jketreno/ai-voicebot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Reload working

Browse Source
This commit is contained in:
James Ketr 2025-09-01 15:59:16 -07:00
parent a313209768
commit e64edf92ca
5 changed files with 227 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
shared/models.py
View File

@ -5,6 +5,7 @@ This module contains all the shared data models used for:
- HTTP API requests and responses
- WebSocket message payloads
- Data persistence structures
Test comment for shared reload detection - updated again
"""
from __future__ import annotations

3
voicebot/entrypoint.sh
View File

@ -21,7 +21,8 @@ export PATH="$VIRTUAL_ENV/bin:$PATH"
# Launch voicebot in production or development mode
if [ "$PRODUCTION" != "true" ]; then
echo "Starting voicebot in development mode with auto-reload..."
python3 scripts/reload_runner.py --watch /voicebot --watch /shared -- uv run main.py \
# Fix: Use single --watch argument with multiple paths instead of multiple --watch arguments
python3 -u scripts/reload_runner.py --watch . /shared --verbose --interval 0.5 -- uv run main.py \
--insecure \
--server-url https://ketrenos.com/ai-voicebot \
--lobby default \

6
voicebot/logger.py
View File

@ -48,9 +48,9 @@ def _setup_logging(level: str=logging_level) -> logging.Logger:
warnings.filterwarnings("ignore", message="n_jobs value 1 overridden")
warnings.filterwarnings("ignore", message=".*websocket.*is deprecated")
logging.getLogger("aiortc").setLevel(logging.INFO)
logging.getLogger("aioice").setLevel(logging.INFO)
logging.getLogger("asyncio").setLevel(logging.INFO)
logging.getLogger("aiortc").setLevel(logging.WARNING)
logging.getLogger("aioice").setLevel(logging.WARNING)
logging.getLogger("asyncio").setLevel(logging.WARNING)
numeric_level = getattr(logging, level.upper(), None)
if not isinstance(numeric_level, int):

