52 lines
1.6 KiB
Python
52 lines
1.6 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
KM Vectorize CronJob 入口腳本 — ADR-073 Phase 4-3
|
|
|
|
每日由 K8s CronJob 呼叫，對新增 KM 條目執行向量化，
|
|
確保 RAG 查詢可存取最新知識（飛輪「學習固化」節點）。
|
|
|
|
2026-04-12 ogt (ADR-073 Phase 4-3, P2-1 重構)
|
|
"""
|
|
|
|
import asyncio
|
|
import os
|
|
import sys
|
|
|
|
import httpx
|
|
|
|
|
|
def _project_headers() -> dict[str, str]:
|
|
"""Return internal API project context headers required by fail-closed RLS."""
|
|
project_id = os.environ.get("KM_PROJECT_ID", "awoooi").strip()
|
|
if not project_id:
|
|
project_id = "awoooi"
|
|
return {"X-Project-ID": project_id}
|
|
|
|
|
|
async def main() -> int:
|
|
api_base = os.environ.get(
|
|
"INTERNAL_API_URL",
|
|
"http://awoooi-api-svc.awoooi-prod.svc.cluster.local:8000",
|
|
)
|
|
url = f"{api_base}/api/v1/knowledge/embed-all"
|
|
|
|
async with httpx.AsyncClient(timeout=1800) as client:
|
|
try:
|
|
resp = await client.post(url, headers=_project_headers())
|
|
print(f"embed-all: {resp.status_code} {resp.text[:200]}")
|
|
if resp.status_code >= 400:
|
|
print(f"ERROR: embed-all returned {resp.status_code}", file=sys.stderr)
|
|
return 1
|
|
result = resp.json()
|
|
if int(result.get("failed", 0)) > 0:
|
|
print(f"ERROR: embed-all failed rows: {result}", file=sys.stderr)
|
|
return 1
|
|
return 0
|
|
except httpx.RequestError as exc:
|
|
print(f"ERROR: request failed — {exc}", file=sys.stderr)
|
|
return 1
|
|
|
|
|
|
if __name__ == "__main__":
|
|
sys.exit(asyncio.run(main()))
|