Files
OSIT-AE-API-FastAPI/app/routers/api_v3_actions_hosted_file.py
Scott Idem 07609bae9a feat(hosted-file): implement hash-based download action and flexible auth
- Adds GET /v3/action/hosted_file/hash/{sha256}/download for direct content-addressable storage access.
- Updates V3 authentication dependencies to support 'api_key' in the query parameter (alias 'api_key').
- Implements auth_method: 'api_key' for machine-to-machine requests that provide a valid key but no user/account context.
- Updates GUIDE__V3_FRONTEND_API.md with the new endpoint and auth options.
2026-02-03 13:44:07 -05:00

418 lines
17 KiB
Python

from fastapi import APIRouter, Depends, File, Form, Header, HTTPException, Path, Query, Response, status, UploadFile
from fastapi.responses import FileResponse, StreamingResponse
import aiofiles
import mimetypes
import os
import pathlib
from typing import Dict, List, Optional, Set, Union
import asyncio
import logging
log = logging.getLogger(__name__)
from app.config import settings
from app.db_sql import redis_lookup_id_random, sql_select, sql_update, sql_delete, get_id_random
from app.methods.hosted_file_methods import (
create_hosted_file_obj, load_hosted_file_obj, save_file,
create_hosted_file_link, delete_hosted_file_link, get_hosted_file_link_rec_list
)
from app.lib_general_v3 import (
AccountContext, get_account_context, get_account_context_optional,
SerializationParams, DelayParams
)
from app.models.hosted_file_models import Hosted_File_Base
from app.models.response_models import Resp_Body_Base, mk_resp
"""
Aether API V3 - Hosted File Action Router
------------------------------------------
Handles specialized binary operations like uploads, downloads, and complex deletions.
These routes complement the standard CRUD metadata routes.
"""
router = APIRouter()
# --- Helpers ---
def validate_file_extension(filename: str, allowed_extensions: List[str]):
"""
Backup check for file extensions.
"""
if not allowed_extensions:
return True
ext = filename.rsplit('.', 1)[-1].lower()
if ext not in [e.lower().strip('.') for e in allowed_extensions]:
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=f"File extension '.{ext}' is not allowed. Allowed: {', '.join(allowed_extensions)}"
)
return True
async def file_streamer(path: str, start: int, end: int):
chunk_size = 8192 # 8KB
async with aiofiles.open(path, mode='rb') as f:
await f.seek(start)
while True:
chunk_start = await f.tell()
if chunk_start >= end:
break
bytes_to_read = min(chunk_size, end - chunk_start)
data = await f.read(bytes_to_read)
if not data:
break
yield data
# --- Routes ---
@router.post('/upload', response_model=Resp_Body_Base)
async def upload_files_action(
file_list: List[UploadFile] = File(...),
account_id: str = Form(..., min_length=11, max_length=22),
link_to_type: str = Form(...),
link_to_id: str = Form(..., min_length=11, max_length=22),
allowed_extensions: Optional[List[str]] = Query(None),
account: AccountContext = Depends(get_account_context),
delay: DelayParams = Depends(),
):
"""
V3 Enhanced Upload Action.
- Handles multiple files.
- Resolves IDs to integers.
- Deduplicates via Hash lookup.
- Returns clean Vision IDs.
"""
if delay.sleep_time_s > 0: await asyncio.sleep(delay.sleep_time_s)
# 1. Resolve Parent IDs
account_id_random = account_id
if res_account_id := redis_lookup_id_random(record_id_random=account_id, table_name='account'):
account_id_int = res_account_id
else:
raise HTTPException(status_code=400, detail="Invalid account_id.")
link_to_id_random = link_to_id
if res_link_id := redis_lookup_id_random(record_id_random=link_to_id, table_name=link_to_type):
link_to_id_int = res_link_id
else:
raise HTTPException(status_code=400, detail=f"Invalid link_to_id for type {link_to_type}.")
