mirror of
https://github.com/blakeblackshear/frigate.git
synced 2026-04-28 23:06:13 +02:00
Early 0.18 work (#22138)
* Update version * Create scaffolding for case management (#21293) * implement case management for export apis (#21295) * refactor vainfo to search for first GPU (#21296) use existing LibvaGpuSelector to pick appropritate libva device * Case management UI (#21299) * Refactor export cards to match existing cards in other UI pages * Show cases separately from exports * Add proper filtering and display of cases * Add ability to edit and select cases for exports * Cleanup typing * Hide if no unassigned * Cleanup hiding logic * fix scrolling * Improve layout * Camera connection quality indicator (#21297) * add camera connection quality metrics and indicator * formatting * move stall calcs to watchdog * clean up * change watchdog to 1s and separately track time for ffmpeg retry_interval * implement status caching to reduce message volume * Export filter UI (#21322) * Get started on export filters * implement basic filter * Implement filtering and adjust api * Improve filter handling * Improve navigation * Cleanup * handle scrolling * Refactor temperature reporting for detectors and implement Hailo temp reading (#21395) * Add Hailo temperature retrieval * Refactor `get_hailo_temps()` to use ctxmanager * Show Hailo temps in system UI * Move hailo_platform import to get_hailo_temps * Refactor temperatures calculations to use within detector block * Adjust webUI to handle new location --------- Co-authored-by: tigattack <10629864+tigattack@users.noreply.github.com> * Camera-specific hwaccel settings for timelapse exports (correct base) (#21386) * added hwaccel_args to camera.record.export config struct * populate camera.record.export.hwaccel_args with a cascade up to camera then global if 'auto' * use new hwaccel args in export * added documentation for camera-specific hwaccel export * fix c/p error * missed an import * fleshed out the docs and comments a bit * ruff lint * separated out the tips in the doc * fix documentation * fix and simplify reference config doc * Add support for GPU and NPU temperatures (#21495) * Add rockchip temps * Add support for GPU and NPU temperatures in the frontend * Add support for Nvidia temperature * Improve separation * Adjust graph scaling * Exports Improvements (#21521) * Add images to case folder view * Add ability to select case in export dialog * Add to mobile review too * Add API to handle deleting recordings (#21520) * Add recording delete API * Re-organize recordings apis * Fix import * Consolidate query types * Add media sync API endpoint (#21526) * add media cleanup functions * add endpoint * remove scheduled sync recordings from cleanup * move to utils dir * tweak import * remove sync_recordings and add config migrator * remove sync_recordings * docs * remove key * clean up docs * docs fix * docs tweak * Media sync API refactor and UI (#21542) * generic job infrastructure * types and dispatcher changes for jobs * save data in memory only for completed jobs * implement media sync job and endpoints * change logs to debug * websocket hook and types * frontend * i18n * docs tweaks * endpoint descriptions * tweak docs * use same logging pattern in sync_recordings as the other sync functions (#21625) * Fix incorrect counting in sync_recordings (#21626) * Update go2rtc to v1.9.13 (#21648) Co-authored-by: Eugeny Tulupov <eugeny.tulupov@spirent.com> * Refactor Time-Lapse Export (#21668) * refactor time lapse creation to be a separate API call with ability to pass arbitrary ffmpeg args * Add CPU fallback * Optimize empty directory cleanup for recordings (#21695) The previous empty directory cleanup did a full recursive directory walk, which can be extremely slow. This new implementation only removes directories which have a chance of being empty due to a recent file deletion. * Implement llama.cpp GenAI Provider (#21690) * Implement llama.cpp GenAI Provider * Add docs * Update links * Fix broken mqtt links * Fix more broken anchors * Remove parents in remove_empty_directories (#21726) The original implementation did a full directory tree walk to find and remove empty directories, so this implementation should remove the parents as well, like the