plot3 0.6.1__tar.gz → 0.6.2__tar.gz

{plot3-0.6.1 → plot3-0.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plot3
-Version: 0.6.1
+Version: 0.6.2
 Summary: ggplot2 grammar of graphics for Python: interactive WebGL figures and journal-ready PNG, SVG, and PDF
 Author: Rigoberto Leyva Salmeron
 License-Expression: MIT
@@ -224,18 +224,27 @@ ggplot2's names and defaults.
 
 ```python
 rng = np.random.default_rng(1)
-trial = pd.DataFrame({"arm": rng.choice(["placebo", "low", "high"], 150),
+arms = ["placebo", "low", "high"]
+trial = pd.DataFrame({"arm": pd.Categorical(rng.choice(arms, 150), categories=arms),
                       "dose": rng.uniform(0, 10, 150)})
-trial["response"] = 2 + 0.6 * trial.dose + 1.5 * (trial.arm == "high") + rng.normal(0, 1.2, 150)
+slope = trial.arm.map({"placebo": 0.05, "low": 0.35, "high": 0.7}).astype(float)
+trial["response"] = 2 + slope * trial.dose + rng.normal(0, 0.9, 150)
 
 a = (ggplot(trial, aes(x="dose", y="response", colour="arm"))
-     + geom_point(alpha=0.7) + geom_smooth(method="lm"))
+     + geom_point(alpha=0.7) + geom_smooth(method="lm", se=False)
+     + scale_colour_hue() + theme_grey() + theme(legend_position="bottom"))
 b = (ggplot(trial, aes(x="arm", y="response", fill="arm"))
-     + geom_violin() + geom_jitter(width=0.1, height=0) + theme(legend_position="none"))
-fig = (a | b) + plot_annotation(title="Response by dose and arm", tag_levels="A")
+     + geom_violin() + geom_jitter(width=0.1, height=0)
+     + scale_fill_hue() + theme_grey() + theme(legend_position="none"))
+fig = ((a | b) + plot_layout(widths=[3, 2])
+       + plot_annotation(title="Response by dose and arm", tag_levels="A"))
 ```
 
-<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_compose.png" alt="Two-panel figure: scatter with linear fits, and violins by arm">
+`scale_colour_hue()` and `theme_grey()` are ggplot2's own defaults, so the
+figure looks as it would from R. plot3's defaults are `theme_bw()` for saved
+files and `theme_dark()` in the viewer.
+
+<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_look.png" alt="Two-panel figure in ggplot2's grey theme: response against dose with one fitted line per arm, and violins of response by arm">
 
 Everything else you'd expect is there, including histograms, densities,
 boxplots, bar positions, `stat_summary`, error bars, heatmaps, hexbins, 2D

{plot3-0.6.1 → plot3-0.6.2}/README.md

@@ -191,18 +191,27 @@ ggplot2's names and defaults.
 
 ```python
 rng = np.random.default_rng(1)
-trial = pd.DataFrame({"arm": rng.choice(["placebo", "low", "high"], 150),
+arms = ["placebo", "low", "high"]
+trial = pd.DataFrame({"arm": pd.Categorical(rng.choice(arms, 150), categories=arms),
                       "dose": rng.uniform(0, 10, 150)})
-trial["response"] = 2 + 0.6 * trial.dose + 1.5 * (trial.arm == "high") + rng.normal(0, 1.2, 150)
+slope = trial.arm.map({"placebo": 0.05, "low": 0.35, "high": 0.7}).astype(float)
+trial["response"] = 2 + slope * trial.dose + rng.normal(0, 0.9, 150)
 
 a = (ggplot(trial, aes(x="dose", y="response", colour="arm"))
-     + geom_point(alpha=0.7) + geom_smooth(method="lm"))
+     + geom_point(alpha=0.7) + geom_smooth(method="lm", se=False)
+     + scale_colour_hue() + theme_grey() + theme(legend_position="bottom"))
 b = (ggplot(trial, aes(x="arm", y="response", fill="arm"))
-     + geom_violin() + geom_jitter(width=0.1, height=0) + theme(legend_position="none"))
-fig = (a | b) + plot_annotation(title="Response by dose and arm", tag_levels="A")
+     + geom_violin() + geom_jitter(width=0.1, height=0)
+     + scale_fill_hue() + theme_grey() + theme(legend_position="none"))
+fig = ((a | b) + plot_layout(widths=[3, 2])
+       + plot_annotation(title="Response by dose and arm", tag_levels="A"))
 ```
 
