Migrate to pytest

.github/workflows/main.yaml

@@ -62,10 +62,14 @@ jobs:
         git clone --depth 1 https://github.com/w3c/json-ld-framing.git _json-ld-framing
         git clone --depth 1 https://github.com/json-ld/normalization.git _normalization
     - name: Test with Python=${{ matrix.python-version }} Loader=${{ matrix.loader }}
+      # run: |
+      #   python tests/runtests.py ./_json-ld-api/tests -l ${{ matrix.loader }}
+      #   python tests/runtests.py ./_json-ld-framing/tests -l ${{ matrix.loader }}
+      #   python tests/runtests.py ./_normalization/tests -l ${{ matrix.loader }}
       run: |
-        python tests/runtests.py ./_json-ld-api/tests -l ${{ matrix.loader }}
-        python tests/runtests.py ./_json-ld-framing/tests -l ${{ matrix.loader }}
-        python tests/runtests.py ./_normalization/tests -l ${{ matrix.loader }}
+        pytest --tests=./_json-ld-api/tests --loader=${{ matrix.loader }}
+        pytest --tests=./_json-ld-framing/tests --loader=${{ matrix.loader }}
+        pytest --tests=./_normalization/tests --loader=${{ matrix.loader }}
       env:
         LOADER: ${{ matrix.loader }}
   #coverage:

requirements-test.txt

@@ -1 +1,2 @@
 flake8
+pytest

tests/conftest.py

@@ -0,0 +1,153 @@
+import os
+import unittest
+import pytest
+
+# Import the existing test runner module so we can reuse Manifest/Test
+# implementations with minimal changes.
+from . import runtests
+
+
+def pytest_addoption(parser):
+    # Do only long options for pytest integration; pytest reserves
+    # lowercase single-letter short options for its own CLI flags.
+    parser.addoption('--tests', nargs='*', default=[], help='A manifest or directory to test')
+    parser.addoption('--earl', dest='earl', help='The filename to write an EARL report to')
+    parser.addoption('--loader', dest='loader', default='requests', help='The remote URL document loader: requests, aiohttp')
+    parser.addoption('--number', dest='number', help='Limit tests to those containing the specified test identifier')
+
+
+def pytest_configure(config):
+    # Apply loader choice and selected test number globally so that the
+    # existing `runtests` helpers behave the same as the CLI runner.
+    loader = config.getoption('loader')
+    if loader == 'requests':
+        runtests.jsonld._default_document_loader = runtests.jsonld.requests_document_loader()
+    elif loader == 'aiohttp':
+        runtests.jsonld._default_document_loader = runtests.jsonld.aiohttp_document_loader()
+
+    number = config.getoption('number')
+    if number:
+        runtests.ONLY_IDENTIFIER = number
+    # If an EARL output file was requested, create a session-level
+    # EarlReport instance we will populate per-test.
+    earl_fn = config.getoption('earl')
+    if earl_fn:
+        config._earl_report = runtests.EarlReport()
+    else:
+        config._earl_report = None
+
+
+def _flatten_suite(suite):
+    """Yield TestCase instances from a unittest TestSuite (recursively)."""
+    if isinstance(suite, unittest.TestSuite):
+        for s in suite:
+            yield from _flatten_suite(s)
+    elif isinstance(suite, unittest.TestCase):
+        yield suite
+
+
+def pytest_generate_tests(metafunc):
+    # Parametrize tests using the existing manifest loader if the test
+    # function needs a `manifest_test` argument.
+    if 'manifest_test' not in metafunc.fixturenames:
+        return
+
+    config = metafunc.config
+    tests_arg = config.getoption('tests') or []
+
+    if len(tests_arg):
+        test_targets = tests_arg
+    else:
+        # Default sibling directories used by the original runner
+        sibling_dirs = [
+            '../specifications/json-ld-api/tests/',
+            '../specifications/json-ld-framing/tests/',
+            '../specifications/normalization/tests/',
+        ]
+        test_targets = []
+        for d in sibling_dirs:
+            if os.path.exists(d):
+                test_targets.append(d)
+
+    if len(test_targets) == 0:
+        pytest.skip('No test manifest or directory specified (use --tests)')
+
+    # Build a root manifest structure with target files and dirs (equivalent to the original runner).
+    root_manifest = {
+        '@context': 'https://w3c.github.io/tests/context.jsonld',
+        '@id': '',
+        '@type': 'mf:Manifest',
+        'description': 'Top level PyLD test manifest',
+        'name': 'PyLD',
+        'sequence': [],
+        'filename': '/'
+    }
+
+    for test in test_targets:
+        if os.path.isfile(test):
+            root, ext = os.path.splitext(test)
+            if ext in ['.json', '.jsonld']:
+                root_manifest['sequence'].append(os.path.abspath(test))
+            else:
+                raise Exception('Unknown test file ext', root, ext)
+        elif os.path.isdir(test):
+            filename = os.path.join(test, 'manifest.jsonld')
+            if os.path.exists(filename):
+                root_manifest['sequence'].append(os.path.abspath(filename))
+
+    # Use the existing Manifest loader to create a TestSuite and flatten it
+    suite = runtests.Manifest(root_manifest, root_manifest['filename']).load()
+    tests = list(_flatten_suite(suite))
+
+    # Parametrize the test function with Test instances and use their
+    # string representation as test ids for readability in pytest output.
+    metafunc.parametrize('manifest_test', tests, ids=[str(t) for t in tests])
+
+
+@pytest.hookimpl(hookwrapper=True, tryfirst=True)
+def pytest_runtest_makereport(item):
+    # Hookwrapper gives us the final test report via `outcome.get_result()`.
+    outcome = yield
+    rep = outcome.get_result()
+
+    # We only handle the main call phase to match
+    # the behaviour of the original runner which only reported passes
+    # and failures/errors.
+    if rep.when not in ('call'):
+        return
+
+    # The parametrized pytest test attaches the original runtests.Test
+    # instance as the `manifest_test` fixture; retrieve it here.
+    manifest_test = item.funcargs.get('manifest_test')
+    if manifest_test is None:
+        return
+
+    # If an EARL report was requested at configure time, add an assertion
+    # for this test based on the pytest outcome.
+    earl_report = getattr(item.config, '_earl_report', None)
+    if earl_report is None:
+        return
+
+    # Map pytest outcomes to whether the test should be recorded as
+    # succeeded or failed. We skip 'skipped' outcomes to avoid polluting
+    # the EARL report with non-asserted tests.
+    if rep.outcome == 'skipped':
+        return
+
+    success = (rep.outcome == 'passed')
+    try:
+        earl_report.add_assertion(manifest_test, success)
+    except Exception:
+        # Don't let EARL bookkeeping break test execution; be quiet on error.
+        pass
+
+
+def pytest_sessionfinish(session, exitstatus):
+    # If the user requested an EARL report, write it using the existing
+    # `EarlReport` helper. We can't collect per-test assertions here
+    # The per-test assertions (if any) were appended to config._earl_report
+    # during test execution; write the report now if present.
+    earl = session.config.getoption('earl')
+    earl_report = getattr(session.config, '_earl_report', None)
+    if earl and earl_report is not None:
+        earl_report.write(os.path.abspath(earl))