Merge pull request #4790 from FBruzzesi/plotly-with-narwhals

feat: make plotly-express dataframe agnostic via narwhals

Author: emilykl

CHANGELOG.md

@@ -8,8 +8,9 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 
 ### Updated
 
-- Updated plotly.py to use base64 encoding of arrays in plotly JSON to improve performance. 
+- Updated plotly.py to use base64 encoding of arrays in plotly JSON to improve performance.
 - Add `subtitle` attribute to all Plotly Express traces
+- Make plotly-express dataframe agnostic via Narwhals [#4790](https://github.com/plotly/plotly.py/pull/4790)
 
 ## [5.24.1] - 2024-09-12
 

packages/python/plotly/_plotly_utils/basevalidators.py

@@ -8,6 +8,7 @@
 import re
 import sys
 import warnings
+import narwhals.stable.v1 as nw
 
 from _plotly_utils.optional_imports import get_module
 
@@ -72,8 +73,6 @@ def copy_to_readonly_numpy_array(v, kind=None, force_numeric=False):
     """
     np = get_module("numpy")
 
-    # Don't force pandas to be loaded, we only want to know if it's already loaded
-    pd = get_module("pandas", should_load=False)
     assert np is not None
 
     # ### Process kind ###
@@ -93,34 +92,26 @@ def copy_to_readonly_numpy_array(v, kind=None, force_numeric=False):
         "O": "object",
     }
 
-    # Handle pandas Series and Index objects
-    if pd and isinstance(v, (pd.Series, pd.Index)):
-        if v.dtype.kind in numeric_kinds:
-            # Get the numeric numpy array so we use fast path below
-            v = v.values
-        elif v.dtype.kind == "M":
-            # Convert datetime Series/Index to numpy array of datetimes
-            if isinstance(v, pd.Series):
-                with warnings.catch_warnings():
-                    warnings.simplefilter("ignore", FutureWarning)
-                    # Series.dt.to_pydatetime will return Index[object]
-                    # https://github.com/pandas-dev/pandas/pull/52459
-                    v = np.array(v.dt.to_pydatetime())
-            else:
-                # DatetimeIndex
-                v = v.to_pydatetime()
-    elif pd and isinstance(v, pd.DataFrame) and len(set(v.dtypes)) == 1:
-        dtype = v.dtypes.tolist()[0]
-        if dtype.kind in numeric_kinds:
-            v = v.values
-        elif dtype.kind == "M":
-            with warnings.catch_warnings():
-                warnings.simplefilter("ignore", FutureWarning)
-                # Series.dt.to_pydatetime will return Index[object]
-                # https://github.com/pandas-dev/pandas/pull/52459
-                v = [
-                    np.array(row.dt.to_pydatetime()).tolist() for i, row in v.iterrows()
-                ]
+    # With `pass_through=True`, the original object will be returned if unable to convert
+    # to a Narwhals DataFrame or Series.
+    v = nw.from_native(v, allow_series=True, pass_through=True)
+
+    if isinstance(v, nw.Series):
+        if v.dtype == nw.Datetime and v.dtype.time_zone is not None:
+            # Remove time zone so that local time is displayed
+            v = v.dt.replace_time_zone(None).to_numpy()
+        else:
+            v = v.to_numpy()
+    elif isinstance(v, nw.DataFrame):
+        schema = v.schema
+        overrides = {}
+        for key, val in schema.items():
+            if val == nw.Datetime and val.time_zone is not None:
+                # Remove time zone so that local time is displayed
+                overrides[key] = nw.col(key).dt.replace_time_zone(None)
+        if overrides:
+            v = v.with_columns(**overrides)
+        v = v.to_numpy()
 
     if not isinstance(v, np.ndarray):
         # v has its own logic on how to convert itself into a numpy array
@@ -193,6 +184,7 @@ def is_homogeneous_array(v):
         np
         and isinstance(v, np.ndarray)
         or (pd and isinstance(v, (pd.Series, pd.Index)))
+        or (isinstance(v, nw.Series))
     ):
         return True
     if is_numpy_convertable(v):

packages/python/plotly/_plotly_utils/tests/validators/test_pandas_series_input.py

@@ -73,12 +73,13 @@ def color_categorical_pandas(request, pandas_type):
 def dates_array(request):
     return np.array(
         [
-            datetime(year=2013, month=10, day=10),
-            datetime(year=2013, month=11, day=10),
-            datetime(year=2013, month=12, day=10),
-            datetime(year=2014, month=1, day=10),
-            datetime(year=2014, month=2, day=10),
-        ]
+            "2013-10-10",
+            "2013-11-10",
+            "2013-12-10",
+            "2014-01-10",
+            "2014-02-10",
+        ],
+        dtype="datetime64[ns]",
     )
 
 
@@ -183,7 +184,7 @@ def test_data_array_validator_dates_series(
     assert isinstance(res, np.ndarray)
 
     # Check dtype
-    assert res.dtype == "object"
+    assert res.dtype == "<M8[ns]"
 
     # Check values
     np.testing.assert_array_equal(res, dates_array)
@@ -200,7 +201,7 @@ def test_data_array_validator_dates_dataframe(
     assert isinstance(res, np.ndarray)
 
     # Check dtype
-    assert res.dtype == "object"
+    assert res.dtype == "<M8[ns]"
 
     # Check values
     np.testing.assert_array_equal(res, dates_array.reshape(len(dates_array), 1))

packages/python/plotly/optional-requirements.txt

@@ -39,6 +39,7 @@ ipython
 
 ## pandas deps for some matplotlib functionality ##
 pandas
+narwhals>=1.13.3
 
 ## scipy deps for some FigureFactory functions ##
 scipy