fix(drainer): 同帳號 worker-to-worker 改走 Service Bindings(修 CF 1042)

Production bug(本地 miniflare 未抓到、真端點才爆):部署後 drainer 對同帳號
arcrun-code/arcrun-kbdb/kbdb-graph-plugin 做 fetch("https://*.leo21c.workers.dev")
→ CF 1042(同帳號 worker 互打限制),被 error handler 包成 "not valid JSON"。

修法=比照 cypher-executor(wrangler.toml [[services]] 綁零件):
- wrangler.toml 加 [[services]] SVC_CODE→arcrun-code / SVC_KBDB→arcrun-kbdb /
  SVC_GRAPH→kbdb-graph-plugin;移除 CODE_WORKER_URL/KBDB_URL/GRAPH_URL vars。
- index.ts:所有 code/kbdb/graph 呼叫改 env.SVC_X.fetch()(svc() 輔助);Gitea
  (git.uncle6.me,非 CF)維持 plain fetch。

驗收(強制對部署後真端點、禁 miniflare):POST /drain?batch=2 回真 JSON、notes 3 卡
冪等全 skip、無 1042;游標前進+回捲;加測試卡→真端點 /webhook 只寫該 delta(entry
create 進 live KBDB)→三模式 curl 命中(含 Arcrun#11 修後語意 owner+type 過濾生效)→
測試卡清除。deploy version 95562b3e。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HJiLCRUU2o3aSpPEzVCt2o
This commit is contained in:
kbdb-cc
2026-07-06 10:51:42 +00:00
parent 88b7fbf409
commit c6bd0d7899
3 changed files with 38 additions and 22 deletions
@@ -57,10 +57,13 @@ printf '%s' "$GITEA_TOKEN" | npx wrangler secret put GITEA_TOKEN
- Secret:若設,同步 `GITEA_WEBHOOK_SECRET` secret(本 Worker 驗 `X-Gitea-Signature` HMAC-SHA256
- Gitea 的 ping event → Worker 回 `{ok:true,pong:true}`(測連通)
## configwrangler.toml [vars]
## config
`REPO=Leo/notes REF=main OWNER=leo CARDS_ROOT=system-dev/wiki/cards BATCH_SIZE=2`
`CODE_WORKER_URL / KBDB_URL / GRAPH_URL / GRAPH_KEY`(皆 leo21c live 服務)
- **vars**`REPO=Leo/notes REF=main OWNER=leo CARDS_ROOT=system-dev/wiki/cards GRAPH_KEY=leo BATCH_SIZE=2`
- **Service Bindings`[[services]]`**`SVC_CODE→arcrun-code``SVC_KBDB→arcrun-kbdb``SVC_GRAPH→kbdb-graph-plugin`
⚠️ 同帳號 worker-to-worker **必須走 Service Binding**`env.SVC_X.fetch()`)——直接 `fetch("https://arcrun-*.leo21c.workers.dev")`
會撞 **CF 1042**(同帳號 worker 互打限制)。Gitea`git.uncle6.me`,非 CF)才用 plain `fetch`
註:`wrangler dev` 本地 miniflare **不強制** 1042,故本地測會假綠 → 一律對部署後真端點驗收。
## 規模備註(5,083 檔)
@@ -28,15 +28,21 @@ export interface Env {
REF: string; // main
OWNER: string; // leoKBDB 租戶 owner_id
CARDS_ROOT: string; // system-dev/wiki/cards
CODE_WORKER_URL: string; // https://arcrun-code.leo21c.workers.dev/
KBDB_URL: string; // https://arcrun-kbdb.leo21c.workers.dev
GRAPH_URL: string; // https://kbdb-graph-plugin.leo21c.workers.dev
// 同帳號 worker-to-worker 必須走 Service Binding(直接 fetch(*.leo21c.workers.dev) 會撞 CF 1042)。
SVC_CODE: Fetcher; // → arcrun-code
SVC_KBDB: Fetcher; // → arcrun-kbdb
SVC_GRAPH: Fetcher; // → kbdb-graph-plugin
GRAPH_KEY: string; // leo
BATCH_SIZE: string; // "2"
GITEA_TOKEN: string; // secret:讀 repo
GITEA_WEBHOOK_SECRET?: string; // secret(選填):驗 webhook 簽章
}
// Service Binding fetch 輔助:host 是 placeholderbinding 直接路由到目標 worker),path/query/method/body 才重要。
function svc(binding: Fetcher, path: string, init?: RequestInit): Promise<Response> {
return binding.fetch(new Request(`https://svc${path}`, init));
}
const CURSOR_TYPE = 'ingest_cursor';
const cursorPageName = (env: Env) => `cursor:km_wiki_ingest_drain:${env.REPO}`;
@@ -73,7 +79,7 @@ async function giteaRaw(env: Env, relPath: string): Promise<string> {
// ── card-processing 核心 ──────────────────────────────────────────────────────
async function parseCard(env: Env, md: string, relPath: string): Promise<any> {
const r = await fetch(env.CODE_WORKER_URL, {
const r = await svc(env.SVC_CODE, '/', {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({
@@ -99,20 +105,20 @@ async function upsertEntry(env: Env, entry: any): Promise<{ action: string; id?:
const pageName = entry.page_name;
const metadata_json = JSON.stringify(entry.metadata);
const newHash = entry.metadata.content_hash;
const lookup = await fetch(
`${env.KBDB_URL}/entries?page_name=${encodeURIComponent(pageName)}&owner_id=${encodeURIComponent(env.OWNER)}`,
const lookup = await svc(env.SVC_KBDB,
`/entries?page_name=${encodeURIComponent(pageName)}&owner_id=${encodeURIComponent(env.OWNER)}`,
