Use common JSON schema validation

Author: vjovanov

substratevm/mx.substratevm/mx_substratevm.py

@@ -38,6 +38,7 @@
 import fnmatch
 import collections
 from io import StringIO
+from typing import Iterable, Tuple
 
 import mx
 import mx_compiler
@@ -525,13 +526,6 @@ def help_stdout_check(output):
 
     with Task('Validate JSON build info', tasks, tags=[GraalTags.helloworld]) as t:
         if t:
-            import json
-            try:
-                from jsonschema import validate as json_validate
-                from jsonschema.exceptions import ValidationError, SchemaError
-            except ImportError:
-                mx.abort('Unable to import jsonschema')
-
             json_and_schema_file_pairs = [
                 ('build-artifacts.json', 'build-artifacts-schema-v0.9.0.json'),
                 ('build-output.json', 'build-output-schema-v0.9.4.json'),
@@ -540,19 +534,7 @@ def help_stdout_check(output):
             build_output_file = join(svmbuild_dir(), 'build-output.json')
             helloworld(['--output-path', svmbuild_dir()] + svm_experimental_options([f'-H:BuildOutputJSONFile={build_output_file}', '-H:+GenerateBuildArtifactsFile']))
 
-            try:
-                for json_file, schema_file in json_and_schema_file_pairs:
-                    with open(join(svmbuild_dir(), json_file)) as f:
-                        json_contents = json.load(f)
-                    with open(join(suite.dir, '..', 'docs', 'reference-manual', 'native-image', 'assets', schema_file)) as f:
-                        schema_contents = json.load(f)
-                    json_validate(json_contents, schema_contents)
-            except IOError as e:
-                mx.abort(f'Unable to load JSON build info: {e}')
-            except ValidationError as e:
-                mx.abort(f'Unable to validate JSON build info against the schema: {e}')
-            except SchemaError as e:
-                mx.abort(f'JSON schema not valid: {e}')
+            validate_dir_files_with_file_schema_pairs(svmbuild_dir(), json_and_schema_file_pairs)
 
     with Task('java agent tests', tasks, tags=[GraalTags.java_agent]) as t:
         if t:
@@ -2494,7 +2476,7 @@ def compute_graal_compiler_flags_map(self):
 
         return graal_compiler_flags_map
 
-def get_jsonschema_validator(schema_path):
+def create_jsonschema_validator(schema_path):
     """Create and return a jsonschema Validator for the schema at the given file path. Abort on missing jsonschema or invalid schema."""
     import json
     try:
@@ -2510,7 +2492,7 @@ def get_jsonschema_validator(schema_path):
     except ImportError as e:
         mx.abort(
             'Python module "jsonschema" is required to validate reachability metadata but was not found. '
-            'Install it with: "python3 -m pip install --user jsonschema" (or "pip3 install jsonschema"). '
+            'Install it with: \n\npython3 -m pip install --user jsonschema \n\n or \n\npip3 install jsonschema). '
             f'Original error: {e}')
     try:
         if hasattr(jsonschema, 'Draft202012Validator'):
@@ -2520,8 +2502,8 @@ def get_jsonschema_validator(schema_path):
     except (jsonschema.exceptions.SchemaError, TypeError) as e:
         mx.abort(f'Invalid reachability metadata schema: {e}')
 
-def validate_json_file_against_schema(validator, file_path):
-    """Validates a JSON file against the provided Validator. Returns a list of error strings; empty if valid."""
+def validate_json_file_with_validator(validator, file_path):
+    """Validates a JSON file against the provided Validator. Returns a list of detailed error strings; empty if valid."""
     import json
     try:
         with open(file_path, 'r', encoding='utf-8') as f:
@@ -2531,12 +2513,55 @@ def validate_json_file_against_schema(validator, file_path):
     except OSError as e:
         mx.abort(f'I/O error: {e}')
 
+    # Use jsonschema's ValidationError string representation to produce messages in the form:
+    # Failed validating '<validator>' in schema[...] :
+    #     { ... failing subschema ... }
+    #
+    # On instance[...] :
+    #     <offending instance>
     errors = []
     for err in validator.iter_errors(data):
-        path = '/'.join([str(p) for p in err.path]) if err.path else '#'
-        errors.append(f'{path}: {err.message}')
+        errors.append(str(err))
     return errors
 
+def validate_dir_files_with_file_schema_pairs(dir_with_json: str, json_and_schema_file_pairs: Iterable[Tuple[str, str]]) -> None:
+    """
+    Validate JSON files in a directory against corresponding JSON Schemas.
+
+    This scans the given directory for each JSON filename listed in json_and_schema_file_pairs.
+    For every file that exists, it validates the file against the associated schema located in
+    the Native Image docs assets directory. Validation uses a pre-built jsonschema.Validator
+    per schema to ensure consistent behavior and error reporting.
+
+    Parameters:
+    - dir_with_json: Directory path containing JSON files to validate.
+    - json_and_schema_file_pairs: Iterable of (json_filename, schema_filename) pairs.
+
+    Behavior:
+    - Logs each validation attempt.
+    - Accumulates all validation errors across files.
+    - Calls mx.abort with a detailed message if any error is found; otherwise returns None.
+    """
+    assets_dir = join(suite.dir, '..', 'docs', 'reference-manual', 'native-image', 'assets')
+
+    # Build validators for all known schema files using CE helper to ensure uniform behavior and error reporting.
+    validators = {}
+    for _, schema_file in json_and_schema_file_pairs:
+        schema_path = join(assets_dir, schema_file)
+        validators[schema_file] = create_jsonschema_validator(schema_path)
+
+    # Validate any present config files
+    validation_errors = []
+    for json_filename, schema_file in json_and_schema_file_pairs:
+        json_path = join(dir_with_json, json_filename)
+        if exists(json_path):
+            mx.log(f'Validating {json_path} against {schema_file}...')
+            errs = validate_json_file_with_validator(validators[schema_file], json_path)
+            validation_errors.extend(errs)
+
+    if validation_errors:
+        mx.abort('Unable to validate JSON file against the schema:\n\n' + '\n\n'.join(validation_errors))
+
 class SubstrateCompilerFlagsBuildTask(mx.ArchivableBuildTask):
     def __init__(self, subject, args):
         mx.ArchivableBuildTask.__init__(self, subject, args, 1)