Spaces:
Sleeping
Sleeping
File size: 12,884 Bytes
b7a3e32 a546147 6cbca40 c0827a3 7f15e1c 6cbca40 4909ef6 b9b869f b7a3e32 a546147 6cbca40 b9b869f 6cbca40 b7a3e32 4909ef6 b7a3e32 6cbca40 b7a3e32 b9b869f b7a3e32 c0827a3 b7a3e32 c0827a3 b7a3e32 c0827a3 b7a3e32 c0827a3 b7a3e32 a546147 7f15e1c a546147 61e4b1e a546147 7f15e1c 61e4b1e a546147 61e4b1e a546147 61e4b1e a546147 61e4b1e 7f15e1c 61e4b1e a546147 61e4b1e a546147 61e4b1e a546147 61e4b1e a546147 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 |
from fastapi import (
APIRouter,
status,
HTTPException,
File,
UploadFile,
Form,
)
from fastapi.responses import JSONResponse, StreamingResponse
from src.utils.logger import logger
from src.agents.role_play.flow import role_play_agent
from src.services.tts_service import tts_service
from pydantic import BaseModel, Field
from typing import Dict, Any, Optional
from src.agents.role_play.scenarios import get_scenarios
import json
import base64
import asyncio
router = APIRouter(prefix="/ai", tags=["AI"])
class RoleplayRequest(BaseModel):
query: str = Field(..., description="User's query for the AI agent")
session_id: str = Field(
..., description="Session ID for tracking user interactions"
)
scenario: Dict[str, Any] = Field(..., description="The scenario for the roleplay")
@router.get("/scenarios", status_code=status.HTTP_200_OK)
async def list_scenarios():
"""Get all available scenarios"""
return JSONResponse(content=get_scenarios())
@router.post("/roleplay", status_code=status.HTTP_200_OK)
async def roleplay(
session_id: str = Form(
..., description="Session ID for tracking user interactions"
),
scenario: str = Form(
..., description="The scenario for the roleplay as JSON string"
),
text_message: Optional[str] = Form(None, description="Text message from user"),
audio_file: Optional[UploadFile] = File(None, description="Audio file from user"),
):
"""Send a message (text or audio) to the roleplay agent"""
logger.info(f"Received roleplay request: {session_id}")
# Validate that at least one input is provided
if not text_message and not audio_file:
raise HTTPException(
status_code=400, detail="Either text_message or audio_file must be provided"
)
# Parse scenario from JSON string
try:
scenario_dict = json.loads(scenario)
except json.JSONDecodeError:
raise HTTPException(status_code=400, detail="Invalid scenario JSON format")
if not scenario_dict:
raise HTTPException(status_code=400, detail="Scenario not provided")
# Prepare message content
message_content = []
# Handle text input
if text_message:
message_content.append({"type": "text", "text": text_message})
# Handle audio input
if audio_file:
try:
# Read audio file content
audio_data = await audio_file.read()
# Convert to base64
audio_base64 = base64.b64encode(audio_data).decode("utf-8")
# Determine mime type based on file extension
file_extension = (
audio_file.filename.split(".")[-1].lower()
if audio_file.filename
else "wav"
)
mime_type_map = {
"wav": "audio/wav",
"mp3": "audio/mpeg",
"ogg": "audio/ogg",
"webm": "audio/webm",
"m4a": "audio/mp4",
}
mime_type = mime_type_map.get(file_extension, "audio/wav")
message_content.append(
{
"type": "audio",
"source_type": "base64",
"data": audio_base64,
"mime_type": mime_type,
}
)
except Exception as e:
logger.error(f"Error processing audio file: {str(e)}")
raise HTTPException(
status_code=400, detail=f"Error processing audio file: {str(e)}"
)
# Create message in the required format
message = {"role": "user", "content": message_content}
try:
response = await role_play_agent().ainvoke(
{
"messages": [message],
"scenario_title": scenario_dict["scenario_title"],
"scenario_description": scenario_dict["scenario_description"],
"scenario_context": scenario_dict["scenario_context"],
"your_role": scenario_dict["your_role"],
"key_vocabulary": scenario_dict["key_vocabulary"],
},
{"configurable": {"thread_id": session_id}},
)
# Extract AI response content
ai_response = response["messages"][-1].content
logger.info(f"AI response: {ai_response}")
return JSONResponse(content={"response": ai_response})
except Exception as e:
logger.error(f"Error in roleplay: {str(e)}")
raise HTTPException(status_code=500, detail=f"Internal server error: {str(e)}")
@router.post("/roleplay/stream", status_code=status.HTTP_200_OK)
async def roleplay_stream(
session_id: str = Form(
..., description="Session ID for tracking user interactions"
),
scenario: str = Form(
..., description="The scenario for the roleplay as JSON string"
),
text_message: Optional[str] = Form(None, description="Text message from user"),
audio_file: Optional[UploadFile] = File(None, description="Audio file from user"),
audio: bool = Form(False, description="Whether to return TTS audio response"),
):
"""Send a message (text or audio) to the roleplay agent with streaming response"""
logger.info(f"Received streaming roleplay request: {session_id}")
