Skip to content

Commit

Permalink
refactor: Collect functions, global into _Registry
Browse files Browse the repository at this point in the history
  • Loading branch information
@dangotbanned
dangotbanned committed Sep 2, 2024
1 parent 5d9fb65 commit a26b3f1
Show file tree
Hide file tree
Showing 2 changed files with 146 additions and 112 deletions.
129 changes: 73 additions & 56 deletions altair/utils/schemapi.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -292,6 +292,74 @@ def specification_with(dialect_id: str, /) -> Specification[Any]:
"""
return _specification_with(dialect_id)

class _Registry:
"""
A cache of `Registry`_ (s).
An instance named ``registry`` is used to wrap the `Registry`_ API,
with a managed cache.
See Also
--------
``_Registry.__call__``
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
"""

_cached: ClassVar[dict[tuple[str, str], Registry[Any]]] = {}

@staticmethod
def compute_key(root: Map, dialect_id: str, /) -> tuple[str, str]:
"""
Generate a simple-minded hash to identify a registry.
Notes
-----
Why the strange hash?
- **All** generated schemas hit the ``"$ref"`` branch.
- ``api.Then`` hits the len(...) 1 branch w/ ``{"type": "object"}``.
- Final branch is only hit by mock schemas in:
- `tests/utils/test_core.py::test_infer_encoding_types`
- `tests/utils/test_schemapi.py`
"""
if "$ref" in root:
k1 = root["$ref"]
elif len(root) == 1:
k1 = "".join(f"{s!s}" for s in chain(*root.items()))
else:
k1 = _HASH_ENCODER.encode(root)
return k1, dialect_id

@classmethod
def update_cached(
cls, root: Map, dialect_id: str, resolver: Resolver[Any]
) -> None:
cls._cached[cls.compute_key(root, dialect_id)] = resolver._registry

def __call__(self, root: Map, dialect_id: str, /) -> Registry[Any]:
"""
Constructs a `Registry`_, adding the `Resource`_ produced by ``rootschema``.
Requires at least ``jsonschema`` `v4.18.0a1`_.
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
.. _Resource:
https://referencing.readthedocs.io/en/stable/api/#referencing.Resource
.. _v4.18.0a1:
https://github.com/python-jsonschema/jsonschema/releases/tag/v4.18.0a1
"""
cache_key = self.compute_key(root, dialect_id)
if (reg := self._cached.get(cache_key, None)) is not None:
return reg
resource = specification_with(dialect_id).create_resource(root)
reg = Registry().with_resource(_VEGA_LITE_ROOT_URI, resource).crawl()
type(self)._cached[cache_key] = reg
return reg

registry: _Registry = _Registry()

def _validator(schema: Map, rootschema: Map | None = None, /) -> Validator:
"""
Constructs a `Validator`_ for future validation.
Expand All @@ -311,37 +379,9 @@ def _validator(schema: Map, rootschema: Map | None = None, /) -> Validator:
# NOTE: This is the current version
uri = _get_schema_dialect_uri(rootschema or schema)
validator = _validator_for(uri)
registry = _registry(rootschema or schema, uri)
return validator(_prepare_references(schema), registry=registry)

def _registry(rootschema: Map, dialect_id: str) -> Registry[Any]:
"""
Constructs a `Registry`_, adding the `Resource`_ produced by ``rootschema``.
Requires at least ``jsonschema`` `v4.18.0a1`_.
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
.. _Resource:
https://referencing.readthedocs.io/en/stable/api/#referencing.Resource
.. _v4.18.0a1:
https://github.com/python-jsonschema/jsonschema/releases/tag/v4.18.0a1
"""
global _REGISTRY_CACHE
cache_key = _registry_comp_key(rootschema, dialect_id)
if (registry := _REGISTRY_CACHE.get(cache_key, None)) is not None:
return registry
else:
specification = specification_with(dialect_id)
resource = specification.create_resource(rootschema)
registry = Registry().with_resource(_VEGA_LITE_ROOT_URI, resource).crawl()
_REGISTRY_CACHE[cache_key] = registry
return registry

