Python: Do not execute Ibis expressions when ibis.options.interactive…

… is True (#5625)

Addresses #5499 by adding a custom inspector for Ibis expressions. This
is very basic, and per #5573 should perhaps live eventually in Ibis
itself.

Ibis is a bit unusual in that its interactive mode causes computation to
be executed when running the `__repr__` method, for nice interactivity
in the console and in Jupyter notebooks. So here we avoid running the
`__repr__` method so we don't accidentally fire off a BigQuery,
Snowflake, or other query which might have unwanted costs or side
effects.

There is a unit test -- Ibis with DuckDB is a minor dependency to pull
in relative to the rest of our test dependencies so I do not think this
is too onerous.

extensions/positron-python/python_files/positron/pinned-test-requirements.txt

@@ -5,6 +5,7 @@ geopandas==0.13.2; python_version < '3.9'
 geopandas==1.0.1; python_version >= '3.9'
 hvplot==0.10.0 ; python_version >= '3.9'
 hvplot==0.8.0 ; python_version < '3.9'
+ibis-framework[duckdb]==9.5.0; python_version >= '3.10'
 ipykernel==6.29.5
 ipywidgets==8.1.5
 matplotlib==3.9.2; python_version >= '3.9'

extensions/positron-python/python_files/positron/positron_ipykernel/inspectors.py

@@ -61,6 +61,12 @@
     except ImportError:
         pass
 
+    try:
+        import ibis  # python >= 3.10
+    except ImportError:
+        pass
+
+
 # General display settings
 TRUNCATE_AT: int = 1024
 PRINT_WIDTH: int = 100
@@ -90,9 +96,35 @@
 }
 
 
-def _get_class_display(value):
+def _remap_ibis_classnames(value):
+    # We will provide more nuanced handling of Ibis expressions in the
+    # inspector class for ibis.Expr and its many subclasses
+
+    import ibis
+
+    if isinstance(value, ibis.Expr):
+        return "ibis.Expr"
+
+    return get_qualname(value)
+
+
+PACKAGE_REMAPPERS = {
+    "ibis": _remap_ibis_classnames,
+}
+
+
+def _get_simplified_qualname(value):
     display_value = get_qualname(value)
-    return SIMPLER_NAMES.get(display_value, display_value)
+
+    if display_value in SIMPLER_NAMES:
+        return SIMPLER_NAMES[display_value]
+
+    top_path = display_value.split(".")[0]
+
+    if top_path in PACKAGE_REMAPPERS:
+        return PACKAGE_REMAPPERS[top_path](value)
+
+    return display_value
 
 
 class PositronInspector(Generic[T]):
@@ -805,7 +837,7 @@ def get_display_value(
         print_width: Optional[int] = PRINT_WIDTH,
         truncate_at: int = TRUNCATE_AT,
     ) -> Tuple[str, bool]:
-        display_value = _get_class_display(self.value)
+        display_value = _get_simplified_qualname(self.value)
         column_values = str(cast(Column, self.value[:100]).to_list())
         display_value = f"{display_value} {column_values}"
 
@@ -821,9 +853,10 @@ def get_size(self) -> int:
 
 
 class PandasSeriesInspector(BaseColumnInspector["pd.Series"]):
+    # Simplified names
     CLASS_QNAME = [
-        "pandas.core.series.Series",
-        "geopandas.geoseries.GeoSeries",
+        "pandas.Series",
+        "geopandas.GeoSeries",
     ]
 
     def get_display_name(self, key: int) -> str:
@@ -900,9 +933,9 @@ def to_plaintext(self) -> str:
 
 
 class PolarsSeriesInspector(BaseColumnInspector["pl.Series"]):
+    # Simplified class names
     CLASS_QNAME = [
-        "polars.series.series.Series",
-        "polars.internals.series.series.Series",
+        "polars.Series",
     ]
 
     def equals(self, value: pl.Series) -> bool:
@@ -956,7 +989,7 @@ def get_display_value(
         print_width: Optional[int] = PRINT_WIDTH,
         truncate_at: int = TRUNCATE_AT,
     ) -> Tuple[str, bool]:
-        display_value = _get_class_display(self.value)
+        display_value = _get_simplified_qualname(self.value)
         if hasattr(self.value, "shape"):
             shape = self.value.shape
             display_value = f"[{shape[0]} rows x {shape[1]} columns] {display_value}"
@@ -970,9 +1003,10 @@ def get_display_value(
 
 
 class PandasDataFrameInspector(BaseTableInspector["pd.DataFrame", "pd.Series"]):
+    # Simplified names
     CLASS_QNAME = [
-        "pandas.core.frame.DataFrame",
-        "geopandas.geodataframe.GeoDataFrame",
+        "pandas.DataFrame",
+        "geopandas.GeoDataFrame",
     ]
 
     def get_display_name(self, key: int) -> str:
@@ -1000,9 +1034,9 @@ def to_plaintext(self) -> str:
 