-<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_compose.png" alt="Two-panel figure: scatter with linear fits, and violins by arm">
+`scale_colour_hue()` and `theme_grey()` are ggplot2's own defaults, so the
+figure looks as it would from R. plot3's defaults are `theme_bw()` for saved
+files and `theme_dark()` in the viewer.
+
+<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_look.png" alt="Two-panel figure in ggplot2's grey theme: response against dose with one fitted line per arm, and violins of response by arm">
 
 Everything else you'd expect is there, including histograms, densities,
 boxplots, bar positions, `stat_summary`, error bars, heatmaps, hexbins, 2D

plot3-0.6.2/plot3/__version__.py

@@ -0,0 +1 @@
+__version__ = "0.6.2"

{plot3-0.6.1 → plot3-0.6.2}/plot3/build.py

@@ -762,7 +762,7 @@ def expand_stat_geom(
                 if len(group_cols) > 1:
                     out_row[colour_col] = key_tuple[1]
                 outlier_rows.append(out_row)
-        frame = pd.DataFrame(rows)
+        frame = _keep_categories(pd.DataFrame(rows), data, [colour_col])
         box_width = float(getattr(geom, "width", 0.75))
         dodge_levels = None
         if len(group_cols) > 1 and not frame.empty:
@@ -920,7 +920,7 @@ def expand_stat_geom(
                 elif colour_col == xcol:
                     row[colour_col] = x_key
                 rows.append(row)
-        frame = pd.DataFrame(rows)
+        frame = _keep_categories(pd.DataFrame(rows), data, [colour_col])
         if frame.empty:
             frame = pd.DataFrame(columns=["x", "y", "group"])
         # Force categorical x scale labels via a helper frame for domains:
@@ -1423,6 +1423,48 @@ def _legend_position_spec(value):
 
 
 
+def _categorical_order(g, resolved, levels) -> list:
+    """Colour levels in the order of a categorical colour column, as ggplot2.
+
+    Stats (density, histogram, smooth, summaries) rebuild their data, so
+    their levels arrive sorted as text. When the plot's colour column is a
+    ``pd.Categorical``, its categories set the order (and so the colours);
+    levels it does not list keep their place after them.
+    """
+    data = getattr(g, "data", None)
+    if data is None:
+        return list(levels)
+    columns = dict.fromkeys(m.get("color") for _geom, m in resolved if isinstance(m.get("color"), str))
+    for column in columns:
+        if not has_column(data, column):
+            continue
+        source = materialize_columns(data, [column])[column]
+        if not isinstance(source.dtype, pd.CategoricalDtype):
+            continue
+        wanted = [str(c) for c in source.cat.categories]
+        present = [str(level) for level in levels]
+        first = [c for c in wanted if c in present]
+        return first + [level for level in present if level not in first]
+    return list(levels)
+
+
+def _keep_categories(frame: pd.DataFrame, data, columns) -> pd.DataFrame:
+    """Give a stat's output columns the category order of the source column.
+
+    Box and violin stats build their output from plain rows, which turns
+    ``pd.Categorical(..., categories=["placebo", "low", "high"])`` into
+    text, sorted alphabetically: the fill colours then disagree with the
+    points of the same column in another panel or layer.
+    """
+    for column in dict.fromkeys(c for c in columns if c):
+        if column not in frame.columns or not has_column(data, column):
+            continue
+        source = materialize_columns(data, [column])[column]
+        if isinstance(source.dtype, pd.CategoricalDtype):
+            frame[column] = pd.Categorical(frame[column], categories=list(source.cat.categories))
+    return frame
+
+
 def _forced_levels(g, ccats) -> list:
     """Colour levels shared by every facet panel, so a group keeps its colour."""
     forced = getattr(g, "_force_color_levels", None)
@@ -2826,6 +2868,7 @@ def build_spec(g: ggplot) -> tuple[dict, list[tuple[str, str]]]:
     # viridis_d scale, else the theme's colours extended past eight groups.
     cat_colours: list[str] = []
     if color_scale is not None and color_scale[0] == "cat":
+        color_scale[1] = _categorical_order(g, resolved, color_scale[1])
         user_scale = getattr(g, "colour_scale", None)
         if user_scale is not None and user_scale.kind == "continuous":
             raise ValueError(

