Instrument regress-backed pattern evaluation

When an alternate RegexImplementation is selected, it will be passed
down to the code which builds a Validator. The Validator is extended
with the keyword validator provided by the RegexImplementation.

Because this uses the `extend()` interface, a test which subclassed a
validator broke -- this is documented in `jsonschema` as unsupported
usage, so the test simply had to be updated to use supported
interfaces.

Author: sirosen

src/check_jsonschema/checker.py

@@ -11,6 +11,7 @@
 from .formats import FormatOptions
 from .instance_loader import InstanceLoader
 from .parsers import ParseError
+from .regex_variants import RegexImplementation
 from .reporter import Reporter
 from .result import CheckResult
 from .schema_loader import SchemaLoaderBase, SchemaParseError, UnsupportedUrlScheme
@@ -29,14 +30,16 @@ def __init__(
         reporter: Reporter,
         *,
         format_opts: FormatOptions,
+        regex_impl: RegexImplementation,
         traceback_mode: str = "short",
         fill_defaults: bool = False,
     ) -> None:
         self._schema_loader = schema_loader
         self._instance_loader = instance_loader
         self._reporter = reporter
 
-        self._format_opts = format_opts if format_opts is not None else FormatOptions()
+        self._format_opts = format_opts
+        self._regex_impl = regex_impl
         self._traceback_mode = traceback_mode
         self._fill_defaults = fill_defaults
 
@@ -51,7 +54,7 @@ def get_validator(
     ) -> jsonschema.protocols.Validator:
         try:
             return self._schema_loader.get_validator(
-                path, doc, self._format_opts, self._fill_defaults
+                path, doc, self._format_opts, self._regex_impl, self._fill_defaults
             )
         except SchemaParseError as e:
             self._fail("Error: schemafile could not be parsed as JSON", e)

src/check_jsonschema/cli/main_command.py

@@ -9,9 +9,10 @@
 
 from ..catalog import CUSTOM_SCHEMA_NAMES, SCHEMA_CATALOG
 from ..checker import SchemaChecker
-from ..formats import KNOWN_FORMATS, RegexVariantName
+from ..formats import KNOWN_FORMATS
 from ..instance_loader import InstanceLoader
 from ..parsers import SUPPORTED_FILE_FORMATS
+from ..regex_variants import RegexImplementation, RegexVariantName
 from ..reporter import REPORTER_BY_NAME, Reporter
 from ..schema_loader import (
     BuiltinSchemaLoader,
@@ -327,6 +328,7 @@ def build_checker(args: ParseResult) -> SchemaChecker:
         instance_loader,
         reporter,
         format_opts=args.format_opts,
+        regex_impl=RegexImplementation(args.regex_variant),
         traceback_mode=args.traceback_mode,
         fill_defaults=args.fill_defaults,
     )

src/check_jsonschema/cli/parse_result.py

@@ -8,7 +8,7 @@
 import jsonschema
 
 from ..formats import FormatOptions
-from ..regex_variants import RegexVariantName
+from ..regex_variants import RegexImplementation, RegexVariantName
 from ..transforms import Transform
 
 if sys.version_info >= (3, 8):
@@ -99,7 +99,7 @@ def set_validator(
     @property
     def format_opts(self) -> FormatOptions:
         return FormatOptions(
+            regex_impl=RegexImplementation(self.regex_variant),
             enabled=not self.disable_all_formats,
-            regex_variant=self.regex_variant,
             disabled_formats=self.disable_formats,
         )

src/check_jsonschema/formats/__init__.py

@@ -5,7 +5,7 @@
 import jsonschema
 import jsonschema.validators
 
-from ..regex_variants import RegexImplementation, RegexVariantName
+from ..regex_variants import RegexImplementation
 from .implementations import validate_rfc3339, validate_time
 
 # all known format strings except for a selection from draft3 which have either
@@ -40,12 +40,12 @@ class FormatOptions:
     def __init__(
         self,
         *,
+        regex_impl: RegexImplementation,
         enabled: bool = True,
-        regex_variant: RegexVariantName = RegexVariantName.default,
         disabled_formats: tuple[str, ...] = (),
     ) -> None:
         self.enabled = enabled
-        self.regex_variant = regex_variant
+        self.regex_impl = regex_impl
         self.disabled_formats = disabled_formats
 
 
@@ -72,8 +72,7 @@ def make_format_checker(
 
     # replace the regex check
     del checker.checkers["regex"]
-    regex_impl = RegexImplementation(opts.regex_variant)
-    checker.checks("regex")(regex_impl.check_format)
+    checker.checks("regex")(opts.regex_impl.check_format)
     checker.checks("date-time")(validate_rfc3339)
     checker.checks("time")(validate_time)
 

src/check_jsonschema/regex_variants.py

@@ -2,6 +2,7 @@
 import re
 import typing as t
 
+import jsonschema
 import regress
 
 
@@ -29,3 +30,32 @@ def check_format(self, instance: t.Any) -> bool:
             return False
 
