Files
next-tjwater-server/app/services/calculation_cleanup.py
T

75 lines
2.8 KiB
Python

"""Record only job-owned native databases so abrupt cancellation is recoverable."""
import asyncio
import json
import os
from pathlib import Path
import re
from uuid import UUID
def _names(directory: Path) -> list[str]:
path = directory / "databases.json"
if not path.exists():
return []
if path.is_symlink() or path.stat().st_size > 65536:
raise ValueError("Invalid calculation database ledger")
names = json.loads(path.read_text())
if not isinstance(names, list) or len(names) > 100 or any(
not isinstance(name, str) or not re.fullmatch(r"tjw_tmp_[a-z0-9_]+_[a-f0-9]{32}", name)
or len(name) > 63 for name in names
):
raise ValueError("Invalid calculation database ledger")
return names
def _write(directory: Path, names: list[str]) -> None:
path = directory / "databases.json"
staging = directory / "databases.json.new"
with staging.open("w") as output:
json.dump(names, output)
output.flush()
os.fsync(output.fileno())
staging.replace(path)
def record_calculation_database(name: str, *, removed: bool = False) -> None:
"""Called before clone creation and after successful normal cleanup."""
root = os.environ.get("TJWATER_CALCULATION_JOB_DIR")
if not root:
return
directory = Path(root)
names = _names(directory)
if removed:
names = [value for value in names if value != name]
elif name not in names:
names.append(name)
_write(directory, names)
async def cleanup_calculation_databases(project_id: str, job_directory: str) -> None:
directory = Path(job_directory)
names = _names(directory)
if not names:
return
from app.infra.db.metadb.database import SessionLocal
from app.infra.db.metadb.repositories.metadata_repository import MetadataRepository
from app.infra.db.project_routing import ActiveProjectRouting, activate_project_routing
from app.native.wndb.core.projects import have_project, delete_project
async with SessionLocal() as session:
metadata = MetadataRepository(session)
project = await metadata.get_project_by_id(UUID(str(project_id)))
route = await metadata.get_project_db_routing(UUID(str(project_id)), "biz_data")
if not project or not route or route.db_type != "postgresql":
raise ValueError("Calculation cleanup routing unavailable")
routing = ActiveProjectRouting(project.code, route.dsn, project_id=project.id)
await session.commit()
def cleanup():
with activate_project_routing(routing):
for name in names[:]:
if have_project(name):
delete_project(name)
names.remove(name)
_write(directory, names)
await asyncio.to_thread(cleanup)