original did. * Implement LLM Chat API with tool calling support (#21731) * Implement initial tools definiton APIs * Add initial chat completion API with tool support * Implement other providers * Cleanup * Offline preview image (#21752) * use latest preview frame for latest image when camera is offline * remove frame extraction logic * tests * frontend * add description to api endpoint * Update to ROCm 7.2.0 (#21753) * Update to ROCm 7.2.0 * ROCm now works properly with JinaV1 * Arcface has compilation error * Add live context tool to LLM (#21754) * Add live context tool * Improve handling of images in request * Improve prompt caching * Add networking options for configuring listening ports (#21779) * feat: add X-Frame-Time when returning snapshot (#21932) Co-authored-by: Florent MORICONI <170678386+fmcloudconsulting@users.noreply.github.com> * Improve jsmpeg player websocket handling (#21943) * improve jsmpeg player websocket handling prevent websocket console messages from appearing when player is destroyed * reformat files after ruff upgrade * Allow API Events to be Detections or Alerts, depending on the Event Label (#21923) * - API created events will be alerts OR detections, depending on the event label, defaulting to alerts - Indefinite API events will extend the recording segment until those events are ended - API event start time is the actual start time, instead of having a pre-buffer of record.event_pre_capture * Instead of checking for indefinite events on a camera before deciding if we should end the segment, only update last_detection_time and last_alert_time if frame_time is greater, which should have the same effect * Add the ability to set a pre_capture number of seconds when creating a manual event via the API. Default behavior unchanged * Remove unnecessary _publish_segment_start() call * Formatting * handle last_alert_time or last_detection_time being None when checking them against the frame_time * comment manual_info["label"].split(": ")[0] for clarity * ffmpeg Preview Segment Optimization for "high" and "very_high" (#21996) * Introduce qmax parameter for ffmpeg preview encoding Added PREVIEW_QMAX_PARAM to control ffmpeg encoding quality. * formatting * Fix spacing in qmax parameters for preview quality * Adapt to new Gemini format * Fix frame time access * Remove exceptions * Cleanup --------- Co-authored-by: Josh Hawkins <32435876+hawkeye217@users.noreply.github.com> Co-authored-by: tigattack <10629864+tigattack@users.noreply.github.com> Co-authored-by: Andrew Roberts <adroberts@gmail.com> Co-authored-by: Eugeny Tulupov <zhekka3@gmail.com> Co-authored-by: Eugeny Tulupov <eugeny.tulupov@spirent.com> Co-authored-by: John Shaw <1753078+johnshaw@users.noreply.github.com> Co-authored-by: Eric Work <work.eric@gmail.com> Co-authored-by: FL42 <46161216+fl42@users.noreply.github.com> Co-authored-by: Florent MORICONI <170678386+fmcloudconsulting@users.noreply.github.com> Co-authored-by: nulledy <254504350+nulledy@users.noreply.github.com>
This commit is contained in:
Nicolas Mowen
GitHub
642
frigate/api/chat.py
Normal file
642
frigate/api/chat.py
Normal file
@@ -0,0 +1,642 @@
|
||||
"""Chat and LLM tool calling APIs."""
|
||||
|
||||
import base64
|
||||
import json
|
||||
import logging
|
||||
from datetime import datetime, timezone
|
||||
from typing import Any, Dict, List, Optional
|
||||
|
||||
import cv2
|
||||
from fastapi import APIRouter, Body, Depends, Request
|
||||
from fastapi.responses import JSONResponse
|
||||
from pydantic import BaseModel
|
||||
|
||||
from frigate.api.auth import (
|
||||
allow_any_authenticated,
|
||||
get_allowed_cameras_for_filter,
|
||||
)
|
||||
from frigate.api.defs.query.events_query_parameters import EventsQueryParams
|
||||
from frigate.api.defs.request.chat_body import ChatCompletionRequest
|
||||
from frigate.api.defs.response.chat_response import (
|
||||
ChatCompletionResponse,
|
||||
ChatMessageResponse,
|
||||
)
|
||||
from frigate.api.defs.tags import Tags
|
||||
from frigate.api.event import events
|
||||
from frigate.genai import get_genai_client
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(tags=[Tags.chat])
|
||||
|
||||
|
||||
class ToolExecuteRequest(BaseModel):
|
||||
"""Request model for tool execution."""