         return True
+
+    def pattern_keyword(
+        self, validator: t.Any, pattern: str, instance: str, schema: t.Any
+    ) -> t.Iterator[jsonschema.ValidationError]:
+        if not validator.is_type(instance, "string"):
+            return
+
+        if self.variant == RegexVariantName.default:
+            try:
+                regress_pattern = regress.Regex(pattern)
+            except regress.RegressError:  # type: ignore[attr-defined]
+                yield jsonschema.ValidationError(
+                    f"pattern {pattern!r} failed to compile"
+                )
+            if not regress_pattern.find(instance):
+                yield jsonschema.ValidationError(
+                    f"{instance!r} does not match {pattern!r}"
+                )
+        else:
+            try:
+                re_pattern = re.compile(pattern)
+            except re.error:
+                yield jsonschema.ValidationError(
+                    f"pattern {pattern!r} failed to compile"
+                )
+            if not re_pattern.search(instance):
+                yield jsonschema.ValidationError(
+                    f"{instance!r} does not match {pattern!r}"
+                )

src/check_jsonschema/schema_loader/main.py

@@ -11,6 +11,7 @@
 from ..builtin_schemas import get_builtin_schema
 from ..formats import FormatOptions, make_format_checker
 from ..parsers import ParserSet
+from ..regex_variants import RegexImplementation
 from ..utils import is_url_ish
 from .errors import UnsupportedUrlScheme
 from .readers import HttpSchemaReader, LocalSchemaReader, StdinSchemaReader
@@ -45,12 +46,23 @@ def set_defaults_then_validate(
     )
 
 
+def _extend_with_pattern_implementation(
+    validator_class: type[jsonschema.protocols.Validator],
+    regex_impl: RegexImplementation,
+) -> type[jsonschema.Validator]:
+    return jsonschema.validators.extend(
+        validator_class,
+        {"pattern": regex_impl.pattern_keyword},
+    )
+
+
 class SchemaLoaderBase:
     def get_validator(
         self,
         path: pathlib.Path | str,
         instance_doc: dict[str, t.Any],
         format_opts: FormatOptions,
+        regex_impl: RegexImplementation,
         fill_defaults: bool,
     ) -> jsonschema.protocols.Validator:
         raise NotImplementedError
@@ -124,14 +136,16 @@ def get_validator(
         path: pathlib.Path | str,
         instance_doc: dict[str, t.Any],
         format_opts: FormatOptions,
+        regex_impl: RegexImplementation,
         fill_defaults: bool,
     ) -> jsonschema.protocols.Validator:
-        return self._get_validator(format_opts, fill_defaults)
+        return self._get_validator(format_opts, regex_impl, fill_defaults)
 
     @functools.lru_cache
     def _get_validator(
         self,
         format_opts: FormatOptions,
+        regex_impl: RegexImplementation,
         fill_defaults: bool,
     ) -> jsonschema.protocols.Validator:
         retrieval_uri = self.get_schema_retrieval_uri()
@@ -168,6 +182,9 @@ def _get_validator(
         if fill_defaults:
             validator_cls = _extend_with_default(validator_cls)
 
+        # set the regex variant for 'pattern' keywords
+        validator_cls = _extend_with_pattern_implementation(validator_cls, regex_impl)
+
         # now that we know it's safe to try to create the validator instance, do it
         validator = validator_cls(
             schema,
@@ -206,6 +223,7 @@ def get_validator(
         path: pathlib.Path | str,
         instance_doc: dict[str, t.Any],
         format_opts: FormatOptions,
+        regex_impl: RegexImplementation,
         fill_defaults: bool,
     ) -> jsonschema.protocols.Validator:
         schema_validator = jsonschema.validators.validator_for(instance_doc)

tests/acceptance/test_custom_validator_class.py

@@ -66,24 +66,32 @@ def _foo_module(mock_module):
         """\
 import jsonschema
 
-class MyValidator:
-    def __init__(self, schema, *args, **kwargs):
-        self.schema = schema
-        self.real_validator = jsonschema.validators.Draft7Validator(
-            schema, *args, **kwargs
-        )
-
-    def iter_errors(self, data, *args, **kwargs):
-        yield from self.real_validator.iter_errors(data, *args, **kwargs)
-        for event in data["events"]:
-            if "Occult" in event["title"]:
+
+def check_occult_properties(validator, properties, instance, schema):
+    if not validator.is_type(instance, "object"):
+        return
+
+    for property, subschema in properties.items():
+        if property in instance:
+            if property == "title" and "Occult" in instance["title"]:
                 yield jsonschema.exceptions.ValidationError(
                     "Error! Occult event detected! Run!",
-                    validator=None,
+                    validator=validator,
                     validator_value=None,
-                    instance=event,
-                    schema=self.schema,
+                    instance=instance,
+                    schema=schema,
                 )
+            yield from validator.descend(
+                instance[property],
+                subschema,
+                path=property,
+                schema_path=property,
+            )
+
+MyValidator = jsonschema.validators.extend(
+    jsonschema.validators.Draft7Validator,
+    {"properties": check_occult_properties},
+)
 """,
     )
 
@@ -115,7 +123,7 @@ def test_custom_validator_class_can_detect_custom_conditions(run_line, tmp_path)
             str(doc),
         ],
     )
-    assert result.exit_code == 1  # fail
+    assert result.exit_code == 1, result.stdout  # fail
     assert "Occult event detected" in result.stdout, result.stdout