feat: validate schemas using pydantic_schemaorg

blurry/__init__.py

@@ -15,6 +15,7 @@
 from jinja2 import select_autoescape
 from livereload import Server
 from rich import print
+from rich.console import Console
 
 from blurry.async_typer import AsyncTyper
 from blurry.cli import print_blurry_name
@@ -25,6 +26,7 @@
 from blurry.open_graph import open_graph_meta_tags
 from blurry.plugins import discovered_html_plugins
 from blurry.plugins import discovered_jinja_filter_plugins
+from blurry.schema_validation import validate_front_matter_as_schema
 from blurry.settings import get_build_directory
 from blurry.settings import get_content_directory
 from blurry.settings import get_templates_directory
@@ -46,6 +48,8 @@ def json_converter_with_dates(item: Any) -> None | str:
         return item.strftime("%Y-%M-%D")
 
 
+warning_console = Console(stderr=True, style="bold yellow")
+
 print_blurry_name()
 print_plugin_table()
 
@@ -147,6 +151,9 @@ async def write_html_file(
         format_schema_data(schema_variables),
         default=json_converter_with_dates,
     )
+
+    validate_front_matter_as_schema(file_data.path, front_matter, warning_console)
+
     schema_type_tag = f'<script type="application/ld+json">{schema_data}</script>'
 
     template_context = {

blurry/schema_validation.py

@@ -0,0 +1,58 @@
+import importlib
+from collections.abc import MutableMapping
+from pathlib import Path
+
+from pydantic.v1 import ValidationError
+from rich.console import Console
+
+from blurry.settings import SETTINGS
+from blurry.settings import update_settings
+
+
+class Config:
+    extra = "forbid"
+
+
+def validate_front_matter_as_schema(
+    path: Path, front_matter: MutableMapping, console: Console
+):
+    """
+    Validates schema data using pydantic_schemaorg, disallowing extra fields
+    """
+    update_settings()
+
+    schema_type = front_matter["@type"]
+
+    # Import pydantic_schemaorg model
+    try:
+        pydantic_schemaorg_model_module = importlib.import_module(
+            f"pydantic_schemaorg.{schema_type}"
+        )
+    except ModuleNotFoundError:
+        console.print(
+            f"{path}: Could not find Schema type for {schema_type}. Skipping."
+        )
+        return
+
+    schema_model = getattr(pydantic_schemaorg_model_module, schema_type)
+
+    # Create new Pydantic model that forbids extra fields
+    class NonExtraSchemaModel(schema_model, extra="forbid"):  # type: ignore
+        pass
+
+    # Validate model and print errors
+    try:
+        non_schema_variable_prefix = SETTINGS["FRONTMATTER_NON_SCHEMA_VARIABLE_PREFIX"]
+        schema_front_matter = {
+            k: v
+            for k, v in front_matter.items()
+            if not k.startswith(non_schema_variable_prefix)
+        }
+        NonExtraSchemaModel(**schema_front_matter)
+    except ValidationError as e:
+        for error in e.errors():
+            msg = error["msg"]
+            loc = error["loc"]
+            console.print(
+                f"{path}: {schema_type} schema validation error: {msg}: {loc}"
+            )