Merge branch 'master' into conda-update

Author: emilykl

CHANGELOG.md

@@ -32,7 +32,7 @@ This project adheres to [Semantic Versioning](http://semver.org/).
   - Drop deprecated `autotick` attributes from cartesian axes [[#7236](https://github.com/plotly/plotly.js/pull/7236)]
   - Drop `transforms` from the API [[#7240](https://github.com/plotly/plotly.js/pull/7240), [#7254](https://github.com/plotly/plotly.js/pull/7254)]
 
-## Fixed
+### Fixed
 - Fix a bug in JupyterLab >= 4 and Jupyter Notebook >= 7 that caused LaTeX to not render in plotly charts [[#4763](https://github.com/plotly/plotly.py/pull/4763)].
 - Fix `go.FigureWidget.show` to return `FigureWidget` instead of displaying `Figure` [[#4869](https://github.com/plotly/plotly.py/pull/4869)]
 

packages/python/plotly/_plotly_utils/utils.py

@@ -43,8 +43,10 @@ def to_typed_array_spec(v):
     """
     v = copy_to_readonly_numpy_array(v)
 
+    # Skip b64 encoding if numpy is not installed,
+    # or if v is not a numpy array, or if v is empty
     np = get_module("numpy", should_load=False)
-    if not np or not isinstance(v, np.ndarray):
+    if not np or not isinstance(v, np.ndarray) or v.size == 0:
         return v
 
     dtype = str(v.dtype)

packages/python/plotly/plotly/express/_core.py

@@ -1168,16 +1168,13 @@ def _escape_col_name(columns, col_name, extra):
     return col_name
 
 
-def to_unindexed_series(x, name=None, native_namespace=None):
-    """Assuming x is list-like or even an existing Series, returns a new Series (with
-    its index reset if pandas-like). Stripping the index from existing pd.Series is
-    required to get things to match up right in the new DataFrame we're building.
-    """
+def to_named_series(x, name=None, native_namespace=None):
+    """Assuming x is list-like or even an existing Series, returns a new Series named `name`."""
     # With `pass_through=True`, the original object will be returned if unable to convert
     # to a Narwhals Series.
     x = nw.from_native(x, series_only=True, pass_through=True)
     if isinstance(x, nw.Series):
-        return nw.maybe_reset_index(x).rename(name)
+        return x.rename(name)
     elif native_namespace is not None:
         return nw.new_series(name=name, values=x, native_namespace=native_namespace)
     else:
@@ -1306,7 +1303,7 @@ def process_args_into_dataframe(
                                 length,
                             )
                         )
-                    df_output[col_name] = to_unindexed_series(
+                    df_output[col_name] = to_named_series(
                         real_argument, col_name, native_namespace
                     )
                 elif not df_provided:
@@ -1343,7 +1340,7 @@ def process_args_into_dataframe(
                     )
                 else:
                     col_name = str(argument)
-                    df_output[col_name] = to_unindexed_series(
+                    df_output[col_name] = to_named_series(
                         df_input.get_column(argument), col_name
                     )
             # ----------------- argument is likely a column / array / list.... -------
@@ -1362,7 +1359,7 @@ def process_args_into_dataframe(
                             argument.name is not None
                             and argument.name in df_input.columns
                             and (
-                                to_unindexed_series(
+                                to_named_series(
                                     argument, argument.name, native_namespace
                                 )
                                 == df_input.get_column(argument.name)
@@ -1380,7 +1377,7 @@ def process_args_into_dataframe(
                         % (field, len_arg, str(list(df_output.keys())), length)
                     )
 
-                df_output[str(col_name)] = to_unindexed_series(
+                df_output[str(col_name)] = to_named_series(
                     x=argument,
                     name=str(col_name),
                     native_namespace=native_namespace,
@@ -2121,16 +2118,21 @@ def process_dataframe_timeline(args):
     if args["x_start"] is None or args["x_end"] is None:
         raise ValueError("Both x_start and x_end are required")
 
-    try:
-        df: nw.DataFrame = args["data_frame"]
-        df = df.with_columns(
-            nw.col(args["x_start"]).str.to_datetime().alias(args["x_start"]),
-            nw.col(args["x_end"]).str.to_datetime().alias(args["x_end"]),
-        )
-    except Exception:
-        raise TypeError(
-            "Both x_start and x_end must refer to data convertible to datetimes."
-        )
+    df: nw.DataFrame = args["data_frame"]
+    schema = df.schema
+    to_convert_to_datetime = [
+        col
+        for col in [args["x_start"], args["x_end"]]
+        if schema[col] != nw.Datetime and schema[col] != nw.Date
+    ]
+
+    if to_convert_to_datetime:
+        try:
+            df = df.with_columns(nw.col(to_convert_to_datetime).str.to_datetime())
+        except Exception as exc:
+            raise TypeError(
+                "Both x_start and x_end must refer to data convertible to datetimes."
+            ) from exc
 
