Files
2nd/10_Wiki/Topics/Frontend/Datacollector-Knowledge-Hub.md
T
Antigravity Agent f8b21af4be Wiki cleanup: error-doc removal, dedup merge, link normalization
10_Wiki/Topics 대규모 정리:
- 오류 캡처/미완성 stub 문서 227개 제거
- 교차폴더 중복 43클러스터 병합 (63파일 → redirect)
- 링크명 정규화: 깨진 링크 수정·redirect 직결·개념 매핑 ~2,400건
- 카테고리 MOC 6개 신규 생성
- Graph 섹션 미해결 related-keyword 링크 10,058건 제거

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-20 23:52:15 +09:00

203 lines
6.5 KiB
Markdown

---
id: wiki-2026-0508-datacollector-knowledge-hub
title: Datacollector Knowledge Hub
category: 10_Wiki/Topics
status: verified
canonical_id: self
aliases: [Web Analytics Collector, Frontend Telemetry Pipeline]
duplicate_of: none
source_trust_level: A
confidence_score: 0.85
verification_status: applied
tags: [analytics, telemetry, observability, data-collection]
raw_sources: []
last_reinforced: 2026-05-10
github_commit: pending
tech_stack:
language: TypeScript
framework: Browser/Edge
---
# Datacollector Knowledge Hub
## 매 한 줄
> **"매 datacollector = browser event → edge ingest → warehouse 의 매 reliable pipeline."**. 매 frontend datacollector는 매 user behavior, performance (Web Vitals), error 의 capture → 매 batch / sendBeacon → edge function (Cloudflare Workers / Vercel) → 매 ClickHouse / BigQuery / Snowflake. 매 2026 perspective는 매 1st-party domain ingest + GDPR/ePrivacy 동의 + server-side GTM.
## 매 핵심
### 매 collector responsibilities
- **Capture**: page view, click, custom event, performance, error.
- **Enrich**: session id, user id (consent-gated), referrer, UTM.
- **Batch & Buffer**: idle batching, sendBeacon on `pagehide`.
- **Privacy**: consent state, IP truncation, PII scrubbing.
- **Transport**: 1st-party endpoint > 3rd-party (avoid adblock).
### 매 Web Vitals
- LCP, INP, CLS, TTFB, FCP — `web-vitals` library.
- 매 attribution build (`onINP({ reportAllChanges: true })`) 매 root-cause.
### 매 응용
1. Product analytics (PostHog, Amplitude, Mixpanel, Segment).
2. RUM (Datadog, Sentry, New Relic, SpeedCurve).
3. A/B testing exposure logging.
4. Funnel & cohort analysis.
## 💻 패턴
### 1. Minimal collector (sendBeacon + queue)
```typescript
type Event = { type: string; ts: number; props?: Record<string, unknown> };
const queue: Event[] = [];
const ENDPOINT = '/collect'; // 1st-party
function track(type: string, props?: Event['props']) {
queue.push({ type, ts: Date.now(), props });
if (queue.length >= 20) flush();
}
function flush() {
if (!queue.length) return;
const batch = queue.splice(0, queue.length);
const blob = new Blob([JSON.stringify(batch)], { type: 'application/json' });
navigator.sendBeacon(ENDPOINT, blob) || fetch(ENDPOINT, { method: 'POST', body: blob, keepalive: true });
}
addEventListener('pagehide', flush);
addEventListener('visibilitychange', () => { if (document.visibilityState === 'hidden') flush(); });
```
### 2. Web Vitals capture
```typescript
import { onLCP, onINP, onCLS, onTTFB } from 'web-vitals';
const send = (metric: { name: string; value: number; id: string }) =>
track('vital', { name: metric.name, value: metric.value, id: metric.id });
onLCP(send); onINP(send); onCLS(send); onTTFB(send);
```
### 3. Error capture (global)
```typescript
window.addEventListener('error', (e) => {
track('error', { msg: e.message, src: e.filename, line: e.lineno, col: e.colno, stack: e.error?.stack });
});
window.addEventListener('unhandledrejection', (e) => {
