mirror of
https://github.com/invoke-ai/InvokeAI.git
synced 2026-01-24 08:38:03 -05:00
c238a7f18b
Upgrade pydantic and fastapi to latest.
- pydantic~=2.4.2
- fastapi~=103.2
- fastapi-events~=0.9.1
**Big Changes**
There are a number of logic changes needed to support pydantic v2. Most changes are very simple, like using the new methods to serialized and deserialize models, but there are a few more complex changes.
**Invocations**
The biggest change relates to invocation creation, instantiation and validation.
Because pydantic v2 moves all validation logic into the rust pydantic-core, we may no longer directly stick our fingers into the validation pie.
Previously, we (ab)used models and fields to allow invocation fields to be optional at instantiation, but required when `invoke()` is called. We directly manipulated the fields and invocation models when calling `invoke()`.
With pydantic v2, this is much more involved. Changes to the python wrapper do not propagate down to the rust validation logic - you have to rebuild the model. This causes problem with concurrent access to the invocation classes and is not a free operation.
This logic has been totally refactored and we do not need to change the model any more. The details are in `baseinvocation.py`, in the `InputField` function and `BaseInvocation.invoke_internal()` method.
In the end, this implementation is cleaner.
**Invocation Fields**
In pydantic v2, you can no longer directly add or remove fields from a model.
Previously, we did this to add the `type` field to invocations.
**Invocation Decorators**
With pydantic v2, we instead use the imperative `create_model()` API to create a new model with the additional field. This is done in `baseinvocation.py` in the `invocation()` wrapper.
A similar technique is used for `invocation_output()`.
**Minor Changes**
There are a number of minor changes around the pydantic v2 models API.
**Protected `model_` Namespace**
All models' pydantic-provided methods and attributes are prefixed with `model_` and this is considered a protected namespace. This causes some conflict, because "model" means something to us, and we have a ton of pydantic models with attributes starting with "model_".
Forunately, there are no direct conflicts. However, in any pydantic model where we define an attribute or method that starts with "model_", we must tell set the protected namespaces to an empty tuple.
```py
class IPAdapterModelField(BaseModel):
model_name: str = Field(description="Name of the IP-Adapter model")
base_model: BaseModelType = Field(description="Base model")
model_config = ConfigDict(protected_namespaces=())
```
**Model Serialization**
Pydantic models no longer have `Model.dict()` or `Model.json()`.
Instead, we use `Model.model_dump()` or `Model.model_dump_json()`.
**Model Deserialization**
Pydantic models no longer have `Model.parse_obj()` or `Model.parse_raw()`, and there are no `parse_raw_as()` or `parse_obj_as()` functions.
Instead, you need to create a `TypeAdapter` object to parse python objects or JSON into a model.
```py
adapter_graph = TypeAdapter(Graph)
deserialized_graph_from_json = adapter_graph.validate_json(graph_json)
deserialized_graph_from_dict = adapter_graph.validate_python(graph_dict)
```
**Field Customisation**
Pydantic `Field`s no longer accept arbitrary args.
Now, you must put all additional arbitrary args in a `json_schema_extra` arg on the field.
**Schema Customisation**
FastAPI and pydantic schema generation now follows the OpenAPI version 3.1 spec.
This necessitates two changes:
- Our schema customization logic has been revised
- Schema parsing to build node templates has been revised
The specific aren't important, but this does present additional surface area for bugs.
**Performance Improvements**
Pydantic v2 is a full rewrite with a rust backend. This offers a substantial performance improvement (pydantic claims 5x to 50x depending on the task). We'll notice this the most during serialization and deserialization of sessions/graphs, which happens very very often - a couple times per node.
I haven't done any benchmarks, but anecdotally, graph execution is much faster. Also, very larges graphs - like with massive iterators - are much, much faster.