|
||||
|
||||
tool_name: str
|
||||
arguments: Dict[str, Any]
|
||||
|
||||
|
||||
def get_tool_definitions() -> List[Dict[str, Any]]:
|
||||
"""
|
||||
Get OpenAI-compatible tool definitions for Frigate.
|
||||
|
||||
Returns a list of tool definitions that can be used with OpenAI-compatible
|
||||
function calling APIs.
|
||||
"""
|
||||
return [
|
||||
{
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": "search_objects",
|
||||
"description": (
|
||||
"Search for detected objects in Frigate by camera, object label, time range, "
|
||||
"zones, and other filters. Use this to answer questions about when "
|
||||
"objects were detected, what objects appeared, or to find specific object detections. "
|
||||
"An 'object' in Frigate represents a tracked detection (e.g., a person, package, car)."
|
||||
),
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"camera": {
|
||||
"type": "string",
|
||||
"description": "Camera name to filter by (optional). Use 'all' for all cameras.",
|
||||
},
|
||||
"label": {
|
||||
"type": "string",
|
||||
"description": "Object label to filter by (e.g., 'person', 'package', 'car').",
|
||||
},
|
||||
"after": {
|
||||
"type": "string",
|
||||
"description": "Start time in ISO 8601 format (e.g., '2024-01-01T00:00:00Z').",
|
||||
},
|
||||
"before": {
|
||||
"type": "string",
|
||||
"description": "End time in ISO 8601 format (e.g., '2024-01-01T23:59:59Z').",
|
||||
},
|
||||
"zones": {
|
||||
"type": "array",
|
||||
"items": {"type": "string"},
|
||||
"description": "List of zone names to filter by.",
|
||||
},
|
||||
"limit": {
|
||||
"type": "integer",
|
||||
"description": "Maximum number of objects to return (default: 10).",
|
||||
"default": 10,
|
||||
},
|
||||
},
|
||||
},
|
||||
"required": [],
|
||||
},
|
||||
},
|
||||
{
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": "get_live_context",
|
||||
"description": (
|
||||
"Get the current detection information for a camera: objects being tracked, "
|
||||
"zones, timestamps. Use this to understand what is visible in the live view. "
|
||||
"Call this when the user has included a live image (via include_live_image) or "
|
||||
"when answering questions about what is happening right now on a specific camera."
|
||||
),
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"camera": {
|
||||
"type": "string",
|
||||
"description": "Camera name to get live context for.",
|
||||
},
|
||||
},
|
||||
"required": ["camera"],
|
||||
},
|
||||
},
|
||||
},
|
||||
]
|
||||
|
||||
|
||||
@router.get(
|
||||
"/chat/tools",
|
||||
dependencies=[Depends(allow_any_authenticated())],
|
||||
summary="Get available tools",
|
||||
description="Returns OpenAI-compatible tool definitions for function calling.",
|
||||
)
|
||||
def get_tools(request: Request) -> JSONResponse:
|
||||
"""Get list of available tools for LLM function calling."""
|
||||
tools = get_tool_definitions()
|
||||
return JSONResponse(content={"tools": tools})
|
||||
|
||||
|
||||
async def _execute_search_objects(
|
||||
request: Request,
|
||||
arguments: Dict[str, Any],
|
||||
allowed_cameras: List[str],
|
||||
) -> JSONResponse:
|
||||
"""
|
||||
Execute the search_objects tool.
|
||||
|
||||
This searches for detected objects (events) in Frigate using the same
|
||||
logic as the events API endpoint.
