Fix px scrambling categorical axis order when data is split into traces

Author: vineoy

CHANGELOG.md

@@ -7,6 +7,7 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 ### Fixed
 - Fix concurrent first access to lazily initialized graph object properties, which could raise `ValueError("Invalid value")` [[#3441](https://github.com/plotly/plotly.py/issues/3441)], with thanks to @hb1915 for the contribution!
 - Fix `px.sunburst`, `px.treemap` and `px.icicle` listing sectors in a different order on every run when `path` is used with a Polars DataFrame; sectors now follow their order of first appearance for all dataframe backends [[#5765](https://github.com/plotly/plotly.py/issues/5765)], with thanks to @Irahan2 for the contribution!
+- Fix Plotly Express scrambling the order of categorical axes when the data is split into several traces (e.g. by `color`); the axis now follows the order of first appearance in the underlying data [[#3198](https://github.com/plotly/plotly.py/issues/3198), [#3291](https://github.com/plotly/plotly.py/issues/3291)]
 
 
 ## [7.1.0] - 2026-09-15

plotly/express/_core.py

@@ -2529,6 +2529,36 @@ def get_groups_and_orders(args, grouper):
     return groups, orders
 
 
+def _default_category_orders_for_axes(args, groups, orders):
+    """
+    Add data-appearance orders for categorical x/y columns.
+
+    When the data is split into several traces (e.g. by `color`), plotly.js
+    orders categories by first appearance across traces, which can differ from
+    the order in the underlying data. Recording the data-appearance order here
+    lets `set_cartesian_axis_opts` emit an explicit `categoryarray`, matching
+    the documented behavior that Plotly Express lays out categorical data in
+    the order in which it appears in the data. See #3198.
+    """
+    if len(groups) <= 1:
+        return orders
+    df = args["data_frame"]
+    for letter in ("x", "y"):
+        col = args.get(letter)
+        if not isinstance(col, str) or col in orders or col not in df.columns:
+            continue
+        uniques = df.get_column(col).unique(maintain_order=True).to_list()
+        categories = [
+            value
+            for value in uniques
+            if value is not None
+            and not (isinstance(value, float) and math.isnan(value))
+        ]
+        if categories and all(isinstance(value, str) for value in categories):
+            orders[col] = categories
+    return orders
+
+
 def make_figure(args, constructor, trace_patch=None, layout_patch=None):
     trace_patch = trace_patch or {}
     layout_patch = layout_patch or {}
@@ -2555,6 +2585,7 @@ def make_figure(args, constructor, trace_patch=None, layout_patch=None):
     )
     grouper = [x.grouper or one_group for x in grouped_mappings] or [one_group]
     groups, orders = get_groups_and_orders(args, grouper)
+    orders = _default_category_orders_for_axes(args, groups, orders)
 
     col_labels = []
     row_labels = []

tests/test_optional/test_px/test_px.py

@@ -361,6 +361,38 @@ def test_category_order_with_category_as_x(backend):
         assert set(trace["x"]) == {"Thur", "Fri", "Sat", "Sun"}
 
 
+def test_category_order_defaults_to_data_order_with_color():
+    # https://github.com/plotly/plotly.py/issues/3198
+    # Splitting into one trace per color must not change the axis order:
+    # it should follow first appearance in the underlying data.
+    fig = px.line(
+        x=["A", "B", None, "B", "C", None, "C", "D", None, "D", "E", None],
+        y=[1, 1, None, 2, 2, None, 3, 3, None, 4, 4, None],
+        color=[1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 1, 1],
+    )
+    assert fig["layout"]["xaxis"]["categoryorder"] == "array"
+    assert fig["layout"]["xaxis"]["categoryarray"] == ("A", "B", "C", "D", "E")
+
+
+def test_category_order_defaults_to_data_order_with_color_bar():
+    # https://github.com/plotly/plotly.py/issues/3291
+    df = pd.DataFrame(
+        {"x": ["a", "b", "c", "d"], "y": [1, 2, 3, 4], "z": ["p", "q", "p", "q"]}
+    )
+    fig = px.bar(df, x="x", y="y", color="z")
+    assert fig["layout"]["xaxis"]["categoryorder"] == "array"
+    assert fig["layout"]["xaxis"]["categoryarray"] == ("a", "b", "c", "d")
+
+
+def test_no_default_category_array_for_single_trace_or_numeric():
+    # A single trace already follows data order, and numeric axes
+    # must never get a category array.
+    fig = px.line(x=["A", "B", "C"], y=[1, 2, 3])
+    assert fig["layout"]["xaxis"]["categoryorder"] is None
+    fig = px.line(x=[1, 2, 3], y=[1, 2, 3], color=["a", "b", "a"])
+    assert fig["layout"]["xaxis"]["categoryorder"] is None
+
+
 def test_permissive_defaults():
     msg = "'PxDefaults' object has no attribute 'should_not_work'"
     with pytest.raises(AttributeError, match=msg):