Add optional ecma-regex backend for strict OpenAPI pattern validation

Author: p1c2u

.github/workflows/python-tests.yml

@@ -62,6 +62,46 @@ jobs:
       - name: Upload coverage
         uses: codecov/codecov-action@v5
 
+  tests_no_extras:
+    name: "py3.14 no extras"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Set up Python 3.14
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.14"
+
+      - name: Get full Python version
+        id: full-python-version
+        run: echo ::set-output name=version::$(python -c "import sys; print('-'.join(str(v) for v in sys.version_info))")
+
+      - name: Set up poetry
+        uses: Gr1N/setup-poetry@v9
+
+      - name: Configure poetry
+        run: poetry config virtualenvs.in-project true
+
+      - name: Set up cache
+        uses: actions/cache@v5
+        id: cache
+        with:
+          path: .venv
+          key: venv-no-extras-${{ steps.full-python-version.outputs.version }}-${{ hashFiles('**/poetry.lock') }}
+
+      - name: Ensure cache is healthy
+        if: steps.cache.outputs.cache-hit == 'true'
+        run: timeout 10s poetry run pip --version || rm -rf .venv
+
+      - name: Install dependencies
+        run: poetry install
+
+      - name: Test fallback regex behavior
+        env:
+          PYTEST_ADDOPTS: "--color=yes"
+        run: poetry run pytest tests/integration/test_validators.py -k pattern
+
   static_checks:
     name: "Static checks"
     runs-on: ubuntu-latest

README.rst

@@ -127,7 +127,6 @@ unresolved-metaschema fallback warnings.
    assert validator_for(schema) is OAS31Validator
    assert validator_for(schema32) is OAS32Validator
 
-
 Binary Data Semantics
 =====================
 
@@ -189,6 +188,19 @@ Quick Reference
      - Fail
      - Same semantics as OAS 3.1
 
+
+Regex Behavior
+==============
+
+By default, ``pattern`` handling follows host Python regex behavior.
+For ECMAScript-oriented regex validation and matching (via ``regress``),
+install the optional extra:
+
+.. code-block:: console
+
+   pip install "openapi-schema-validator[ecma-regex]"
+
+
 For more details read about `Validation <https://openapi-schema-validator.readthedocs.io/en/latest/validation.html>`__.
 
 

docs/validation.rst

@@ -138,7 +138,8 @@ Malformed schema values (for example an invalid regex in ``pattern``) raise
 If you instantiate a validator class directly and call ``.validate(...)``,
 schema checking is not performed automatically, matching
 ``jsonschema`` validator-class behavior.
-For malformed regex patterns this may raise a lower-level regex error.
+For malformed regex patterns this may raise a lower-level regex error
+(default mode) or ``ValidationError`` from the validator (ECMAScript mode).
 
 Use ``<ValidatorClass>.check_schema(schema)`` first when you need deterministic
 schema-validation errors with direct validator usage.
@@ -245,6 +246,21 @@ Quick Reference
      - Fail
      - Same semantics as OAS 3.1
 
+Regex Behavior
+--------------
+
+Pattern validation follows one of two modes:
+
+- default installation: follows host Python regex behavior
+- ``ecma-regex`` extra installed: uses ``regress`` for ECMAScript-oriented
+  regex validation and matching
+
+Install optional ECMAScript regex support with:
+
+.. code-block:: console
+
+   pip install "openapi-schema-validator[ecma-regex]"
+
 Example usage:
 
 .. code-block:: python

openapi_schema_validator/_format.py

@@ -5,6 +5,8 @@
 
 from jsonschema._format import FormatChecker
 
+from openapi_schema_validator._regex import is_valid_regex
+
 
 def is_int32(instance: object) -> bool:
     # bool inherits from int, so ensure bools aren't reported as ints
@@ -82,6 +84,12 @@ def is_password(instance: object) -> bool:
     return True
 
 
+def is_regex(instance: object) -> bool:
+    if not isinstance(instance, str):
+        return True
+    return is_valid_regex(instance)
+
+
 oas30_format_checker = FormatChecker()
 oas30_format_checker.checks("int32")(is_int32)
 oas30_format_checker.checks("int64")(is_int64)
@@ -90,6 +98,7 @@ def is_password(instance: object) -> bool:
 oas30_format_checker.checks("binary")(is_binary_pragmatic)
 oas30_format_checker.checks("byte", (binascii.Error, TypeError))(is_byte)
 oas30_format_checker.checks("password")(is_password)
+oas30_format_checker.checks("regex")(is_regex)
 
 oas30_strict_format_checker = FormatChecker()
 oas30_strict_format_checker.checks("int32")(is_int32)
@@ -101,13 +110,15 @@ def is_password(instance: object) -> bool:
     is_byte
 )
 oas30_strict_format_checker.checks("password")(is_password)
+oas30_strict_format_checker.checks("regex")(is_regex)
 
 oas31_format_checker = FormatChecker()
 oas31_format_checker.checks("int32")(is_int32)
 oas31_format_checker.checks("int64")(is_int64)
 oas31_format_checker.checks("float")(is_float)
 oas31_format_checker.checks("double")(is_double)
 oas31_format_checker.checks("password")(is_password)
+oas31_format_checker.checks("regex")(is_regex)
 
