Remove references to RefResolver (#1215)

* Remove references to `RefResolver` * Add dynamic fetching of remote schemas * Try to parse relative refs * Replace non-http ids * Let registry handle relative ref links * Undo error list changes * Remove LocalValidator class and populate schema_cache with local schemas instead * Fix linting * Rewrite cassettes * Rewrite cassettes * Just rewrite casettes that needs rewrite * Remove cast * Importable even if jsonschema not importable * Add back LocalValidator but make it deprecated * Update changelog * Deprecate global variables --------- Co-authored-by: Pete Gadomski <pete.gadomski@gmail.com>
stac-utils · Sep 19, 2023 · e4a17c1 · e4a17c1
1 parent de6cfc2
commit e4a17c1
Show file tree

Hide file tree

Showing 6 changed files with 247 additions and 63 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,6 +2,15 @@
 
 ## [Unreleased]
 
+### Fixed
+
+- Update usage of jsonschema ([#1215](https://github.com/stac-utils/pystac/pull/1215))
+
+### Deprecated
+
+- `pystac.validation.local_validator.LocalValidator` ([#1215](https://github.com/stac-utils/pystac/pull/1215))
+
+
 ## [v1.8.3] - 2023-07-12
 
 ### Added

diff --git a/pyproject.toml b/pyproject.toml
@@ -49,7 +49,7 @@ test = [
     "doc8~=1.1",
     "html5lib~=1.1",
     "jinja2<4.0",
-    "jsonschema>=4.0.1,<4.18",
+    "jsonschema~=4.18",
     "mypy~=1.2",
     "orjson~=3.8",
     "pre-commit~=3.2",
@@ -64,8 +64,7 @@ test = [
     "types-urllib3~=1.26",
 ]
 urllib3 = ["urllib3>=1.26"]
-# jsonschema v4.18.2 breaks validation, and it feels safer to set a ceiling rather than just skip this version. The ceiling should be removed when the v4.18 lineage has settled down and feels safer.
-validation = ["jsonschema>=4.0.1,<4.18"]
+validation = ["jsonschema~=4.18"]
 
 [project.urls]
 homepage = "https://github.com/stac-utils/pystac"
@@ -88,9 +87,6 @@ select = ["E", "F", "I"]
 [tool.pytest.ini_options]
 filterwarnings = [
     "error",
-    # Allows jsonschema's RefResolver deprecation warning through until we're
-    # updated to support jsonschema v4.18
-    "default::DeprecationWarning:pystac.validation.*",
 ]
 
 [build-system]

diff --git a/pystac/validation/__init__.py b/pystac/validation/__init__.py
@@ -4,11 +4,12 @@
 
 import pystac
 from pystac.serialization.identify import STACVersionID, identify_stac_object
+from pystac.stac_object import STACObjectType
 from pystac.utils import make_absolute_href
 from pystac.validation.schema_uri_map import OldExtensionSchemaUriMap
 
 if TYPE_CHECKING:
-    from pystac.stac_object import STACObject, STACObjectType
+    from pystac.stac_object import STACObject
 
 
 # Import after above class definition

diff --git a/pystac/validation/local_validator.py b/pystac/validation/local_validator.py
@@ -1,8 +1,10 @@
 import json
 import sys
+import warnings
 from typing import Any, Dict, List, cast
 
-from jsonschema import Draft7Validator, RefResolver, ValidationError
+from jsonschema import Draft7Validator, ValidationError
+from referencing import Registry, Resource
 
 from pystac.errors import STACLocalValidationError
 from pystac.version import STACVersion
@@ -13,27 +15,93 @@
     from importlib.resources import files as importlib_resources_files
 
