Normalize mime types for validation. RDF/XML support

Author: avillar

ogc/bblocks/known-mimetypes.yaml

@@ -36,3 +36,12 @@ xml:
     - application/xml
   extensions:
     - xml
+
+yaml:
+  label: YAML
+  mimeType: application/x-yaml
+  aliases:
+    - text/yaml
+  extensions:
+    - yaml
+    - yml

ogc/bblocks/validate.py

@@ -13,6 +13,7 @@
 from mako import exceptions as mako_exceptions, template as mako_template
 from ogc.na.util import is_url
 
+from ogc.bblocks import mimetypes
 from ogc.bblocks.models import BuildingBlock, BuildingBlockRegister
 from ogc.bblocks.util import sanitize_filename
 from ogc.bblocks.validation import Validator, ValidationItemSourceType, ValidationReportSection, ValidationItemSource, \
@@ -151,7 +152,8 @@ def _validate_resource(bblock: BuildingBlock,
                        require_fail: bool | None = None,
                        resource_url: str | None = None,
                        example_index: tuple[int, int] | None = None,
-                       prefixes: dict[str, str] | None = None) -> ValidationReportItem | None:
+                       prefixes: dict[str, str] | None = None,
+                       file_format: str | None = None) -> ValidationReportItem | None:
     if require_fail is None:
         require_fail = filename.stem.endswith('-fail') and not example_index
 
@@ -165,7 +167,7 @@ def _validate_resource(bblock: BuildingBlock,
             filename=output_filename,
             example_index=int(example_idx),
             snippet_index=int(snippet_idx),
-            language=filename.suffix[1:],
+            language=file_format,
             source_url=resource_url,
         )
     else:
@@ -186,7 +188,8 @@ def _validate_resource(bblock: BuildingBlock,
                                         base_uri=base_uri,
                                         additional_shacl_closures=additional_shacl_closures,
                                         schema_ref=schema_ref,
-                                        prefixes=prefixes)
+                                        prefixes=prefixes,
+                                        file_format=file_format)
             any_validator_run = any_validator_run or (result is not False)
 
     except Exception as unknown_exc:
@@ -279,6 +282,7 @@ def validate_test_resources(bblock: BuildingBlock,
             resource_contents=extra_test_resource['contents'],
             require_fail=extra_test_resource.get('require-fail', False),
             resource_url=extra_test_resource['ref'] if isinstance(extra_test_resource['ref'], str) else None,
+            file_format=mimetypes.from_extension(fn.suffix[1:]),
         )
         if test_result:
             all_results.append(test_result)
@@ -301,8 +305,9 @@ def validate_test_resources(bblock: BuildingBlock,
                 elif not add_snippets_formats:
                     add_snippets_formats = []
 
+                extension = FORMAT_ALIASES.get(snippet['language'], snippet['language']).replace('/', '.')
                 fn = bblock.files_path / (f"example_{example_id + 1}_{snippet_id + 1}"
-                                          f".{FORMAT_ALIASES.get(snippet['language'], snippet['language'])}")
+                                          f".{extension}")
 
                 output_fn = output_dir / sanitize_filename(example.get('base-output-filename', fn.name))
                 i = 0
@@ -314,6 +319,9 @@ def validate_test_resources(bblock: BuildingBlock,
                     f.write(code)
 
                 snippet['path'] = output_fn
+                snippet_language = snippet.get('language')
+                if snippet_language:
+                    snippet_language = mimetypes.normalize(snippet_language)
 
                 example_result = _validate_resource(
                     bblock=bblock,
@@ -327,6 +335,7 @@ def validate_test_resources(bblock: BuildingBlock,
                     resource_url=snippet.get('ref'),
                     require_fail=False,
                     prefixes=example.get('prefixes'),
+                    file_format=snippet_language,
                 )
                 if example_result:
                     all_results.append(example_result)

ogc/bblocks/validation/__init__.py

@@ -16,14 +16,17 @@ def __init__(self, bblock: BuildingBlock, register: BuildingBlockRegister):
         self.register = register
 
