omle-convert 0.5.0__tar.gz

omle_convert-0.5.0/.github/workflows/publish.yml

@@ -0,0 +1,216 @@
+name: Publish
+
+# Publishes omle-convert to PyPI.
+#
+# Two ways to run:
+#   1. Push a version tag  →  builds, verifies, publishes to PyPI.
+#        git tag v0.1.0 && git push origin v0.1.0
+#   2. Run manually        →  builds, verifies, publishes to TestPyPI
+#        (Actions → Publish → Run workflow)
+#
+# ORDERING: this package depends on `omle>=0.1.0`, so the verify jobs cannot
+# install the built wheel until omle itself is on PyPI. Publish omle
+# first; until then every run of this workflow fails at "Install the wheel".
+#
+# Authentication uses PyPI Trusted Publishing (OIDC) — no API tokens or
+# secrets. Before the first run, register this workflow as a trusted publisher:
+#   PyPI:     https://pypi.org/manage/account/publishing/
+#   TestPyPI: https://test.pypi.org/manage/account/publishing/
+# with owner "openmle", repository "omle-convert", workflow "publish.yml",
+# and environment "pypi" / "testpypi" respectively.
+
+on:
+  push:
+    tags: ["v*"]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  build:
+    name: Build distributions
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4.2.2
+        with:
+          # setuptools-scm derives the version from git tags — a shallow
+          # checkout yields a dev version with a local segment.
+          fetch-depth: 0
+
+      - name: Set up Python
+        uses: actions/setup-python@v5.6.0
+        with:
+          python-version: "3.12"
+
+      - name: Install build tooling
+        run: pip install build twine
+
+      - name: Build sdist and wheel
+        run: |
+          rm -rf build dist
+          python -m build
+
+      - name: Check metadata
+        run: twine check dist/*
+
+      - name: Reject versions PyPI will not accept
+        run: |
+          python - <<'PY'
+          import glob
+          import os
+          import sys
+
+          wheels = glob.glob("dist/*.whl")
+          if len(wheels) != 1:
+              sys.exit(f"::error::Expected exactly one wheel in dist/, found {wheels}")
+
+          # Wheel filenames are {name}-{version}-{python}-{abi}-{platform}.whl;
+          # parse the basename so a '-' in the workspace path cannot shift fields.
+          version = os.path.basename(wheels[0]).split("-")[1]
+          print(f"Built version: {version}")
+
+          if "+" in version:
+              print(
+                  f"::error::Version '{version}' carries a local segment, which PyPI "
+                  "rejects. Publish from a checkout of a version tag (e.g. git tag v0.1.0)."
+              )
+              sys.exit(1)
+          PY
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+  verify:
+    name: Verify install (${{ matrix.python-version }})
+    needs: build
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.10", "3.11", "3.12", "3.13", "3.14"]
+    steps:
+      - uses: actions/checkout@v4.2.2
+
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5.6.0
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install the wheel
+        # Pulls omle from PyPI — see the ORDERING note at the top of this file.
+        run: pip install pytest scikit-learn pandas numpy dist/*.whl
+
+      - name: Run the scikit-learn suite against the installed wheel
+        # -o pythonpath= drops the src/ entries from pyproject.toml so the tests
+        # exercise the wheel that is about to be published, not the checkout.
+        # test_cli.py skips its XGBoost/LightGBM cases via importorskip.
+        run: |
+          python -c "import omle_convert; print('testing:', omle_convert.__file__)"
+          pytest tests/test_sklearn.py tests/test_cli.py -o pythonpath=
+
+      - name: Smoke-test an end-to-end conversion
+        # Run from a scratch directory so the repo checkout cannot satisfy imports.
+        run: |
+          mkdir -p /tmp/smoke && cd /tmp/smoke
+          python - <<'PY'
+          from sklearn.datasets import make_classification
+          from sklearn.ensemble import RandomForestClassifier
+          from sklearn.pipeline import make_pipeline
+          from sklearn.preprocessing import StandardScaler
+
+          import omle
+          from omle_convert import export_omle
+
+          X, y = make_classification(n_samples=64, n_features=6, random_state=0)
+          pipeline = make_pipeline(StandardScaler(), RandomForestClassifier(n_estimators=5, random_state=0))
+          pipeline.fit(X, y)
+
+          export_omle(pipeline, "smoke.omle", X=X)
+
+          model = omle.load("smoke.omle")
+          result = omle.validate(model)
+          assert result.is_valid, result
+          print("converted and validated:", [f"{n.domain}.{n.op}" for n in model.nodes])
+          PY
+          omle-convert --help > /dev/null && echo "CLI ok"
+
+  verify-frameworks:
+    name: Verify install (gradient boosting frameworks)
+    needs: build
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4.2.2
+
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Set up JDK 17
+        # For the Spark suites, which fit models through a real SparkSession.
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: "17"
+
+      - name: Set up Python
+        uses: actions/setup-python@v5.6.0
+        with:
+          python-version: "3.11"
+
+      - name: Install the wheel with the supported frameworks
+        run: |
+          pip install pytest scikit-learn pandas numpy category_encoders \
+            xgboost lightgbm catboost pyspark synapseml dist/*.whl
+
+      - name: Run the framework suites against the installed wheel
+        # The whole suite, Spark included: the JDK above plus these packages are
+        # all it needs, and the release gate should exercise every converter the
+        # wheel claims to ship.
+        run: pytest tests/ -o pythonpath=
+
+  publish-testpypi:
+    name: Publish to TestPyPI
+    if: github.event_name == 'workflow_dispatch'
+    needs: [verify, verify-frameworks]
+    runs-on: ubuntu-latest
+    environment:
+      name: testpypi
+      url: https://test.pypi.org/p/omle-convert
+    permissions:
+      id-token: write
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          repository-url: https://test.pypi.org/legacy/
+          skip-existing: true
+
+  publish-pypi:
+    name: Publish to PyPI
+    if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/v')
+    needs: [verify, verify-frameworks]
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/p/omle-convert
+    permissions:
+      id-token: write
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - uses: pypa/gh-action-pypi-publish@release/v1

omle_convert-0.5.0/.github/workflows/site-rebuild.yml

@@ -0,0 +1,43 @@
+name: Rebuild site
+
+# docs/ holds the per-framework support references, which the project site pulls
+# in at build time (openmle.github.io/scripts/sync_content.py → docs/converters/).
+# Nothing rebuilds the site on its own, so without this a corrected support table
+# stays invisible until some other repository happens to trigger a build.
+#
+# The same workflow exists in omle for the same reason. Both are no-ops when
+# SITE_DISPATCH_TOKEN is unset, so a fork does not fail on it.
+on:
+  push:
+    branches: [main]
+    paths:
+      - "docs/**"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  rebuild-site:
+    name: Trigger site rebuild
+    runs-on: ubuntu-latest
+    steps:
+      - name: Dispatch content-updated to openmle.github.io
+        env:
+          # A PAT with `contents: write` on openmle/openmle.github.io. The
+          # default GITHUB_TOKEN is scoped to this repository and cannot
+          # dispatch to another one.
+          TOKEN: ${{ secrets.SITE_DISPATCH_TOKEN }}
+        run: |
+          if [ -z "$TOKEN" ]; then
+            echo "::warning::SITE_DISPATCH_TOKEN is not set; skipping site rebuild."
+            exit 0
+          fi
+          curl --fail-with-body -sS -X POST \
+            -H "Authorization: Bearer $TOKEN" \
+            -H "Accept: application/vnd.github+json" \
+            -H "X-GitHub-Api-Version: 2022-11-28" \
+            https://api.github.com/repos/openmle/openmle.github.io/dispatches \
+            -d "{\"event_type\":\"content-updated\",
+                 \"client_payload\":{\"source\":\"omle-convert\",\"sha\":\"${GITHUB_SHA}\"}}"
+          echo "Requested a site rebuild for omle-convert ${GITHUB_SHA}."

omle_convert-0.5.0/.github/workflows/test.yml

@@ -0,0 +1,163 @@
+name: Tests
+
+on:
+  push:
+    branches: ["**"]
+  pull_request:
+    branches: ["**"]
+
+jobs:
+  # Runs .pre-commit-config.yaml, the same file the local git hook runs, so the
+  # two cannot drift. --all-files because a hook installed locally only ever
+  # sees staged files, and nothing guarantees a contributor installed it at all.
+  # No `pip install -e .` here: every hook brings its own pinned environment, so
+  # the job skips the scikit-learn/xgboost/lightgbm install the test jobs need.
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4.2.2
+      - uses: actions/setup-python@v5.6.0
+        with:
+          python-version: "3.12"
+      - name: Cache pre-commit environments
+        uses: actions/cache@v4
+        with:
+          path: ~/.cache/pre-commit
+          key: pre-commit-${{ runner.os }}-${{ hashFiles('.pre-commit-config.yaml') }}
+      - run: pip install pre-commit
+      # --show-diff-on-failure prints the patch the autofixing hooks would have
+      # applied, so a red build hands back the fix instead of just a rule name.
+      - run: pre-commit run --all-files --show-diff-on-failure
+
+  test:
+    name: ${{ matrix.label }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        # Each row pins a DIFFERENT framework combination. Left to pip, every row
+        # would resolve to "newest that supports this interpreter", which both
+        # duplicates (scikit-learn 1.9 would run on three rows, lightgbm 4.7 on
+        # four) and leaves whole series untested — and the coverage would drift
+        # silently every time upstream publishes.
+        #
+        # Pins stay inside series upstream still maintains; anything older is not
+        # worth a CI slot. Revisit when a pinned series goes stale.
+        #
+        # Collectively: scikit-learn 1.7-1.9, xgboost 3.0-3.4, lightgbm 4.7,
+        # category_encoders 2.8-2.11, Spark 3.5 and 4.x, plus one unpinned row.
+        #
+        # The dev-extra floors in pyproject.toml (scikit-learn 1.6, xgboost 2.1,
+        # lightgbm 4.6, catboost 1.2.7, category_encoders 2.6.4) all sit below
+        # the oldest row here, so no row exercises them. Raise them to match
+        # this matrix, or accept that the declared minimums are untested.
+        #
+        # catboost is pinned by PATCH, not minor: it ships one long-lived series
+        # (1.1 is 47 months stale, so 1.2 is effectively the only one) but that
+        # series has 11 releases spanning 41 months, which is more drift than
+        # lightgbm has across its minors. Wheel coverage bounds the choices:
+        # cp312 needs >=1.2.3, cp313 needs >=1.2.8, cp314 needs >=1.2.9.
+        #
+        # category_encoders is pinned by PATCH for the same reason its series
+        # boundaries are interpreter-driven: 2.8.1 is the ceiling on Python 3.10
+        # so that row is already determined. Everything from 2.9 requires 3.11+,
+        # which would collapse the remaining rows onto whatever is newest;
+        # pinning 2.9, 2.10 and 2.11 keeps each post-3.11 series covered and
+        # leaves the canary unpinned.
+        include:
+          # Oldest supported interpreter, and the oldest framework combination
+          # still under test. scikit-learn 1.7 and lightgbm 4.7 both require
+          # 3.10+, so this is the earliest interpreter that can carry them.
+          - label: "py3.10 / oldest maintained"
+            python-version: "3.10"
+            pins: >-
+              scikit-learn==1.7.*
+              xgboost==3.0.*
+              lightgbm==4.7.*
+              catboost==1.2.8
+              category_encoders==2.8.1
+              pyspark==3.5.*
+
+          - label: "py3.11 / sklearn 1.8, xgb 3.2"
+            python-version: "3.11"
+            pins: >-
+              scikit-learn==1.8.*
+              xgboost==3.2.*
+              lightgbm==4.7.*
+              catboost==1.2.9
+              category_encoders==2.9.0
+              pyspark==3.5.*
+
+          # xgboost 3.3+ requires 3.12, so this is the earliest row that can test
+          # it. Spark 4 rides the newer interpreters; 3.5 is covered above.
+          - label: "py3.12 / sklearn 1.9, xgb 3.3, Spark 4"
+            python-version: "3.12"
+            pins: >-
+              scikit-learn==1.9.*
+              xgboost==3.3.*
+              lightgbm==4.7.*
+              catboost==1.2.10
+              category_encoders==2.10.0
+
+          # Newly pinned now that 3.14 is the canary: without pins this row and
+          # the one below would both resolve to "all latest" and cover the same
+          # ground. xgboost 3.4 and category_encoders 2.11 are only reachable
+          # here and above.
+          - label: "py3.13 / xgb 3.4, ce 2.11"
+            python-version: "3.13"
+            pins: >-
+              xgboost==3.4.*
+              category_encoders==2.11.*
+
+          # Deliberately unpinned: the canary that catches a new upstream release
+          # breaking us before anyone has pinned it.
+          - label: "py3.14 / all latest"
+            python-version: "3.14"
+
+    steps:
+      - uses: actions/checkout@v4.2.2
+
+      - name: Set up JDK 17
+        # Every row needs a JVM: pyspark runs one for all of its tests, not just
+        # the suite that loads third-party JARs. Spark 4 requires 17+ and Spark
+        # 3.5 supports it, so one JDK covers the whole matrix.
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: "17"
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5.6.0
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install package with dev dependencies
+        run: pip install -e ".[dev]"
+
+      - name: Apply framework pins for this matrix cell
+        if: matrix.pins != ''
+        run: pip install ${{ matrix.pins }}
+
+      - name: Record the framework versions under test
+        # Printed so a failure report names the combination that broke, rather
+        # than leaving it to be reconstructed from the interpreter version.
+        run: |
+          java -version
+          python - <<'PY'
+          import importlib.metadata as md
+          for pkg in ("scikit-learn", "xgboost", "lightgbm", "catboost",
+                      "category_encoders", "synapseml", "pyspark", "numpy",
+                      "pandas", "pyarrow"):
+              try:
+                  print(f"{pkg:18s} {md.version(pkg)}")
+              except md.PackageNotFoundError:
+                  print(f"{pkg:18s} (not installed)")
+          PY
+
+      - name: Run tests
+        # Including tests/test_spark_jvm.py. It used to be excluded because it
+        # needed third-party JARs downloaded by hand; it no longer does. The
+        # `sparkxgb` tests are gone and SynapseML resolves its JARs from Maven
+        # at session start, so `pip install -e ".[dev]"` plus the JDK above is
+        # the entire setup.
+        run: pytest tests/

omle_convert-0.5.0/.gitignore

@@ -0,0 +1,214 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+#poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+#pdm.lock
+#pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+#pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# Abstra
+# Abstra is an AI-powered process automation framework.
+# Ignore directories containing user credentials, local state, and settings.
+# Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore
+#  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#  and can be added to the global gitignore or merged into this file. However, if you prefer,
+#  you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+/catboost_info
+/tests/models
+/tests/catboost_info
+
+# setuptools-scm generated version file
+/src/omle_convert/_version.py

omle_convert-0.5.0/.pre-commit-config.yaml

@@ -0,0 +1,25 @@
+# Optional but recommended:
+#   pip install pre-commit && pre-commit install
+#
+# `ruff check` enforces the Google Python Style Guide rules configured in
+# pyproject.toml. No autoformatter: `ruff format` would rewrite the aligned
+# dict literals and section dividers this codebase uses.
+
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+      - id: trailing-whitespace
+      - id: end-of-file-fixer
+      - id: check-merge-conflict
+      - id: check-case-conflict
+      - id: check-yaml
+      - id: check-toml
+      - id: check-added-large-files
+        args: [--maxkb=1024]
+
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.16.7
+    hooks:
+      - id: ruff-check
+        args: [--fix]