|
||||
"""
|
||||
# Parse ISO 8601 timestamps to Unix timestamps if provided
|
||||
after = arguments.get("after")
|
||||
before = arguments.get("before")
|
||||
|
||||
if after:
|
||||
try:
|
||||
after_dt = datetime.fromisoformat(after.replace("Z", "+00:00"))
|
||||
after = after_dt.timestamp()
|
||||
except (ValueError, AttributeError):
|
||||
logger.warning(f"Invalid 'after' timestamp format: {after}")
|
||||
after = None
|
||||
|
||||
if before:
|
||||
try:
|
||||
before_dt = datetime.fromisoformat(before.replace("Z", "+00:00"))
|
||||
before = before_dt.timestamp()
|
||||
except (ValueError, AttributeError):
|
||||
logger.warning(f"Invalid 'before' timestamp format: {before}")
|
||||
before = None
|
||||
|
||||
# Convert zones array to comma-separated string if provided
|
||||
zones = arguments.get("zones")
|
||||
if isinstance(zones, list):
|
||||
zones = ",".join(zones)
|
||||
elif zones is None:
|
||||
zones = "all"
|
||||
|
||||
# Build query parameters compatible with EventsQueryParams
|
||||
query_params = EventsQueryParams(
|
||||
camera=arguments.get("camera", "all"),
|
||||
cameras=arguments.get("camera", "all"),
|
||||
label=arguments.get("label", "all"),
|
||||
labels=arguments.get("label", "all"),
|
||||
zones=zones,
|
||||
zone=zones,
|
||||
after=after,
|
||||
before=before,
|
||||
limit=arguments.get("limit", 10),
|
||||
)
|
||||
|
||||
try:
|
||||
# Call the events endpoint function directly
|
||||
# The events function is synchronous and takes params and allowed_cameras
|
||||
response = events(query_params, allowed_cameras)
|
||||
|
||||
# The response is already a JSONResponse with event data
|
||||
# Return it as-is for the LLM
|
||||
return response
|
||||
except Exception as e:
|
||||
logger.error(f"Error executing search_objects: {e}", exc_info=True)
|
||||
return JSONResponse(
|
||||
content={
|
||||
"success": False,
|
||||
"message": "Error searching objects",
|
||||
},
|
||||
status_code=500,
|
||||
)
|
||||
|
||||
|
||||
@router.post(
|
||||
"/chat/execute",
|
||||
dependencies=[Depends(allow_any_authenticated())],
|
||||
summary="Execute a tool",
|
||||
description="Execute a tool function call from an LLM.",
|
||||
)
|
||||
async def execute_tool(
|
||||
request: Request,
|
||||
body: ToolExecuteRequest = Body(...),
|
||||
allowed_cameras: List[str] = Depends(get_allowed_cameras_for_filter),
|
||||
) -> JSONResponse:
|
||||
"""
|
||||
Execute a tool function call.
|
||||
|
||||
This endpoint receives tool calls from LLMs and executes the corresponding
|
||||
Frigate operations, returning results in a format the LLM can understand.