hosted_file_list = []
for file_obj in file_list:
# 2. Extension Validation
validate_file_extension(file_obj.filename, allowed_extensions)
# 3. Physical Save
file_info = await save_file(
file = file_obj,
account_id = account_id_int,
account_id_random = account_id_random,
link_to_type = link_to_type,
link_to_id = link_to_id_int,
link_to_id_random = link_to_id_random,
check_allowed_extension = False, # Handled by validate_file_extension above
)
if not file_info.get('saved'):
log.error(f"Failed to save file: {file_obj.filename}")
continue
hosted_file_id_int = None
hosted_file_dict = {}
# 4. Database Synchronization (Deduplication)
log.info(f"Syncing DB record for hash: {file_info['hash_sha256']}")
if existing_rec := sql_select(
table_name = 'hosted_file',
field_name = 'hash_sha256',
field_value = file_info['hash_sha256'],
):
# Use existing record
hosted_file_id_int = existing_rec.get('id')
# Migration check: Update subdirectory if missing
if not existing_rec.get('subdirectory_path') and file_info.get('subdirectory_path'):
sql_update(
table_name = 'hosted_file',
data = {'id': hosted_file_id_int, 'subdirectory_path': file_info['subdirectory_path']}
)
hosted_file_dict = load_hosted_file_obj(hosted_file_id=hosted_file_id_int, model_as_dict=True)
else:
# Create new record
file_info['account_id'] = account_id_int
file_info['account_id_random'] = account_id_random
new_hosted_file_obj = Hosted_File_Base(**file_info)
if res_new_id := create_hosted_file_obj(hosted_file_obj_new=new_hosted_file_obj):
hosted_file_id_int = res_new_id
hosted_file_dict = load_hosted_file_obj(hosted_file_id=hosted_file_id_int, model_as_dict=True)
else:
log.error("Database insertion failed for hosted_file.")
continue
# 5. Relational Linking
if hosted_file_id_int:
create_hosted_file_link(
account_id = account_id_int,
hosted_file_id = hosted_file_id_int,
link_to_type = link_to_type,
link_to_id = link_to_id_int
)
# 6. Response Preparation (Vision IDs)
# Add metadata flags
hosted_file_dict['already_exists'] = file_info.get('already_exists')
hosted_file_dict['saved'] = file_info.get('saved')
hosted_file_dict['copy_timer'] = file_info.get('copy_timer')
# Ensure ID is a random string for the frontend
if not isinstance(hosted_file_dict.get('id'), str):
rid = get_id_random(hosted_file_id_int, table_name='hosted_file')
hosted_file_dict['id'] = rid
hosted_file_dict['hosted_file_id'] = rid
hosted_file_list.append(hosted_file_dict)
return mk_resp(data=hosted_file_list, status_message=f"Successfully processed {len(hosted_file_list)} files.")
@router.get('/{hosted_file_id}/download')
async def download_file_action(
response: Response,
hosted_file_id: str = Path(min_length=11, max_length=22),
filename: Optional[str] = Query(None, min_length=4, max_length=255),
site_key: Optional[str] = Query(None), # Bypass API Key/JWT if valid site key provided
range: Optional[str] = Header(None),
account: AccountContext = Depends(get_account_context_optional),
delay: DelayParams = Depends(),
):
"""
Enhanced download/streaming logic.
Supports byte-range seeking, delay simulation, and site_key bypass.
"""
if delay.sleep_time_s > 0: await asyncio.sleep(delay.sleep_time_s)
# 1. Auth Bypass Logic (site_key)
is_authorized = False
if account.auth_method != 'guest':
is_authorized = True
elif site_key:
# Verify site key existence and status
sql = "SELECT id FROM site WHERE auth_key = :key AND enable = true LIMIT 1"
if site_res := sql_select(sql=sql, data={'key': site_key}):
is_authorized = True
log.info(f"Auth Bypass: Download authorized via site_key.")
if not is_authorized:
raise HTTPException(status_code=status.HTTP_403_FORBIDDEN, detail="Authentication required or invalid site_key.")
# 2. Resolve File Record
# ID Vision: Attempt to resolve the ID.
# If not found in hosted_file, check if it's an event_file ID that we can resolve to a hosted_file.
resolved_id = redis_lookup_id_random(record_id_random=hosted_file_id, table_name='hosted_file')
if not resolved_id:
log.info(f"ID {hosted_file_id} not found in hosted_file. Checking event_file...")
if ef_id := redis_lookup_id_random(record_id_random=hosted_file_id, table_name='event_file'):
if ef_rec := sql_select(sql="SELECT hosted_file_id FROM event_file WHERE id = :id", data={'id': ef_id}):
resolved_id = ef_rec.get('hosted_file_id')
log.info(f"Resolved event_file {hosted_file_id} to hosted_file {resolved_id}")
if not resolved_id:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Hosted file record not found.")
hosted_file_obj = load_hosted_file_obj(hosted_file_id=resolved_id)
if not hosted_file_obj:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Hosted file data could not be loaded.")
# 3. Path Resolution
hosted_files_path = settings.FILES_PATH['hosted_files_root']
subdir_path = hosted_file_obj.subdirectory_path
hash_sha256 = hosted_file_obj.hash_sha256
hash_filename = f"{hash_sha256}.file"
if subdir_path:
full_file_path = os.path.join(hosted_files_path, subdir_path, hash_filename)
else:
full_file_path = os.path.join(hosted_files_path, hash_filename)
if not os.path.exists(full_file_path):
log.error(f"File not found on disk: {full_file_path}")
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Physical file not found on server.")
# 4. Streaming / Download logic
target_filename = filename or hosted_file_obj.filename
media_type = mimetypes.guess_type(target_filename)[0] or 'application/octet-stream'
if range:
file_size = os.stat(full_file_path).st_size
try:
range_parts = range.replace('bytes=', '').split('-')
start = int(range_parts[0])
end = int(range_parts[1]) if len(range_parts) > 1 and range_parts[1] else file_size - 1
except (ValueError, IndexError):
raise HTTPException(status_code=status.HTTP_416_REQUESTED_RANGE_NOT_SATISFIABLE)
if start >= file_size:
raise HTTPException(status_code=status.HTTP_416_REQUESTED_RANGE_NOT_SATISFIABLE)
end = min(end, file_size - 1)
content_length = end - start + 1
return StreamingResponse(
file_streamer(full_file_path, start, end + 1),
media_type = media_type,
status_code = status.HTTP_206_PARTIAL_CONTENT,
headers = {
'Accept-Ranges': 'bytes',
'Content-Range': f'bytes {start}-{end}/{file_size}',
'Content-Length': str(content_length),
'Content-Disposition': f'attachment; filename="{target_filename}"'
}
)
return FileResponse(full_file_path, filename=target_filename, media_type=media_type)
@router.get('/hash/{sha256}/download')
async def download_file_by_hash_action(
response: Response,
sha256: str = Path(min_length=64, max_length=64, regex='^[a-f0-9]{64}$'),
filename: Optional[str] = Query(None, min_length=4, max_length=255),
account: AccountContext = Depends(get_account_context_optional),
delay: DelayParams = Depends(),
):
"""
Direct hash-based download (Content-Addressable).
- Skips DB lookup for path resolution.
- Requires a valid API Key (via header or ?api_key=).
- Ideal for local caching systems like Events Launcher.
"""
if delay.sleep_time_s > 0: await asyncio.sleep(delay.sleep_time_s)
# 1. Mandatory Auth Check
# For now, we strictly require a valid machine API key (auth_method will not be 'guest')
if account.auth_method == 'guest':
raise HTTPException(
status_code=status.HTTP_403_FORBIDDEN,
detail="Valid API Key required for hash-based downloads."
)
# 2. Path Resolution (Deterministic)
hosted_files_path = settings.FILES_PATH['hosted_files_root']
subdir = sha256[0:2]
hash_filename = f"{sha256}.file"
full_file_path = os.path.join(hosted_files_path, subdir, hash_filename)
if not os.path.exists(full_file_path):
# Fallback to root (legacy structure)
full_file_path = os.path.join(hosted_files_path, hash_filename)
if not os.path.exists(full_file_path):
log.error(f"Hash-based file not found: {sha256}")
raise HTTPException(status_code=404, detail="File not found on server.")
