chore: Remove excess test data from bidsschematools installation dire…

…ctories (#1985)

* chore: Move bidsschematools to src-layout

* chore: Update locations in .pre-commit-config.yaml

* chore: Ignore tests/data/bids*-examples

* test: Save test data into tests/data

* chore: Update package data discovery

* chore(ci): Make sure schemacode tests are triggered on maintenance branches

* chore(ci): Update pytest invocation

* chore: Ensure coverage is collected from tests and installed modules

.github/workflows/schemacode_ci.yml

@@ -4,10 +4,13 @@ on:
   push:
     branches:
       - "master"
+      - "maint/*"
     tags:
       - "schema-*"
   pull_request:
     branches:
+      - "master"
+      - "maint/*"
       - "*"
 
 concurrency:
@@ -87,14 +90,15 @@ jobs:
 
       - name: "Run tests"
         run: |
-          python -m pytest -vs --pyargs bidsschematools -m "not validate_schema" \
-              --cov-append --cov-report=xml --cov=bidsschematools --doctest-modules
+          python -m pytest -vs --doctest-modules -m "not validate_schema" \
+              --cov-append --cov-report=xml --cov-report=term --cov=src/bidsschematools
+        working-directory: tools/schemacode
 
       - name: Upload artifacts
         uses: actions/upload-artifact@v4
         with:
           name: unit_${{ matrix.os }}_${{ matrix.python-version }}
-          path: coverage.xml
+          path: tools/schemacode/coverage.xml
         if: success()
 
   publish:
@@ -143,13 +147,16 @@ jobs:
           python -m pip install -e ./tools/schemacode[all]
 
       - name: Run schema validation tests
-        run: python -m pytest --pyargs bidsschematools -m "validate_schema" --cov-append --cov-report=xml --cov=bidsschematools
+        run: |
+          python -m pytest -vs --doctest-modules -m "not validate_schema" \
+              --cov-append --cov-report=xml --cov-report=term --cov=src/bidsschematools
+        working-directory: tools/schemacode
 
       - name: Upload artifacts
         uses: actions/upload-artifact@v4
         with:
           name: schema_validation
-          path: coverage.xml
+          path: tools/schemacode/coverage.xml
         if: success()
 
   upload_to_codecov:

.pre-commit-config.yaml

@@ -2,7 +2,7 @@
 # See https://pre-commit.com/hooks.html for more hooks
 ci:
   skip: [shellcheck]
-exclude: 'tools/schemacode/bidsschematools/tests/data/broken_dataset_description.json'
+exclude: 'tools/schemacode/src/bidsschematools/tests/data/broken_dataset_description.json'
 repos:
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.6.0
@@ -83,7 +83,7 @@ repos:
           - types-jsonschema
           - jsonschema
           - httpx
-        args: ["tools/schemacode/bidsschematools"]
+        args: ["tools/schemacode/src"]
         pass_filenames: false
   - repo: https://github.com/koalaman/shellcheck-precommit
     rev: v0.10.0

tools/schemacode/MANIFEST.in

@@ -0,0 +1 @@
+recursive-include tests *

tools/schemacode/pyproject.toml

@@ -30,9 +30,11 @@ markers = [
   "validate_schema: tests that validate the schema itself",
 ]
 
+[tool.coverage.paths]
+source = [
+  "src/bidsschematools",
+  "**/site-packages/bidsschematools",
+]
+
 [tool.coverage.run]
 parallel = true
-omit = [
-  "*/*/tests/*",
-  "**/tests/*"
-]

tools/schemacode/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = bidsschematools
-version = file:bidsschematools/data/schema/SCHEMA_VERSION
+version = file:src/bidsschematools/data/schema/SCHEMA_VERSION
 url = https://github.com/bids-standard/bids-specification
 author = bids-standard developers
 author_email = bids.maintenance@gmail.com
@@ -25,8 +25,6 @@ install_requires =
     pyyaml
     importlib_resources; python_version < "3.9"
     jsonschema
-packages = find:
-include_package_data = false
 zip_safe = false
 
 [options.extras_require]
@@ -59,8 +57,7 @@ bidsschematools =
     data/schema/BIDS_VERSION
     data/schema/SCHEMA_VERSION
     data/schema/**/*.yaml
-    tests/data/**/*
-    tests/data/**/.bidsignore
+    tests/data/*
 
 [options.entry_points]
 console_scripts =

tools/schemacode/bidsschematools/conftest.py → tools/schemacode/src/bidsschematools/conftest.py

@@ -1,12 +1,8 @@
 import logging
 import tempfile
+from pathlib import Path
 from subprocess import run
 
-try:
-    from importlib.resources import as_file, files
-except ImportError:  # PY<3.9
-    from importlib_resources import as_file, files
-
 import pytest
 
 lgr = logging.getLogger()
@@ -35,23 +31,37 @@
 ]
 
 
+@pytest.fixture(scope="session")
+def tests_data_dir():
+    try:
+        this_file = Path(__file__)
+    except NameError:
+        return None
+
+    data_dir = this_file.parent.parent.parent / "tests" / "data"
+
+    if data_dir.exists():
+        return data_dir
+
+
 def get_gitrepo_fixture(url, whitelist):
     @pytest.fixture(scope="session")
-    def fixture():
+    def fixture(tests_data_dir):
+        if tests_data_dir is None:
+            pytest.skip("No test data directory found; probably in an installed package")
         archive_name = url.rsplit("/", 1)[-1]
-        testdata_dir = files("bidsschematools.tests.data") / archive_name
-        if testdata_dir.is_dir():
+        archive_dir = tests_data_dir / archive_name
+        if archive_dir.is_dir():
             lgr.info(
-                f"Found static testdata archive under `{testdata_dir}`. "
+                f"Found static testdata archive under `{archive_dir}`. "
                 "Not downloading latest data from version control."
             )
-            with as_file(testdata_dir) as path:
-                yield path
+            yield archive_dir
         else:
             lgr.info(
                 "No static testdata available under `%s`. "
                 "Attempting to fetch live data from version control.",
-                testdata_dir,
+                archive_dir,
             )
             with tempfile.TemporaryDirectory() as path:
                 lgr.debug("Cloning %r into %r", url, path)

tools/schemacode/src/bidsschematools/data/metaschema.json

@@ -0,0 +1 @@
+../../../../../src/metaschema.json

tools/schemacode/src/bidsschematools/data/schema

@@ -0,0 +1 @@
+../../../../../src/schema

tools/schemacode/src/bidsschematools/tests/test_make_testdata.py

@@ -0,0 +1,38 @@
+import os
+import shutil
+
+import pytest
+
+
+def require_env(var):
+    env = os.environ.get(var)
+    return pytest.mark.skipif(
+        not env, reason=f"To activate this test/feature `export {var}=1` and re-run."
+    )
+
+
+@require_env("BIDSSCHEMATOOLS_RELEASE")
+def test_make_archive(tests_data_dir, bids_examples, bids_error_examples):
+    """
+    ATTENTION! This is not a test!
+    Create static testdata archive containing the bidsschematools data reference whitelist.
+
+    Notes
+    -----
+    Due to intricacies arising from:
+
+    (1) fixtures not working outside of pytest
+    (2) implicit teardown leveraging tempdata removal (while held open by yield)
+    (3) wrappers evaluating the yield statement
+    (4) the desire to not download testdata twice for archive creation
+
+    testdata archive creation is now inconspicuously posing as a test.
+    """
+
+    ignore_git = shutil.ignore_patterns(".git*")
+    target_examples = tests_data_dir / "bids-examples"
+    target_error_examples = tests_data_dir / "bids-error-examples"
+    if bids_examples != target_examples:
+        shutil.copytree(bids_examples, target_examples, ignore=ignore_git)
+    if bids_error_examples != target_error_examples:
+        shutil.copytree(bids_error_examples, target_error_examples, ignore=ignore_git)

tools/schemacode/tests/data/.gitignore

@@ -0,0 +1 @@
+bids*-examples