{plot3-0.6.1 → plot3-0.6.2}/plot3/static.py

@@ -2196,13 +2196,15 @@ def _paint_legend(commands, origin, metrics, theme, fonts) -> None:
     surface = theme.get("surface") or "#0b1020"
     grid = theme.get("grid") or "#1c2742"
     if _rgb(grid) == _rgb(surface):
-        grid = theme.get("muted") or "#898781"
+        # theme_grey / theme_classic: white grid on a white page. ggplot2
+        # draws no legend box, so neither do we (not a dark one instead).
+        grid = None
     tick = fonts[0]
     row_h = metrics["row_h"]
     lx, ly = origin
     # Opaque, so a legend inside the panel does not fade the marks under it.
     commands.append((
-        "rect", lx, ly, metrics["w"], metrics["h"], surface, grid, 1, 1.0,
+        "rect", lx, ly, metrics["w"], metrics["h"], surface, grid, 1 if grid else 0, 1.0,
     ))
     cursor = ly + 6
     for row in metrics["rows"]:

{plot3-0.6.1 → plot3-0.6.2}/plot3.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plot3
-Version: 0.6.1
+Version: 0.6.2
 Summary: ggplot2 grammar of graphics for Python: interactive WebGL figures and journal-ready PNG, SVG, and PDF
 Author: Rigoberto Leyva Salmeron
 License-Expression: MIT
@@ -224,18 +224,27 @@ ggplot2's names and defaults.
 
 ```python
 rng = np.random.default_rng(1)
-trial = pd.DataFrame({"arm": rng.choice(["placebo", "low", "high"], 150),
+arms = ["placebo", "low", "high"]
+trial = pd.DataFrame({"arm": pd.Categorical(rng.choice(arms, 150), categories=arms),
                       "dose": rng.uniform(0, 10, 150)})
-trial["response"] = 2 + 0.6 * trial.dose + 1.5 * (trial.arm == "high") + rng.normal(0, 1.2, 150)
+slope = trial.arm.map({"placebo": 0.05, "low": 0.35, "high": 0.7}).astype(float)
+trial["response"] = 2 + slope * trial.dose + rng.normal(0, 0.9, 150)
 
 a = (ggplot(trial, aes(x="dose", y="response", colour="arm"))
-     + geom_point(alpha=0.7) + geom_smooth(method="lm"))
+     + geom_point(alpha=0.7) + geom_smooth(method="lm", se=False)
+     + scale_colour_hue() + theme_grey() + theme(legend_position="bottom"))
 b = (ggplot(trial, aes(x="arm", y="response", fill="arm"))
-     + geom_violin() + geom_jitter(width=0.1, height=0) + theme(legend_position="none"))
-fig = (a | b) + plot_annotation(title="Response by dose and arm", tag_levels="A")
+     + geom_violin() + geom_jitter(width=0.1, height=0)
+     + scale_fill_hue() + theme_grey() + theme(legend_position="none"))
+fig = ((a | b) + plot_layout(widths=[3, 2])
+       + plot_annotation(title="Response by dose and arm", tag_levels="A"))
 ```
 