 VERSION = STACVersion.DEFAULT_STAC_VERSION
-ITEM_SCHEMA_URI = (
+
+
+def _read_schema(file_name: str) -> Dict[str, Any]:
+    with importlib_resources_files("pystac.validation.jsonschemas").joinpath(
+        file_name
+    ).open("r") as f:
+        return cast(Dict[str, Any], json.load(f))
+
+
+def get_local_schema_cache() -> Dict[str, Dict[str, Any]]:
+    return {
+        **{
+            (
+                f"https://schemas.stacspec.org/v{VERSION}/"
+                f"{name}-spec/json-schema/{name}.json"
+            ): _read_schema(f"stac-spec/v{VERSION}/{name}.json")
+            for name in ("item", "catalog", "collection")
+        },
+        **{
+            f"https://geojson.org/schema/{name}.json": _read_schema(
+                f"geojson/{name}.json"
+            )
+            for name in ("Feature", "Geometry")
+        },
+        **{
+            (
+                f"https://schemas.stacspec.org/v{VERSION}/"
+                f"item-spec/json-schema/{name}.json"
+            ): _read_schema(f"stac-spec/v{VERSION}/{name}.json")
+            for name in (
+                "basics",
+                "datetime",
+                "instrument",
+                "licensing",
+                "provider",
+            )
+        },
+    }
+
+
+############################### DEPRECATED #################################
+
+_deprecated_ITEM_SCHEMA_URI = (
     f"https://schemas.stacspec.org/v{VERSION}/item-spec/json-schema/item.json"
 )
-COLLECTION_SCHEMA_URI = (
+_deprecated_COLLECTION_SCHEMA_URI = (
     f"https://schemas.stacspec.org/v{VERSION}/"
     "collection-spec/json-schema/collection.json"
 )
-CATALOG_SCHEMA_URI = (
+_deprecated_CATALOG_SCHEMA_URI = (
     f"https://schemas.stacspec.org/v{VERSION}/catalog-spec/json-schema/catalog.json"
 )
 
+deprecated_names = ["ITEM_SCHEMA_URI", "COLLECTION_SCHEMA_URI", "CATALOG_SCHEMA_URI"]
+
+
+def __getattr__(name: str) -> Any:
+    if name in deprecated_names:
+        warnings.warn(f"{name} is deprecated and will be removed in v2.", FutureWarning)
+        return globals()[f"_deprecated_{name}"]
+    raise AttributeError(f"module {__name__} has no attribute {name}")
+
 
 class LocalValidator:
+    def __init__(self) -> None:
+        """DEPRECATED"""
+        warnings.warn(
+            "``LocalValidator`` is deprecated and will be removed in v2.",
+            DeprecationWarning,
+        )
+        self.schema_cache = get_local_schema_cache()
+
+    def registry(self) -> Any:
+        return Registry().with_resources(
+            [
+                (k, Resource.from_contents(v)) for k, v in self.schema_cache.items()
+            ]  # type: ignore
+        )
+
     def _validate_from_local(
         self, schema_uri: str, stac_dict: Dict[str, Any]
     ) -> List[ValidationError]:
-        if schema_uri == ITEM_SCHEMA_URI:
+        if schema_uri == _deprecated_ITEM_SCHEMA_URI:
             validator = self.item_validator(VERSION)
-        elif schema_uri == COLLECTION_SCHEMA_URI:
+        elif schema_uri == _deprecated_COLLECTION_SCHEMA_URI:
             validator = self.collection_validator(VERSION)
-        elif schema_uri == CATALOG_SCHEMA_URI:
+        elif schema_uri == _deprecated_CATALOG_SCHEMA_URI:
             validator = self.catalog_validator(VERSION)
         else:
             raise STACLocalValidationError(
@@ -43,22 +111,7 @@ def _validate_from_local(
 
     def _validator(self, stac_type: str, version: str) -> Draft7Validator:
         schema = _read_schema(f"stac-spec/v{version}/{stac_type}.json")
-        resolver = RefResolver.from_schema(schema)
-        resolver.store[
-            f"https://schemas.stacspec.org/v{version}/collection-spec/json-schema/collection.json"
-        ] = _read_schema(f"stac-spec/v{version}/collection.json")
-        resolver.store[
-            f"https://schemas.stacspec.org/v{version}/item-spec/json-schema/item.json"
-        ] = _read_schema(f"stac-spec/v{version}/item.json")
-        for name in ("Feature", "Geometry"):
-            resolver.store[f"https://geojson.org/schema/{name}.json"] = _read_schema(
-                f"geojson/{name}.json"
-            )
-        for name in ("basics", "datetime", "instrument", "licensing", "provider"):
-            resolver.store[
-                f"https://schemas.stacspec.org/v{version}/item-spec/json-schema/{name}.json"
-            ] = _read_schema(f"stac-spec/v{version}/{name}.json")
-        return Draft7Validator(schema, resolver=resolver)
+        return Draft7Validator(schema, registry=self.registry)
 
     def catalog_validator(self, version: str = VERSION) -> Draft7Validator:
         return self._validator("catalog", version)
@@ -68,10 +121,3 @@ def collection_validator(self, version: str = VERSION) -> Draft7Validator:
 
