Split make_assets scripts by plot functions (#247)

* migrate some changes

* put under dir first

* split coordination

* split histogram

* split phonon

* splig rdf

* splig scatter

* splig structure_viz and scatter

* split the rest

* split ptable

* copy random data generator

* rename powerups plotly

* remove plotly example from ptable mpl

* move assets/(asset_makers->scripts)

* mv assets/*.svg assets/svg/*.svg

* change save_and_compress_svg default output directory to assets/svg/

* restore examples/ ruff per-file-ignores

* rm assets/scripts/(scatter|uncertainty)/_random_regression_data.py

unused, left from bad merge conflict resolve

* revert ptable_heatmap_splits

* add tests/data/regression.py

* fix with_marginal_hist creating two nested coordinated axes in density_hexbin_(hex_)with_hist

* make dummy data generator modular a bit

* add config_matplotlib

* remove np_rng

* enhance description of dummy data with GPT

* migrate ptable

* migrate phonons

* relocate ptable dash app

* migrate the rest

* mv pymatviz/test/data.py to pymatviz/data/__init__.py

mv tests/test/test_data.py to tests/test_data.py

* fix test.yml find-scripts looking in wrong folder

* merge pymatviz/test/config.py into pymatviz/utils.py and rename config_matplotlib to apply_matplotlib_template

add unit test for apply_matplotlib_template()
remove unneeded plotly templates from matplotlib asset scripts

* fix asset scripts failing in CI, delete duplicate assets/scripts/powerups/powerups.py

remove unused pmv.set_plotly_template

---------

Co-authored-by: Janosh Riebesell <[email protected]>

Author: DanielYang59

.github/workflows/test.yml

@@ -10,6 +10,9 @@ on:
   workflow_dispatch:
   workflow_call:
 
+env:
+  MPLBACKEND: Agg # non-interactive backend for matplotlib
+
 jobs:
   tests:
     strategy:
@@ -36,7 +39,7 @@ jobs:
       - name: Find Python scripts
         id: set-matrix
         run: |
-          SCRIPTS=$(find examples/make_assets -name "*.py" | jq -R -s -c 'split("\n")[:-1]')
+          SCRIPTS=$(find assets/scripts -name "*.py" | jq -R -s -c 'split("\n")[:-1]')
           echo "script_list=$SCRIPTS" >> $GITHUB_OUTPUT
 
   test-scripts:

assets/density-scatter-with-hist.svg

@@ -8,7 +8,6 @@
 
 import pymatviz as pmv
 from pymatviz.coordination import CnSplitMode
-from pymatviz.enums import Key
 from pymatviz.utils import TEST_FILES
 
 
@@ -190,44 +189,3 @@
 )
 fig.show()
 pmv.io.save_and_compress_svg(fig, "coordination-hist-all-combined-side-by-side")
-
-
-# %% Coordination vs Cutoff example for a single structure
-fig = pmv.coordination_vs_cutoff_line(structures[key1])
-fig.layout.title = dict(text=f"Coordination vs Cutoff: {key1}", x=0.5, y=0.98)
-fig.layout.margin.t = 50
-fig.show()
-pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-single")
-
-
-# %% Coordination vs Cutoff example for multiple structures
-fig = pmv.coordination_vs_cutoff_line(
-    {key1: structures[key1], key2: structures[key2], key3: structures[key3]},
-    strategy=(1.0, 6.0),
-    num_points=100,
-)
-fig.layout.title = dict(
-    text="Coordination vs Cutoff: Multiple Structures", x=0.5, y=0.98
-)
-fig.layout.legend.update(x=0, y=1, bgcolor="rgba(0,0,0,0)")
-fig.layout.margin.t = 50
-fig.show()
-pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-multiple")
-
-
-# %% Coordination vs Cutoff example with custom color scheme
-custom_colors = {"Pd": "red", "Zr": "blue"}
-fig = pmv.coordination_vs_cutoff_line(
-    structures[key1], element_color_scheme=custom_colors
-)
-fig.layout.margin.t = 25
-fig.layout.legend.update(x=0, y=1)
-fig.show()
-pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-custom-colors")
-
-
-# %% plot first 10 structures
-fig = pmv.coordination_vs_cutoff_line(df_phonon[Key.structure][:4].tolist())
-fig.layout.margin.t = 25
-fig.layout.legend.update(x=0, y=1)
-fig.show()

examples/make_assets/coordination.py renamed to assets/scripts/coordination/coordination_hist.py