 
 class PolarsDataFrameInspector(BaseTableInspector["pl.DataFrame", "pl.Series"]):
+    # Simplified class name
     CLASS_QNAME = [
-        "polars.dataframe.frame.DataFrame",
-        "polars.internals.dataframe.frame.DataFrame",
+        "polars.DataFrame",
     ]
 
     def get_children(self):
@@ -1013,7 +1047,7 @@ def get_display_value(
         print_width: Optional[int] = PRINT_WIDTH,
         truncate_at: int = TRUNCATE_AT,
     ) -> Tuple[str, bool]:
-        qualname = _get_class_display(self.value)
+        qualname = _get_simplified_qualname(self.value)
         shape = self.value.shape
         display_value = f"[{shape[0]} rows x {shape[1]} columns] {qualname}"
         return (display_value, True)
@@ -1079,6 +1113,35 @@ def _is_active(self, value) -> bool:
         return True
 
 
+class IbisExprInspector(PositronInspector["ibis.Expr"]):
+    def has_children(self) -> bool:
+        return False
+
+    def get_length(self) -> int:
+        return 0
+
+    def is_mutable(self):
+        return False
+
+    def get_display_value(
+        self,
+        print_width: Optional[int] = PRINT_WIDTH,
+        truncate_at: int = TRUNCATE_AT,
+    ) -> Tuple[str, bool]:
+        # Just use the default object.__repr__ for now
+        simplified_name = get_qualname(self.value)
+        return (f"{simplified_name}", True)
+
+    def get_display_type(self) -> str:
+        return "ibis.Expr"
+
+    def to_html(self) -> str:
+        return self.get_display_value()[0]
+
+    def to_plaintext(self) -> str:
+        return self.get_display_value()[0]
+
+
 INSPECTOR_CLASSES: Dict[str, Type[PositronInspector]] = {
     **dict.fromkeys(PandasDataFrameInspector.CLASS_QNAME, PandasDataFrameInspector),
     **dict.fromkeys(PandasSeriesInspector.CLASS_QNAME, PandasSeriesInspector),
@@ -1092,6 +1155,7 @@ def _is_active(self, value) -> bool:
     DatetimeInspector.CLASS_QNAME: DatetimeInspector,
     **dict.fromkeys(SQLiteConnectionInspector.CLASS_QNAME, SQLiteConnectionInspector),
     **dict.fromkeys(SQLAlchemyEngineInspector.CLASS_QNAME, SQLAlchemyEngineInspector),
+    "ibis.Expr": IbisExprInspector,
     "boolean": BooleanInspector,
     "bytes": BytesInspector,
     "class": ClassInspector,
@@ -1117,7 +1181,7 @@ def get_inspector(value: T) -> PositronInspector[T]:
     elif isinstance(value, property):
         qualname = "property"
     else:
-        qualname = get_qualname(value)
+        qualname = _get_simplified_qualname(value)
     inspector_cls = INSPECTOR_CLASSES.get(qualname, None)
 
     if inspector_cls is None:

extensions/positron-python/python_files/positron/positron_ipykernel/tests/test_inspectors.py

@@ -8,6 +8,7 @@
 import pprint
 import random
 import string
+import sys
 import types
 from typing import Any, Callable, Iterable, Optional, Tuple
 
@@ -914,6 +915,46 @@ def test_get_child(value: Any, key: Any, expected: Any) -> None:
     assert get_inspector(child).equals(expected)
 
 
+@pytest.mark.skipif(sys.version_info < (3, 10), reason="requires Python 3.10 or higher")
+def test_inspect_ibis_exprs() -> None:
+    import ibis
+
+    # Make sure we don't return an executed repr
+    ibis.options.interactive = True
+
+    df = pd.DataFrame({"a": [1, 2, 1, 1, 2], "b": ["foo", "bar", "baz", "qux", None]})
+
+    t = ibis.memtable(df, name="df")
+    table_type = "ibis.expr.types.relations.Table"
+
+    verify_inspector(
+        value=t,
+        display_value=table_type,
+        kind=VariableKind.Other,
+        display_type=f"ibis.Expr",
+        type_info=get_type_as_str(t),
+        has_children=False,
+        is_truncated=True,
+        length=0,
+        mutable=False,
+    )
+
+    a_sum = t["a"].sum()  # type: ignore
+    int_type = "ibis.expr.types.numeric.IntegerScalar"
+
+    verify_inspector(
+        value=a_sum,
+        display_value=int_type,
+        kind=VariableKind.Other,
+        display_type=f"ibis.Expr",
+        type_info=get_type_as_str(a_sum),
+        has_children=False,
+        is_truncated=True,
+        length=0,
+        mutable=False,
+    )
+
+
 # TODO(wesm): these size values are only currently used for computing
 # comparison costs. We should align on # of cells vs. # of bytes for
 # these comparisons (possibly based on more experiments)

extensions/positron-python/python_files/positron/test-requirements.txt

@@ -2,6 +2,7 @@ bokeh
 fastcore
 geopandas
 holoviews
+ibis-framework[duckdb]; python_version >= '3.10'
 ipykernel
 ipywidgets
 matplotlib