from __future__ import annotations
import logging
from enum import Enum
from typing import TYPE_CHECKING, Any, Dict, List, Optional
from langchain_core.chat_history import BaseChatMessageHistory
from langchain_core.messages import (
AIMessage,
BaseMessage,
HumanMessage,
SystemMessage,
)
if TYPE_CHECKING:
from zep_python import Memory, MemorySearchResult, Message, NotFoundError
logger = logging.getLogger(__name__)
[docs]class SearchScope(str, Enum):
"""Which documents to search. Messages or Summaries?"""
messages = "messages"
"""Search chat history messages."""
summary = "summary"
"""Search chat history summaries."""
[docs]class SearchType(str, Enum):
"""Enumerator of the types of search to perform."""
similarity = "similarity"
"""Similarity search."""
mmr = "mmr"
"""Maximal Marginal Relevance reranking of similarity search."""
[docs]class ZepChatMessageHistory(BaseChatMessageHistory):
"""Chat message history that uses Zep as a backend.
Recommended usage::
# Set up Zep Chat History
zep_chat_history = ZepChatMessageHistory(
session_id=session_id,
url=ZEP_API_URL,
api_key=<your_api_key>,
)
# Use a standard ConversationBufferMemory to encapsulate the Zep chat history
memory = ConversationBufferMemory(
memory_key="chat_history", chat_memory=zep_chat_history
)
Zep provides long-term conversation storage for LLM apps. The server stores,
summarizes, embeds, indexes, and enriches conversational AI chat
histories, and exposes them via simple, low-latency APIs.
For server installation instructions and more, see:
https://docs.getzep.com/deployment/quickstart/
This class is a thin wrapper around the zep-python package. Additional
Zep functionality is exposed via the `zep_summary` and `zep_messages`
properties.
For more information on the zep-python package, see:
https://github.com/getzep/zep-python
"""
[docs] def __init__(
self,
session_id: str,
url: str = "http://localhost:8000",
api_key: Optional[str] = None,
) -> None:
try:
from zep_python import ZepClient
except ImportError:
raise ImportError(
"Could not import zep-python package. "
"Please install it with `pip install zep-python`."
)
self.zep_client = ZepClient(base_url=url, api_key=api_key)
self.session_id = session_id
@property
def messages(self) -> List[BaseMessage]: # type: ignore
"""Retrieve messages from Zep memory"""
zep_memory: Optional[Memory] = self._get_memory()
if not zep_memory:
return []
messages: List[BaseMessage] = []
# Extract summary, if present, and messages
if zep_memory.summary:
if len(zep_memory.summary.content) > 0:
messages.append(SystemMessage(content=zep_memory.summary.content))
if zep_memory.messages:
msg: Message
for msg in zep_memory.messages:
metadata: Dict = {
"uuid": msg.uuid,
"created_at": msg.created_at,
"token_count": msg.token_count,
"metadata": msg.metadata,
}
if msg.role == "ai":
messages.append(
AIMessage(content=msg.content, additional_kwargs=metadata)
)
else:
messages.append(
HumanMessage(content=msg.content, additional_kwargs=metadata)
)
return messages
@property
def zep_messages(self) -> List[Message]:
"""Retrieve summary from Zep memory"""
zep_memory: Optional[Memory] = self._get_memory()
if not zep_memory:
return []
return zep_memory.messages
@property
def zep_summary(self) -> Optional[str]:
"""Retrieve summary from Zep memory"""
zep_memory: Optional[Memory] = self._get_memory()
if not zep_memory or not zep_memory.summary:
return None
return zep_memory.summary.content
def _get_memory(self) -> Optional[Memory]:
"""Retrieve memory from Zep"""
from zep_python import NotFoundError
try:
zep_memory: Memory = self.zep_client.memory.get_memory(self.session_id)
except NotFoundError:
logger.warning(
f"Session {self.session_id} not found in Zep. Returning None"
)
return None
return zep_memory
[docs] def add_user_message( # type: ignore[override]
self, message: str, metadata: Optional[Dict[str, Any]] = None
) -> None:
"""Convenience method for adding a human message string to the store.
Args:
message: The string contents of a human message.
metadata: Optional metadata to attach to the message.
"""
self.add_message(HumanMessage(content=message), metadata=metadata)
[docs] def add_ai_message( # type: ignore[override]
self, message: str, metadata: Optional[Dict[str, Any]] = None
) -> None:
"""Convenience method for adding an AI message string to the store.
Args:
message: The string contents of an AI message.
metadata: Optional metadata to attach to the message.
"""
self.add_message(AIMessage(content=message), metadata=metadata)
[docs] def add_message(
self, message: BaseMessage, metadata: Optional[Dict[str, Any]] = None
) -> None:
"""Append the message to the Zep memory history"""
from zep_python import Memory, Message
zep_message = Message(
content=message.content, role=message.type, metadata=metadata
)
zep_memory = Memory(messages=[zep_message])
self.zep_client.memory.add_memory(self.session_id, zep_memory)
[docs] def search(
self,
query: str,
metadata: Optional[Dict] = None,
search_scope: SearchScope = SearchScope.messages,
search_type: SearchType = SearchType.similarity,
mmr_lambda: Optional[float] = None,
limit: Optional[int] = None,
) -> List[MemorySearchResult]:
"""Search Zep memory for messages matching the query"""
from zep_python import MemorySearchPayload
payload = MemorySearchPayload(
text=query,
metadata=metadata,
search_scope=search_scope,
search_type=search_type,
mmr_lambda=mmr_lambda,
)
return self.zep_client.memory.search_memory(
self.session_id, payload, limit=limit
)
[docs] def clear(self) -> None:
"""Clear session memory from Zep. Note that Zep is long-term storage for memory
and this is not advised unless you have specific data retention requirements.
"""
try:
self.zep_client.memory.delete_memory(self.session_id)
except NotFoundError:
logger.warning(
f"Session {self.session_id} not found in Zep. Skipping delete."
)