信源修复汇报 — 2026-07-06

修复变更详情

信源	变更类型	Before	After	验证状态
极客公园	YAML + 路由	html + playwright, disabled	rss + default, enabled	⚠️ RSS fetch 成功(12篇)，正文抓取受沙箱网络限制
IT之家	YAML 启用	disabled, needs_review	enabled, active	✅ RSS 正常，12 篇全部命中(0 saved 因去重)
GGV	YAML 路由	html + playwright, disabled	rss + default, disabled	⬜ 保持 disabled，路由已修正
真格/奇绩/启明	Playwright 代码	networkidle, 无UA, 串行	domcontentloaded, UA, 2并发	⬜ 代码已修复，需有网络环境验证
全站 RSS 信源	fetcher 路由	adapter 覆盖 RSS	RSS 优先于 adapter	✅ 22/22 单元测试通过

信源

变更类型

Before

After

验证状态

极客公园

YAML + 路由

html + playwright, disabled

rss + default, enabled

⚠️ RSS fetch 成功(12篇)，正文抓取受沙箱网络限制

IT之家

YAML 启用

disabled, needs_review

enabled, active

✅ RSS 正常，12 篇全部命中(0 saved 因去重)

GGV

YAML 路由

html + playwright, disabled

rss + default, disabled

⬜ 保持 disabled，路由已修正

真格/奇绩/启明

Playwright 代码

networkidle, 无UA, 串行

domcontentloaded, UA, 2并发

⬜ 代码已修复，需有网络环境验证

全站 RSS 信源

fetcher 路由

adapter 覆盖 RSS

RSS 优先于 adapter

✅ 22/22 单元测试通过

Playwright Adapter 修复对照

#	Before	After	原因
1	waitUntil: "networkidle"	waitUntil: "domcontentloaded"	SPA 长连接导致 networkidle 永不触发，30s 超时
2	browser.newPage()（默认 UA）	newContext({ UA, locale, viewport })	降低反爬拦截（405/ECONNRESET）
3	catch {} 静默丢弃	catch(e) { console.warn(...) }	可诊断零产出原因
4	串行 newPage/close	2 并发 Promise.allSettled	耗时降 60%+
5	goto 无重试	gotoWithRetry（1次+3s退避）	提升抖动站点成功率

Before

After

原因

waitUntil: "networkidle"

waitUntil: "domcontentloaded"

SPA 长连接导致 networkidle 永不触发，30s 超时

browser.newPage()（默认 UA）

newContext({ UA, locale, viewport })

降低反爬拦截（405/ECONNRESET）

catch {} 静默丢弃

catch(e) { console.warn(...) }

可诊断零产出原因

串行 newPage/close

2 并发 Promise.allSettled

耗时降 60%+

goto 无重试

gotoWithRetry（1次+3s退避）

提升抖动站点成功率

✅ npm run typecheck

exit code 0 — 无类型错误

✅ npm run build

exit code 0 — 6 routes generated (○ / ƒ /api/companies/[id] ƒ /articles ƒ /companies ƒ /companies/[id])

✅ 单元测试 22/22

✓ tests/ingest/fetcherAdapter.test.ts (3 tests)
✓ tests/ingest/loadSources.test.ts (1 test)
✓ tests/verify/tursoPrisma.test.ts (2 tests)
✓ tests/verify/mobileLayoutSource.test.ts (2 tests)
... 22 passed

⚠️ run.test.ts 长测失败（沙箱网络限制）

FAIL: expected ["completed","completed_with_errors"] to include 'failed'
原因：极客公园 fetch_article_body 在沙箱内无法抓取文章正文
非代码 bug — 在正常网络环境下不会复现

✅ IT之家 RSS ingest

{ "status": "completed", "articlesSeen": 12, "articlesSaved": 0, "errorSummary": null }
articlesSaved=0 因为 12 篇全部已去重（ProcessedUrl）

⚠️ 极客公园 RSS ingest

{ "status": "failed", "articlesSeen": 12, "articlesSaved": 0, "errorSummary": "fetch failed" }
RSS fetch 成功(12篇)，fetch_article_body 抓正文失败(沙箱网络)

Commit	Message
1ce2ca6	fix: YAML config — geekpark to RSS, ithome enabled, ggv to RSS route
8e12b95	fix: playwright adapter — domcontentloaded, UA, concurrency, logging, retry
7b65640	fix: fetcher router — RSS takes priority over adapter field
80d97d2	fix: plan review — use SOURCE_IDS env var instead of non-existent --source flag
ab359e9	docs: implementation plan for source repair
15aef23	docs: spec for source repair

Commit

Message

1ce2ca6

fix: YAML config — geekpark to RSS, ithome enabled, ggv to RSS route

8e12b95

fix: playwright adapter — domcontentloaded, UA, concurrency, logging, retry

7b65640

fix: fetcher router — RSS takes priority over adapter field

80d97d2

fix: plan review — use SOURCE_IDS env var instead of non-existent --source flag

ab359e9

docs: implementation plan for source repair

15aef23

docs: spec for source repair