-<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_compose.png" alt="Two-panel figure: scatter with linear fits, and violins by arm">
+`scale_colour_hue()` and `theme_grey()` are ggplot2's own defaults, so the
+figure looks as it would from R. plot3's defaults are `theme_bw()` for saved
+files and `theme_dark()` in the viewer.
+
+<img src="https://raw.githubusercontent.com/rleyvasal/plot3/main/docs/img/ggplot2_look.png" alt="Two-panel figure in ggplot2's grey theme: response against dose with one fitted line per arm, and violins of response by arm">
 
 Everything else you'd expect is there, including histograms, densities,
 boxplots, bar positions, `stat_summary`, error bars, heatmaps, hexbins, 2D

{plot3-0.6.1 → plot3-0.6.2}/tests/test_grammar.py

@@ -135,3 +135,45 @@ def test_geom_smooth_takes_a_linetype():
     assert [layer.get("dash") for layer in spec["layers"] if layer["kind"] == "line"] == [[4.0, 4.0]]
     with pytest.raises(ValueError):
         geom_smooth(linetype="wavy")
+
+
+@pytest.mark.parametrize("kind", ["violin", "boxplot", "density", "histogram", "bar", "smooth", "summary"])
+def test_categorical_colour_order_survives_stats(kind):
+    # pd.Categorical(categories=["placebo", "low", "high"]) sets the colour
+    # order for every geom, as in ggplot2; stats used to sort it as text, so
+    # a violin and a scatter of the same column disagreed on colours.
+    import numpy as np
+    import pandas as pd
+
+    from plot3 import (aes, geom_bar, geom_boxplot, geom_density, geom_histogram,
+                       geom_smooth, geom_violin, ggplot, stat_summary)
+    from plot3.build import build_spec
+
+    rng = np.random.default_rng(0)
+    arms = ["placebo", "low", "high"]
+    df = pd.DataFrame({"arm": pd.Categorical(rng.choice(arms, 90), categories=arms),
+                       "x": rng.uniform(0, 10, 90), "y": rng.normal(size=90),
+                       "sex": rng.choice(["F", "M"], 90)})
+    layer = {
+        "violin": (aes(x="arm", y="y", fill="arm"), geom_violin()),
+        "boxplot": (aes(x="sex", y="y", fill="arm"), geom_boxplot()),
+        "density": (aes(x="y", colour="arm"), geom_density()),
+        "histogram": (aes(x="y", fill="arm"), geom_histogram(bins=10)),
+        "bar": (aes(x="sex", fill="arm"), geom_bar()),
+        "smooth": (aes(x="x", y="y", colour="arm"), geom_smooth(method="lm")),
+        "summary": (aes(x="sex", y="y", colour="arm"), stat_summary()),
+    }[kind]
+    spec, _ = build_spec(ggplot(df, layer[0]) + layer[1])
+    assert spec["color"]["cats"] == arms
+    assert [entry["label"] for entry in spec["legend"]] == arms
+
+
+def test_theme_grey_legend_has_no_dark_frame(tmp_path):
+    import pandas as pd
+
+    from plot3 import aes, geom_point, ggplot, ggsave, theme_grey
+
+    df = pd.DataFrame({"x": [1, 2], "y": [1, 2], "g": ["a", "b"]})
+    out = tmp_path / "g.svg"
+    ggsave(str(out), ggplot(df, aes(x="x", y="y", colour="g")) + geom_point() + theme_grey())
+    assert 'stroke="#4d4d4d"' not in out.read_text()

plot3-0.6.1/plot3/__version__.py

@@ -1 +0,0 @@
-__version__ = "0.6.1"