# 3. Serve File
target_filename = filename or f"file_{sha256[:8]}.bin"
media_type = mimetypes.guess_type(target_filename)[0] or 'application/octet-stream'
return FileResponse(full_file_path, filename=target_filename, media_type=media_type)
@router.delete('/{hosted_file_id}', response_model=Resp_Body_Base)
async def delete_file_action(
hosted_file_id: str = Path(min_length=11, max_length=22),
link_to_type: Optional[str] = Query(None),
link_to_id: Optional[str] = Query(None),
method: str = Query('hide', regex='^(hide|disable|delete)$'),
rm_orphan: bool = Query(False),
fake_delete: bool = Query(False), # Testing mode
account: AccountContext = Depends(get_account_context),
delay: DelayParams = Depends(),
):
"""
Intelligent relational deletion.
- Removes specified link.
- Counts remaining links.
- Optionally cleans up orphans.
- Supports fake_delete for testing.
"""
if delay.sleep_time_s > 0: await asyncio.sleep(delay.sleep_time_s)
# 1. Resolve IDs
file_id_int = redis_lookup_id_random(record_id_random=hosted_file_id, table_name='hosted_file')
if not file_id_int:
raise HTTPException(status_code=404, detail="Hosted file record not found.")
link_id_int = None
if link_to_type and link_to_id:
link_id_int = redis_lookup_id_random(record_id_random=link_to_id, table_name=link_to_type)
if not link_id_int:
raise HTTPException(status_code=404, detail=f"Linked object {link_to_id} not found.")
# 2. Verify State (Existence Checks)
hosted_file_obj = load_hosted_file_obj(hosted_file_id=file_id_int)
if not hosted_file_obj:
raise HTTPException(status_code=404, detail="File metadata not found.")
# Path check
hosted_files_path = settings.FILES_PATH['hosted_files_root']
file_path = os.path.join(hosted_files_path, hosted_file_obj.subdirectory_path or '', f"{hosted_file_obj.hash_sha256}.file")
file_exists_on_disk = os.path.exists(file_path)
# Link check
links = get_hosted_file_link_rec_list(hosted_file_id=file_id_int)
link_found = any(l.get('link_to_type') == link_to_type and l.get('link_to_id') == link_id_int for l in links)
if fake_delete:
log.info(f"Fake Delete active. Verifying existence...")
return mk_resp(data={
"hosted_file_exists": True,
"file_on_disk": file_exists_on_disk,
"link_exists": link_found,
"fake_delete": True
}, status_message="Fake delete successful. No data was modified.")
# 3. Execution Phase
# A. Remove the Link
if link_id_int:
delete_hosted_file_link(
account_id = account.account_id,
hosted_file_id = file_id_int,
link_to_type = link_to_type,
link_to_id = link_id_int
)
log.info(f"Deleted link between file {file_id_int} and {link_to_type}:{link_id_int}")
# B. Orphan Check & Physical Cleanup
remaining_links = get_hosted_file_link_rec_list(hosted_file_id=file_id_int)
is_orphan = (len(remaining_links) == 0)
physical_removed = False
record_removed = False
if rm_orphan and is_orphan:
log.info(f"File {file_id_int} is an orphan. Cleaning up...")
# Method Handling
if method == 'delete':
# Hard delete: Record + Disk
if file_exists_on_disk:
pathlib.Path(file_path).unlink()
physical_removed = True
sql_delete(table_name='hosted_file', record_id=file_id_int)
record_removed = True
elif method == 'hide':
sql_update(table_name='hosted_file', data={'id': file_id_int, 'hide': True})
elif method == 'disable':
sql_update(table_name='hosted_file', data={'id': file_id_int, 'enable': False})
return mk_resp(data={
"link_removed": link_found if link_id_int else False,
"is_orphan": is_orphan,
"physical_removed": physical_removed,
"record_removed": record_removed,
"method": method
}, status_message="Deletion process complete.")