Fix multi-database entity resolution capture validation

src/viano_bug_analysis/analysis/entity_resolution.py

@@ -436,8 +436,16 @@ class EntityResolver:
         captures=[]
         for s,ref in zip(plan['searches'],search['capture_refs'],strict=True):
             c=read_json(reader,ref,'project-capture',self.owner.limits.max_context_bytes)
-            if (ref != self.base(context)+'/captures/'+hashlib.sha256(reader.read_bytes(ref)).hexdigest()+'.json' or c['origin'] != initial['binding']
-                    or c['query'] != query_for(s,initial['policy'])):
+            query=query_for(s,initial['policy'])
+            # Reconstruct the query-specific origin from the frozen binding, so
+            # completed cases remain verifiable without a live database provider.
+            binding=copy.deepcopy(initial['binding'])
+            if binding.get('kind') == 'mariadb' and (binding.get('additional_databases') or '.' in query['table']):
+                parts=query['table'].split('.')
+                database,table=(binding['database'],parts[0]) if len(parts)==1 else parts
+                binding.update(selected_database=database,selected_table=table)
+            if (ref != self.base(context)+'/captures/'+hashlib.sha256(reader.read_bytes(ref)).hexdigest()+'.json' or c['origin'] != binding
+                    or c['query'] != query):
                 raise InterpretationError('ENTITY_RESOLUTION_CAPTURE_CHANGED')
             captures.append(c)
         if search['candidates'] != candidates_for(plan,captures,initial['policy']):

tests/test_entity_resolution.py

@@ -126,6 +126,39 @@ def context_of(store):
         return build_context(json.loads(reader.read_bytes(ref)),ref,InterpretationLimits())
 
 
+@pytest.mark.parametrize('qualified', [False, True])
+@pytest.mark.parametrize('wrong_origin', [False, True])
+def test_multidatabase_capture_origin_and_portable_replay(built_case, tmp_path, qualified, wrong_origin):
+    class MultiDatabase(Database):
+        def binding(self):
+            return {'kind':'mariadb', 'database':'world', 'snapshot_id':self.snapshot,
+                    'tables':['components'], 'additional_databases':{'characters':['components']}}
+
+        def binding_for(self, query):
+            database, table = query['table'].split('.') if '.' in query['table'] else ('world', query['table'])
+            return {**self.binding(), 'selected_database':'wrong' if wrong_origin else database,
+                    'selected_table':table}
+
+        def validate(self, query):
+            assert query['table'] in {'components','characters.components'}
+
+    store=built_case[0]
+    model=Model(); service=service_for(store,MultiDatabase())
+    if qualified:
+        service.entity_resolution['catalogs'][0]['table']='characters.components'
+    if wrong_origin:
+        with pytest.raises(InterpretationError, match='ENTITY_RESOLUTION_CAPTURE_CHANGED'):
+            run(store,model,service)
+        return
+    result=run(store,model,service)
+    assert result.status=='COMPLETED'
+    assert PackageValidator().validate(result.package).valid
+    receiving=CheckpointStore(type(store.workspace)(tmp_path/'portable-multidb'))
+    receiving.intake(result.package)
+    again=run(receiving,model)
+    assert again.reused and len(model.requests)==3
+
+
 def test_resolves_unrecorded_subject_preserves_case_and_portable_replay(built_case, tmp_path):
     store, _, built, _, _ = built_case
     with PackageReader(built.package) as reader: