Merge main: combine single-parser testing with schema generation testing

Author: nathan-stender

.github/workflows/test.yml

@@ -12,11 +12,13 @@ permissions:
   contents: read
 
 jobs:
+  # Determine which tests to run based on changed files
   detect-changes:
     runs-on: ubuntu-latest
     outputs:
       single-parser: ${{ steps.detect-parser.outputs.parser }}
       run-all-tests: ${{ steps.detect-parser.outputs.run-all }}
+      schema-generation: ${{ steps.filter.outputs.schema-generation }}
     steps:
     - uses: actions/checkout@v4
     - uses: dorny/paths-filter@v3
@@ -33,6 +35,16 @@ jobs:
             - 'pyproject.toml'
             - 'scripts/**'
             - '.github/**'
+          schema-generation:
+            - 'src/allotropy/allotrope/schemas/**'
+            - 'src/allotropy/allotrope/schema_parser/**'
+            - 'src/allotropy/allotrope/models/**'
+            - 'src/allotropy/allotrope/schemas.py'
+            - 'src/allotropy/exceptions.py'
+            - 'scripts/generate_schemas.py'
+            - 'scripts/download_schema.py'
+            - 'tests/allotrope/schema_parser/**'
+            - 'pyproject.toml'
     - name: Detect single parser changes
       id: detect-parser
       run: |
@@ -93,11 +105,11 @@ jobs:
       run: pip install "hatch>=1.13.0"
     - name: Install click
       run: pip install click!=8.3.0
-    - name: Run Tests
+    - name: Run Tests (excluding schema generation)
       run: |
         if [[ "${{ needs.detect-changes.outputs.run-all }}" == "true" ]]; then
           echo "Running all tests"
-          hatch run test_all.py3.10:pytest -n 2 tests
+          hatch run test_all.py3.10:pytest -n 2 tests --ignore=tests/allotrope/schema_parser/generate_schemas_test.py
         else
           echo "Running tests for parser: ${{ needs.detect-changes.outputs.single-parser }}"
           hatch run test_all.py3.10:pytest -n 2 tests/parsers/${{ needs.detect-changes.outputs.single-parser }} tests/parser_factory_test.py::test_table_contents
@@ -121,11 +133,11 @@ jobs:
     # NOTE: due to bug: https://github.com/pallets/click/issues/3066
     - name: Install click
       run: pip install click!=8.3.0
-    - name: Run Tests
+    - name: Run Tests (excluding schema generation)
       run: |
         if [[ "${{ needs.detect-changes.outputs.run-all }}" == "true" ]]; then
           echo "Running all tests"
-          hatch run test_all.py3.11:pytest -n 2 tests
+          hatch run test_all.py3.11:pytest -n 2 tests --ignore=tests/allotrope/schema_parser/generate_schemas_test.py
         else
           echo "Running tests for parser: ${{ needs.detect-changes.outputs.single-parser }}"
           hatch run test_all.py3.11:pytest -n 2 tests/parsers/${{ needs.detect-changes.outputs.single-parser }} tests/parser_factory_test.py::test_table_contents
@@ -148,17 +160,43 @@ jobs:
       run: pip install "hatch>=1.13.0"
     - name: Install click
       run: pip install click!=8.3.0
-    - name: Run Tests
+    - name: Run Tests (excluding schema generation)
       run: |
         if [[ "${{ needs.detect-changes.outputs.run-all }}" == "true" ]]; then
           echo "Running all tests"
-          hatch run test_all.py3.12:pytest -n 2 tests
+          hatch run test_all.py3.12:pytest -n 2 tests --ignore=tests/allotrope/schema_parser/generate_schemas_test.py
         else
           echo "Running tests for parser: ${{ needs.detect-changes.outputs.single-parser }}"
           hatch run test_all.py3.12:pytest -n 2 tests/parsers/${{ needs.detect-changes.outputs.single-parser }} tests/parser_factory_test.py::test_table_contents
         fi
       timeout-minutes: 15
 
+  # Schema generation test - only runs when relevant files change or on main branch
+  schema_generation_test:
+    needs: detect-changes
+    # Always run on main branch, otherwise only when schema files change
+    if: ${{ github.ref == 'refs/heads/main' || needs.detect-changes.outputs.schema-generation == 'true' }}
+    runs-on: ubuntu-latest
+    name: Schema Generation Test
+    strategy:
+      matrix:
+        python-version: ["3.10"]  # Only run on one Python version since output is the same
+
+    steps:
+    - uses: actions/checkout@v4
+      with:
+        lfs: true
+    - uses: actions/setup-python@v5
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Install hatch
+      run: pip install "hatch>=1.13.0"
+    - name: Install click
+      run: pip install click!=8.3.0
+    - name: Run Schema Generation Test
+      run: hatch run test_all.py${{ matrix.python-version }}:pytest tests/allotrope/schema_parser/generate_schemas_test.py -v
+      timeout-minutes: 5
+
   lint:
     runs-on: ubuntu-latest
     name: Quality Checks

src/allotropy/allotrope/schema_parser/backup_manager.py

@@ -3,8 +3,7 @@
 from itertools import zip_longest
 from pathlib import Path
 import shutil
-
-from allotropy.parsers.utils.uuids import random_uuid_str
+import uuid
 
 PathType = Path | str
 
@@ -19,7 +18,7 @@ def _files_equal(path1: PathType, path2: PathType) -> bool:
 
 def _get_backup_path(path: PathType) -> Path:
     _path = Path(path)
-    return Path(_path.parent, f".{_path.stem}.{random_uuid_str()}.bak{_path.suffix}")
+    return Path(_path.parent, f".{_path.stem}.{uuid.uuid4()!s}.bak{_path.suffix}")
 
 
 def get_original_path(path: PathType) -> Path:

src/allotropy/allotrope/schema_parser/model_class_editor.py

@@ -17,7 +17,6 @@
     get_all_schema_components,
     get_schema_definitions_mapping,
 )
-from allotropy.parsers.utils.values import assert_not_none
 
 SCHEMA_DIR_PATH = "src/allotropy/allotrope/schemas"
 SHARED_FOLDER_MODULE = "allotropy.allotrope.models.shared"
@@ -196,7 +195,10 @@ def create(contents: str) -> DataclassField:
         type_string, default_value = (
             content.split("=") if "=" in content else (content, None)
         )
-        types = _parse_field_types(assert_not_none(type_string))
+        if type_string is None:
+            msg = f"Expected non-null type string for field {name}."
+            raise ValueError(msg)
+        types = _parse_field_types(type_string)
 
         return DataclassField(name, default_value, types)
 

src/allotropy/allotrope/schema_parser/path_util.py

@@ -4,8 +4,6 @@
 import re
 from typing import Any
 
-from allotropy.exceptions import AllotropeValidationError
-
 ALLOTROPE_DIR: Path = Path(__file__).parent.parent
 ALLOTROPY_DIR: Path = ALLOTROPE_DIR.parent
 ROOT_DIR: Path = ALLOTROPE_DIR.parent.parent.parent
@@ -74,10 +72,10 @@ def get_schema_path_from_manifest(manifest: str) -> Path:
 def get_schema_path_from_asm(asm_dict: Mapping[str, Any]) -> Path:
     if "$asm.manifest" not in asm_dict:
         msg = "File is not valid ASM - missing $asm.manifest field"
-        raise AllotropeValidationError(msg)
+        raise ValueError(msg)
     if not isinstance(asm_dict["$asm.manifest"], str):
         msg = f"File is not valid ASM - $asm.manifest is not a string: {asm_dict['$asm.manifest']}"
-        raise AllotropeValidationError(msg)
+        raise ValueError(msg)
     return get_schema_path_from_manifest(asm_dict["$asm.manifest"])
 
 

src/allotropy/allotrope/schemas.py

@@ -12,9 +12,10 @@
     get_schema_path_from_manifest,
     SHARED_SCHEMAS_DEFINITIONS_PATH,
 )
-from allotropy.constants import DEFAULT_ENCODING
 from allotropy.exceptions import AllotropeSerializationError, AllotropeValidationError
 
+DEFAULT_ENCODING = "UTF-8"
+
 # Override format checker to remove "uri-reference" check, which ASM schemas fail against.
 FORMAT_CHECKER = copy.deepcopy(
     jsonschema.validators.Draft202012Validator.FORMAT_CHECKER

tests/allotrope/schema_parser/generate_schemas_test.py

@@ -1,5 +1,4 @@
 from pathlib import Path
-import re
 
 import pytest
 
@@ -20,33 +19,11 @@ def _get_schema_paths() -> list[Path]:
     ]
 
 
-def _pick_subset(paths: list[Path], count: int = 8) -> list[Path]:
-    if not paths:
-        return []
-    if len(paths) <= count:
-        return paths
-    step = max(1, len(paths) // count)
-    # Evenly sample across the set for broad coverage
-    return [paths[i] for i in range(0, len(paths), step)][:count]
-
-
-def test_generate_schemas_smoke_subset() -> None:
-    """Fast smoke test over a representative subset of schemas."""
-    paths = _get_schema_paths()
-    subset = _pick_subset(paths, count=8)
-    # Build a regex that matches exactly any of the chosen relative schema paths
-    escaped = [re.escape(str(p)) for p in subset]
-    pattern = rf"^({'|'.join(escaped)})$"
-    models_changed = generate_schemas(dry_run=True, schema_regex=pattern)
-    assert (
-        not models_changed
-    ), f"Expected no models files to have changed by generate-schemas script, found changes in: {models_changed}.\nPlease run 'hatch run scripts:generate-schemas' and validate the changes."
-
-
 @pytest.mark.long
-def test_generate_schemas_runs_to_completion() -> None:
-    """Full run once across all schemas. Marked long for optional execution."""
-    models_changed = generate_schemas(dry_run=True)
+@pytest.mark.parametrize("schema_path", _get_schema_paths())
+def test_generate_schemas_runs_to_completion(schema_path: Path) -> None:
+    """Test that schema generation produces no unexpected changes."""
+    models_changed = generate_schemas(dry_run=True, schema_regex=str(schema_path))
     assert (
         not models_changed
     ), f"Expected no models files to have changed by generate-schemas script, found changes in: {models_changed}.\nPlease run 'hatch run scripts:generate-schemas' and validate the changes."