|
||||
"""
|
||||
tool_name = body.tool_name
|
||||
arguments = body.arguments
|
||||
|
||||
logger.debug(f"Executing tool: {tool_name} with arguments: {arguments}")
|
||||
|
||||
if tool_name == "search_objects":
|
||||
return await _execute_search_objects(request, arguments, allowed_cameras)
|
||||
|
||||
return JSONResponse(
|
||||
content={
|
||||
"success": False,
|
||||
"message": f"Unknown tool: {tool_name}",
|
||||
"tool": tool_name,
|
||||
},
|
||||
status_code=400,
|
||||
)
|
||||
|
||||
|
||||
async def _execute_get_live_context(
|
||||
request: Request,
|
||||
camera: str,
|
||||
allowed_cameras: List[str],
|
||||
) -> Dict[str, Any]:
|
||||
if camera not in allowed_cameras:
|
||||
return {
|
||||
"error": f"Camera '{camera}' not found or access denied",
|
||||
}
|
||||
|
||||
if camera not in request.app.frigate_config.cameras:
|
||||
return {
|
||||
"error": f"Camera '{camera}' not found",
|
||||
}
|
||||
|
||||
try:
|
||||
frame_processor = request.app.detected_frames_processor
|
||||
camera_state = frame_processor.camera_states.get(camera)
|
||||
|
||||
if camera_state is None:
|
||||
return {
|
||||
"error": f"Camera '{camera}' state not available",
|
||||
}
|
||||
|
||||
tracked_objects_dict = {}
|
||||
with camera_state.current_frame_lock:
|
||||
tracked_objects = camera_state.tracked_objects.copy()
|
||||
frame_time = camera_state.current_frame_time
|
||||
|
||||
for obj_id, tracked_obj in tracked_objects.items():
|
||||
obj_dict = tracked_obj.to_dict()
|
||||
if obj_dict.get("frame_time") == frame_time:
|
||||
tracked_objects_dict[obj_id] = {
|
||||
"label": obj_dict.get("label"),
|
||||
"zones": obj_dict.get("current_zones", []),
|
||||
"sub_label": obj_dict.get("sub_label"),
|
||||
"stationary": obj_dict.get("stationary", False),
|
||||
}
|
||||
|
||||
return {
|
||||
"camera": camera,
|
||||
"timestamp": frame_time,
|
||||
"detections": list(tracked_objects_dict.values()),
|
||||
}
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Error executing get_live_context: {e}", exc_info=True)
|
||||
return {
|
||||
"error": "Error getting live context",
|
||||
}
|
||||
|
||||
|
||||
async def _get_live_frame_image_url(
|
||||
request: Request,
|
||||
camera: str,
|
||||
allowed_cameras: List[str],
|
||||
) -> Optional[str]:
|
||||
"""
|
||||
Fetch the current live frame for a camera as a base64 data URL.
|
||||
|
||||
Returns None if the frame cannot be retrieved. Used when include_live_image
|
||||
is set to attach the image to the first user message.
|
||||
"""
|
||||
if (
|
||||
camera not in allowed_cameras
|
||||
or camera not in request.app.frigate_config.cameras
|
||||
):
|
||||
return None
|
||||
try:
|
||||
frame_processor = request.app.detected_frames_processor
|
||||
if camera not in frame_processor.camera_states:
|
||||
return None
|
||||
frame = frame_processor.get_current_frame(camera, {})
|
||||
if frame is None:
|
||||
return None
|
||||
height, width = frame.shape[:2]
|
||||
max_dimension = 1024
|
||||
if height > max_dimension or width > max_dimension:
|
||||
scale = max_dimension / max(height, width)
|
||||
frame = cv2.resize(
|
||||
frame,
|
||||
(int(width * scale), int(height * scale)),
|
||||
interpolation=cv2.INTER_AREA,
|
||||
)
|
||||
_, img_encoded = cv2.imencode(".jpg", frame, [cv2.IMWRITE_JPEG_QUALITY, 85])
|
||||
b64 = base64.b64encode(img_encoded.tobytes()).decode("utf-8")
|
||||
return f"data:image/jpeg;base64,{b64}"
|
||||
except Exception as e:
|
||||
logger.debug("Failed to get live frame for %s: %s", camera, e)
|
||||
return None
|
||||
|
||||
|
||||
async def _execute_tool_internal(
|
||||
tool_name: str,
|
||||
arguments: Dict[str, Any],
|
||||
request: Request,
|
||||
allowed_cameras: List[str],
|
||||
) -> Dict[str, Any]:
|
||||
"""
|
||||
Internal helper to execute a tool and return the result as a dict.
|
||||
|
||||
This is used by the chat completion endpoint to execute tools.
