Files
AstrBot/astrbot/core/agent/message.py

362 lines
11 KiB
Python

# Inspired by MoonshotAI/kosong, credits to MoonshotAI/kosong authors for the original implementation.
# License: Apache License 2.0
from typing import Any, ClassVar, Literal, TypeVar, cast
from pydantic import (
BaseModel,
GetCoreSchemaHandler,
PrivateAttr,
ValidationError,
model_serializer,
model_validator,
)
from pydantic_core import core_schema
from typing_extensions import Self
ContentPartT = TypeVar("ContentPartT", bound="ContentPart")
class ContentPart(BaseModel):
"""A part of the content in a message."""
__content_part_registry: ClassVar[dict[str, type["ContentPart"]]] = {}
type: Literal["text", "think", "image_url", "audio_url"]
_no_save: bool = PrivateAttr(default=False)
def __init_subclass__(cls, **kwargs: Any) -> None:
super().__init_subclass__(**kwargs)
invalid_subclass_error_msg = f"ContentPart subclass {cls.__name__} must have a `type` field of type `str`"
type_value = getattr(cls, "type", None)
if type_value is None or not isinstance(type_value, str):
raise ValueError(invalid_subclass_error_msg)
cls.__content_part_registry[type_value] = cls
@classmethod
def __get_pydantic_core_schema__(
cls,
source_type: Any,
handler: GetCoreSchemaHandler,
) -> core_schema.CoreSchema:
# If we're dealing with the base ContentPart class, use custom validation
if cls.__name__ == "ContentPart":
def validate_content_part(value: Any) -> Any:
# if it's already an instance of a ContentPart subclass, return it
if hasattr(value, "__class__") and issubclass(value.__class__, cls):
return value
# if it's a dict with a type field, dispatch to the appropriate subclass
if isinstance(value, dict) and "type" in value:
type_value: Any | None = cast("dict[str, Any]", value).get("type")
if not isinstance(type_value, str):
raise ValueError(f"Cannot validate {value} as ContentPart")
target_class = cls.__content_part_registry[type_value]
part = target_class.model_validate(value)
if cast("dict[str, Any]", value).get("_no_save"):
part._no_save = True
return part
raise ValueError(f"Cannot validate {value} as ContentPart")
return core_schema.no_info_plain_validator_function(validate_content_part)
# for subclasses, use the default schema
return handler(source_type)
def mark_as_temp(self) -> Self:
"""Mark this content part as provider-facing only, not persisted."""
self._no_save = True
return self
def model_dump_for_context(self) -> dict[str, Any]:
data = self.model_dump()
if self._no_save:
data["_no_save"] = True
return data
class TextPart(ContentPart):
"""TextPart(text="Hello, world!").model_dump()
{'type': 'text', 'text': 'Hello, world!'}
"""
type: str = "text"
text: str
class ThinkPart(ContentPart):
"""ThinkPart(think="I think I need to think about this.").model_dump()
{'type': 'think', 'think': 'I think I need to think about this.', 'encrypted': None}
"""
type: str = "think"
think: str
encrypted: str | None = None
"""Encrypted thinking content, or signature."""
def merge_in_place(self, other: Any) -> bool:
if not isinstance(other, ThinkPart):
return False
if self.encrypted:
return False
self.think += other.think
if other.encrypted:
self.encrypted = other.encrypted
return True
class ImageURLPart(ContentPart):
"""ImageURLPart(image_url="http://example.com/image.jpg").model_dump()
{'type': 'image_url', 'image_url': 'http://example.com/image.jpg'}
"""
class ImageURL(BaseModel):
url: str
"""The URL of the image, can be data URI scheme like `data:image/png;base64,...`."""
id: str | None = None
"""The ID of the image, to allow LLMs to distinguish different images."""
type: str = "image_url"
image_url: ImageURL
class AudioURLPart(ContentPart):
"""AudioURLPart(audio_url=AudioURLPart.AudioURL(url="https://example.com/audio.mp3")).model_dump()
{'type': 'audio_url', 'audio_url': {'url': 'https://example.com/audio.mp3', 'id': None}}
"""
class AudioURL(BaseModel):
url: str
"""The URL of the audio, can be data URI scheme like `data:audio/aac;base64,...`."""
id: str | None = None
"""The ID of the audio, to allow LLMs to distinguish different audios."""
type: str = "audio_url"
audio_url: AudioURL
class ToolCall(BaseModel):
"""A tool call requested by the assistant.
ToolCall(
... id="123",
... function=ToolCall.FunctionBody(
... name="function",
... arguments="{}"
... ),
... ).model_dump()
{'type': 'function', 'id': '123', 'function': {'name': 'function', 'arguments': '{}'}}
"""
class FunctionBody(BaseModel):
name: str
arguments: str | None
type: Literal["function"] = "function"
id: str
"""The ID of the tool call."""
function: FunctionBody
"""The function body of the tool call."""
extra_content: dict[str, Any] | None = None
"""Extra metadata for the tool call."""
