        logger.info("Taxo delete, list: %s", to_delete)
+        sql = text("SELECT DISTINCT objid FROM obj_head WHERE classif_id = ANY (:een)")
+        res: Result = session.execute(sql, {"een": list(to_delete)})
+        prevent_obj_head = {an_id for an_id, in res}
+        if len(prevent_obj_head) > 0:
+            logger.error("Unsafe deletion due to objects %s", prevent_obj_head)
+        sql = text("SELECT DISTINCT objid FROM objectsclassifhisto WHERE classif_id = ANY (:een)")
+        res: Result = session.execute(sql, {"een": list(to_delete)})
+        prevent_obj_histo = {an_id for an_id, in res}
+        if len(prevent_obj_histo) > 0:
+            logger.error("Unsafe deletion due to objects history %s", prevent_obj_histo)
+        # assert len(prevent_obj_head) == 0 and len(prevent_obj_histo) == 0, "Cannot achieve safe deletion"
+        logger.info("deleting categories")
+        session.execute(text("alter table taxonomy disable trigger all"))
+        try:
+            for a_taxon in to_delete:
+                logger.info("deleting category %s", a_taxon)
+                taxon = session.query(Taxonomy).get(a_taxon)
+                session.delete(taxon)
+        finally:
+            session.execute(text("alter table taxonomy enable trigger all"))
+
+
 
 class TaxonBOSet(object):
     """