Spaces:
Running
Running
File size: 3,595 Bytes
9700f95 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 |
# src/utils/conversation_manager.py
from typing import List, Dict, Optional
import tiktoken
from datetime import datetime
class ConversationManager:
def __init__(
self,
max_tokens: int = 4000,
max_messages: int = 10,
model: str = "gpt-3.5-turbo"
):
"""
Initialize conversation manager
Args:
max_tokens (int): Maximum tokens to keep in context
max_messages (int): Maximum number of messages to keep
model (str): Model name for token counting
"""
self.max_tokens = max_tokens
self.max_messages = max_messages
self.encoding = tiktoken.encoding_for_model(model)
def format_messages(self, messages: List[Dict]) -> str:
"""Format messages into a conversation string"""
formatted = []
for msg in messages:
role = msg.get('role', 'unknown')
content = msg.get('content', '')
formatted.append(f"{role.capitalize()}: {content}")
return "\n".join(formatted)
def count_tokens(self, text: str) -> int:
"""Count tokens in text"""
return len(self.encoding.encode(text))
def get_relevant_history(
self,
messages: List[Dict],
current_query: str,
max_tokens: Optional[int] = None
) -> List[Dict]:
"""
Get relevant conversation history within token limit
Args:
messages (List[Dict]): Full message history
current_query (str): Current user query
max_tokens (Optional[int]): Override default max tokens
Returns:
List[Dict]: Relevant message history
"""
max_tokens = max_tokens or self.max_tokens
current_tokens = self.count_tokens(current_query)
# Keep track of tokens and messages
history = []
total_tokens = current_tokens
# Process messages from most recent to oldest
for msg in reversed(messages[-self.max_messages:]):
msg_text = f"{msg['role']}: {msg['content']}\n"
msg_tokens = self.count_tokens(msg_text)
# Check if adding this message would exceed token limit
if total_tokens + msg_tokens > max_tokens:
break
total_tokens += msg_tokens
history.append(msg)
# Reverse back to chronological order
return list(reversed(history))
def generate_prompt_with_history(
self,
current_query: str,
history: List[Dict],
context_docs: List[str]
) -> str:
"""
Generate a prompt that includes conversation history and context
Args:
current_query (str): Current user query
history (List[Dict]): Relevant conversation history
context_docs (List[str]): Retrieved context documents
Returns:
str: Formatted prompt
"""
# Format conversation history
conversation_context = self.format_messages(history)
# Format context documents
context_str = "\n\n".join(context_docs)
prompt = f"""
Previous Conversation:
{conversation_context}
Relevant Context:
{context_str}
Current Query: {current_query}
Based on the previous conversation and the provided context, please provide a comprehensive and accurate response that maintains continuity with the conversation history."""
return prompt |