@model_serializer(mode="wrap")
def serialize(self, handler):
data = handler(self)
if self.extra_content is None:
data.pop("extra_content", None)
return data
class ToolCallPart(BaseModel):
"""A part of the tool call."""
arguments_part: str | None = None
"""A part of the arguments of the tool call."""
class CheckpointData(BaseModel):
"""Internal checkpoint data for linking LLM turns to platform history."""
id: str
CHECKPOINT_ROLE = "_checkpoint"
class Message(BaseModel):
"""A message in a conversation."""
role: Literal[
"system",
"user",
"assistant",
"tool",
"_checkpoint",
]
content: str | list[ContentPart] | CheckpointData | None = None
"""The content of the message."""
tool_calls: list[ToolCall] | list[dict] | None = None
"""The tool calls of the message."""
tool_call_id: str | None = None
"""The ID of the tool call."""
_no_save: bool = PrivateAttr(default=False)
_checkpoint_after: CheckpointData | None = PrivateAttr(default=None)
@model_validator(mode="after")
def check_content_required(self):
if self.role == CHECKPOINT_ROLE:
if not isinstance(self.content, CheckpointData):
raise ValueError("checkpoint message content must be CheckpointData")
return self
if isinstance(self.content, CheckpointData):
raise ValueError("CheckpointData is only allowed for role='_checkpoint'")
# assistant + tool_calls is not None: allow content to be None
if self.role == "assistant" and self.tool_calls is not None:
return self
# other all cases: content is required
if self.content is None:
raise ValueError(
"content is required unless role='assistant' and tool_calls is not None",
)
return self
@model_serializer(mode="wrap")
def serialize(self, handler):
data = handler(self)
if self.tool_calls is None:
data.pop("tool_calls", None)
if self.tool_call_id is None:
data.pop("tool_call_id", None)
return data
class AssistantMessageSegment(Message):
"""A message segment from the assistant."""
role: Literal["assistant"] = "assistant"
class ToolCallMessageSegment(Message):
"""A message segment representing a tool call."""
role: Literal["tool"] = "tool"
class UserMessageSegment(Message):
"""A message segment from the user."""
role: Literal["user"] = "user"
class SystemMessageSegment(Message):
"""A message segment from the system."""
role: Literal["system"] = "system"
class CheckpointMessageSegment(Message):
"""Internal checkpoint segment for persisted conversation history."""
role: Literal["_checkpoint"] = "_checkpoint"
content: CheckpointData | None = None
def is_checkpoint_message(message: Message | dict) -> bool:
"""Return whether a message is an internal checkpoint."""
if isinstance(message, Message):
return message.role == CHECKPOINT_ROLE
return isinstance(message, dict) and message.get("role") == CHECKPOINT_ROLE
def get_checkpoint_id(message: Message | dict) -> str | None:
"""Return the checkpoint id from an internal checkpoint message."""
if not is_checkpoint_message(message):
return None
content = (
message.content if isinstance(message, Message) else message.get("content")
)
if isinstance(content, CheckpointData):
return content.id
if isinstance(content, dict):
checkpoint_id = content.get("id")
return (
checkpoint_id if isinstance(checkpoint_id, str) and checkpoint_id else None
)
return None
def strip_checkpoint_messages(history: list[dict]) -> list[dict]:
"""Remove internal checkpoint messages from provider-facing history."""
return [message for message in history if not is_checkpoint_message(message)]
def _get_checkpoint_data(message: Message | dict) -> CheckpointData | None:
if not is_checkpoint_message(message):
return None
content = (
message.content if isinstance(message, Message) else message.get("content")
)
if isinstance(content, CheckpointData):
return content
if isinstance(content, dict):
try:
return CheckpointData.model_validate(content)
except ValidationError:
return None
return None
def bind_checkpoint_messages(history: list[dict]) -> list[Message]:
"""Load persisted history and bind checkpoint segments to prior messages."""
messages: list[Message] = []
for item in history:
if is_checkpoint_message(item):
checkpoint = _get_checkpoint_data(item)
if checkpoint is not None and messages:
messages[-1]._checkpoint_after = checkpoint
continue
message = Message.model_validate(item)
if item.get("_no_save"):
message._no_save = True
messages.append(message)
return messages
def dump_messages_with_checkpoints(messages: list[Message]) -> list[dict]:
"""Dump runtime messages and reinsert bound checkpoint segments."""
dumped: list[dict] = []
for message in messages:
message_data = message.model_dump()
if isinstance(message.content, list):
message_data["content"] = [
part.model_dump()
for part in message.content
if not getattr(part, "_no_save", False)
]
dumped.append(message_data)
if message._checkpoint_after is not None:
dumped.append(
CheckpointMessageSegment(
content=message._checkpoint_after,
).model_dump(),
)
return dumped