|
||||
"""
|
||||
if tool_name == "search_objects":
|
||||
response = await _execute_search_objects(request, arguments, allowed_cameras)
|
||||
try:
|
||||
if hasattr(response, "body"):
|
||||
body_str = response.body.decode("utf-8")
|
||||
return json.loads(body_str)
|
||||
elif hasattr(response, "content"):
|
||||
return response.content
|
||||
else:
|
||||
return {}
|
||||
except (json.JSONDecodeError, AttributeError) as e:
|
||||
logger.warning(f"Failed to extract tool result: {e}")
|
||||
return {"error": "Failed to parse tool result"}
|
||||
elif tool_name == "get_live_context":
|
||||
camera = arguments.get("camera")
|
||||
if not camera:
|
||||
return {"error": "Camera parameter is required"}
|
||||
return await _execute_get_live_context(request, camera, allowed_cameras)
|
||||
else:
|
||||
return {"error": f"Unknown tool: {tool_name}"}
|
||||
|
||||
|
||||
@router.post(
|
||||
"/chat/completion",
|
||||
response_model=ChatCompletionResponse,
|
||||
dependencies=[Depends(allow_any_authenticated())],
|
||||
summary="Chat completion with tool calling",
|
||||
description=(
|
||||
"Send a chat message to the configured GenAI provider with tool calling support. "
|
||||
"The LLM can call Frigate tools to answer questions about your cameras and events."
|
||||
),
|
||||
)
|
||||
async def chat_completion(
|
||||
request: Request,
|
||||
body: ChatCompletionRequest = Body(...),
|
||||
allowed_cameras: List[str] = Depends(get_allowed_cameras_for_filter),
|
||||
) -> JSONResponse:
|
||||
"""
|
||||
Chat completion endpoint with tool calling support.
|
||||
|
||||
This endpoint:
|
||||
1. Gets the configured GenAI client
|
||||
2. Gets tool definitions
|
||||
3. Sends messages + tools to LLM
|
||||
4. Handles tool_calls if present
|
||||
5. Executes tools and sends results back to LLM
|
||||
6. Repeats until final answer
|
||||
7. Returns response to user
|
||||
"""
|
||||
genai_client = get_genai_client(request.app.frigate_config)
|
||||
if not genai_client:
|
||||
return JSONResponse(
|
||||
content={
|
||||
"error": "GenAI is not configured. Please configure a GenAI provider in your Frigate config.",
|
||||
},
|
||||
status_code=400,
|
||||
)
|
||||
|
||||
tools = get_tool_definitions()
|
||||
conversation = []
|
||||
|
||||
current_datetime = datetime.now(timezone.utc)
|
||||
current_date_str = current_datetime.strftime("%Y-%m-%d")
|
||||
current_time_str = current_datetime.strftime("%H:%M:%S %Z")
|
||||
|
||||
cameras_info = []
|
||||
config = request.app.frigate_config
|
||||
for camera_id in allowed_cameras:
|
||||
if camera_id not in config.cameras:
|
||||
continue
|
||||
camera_config = config.cameras[camera_id]
|
||||
friendly_name = (
|
||||
camera_config.friendly_name
|
||||
if camera_config.friendly_name
|
||||
else camera_id.replace("_", " ").title()
|
||||
)
|
||||
cameras_info.append(f" - {friendly_name} (ID: {camera_id})")
|
||||
|
||||
cameras_section = ""
|
||||
if cameras_info:
|
||||
cameras_section = (
|
||||
"\n\nAvailable cameras:\n"
|
||||
+ "\n".join(cameras_info)
|
||||
+ "\n\nWhen users refer to cameras by their friendly name (e.g., 'Back Deck Camera'), use the corresponding camera ID (e.g., 'back_deck_cam') in tool calls."
|
||||
)
|
||||
|
||||
live_image_note = ""
|
||||
if body.include_live_image:
|
||||
live_image_note = (
|
||||
f"\n\nThe first user message includes a live image from camera "
|
||||
f"'{body.include_live_image}'. Use get_live_context for that camera to get "
|
||||
"current detection details (objects, zones) to aid in understanding the image."
|
||||
)
|
||||
|
||||
system_prompt = f"""You are a helpful assistant for Frigate, a security camera NVR system. You help users answer questions about their cameras, detected objects, and events.
|
||||
|
||||
Current date and time: {current_date_str} at {current_time_str} (UTC)
|
||||
|
||||
When users ask questions about "today", "yesterday", "this week", etc., use the current date above as reference.
|
||||
When searching for objects or events, use ISO 8601 format for dates (e.g., {current_date_str}T00:00:00Z for the start of today).
|
||||
Always be accurate with time calculations based on the current date provided.{cameras_section}{live_image_note}"""
|
||||
|
||||
conversation.append(
|
||||
{
|
||||
"role": "system",
|
||||
"content": system_prompt,
|
||||
}
|
||||
)
|
||||
|
||||
first_user_message_seen = False
|
||||
for msg in body.messages:
|
||||
msg_dict = {
|
||||
"role": msg.role,
|
||||
"content": msg.content,
|
||||
}
|
||||
if msg.tool_call_id:
|
||||
msg_dict["tool_call_id"] = msg.tool_call_id
|
||||
if msg.name:
|
||||
msg_dict["name"] = msg.name
|
||||
|
||||
if (
|
||||
msg.role == "user"
|
||||
and not first_user_message_seen
|
||||
and body.include_live_image
|
||||
):
|
||||
first_user_message_seen = True
|
||||
image_url = await _get_live_frame_image_url(
|
||||
request, body.include_live_image, allowed_cameras
|
||||
)
|
||||
if image_url:
|
||||
msg_dict["content"] = [
|
||||
{"type": "text", "text": msg.content},
|
||||
{"type": "image_url", "image_url": {"url": image_url}},
|
||||
]
|
||||
|
||||
conversation.append(msg_dict)
|
||||
|
||||
tool_iterations = 0
|
||||
max_iterations = body.max_tool_iterations
|
||||
|
||||
logger.debug(
|
||||
f"Starting chat completion with {len(conversation)} message(s), "
|
||||
f"{len(tools)} tool(s) available, max_iterations={max_iterations}"
|
||||
)
|
||||
|
||||
try:
|
||||
while tool_iterations < max_iterations:
|
||||
logger.debug(
|
||||
f"Calling LLM (iteration {tool_iterations + 1}/{max_iterations}) "
|
||||
f"with {len(conversation)} message(s) in conversation"
|
||||
)
|
||||
response = genai_client.chat_with_tools(
|
||||
messages=conversation,
|
||||
tools=tools if tools else None,
|
||||
tool_choice="auto",
|
||||
)
|
||||
|
||||
if response.get("finish_reason") == "error":
|
||||
logger.error("GenAI client returned an error")
|
||||
return JSONResponse(
|
||||
content={
|
||||
"error": "An error occurred while processing your request.",
|
||||
},
|
||||
status_code=500,
|
||||
)
|
||||
|
||||
assistant_message = {
|
||||
"role": "assistant",
|
||||
"content": response.get("content"),
|
||||
}
|
||||
if response.get("tool_calls"):
|
||||
assistant_message["tool_calls"] = [
|
||||
{
|
||||
"id": tc["id"],
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": tc["name"],
|
||||
"arguments": json.dumps(tc["arguments"]),
|
||||
},
|
||||
}
|
||||
for tc in response["tool_calls"]
|
||||
]
|
||||
conversation.append(assistant_message)
|
||||
|
||||
tool_calls = response.get("tool_calls")
|
||||
if not tool_calls:
|
||||
logger.debug(
|
||||
f"Chat completion finished with final answer (iterations: {tool_iterations})"
|
||||
)
|
||||
return JSONResponse(
|
||||
content=ChatCompletionResponse(
|
||||
message=ChatMessageResponse(
|
||||
role="assistant",
|
||||
content=response.get("content"),
|
||||
tool_calls=None,
|
||||
),
|
||||
finish_reason=response.get("finish_reason", "stop"),
|
||||
tool_iterations=tool_iterations,
|
||||
).model_dump(),
|
||||
)
|
||||
|
||||
# Execute tools
|