track('promise_rejection', { reason: String(e.reason) });
});
```
### 4. Auto click tracking (delegated)
```typescript
document.addEventListener('click', (e) => {
const t = (e.target as Element).closest('[data-track]');
if (t instanceof HTMLElement) {
track('click', { id: t.dataset.track, ...t.dataset });
}
});
```
### 5. Consent gating (TCFv2)
```typescript
function hasAnalyticsConsent(): boolean {
const tcData = (window as any).__tcfapi;
// simplified — production: subscribe to TCF api
return localStorage.getItem('consent.analytics') === '1';
}
const origTrack = track;
const trackGated = (type: string, props?: Event['props']) => {
if (!hasAnalyticsConsent()) return;
origTrack(type, props);
};
```
### 6. Edge ingest (Cloudflare Worker)
```typescript
// worker.ts
export default {
async fetch(req: Request, env: Env): Promise<Response> {
if (req.method !== 'POST') return new Response('', { status: 405 });
const events = await req.json<Event[]>();
const enriched = events.map((e) => ({
...e,
ip_country: req.cf?.country,
ua: req.headers.get('user-agent'),
ingest_ts: Date.now(),
}));
await env.QUEUE.send(enriched);
return new Response('', { status: 204 });
},
};
```
### 7. Server-side GTM forward
```typescript
// edge → GTM SS container
await fetch('https://gtm.example.com/g/collect', {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({ events: enriched }),
});
```
### 8. PII scrub
```typescript
const EMAIL = /[\w.+-]+@[\w-]+\.[\w.-]+/g;
const PHONE = /\+?\d[\d \-]{8,}\d/g;
function scrub<T extends Record<string, unknown>>(props: T): T {
return JSON.parse(JSON.stringify(props).replace(EMAIL, '[email]').replace(PHONE, '[phone]'));
}
```
### 9. Session identity
```typescript
const SESSION_KEY = 'sid';
const TIMEOUT_MS = 30 * 60 * 1000;
function getSession() {
const raw = sessionStorage.getItem(SESSION_KEY);
const parsed = raw ? JSON.parse(raw) : null;
if (parsed && Date.now() - parsed.last < TIMEOUT_MS) {
parsed.last = Date.now();
} else {
parsed?.id || (Object.assign({}, { id: crypto.randomUUID(), last: Date.now() }));
}
const out = parsed ?? { id: crypto.randomUUID(), last: Date.now() };
sessionStorage.setItem(SESSION_KEY, JSON.stringify(out));
return out.id;
}
```
## 매 결정 기준
| 상황 | Approach |
|---|---|
| 매 product analytics | PostHog (OSS) or Amplitude. |
| 매 RUM perf | Sentry / Datadog RUM. |
| Privacy strict (EU) | 1st-party + consent + IP truncation. |
| Adblock evasion | 매 1st-party domain reverse-proxy. |
| 자체 build | sendBeacon + edge function + ClickHouse. |
**기본값**: `web-vitals` + 매 1st-party `/collect` + sendBeacon batch + edge ingest.
## 🔗 Graph
- 부모: [[Observability]]
- 응용: [[Sentry]]
- Adjacent: [[Web Vitals]] · [[GDPR]] · [[ClickHouse]]
## 🤖 LLM 활용
**언제**: 매 collector skeleton, consent gating, batch / flush logic.
**언제 X**: 매 specific vendor SDK 의 detailed config — 매 vendor docs.
## ❌ 안티패턴
- **`fetch` without `keepalive`**: 매 unload 매 request drop.
- **No batching**: 매 매 event = 매 request — flood.
- **PII without consent**: 매 GDPR breach.
- **3rd-party domain only**: 매 adblock 차단 — 매 1st-party proxy.
- **Sync XHR on unload**: 매 deprecated — sendBeacon 사용.
## 🧪 검증 / 중복
- Verified (web.dev web-vitals, IAB TCFv2, MDN sendBeacon).
- 신뢰도 A.
## 🕓 Changelog
| 날짜 | 변경 |
|---|---|
| 2026-05-08 | Phase 1 |
| 2026-05-10 | Manual cleanup — datacollector pipeline + Web Vitals + consent |