This commit is contained in:
oib
2025-07-20 09:26:07 +02:00
parent da28b205e5
commit ab9d93d913
19 changed files with 1207 additions and 419 deletions

266
upload.py
View File

@ -6,11 +6,13 @@ from slowapi.util import get_remote_address
from slowapi.errors import RateLimitExceeded
from pathlib import Path
import json
import requests
from datetime import datetime
from convert_to_opus import convert_to_opus
from models import UploadLog, UserQuota, User
from sqlalchemy import select
from models import UploadLog, UserQuota, User, PublicStream
from sqlalchemy import select, or_
from database import get_db
from sqlalchemy.orm import Session
limiter = Limiter(key_func=get_remote_address)
router = APIRouter()
@ -23,55 +25,63 @@ DATA_ROOT = Path("./data")
@router.post("/upload")
async def upload(request: Request, db = Depends(get_db), uid: str = Form(...), file: UploadFile = Form(...)):
from log import log_violation
import time
# Generate a unique request ID for this upload
request_id = str(int(time.time()))
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] Starting upload of {file.filename}")
try:
user_dir = DATA_ROOT / uid
user_dir.mkdir(parents=True, exist_ok=True)
raw_path = user_dir / ("raw." + file.filename.split(".")[-1])
import uuid
unique_name = str(uuid.uuid4()) + ".opus"
# Save temp upload FIRST
with open(raw_path, "wb") as f:
f.write(await file.read())
# Block music/singing via Ollama prompt
import requests
try:
with open(raw_path, "rb") as f:
audio = f.read()
res = requests.post("http://localhost:11434/api/generate", json={
"model": "whisper",
"prompt": "Does this audio contain music or singing? Answer yes or no only.",
"audio": audio
}, timeout=10)
resp = res.json().get("response", "").lower()
if "yes" in resp:
raw_path.unlink(missing_ok=True)
raise HTTPException(status_code=403, detail="Upload rejected: music or singing detected")
except Exception as ollama_err:
# fallback: allow, log if needed
pass
processed_path = user_dir / unique_name
# Block unconfirmed users (use ORM)
# First, verify the user exists and is confirmed
user = db.exec(select(User).where((User.username == uid) | (User.email == uid))).first()
# If result is a Row or tuple, extract the User object
if user is not None and not isinstance(user, User) and hasattr(user, "__getitem__"):
user = user[0]
from log import log_violation
log_violation("UPLOAD", request.client.host, uid, f"DEBUG: Incoming uid={uid}, user found={user}, confirmed={getattr(user, 'confirmed', None)}")
log_violation("UPLOAD", request.client.host, uid, f"DEBUG: After unpack, user={user}, type={type(user)}, confirmed={getattr(user, 'confirmed', None)}")
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] User check - found: {user is not None}, confirmed: {getattr(user, 'confirmed', False) if user else 'N/A'}")
if not user or not hasattr(user, "confirmed") or not user.confirmed:
raw_path.unlink(missing_ok=True)
raise HTTPException(status_code=403, detail="Account not confirmed")
# DB-based quota check
quota = db.get(UserQuota, uid)
if quota and quota.storage_bytes >= 100 * 1024 * 1024:
raw_path.unlink(missing_ok=True)
# Check quota before doing any file operations
quota = db.get(UserQuota, uid) or UserQuota(uid=uid, storage_bytes=0)
if quota.storage_bytes >= 100 * 1024 * 1024:
raise HTTPException(status_code=400, detail="Quota exceeded")
# Create user directory if it doesn't exist
user_dir = DATA_ROOT / uid
user_dir.mkdir(parents=True, exist_ok=True)
# Generate a unique filename for the processed file first
import uuid
unique_name = f"{uuid.uuid4()}.opus"
raw_ext = file.filename.split(".")[-1].lower()
raw_path = user_dir / ("raw." + raw_ext)
processed_path = user_dir / unique_name
# Clean up any existing raw files first (except the one we're about to create)
for old_file in user_dir.glob('raw.*'):
try:
if old_file != raw_path: # Don't delete the file we're about to create
old_file.unlink(missing_ok=True)
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] Cleaned up old file: {old_file}")
except Exception as e:
log_violation("UPLOAD_ERROR", request.client.host, uid, f"[{request_id}] Failed to clean up {old_file}: {e}")
# Save the uploaded file temporarily
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] Saving temporary file to {raw_path}")
try:
with open(raw_path, "wb") as f:
content = await file.read()
if not content:
raise ValueError("Uploaded file is empty")
f.write(content)
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] Successfully wrote {len(content)} bytes to {raw_path}")
except Exception as e:
log_violation("UPLOAD_ERROR", request.client.host, uid, f"[{request_id}] Failed to save {raw_path}: {e}")
raise HTTPException(status_code=500, detail=f"Failed to save uploaded file: {e}")
# Ollama music/singing check is disabled for this release
log_violation("UPLOAD", request.client.host, uid, f"[{request_id}] Ollama music/singing check is disabled")
try:
convert_to_opus(str(raw_path), str(processed_path))
@ -82,44 +92,96 @@ async def upload(request: Request, db = Depends(get_db), uid: str = Form(...), f
original_size = raw_path.stat().st_size
raw_path.unlink(missing_ok=True) # cleanup
# First, verify the file was created and has content
if not processed_path.exists() or processed_path.stat().st_size == 0:
raise HTTPException(status_code=500, detail="Failed to process audio file")
# Concatenate all .opus files in random order to stream.opus for public playback
# This is now done after the file is in its final location with log ID
from concat_opus import concat_opus_files
try:
concat_opus_files(user_dir, user_dir / "stream.opus")
except Exception as e:
