Support "segment" cursor style

Author: mdesmet

tests/integration/test_dbapi_integration.py

@@ -1838,6 +1838,26 @@ def test_select_query_spooled_segments(trino_connection):
     assert len(rows) > 0
 
 
+@pytest.mark.skipif(
+    trino_version() <= '464',
+    reason="spooled protocol was introduced in version 464"
+)
+def test_segments_cursor(trino_connection):
+    if trino_connection._client_session.encoding is None:
+        with pytest.raises(ValueError, match=".*encoding.*"):
+            trino_connection.cursor("segment")
+        return
+    cur = trino_connection.cursor("segment")
+    cur.execute("""SELECT l.*
+    FROM tpch.tiny.lineitem l, TABLE(sequence(
+        start => 1,
+        stop => 5,
+        step => 1)) n""")
+    rows = cur.fetchall()
+    # TODO: improve test
+    assert len(rows) > 0
+
+
 def get_cursor(legacy_prepared_statements, run_trino):
     host, port = run_trino
 

trino/client.py

@@ -63,7 +63,16 @@
 from trino._version import __version__
 from trino.mapper import RowMapper, RowMapperFactory
 
-__all__ = ["ClientSession", "TrinoQuery", "TrinoRequest", "PROXIES"]
+__all__ = [
+    "ClientSession",
+    "TrinoQuery",
+    "TrinoRequest",
+    "PROXIES",
+    "SpooledData",
+    "SpooledSegment",
+    "InlineSegment",
+    "Segment"
+]
 
 logger = trino.logging.get_logger(__name__)
 
@@ -753,6 +762,7 @@ def __init__(
             request: TrinoRequest,
             query: str,
             legacy_primitive_types: bool = False,
+            fetch_mode: Literal["mapped", "segments"] = "mapped"
     ) -> None:
         self._query_id: Optional[str] = None
         self._stats: Dict[Any, Any] = {}
@@ -769,6 +779,7 @@ def __init__(
         self._result: Optional[TrinoResult] = None
         self._legacy_primitive_types = legacy_primitive_types
         self._row_mapper: Optional[RowMapper] = None
+        self._fetch_mode = fetch_mode
 
     @property
     def query_id(self) -> Optional[str]:
@@ -869,10 +880,12 @@ def fetch(self) -> List[Union[List[Any]], Any]:
             return []
 
         rows = status.rows
-        if isinstance(rows, dict):
+        if isinstance(status.rows, dict):
             # spooled protocol
             rows = cast(_SpooledProtocolResponseTO, rows)
             segments = self._to_segments(rows)
+            if self._fetch_mode == "segments":
+                return segments
             return list(SegmentIterator(segments))
         else:
             return self._row_mapper.map(rows)

trino/dbapi.py

@@ -251,7 +251,7 @@ def _create_request(self):
             self.request_timeout,
         )
 
-    def cursor(self, legacy_primitive_types: bool = None):
+    def cursor(self, cursor_style: str = "row", legacy_primitive_types: bool = None):
         """Return a new :py:class:`Cursor` object using the connection."""
         if self.isolation_level != IsolationLevel.AUTOCOMMIT:
             if self.transaction is None:
@@ -260,11 +260,21 @@ def cursor(self, legacy_primitive_types: bool = None):
             request = self.transaction.request
         else:
             request = self._create_request()
-        return Cursor(
+
+        cursor_class = {
+            # Add any custom Cursor classes here
+            "segment": SegmentCursor,
+            "row": Cursor
+        }.get(cursor_style.lower(), Cursor)
+
+        return cursor_class(
             self,
             request,
-            # if legacy params are not explicitly set in Cursor, take them from Connection
-            legacy_primitive_types if legacy_primitive_types is not None else self.legacy_primitive_types
+            legacy_primitive_types=(
+                legacy_primitive_types
+                if legacy_primitive_types is not None
+                else self.legacy_primitive_types
+            )
         )
 
     def _use_legacy_prepared_statements(self):
@@ -697,6 +707,28 @@ def close(self):
         #  but also any other outstanding queries executed through this cursor.
 
 
+class SegmentCursor(Cursor):
+    def __init__(
+            self,
+            connection,
+            request,
+            legacy_primitive_types: bool = False):
+        super().__init__(connection, request, legacy_primitive_types=legacy_primitive_types)
+        if self.connection._client_session.encoding is None:
+            raise ValueError("SegmentCursor can only be used if encoding is set on the connection")
+
+    def execute(self, operation, params=None):
+        if params:
+            # TODO: refactor code to allow for params to be supported
+            raise ValueError("params not supported")
+
+        self._query = trino.client.TrinoQuery(self._request, query=operation,
+                                              legacy_primitive_types=self._legacy_primitive_types,
+                                              fetch_mode="segments")
+        self._iterator = iter(self._query.execute())
+        return self
+
+
 Date = datetime.date
 Time = datetime.time
 Timestamp = datetime.datetime