gitadmin/signalscout-backend
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

暂停情报采集任务

Browse Source
This commit is contained in:
Codex
2026-07-10 13:27:53 +08:00
parent 0a045a0c4b
commit bc4bbe0dde
8 changed files with 45 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.env.example
+2 -1
View File
@@ -3,7 +3,7 @@ SIGNALSCOUT_ENV=dev
SIGNALSCOUT_DATABASE_URL=mysql+pymysql://signalscout:signalscout@127.0.0.1:3306/signalscout?charset=utf8mb4
SIGNALSCOUT_LLM_BASE_URL=https://api.zayuapi.com/v1
SIGNALSCOUT_LLM_API_KEY=
SIGNALSCOUT_LLM_MODEL=grok-4.3
SIGNALSCOUT_LLM_MODEL=grok-4.5
SIGNALSCOUT_LLM_TIMEOUT_SECONDS=180
SIGNALSCOUT_LLM_MAX_TOKENS=8000
SIGNALSCOUT_NEWS_RECENT_DAYS=3
@@ -13,6 +13,7 @@ SIGNALSCOUT_GITHUB_RECENT_DAYS=30
SIGNALSCOUT_GITHUB_MIN_STARS=20
SIGNALSCOUT_GITHUB_MAX_PROJECTS=40
SIGNALSCOUT_AGENT_TIMEZONE=Asia/Shanghai
SIGNALSCOUT_AGENT_ENABLED=false
SIGNALSCOUT_AGENT_INTERVAL_MINUTES=30
SIGNALSCOUT_REPORT_DIR=reports
SIGNALSCOUT_LOG_FILE=logs/app.log
README.md
+4 -1
View File
@@ -28,7 +28,7 @@ cp .env.example .env
SIGNALSCOUT_DATABASE_URL=mysql+pymysql://user:password@host:3306/signalscout?charset=utf8mb4
SIGNALSCOUT_LLM_BASE_URL=https://api.zayuapi.com/v1
SIGNALSCOUT_LLM_API_KEY=你的中转站 key
SIGNALSCOUT_LLM_MODEL=grok-4.3
SIGNALSCOUT_LLM_MODEL=grok-4.5
SIGNALSCOUT_LLM_TIMEOUT_SECONDS=180
SIGNALSCOUT_LLM_MAX_TOKENS=8000
SIGNALSCOUT_NEWS_RECENT_DAYS=3
@@ -37,11 +37,14 @@ SIGNALSCOUT_NEWS_SEARCH_MAX_RECORDS=100
SIGNALSCOUT_GITHUB_RECENT_DAYS=30
SIGNALSCOUT_GITHUB_MIN_STARS=20
SIGNALSCOUT_GITHUB_MAX_PROJECTS=40
SIGNALSCOUT_AGENT_ENABLED=false
SIGNALSCOUT_AGENT_INTERVAL_MINUTES=30
```
单次运行的目标上限是新闻 60 条、GitHub 项目 40 条；实际入库数量会根据候选质量、链接去重和模型筛选结果浮动。
设置 `SIGNALSCOUT_AGENT_ENABLED=true` 后，服务才会启用定时和手动情报采集。
## 本机运行
按项目规则，本机 Python 使用：
app/agent/scheduler.py
+5 -1
View File
@@ -16,8 +16,12 @@ def run_scheduled_job(settings: Settings) -> None:
def build_scheduler(settings: Settings) -> BackgroundScheduler:
# APScheduler负责按固定间隔自动运行，避免服务重启时连续触发外部采集限流。
# 调度器始终可供手动任务复用，自动采集关闭时不注册任何周期任务。
scheduler = BackgroundScheduler(timezone=settings.agent_timezone)
if not settings.agent_enabled:
return scheduler
# APScheduler负责按固定间隔自动运行，避免服务重启时连续触发外部采集限流。
scheduler.add_job(
run_scheduled_job,
trigger="interval",
app/api/routes.py
+4 -1
View File
@@ -3,7 +3,7 @@ from __future__ import annotations
from datetime import datetime, timedelta
from typing import Literal, Optional
from fastapi import APIRouter, Depends, Request
from fastapi import APIRouter, Depends, HTTPException, Request
from fastapi.responses import HTMLResponse
from sqlalchemy import select
from sqlalchemy.orm import Session
@@ -50,6 +50,9 @@ def trigger_daily_run(
settings: Settings = Depends(get_settings),
db: Session = Depends(get_db),
) -> RunTriggerRead:
# 暂停期间拒绝所有入口的采集请求，保证不会产生外部模型调用。
if not settings.agent_enabled:
raise HTTPException(status_code=503, detail="情报采集当前已暂停")
# 手动搜索投递到调度器线程，避免HTTP请求承载长时间Grok搜索任务。
running = db.scalar(
select(AgentRun).where(AgentRun.status == "running").order_by(AgentRun.id.desc()).limit(1)
app/core/config.py
+3 -1
View File
@@ -14,7 +14,7 @@ class Settings(BaseSettings):
)
llm_base_url: str = "https://api.zayuapi.com/v1"
llm_api_key: str = ""
llm_model: str = "grok-4.3"
llm_model: str = "grok-4.5"
llm_timeout_seconds: int = 180
llm_max_tokens: int = 8000
news_recent_days: int = 3
@@ -24,6 +24,8 @@ class Settings(BaseSettings):
github_min_stars: int = 20
github_max_projects: int = 40
agent_timezone: str = "Asia/Shanghai"
# 采集开关统一控制定时任务和手动触发，暂停期默认不调用外部模型。
agent_enabled: bool = False
agent_interval_minutes: int = 30
report_dir: Path = Path("reports")
log_file: Path = Path("logs/app.log")
app/ui/home.py
-53
View File
@@ -162,30 +162,6 @@ HOME_PAGE_HTML = """
margin-top: 5px;
}
.actions {
display: flex;
justify-content: flex-end;
margin-top: 12px;
}
.action-button {
border: 1px solid oklch(25% 0.02 210);
border-radius: 8px;
padding: 9px 13px;
background: var(--ink);
color: white;
font: inherit;
font-size: 13px;
font-weight: 700;
cursor: pointer;
min-width: 96px;
}
.action-button:disabled {
cursor: wait;
opacity: 0.66;
}
.layout {
display: grid;
grid-template-columns: minmax(0, 1.2fr) minmax(340px, 0.8fr);
@@ -434,9 +410,6 @@ HOME_PAGE_HTML = """
<div class="pulse">
<strong id="todayCountText">读取中</strong>
<span id="todayCountDetail">正在整理新闻与项目。</span>
<div class="actions">
<button class="action-button" id="manualRunButton" type="button">手动搜索</button>
</div>
</div>
</section>
@@ -511,7 +484,6 @@ HOME_PAGE_HTML = """
const runCountFull = document.querySelector("#runCountFull");
const todayCountText = document.querySelector("#todayCountText");
const todayCountDetail = document.querySelector("#todayCountDetail");
const manualRunButton = document.querySelector("#manualRunButton");
const stateText = {
completed: "完成",
@@ -695,34 +667,9 @@ HOME_PAGE_HTML = """
renderRuns(data.runs || []);
}
async function triggerManualRun() {
manualRunButton.disabled = true;
manualRunButton.textContent = "搜索中";
try {
const response = await fetch("/agent/runs/daily", {
method: "POST",
headers: { Accept: "application/json" }
});
if (!response.ok) throw new Error("手动搜索失败");
await loadDashboard();
} finally {
window.setTimeout(() => {
manualRunButton.disabled = false;
manualRunButton.textContent = "手动搜索";
loadDashboard().catch(() => {});
}, 3000);
}
}
navButtons.forEach((button) => {
button.addEventListener("click", () => activateView(button.dataset.viewTarget));
});
manualRunButton.addEventListener("click", () => {
triggerManualRun().catch(() => {
manualRunButton.disabled = false;
manualRunButton.textContent = "手动搜索";
});
});
const initialView = window.location.hash.replace("#", "");
