fix(datatypes): return pd.Timestamp or pd.Series[datetime64] for date.to_pandas()

docs/contribute/02_workflow.qmd

@@ -274,3 +274,11 @@ you are going only up).
 ```bash
 $ colima delete
 ```
+
+### `x86_64` or `amd64` based containers
+
+While starting the containers based on `x86_64` / `amd64`, the architecture flag needs to be set in two places:
+1. Add `platform: linux/amd64` for the service in `compose.yaml`.
+2. Set the `--arch` flag while starting the VM `colima start --arch x86_64`
+
+For instance, this step is necessary for the `oracle` service in `compose.yaml`. Otherwise, the container will fail shortly after getting started.

ibis/backends/oracle/converter.py

@@ -1,19 +1,19 @@
 from __future__ import annotations
 
-import datetime
+import pandas as pd
 
 from ibis.formats.pandas import PandasData
 
 
 class OraclePandasData(PandasData):
     @classmethod
     def convert_Timestamp_element(cls, dtype):
-        return datetime.datetime.fromisoformat
+        return pd.Timestamp.fromisoformat
 
     @classmethod
     def convert_Date_element(cls, dtype):
-        return datetime.date.fromisoformat
+        return pd.Timestamp.fromisoformat
 
     @classmethod
     def convert_Time_element(cls, dtype):
-        return datetime.time.fromisoformat
+        return pd.Timestamp.fromisoformat

ibis/backends/snowflake/converter.py

@@ -1,9 +1,9 @@
 from __future__ import annotations
 
-import datetime
 import json
 from typing import TYPE_CHECKING
 
+import pandas as pd
 import pyarrow as pa
 
 from ibis.formats.pandas import PandasData
@@ -52,15 +52,15 @@
 class SnowflakePandasData(PandasData):
     @classmethod
     def convert_Timestamp_element(cls, dtype):
-        return datetime.datetime.fromisoformat
+        return pd.Timestamp.fromisoformat
 
     @classmethod
     def convert_Date_element(cls, dtype):
-        return datetime.date.fromisoformat
+        return pd.Timestamp.fromisoformat
 
     @classmethod
     def convert_Time_element(cls, dtype):
-        return datetime.time.fromisoformat
+        return pd.Timestamp.fromisoformat
 
     @classmethod
     def convert_JSON(cls, s, dtype, pandas_type):

ibis/backends/sqlite/tests/test_types.py

@@ -92,6 +92,8 @@ def test_type_map(db):
     sol = pd.DataFrame(
         {"str_col": ["a"], "date_col": pd.Series([date(2022, 1, 1)], dtype="object")}
     )
+    sol["date_col"] = sol["date_col"].astype(res["date_col"].dtype)
+
     assert res.equals(sol)
 
 

ibis/backends/tests/test_aggregation.py

@@ -1251,7 +1251,7 @@ def test_string_quantile(alltypes, func):
 )
 def test_date_quantile(alltypes):
     expr = alltypes.timestamp_col.date().quantile(0.5)
-    result = expr.execute()
+    result = expr.execute().to_pydatetime().date()
     assert result == date(2009, 12, 31)
 
 

ibis/backends/tests/test_array.py

@@ -22,11 +22,10 @@
     GoogleBadRequest,
     MySQLOperationalError,
     PolarsComputeError,
-    PsycoPg2ArraySubscriptError,
     PsycoPg2IndeterminateDatatype,
     PsycoPg2InternalError,
     PsycoPg2ProgrammingError,
-    PsycoPg2SyntaxError,
+    PsycoPgInvalidTextRepresentation,
     PsycoPgSyntaxError,
     Py4JJavaError,
     PyAthenaDatabaseError,
@@ -1118,7 +1117,7 @@ def test_unnest_struct(con):
 
 
 @builtin_array
-@pytest.mark.notimpl(["postgres"], raises=PsycoPg2SyntaxError)
+@pytest.mark.notimpl(["postgres"], raises=PsycoPgSyntaxError)
 @pytest.mark.notimpl(["risingwave"], raises=PsycoPg2InternalError)
 @pytest.mark.notimpl(
     ["trino"], reason="inserting maps into structs doesn't work", raises=TrinoUserError
@@ -1209,7 +1208,7 @@ def test_zip_null(con, fn):
 