     # note that we are not adding any columns to the data frame here, so no risk of overwrite
     args["data_frame"] = df.with_columns(

packages/python/plotly/plotly/tests/test_io/test_to_from_json.py

@@ -2,6 +2,7 @@
 import plotly.io as pio
 import pytest
 import plotly
+import numpy as np
 import json
 import os
 import tempfile
@@ -259,3 +260,16 @@ def test_write_json_from_file_string(fig1, pretty, remove_uids):
         # Check contents that were written
         expected = pio.to_json(fig1, pretty=pretty, remove_uids=remove_uids)
         assert result == expected
+
+
+def test_to_dict_empty_np_array_int64():
+    fig = go.Figure(
+        [
+            go.Bar(
+                x=np.array([], dtype="str"),
+                y=np.array([], dtype="int64"),
+            )
+        ]
+    )
+    # to_dict() should not raise an exception
+    fig.to_dict()

packages/python/plotly/plotly/tests/test_optional/test_px/test_px.py

@@ -360,3 +360,18 @@ def test_render_mode(backend):
     )
     assert fig.data[0].type == "histogram2dcontour"
     assert fig.data[1].type == "scatter"
+
+
+def test_empty_df_int64(backend):
+    # Load px data, then filter it such that the dataframe is empty
+    df = px.data.tips(return_type=backend)
+    df = nw.from_native(px.data.tips(return_type=backend))
+    df_empty = df.filter(nw.col("day") == "banana").to_native()
+
+    fig = px.scatter(
+        df_empty,
+        x="total_bill",
+        y="size",  # size is an int64 column
+    )
+    # to_dict() should not raise an exception
+    fig.to_dict()

packages/python/plotly/plotly/tests/test_optional/test_px/test_px_functions.py

@@ -561,3 +561,29 @@ def test_timeline(constructor):
     msg = "Both x_start and x_end must refer to data convertible to datetimes."
     with pytest.raises(TypeError, match=msg):
         px.timeline(df, x_start="Start", x_end=["a", "b", "c"], y="Task", color="Task")
+
+
+@pytest.mark.parametrize(
+    "datetime_columns",
+    [
+        ["Start"],
+        ["Start", "Finish"],
+        ["Finish"],
+    ],
+)
+def test_timeline_cols_already_temporal(constructor, datetime_columns):
+    # https://github.com/plotly/plotly.py/issues/4913
+    data = {
+        "Task": ["Job A", "Job B", "Job C"],
+        "Start": ["2009-01-01", "2009-03-05", "2009-02-20"],
+        "Finish": ["2009-02-28", "2009-04-15", "2009-05-30"],
+    }
+    df = (
+        nw.from_native(constructor(data))
+        .with_columns(nw.col(datetime_columns).str.to_datetime(format="%Y-%m-%d"))
+        .to_native()
+    )
+    fig = px.timeline(df, x_start="Start", x_end="Finish", y="Task", color="Task")
+    assert len(fig.data) == 3
+    assert fig.layout.xaxis.type == "date"
+    assert fig.layout.xaxis.title.text is None

packages/python/plotly/plotly/tests/test_optional/test_px/test_px_input.py

@@ -435,6 +435,29 @@ def test_splom_case(backend):
     assert np.all(fig.data[0].dimensions[0].values == ar[:, 0])
 
 
+def test_scatter_matrix_indexed_pandas():
+    # https://github.com/plotly/plotly.py/issues/4917
+    # https://github.com/plotly/plotly.py/issues/4788
+    df = pd.DataFrame(
+        {
+            "x": [1, 2, 3, 4],
+            "y": [10, 20, 10, 20],
+            "z": [-1, -2, -3, -4],
+            "color": [1, 2, 3, 4],
+        }
+    )
+    df.index = pd.DatetimeIndex(
+        [
+            "1/1/2020 10:00:00+00:00",
+            "2/1/2020 11:00:00+00:00",
+            "3/1/2020 10:00:00+00:00",
+            "4/1/2020 11:00:00+00:00",
+        ]
+    )
+    fig = px.scatter_matrix(df, color="color")
+    assert np.all(fig.data[0].marker["color"] == np.array([1, 2, 3, 4]))
+
+
 def test_int_col_names(constructor):
     # DataFrame with int column names
     lengths = constructor({"0": np.random.random(100)})

packages/python/plotly/requires-install.txt

@@ -6,5 +6,5 @@
 ###################################################
 
 ## dataframe agnostic layer ##
-narwhals>=1.13.3
+narwhals>=1.15.1
 packaging

packages/python/plotly/test_requirements/requirements_310_core.txt

@@ -1,3 +1,3 @@
 requests==2.25.1
 pytest==7.4.4
-narwhals>=1.13.3
+narwhals>=1.15.1

packages/python/plotly/test_requirements/requirements_310_optional.txt

@@ -20,5 +20,5 @@ kaleido
 orjson==3.8.12
 polars[timezone]
 pyarrow
-narwhals>=1.13.3
+narwhals>=1.15.1
 anywidget==0.9.13