@@ -0,0 +1,85 @@
+# %%
+from glob import glob
+
+from matminer.datasets import load_dataset
+from pymatgen.core import Structure
+from pymatgen.symmetry.analyzer import SpacegroupAnalyzer
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+from pymatviz.utils import TEST_FILES
+
+
+pmv.set_plotly_template("pymatviz_white")
+
+df_phonon = load_dataset(data_name := "matbench_phonons")
+
+
+# %%
+formula_spg_str = (
+    lambda struct: f"{struct.formula} ({struct.get_space_group_info()[1]})"
+)
+structures = {
+    formula_spg_str(struct := Structure.from_file(file)): struct
+    for file in (
+        glob(f"{TEST_FILES}/structures/*.json.gz") + glob(f"{TEST_FILES}/xrd/*.cif")
+    )
+}
+key1, key2, key3, *_ = structures
+
+for struct in structures.values():
+    spga = SpacegroupAnalyzer(struct)
+    sym_struct = spga.get_symmetrized_structure()
+    # add wyckoff symbols to each site
+    struct.add_oxidation_state_by_guess()
+    wyckoff_symbols = ["n/a"] * len(struct)
+    for indices, symbol in zip(
+        sym_struct.equivalent_indices, sym_struct.wyckoff_symbols, strict=True
+    ):
+        for index in indices:
+            wyckoff_symbols[index] = symbol
+    if any(sym == "n/a" for sym in wyckoff_symbols):
+        raise ValueError(f"{struct.formula} has n/a {wyckoff_symbols=}")
+
+    struct.add_site_property("wyckoff", wyckoff_symbols)
+
+
+# %% Coordination vs Cutoff example for a single structure
+fig = pmv.coordination_vs_cutoff_line(structures[key1])
+fig.layout.title = dict(text=f"Coordination vs Cutoff: {key1}", x=0.5, y=0.98)
+fig.layout.margin.t = 50
+fig.show()
+pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-single")
+
+
+# %% Coordination vs Cutoff example for multiple structures
+fig = pmv.coordination_vs_cutoff_line(
+    {key1: structures[key1], key2: structures[key2], key3: structures[key3]},
+    strategy=(1.0, 6.0),
+    num_points=100,
+)
+fig.layout.title = dict(
+    text="Coordination vs Cutoff: Multiple Structures", x=0.5, y=0.98
+)
+fig.layout.legend.update(x=0, y=1, bgcolor="rgba(0,0,0,0)")
+fig.layout.margin.t = 50
+fig.show()
+pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-multiple")
+
+
+# %% Coordination vs Cutoff example with custom color scheme
+custom_colors = {"Pd": "red", "Zr": "blue"}
+fig = pmv.coordination_vs_cutoff_line(
+    structures[key1], element_color_scheme=custom_colors
+)
+fig.layout.margin.t = 25
+fig.layout.legend.update(x=0, y=1)
+fig.show()
+pmv.io.save_and_compress_svg(fig, "coordination-vs-cutoff-custom-colors")
+
+
+# %% plot first 10 structures
+fig = pmv.coordination_vs_cutoff_line(df_phonon[Key.structure][:4].tolist())
+fig.layout.margin.t = 25
+fig.layout.legend.update(x=0, y=1)
+fig.show()

assets/scripts/coordination/coordination_vs_cutoff_line.py

@@ -0,0 +1,14 @@
+# %%
+from matminer.datasets import load_dataset
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+# %% Histogram Plots
+df_expt_gap = load_dataset("matbench_expt_gap")
+
+ax = pmv.elements_hist(
+    df_expt_gap[Key.composition], keep_top=15, v_offset=200, rotation=0, fontsize=12
+)
+pmv.io.save_and_compress_svg(ax, "elements-hist")

assets/scripts/histogram/elements_hist.py

@@ -0,0 +1,21 @@
+# %%
+import numpy as np
+
+import pymatviz as pmv
+
+
+pmv.set_plotly_template("pymatviz_white")
+
+
+# %% plot 2 Gaussians and their cumulative distribution functions
+rand_regression_size = 500
+np_rng = np.random.default_rng(seed=0)
+gauss1 = np_rng.normal(5, 4, rand_regression_size)
+gauss2 = np_rng.normal(10, 2, rand_regression_size)
+
+fig = pmv.histogram({"Gaussian 1": gauss1, "Gaussian 2": gauss2}, bins=100)
+for idx in range(len(fig.data)):
+    pmv.powerups.add_ecdf_line(fig, trace_idx=idx)
+fig.show()
+
+pmv.io.save_and_compress_svg(fig, "histogram-ecdf")

assets/scripts/histogram/histogram.py

@@ -0,0 +1,27 @@
+# %%
+from matminer.datasets import load_dataset
+from tqdm import tqdm
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+pmv.set_plotly_template("pymatviz_white")
+
+
+# %%
+df_phonons = load_dataset("matbench_phonons")
+
+df_phonons[[Key.spg_symbol, Key.spg_num]] = [
+    struct.get_space_group_info()
+    for struct in tqdm(df_phonons[Key.structure], desc="Getting spacegroups")
+]
+
+
+# %% Spacegroup histograms
+for backend in pmv.BACKENDS:
+    fig = pmv.spacegroup_bar(df_phonons[Key.spg_num], backend=backend)
+    pmv.io.save_and_compress_svg(fig, f"spg-num-hist-{backend}")
+
+    fig = pmv.spacegroup_bar(df_phonons[Key.spg_symbol], backend=backend)
+    pmv.io.save_and_compress_svg(fig, f"spg-symbol-hist-{backend}")

assets/scripts/histogram/spacegroup_bar.py

@@ -0,0 +1,38 @@
+# %%
+import json
+from glob import glob
+
+from monty.io import zopen
+from monty.json import MontyDecoder
+from pymatgen.phonon.bandstructure import PhononBandStructureSymmLine as PhononBands
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+# TODO: ffonons not working properly (see #195)
+try:
+    import ffonons  # noqa: F401
+except ImportError:
+    raise SystemExit(0) from None  # install ffonons to run this script
+
+
+# %% Plot phonon bands and DOS
+for mp_id, formula in (
+    ("mp-2758", "Sr4Se4"),
+    ("mp-23907", "H2"),
+):
+    docs = {}
+    for path in glob(f"{pmv.utils.TEST_FILES}/phonons/{mp_id}-{formula}-*.json.lzma"):
+        key = path.split("-")[-1].split(".")[0]
+        with zopen(path) as file:
+            docs[key] = json.loads(file.read(), cls=MontyDecoder)
+
+    ph_bands: dict[str, PhononBands] = {
+        key: getattr(doc, Key.ph_band_structure) for key, doc in docs.items()
+    }
+
+    fig = pmv.phonon_bands(ph_bands)
+    fig.layout.title = dict(text=f"Phonon Bands of {formula} ({mp_id})", x=0.5, y=0.98)
+    fig.layout.margin = dict(l=0, r=0, b=0, t=40)
+    pmv.io.save_and_compress_svg(fig, f"phonon-bands-{mp_id}")

assets/scripts/phonons/phonon_bands.py

@@ -36,16 +36,6 @@
         key: getattr(doc, Key.ph_dos) for key, doc in docs.items()
     }
 
-    fig = pmv.phonon_bands(ph_bands)
-    fig.layout.title = dict(text=f"Phonon Bands of {formula} ({mp_id})", x=0.5, y=0.98)
-    fig.layout.margin = dict(l=0, r=0, b=0, t=40)
-    pmv.io.save_and_compress_svg(fig, f"phonon-bands-{mp_id}")
-
-    fig = pmv.phonon_dos(ph_doses)
-    fig.layout.title = dict(text=f"Phonon DOS of {formula} ({mp_id})", x=0.5, y=0.98)
-    fig.layout.margin = dict(l=0, r=0, b=0, t=40)
-    pmv.io.save_and_compress_svg(fig, f"phonon-dos-{mp_id}")
-
     fig = pmv.phonon_bands_and_dos(ph_bands, ph_doses)
     fig.layout.title = dict(
         text=f"Phonon Bands and DOS of {formula} ({mp_id})", x=0.5, y=0.98

examples/make_assets/phonons.py renamed to assets/scripts/phonons/phonon_bands_and_dos.py

@@ -0,0 +1,38 @@
+# %%
+import json
+from glob import glob
+
+from monty.io import zopen
+from monty.json import MontyDecoder
+from pymatgen.phonon.dos import PhononDos
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+# TODO: ffonons not working properly (see #195)
+try:
+    import ffonons  # noqa: F401
+except ImportError:
+    raise SystemExit(0) from None  # install ffonons to run this script
+
+
+# %% Plot phonon bands and DOS
+for mp_id, formula in (
+    ("mp-2758", "Sr4Se4"),
+    ("mp-23907", "H2"),
+):
+    docs = {}
+    for path in glob(f"{pmv.utils.TEST_FILES}/phonons/{mp_id}-{formula}-*.json.lzma"):
+        key = path.split("-")[-1].split(".")[0]
+        with zopen(path) as file:
+            docs[key] = json.loads(file.read(), cls=MontyDecoder)
+
+    ph_doses: dict[str, PhononDos] = {
+        key: getattr(doc, Key.ph_dos) for key, doc in docs.items()
+    }
+
+    fig = pmv.phonon_dos(ph_doses)
+    fig.layout.title = dict(text=f"Phonon DOS of {formula} ({mp_id})", x=0.5, y=0.98)
+    fig.layout.margin = dict(l=0, r=0, b=0, t=40)
+    pmv.io.save_and_compress_svg(fig, f"phonon-dos-{mp_id}")

assets/scripts/phonons/phonon_dos.py

@@ -0,0 +1,44 @@
+# %%
+from matminer.datasets import load_dataset
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+# %%
+df_expt_gap = load_dataset("matbench_expt_gap")
+
+
+# %% Elemental Plots
+ax = pmv.ptable_heatmap(
+    pmv.count_elements(df_expt_gap[Key.composition]),
+    log=True,
+    return_type="axes",  # TODO: change to return Figure after 2025-07-01
+)
+title = (
+    f"Elements in Matbench Experimental Band Gap ({len(df_expt_gap):,} compositions)"
+)
+ax.set_title(title, x=0.75, y=2.5, fontsize=18, fontweight="bold")
+pmv.io.save_and_compress_svg(ax, "ptable-heatmap")
+
+
+# %%
+fig = pmv.ptable_heatmap(pmv.df_ptable[Key.atomic_mass], return_type="figure")
+fig.suptitle("Atomic Mass Heatmap", y=0.96, fontsize=20, fontweight="bold")
+pmv.io.save_and_compress_svg(fig, "ptable-heatmap-atomic-mass")
+
+
+# %%
+# Filter out near-zero entries
+ptable_data = pmv.count_elements(df_expt_gap[Key.composition])
+ptable_data = ptable_data[ptable_data > 0.01]
+
+fig = pmv.ptable_heatmap(
+    ptable_data,
+    value_show_mode="percent",
+    exclude_elements=["O"],
+    return_type="figure",
+)
+title = "Elements in Matbench Experimental Band Gap (percent)"
+fig.suptitle(title, y=0.96, fontsize=20, fontweight="bold")
+pmv.io.save_and_compress_svg(fig, "ptable-heatmap-percent")

examples/make_assets/powerups.py renamed to assets/scripts/powerups/plotly_powerups.py

@@ -0,0 +1,19 @@
+# %%
+from matminer.datasets import load_dataset
+
+import pymatviz as pmv
+from pymatviz.enums import Key
+
+
+# %%
+df_expt_gap = load_dataset("matbench_expt_gap")
+df_steels = load_dataset("matbench_steels")
+
+
+# %%
+fig = pmv.ptable_heatmap_ratio(
+    df_expt_gap[Key.composition], df_steels[Key.composition], log=True, value_fmt=".4g"
+)
+title = "Element ratios in Matbench Experimental Band Gap vs Matbench Steel"
+fig.suptitle(title, y=0.96, fontsize=16, fontweight="bold")
+pmv.io.save_and_compress_svg(fig, "ptable-heatmap-ratio")