);
const existing = ((await lookup.json()) as { entries?: any[] }).entries?.[0];
if (existing) {
if (metaHash(existing) === newHash) return { action: 'skip', id: existing.id };
const r = await fetch(`${env.KBDB_URL}/entries/${existing.id}`, {
const r = await svc(env.SVC_KBDB, `/entries/${existing.id}`, {
method: 'PATCH',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({ content: entry.content, tags_json: JSON.stringify(entry.tags || []), metadata_json }),
});
return { action: 'patch', id: existing.id, ...(r.ok ? {} : { action: 'patch-failed' }) };
}
const r = await fetch(`${env.KBDB_URL}/entries`, {
const r = await svc(env.SVC_KBDB, `/entries`, {
method: 'POST',
headers: { 'content-type': 'application/json' },
body: JSON.stringify({
@@ -134,7 +140,7 @@ async function postEnvelopes(env: Env, envelopes: any[]): Promise<any[]> {
for (const envp of envelopes) {
// envelopes 已由 code 節點剝除 _-鍵;此處再保險剝一次。
const clean = Object.fromEntries(Object.entries(envp).filter(([k]) => !k.startsWith('_')));
const r = await fetch(`${env.GRAPH_URL}/triplets/ingest?owner_id=${encodeURIComponent(env.OWNER)}`, {
const r = await svc(env.SVC_GRAPH, `/triplets/ingest?owner_id=${encodeURIComponent(env.OWNER)}`, {
method: 'POST',
headers: { 'content-type': 'application/json', 'X-Arcrun-API-Key': env.GRAPH_KEY },
body: JSON.stringify(clean),
@@ -168,8 +174,8 @@ async function processCard(env: Env, relPath: string): Promise<any> {
interface CursorState { last_path: string; cycle: number; processed_total: number; updated_at: number; }
async function getCursor(env: Env): Promise<{ id?: string; state: CursorState }> {
const r = await fetch(
`${env.KBDB_URL}/entries?page_name=${encodeURIComponent(cursorPageName(env))}&owner_id=${encodeURIComponent(env.OWNER)}`,
const r = await svc(env.SVC_KBDB,
`/entries?page_name=${encodeURIComponent(cursorPageName(env))}&owner_id=${encodeURIComponent(env.OWNER)}`,
);
const e = ((await r.json()) as { entries?: any[] }).entries?.[0];
if (e) {
@@ -182,11 +188,11 @@ async function getCursor(env: Env): Promise<{ id?: string; state: CursorState }>
async function setCursor(env: Env, cur: { id?: string; state: CursorState }): Promise<void> {
const content = JSON.stringify(cur.state);
if (cur.id) {
await fetch(`${env.KBDB_URL}/entries/${cur.id}`, {
await svc(env.SVC_KBDB, `/entries/${cur.id}`, {
method: 'PATCH', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ content }),
});
} else {
await fetch(`${env.KBDB_URL}/entries`, {
await svc(env.SVC_KBDB, `/entries`, {
method: 'POST', headers: { 'content-type': 'application/json' },
body: JSON.stringify({ entry_type: CURSOR_TYPE, page_name: cursorPageName(env), owner_id: env.OWNER, content,
metadata_json: JSON.stringify({ kind: 'ingest_cursor', embed: false }) }),
@@ -3,20 +3,27 @@ main = "src/index.ts"
compatibility_date = "2025-02-19"
workers_dev = true
# ⚠️ 維持「非-auto」:預設不註冊 cron trigger(不自走排程、不無人值守亂寫)。
# drain 以 POST /drain 手動觸發(與 scheduled handler 同核心)。
# 要開自走 Phase 0 backfillleo 過閘後,取消下面兩行註解再 deployscheduled handler 已在 code)。
# ⚠️ 維持「非-auto」:預設不註冊 cron trigger(不自走排程)。drain 走 POST /drain(與 scheduled 同核心)。
# 要開自走 Phase 0 backfillleo 過閘後取消下面兩行註解再 deploy(scheduled handler 已在 code)。
# [triggers]
# crons = ["*/5 * * * *"]
# ── Service Bindings(同帳號 worker-to-worker 正解;直接 fetch(*.leo21c.workers.dev) 會撞 CF 1042)──
[[services]]
binding = "SVC_CODE"
service = "arcrun-code"
[[services]]
binding = "SVC_KBDB"
service = "arcrun-kbdb"
[[services]]
binding = "SVC_GRAPH"
service = "kbdb-graph-plugin"
[vars]
REPO = "Leo/notes"
REF = "main"
OWNER = "leo"
CARDS_ROOT = "system-dev/wiki/cards"
CODE_WORKER_URL = "https://arcrun-code.leo21c.workers.dev/"
KBDB_URL = "https://arcrun-kbdb.leo21c.workers.dev"
GRAPH_URL = "https://kbdb-graph-plugin.leo21c.workers.dev"
GRAPH_KEY = "leo"
BATCH_SIZE = "2"
# secretwrangler secret put):GITEA_TOKEN(必),GITEA_WEBHOOK_SECRET(選)