mirror of
https://github.com/tiennm99/DocsGPT.git
synced 2026-10-19 10:14:46 +00:00
Alex
574f96341e
The backend import package is now docsgpt, the name it will carry on PyPI; application was far too generic to install into anyone's site-packages. git mv plus a mechanical rewrite of every import, dotted string and path reference: 734 Python files, the compose files, Dockerfile, workflows, docs, setup scripts, devcontainer, k8s manifests, vscode config, pytest and coverage config, .gitignore. Behaviour is unchanged. Kept for one release: - A top-level application package whose meta-path finder resolves application.x.y to the already-imported docsgpt.x.y object, so old imports and entry points (celery -A application.app.celery, uvicorn application.asgi:asgi_app) keep working with a FutureWarning. - Celery registers every application.* task name as an alias of its docsgpt.* task on start-up, so messages queued by the previous release still run. The redbeat key prefix moves to redbeat:docsgpt:v2: so schedule entries the previous release wrote are left unread instead of firing twice. The backend image builds from the repository root (docker build -f docsgpt/Dockerfile .) so it can ship the alias package; a root .dockerignore allow-lists docsgpt/ and application/ and keeps caches, local data, .env files, the sample index files and the Dockerfile out. Compose and the image workflows point at the new context.
57 lines
2.3 KiB
Python
57 lines
2.3 KiB
Python
"""0018 tool_call_attempts attribution — record user/agent on the journal.
|
|
|
|
Adds ``tool_call_attempts.user_id`` / ``agent_id`` so tool analytics can
|
|
attribute attempts that never get a ``message_id`` — headless runs
|
|
(scheduled / webhook) execute tools before any conversation message
|
|
exists, and parse-failure rows never reach one at all.
|
|
|
|
DDL only — a whole-table UPDATE here would hold the ALTERs' ACCESS
|
|
EXCLUSIVE lock across the rewrite and stall live tool journaling. The
|
|
backfill lives in ``scripts/db/backfill_tool_attempts_attribution.py``;
|
|
until it runs, the analytics reader falls back to the parent message's
|
|
user for unstamped rows.
|
|
|
|
The index is built ``CONCURRENTLY`` in an autocommit block: the
|
|
``ADD COLUMN`` ALTERs commit first (a fast metadata-only change), so the
|
|
O(table-size) index build never runs while their ACCESS EXCLUSIVE lock
|
|
is held and never blocks live journaling.
|
|
|
|
Revision ID: 0018_tool_attempts_attribution
|
|
Revises: 0017_oidc_scim
|
|
"""
|
|
|
|
from typing import Sequence, Union
|
|
|
|
from alembic import op
|
|
|
|
|
|
revision: str = "0018_tool_attempts_attribution"
|
|
down_revision: Union[str, None] = "0017_oidc_scim"
|
|
branch_labels: Union[str, Sequence[str], None] = None
|
|
depends_on: Union[str, Sequence[str], None] = None
|
|
|
|
|
|
def upgrade() -> None:
|
|
op.execute("ALTER TABLE tool_call_attempts ADD COLUMN user_id TEXT;")
|
|
op.execute("ALTER TABLE tool_call_attempts ADD COLUMN agent_id UUID;")
|
|
# CONCURRENTLY can't run inside a transaction; the autocommit block
|
|
# commits the ALTERs above (releasing their lock) before building.
|
|
# DROP IF EXISTS first so a failed concurrent build (which leaves an
|
|
# INVALID index) is cleaned up on re-run.
|
|
with op.get_context().autocommit_block():
|
|
op.execute("DROP INDEX IF EXISTS tool_call_attempts_user_ts_idx;")
|
|
op.execute(
|
|
"CREATE INDEX CONCURRENTLY tool_call_attempts_user_ts_idx "
|
|
"ON tool_call_attempts (user_id, attempted_at DESC) "
|
|
"WHERE user_id IS NOT NULL;"
|
|
)
|
|
|
|
|
|
def downgrade() -> None:
|
|
with op.get_context().autocommit_block():
|
|
op.execute(
|
|
"DROP INDEX CONCURRENTLY IF EXISTS tool_call_attempts_user_ts_idx;"
|
|
)
|
|
op.execute("ALTER TABLE tool_call_attempts DROP COLUMN IF EXISTS agent_id;")
|
|
op.execute("ALTER TABLE tool_call_attempts DROP COLUMN IF EXISTS user_id;")
|