mirror of
https://github.com/invoke-ai/InvokeAI
synced 2024-08-30 20:32:17 +00:00
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.
292 lines
9.9 KiB
Python
292 lines
9.9 KiB
Python
# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)
|
|
|
|
from typing import Literal
|
|
|
|
import numpy as np
|
|
from pydantic import field_validator
|
|
|
|
from invokeai.app.invocations.primitives import FloatOutput, IntegerOutput
|
|
|
|
from .baseinvocation import BaseInvocation, FieldDescriptions, InputField, InvocationContext, invocation
|
|
|
|
|
|
@invocation("add", title="Add Integers", tags=["math", "add"], category="math", version="1.0.0")
|
|
class AddInvocation(BaseInvocation):
|
|
"""Adds two numbers"""
|
|
|
|
a: int = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: int = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
return IntegerOutput(value=self.a + self.b)
|
|
|
|
|
|
@invocation("sub", title="Subtract Integers", tags=["math", "subtract"], category="math", version="1.0.0")
|
|
class SubtractInvocation(BaseInvocation):
|
|
"""Subtracts two numbers"""
|
|
|
|
a: int = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: int = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
return IntegerOutput(value=self.a - self.b)
|
|
|
|
|
|
@invocation("mul", title="Multiply Integers", tags=["math", "multiply"], category="math", version="1.0.0")
|
|
class MultiplyInvocation(BaseInvocation):
|
|
"""Multiplies two numbers"""
|
|
|
|
a: int = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: int = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
return IntegerOutput(value=self.a * self.b)
|
|
|
|
|
|
@invocation("div", title="Divide Integers", tags=["math", "divide"], category="math", version="1.0.0")
|
|
class DivideInvocation(BaseInvocation):
|
|
"""Divides two numbers"""
|
|
|
|
a: int = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: int = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
return IntegerOutput(value=int(self.a / self.b))
|
|
|
|
|
|
@invocation(
|
|
"rand_int",
|
|
title="Random Integer",
|
|
tags=["math", "random"],
|
|
category="math",
|
|
version="1.0.0",
|
|
use_cache=False,
|
|
)
|
|
class RandomIntInvocation(BaseInvocation):
|
|
"""Outputs a single random integer."""
|
|
|
|
low: int = InputField(default=0, description=FieldDescriptions.inclusive_low)
|
|
high: int = InputField(default=np.iinfo(np.int32).max, description=FieldDescriptions.exclusive_high)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
return IntegerOutput(value=np.random.randint(self.low, self.high))
|
|
|
|
|
|
@invocation(
|
|
"rand_float",
|
|
title="Random Float",
|
|
tags=["math", "float", "random"],
|
|
category="math",
|
|
version="1.0.1",
|
|
use_cache=False,
|
|
)
|
|
class RandomFloatInvocation(BaseInvocation):
|
|
"""Outputs a single random float"""
|
|
|
|
low: float = InputField(default=0.0, description=FieldDescriptions.inclusive_low)
|
|
high: float = InputField(default=1.0, description=FieldDescriptions.exclusive_high)
|
|
decimals: int = InputField(default=2, description=FieldDescriptions.decimal_places)
|
|
|
|
def invoke(self, context: InvocationContext) -> FloatOutput:
|
|
random_float = np.random.uniform(self.low, self.high)
|
|
rounded_float = round(random_float, self.decimals)
|
|
return FloatOutput(value=rounded_float)
|
|
|
|
|
|
@invocation(
|
|
"float_to_int",
|
|
title="Float To Integer",
|
|
tags=["math", "round", "integer", "float", "convert"],
|
|
category="math",
|
|
version="1.0.0",
|
|
)
|
|
class FloatToIntegerInvocation(BaseInvocation):
|
|
"""Rounds a float number to (a multiple of) an integer."""