112
voicebot/main.py
View File

@ -3,6 +3,7 @@ WebRTC Media Agent for Python
This module provides WebRTC signaling server communication and peer connection management.
Synthetic audio/video track creation is handled by the synthetic_media module.
Test change to trigger reload - TESTING RELOAD NOW
"""
from __future__ import annotations
@ -20,6 +21,7 @@ from typing import (
cast,
)
# test
from dataclasses import dataclass, field
from pydantic import ValidationError
@ -60,7 +62,6 @@ from synthetic_media import create_synthetic_tracks
# import debug_aioice
# Generic message payload type
MessageData = dict[str, object]
@ -237,9 +238,9 @@ class WebRTCSignalingClient:
ws = cast(WebSocketProtocol, self.websocket)
try:
logger.info(f"_send_message: Sending {message_type} with data: {data}")
logger.debug(f"_send_message: Sending {message_type} with data: {data}")
await ws.send(json.dumps(message))
logger.info(f"_send_message: Sent message: {message_type}")
logger.debug(f"_send_message: Sent message: {message_type}")
except Exception as e:
logger.error(
f"_send_message: Failed to send {message_type}: {e}", exc_info=True
@ -250,7 +251,7 @@ class WebRTCSignalingClient:
try:
ws = cast(WebSocketProtocol, self.websocket)
async for message in ws:
logger.info(f"_handle_messages: Received raw message: {message}")
logger.debug(f"_handle_messages: Received raw message: {message}")
try:
data = cast(MessageData, json.loads(message))
except Exception as e:
@ -275,7 +276,7 @@ class WebRTCSignalingClient:
logger.error(f"Invalid message structure: {e}", exc_info=True)
return
logger.info(
logger.debug(
f"_process_message: Received message type: {msg_type} with data: {data}"
)
@ -350,14 +351,14 @@ class WebRTCSignalingClient:
logger.info(
f"Adding peer: {peer_name} (should_create_offer: {should_create_offer})"
)
logger.info(
logger.debug(
f"_handle_add_peer: peer_id={peer_id}, peer_name={peer_name}, should_create_offer={should_create_offer}"
)
# Check if peer already exists
if peer_id in self.peer_connections:
pc = self.peer_connections[peer_id]
logger.info(
logger.debug(
f"_handle_add_peer: Existing connection state: {pc.connectionState}"
)
if pc.connectionState in ["new", "connected", "connecting"]:
@ -365,7 +366,7 @@ class WebRTCSignalingClient:
return
else:
# Clean up stale connection
logger.info(
logger.debug(
f"_handle_add_peer: Closing stale connection for {peer_name}"
)
await pc.close()
@ -390,7 +391,7 @@ class WebRTCSignalingClient:
RTCIceServer(urls="stun:stun.l.google.com:19302"),
],
)
logger.info(
logger.debug(
f"_handle_add_peer: Creating RTCPeerConnection for {peer_name} with config: {config}"
)
pc = RTCPeerConnection(configuration=config)
@ -400,7 +401,9 @@ class WebRTCSignalingClient:
logger.info(f"ICE gathering state: {pc.iceGatheringState}")
# Debug: Check if we have any local candidates when gathering is complete
if pc.iceGatheringState == "complete":
logger.info(f"ICE gathering complete for {peer_name} - checking if candidates were generated...")
logger.info(
f"ICE gathering complete for {peer_name} - checking if candidates were generated..."
)
pc.on("icegatheringstatechange")(on_ice_gathering_state_change)
@ -425,9 +428,13 @@ class WebRTCSignalingClient:
def on_ice_candidate(candidate: Optional[RTCIceCandidate]) -> None:
logger.info(f"on_ice_candidate: {candidate}")
logger.info(f"on_ice_candidate CALLED for {peer_name}: candidate={candidate}")
logger.info(
f"on_ice_candidate CALLED for {peer_name}: candidate={candidate}"
)
if not candidate:
logger.info(f"on_ice_candidate: End of candidates signal for {peer_name}")
logger.info(
f"on_ice_candidate: End of candidates signal for {peer_name}"
)
return
# Raw SDP fragment for the candidate
@ -467,7 +474,7 @@ class WebRTCSignalingClient:
# Add local tracks
for track in self.local_tracks.values():
logger.info(
logger.debug(
f"_handle_add_peer: Adding local track {track.kind} to {peer_name}"
)
pc.addTrack(track)
@ -479,34 +486,42 @@ class WebRTCSignalingClient:
self.is_negotiating[peer_id] = True
try:
logger.info(f"_handle_add_peer: Creating offer for {peer_name}")
logger.debug(f"_handle_add_peer: Creating offer for {peer_name}")
offer = await pc.createOffer()
logger.info(f"_handle_add_peer: Offer created for {peer_name}: {offer}")
logger.debug(
f"_handle_add_peer: Offer created for {peer_name}: {offer}"
)
await pc.setLocalDescription(offer)
logger.info(f"_handle_add_peer: Local description set for {peer_name}")
logger.debug(f"_handle_add_peer: Local description set for {peer_name}")
# WORKAROUND for aiortc icecandidate event not firing (GitHub issue #1344)
# Use Method 2: Complete SDP approach to extract ICE candidates
logger.info(f"_handle_add_peer: Waiting for ICE gathering to complete for {peer_name}")
logger.debug(
f"_handle_add_peer: Waiting for ICE gathering to complete for {peer_name}"
)
while pc.iceGatheringState != "complete":
await asyncio.sleep(0.1)
logger.info(f"_handle_add_peer: ICE gathering complete, extracting candidates from SDP for {peer_name}")
logger.debug(
f"_handle_add_peer: ICE gathering complete, extracting candidates from SDP for {peer_name}"
)
# Parse ICE candidates from the local SDP
sdp_lines = pc.localDescription.sdp.split('\n')
candidate_lines = [line for line in sdp_lines if line.startswith('a=candidate:')]
sdp_lines = pc.localDescription.sdp.split("\n")
candidate_lines = [
line for line in sdp_lines if line.startswith("a=candidate:")
]
# Track which media section we're in to determine sdpMid and sdpMLineIndex
current_media_index = -1
current_mid = None
for line in sdp_lines:
if line.startswith('m='): # Media section
if line.startswith("m="): # Media section
current_media_index += 1
elif line.startswith('a=mid:'): # Media ID
current_mid = line.split(':', 1)[1].strip()
elif line.startswith('a=candidate:'):
elif line.startswith("a=mid:"): # Media ID
current_mid = line.split(":", 1)[1].strip()
elif line.startswith("a=candidate:"):
candidate_sdp = line[2:] # Remove 'a=' prefix
candidate_model = ICECandidateDictModel(
@ -520,7 +535,9 @@ class WebRTCSignalingClient:
candidate=candidate_model,
)
logger.info(f"_handle_add_peer: Sending extracted ICE candidate for {peer_name}: {candidate_sdp[:60]}...")
logger.debug(
f"_handle_add_peer: Sending extracted ICE candidate for {peer_name}: {candidate_sdp[:60]}..."
)
await self._send_message(
"relayICECandidate", payload_candidate.model_dump()
)
@ -534,10 +551,14 @@ class WebRTCSignalingClient:
payload_end = IceCandidateModel(
peer_id=peer_id, peer_name=peer_name, candidate=end_candidate_model
)
logger.info(f"_handle_add_peer: Sending end-of-candidates signal for {peer_name}")
logger.debug(
f"_handle_add_peer: Sending end-of-candidates signal for {peer_name}"
)
await self._send_message("relayICECandidate", payload_end.model_dump())
logger.info(f"_handle_add_peer: Sent {len(candidate_lines)} ICE candidates to {peer_name}")
logger.debug(
f"_handle_add_peer: Sent {len(candidate_lines)} ICE candidates to {peer_name}"
)
session_desc_typed = SessionDescriptionTypedModel(
type=offer.type, sdp=offer.sdp
@ -663,26 +684,32 @@ class WebRTCSignalingClient:
# WORKAROUND for aiortc icecandidate event not firing (GitHub issue #1344)
# Use Method 2: Complete SDP approach to extract ICE candidates
logger.info(f"_handle_session_description: Waiting for ICE gathering to complete for {peer_name} (answer)")
logger.debug(
f"_handle_session_description: Waiting for ICE gathering to complete for {peer_name} (answer)"
)
while pc.iceGatheringState != "complete":
await asyncio.sleep(0.1)
logger.info(f"_handle_session_description: ICE gathering complete, extracting candidates from SDP for {peer_name} (answer)")
logger.debug(
f"_handle_session_description: ICE gathering complete, extracting candidates from SDP for {peer_name} (answer)"
)
# Parse ICE candidates from the local SDP
sdp_lines = pc.localDescription.sdp.split('\n')
candidate_lines = [line for line in sdp_lines if line.startswith('a=candidate:')]
sdp_lines = pc.localDescription.sdp.split("\n")
candidate_lines = [
line for line in sdp_lines if line.startswith("a=candidate:")
]
# Track which media section we're in to determine sdpMid and sdpMLineIndex
current_media_index = -1
current_mid = None
for line in sdp_lines:
if line.startswith('m='): # Media section
if line.startswith("m="): # Media section
current_media_index += 1
elif line.startswith('a=mid:'): # Media ID
current_mid = line.split(':', 1)[1].strip()
elif line.startswith('a=candidate:'):
elif line.startswith("a=mid:"): # Media ID
current_mid = line.split(":", 1)[1].strip()
elif line.startswith("a=candidate:"):
candidate_sdp = line[2:] # Remove 'a=' prefix
candidate_model = ICECandidateDictModel(
@ -696,7 +723,9 @@ class WebRTCSignalingClient:
candidate=candidate_model,
)
logger.info(f"_handle_session_description: Sending extracted ICE candidate for {peer_name} (answer): {candidate_sdp[:60]}...")
logger.debug(
f"_handle_session_description: Sending extracted ICE candidate for {peer_name} (answer): {candidate_sdp[:60]}..."
)
await self._send_message(
"relayICECandidate", payload_candidate.model_dump()
)
@ -710,10 +739,14 @@ class WebRTCSignalingClient:
payload_end = IceCandidateModel(
peer_id=peer_id, peer_name=peer_name, candidate=end_candidate_model
)
logger.info(f"_handle_session_description: Sending end-of-candidates signal for {peer_name} (answer)")
logger.debug(
f"_handle_session_description: Sending end-of-candidates signal for {peer_name} (answer)"
)
await self._send_message("relayICECandidate", payload_end.model_dump())
logger.info(f"_handle_session_description: Sent {len(candidate_lines)} ICE candidates to {peer_name} (answer)")
logger.debug(
f"_handle_session_description: Sent {len(candidate_lines)} ICE candidates to {peer_name} (answer)"
)
session_desc_typed = SessionDescriptionTypedModel(
type=answer.type, sdp=answer.sdp
@ -1056,3 +1089,6 @@ if __name__ == "__main__":
# pip install aiortc websockets opencv-python numpy
asyncio.run(main())
# test modification
# Test comment Mon Sep 1 03:48:19 PM PDT 2025
# Test change at Mon Sep 1 03:52:13 PM PDT 2025