 # OAS 3.2 uses the same format checks as OAS 3.1
 oas32_format_checker = FormatChecker()
@@ -116,3 +127,4 @@ def is_password(instance: object) -> bool:
 oas32_format_checker.checks("float")(is_float)
 oas32_format_checker.checks("double")(is_double)
 oas32_format_checker.checks("password")(is_password)
+oas32_format_checker.checks("regex")(is_regex)

openapi_schema_validator/_keywords.py

@@ -6,12 +6,17 @@
 from jsonschema._keywords import allOf as _allOf
 from jsonschema._keywords import anyOf as _anyOf
 from jsonschema._keywords import oneOf as _oneOf
+from jsonschema._keywords import pattern as _pattern
 from jsonschema._utils import extras_msg
 from jsonschema._utils import find_additional_properties
 from jsonschema.exceptions import FormatError
 from jsonschema.exceptions import ValidationError
 from jsonschema.exceptions import _WrappedReferencingError
 
+from openapi_schema_validator._regex import ECMARegexSyntaxError
+from openapi_schema_validator._regex import has_ecma_regex
+from openapi_schema_validator._regex import search as regex_search
+
 
 def handle_discriminator(
     validator: Any, _: Any, instance: Any, schema: Mapping[str, Any]
@@ -159,6 +164,34 @@ def strict_type(
         yield ValidationError(f"{instance!r} is not of type {data_repr}")
 
 
+def pattern(
+    validator: Any,
+    patrn: str,
+    instance: Any,
+    schema: Mapping[str, Any],
+) -> Iterator[ValidationError]:
+    if not has_ecma_regex():
+        yield from cast(
+            Iterator[ValidationError],
+            _pattern(validator, patrn, instance, schema),
+        )
+        return
+
+    if not validator.is_type(instance, "string"):
+        return
+
+    try:
+        matches = regex_search(patrn, instance)
+    except ECMARegexSyntaxError as exc:
+        yield ValidationError(
+            f"{patrn!r} is not a valid regular expression ({exc})"
+        )
+        return
+
+    if not matches:
+        yield ValidationError(f"{instance!r} does not match {patrn!r}")
+
+
 def format(
     validator: Any,
     format: str,

openapi_schema_validator/_regex.py

@@ -0,0 +1,44 @@
+import re
+from typing import Any
+
+_REGEX_CLASS: Any = None
+_REGRESS_ERROR: type[Exception] = Exception
+
+try:
+    from regress import Regex as _REGEX_CLASS
+    from regress import RegressError as _REGRESS_ERROR
+except ImportError:  # pragma: no cover - optional dependency
+    pass
+
+
+class ECMARegexSyntaxError(ValueError):
+    pass
+
+
+def has_ecma_regex() -> bool:
+    return _REGEX_CLASS is not None
+
+
+def is_valid_regex(pattern: str) -> bool:
+    if _REGEX_CLASS is None:
+        try:
+            re.compile(pattern)
+        except re.error:
+            return False
+        return True
+
+    try:
+        _REGEX_CLASS(pattern)
+    except _REGRESS_ERROR:
+        return False
+    return True
+
+
+def search(pattern: str, instance: str) -> bool:
+    if _REGEX_CLASS is None:
+        return re.search(pattern, instance) is not None
+
+    try:
+        return _REGEX_CLASS(pattern).find(instance) is not None
+    except _REGRESS_ERROR as exc:
+        raise ECMARegexSyntaxError(str(exc)) from exc

openapi_schema_validator/validators.py

@@ -62,7 +62,7 @@ def _oas30_id_of(schema: Any) -> str:
         "minimum": _legacy_keywords.minimum_draft3_draft4,
         "maxLength": _keywords.maxLength,
         "minLength": _keywords.minLength,
-        "pattern": _keywords.pattern,
+        "pattern": oas_keywords.pattern,
         "maxItems": _keywords.maxItems,
         "minItems": _keywords.minItems,
         "uniqueItems": _keywords.uniqueItems,
@@ -118,6 +118,7 @@ def _build_oas31_validator() -> Any:
             "allOf": oas_keywords.allOf,
             "oneOf": oas_keywords.oneOf,
             "anyOf": oas_keywords.anyOf,
+            "pattern": oas_keywords.pattern,
             "description": oas_keywords.not_implemented,
             # fixed OAS fields
             "discriminator": oas_keywords.not_implemented,
@@ -180,5 +181,6 @@ def _build_oas32_validator() -> Any:
 OAS31Validator = _build_oas31_validator()
 OAS32Validator = _build_oas32_validator()
 
+OAS30Validator.check_schema = classmethod(check_openapi_schema)
 OAS31Validator.check_schema = classmethod(check_openapi_schema)
 OAS32Validator.check_schema = classmethod(check_openapi_schema)