Handle decimals and date types (#53)

* handle decimals

* lint

* Cast to timestamp

Author: ahuang11

src/panel_gwalker/_gwalker.py

@@ -35,6 +35,7 @@
     SPECTYPES,
     SpecType,
     _raw_fields,
+    cast_to_supported_dtypes,
     configure_debug_log_level,
     logger,
     process_spec,
@@ -432,6 +433,10 @@ def _compute(self, payload):
             result = pd.DataFrame()
 
         df = pd.DataFrame.from_records(result)
+
+        # Convert any Decimal objects to float
+        df = cast_to_supported_dtypes(df)
+
         logger.debug("response:\n%s", df)
         return {col: df[col].values for col in df.columns}
 

src/panel_gwalker/_tabular_data.py

@@ -15,6 +15,8 @@
 from narwhals.typing import FrameT, IntoFrame
 from panel.io.datamodel import PARAM_MAPPING
 
+from ._utils import cast_to_supported_dtypes
+
 TabularDataType = IntoFrame
 
 
@@ -45,7 +47,7 @@ def _validate(self, val):
 def _column_datasource_from_tabular_df(data: FrameT):
     if isinstance(data, nw.LazyFrame):
         data = data.collect()
-    data = data.to_pandas()
+    data = cast_to_supported_dtypes(data.to_pandas())
     return ColumnDataSource._data_from_df(data)
 
 

src/panel_gwalker/_utils.py

@@ -1,8 +1,11 @@
+import datetime
+import decimal
 import json
 import logging
 import os
 import sys
 from pathlib import Path
+from typing import Any, Union
 
 import narwhals as nw
 import pandas as pd
@@ -11,6 +14,35 @@
 from narwhals.dataframe import LazyFrame
 from narwhals.typing import FrameT
 
+
+def cast_to_supported_dtypes(df: pd.DataFrame, sample: int = 100) -> pd.DataFrame:
+    """
+    Convert decimal.Decimal to float in a pandas DataFrame, as
+    Bokeh ColumnDataSource does not support decimal.Decimal.
+    Samples only a subset of the DataFrame to check for decimal.Decimal
+
+    Arguments
+    ---------
+    df (pd.DataFrame):
+      the DataFrame to convert
+    sample (int):
+      number of rows to sample to check for decimal.Decimal
+    """
+    df = df.copy()
+    for col in df.select_dtypes(include=["object"]).columns:
+        df_col_sample = df[col].sample(min(sample, len(df)))
+        try:
+            if df_col_sample.apply(lambda x: isinstance(x, decimal.Decimal)).any():
+                df[col] = pd.to_numeric(df[col], errors="coerce")
+            if df_col_sample.apply(
+                lambda x: isinstance(x, (datetime.datetime, datetime.date))
+            ).any():
+                df[col] = pd.to_datetime(df[col], errors="coerce")
+        except Exception:
+            df[col] = df[col].astype(str)
+    return df
+
+
 logger = logging.getLogger("panel-graphic-walker")
 FORMAT = "%(asctime)s | %(levelname)s | %(name)s | %(message)s"
 from narwhals.typing import FrameT
@@ -90,7 +122,7 @@ def _raw_fields(data: FrameT) -> list[dict]:
         except Exception as ex:
             pass
 
-    pandas_data = data.to_pandas()
+    pandas_data = cast_to_supported_dtypes(data.to_pandas())
     return _raw_fields_core(pandas_data)
 
 

tests/test_tabular_data.py

@@ -1,7 +1,13 @@
+import datetime
+import decimal
+
+import numpy as np
+import pandas as pd
 import param
 import pytest
 
 from panel_gwalker._tabular_data import TabularData, _column_datasource_from_tabular_df
+from panel_gwalker._utils import cast_to_supported_dtypes
 
 
 class MyClass(param.Parameterized):
@@ -20,3 +26,38 @@ def test_tabular_data_raises():
 
 def test_column_datasource_from_tabular_df(data):
     assert _column_datasource_from_tabular_df(data)
+
+
+def test_decimal_conversion():
+    df = pd.DataFrame(
+        {
+            "price": [decimal.Decimal("10.50"), decimal.Decimal("25.75")],
+            "qty": [5, 10],
+            "name": ["Item A", "Item B"],
+        }
+    )
+
+    converted_df = cast_to_supported_dtypes(df)
+
+    assert isinstance(converted_df["price"][0], float)
+    assert not isinstance(converted_df["price"][0], decimal.Decimal)
+    assert converted_df["price"][0] == 10.5
+
+
+def test_date_conversion():
+    df = pd.DataFrame(
+        {
+            "date": pd.to_datetime(
+                [
+                    datetime.date(2020, 1, 1) + datetime.timedelta(days=i)
+                    for i in range(3)
+                ]
+            ),
+            "value": np.random.randn(3).cumsum(),
+        }
+    )
+    converted_df = cast_to_supported_dtypes(df)
+    print(df)
+
+    assert isinstance(converted_df["date"][0], pd.Timestamp)
+    assert converted_df["date"][0] == pd.Timestamp("2020-01-01")