     @abstractmethod
-    def validate(self, filename: Path, output_filename: Path,
+    def validate(self,
+                 filename: Path,
+                 output_filename: Path,
                  report: ValidationReportItem,
                  contents: str | None = None,
                  schema_ref: str | None = None,
                  base_uri: str | None = None,
                  resource_url: str | None = None,
                  require_fail: bool | None = None,
                  prefixes: dict[str, str] | None = None,
+                 file_format: str | None = None,
                  **kwargs) -> bool | None:
         raise NotImplementedError
 

ogc/bblocks/validation/json_.py

@@ -88,9 +88,11 @@ def __init__(self, bblock: BuildingBlock, register: BuildingBlockRegister):
     def validate(self, filename: Path, output_filename: Path, report: ValidationReportItem,
                  contents: str | None = None,
                  schema_ref: str | None = None,
+                 file_format: str | None = None,
                  **kwargs) -> bool | None:
 
-        if filename.suffix not in ('.json', '.jsonld', '.yaml', '.yml'):
+        if filename.suffix not in ('.json', '.jsonld', '.yaml', '.yml')\
+                and file_format not in ('application/json', 'application/x-yaml'):
             return False
 
         file_from = 'examples' if report.source.type == ValidationItemSourceType.EXAMPLE else 'test resources'

ogc/bblocks/validation/rdf.py

@@ -20,6 +20,13 @@
 from ogc.bblocks.validation.uplift import Uplifter
 
 
+NATIVE_RDF_LANGS = {
+    'application/ld+json': 'jsonld',
+    'text/turtle': 'ttl',
+    'application/rdf+xml': 'xml',
+}
+
+
 def shacl_validate(g: Graph, s: Graph, ont_graph: Graph | None = None) \
         -> tuple[bool, Graph, str, dict[pyshacl.Shape, Sequence[Node]]]:
     validator = pyshacl.Validator(g, shacl_graph=s, ont_graph=ont_graph, options={
@@ -131,9 +138,10 @@ def __init__(self, bblock: BuildingBlock, register: BuildingBlockRegister):
     def _load_graph(self, filename: Path, output_filename: Path, report: ValidationReportItem,
                     contents: str | None = None,
                     base_uri: str | None = None,
-                    prefixes: dict[str, str] | None = None) -> Graph | None | bool:
+                    prefixes: dict[str, str] | None = None,
+                    file_format: str | None = None) -> Graph | None | bool:
         graph = False
-        if filename.suffix == '.json':
+        if filename.suffix == '.json' or file_format == 'application/json':
             if self.jsonld_error:
                 report.add_entry(ValidationReportEntry(
                     section=ValidationReportSection.JSON_LD,
@@ -231,7 +239,7 @@ def _load_graph(self, filename: Path, output_filename: Path, report: ValidationR
             if graph:
                 graph = self.uplifter.post_uplift(report, graph)
 
-        elif output_filename.suffix == '.jsonld':
+        elif output_filename.suffix == '.jsonld' or file_format == 'application/ld+json':
 
             if contents:
                 jsonld_doc = load_yaml(content=contents)
@@ -258,9 +266,11 @@ def _load_graph(self, filename: Path, output_filename: Path, report: ValidationR
             graph = Graph().parse(data=json.dumps(jsonld_doc), format='json-ld', base=base_uri)
             graph = self.uplifter.post_uplift(report, graph)
 
-        elif output_filename.suffix in ('.ttl', '.jsonld'):
+        elif (output_filename.suffix in ('.ttl', '.jsonld', '.rdf')
+              or file_format in NATIVE_RDF_LANGS):
             file_from = 'examples' if report.source.type == ValidationItemSourceType.EXAMPLE else 'test resources'
-            rdf_format = 'ttl' if output_filename.suffix == '.ttl' else 'json-ld'
+            rdf_format = NATIVE_RDF_LANGS.get(file_format,
+                                              'ttl' if output_filename.suffix == '.ttl' else 'json-ld')
             try:
                 if contents:
                     # Prepend prefixes
@@ -299,8 +309,10 @@ def validate(self, filename: Path, output_filename: Path, report: ValidationRepo
                  base_uri: str | None = None,
                  additional_shacl_closures: list[str | Path] | None = None,
                  prefixes: dict[str, str] | None = None,
+                 file_format: str | None = None,
                  **kwargs) -> bool | None:
-        graph = self._load_graph(filename, output_filename, report, contents, base_uri, prefixes)
+        graph = self._load_graph(filename, output_filename, report,
+                                 contents, base_uri, prefixes, file_format=file_format)
 
         if graph is False:
             return False