# fallback: just use the latest processed file if concat fails
import shutil
stream_path = user_dir / "stream.opus"
shutil.copy2(processed_path, stream_path)
# Create a log entry with the original filename
log = UploadLog(
uid=uid,
ip=request.client.host,
filename=file.filename, # Store original filename
processed_filename=unique_name, # Store the processed filename
size_bytes=original_size
)
db.add(log)
db.commit()
db.refresh(log)
# Rename the processed file to include the log ID for better tracking
processed_with_id = user_dir / f"{log.id}_{unique_name}"
processed_path.rename(processed_with_id)
processed_path = processed_with_id
# Store updated quota
def update_stream_opus():
try:
concat_opus_files(user_dir, user_dir / "stream.opus")
except Exception as e:
# fallback: just use the latest processed file if concat fails
import shutil
stream_path = user_dir / "stream.opus"
shutil.copy2(processed_path, stream_path)
log_violation("STREAM_UPDATE", request.client.host, uid,
f"[fallback] Updated stream.opus with {processed_path}")
# We'll call this after the file is in its final location
# Get the final file size
size = processed_path.stat().st_size
quota = db.get(UserQuota, uid)
if not quota:
quota = UserQuota(uid=uid)
db.add(quota)
quota.storage_bytes += size
db.commit()
# Update public streams list
update_public_streams(uid, quota.storage_bytes)
# Start a transaction
try:
# Create a log entry with the original filename
log = UploadLog(
uid=uid,
ip=request.client.host,
filename=file.filename, # Store original filename
processed_filename=unique_name, # Store the processed filename
size_bytes=size
)
db.add(log)
db.flush() # Get the log ID without committing
# Rename the processed file to include the log ID for better tracking
processed_with_id = user_dir / f"{log.id}_{unique_name}"
if processed_path.exists():
# First check if there's already a file with the same UUID but different prefix
for existing_file in user_dir.glob(f"*_{unique_name}"):
if existing_file != processed_path:
log_violation("CLEANUP", request.client.host, uid,
f"[UPLOAD] Removing duplicate file: {existing_file}")
existing_file.unlink(missing_ok=True)
# Now do the rename
if processed_path != processed_with_id:
if processed_with_id.exists():
processed_with_id.unlink(missing_ok=True)
processed_path.rename(processed_with_id)
processed_path = processed_with_id
# Only clean up raw.* files, not previously uploaded opus files
for old_temp_file in user_dir.glob('raw.*'):
try:
old_temp_file.unlink(missing_ok=True)
log_violation("CLEANUP", request.client.host, uid, f"[{request_id}] Cleaned up temp file: {old_temp_file}")
except Exception as e:
log_violation("CLEANUP_ERROR", request.client.host, uid, f"[{request_id}] Failed to clean up {old_temp_file}: {e}")
# Get or create quota
quota = db.query(UserQuota).filter(UserQuota.uid == uid).first()
if not quota:
quota = UserQuota(uid=uid, storage_bytes=0)
db.add(quota)
# Update quota with the new file size
quota.storage_bytes = sum(
f.stat().st_size
for f in user_dir.glob('*.opus')
if f.name != 'stream.opus' and f != processed_path
) + size
# Update public streams
update_public_streams(uid, quota.storage_bytes, db)
# Commit the transaction
db.commit()
# Now that the transaction is committed and files are in their final location,
# update the stream.opus file to include all files
update_stream_opus()
except Exception as e:
db.rollback()
# Clean up the processed file if something went wrong
if processed_path.exists():
processed_path.unlink(missing_ok=True)
raise HTTPException(status_code=500, detail=f"Database error: {str(e)}")
return {
"filename": file.filename,
@ -142,37 +204,33 @@ async def upload(request: Request, db = Depends(get_db), uid: str = Form(...), f
return {"detail": f"Server error: {type(e).__name__}: {str(e)}"}
def update_public_streams(uid: str, storage_bytes: int, db = Depends(get_db)):
def update_public_streams(uid: str, storage_bytes: int, db: Session):
"""Update the public streams list in the database with the latest user upload info"""
try:
from models import PublicStream
# Get or create the public stream record
public_stream = db.get(PublicStream, uid)
current_time = datetime.utcnow()
if public_stream is None:
# Create a new record if it doesn't exist
public_stream = PublicStream(
uid=uid,
size=storage_bytes,
mtime=int(current_time.timestamp()),
created_at=current_time,
updated_at=current_time
)
# Get the user's info
user = db.query(User).filter(User.username == uid).first()
if not user:
print(f"[WARNING] User {uid} not found when updating public streams")
return
# Try to get existing public stream or create new one
public_stream = db.query(PublicStream).filter(PublicStream.uid == uid).first()
if not public_stream:
public_stream = PublicStream(uid=uid)
db.add(public_stream)
else:
# Update existing record
public_stream.size = storage_bytes
public_stream.mtime = int(current_time.timestamp())
public_stream.updated_at = current_time
# Update the public stream info
public_stream.username = user.username
public_stream.display_name = user.display_name or user.username
public_stream.storage_bytes = storage_bytes
public_stream.last_updated = datetime.utcnow()
db.commit()
db.refresh(public_stream)
# Don't commit here - let the caller handle the transaction
db.flush()
except Exception as e:
db.rollback()
# Just log the error and let the caller handle the rollback
print(f"[ERROR] Error updating public streams: {e}")
import traceback
print(f"Error updating public streams in database: {e}")
print(traceback.format_exc())
raise
traceback.print_exc()
raise # Re-raise to let the caller handle the error