|
|
|
|
value: float = InputField(default=0, description="The value to round")
|
|
multiple: int = InputField(default=1, ge=1, title="Multiple of", description="The multiple to round to")
|
|
method: Literal["Nearest", "Floor", "Ceiling", "Truncate"] = InputField(
|
|
default="Nearest", description="The method to use for rounding"
|
|
)
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
if self.method == "Nearest":
|
|
return IntegerOutput(value=round(self.value / self.multiple) * self.multiple)
|
|
elif self.method == "Floor":
|
|
return IntegerOutput(value=np.floor(self.value / self.multiple) * self.multiple)
|
|
elif self.method == "Ceiling":
|
|
return IntegerOutput(value=np.ceil(self.value / self.multiple) * self.multiple)
|
|
else: # self.method == "Truncate"
|
|
return IntegerOutput(value=int(self.value / self.multiple) * self.multiple)
|
|
|
|
|
|
@invocation("round_float", title="Round Float", tags=["math", "round"], category="math", version="1.0.0")
|
|
class RoundInvocation(BaseInvocation):
|
|
"""Rounds a float to a specified number of decimal places."""
|
|
|
|
value: float = InputField(default=0, description="The float value")
|
|
decimals: int = InputField(default=0, description="The number of decimal places")
|
|
|
|
def invoke(self, context: InvocationContext) -> FloatOutput:
|
|
return FloatOutput(value=round(self.value, self.decimals))
|
|
|
|
|
|
INTEGER_OPERATIONS = Literal[
|
|
"ADD",
|
|
"SUB",
|
|
"MUL",
|
|
"DIV",
|
|
"EXP",
|
|
"MOD",
|
|
"ABS",
|
|
"MIN",
|
|
"MAX",
|
|
]
|
|
|
|
|
|
INTEGER_OPERATIONS_LABELS = dict(
|
|
ADD="Add A+B",
|
|
SUB="Subtract A-B",
|
|
MUL="Multiply A*B",
|
|
DIV="Divide A/B",
|
|
EXP="Exponentiate A^B",
|
|
MOD="Modulus A%B",
|
|
ABS="Absolute Value of A",
|
|
MIN="Minimum(A,B)",
|
|
MAX="Maximum(A,B)",
|
|
)
|
|
|
|
|
|
@invocation(
|
|
"integer_math",
|
|
title="Integer Math",
|
|
tags=[
|
|
"math",
|
|
"integer",
|
|
"add",
|
|
"subtract",
|
|
"multiply",
|
|
"divide",
|
|
"modulus",
|
|
"power",
|
|
"absolute value",
|
|
"min",
|
|
"max",
|
|
],
|
|
category="math",
|
|
version="1.0.0",
|
|
)
|
|
class IntegerMathInvocation(BaseInvocation):
|
|
"""Performs integer math."""
|
|
|
|
operation: INTEGER_OPERATIONS = InputField(
|
|
default="ADD", description="The operation to perform", ui_choice_labels=INTEGER_OPERATIONS_LABELS
|
|
)
|
|
a: int = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: int = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
@field_validator("b")
|
|
def no_unrepresentable_results(cls, v, values):
|
|
if values["operation"] == "DIV" and v == 0:
|
|
raise ValueError("Cannot divide by zero")
|
|
elif values["operation"] == "MOD" and v == 0:
|
|
raise ValueError("Cannot divide by zero")
|
|
elif values["operation"] == "EXP" and v < 0:
|
|
raise ValueError("Result of exponentiation is not an integer")
|
|
return v
|
|
|
|
def invoke(self, context: InvocationContext) -> IntegerOutput:
|
|
# Python doesn't support switch statements until 3.10, but InvokeAI supports back to 3.9
|
|
if self.operation == "ADD":
|
|
return IntegerOutput(value=self.a + self.b)
