Make sure the type reference is removed from the seen references #11143
+40
−1
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Deploying pydantic-docs with
|
| Latest commit: |
75241c2
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://95df4efe.pydantic-docs.pages.dev |
| Branch Preview URL: | https://11024.pydantic-docs.pages.dev |
CodSpeed Performance ReportMerging #11143 will improve performances by 6.26%Comparing Summary
Benchmarks breakdown
|
Coverage report
This PR does not seem to contain any modification to coverable code. |
|
I'll note, the current implementation is confusing and I think this is what led to the oversight. Here is a simpler implementation: _generic_recursion_cache: ContextVar[set[str]] = ContextVar('_generic_recursion_cache', default=set())
@contextmanager
def generic_recursion_self_type(
origin: type[BaseModel], args: tuple[Any, ...]
) -> Iterator[PydanticRecursiveRef | None]:
"""This contextmanager should be placed around the recursive calls used to build a generic type,
and accept as arguments the generic origin type and the type arguments being passed to it.
If the same origin and arguments are observed twice, it implies that a self-reference placeholder
can be used while building the core schema, and will produce a schema_ref that will be valid in the
final parent schema.
"""
previously_seen_type_refs = _generic_recursion_cache.get()
type_ref = get_type_ref(origin, args_override=args)
if type_ref in previously_seen_type_refs:
self_type = PydanticRecursiveRef(type_ref=type_ref)
yield self_type
else:
try:
previously_seen_type_refs.add(type_ref)
yield
finally:
previously_seen_type_refs.remove(type_ref)Diffdiff --git a/pydantic/_internal/_generics.py b/pydantic/_internal/_generics.py
index 7b57fc71c..dc474861a 100644
--- a/pydantic/_internal/_generics.py
+++ b/pydantic/_internal/_generics.py
@@ -401,7 +401,7 @@ def check_parameters_count(cls: type[BaseModel], parameters: tuple[Any, ...]) ->
raise TypeError(f'Too {description} parameters for {cls}; actual {actual}, expected {expected}')
-_generic_recursion_cache: ContextVar[set[str] | None] = ContextVar('_generic_recursion_cache', default=None)
+_generic_recursion_cache: ContextVar[set[str]] = ContextVar('_generic_recursion_cache', default=set())
@contextmanager
@@ -416,24 +416,17 @@ def generic_recursion_self_type(
final parent schema.
"""
previously_seen_type_refs = _generic_recursion_cache.get()
- if previously_seen_type_refs is None:
- previously_seen_type_refs = set()
- token = _generic_recursion_cache.set(previously_seen_type_refs)
- else:
- token = None
+ type_ref = get_type_ref(origin, args_override=args)
- try:
- type_ref = get_type_ref(origin, args_override=args)
- if type_ref in previously_seen_type_refs:
- self_type = PydanticRecursiveRef(type_ref=type_ref)
- yield self_type
- else:
+ if type_ref in previously_seen_type_refs:
+ self_type = PydanticRecursiveRef(type_ref=type_ref)
+ yield self_type
+ else:
+ try:
previously_seen_type_refs.add(type_ref)
yield
+ finally:
previously_seen_type_refs.remove(type_ref)
- finally:
- if token:
- _generic_recursion_cache.reset(token)
def recursively_defined_type_refs() -> set[str]:Edit: turns out this is not going to work well as context variables shouldn't have mutable defaults. |
sydney-runkle
approved these changes
Dec 18, 2024
36 tasks
When parametrizing generic models, the `generic_recursion_self_type` context manager is entered, and is used to avoid infinite recursions if the same generic model happens to be parametrized again with the same args during the first parametrization. However, upon exiting the context manager (and thus when the parametrized model is fully created), we forgot to remove the type ref from the set. This happened only if we were already in the process of parametrizing another model, as otherwise the `_generic_recursion_cache` would be reset (see the `if token` condition). In theory, this couldn't cause issues because parametrized models are cached, and the cache is checked *before* entering the context manager. However, because we have custom `mro()` implementation on the `BaseModel` metaclass, this ends up causing issues is some really specific scenarios.
|
Well credits goes to @kc0506 for this one |
Viicos
added a commit
that referenced
this pull request
Dec 18, 2024
) When parametrizing generic models, the `generic_recursion_self_type` context manager is entered, and is used to avoid infinite recursions if the same generic model happens to be parametrized again with the same args during the first parametrization. However, upon exiting the context manager (and thus when the parametrized model is fully created), we forgot to remove the type ref from the set. This happened only if we were already in the process of parametrizing another model, as otherwise the `_generic_recursion_cache` would be reset (see the `if token` condition). In theory, this couldn't cause issues because parametrized models are cached, and the cache is checked *before* entering the context manager. However, because we have custom `mro()` implementation on the `BaseModel` metaclass, this ends up causing issues is some really specific scenarios. Co-authored-by: kc0506 <kchong0506@gmail.com>
5 tasks
5 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As per #11037. Fixes #11024.
When parametrizing generic models, the
generic_recursion_self_typecontext manager is entered, and is used to avoid infinite recursions if the same generic model happens to be parametrized again with the same args during the first parametrization.However, upon exiting the context manager (and thus when the parametrized model is fully created), we forgot to remove the type ref from the set. This happened only if we were already in the process of parametrizing another model, as otherwise the
_generic_recursion_cachewould be reset (see theif tokencondition).In theory, this couldn't cause issues because parametrized models are cached, and the cache is checked before entering the context manager. However, because we have custom
mro()implementation on theBaseModelmetaclass, this ends up causing issues is some really specific scenarios.Change Summary
Related issue number
Checklist