feat(perf): avoid materializing the input table if given a duckdb-backed ibis.Table

pyproject.toml

@@ -23,6 +23,7 @@ dev-dependencies = [
     "ruff>=0.6.2",
     "polars>=1.5.0",
     "vega-datasets>=0.9.0",
+    "ibis-framework[duckdb]>=6; python_version >= '3.9'",
 ]
 
 [tool.ruff.lint]

src/quak/_util.py

@@ -2,11 +2,13 @@
 
 import sys
 import typing
+import uuid
 
 DataFrameObject = typing.Any
 
 if typing.TYPE_CHECKING:
     import duckdb
+    import ibis
     import polars as pl
     import pyarrow as pa
 
@@ -104,3 +106,17 @@ def is_polars(obj: object) -> typing.TypeGuard[pl.DataFrame]:
     """Check if an object is a Polars DataFrame."""
     polars = sys.modules.get("polars")
     return polars is not None and isinstance(obj, polars.DataFrame)
+
+
+def is_ibis_duckdb(obj: object) -> typing.TypeGuard[ibis.Table]:
+    """Check if an object is an ibis Table backed by DuckDB."""
+    ibis = sys.modules.get("ibis")
+    if ibis is None or not isinstance(obj, ibis.Table):
+        return False
+    backend = obj.get_backend()
+    return backend.name == "duckdb"
+
+
+def gen_unique_name(prefix: str = "quak") -> str:
+    """Generate a unique name with a given prefix."""
+    return f"{prefix}_{uuid.uuid4().hex[:8]}"

src/quak/_widget.py

@@ -12,10 +12,12 @@
 from ._util import (
     arrow_table_from_dataframe_protocol,
     arrow_table_from_ipc,
+    gen_unique_name,
     get_columns,
     has_pycapsule_stream_interface,
     is_arrow_ipc,
     is_dataframe_api_obj,
+    is_ibis_duckdb,
     is_polars,
     table_to_ipc,
 )
@@ -42,6 +44,17 @@ class Widget(anywidget.AnyWidget):
     def __init__(self, data, *, table: str = "df"):
         if isinstance(data, duckdb.DuckDBPyConnection):
             conn = data
+        elif is_ibis_duckdb(data):
+            backend = data.get_backend()
+            conn = backend.con
+            table = gen_unique_name()
+            # if the given table is a memtable, need to register it so that the
+            # following backend.compile(data) references a table that actually exists.
+            backend._register_in_memory_tables(data)
+            # create a view, not a table, so that if you have an eg postgres
+            # database attached to duckdb, you can query it without
+            # materializing the entire table from postgres into duckdb.
+            backend.raw_sql(f"CREATE TEMP VIEW {table} AS ({backend.compile(data)})")
         else:
             conn = duckdb.connect(":memory:")
             if is_polars(data):