|
|
elif self.operation == "SUB":
|
|
return IntegerOutput(value=self.a - self.b)
|
|
elif self.operation == "MUL":
|
|
return IntegerOutput(value=self.a * self.b)
|
|
elif self.operation == "DIV":
|
|
return IntegerOutput(value=int(self.a / self.b))
|
|
elif self.operation == "EXP":
|
|
return IntegerOutput(value=self.a**self.b)
|
|
elif self.operation == "MOD":
|
|
return IntegerOutput(value=self.a % self.b)
|
|
elif self.operation == "ABS":
|
|
return IntegerOutput(value=abs(self.a))
|
|
elif self.operation == "MIN":
|
|
return IntegerOutput(value=min(self.a, self.b))
|
|
else: # self.operation == "MAX":
|
|
return IntegerOutput(value=max(self.a, self.b))
|
|
|
|
|
|
FLOAT_OPERATIONS = Literal[
|
|
"ADD",
|
|
"SUB",
|
|
"MUL",
|
|
"DIV",
|
|
"EXP",
|
|
"ABS",
|
|
"SQRT",
|
|
"MIN",
|
|
"MAX",
|
|
]
|
|
|
|
|
|
FLOAT_OPERATIONS_LABELS = dict(
|
|
ADD="Add A+B",
|
|
SUB="Subtract A-B",
|
|
MUL="Multiply A*B",
|
|
DIV="Divide A/B",
|
|
EXP="Exponentiate A^B",
|
|
ABS="Absolute Value of A",
|
|
SQRT="Square Root of A",
|
|
MIN="Minimum(A,B)",
|
|
MAX="Maximum(A,B)",
|
|
)
|
|
|
|
|
|
@invocation(
|
|
"float_math",
|
|
title="Float Math",
|
|
tags=["math", "float", "add", "subtract", "multiply", "divide", "power", "root", "absolute value", "min", "max"],
|
|
category="math",
|
|
version="1.0.0",
|
|
)
|
|
class FloatMathInvocation(BaseInvocation):
|
|
"""Performs floating point math."""
|
|
|
|
operation: FLOAT_OPERATIONS = InputField(
|
|
default="ADD", description="The operation to perform", ui_choice_labels=FLOAT_OPERATIONS_LABELS
|
|
)
|
|
a: float = InputField(default=0, description=FieldDescriptions.num_1)
|
|
b: float = InputField(default=0, description=FieldDescriptions.num_2)
|
|
|
|
@field_validator("b")
|
|
def no_unrepresentable_results(cls, v, values):
|
|
if values["operation"] == "DIV" and v == 0:
|
|
raise ValueError("Cannot divide by zero")
|
|
elif values["operation"] == "EXP" and values["a"] == 0 and v < 0:
|
|
raise ValueError("Cannot raise zero to a negative power")
|
|
elif values["operation"] == "EXP" and type(values["a"] ** v) is complex:
|
|
raise ValueError("Root operation resulted in a complex number")
|
|
return v
|
|
|
|
def invoke(self, context: InvocationContext) -> FloatOutput:
|
|
# Python doesn't support switch statements until 3.10, but InvokeAI supports back to 3.9
|
|
if self.operation == "ADD":
|
|
return FloatOutput(value=self.a + self.b)
|
|
elif self.operation == "SUB":
|
|
return FloatOutput(value=self.a - self.b)
|
|
elif self.operation == "MUL":
|
|
return FloatOutput(value=self.a * self.b)
|
|
elif self.operation == "DIV":
|
|
return FloatOutput(value=self.a / self.b)
|
|
elif self.operation == "EXP":
|
|
return FloatOutput(value=self.a**self.b)
|
|
elif self.operation == "SQRT":
|
|
return FloatOutput(value=np.sqrt(self.a))
|
|
elif self.operation == "ABS":
|
|
return FloatOutput(value=abs(self.a))
|
|
elif self.operation == "MIN":
|
|
return FloatOutput(value=min(self.a, self.b))
|
|
else: # self.operation == "MAX":
|
|
return FloatOutput(value=max(self.a, self.b))
|