     def item_validator(self, version: str = VERSION) -> Draft7Validator:
         return self._validator("item", version)
-
-
-def _read_schema(file_name: str) -> Dict[str, Any]:
-    with importlib_resources_files("pystac.validation.jsonschemas").joinpath(
-        file_name
-    ).open("r") as f:
-        return cast(Dict[str, Any], json.load(f))
diff --git a/pystac/validation/stac_validator.py b/pystac/validation/stac_validator.py
@@ -1,20 +1,22 @@
 import json
 import logging
+import warnings
 from abc import ABC, abstractmethod
 from typing import Any, Dict, List, Optional, Tuple
 
 import pystac
 import pystac.utils
-from pystac.errors import STACLocalValidationError, STACValidationError
+from pystac.errors import STACValidationError
 from pystac.stac_object import STACObjectType
 from pystac.validation.schema_uri_map import DefaultSchemaUriMap, SchemaUriMap
 
 try:
     import jsonschema
     import jsonschema.exceptions
     import jsonschema.validators
+    from referencing import Registry, Resource
 
-    from pystac.validation.local_validator import LocalValidator
+    from pystac.validation.local_validator import get_local_schema_cache
 
     HAS_JSONSCHEMA = True
 except ImportError:
@@ -149,20 +151,35 @@ def __init__(self, schema_uri_map: Optional[SchemaUriMap] = None) -> None:
         else:
             self.schema_uri_map = DefaultSchemaUriMap()
 
-        self.schema_cache = {}
+        self.schema_cache = get_local_schema_cache()
 
-    def get_schema_from_uri(self, schema_uri: str) -> Tuple[Dict[str, Any], Any]:
+    def _get_schema(self, schema_uri: str) -> Dict[str, Any]:
         if schema_uri not in self.schema_cache:
             s = json.loads(pystac.StacIO.default().read_text(schema_uri))
             self.schema_cache[schema_uri] = s
-
-        schema = self.schema_cache[schema_uri]
-
-        resolver = jsonschema.validators.RefResolver(
-            base_uri=schema_uri, referrer=schema, store=self.schema_cache
+            id_field = "$id" if "$id" in s else "id"
+            if not s[id_field].startswith("http"):
+                s[id_field] = schema_uri
+        return self.schema_cache[schema_uri]
+
+    @property
+    def registry(self) -> Any:
+        def retrieve(schema_uri: str) -> Resource[Dict[str, Any]]:
+            return Resource.from_contents(self._get_schema(schema_uri))
+
+        return Registry(retrieve=retrieve).with_resources(  # type: ignore
+            [
+                (k, Resource.from_contents(v)) for k, v in self.schema_cache.items()
+            ]  # type: ignore
         )
 
-        return schema, resolver
+    def get_schema_from_uri(self, schema_uri: str) -> Tuple[Dict[str, Any], Any]:
+        """DEPRECATED"""
+        warnings.warn(
+            "get_schema_from_uri is deprecated and will be removed in v2.",
+            DeprecationWarning,
+        )
+        return self._get_schema(schema_uri), self.registry
 
     def _validate_from_uri(
         self,
@@ -172,17 +189,13 @@ def _validate_from_uri(
         href: Optional[str] = None,
     ) -> None:
         try:
-            resolver = None
-            try:
-                errors = LocalValidator()._validate_from_local(schema_uri, stac_dict)
-            except STACLocalValidationError:
-                schema, resolver = self.get_schema_from_uri(schema_uri)
-                # This block is cribbed (w/ change in error handling) from
-                # jsonschema.validate
-                cls = jsonschema.validators.validator_for(schema)
-                cls.check_schema(schema)
-                validator = cls(schema, resolver=resolver)
-                errors = list(validator.iter_errors(stac_dict))
+            schema = self._get_schema(schema_uri)
+            # This block is cribbed (w/ change in error handling) from
+            # jsonschema.validate
+            cls = jsonschema.validators.validator_for(schema)
+            cls.check_schema(schema)
+            validator = cls(schema, registry=self.registry)
+            errors = list(validator.iter_errors(stac_dict))
         except Exception as e:
             logger.error(f"Exception while validating {stac_object_type} href: {href}")
             logger.exception(e)
@@ -199,11 +212,6 @@ def _validate_from_uri(
             best = jsonschema.exceptions.best_match(errors)
             raise STACValidationError(msg, source=errors) from best
 
-        if resolver is not None:
-            for uri in resolver.store:
-                if uri not in self.schema_cache:
-                    self.schema_cache[uri] = resolver.store[uri]
-
     def validate_core(
         self,
         stac_dict: Dict[str, Any],