fix(vector): harden packaged migrations

harness/tht/adapters/vector/pgvector.py

@@ -27,7 +27,7 @@ COLLECTION_KINDS = {
 }
 ALLOWED_COLLECTIONS = frozenset(COLLECTION_KINDS)
 ALLOWED_KINDS = frozenset().union(*COLLECTION_KINDS.values())
-_VECTOR_DIMENSION = re.compile(r"^vector\((\d+)\)$")
+_VECTOR_DIMENSION = re.compile(r"^(?:[a-z_][a-z0-9_]*\.)?vector\((\d+)\)$")
 
 
 def _collection(schema: str, name: str) -> sql.Identifier:
@@ -40,6 +40,14 @@ def _vector_literal(values: list[float]) -> str:
     return "[" + ",".join(str(float(value)) for value in values) + "]"
 
 
+def _vector_type(schema: str) -> sql.Identifier:
+    return sql.Identifier(schema, "vector")
+
+
+def _cosine_operator(schema: str) -> sql.Composed:
+    return sql.SQL("OPERATOR({}.<=>)").format(sql.Identifier(schema))
+
+
 def _validate_collection_kinds(collection: str, kinds: list[str]) -> None:
     invalid = set(kinds) - COLLECTION_KINDS[collection]
     if invalid:
@@ -248,9 +256,16 @@ class PgVectorStore:
                         continue
                     where = sql.SQL(" WHERE kind = ANY(%s)") if collection_kinds else sql.SQL("")
                     query = sql.SQL(
-                        "SELECT metadata, 1 - (embedding <=> %s::vector) AS similarity "
-                        "FROM {}{} ORDER BY embedding <=> %s::vector LIMIT %s"
-                    ).format(table, where)
+                        "SELECT metadata, 1 - (embedding {} %s::{}) AS similarity "
+                        "FROM {}{} ORDER BY embedding {} %s::{} LIMIT %s"
+                    ).format(
+                        _cosine_operator(self._schema),
+                        _vector_type(self._schema),
+                        table,
+                        where,
+                        _cosine_operator(self._schema),
+                        _vector_type(self._schema),
+                    )
                     params = [_vector_literal(embedding)]
                     if collection_kinds:
                         params.append(collection_kinds)
@@ -295,13 +310,13 @@ class PgVectorStore:
                 raise VectorStoreError("Embedding dimension does not match configured dimension")
         insert = sql.SQL(
             "INSERT INTO {} (record_key, kind, content_hash, metadata, embedding) "
-            "VALUES (%s, %s, %s, %s::jsonb, %s::vector) "
+            "VALUES (%s, %s, %s, %s::jsonb, %s::{}) "
             "ON CONFLICT (record_key) DO NOTHING"
-        ).format(table)
+        ).format(table, _vector_type(self._schema))
         update = sql.SQL(
             "UPDATE {} SET kind = %s, content_hash = %s, metadata = %s::jsonb, "
-            "embedding = %s::vector, indexed_at = now() WHERE record_key = %s"
-        ).format(table)
+            "embedding = %s::{}, indexed_at = pg_catalog.now() WHERE record_key = %s"
+        ).format(table, _vector_type(self._schema))
         raw = engine.raw_connection()
         try:
             with raw.cursor() as cursor: