[ci] test the python wheels by running all tutorials

Author: siliataider

.github/workflows/cibuildwheel-impl/action.yml

@@ -8,10 +8,20 @@ inputs:
 runs:
   using: "composite"
   steps:
+    - name: Install system dependencies
+      shell: bash
+      run: |
+        sudo apt install -y binutils cmake dpkg-dev g++ gcc libssl-dev git libx11-dev \
+        libxext-dev libxft-dev libxpm-dev libtbb-dev libvdt-dev libgif-dev libfftw3-dev
+
     - name: Build wheel
       uses: pypa/cibuildwheel@v3.0.1
       env:
+        PIP_NO_CACHE_DIR: "1"
         CIBW_BUILD: ${{ inputs.build-tag }}
+        CIBW_TEST_REQUIRES: "-r test_tutorials/requirements.txt"
+        CIBW_TEST_SOURCES: "test_tutorials"
+        CIBW_TEST_COMMAND: "pytest -vv -rF --show-capture=all test_tutorials"
 
     - name: Upload wheel
       uses: actions/upload-artifact@v4

.github/workflows/python_wheel_build.yml

@@ -11,7 +11,7 @@ on:
   schedule:
     - cron: '01 1 * * *'
   pull_request:
-    types: [labeled]
+    types: [opened, synchronize, reopened, labeled]
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.run_id }}
@@ -25,6 +25,7 @@ jobs:
       contains(github.event.pull_request.labels.*.name, 'build-python-wheels')
     runs-on: ubuntu-latest
     strategy:
+      fail-fast: false
       matrix:
         target: [cp38-manylinux_x86_64, cp39-manylinux_x86_64, cp310-manylinux_x86_64, cp311-manylinux_x86_64, cp312-manylinux_x86_64, cp313-manylinux_x86_64]
     name: ${{ matrix.target }}

test_tutorials/requirements.txt

@@ -0,0 +1,49 @@
+# ROOT requirements for third-party Python packages
+
+# PyROOT: Interoperability with numpy arrays
+numpy
+pandas
+
+# TMVA: SOFIE
+# dm-sonnet # used for GNNs
+# graph_nets
+# onnx
+
+# TMVA: PyMVA interfaces
+# scikit-learn
+# tensorflow ; python_version < "3.13" # TensorFlow doesn't support Python 3.13 yet
+# torch
+# xgboost
+
+# PyROOT: ROOT.Numba.Declare decorator
+numba>=0.48
+cffi>=1.9.1
+
+# Notebooks: ROOT C++ kernel
+# IPython
+# jupyter
+# metakernel>=0.20.0
+# notebook>=4.4.1
+
+# Distributed RDataFrame
+# pyspark>=2.4 # Spark backend
+# dask>=2022.08.1 # Dask backend
+# distributed>=2022.08.1 # Dask backend
+
+# JsMVA: Jupyter notebook magic for TMVA
+# ipywidgets
+
+# Unified Histogram Interface (UHI)
+uhi
+matplotlib
+mplhep
+
+# For testing
+# nbconvert>=7.4.0
+pytest
+# setuptools
+
+scikit-learn
+
+# Look for CPU-only versions of PyTorch to avoid pulling CUDA in the CI docker images.
+# -f https://download.pytorch.org/whl/cpu/torch_stable.html

test_tutorials/test_tutorials.py

@@ -0,0 +1,87 @@
+import subprocess
+import sys
+import pathlib
+import ROOT
+import os
+import pytest
+import signal
+
+ROOT.gROOT.SetBatch(True)
+
+tutorial_dir = pathlib.Path(str(ROOT.gROOT.GetTutorialDir()))
+
+subdirs = [
+    "analysis/dataframe",
+    "analysis/tree",
+    "hist",
+    "io/ntuple",
+    "roofit/roofit"
+]
+
+# ----------------------
+# Python tutorials tests
+# ----------------------
+py_tutorials = []
+for sub in subdirs:
+    sub_path = tutorial_dir / sub
+    # py_tutorials.extend(sub_path.rglob("*.py"))
+    for f in sub_path.rglob("*.py"):
+        # skip distrdf tutorials for now
+        if "distrdf" in f.name:
+            continue
+        py_tutorials.append(f)
+
+def test_tutorials_are_detected():
+    assert len(py_tutorials) > 0
+
+@pytest.mark.parametrize("tutorial", py_tutorials, ids=lambda p: p.name)
+def test_tutorial(tutorial):
+    env = dict(**os.environ)
+    # force matplotlib to use a non-GUI backend
+    env["MPLBACKEND"] = "Agg"
+    print("Test env:", env)
+    try:
+        result = subprocess.run(
+            [sys.executable, str(tutorial)],
+            check=True,
+            env=env,
+            timeout=60,
+        )
+        print("Test stderr:", result.stderr)
+    except subprocess.TimeoutExpired:
+        pytest.skip(f"Tutorial {tutorial} timed out")
+    except subprocess.CalledProcessError as e:
+        # read stderr to see if EOFError occurred
+        if "EOFError" in e.stderr:
+            pytest.skip(f"Skipping {tutorial.name} (requires user input)")
+        raise
+
+# ----------------------
+# C++ tutorials tests
+# ----------------------
+cpp_tutorials = []
+for sub in subdirs:
+    sub_path = tutorial_dir / sub
+    cpp_tutorials.extend(sub_path.rglob("*.C"))
+
+def test_cpp_tutorials_are_detected():
+    assert len(cpp_tutorials) > 0
+
+@pytest.mark.parametrize("tutorial", cpp_tutorials, ids=lambda p: p.name)
+def test_cpp_tutorial(tutorial):
+    try:
+        result = subprocess.run(
+            [sys.executable, "-c", f'import ROOT; ROOT.gROOT.ProcessLine(".x {tutorial}")'],
+            check=True,
+            timeout=60,
+            capture_output=True,
+            text=True
+        )
+    except subprocess.TimeoutExpired:
+        pytest.skip(f"Tutorial {tutorial} timed out")
+    except subprocess.CalledProcessError as e:
+        if e.returncode == -signal.SIGILL or e.returncode == 132:
+            pytest.fail(f"Failing {tutorial.name} (illegal instruction on this platform)")
+        elif "EOFError" in e.stderr:
+            pytest.skip(f"Skipping {tutorial.name} (requires user input)")
+        raise

tutorials/analysis/dataframe/df000_simple_dbg1.py

@@ -0,0 +1,19 @@
+## \file
+## \ingroup tutorial_dataframe
+## \notebook -draw
+## Simple RDataFrame example in Python.
+##
+## This tutorial shows a minimal example of RDataFrame. It starts without input
+## data, generates a new column `x` with random numbers, and finally draws
+## a histogram for `x`.
+##
+## \macro_code
+## \macro_output
+##
+## \date September 2021
+## \author Enric Tejedor (CERN)
+
+import ROOT
+
+# Create a data frame with 100 rows
+rdf = ROOT.RDataFrame(100)

tutorials/analysis/dataframe/df000_simple_dbg2.py

@@ -0,0 +1,22 @@
+## \file
+## \ingroup tutorial_dataframe
+## \notebook -draw
+## Simple RDataFrame example in Python.
+##
+## This tutorial shows a minimal example of RDataFrame. It starts without input
+## data, generates a new column `x` with random numbers, and finally draws
+## a histogram for `x`.
+##
+## \macro_code
+## \macro_output
+##
+## \date September 2021
+## \author Enric Tejedor (CERN)
+
+import ROOT
+
+# Create a data frame with 100 rows
+rdf = ROOT.RDataFrame(100)
+
+# Define a new column `x` that contains random numbers
+rdf_x = rdf.Define("x", "gRandom->Rndm()")

tutorials/analysis/dataframe/df000_simple_dbg3.py

@@ -0,0 +1,22 @@
+## \file
+## \ingroup tutorial_dataframe
+## \notebook -draw
+## Simple RDataFrame example in Python.
+##
+## This tutorial shows a minimal example of RDataFrame. It starts without input
+## data, generates a new column `x` with random numbers, and finally draws
+## a histogram for `x`.
+##
+## \macro_code
+## \macro_output
+##
+## \date September 2021
+## \author Enric Tejedor (CERN)
+
+import ROOT
+
+# Create a data frame with 100 rows
+rdf = ROOT.RDataFrame(100)
+
+# Define a new column `x` that contains random numbers
+rdf_x = rdf.Define("x", "42")