639562593e
之前 service.translate 写 cache 无条件,导致:
- 第一次翻译失败时,'[翻译失败: ...]' 占位符被写进 cache
- 30 天内相同文本的请求(新文章 title 与老文章 title 相同时)全部返回占位符
- 触发 200+ 文章 title_zh 字段被永久污染
修法:仅在 engine ∈ {tencent, nllb, cache} 且文本不含错误标记时,才写 cache。
53 lines
2.7 KiB
Python
53 lines
2.7 KiB
Python
"""pull + 重建 worker + 扫描 DB 把翻译失败的改回 pending + 看新 worker 跑起来。"""
|
|
import os, paramiko, json, time
|
|
PW = os.environ["REMOTE_PASS"]
|
|
c = paramiko.SSHClient()
|
|
c.set_missing_host_key_policy(paramiko.AutoAddPolicy())
|
|
c.connect("207.57.129.228", port=19717, username="root", password=PW, timeout=15, allow_agent=False, look_for_keys=False)
|
|
def run(cmd, t=60):
|
|
si, so, se = c.exec_command(cmd, timeout=t)
|
|
out = so.read().decode("utf-8", "replace")
|
|
err = se.read().decode("utf-8", "replace")
|
|
rc = so.channel.recv_exit_status()
|
|
if out: print(out, end="")
|
|
if err and "Warning" not in err and "warn" not in err: print(err, end="", file=__import__("sys").stderr)
|
|
return out
|
|
|
|
# 1) pull
|
|
print("--- pull ---")
|
|
run("cd /srv/news && sudo -u news git pull --rebase 2>&1 | tail -3")
|
|
|
|
# 2) 重建 worker
|
|
print("--- 重建 worker ---")
|
|
run("cd /srv/news && docker compose up -d --force-recreate --no-deps --build worker 2>&1 | tail -5", t=120)
|
|
time.sleep(5)
|
|
|
|
# 3) 扫描 DB:title_zh/body_zh_text 含 '翻译失败' 改回 pending
|
|
print("--- DB 扫描 ---")
|
|
out = run("docker exec news-aggregator-postgres-1 psql -U news -d news -tA -c \"SELECT count(*) FROM articles WHERE title_zh LIKE '%[翻译失败:%' OR body_zh_text LIKE '%[翻译失败:%' OR body_zh_html LIKE '%[翻译失败:%';\"")
|
|
print(f" 含 '翻译失败' 占位符的文章数: {out.strip()}")
|
|
n = run("docker exec news-aggregator-postgres-1 psql -U news -d news -tA -c \"UPDATE articles SET translation_status='pending', title_zh=NULL, body_zh_text=NULL, body_zh_html=NULL, translated_at=NULL, translation_engine=NULL, translation_chars=0 WHERE title_zh LIKE '%[翻译失败:%' OR body_zh_text LIKE '%[翻译失败:%' OR body_zh_html LIKE '%[翻译失败:%';\"")
|
|
print(f" UPDATE 状态: {n.strip()}")
|
|
|
|
# 4) 看 worker 是否在跑 translation_loop
|
|
print("\n--- worker 日志(看 translation_loop 启动 + 节奏)---")
|
|
time.sleep(15)
|
|
out = run("docker logs --tail=50 news-aggregator-worker-1 2>&1 | tail -30", t=15)
|
|
print(out)
|
|
|
|
# 5) 等 30 秒再看(应该已经翻译 30 篇左右)
|
|
print("\n--- 等 30 秒看翻译进度 ---")
|
|
time.sleep(30)
|
|
out = run("docker exec news-aggregator-postgres-1 psql -U news -d news -c \"SELECT translation_status, count(*) FROM articles GROUP BY 1 ORDER BY 1;\"")
|
|
print(out)
|
|
|
|
# 6) redis usage
|
|
rpw = run("grep ^REDIS_PASSWORD /srv/news/.env | cut -d= -f2").strip()
|
|
out = run("docker exec news-aggregator-redis-1 redis-cli -a '" + rpw + "' GET translation:month:202606 2>/dev/null")
|
|
print(f"\n--- redis usage: {out.strip()}")
|
|
|
|
# 7) 验证 fetch_one_source 不再自动翻译(看 worker 日志确认)
|
|
print("\n--- worker 进程信息 ---")
|
|
run("docker ps --filter 'name=news-aggregator-worker' --format 'table {{.Names}}\\t{{.Status}}'")
|
|
c.close()
|