feat: Remove support for multiple MLT ids (#64)

Author: isaacvando

paradedb/sqlalchemy/search.py

@@ -5,7 +5,7 @@
 from collections.abc import Sequence
 from typing import Any
 
-from sqlalchemy import Text, cast, func, literal, literal_column, or_
+from sqlalchemy import Text, cast, func, literal, literal_column
 from sqlalchemy.dialects.postgresql import ARRAY, array
 from sqlalchemy.sql import operators
 from sqlalchemy.sql.elements import ClauseElement, ColumnElement
@@ -386,7 +386,6 @@ def more_like_this(
     field: ColumnElement,
     *,
     document_id: Any | None = None,
-    document_ids: list[Any] | None = None,
     document: dict[str, Any] | str | None = None,
     fields: list[str] | None = None,
     min_term_frequency: int | None = None,
@@ -400,17 +399,13 @@ def more_like_this(
 ) -> ColumnElement[bool]:
     error_cls = InvalidMoreLikeThisOptionsError
 
-    sources_provided = sum(x is not None for x in (document_id, document_ids, document))
+    sources_provided = sum(x is not None for x in (document_id, document))
     if sources_provided != 1:
-        raise error_cls("exactly one of document_id, document_ids, or document must be provided")
-    if document_ids is not None and len(document_ids) == 0:
-        raise error_cls("document_ids must not be empty")
+        raise error_cls("exactly one of document_id, or document must be provided")
     if document is not None and fields is not None:
-        raise error_cls("fields can only be used with document_id or document_ids")
+        raise error_cls("fields can only be used with document_id")
     if fields is not None:
         require_non_empty_sequence(fields, field_name="fields", error_cls=error_cls)
-    if document_ids is not None and any(doc_id is None for doc_id in document_ids):
-        raise error_cls("document_ids entries cannot be null")
     if isinstance(document, dict) and not document:
         raise error_cls("document must not be empty")
 
@@ -469,12 +464,6 @@ def _build_mlt_call(source_arg: ClauseElement, *, include_fields: bool) -> Claus
             positional_args.append(_text_array(fields))
         return PDBFunctionWithNamedArgs("more_like_this", positional_args, named_options)
 
-    if document_ids is not None:
-        clauses = [
-            field.operate(_QUERY, _build_mlt_call(literal(doc_id), include_fields=True)) for doc_id in document_ids
-        ]
-        return or_(*clauses)
-
     if document_id is not None:
         return field.operate(_QUERY, _build_mlt_call(literal(document_id), include_fields=True))
 

tests/integration/test_advanced_search_integration.py

@@ -78,32 +78,6 @@ def test_more_like_this_rejects_invalid_numeric_options():
         search.more_like_this(Product.id, document_id=1, boost_factor=-1.0)
 
 
-def test_more_like_this_by_document_ids(session):
-    """document_ids ORs results from multiple individual MLT queries."""
-    stmt_combined = (
-        select(Product.id)
-        .where(search.more_like_this(Product.id, document_ids=[1, 3], fields=["description"]))
-        .order_by(Product.id)
-    )
-    stmt_id1 = (
-        select(Product.id)
-        .where(search.more_like_this(Product.id, document_id=1, fields=["description"]))
-        .order_by(Product.id)
-    )
-    stmt_id3 = (
-        select(Product.id)
-        .where(search.more_like_this(Product.id, document_id=3, fields=["description"]))
-        .order_by(Product.id)
-    )
-    assert_uses_paradedb_scan(session, stmt_combined)
-    ids_combined = set(session.scalars(stmt_combined))
-    ids_1 = set(session.scalars(stmt_id1))
-    ids_3 = set(session.scalars(stmt_id3))
-    # Combined should be union of individual results
-    assert ids_1.issubset(ids_combined)
-    assert ids_3.issubset(ids_combined)
-
-
 def test_proximity(session):
     prox = search.prox_str("sleek").within(3, "shoes")
     stmt = select(Product.id).where(search.proximity(Product.description, prox)).order_by(Product.id)

tests/integration/test_paradedb_queries_integration.py

@@ -418,22 +418,6 @@ def test_more_like_this_by_document_id(mock_session):
     assert ids == MLT_RUNNING_SHOES_IDS
 
 
-def test_more_like_this_by_document_ids(mock_session):
-    """MLT with multiple document_ids ORs the results together."""
-    stmt_combined = select(MockItem.id).where(
-        search.more_like_this(MockItem.id, document_ids=[3, 12], fields=["description"])
-    )
-    stmt_id1 = select(MockItem.id).where(search.more_like_this(MockItem.id, document_id=3, fields=["description"]))
-    stmt_id2 = select(MockItem.id).where(search.more_like_this(MockItem.id, document_id=12, fields=["description"]))
-    assert_uses_paradedb_scan(mock_session, stmt_combined, index_name="mock_items_bm25_idx")
-    ids_combined = _ids(mock_session, stmt_combined)
-    ids_1 = _ids(mock_session, stmt_id1)
-    ids_2 = _ids(mock_session, stmt_id2)
-    assert ids_1 == MLT_RUNNING_SHOES_IDS
-    assert ids_2 == WIRELESS_IDS
-    assert ids_combined == MLT_COMBINED_IDS
-
-
 def test_more_like_this_by_document_payload(mock_session):
     """MLT with document dict finds similar documents."""
     stmt = select(MockItem.id).where(
@@ -476,12 +460,6 @@ def test_more_like_this_with_min_word_length(mock_session):
     assert ids_filtered == RUNNING_IDS
 
 
-def test_more_like_this_document_ids_empty_raises(mock_session):
-    """document_ids=[] raises an error."""
-    with pytest.raises(InvalidMoreLikeThisOptionsError, match="document_ids must not be empty"):
-        search.more_like_this(MockItem.id, document_ids=[])
-
-
 def test_more_like_this_multiple_sources_raises(mock_session):
     """Providing more than one input source raises an error."""
     with pytest.raises(InvalidMoreLikeThisOptionsError):