132 lines
4.6 KiB
Python
132 lines
4.6 KiB
Python
from collections import OrderedDict
|
|
from dataclasses import dataclass, field
|
|
from threading import Lock
|
|
from typing import Optional, Union
|
|
|
|
from invokeai.app.invocations.baseinvocation import BaseInvocation, BaseInvocationOutput
|
|
from invokeai.app.services.invocation_cache.invocation_cache_base import InvocationCacheBase
|
|
from invokeai.app.services.invocation_cache.invocation_cache_common import InvocationCacheStatus
|
|
from invokeai.app.services.invoker import Invoker
|
|
|
|
|
|
@dataclass(order=True)
|
|
class CachedItem:
|
|
invocation_output: BaseInvocationOutput = field(compare=False)
|
|
invocation_output_json: str = field(compare=False)
|
|
|
|
|
|
class MemoryInvocationCache(InvocationCacheBase):
|
|
_cache: OrderedDict[Union[int, str], CachedItem]
|
|
_max_cache_size: int
|
|
_disabled: bool
|
|
_hits: int
|
|
_misses: int
|
|
_invoker: Invoker
|
|
_lock: Lock
|
|
|
|
def __init__(self, max_cache_size: int = 0) -> None:
|
|
self._cache = OrderedDict()
|
|
self._max_cache_size = max_cache_size
|
|
self._disabled = False
|
|
self._hits = 0
|
|
self._misses = 0
|
|
self._lock = Lock()
|
|
|
|
def start(self, invoker: Invoker) -> None:
|
|
self._invoker = invoker
|
|
if self._max_cache_size == 0:
|
|
return
|
|
self._invoker.services.images.on_deleted(self._delete_by_match)
|
|
self._invoker.services.latents.on_deleted(self._delete_by_match)
|
|
|
|
def get(self, key: Union[int, str]) -> Optional[BaseInvocationOutput]:
|
|
with self._lock:
|
|
if self._max_cache_size == 0 or self._disabled:
|
|
return None
|
|
item = self._cache.get(key, None)
|
|
if item is not None:
|
|
self._hits += 1
|
|
self._cache.move_to_end(key)
|
|
return item.invocation_output
|
|
self._misses += 1
|
|
return None
|
|
|
|
def save(self, key: Union[int, str], invocation_output: BaseInvocationOutput) -> None:
|
|
with self._lock:
|
|
if self._max_cache_size == 0 or self._disabled or key in self._cache:
|
|
return
|
|
# If the cache is full, we need to remove the least used
|
|
number_to_delete = len(self._cache) + 1 - self._max_cache_size
|
|
self._delete_oldest_access(number_to_delete)
|
|
self._cache[key] = CachedItem(
|
|
invocation_output,
|
|
invocation_output.model_dump_json(
|
|
warnings=False, exclude_defaults=True, exclude_unset=True, include={"type"}
|
|
),
|
|
)
|
|
|
|
def _delete_oldest_access(self, number_to_delete: int) -> None:
|
|
number_to_delete = min(number_to_delete, len(self._cache))
|
|
for _ in range(number_to_delete):
|
|
self._cache.popitem(last=False)
|
|
|
|
def _delete(self, key: Union[int, str]) -> None:
|
|
if self._max_cache_size == 0:
|
|
return
|
|
if key in self._cache:
|
|
del self._cache[key]
|
|
|
|
def delete(self, key: Union[int, str]) -> None:
|
|
with self._lock:
|
|
return self._delete(key)
|
|
|
|
def clear(self, *args, **kwargs) -> None:
|
|
with self._lock:
|
|
if self._max_cache_size == 0:
|
|
return
|
|
self._cache.clear()
|
|
self._misses = 0
|
|
self._hits = 0
|
|
|
|
@staticmethod
|
|
def create_key(invocation: BaseInvocation) -> int:
|
|
return hash(invocation.model_dump_json(exclude={"id"}, warnings=False))
|
|
|
|
def disable(self) -> None:
|
|
with self._lock:
|
|
if self._max_cache_size == 0:
|
|
return
|
|
self._disabled = True
|
|
|
|
def enable(self) -> None:
|
|
with self._lock:
|
|
if self._max_cache_size == 0:
|
|
return
|
|
self._disabled = False
|
|
|
|
def get_status(self) -> InvocationCacheStatus:
|
|
with self._lock:
|
|
return InvocationCacheStatus(
|
|
hits=self._hits,
|
|
misses=self._misses,
|
|
enabled=not self._disabled and self._max_cache_size > 0,
|
|
size=len(self._cache),
|
|
max_size=self._max_cache_size,
|
|
)
|
|
|
|
def _delete_by_match(self, to_match: str) -> None:
|
|
with self._lock:
|
|
if self._max_cache_size == 0:
|
|
return
|
|
keys_to_delete = set()
|
|
for key, cached_item in self._cache.items():
|
|
if to_match in cached_item.invocation_output_json:
|
|
keys_to_delete.add(key)
|
|
if not keys_to_delete:
|
|
return
|
|
for key in keys_to_delete:
|
|
self._delete(key)
|
|
self._invoker.services.logger.debug(
|
|
f"Deleted {len(keys_to_delete)} cached invocation outputs for {to_match}"
|
|
)
|