 
 @builtin_array
-@pytest.mark.notimpl(["postgres"], raises=PsycoPg2SyntaxError)
+@pytest.mark.notimpl(["postgres"], raises=PsycoPgSyntaxError)
 @pytest.mark.notimpl(["risingwave"], raises=PsycoPg2ProgrammingError)
 @pytest.mark.notimpl(["datafusion"], raises=Exception, reason="not yet supported")
 @pytest.mark.notimpl(
@@ -1769,7 +1768,7 @@ def test_table_unnest_column_expr(backend):
 @pytest.mark.notimpl(["datafusion", "polars"], raises=com.OperationNotDefinedError)
 @pytest.mark.notimpl(["trino"], raises=TrinoUserError)
 @pytest.mark.notimpl(["athena"], raises=PyAthenaOperationalError)
-@pytest.mark.notimpl(["postgres"], raises=PsycoPg2SyntaxError)
+@pytest.mark.notimpl(["postgres"], raises=PsycoPgSyntaxError)
 @pytest.mark.notimpl(["risingwave"], raises=PsycoPg2ProgrammingError)
 @pytest.mark.notyet(
     ["risingwave"], raises=PsycoPg2InternalError, reason="not supported in risingwave"
@@ -1890,7 +1889,7 @@ def test_array_agg_bool(con, data, agg, baseline_func):
 
 @pytest.mark.notyet(
     ["postgres"],
-    raises=PsycoPg2ArraySubscriptError,
+    raises=PsycoPgInvalidTextRepresentation,
     reason="all dimensions must match in size",
 )
 @pytest.mark.notimpl(["risingwave", "flink"], raises=com.OperationNotDefinedError)

ibis/backends/tests/test_client.py

@@ -32,7 +32,7 @@
     ImpalaHiveServer2Error,
     OracleDatabaseError,
     PsycoPg2InternalError,
-    PsycoPg2UndefinedObject,
+    PsycoPgUndefinedObject,
     Py4JJavaError,
     PyAthenaDatabaseError,
     PyODBCProgrammingError,
@@ -725,7 +725,7 @@ def test_list_database_contents(con):
 @pytest.mark.notyet(["databricks"], raises=DatabricksServerOperationError)
 @pytest.mark.notyet(["bigquery"], raises=com.UnsupportedBackendType)
 @pytest.mark.notyet(
-    ["postgres"], raises=PsycoPg2UndefinedObject, reason="no unsigned int types"
+    ["postgres"], raises=PsycoPgUndefinedObject, reason="no unsigned int types"
 )
 @pytest.mark.notyet(
     ["oracle"], raises=OracleDatabaseError, reason="no unsigned int types"

ibis/backends/tests/test_generic.py

@@ -25,7 +25,7 @@
     OracleDatabaseError,
     PolarsInvalidOperationError,
     PsycoPg2InternalError,
-    PsycoPg2SyntaxError,
+    PsycoPgSyntaxError,
     Py4JJavaError,
     PyAthenaDatabaseError,
     PyAthenaOperationalError,
@@ -1736,7 +1736,7 @@ def hash_256(col):
                 pytest.mark.notimpl(["flink"], raises=Py4JJavaError),
                 pytest.mark.notimpl(["druid"], raises=PyDruidProgrammingError),
                 pytest.mark.notimpl(["oracle"], raises=OracleDatabaseError),
-                pytest.mark.notimpl(["postgres"], raises=PsycoPg2SyntaxError),
+                pytest.mark.notimpl(["postgres"], raises=PsycoPgSyntaxError),
                 pytest.mark.notimpl(["risingwave"], raises=PsycoPg2InternalError),
                 pytest.mark.notimpl(["snowflake"], raises=AssertionError),
                 pytest.mark.never(

ibis/backends/tests/test_struct.py

@@ -13,7 +13,7 @@
     DatabricksServerOperationError,
     PolarsColumnNotFoundError,
     PsycoPg2InternalError,
-    PsycoPg2SyntaxError,
+    PsycoPgSyntaxError,
     Py4JJavaError,
     PyAthenaDatabaseError,
     PyAthenaOperationalError,
@@ -138,7 +138,7 @@ def test_collect_into_struct(alltypes):
 
 
 @pytest.mark.notimpl(
-    ["postgres"], reason="struct literals not implemented", raises=PsycoPg2SyntaxError
+    ["postgres"], reason="struct literals not implemented", raises=PsycoPgSyntaxError
 )
 @pytest.mark.notimpl(
     ["risingwave"],
@@ -155,7 +155,7 @@ def test_field_access_after_case(con):
 
 
 @pytest.mark.notimpl(
-    ["postgres"], reason="struct literals not implemented", raises=PsycoPg2SyntaxError
+    ["postgres"], reason="struct literals not implemented", raises=PsycoPgSyntaxError
 )
 @pytest.mark.notimpl(["flink"], raises=IbisError, reason="not implemented in ibis")
 @pytest.mark.parametrize(
@@ -242,7 +242,7 @@ def test_keyword_fields(con, nullable):
 