# Validate that at least one input is provided
if not text_message and not audio_file:
raise HTTPException(
status_code=400, detail="Either text_message or audio_file must be provided"
)
# Parse scenario from JSON string
try:
scenario_dict = json.loads(scenario)
except json.JSONDecodeError:
raise HTTPException(status_code=400, detail="Invalid scenario JSON format")
if not scenario_dict:
raise HTTPException(status_code=400, detail="Scenario not provided")
# Prepare message content
message_content = []
# Handle text input
if text_message:
message_content.append({"type": "text", "text": text_message})
# Handle audio input
if audio_file:
try:
# Read audio file content
audio_data = await audio_file.read()
# Convert to base64
audio_base64 = base64.b64encode(audio_data).decode("utf-8")
# Determine mime type based on file extension
file_extension = (
audio_file.filename.split(".")[-1].lower()
if audio_file.filename
else "wav"
)
mime_type_map = {
"wav": "audio/wav",
"mp3": "audio/mpeg",
"ogg": "audio/ogg",
"webm": "audio/webm",
"m4a": "audio/mp4",
}
mime_type = mime_type_map.get(file_extension, "audio/wav")
message_content.append(
{
"type": "audio",
"source_type": "base64",
"data": audio_base64,
"mime_type": mime_type,
}
)
except Exception as e:
logger.error(f"Error processing audio file: {str(e)}")
raise HTTPException(
status_code=400, detail=f"Error processing audio file: {str(e)}"
)
# Create message in the required format
message = {"role": "user", "content": message_content}
async def generate_stream():
"""Generator function for streaming responses"""
accumulated_content = ""
conversation_ended = False
try:
input_graph = {
"messages": [message],
"scenario_title": scenario_dict["scenario_title"],
"scenario_description": scenario_dict["scenario_description"],
"scenario_context": scenario_dict["scenario_context"],
"your_role": scenario_dict["your_role"],
"key_vocabulary": scenario_dict["key_vocabulary"],
}
config = {"configurable": {"thread_id": session_id}}
async for event in role_play_agent().astream(
input=input_graph,
stream_mode=["messages", "values"],
config=config,
subgraphs=True,
):
_, event_type, message_chunk = event
if event_type == "messages":
# message_chunk is a tuple, get the first element which is the actual AIMessageChunk
if isinstance(message_chunk, tuple) and len(message_chunk) > 0:
actual_message = message_chunk[0]
content = getattr(actual_message, "content", "")
else:
actual_message = message_chunk
content = getattr(message_chunk, "content", "")
# Check if this is a tool call message and if it's an end conversation tool call
if (
hasattr(actual_message, "tool_calls")
and actual_message.tool_calls
):
# Check if any tool call is for ending the conversation
for tool_call in actual_message.tool_calls:
if (
isinstance(tool_call, dict)
and tool_call.get("name") == "end_conversation"
):
# Send a special termination message to the client
termination_data = {
"type": "termination",
"content": "Conversation ended",
"reason": tool_call.get("args", {}).get("reason", "Unknown reason")
}
yield f"data: {json.dumps(termination_data)}\n\n"
conversation_ended = True
break
if content and not conversation_ended:
# Accumulate content for TTS
accumulated_content += content
# Create SSE-formatted response
response_data = {
"type": "message_chunk",
"content": content,
"metadata": {
"agent": getattr(actual_message, "name", "unknown"),
"id": getattr(actual_message, "id", ""),
"usage_metadata": getattr(
actual_message, "usage_metadata", {}
),
},
}
yield f"data: {json.dumps(response_data)}\n\n"
# Small delay to prevent overwhelming the client
await asyncio.sleep(0.01)
# Only send completion signal if conversation wasn't ended by tool call
if not conversation_ended:
# Generate TTS audio if requested
audio_data = None
if audio and accumulated_content.strip():
try:
logger.info(
f"Generating TTS for accumulated content: {len(accumulated_content)} chars"
)
audio_result = await tts_service.text_to_speech(accumulated_content)
if audio_result:
audio_data = {
"audio_data": audio_result["audio_data"],
"mime_type": audio_result["mime_type"],
"format": audio_result["format"],
}
logger.info("TTS audio generated successfully")
else:
logger.warning("TTS generation failed")
except Exception as tts_error:
logger.error(f"TTS generation error: {str(tts_error)}")
# Send completion signal with optional audio
completion_data = {"type": "completion", "content": "", "audio": audio_data}
yield f"data: {json.dumps(completion_data)}\n\n"
except Exception as e:
logger.error(f"Error in streaming roleplay: {str(e)}")
error_data = {"type": "error", "content": str(e)}
yield f"data: {json.dumps(error_data)}\n\n"
return StreamingResponse(
generate_stream(),
media_type="text/plain",
headers={
"Cache-Control": "no-cache",
"Connection": "keep-alive",
"Content-Type": "text/event-stream",
},
)
|