||||
tool_iterations += 1
|
||||
logger.debug(
|
||||
f"Tool calls detected (iteration {tool_iterations}/{max_iterations}): "
|
||||
f"{len(tool_calls)} tool(s) to execute"
|
||||
)
|
||||
tool_results = []
|
||||
|
||||
for tool_call in tool_calls:
|
||||
tool_name = tool_call["name"]
|
||||
tool_args = tool_call["arguments"]
|
||||
tool_call_id = tool_call["id"]
|
||||
|
||||
logger.debug(
|
||||
f"Executing tool: {tool_name} (id: {tool_call_id}) with arguments: {json.dumps(tool_args, indent=2)}"
|
||||
)
|
||||
|
||||
try:
|
||||
tool_result = await _execute_tool_internal(
|
||||
tool_name, tool_args, request, allowed_cameras
|
||||
)
|
||||
|
||||
if isinstance(tool_result, dict):
|
||||
result_content = json.dumps(tool_result)
|
||||
result_summary = tool_result
|
||||
if isinstance(tool_result, dict) and isinstance(
|
||||
tool_result.get("content"), list
|
||||
):
|
||||
result_count = len(tool_result.get("content", []))
|
||||
result_summary = {
|
||||
"count": result_count,
|
||||
"sample": tool_result.get("content", [])[:2]
|
||||
if result_count > 0
|
||||
else [],
|
||||
}
|
||||
logger.debug(
|
||||
f"Tool {tool_name} (id: {tool_call_id}) completed successfully. "
|
||||
f"Result: {json.dumps(result_summary, indent=2)}"
|
||||
)
|
||||
elif isinstance(tool_result, str):
|
||||
result_content = tool_result
|
||||
logger.debug(
|
||||
f"Tool {tool_name} (id: {tool_call_id}) completed successfully. "
|
||||
f"Result length: {len(result_content)} characters"
|
||||
)
|
||||
else:
|
||||
result_content = str(tool_result)
|
||||
logger.debug(
|
||||
f"Tool {tool_name} (id: {tool_call_id}) completed successfully. "
|
||||
f"Result type: {type(tool_result).__name__}"
|
||||
)
|
||||
|
||||
tool_results.append(
|
||||
{
|
||||
"role": "tool",
|
||||
"tool_call_id": tool_call_id,
|
||||
"content": result_content,
|
||||
}
|
||||
)
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"Error executing tool {tool_name} (id: {tool_call_id}): {e}",
|
||||
exc_info=True,
|
||||
)
|
||||
error_content = json.dumps({"error": "Tool execution failed"})
|
||||
tool_results.append(
|
||||
{
|
||||
"role": "tool",
|
||||
"tool_call_id": tool_call_id,
|
||||
"content": error_content,
|
||||
}
|
||||
)
|
||||
logger.debug(
|
||||
f"Tool {tool_name} (id: {tool_call_id}) failed. Error result added to conversation."
|
||||
)
|
||||
|
||||
conversation.extend(tool_results)
|
||||
logger.debug(
|
||||
f"Added {len(tool_results)} tool result(s) to conversation. "
|
||||
f"Continuing with next LLM call..."
|
||||
)
|
||||
|
||||
logger.warning(
|
||||
f"Max tool iterations ({max_iterations}) reached. Returning partial response."
|
||||
)
|
||||
return JSONResponse(
|
||||
content=ChatCompletionResponse(
|
||||
message=ChatMessageResponse(
|
||||
role="assistant",
|
||||
content="I reached the maximum number of tool call iterations. Please try rephrasing your question.",
|
||||
tool_calls=None,
|
||||
),
|
||||
finish_reason="length",
|
||||
tool_iterations=tool_iterations,
|
||||
).model_dump(),
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Error in chat completion: {e}", exc_info=True)
|
||||
return JSONResponse(
|
||||
content={
|
||||
"error": "An error occurred while processing your request.",
|
||||
},
|
||||
status_code=500,
|
||||
)
|
||||
Reference in New Issue
Block a user