feat(duckdb): use delta_scan instead of reading pyarrow datasets (#…

…9566)

ibis/backends/duckdb/__init__.py

@@ -885,23 +885,20 @@ def read_in_memory(
         return self.table(table_name)
 
     def read_delta(
-        self,
-        source_table: str,
-        table_name: str | None = None,
-        **kwargs: Any,
+        self, source_table: str, table_name: str | None = None, **kwargs: Any
     ) -> ir.Table:
         """Register a Delta Lake table as a table in the current database.
 
         Parameters
         ----------
         source_table
-            The data source. Must be a directory
-            containing a Delta Lake table.
+            The data source. Must be a directory containing a Delta Lake table.
         table_name
-            An optional name to use for the created table. This defaults to
-            a sequentially generated name.
-        **kwargs
-            Additional keyword arguments passed to deltalake.DeltaTable.
+            An optional name to use for the created table. This defaults to a
+            generated name.
+        kwargs
+            Additional keyword arguments passed to the `delta` extension's
+            `delta_scan` function.
 
         Returns
         -------
@@ -913,21 +910,25 @@ def read_delta(
 
         table_name = table_name or util.gen_name("read_delta")
 
-        try:
-            from deltalake import DeltaTable
-        except ImportError:
-            raise ImportError(
-                "The deltalake extra is required to use the "
-                "read_delta method. You can install it using pip:\n\n"
-                "pip install 'ibis-framework[deltalake]'\n"
-            )
+        # always try to load the delta extension
+        extensions = ["delta"]
+
+        # delta handles s3 itself, not with httpfs
+        if source_table.startswith(("http://", "https://")):
+            extensions.append("httpfs")
 
-        delta_table = DeltaTable(source_table, **kwargs)
+        self._load_extensions(extensions)
 
-        return self.read_in_memory(
-            delta_table.to_pyarrow_dataset(), table_name=table_name
+        options = [
+            sg.to_identifier(key).eq(sge.convert(val)) for key, val in kwargs.items()
+        ]
+        self._create_temp_view(
+            table_name,
+            sg.select(STAR).from_(self.compiler.f.delta_scan(source_table, *options)),
         )
 
+        return self.table(table_name)
+
     def list_tables(
         self,
         like: str | None = None,

ibis/backends/tests/test_export.py

@@ -405,7 +405,7 @@ def test_to_pyarrow_decimal(backend, dtype, pyarrow_dtype):
 )
 @pytest.mark.notyet(["clickhouse"], raises=Exception)
 @pytest.mark.notyet(["mssql", "pandas"], raises=PyDeltaTableError)
-def test_roundtrip_delta(backend, con, alltypes, tmp_path, monkeypatch):
+def test_roundtrip_delta(backend, con, alltypes, tmp_path):
     if con.name == "pyspark":
         pytest.importorskip("delta")
     else:
@@ -416,11 +416,10 @@ def test_roundtrip_delta(backend, con, alltypes, tmp_path, monkeypatch):
     path = tmp_path / "test.delta"
     t.to_delta(path)
 
-    monkeypatch.setattr(ibis.options, "default_backend", con)
-    dt = ibis.read_delta(path)
-    result = dt.to_pandas()
+    dt = con.read_delta(path)
+    result = con.to_pandas(dt)
 
-    backend.assert_frame_equal(result, expected)
+    backend.assert_frame_equal(result, expected, check_dtype=con.name != "duckdb")
 
 
 @pytest.mark.notimpl(