if (["signals", "report", "runs"].includes(initialView)) {
tests/test_api_health.py
+18 -3
View File
@@ -36,15 +36,16 @@ def test_home_page_returns_dashboard_shell(monkeypatch) -> None:
assert response.status_code == 200
assert "SignalScout" in response.text
assert "今日 AI 信号" in response.text
assert "手动搜索" in response.text
assert "手动搜索" not in response.text
assert "renderMarkdownReport" in response.text
assert "自动记录" not in response.text
assert "采集今日情报" not in response.text
def test_manual_agent_run_endpoint_starts_background_job(monkeypatch) -> None:
# 手动搜索入口复用调度任务，方便验证每轮候选和入库数量。
def test_manual_agent_run_endpoint_starts_background_job_when_enabled(monkeypatch) -> None:
# 启用采集后手动入口复用调度任务，方便验证每轮候选和入库数量。
monkeypatch.setenv("SIGNALSCOUT_ENV", "test")
monkeypatch.setenv("SIGNALSCOUT_AGENT_ENABLED", "true")
get_settings.cache_clear()
started = []
engine = create_engine(
@@ -82,6 +83,7 @@ def test_manual_agent_run_endpoint_starts_background_job(monkeypatch) -> None:
def test_manual_agent_run_endpoint_uses_scheduler_when_available(monkeypatch) -> None:
# 生产环境手动搜索投递到常驻调度器，避免HTTP请求承载长时间联网搜索。
monkeypatch.setenv("SIGNALSCOUT_ENV", "test")
monkeypatch.setenv("SIGNALSCOUT_AGENT_ENABLED", "true")
get_settings.cache_clear()
engine = create_engine(
"sqlite:///:memory:",
@@ -121,6 +123,19 @@ def test_manual_agent_run_endpoint_uses_scheduler_when_available(monkeypatch) ->
assert scheduled[0]["run_date"].tzinfo is not None
def test_manual_agent_run_endpoint_rejects_requests_when_disabled(monkeypatch) -> None:
# 采集暂停时接口拒绝请求，避免其他调用方绕过页面入口消耗模型额度。
monkeypatch.setenv("SIGNALSCOUT_ENV", "test")
monkeypatch.setenv("SIGNALSCOUT_AGENT_ENABLED", "false")
get_settings.cache_clear()
with TestClient(create_app()) as client:
response = client.post("/agent/runs/daily")
get_settings.cache_clear()
assert response.status_code == 503
assert response.json()["detail"] == "情报采集当前已暂停"
def test_dashboard_returns_news_and_github_signals_separately(monkeypatch) -> None:
# 仪表盘接口直接返回分组后的业务数据，前端不再用标题或来源名自行猜测分类。
monkeypatch.setenv("SIGNALSCOUT_ENV", "test")
tests/test_scheduler.py
+9 -2
View File
@@ -3,10 +3,17 @@ from app.core.config import Settings
def test_scheduler_runs_on_interval() -> None:
# Agent默认按固定间隔自动运行，避免产品依赖用户手动触发。
scheduler = build_scheduler(Settings(agent_interval_minutes=30))
# 启用采集后按固定间隔自动运行，避免产品依赖用户手动触发。
scheduler = build_scheduler(Settings(agent_enabled=True, agent_interval_minutes=30))
job = scheduler.get_job("continuous-intelligence")
assert job is not None
assert job.trigger.interval.total_seconds() == 1800
assert job.max_instances == 1
def test_scheduler_skips_periodic_job_when_agent_is_disabled() -> None:
# 采集暂停时不注册任务，确保调度器没有可执行的外部模型调用。
scheduler = build_scheduler(Settings(agent_enabled=False))
assert scheduler.get_job("continuous-intelligence") is None