Use regex variant when checking schema validity

Rather than relying directly on `validator_cls.check_schema`,
reimplement this functionality in-tree here to allow for
customizations to the validator class before it is run on the user's
schema.

A new test uses an invalid regex under the regress unicode engine
which is valid in the python engine to ensure that consistent checking
is applied.

Manual testing revealed that the `_fail()` message production for
SchemaError was showing error information twice, which is fixed
without a new test to guarantee the new behavior.

Author: sirosen

src/check_jsonschema/checker.py

@@ -59,7 +59,7 @@ def get_validator(
         except SchemaParseError as e:
             self._fail("Error: schemafile could not be parsed as JSON", e)
         except jsonschema.SchemaError as e:
-            self._fail(f"Error: schemafile was not valid: {e}\n", e)
+            self._fail("Error: schemafile was not valid\n", e)
         except UnsupportedUrlScheme as e:
             self._fail(f"Error: {e}\n", e)
         except Exception as e:

src/check_jsonschema/regex_variants.py

@@ -70,10 +70,7 @@ def pattern_keyword(
         if not validator.is_type(instance, "string"):
             return
 
-        try:
-            regress_pattern = self._compile_pattern(pattern)
-        except regress.RegressError:
-            yield jsonschema.ValidationError(f"pattern {pattern!r} failed to compile")
+        regress_pattern = self._compile_pattern(pattern)
         if not regress_pattern.find(instance):
             yield jsonschema.ValidationError(f"{instance!r} does not match {pattern!r}")
 
@@ -120,10 +117,7 @@ def pattern_keyword(
         if not validator.is_type(instance, "string"):
             return
 
-        try:
-            re_pattern = re.compile(pattern)
-        except re.error:
-            yield jsonschema.ValidationError(f"pattern {pattern!r} failed to compile")
+        re_pattern = re.compile(pattern)
         if not re_pattern.search(instance):
             yield jsonschema.ValidationError(f"{instance!r} does not match {pattern!r}")
 

src/check_jsonschema/schema_loader/main.py

@@ -170,7 +170,12 @@ def _get_validator(
         if self.validator_class is None:
             # get the correct validator class and check the schema under its metaschema
             validator_cls = jsonschema.validators.validator_for(schema)
-            validator_cls.check_schema(schema, format_checker=format_checker)
+            _check_schema(
+                validator_cls,
+                schema,
+                format_checker=format_checker,
+                regex_impl=regex_impl,
+            )
         else:
             # for a user-provided validator class, don't check_schema
             # on the grounds that it might *not* be valid but the user wants to use
@@ -197,6 +202,32 @@ def _get_validator(
         return t.cast(jsonschema.protocols.Validator, validator)
 
 
+def _check_schema(
+    validator_cls: type[jsonschema.protocols.Validator],
+    schema: dict[str, t.Any],
+    *,
+    format_checker: jsonschema.FormatChecker | None,
+    regex_impl: RegexImplementation,
+) -> None:
+    """A variant definition of Validator.check_schema which uses the regex
+    implementation and format checker specified."""
+    schema_validator_cls = jsonschema.validators.validator_for(
+        validator_cls.META_SCHEMA, default=validator_cls
+    )
+    schema_validator_cls = _extend_with_pattern_implementation(
+        schema_validator_cls, regex_impl
+    )
+
+    if format_checker is None:
+        format_checker = schema_validator_cls.FORMAT_CHECKER
+
+    schema_validator = schema_validator_cls(
+        validator_cls.META_SCHEMA, format_checker=format_checker
+    )
+    for error in schema_validator.iter_errors(schema):
+        raise jsonschema.exceptions.SchemaError.create_from(error)
+
+
 class BuiltinSchemaLoader(SchemaLoader):
     def __init__(self, schema_name: str, *, base_uri: str | None = None) -> None:
         self.schema_name = schema_name

tests/acceptance/test_invalid_schema_files.py

@@ -29,3 +29,15 @@ def test_checker_invalid_schemafile_scheme(run_line, tmp_path):
     res = run_line(["check-jsonschema", "--schemafile", f"ftp://{foo}", str(bar)])
     assert res.exit_code == 1
     assert "only supports http, https" in res.stderr
+
+
+def test_checker_invalid_schemafile_due_to_bad_regex(run_line, tmp_path):
+    foo = tmp_path / "foo.json"
+    bar = tmp_path / "bar.json"
+    # too many backslash escapes -- not a valid Unicode-mode regex
+    foo.write_text(r'{"properties": {"foo": {"pattern": "\\\\p{N}"}}}')
+    bar.write_text("{}")
+
+    res = run_line(["check-jsonschema", "--schemafile", str(foo), str(bar)])
+    assert res.exit_code == 1
+    assert "schemafile was not valid" in res.stderr