 @pytest.mark.notyet(
     ["postgres"],
-    raises=PsycoPg2SyntaxError,
+    raises=PsycoPgSyntaxError,
     reason="sqlglot doesn't implement structs for postgres correctly",
 )
 @pytest.mark.notyet(

ibis/backends/tests/test_temporal.py

@@ -650,9 +650,7 @@ def convert_to_offset(x):
             "ignore", category=(UserWarning, pd.errors.PerformanceWarning)
         )
         expected = (
-            pd.to_datetime(df.date_string_col)
-            .add(offset)
-            .map(lambda ts: ts.normalize().date(), na_action="ignore")
+            pd.to_datetime(df.date_string_col).add(offset).astype("datetime64[s]")
         )
 
     expected = backend.default_series_rename(expected)
@@ -727,12 +725,7 @@ def convert_to_offset(x):
         ),
         param(
             lambda t, _: t.timestamp_col.date() + ibis.interval(days=4),
-            lambda t, _: (
-                t.timestamp_col.dt.floor("d")
-                .add(pd.Timedelta(days=4))
-                .dt.normalize()
-                .dt.date
-            ),
+            lambda t, _: t.timestamp_col.dt.floor("d").add(pd.Timedelta(days=4)),
             id="date-add-interval",
             marks=[
                 pytest.mark.notimpl(
@@ -743,12 +736,7 @@ def convert_to_offset(x):
         ),
         param(
             lambda t, _: t.timestamp_col.date() - ibis.interval(days=14),
-            lambda t, _: (
-                t.timestamp_col.dt.floor("d")
-                .sub(pd.Timedelta(days=14))
-                .dt.normalize()
-                .dt.date
-            ),
+            lambda t, _: t.timestamp_col.dt.floor("d").sub(pd.Timedelta(days=14)),
             id="date-subtract-interval",
             marks=[
                 pytest.mark.notimpl(
@@ -1013,14 +1001,15 @@ def test_interval_add_cast_column(backend, alltypes, df):
     delta = alltypes.bigint_col.cast("interval('D')")
     expr = alltypes.select("id", (timestamp_date + delta).name("tmp"))
     result = expr.execute().sort_values("id").reset_index().tmp
+
     df = df.sort_values("id").reset_index(drop=True)
     expected = (
         df["timestamp_col"]
         .dt.normalize()
         .add(df.bigint_col.astype("timedelta64[D]"))
         .rename("tmp")
-        .dt.date
     )
+
     backend.assert_series_equal(result, expected.astype(result.dtype))
 
 
@@ -2296,20 +2285,18 @@ def test_time_literal_sql(dialect, snapshot, micros):
 )
 def test_date_scalar(con, value, func):
     expr = ibis.date(func(value)).name("tmp")
-
     result = con.execute(expr)
 
-    assert not isinstance(result, datetime.datetime)
-    assert isinstance(result, datetime.date)
-
-    assert result == datetime.date.fromisoformat(value)
+    assert isinstance(result, pd.Timestamp)
+    assert result == pd.Timestamp.fromisoformat(value)
 
 
 @pytest.mark.notyet(
     ["datafusion", "druid", "exasol"], raises=com.OperationNotDefinedError
 )
 def test_simple_unix_date_offset(con):
-    d = ibis.date("2023-04-07")
+    s = "2023-04-07"
+    d = ibis.date(s)
     expr = d.epoch_days()
     result = con.execute(expr)
     delta = datetime.date(2023, 4, 7) - datetime.date(1970, 1, 1)

ibis/formats/pandas.py

@@ -222,17 +222,20 @@
     def convert_Date(cls, s, dtype, pandas_type):
         if isinstance(s.dtype, pd.DatetimeTZDtype):
             s = s.dt.tz_convert("UTC").dt.tz_localize(None)
+
         try:
-            return s.astype(pandas_type).dt.date
+            return s.astype(pandas_type)
         except (ValueError, TypeError, pd._libs.tslibs.OutOfBoundsDatetime):
 
             def try_date(v):
-                if isinstance(v, datetime.datetime):
-                    return v.date()
+                if isinstance(v, datetime.date):
+                    return pd.Timestamp(v)
                 elif isinstance(v, str):
                     if v.endswith("Z"):
-                        return datetime.datetime.fromisoformat(v[:-1]).date()
-                    return datetime.date.fromisoformat(v)
+                        datetime_obj = datetime.datetime.fromisoformat(v[:-1])
+                    else:
+                        datetime_obj = datetime.datetime.fromisoformat(v)
+                    return pd.Timestamp(datetime_obj)
                 else:
                     return v