134
voicebot/scripts/reload_runner.py
View File

@ -11,6 +11,7 @@ and inside containers without installing extra packages.
from __future__ import annotations
import argparse
import hashlib
import os
import signal
import subprocess
@ -21,28 +22,90 @@ from types import FrameType
def scan_py_mtimes(paths: List[str]) -> Dict[str, float]:
# Directories to skip during scanning
SKIP_DIRS = {
".venv",
"__pycache__",
".git",
"node_modules",
".mypy_cache",
".pytest_cache",
"build",
"dist",
}
mtimes: Dict[str, float] = {}
for p in paths:
if os.path.isfile(p) and p.endswith('.py'):
try:
mtimes[p] = os.path.getmtime(p)
# Use both mtime and ctime to catch more changes in Docker environments
stat = os.stat(p)
mtimes[p] = max(stat.st_mtime, stat.st_ctime)
except OSError:
pass
continue
for root, _, files in os.walk(p):
for root, dirs, files in os.walk(p):
# Skip common directories that shouldn't trigger reloads
dirs[:] = [d for d in dirs if d not in SKIP_DIRS]
for f in files:
if not f.endswith('.py'):
continue
fp = os.path.join(root, f)
try:
mtimes[fp] = os.path.getmtime(fp)
# Use both mtime and ctime to catch more changes in Docker environments
stat = os.stat(fp)
mtimes[fp] = max(stat.st_mtime, stat.st_ctime)
except OSError:
# file might disappear between walk and stat
pass
return mtimes
def scan_py_hashes(paths: List[str]) -> Dict[str, str]:
"""Fallback method: scan file content hashes for change detection."""
# Directories to skip during scanning
SKIP_DIRS = {
".venv",
"__pycache__",
".git",
"node_modules",
".mypy_cache",
".pytest_cache",
"build",
"dist",
}
hashes: Dict[str, str] = {}
for p in paths:
if os.path.isfile(p) and p.endswith(".py"):
try:
with open(p, "rb") as f:
content = f.read()
hashes[p] = hashlib.md5(content).hexdigest()
except OSError:
pass
continue
for root, dirs, files in os.walk(p):
# Skip common directories that shouldn't trigger reloads
dirs[:] = [d for d in dirs if d not in SKIP_DIRS]
for f in files:
if not f.endswith(".py"):
continue
fp = os.path.join(root, f)
try:
with open(fp, "rb") as file:
content = file.read()
hashes[fp] = hashlib.md5(content).hexdigest()
except OSError:
# file might disappear between walk and read
pass
return hashes
def start_process(cmd: List[str]) -> subprocess.Popen[bytes]:
print("Starting:", " ".join(cmd))
return subprocess.Popen(cmd)
@ -66,10 +129,20 @@ def terminate_process(p: subprocess.Popen[bytes], timeout: float = 5.0) -> None:
def main() -> int:
parser = argparse.ArgumentParser(description="Restart a command when .py files change")
parser.add_argument("--watch", "-w", nargs="+", default=["."], help="Directories or files to watch")
parser.add_argument("--interval", "-i", type=float, default=1.0, help="Polling interval in seconds")
parser.add_argument(
"--interval", "-i", type=float, default=0.5, help="Polling interval in seconds"
)
parser.add_argument("--delay-restart", type=float, default=0.1, help="Delay after change before restarting")
parser.add_argument("--no-restart-on-exit", action="store_true", help="Don't restart if the process exits on its own")
parser.add_argument("--pass-sigterm", action="store_true", help="Forward SIGTERM to child and exit when received")
parser.add_argument(
"--verbose", "-v", action="store_true", help="Enable verbose logging"
)
parser.add_argument(
"--use-hash-fallback",
action="store_true",
help="Use content hashing as fallback for Docker environments",
)
# Accept the command to run as a positional "remainder" so callers can
# separate options with `--` and have everything after it treated as the
# command. Defining an option named "--" doesn't work reliably with
@ -93,6 +166,20 @@ def main() -> int:
watch_paths = args.watch
last_mtimes = scan_py_mtimes(watch_paths)
last_hashes = scan_py_hashes(watch_paths) if args.use_hash_fallback else {}
if args.verbose:
print(f"Watching {len(last_mtimes)} Python files in paths: {watch_paths}")
print(f"Working directory: {os.getcwd()}")
print(f"Resolved watch paths: {[os.path.abspath(p) for p in watch_paths]}")
print(f"Polling interval: {args.interval}s")
if args.use_hash_fallback:
print("Using content hash fallback for change detection")
print("Sample files being watched:")
for fp in sorted(last_mtimes.keys())[:5]:
print(f" {fp}")
if len(last_mtimes) > 5:
print(f" ... and {len(last_mtimes) - 5} more")
child = start_process(cmd)
@ -125,31 +212,70 @@ def main() -> int:
# else restart immediately
child = start_process(cmd)
last_mtimes = scan_py_mtimes(watch_paths)
last_hashes = (
scan_py_hashes(watch_paths) if args.use_hash_fallback else {}
)
continue
# Check for source changes
current = scan_py_mtimes(watch_paths)
changed = False
change_reason = ""
# Check for new or changed files
for fp, m in current.items():
if fp not in last_mtimes or last_mtimes.get(fp) != m:
print("Detected change in:", fp)
if args.verbose:
old_mtime = last_mtimes.get(fp, 0)
print(f" Old mtime: {old_mtime}, New mtime: {m}")
changed = True
change_reason = f"mtime change in {fp}"
break
# Hash-based fallback check if mtime didn't detect changes
if not changed and args.use_hash_fallback:
current_hashes = scan_py_hashes(watch_paths)
for fp, h in current_hashes.items():
if fp not in last_hashes or last_hashes.get(fp) != h:
print("Detected content change in:", fp)
if args.verbose:
print(
f" Hash changed: {last_hashes.get(fp, 'None')} -> {h}"
)
changed = True
change_reason = f"content change in {fp}"
break
# Update hash cache
last_hashes = current_hashes
# Check for deleted files
if not changed:
for fp in list(last_mtimes.keys()):
if fp not in current:
print("Detected deleted file:", fp)
changed = True
change_reason = f"deleted file {fp}"
break
# Additional debug output
if args.verbose and not changed:
num_files = len(current)
if num_files != len(last_mtimes):
print(f"File count changed: {len(last_mtimes)} -> {num_files}")
changed = True
change_reason = "file count change"
if changed:
if args.verbose:
print(f"Restarting due to: {change_reason}")
# Small debounce
time.sleep(args.delay_restart)
terminate_process(child)
child = start_process(cmd)
last_mtimes = scan_py_mtimes(watch_paths)
if args.use_hash_fallback:
last_hashes = scan_py_hashes(watch_paths)
except KeyboardInterrupt:
print("Interrupted, shutting down.")