def _registry_update(root: Map, dialect_id: str, resolver: Resolver[Any]) -> None:
global _REGISTRY_CACHE
cache_key = _registry_comp_key(root, dialect_id)
_REGISTRY_CACHE[cache_key] = resolver._registry
return validator(
_prepare_references(schema), registry=registry(rootschema or schema, uri)
)

def _resolve_references(schema: Map, rootschema: Map) -> Map:
"""
Expand All @@ -359,36 +399,13 @@ def _resolve_references(schema: Map, rootschema: Map) -> Map:
if ("$ref" not in root) or ("$ref" not in schema):
return schema
uri = _get_schema_dialect_uri(rootschema)
registry = _registry(root, uri)
resolver = registry.resolver(_VEGA_LITE_ROOT_URI)
resolver = registry(root, uri).resolver(_VEGA_LITE_ROOT_URI)
while "$ref" in schema:
resolved = resolver.lookup(schema["$ref"])
schema = resolved.contents
_registry_update(root, uri, resolved.resolver)
registry.update_cached(root, uri, resolved.resolver)
return schema

def _registry_comp_key(root: Map, dialect_id: str, /) -> tuple[str, str]:
"""
Generate a simple-minded hash to identify a registry.
Notes
-----
Why the strange hash?
- **All** generated schemas hit the ``"$ref"`` branch.
- ``api.Then`` hits the len(...) 1 branch w/ ``{"type": "object"}``.
- Final branch is only hit by mock schemas in:
- `tests/utils/test_core.py::test_infer_encoding_types`
- `tests/utils/test_schemapi.py`
"""
if "$ref" in root:
k1 = root["$ref"]
elif len(root) == 1:
k1 = "".join(f"{s!s}" for s in chain(*root.items()))
else:
k1 = _HASH_ENCODER.encode(root)
return k1, dialect_id

_REGISTRY_CACHE: dict[tuple[str, str], Registry[Any]] = {}

else:

Expand Down
129 changes: 73 additions & 56 deletions tools/schemapi/schemapi.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -290,6 +290,74 @@ def specification_with(dialect_id: str, /) -> Specification[Any]:
"""
return _specification_with(dialect_id)

class _Registry:
"""
A cache of `Registry`_ (s).
An instance named ``registry`` is used to wrap the `Registry`_ API,
with a managed cache.
See Also
--------
``_Registry.__call__``
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
"""

_cached: ClassVar[dict[tuple[str, str], Registry[Any]]] = {}

@staticmethod
def compute_key(root: Map, dialect_id: str, /) -> tuple[str, str]:
"""
Generate a simple-minded hash to identify a registry.
Notes
-----
Why the strange hash?
- **All** generated schemas hit the ``"$ref"`` branch.
- ``api.Then`` hits the len(...) 1 branch w/ ``{"type": "object"}``.
- Final branch is only hit by mock schemas in:
- `tests/utils/test_core.py::test_infer_encoding_types`
- `tests/utils/test_schemapi.py`
"""
if "$ref" in root:
k1 = root["$ref"]
elif len(root) == 1:
k1 = "".join(f"{s!s}" for s in chain(*root.items()))
else:
k1 = _HASH_ENCODER.encode(root)
return k1, dialect_id

@classmethod
def update_cached(
cls, root: Map, dialect_id: str, resolver: Resolver[Any]
) -> None:
cls._cached[cls.compute_key(root, dialect_id)] = resolver._registry

def __call__(self, root: Map, dialect_id: str, /) -> Registry[Any]:
"""
Constructs a `Registry`_, adding the `Resource`_ produced by ``rootschema``.
Requires at least ``jsonschema`` `v4.18.0a1`_.
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
.. _Resource:
https://referencing.readthedocs.io/en/stable/api/#referencing.Resource
.. _v4.18.0a1:
https://github.com/python-jsonschema/jsonschema/releases/tag/v4.18.0a1
"""
cache_key = self.compute_key(root, dialect_id)
if (reg := self._cached.get(cache_key, None)) is not None:
return reg
resource = specification_with(dialect_id).create_resource(root)
reg = Registry().with_resource(_VEGA_LITE_ROOT_URI, resource).crawl()
type(self)._cached[cache_key] = reg
return reg

registry: _Registry = _Registry()

def _validator(schema: Map, rootschema: Map | None = None, /) -> Validator:
"""
Constructs a `Validator`_ for future validation.
Expand All @@ -309,37 +377,9 @@ def _validator(schema: Map, rootschema: Map | None = None, /) -> Validator:
# NOTE: This is the current version
uri = _get_schema_dialect_uri(rootschema or schema)
validator = _validator_for(uri)
registry = _registry(rootschema or schema, uri)
return validator(_prepare_references(schema), registry=registry)

def _registry(rootschema: Map, dialect_id: str) -> Registry[Any]:
"""
Constructs a `Registry`_, adding the `Resource`_ produced by ``rootschema``.
Requires at least ``jsonschema`` `v4.18.0a1`_.
.. _Registry:
https://referencing.readthedocs.io/en/stable/api/#referencing.Registry
.. _Resource:
https://referencing.readthedocs.io/en/stable/api/#referencing.Resource
.. _v4.18.0a1:
https://github.com/python-jsonschema/jsonschema/releases/tag/v4.18.0a1
"""
global _REGISTRY_CACHE
cache_key = _registry_comp_key(rootschema, dialect_id)
if (registry := _REGISTRY_CACHE.get(cache_key, None)) is not None:
return registry
else:
specification = specification_with(dialect_id)
resource = specification.create_resource(rootschema)
registry = Registry().with_resource(_VEGA_LITE_ROOT_URI, resource).crawl()
_REGISTRY_CACHE[cache_key] = registry
return registry

def _registry_update(root: Map, dialect_id: str, resolver: Resolver[Any]) -> None:
global _REGISTRY_CACHE
cache_key = _registry_comp_key(root, dialect_id)
_REGISTRY_CACHE[cache_key] = resolver._registry
return validator(
_prepare_references(schema), registry=registry(rootschema or schema, uri)
)

def _resolve_references(schema: Map, rootschema: Map) -> Map:
"""
Expand All @@ -357,36 +397,13 @@ def _resolve_references(schema: Map, rootschema: Map) -> Map:
if ("$ref" not in root) or ("$ref" not in schema):
return schema
uri = _get_schema_dialect_uri(rootschema)
registry = _registry(root, uri)
resolver = registry.resolver(_VEGA_LITE_ROOT_URI)
resolver = registry(root, uri).resolver(_VEGA_LITE_ROOT_URI)
while "$ref" in schema:
resolved = resolver.lookup(schema["$ref"])
schema = resolved.contents
_registry_update(root, uri, resolved.resolver)
registry.update_cached(root, uri, resolved.resolver)
return schema

def _registry_comp_key(root: Map, dialect_id: str, /) -> tuple[str, str]:
"""
Generate a simple-minded hash to identify a registry.
Notes
-----
Why the strange hash?
- **All** generated schemas hit the ``"$ref"`` branch.
- ``api.Then`` hits the len(...) 1 branch w/ ``{"type": "object"}``.
- Final branch is only hit by mock schemas in:
- `tests/utils/test_core.py::test_infer_encoding_types`
- `tests/utils/test_schemapi.py`
"""
if "$ref" in root:
k1 = root["$ref"]
elif len(root) == 1:
k1 = "".join(f"{s!s}" for s in chain(*root.items()))
else:
k1 = _HASH_ENCODER.encode(root)
return k1, dialect_id

_REGISTRY_CACHE: dict[tuple[str, str], Registry[Any]] = {}

else:

Expand Down

0 comments on commit a26b3f1

Please sign in to comment.