From 9e159910cab3150630024c93eb60ffc18ef2fad6 Mon Sep 17 00:00:00 2001 From: tangsongdayo Date: Sun, 19 Jul 2026 05:43:03 +0800 Subject: [PATCH] feat: batch large source imports --- CHANGELOG.md | 1 + app/api/sources/[id]/manage/route.ts | 6 +++--- app/dashboard/page.tsx | 8 ++++---- lib/db.ts | 6 ++++++ lib/memos.ts | 16 +++++++++------- lib/sync.ts | 2 +- tests/sync.test.ts | 2 +- worker/index.ts | 28 +++++++++++++++------------- 8 files changed, 40 insertions(+), 29 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 0a68959..ea20ba9 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -24,6 +24,7 @@ ### Added +- Added resumable, page-token-based Memos imports with configurable batch size and first-import limit; posts are only hidden after a complete scan. - Added owner/member-only JSON and Markdown exports for individual posts and complete sources. - Added read-only RSS sources: validate a feed URL, queue recurring imports, and show imported posts under their RSS source rather than a Hub account. diff --git a/app/api/sources/[id]/manage/route.ts b/app/api/sources/[id]/manage/route.ts index 612e638..446f690 100644 --- a/app/api/sources/[id]/manage/route.ts +++ b/app/api/sources/[id]/manage/route.ts @@ -21,9 +21,9 @@ export async function POST(req: Request, { params }: { params: Promise<{ id: str } else if (action === "set-sync-rules") { if (!owner) throw new Error("Only the owner can change sync rules"); const tags = String(form.get("tags") || "").split(",").map((tag) => tag.trim().replace(/^#/, "")).filter(Boolean).slice(0, 20); - const from = String(form.get("from") || ""); const to = String(form.get("to") || ""); const attachmentMode = String(form.get("attachmentMode") || "all"); - if ((from && !/^\d{4}-\d{2}-\d{2}$/.test(from)) || (to && !/^\d{4}-\d{2}-\d{2}$/.test(to)) || (from && to && from > to) || !["all", "images", "none"].includes(attachmentMode)) throw new Error("Invalid sync rules"); - db.prepare("UPDATE sources SET sync_tags_json=?,sync_from=?,sync_to=?,sync_attachment_mode=? WHERE id=?").run(JSON.stringify(tags), from || null, to || null, attachmentMode, id); + const from = String(form.get("from") || ""); const to = String(form.get("to") || ""); const attachmentMode = String(form.get("attachmentMode") || "all"); const batchSize = Number(form.get("batchSize") || 100); const maxPosts = Number(form.get("maxPosts") || 0); + if ((from && !/^\d{4}-\d{2}-\d{2}$/.test(from)) || (to && !/^\d{4}-\d{2}-\d{2}$/.test(to)) || (from && to && from > to) || !["all", "images", "none"].includes(attachmentMode) || !Number.isInteger(batchSize) || batchSize < 10 || batchSize > 100 || !Number.isInteger(maxPosts) || maxPosts < 0 || maxPosts > 100000) throw new Error("Invalid sync rules"); + db.prepare("UPDATE sources SET sync_tags_json=?,sync_from=?,sync_to=?,sync_attachment_mode=?,sync_batch_size=?,sync_max_posts=?,sync_cursor=NULL,sync_imported_count=0,sync_run_id=NULL WHERE id=?").run(JSON.stringify(tags), from || null, to || null, attachmentMode, batchSize, maxPosts || null, id); queuePull(id, "manual"); } else if (action === "set-attachment-storage") { if (!owner) throw new Error("Only the owner can change attachment storage"); const mode = String(form.get("mode") || "remote"); const quotaMiB = Number(form.get("quotaMiB") || 100); diff --git a/app/dashboard/page.tsx b/app/dashboard/page.tsx index 18cceef..e211f19 100644 --- a/app/dashboard/page.tsx +++ b/app/dashboard/page.tsx @@ -4,14 +4,14 @@ import { db } from "@/lib/db"; import { PublishForm } from "./publish-form"; import { WebhookControl } from "./webhook-control"; -type Source = { id: number; name: string; base_url: string; integration_type: "memos" | "rss"; rss_feed_url: string | null; sync_status: string; last_synced_at: string | null; last_error: string | null; webhook_secret_hash: string | null; webhook_mode: string; last_webhook_at: string | null; owner_id: number; is_enabled: number; disabled_at: string | null; sync_tags_json: string; sync_from: string | null; sync_to: string | null; sync_attachment_mode: "all" | "images" | "none"; attachment_storage_mode: "remote" | "images" | "all"; attachment_cache_limit_bytes: number; attachment_cache_error: string | null; remote_display_name: string | null; remote_avatar_url: string | null; last_connection_at: string | null; last_connection_error: string | null }; +type Source = { id: number; name: string; base_url: string; integration_type: "memos" | "rss"; rss_feed_url: string | null; sync_status: string; last_synced_at: string | null; last_error: string | null; webhook_secret_hash: string | null; webhook_mode: string; last_webhook_at: string | null; owner_id: number; is_enabled: number; disabled_at: string | null; sync_tags_json: string; sync_from: string | null; sync_to: string | null; sync_attachment_mode: "all" | "images" | "none"; sync_batch_size: number; sync_max_posts: number | null; sync_cursor: string | null; sync_imported_count: number; attachment_storage_mode: "remote" | "images" | "all"; attachment_cache_limit_bytes: number; attachment_cache_error: string | null; remote_display_name: string | null; remote_avatar_url: string | null; last_connection_at: string | null; last_connection_error: string | null }; type Job = { id: number; kind: string; trigger: string | null; status: string; attempts: number; last_error: string | null; created_at: string; finished_at: string | null }; export const dynamic = "force-dynamic"; export default async function Dashboard({ searchParams }: { searchParams: Promise<{ error?: string; source?: string; sync?: string }> }) { const query = await searchParams; const user = await getSession(); if (!user) redirect("/login"); - const sourceRows = db.prepare("SELECT s.id,s.name,s.base_url,s.integration_type,s.rss_feed_url,s.sync_status,s.last_synced_at,s.last_error,s.webhook_secret_hash,s.webhook_mode,s.last_webhook_at,s.user_id AS owner_id,s.is_enabled,s.disabled_at,s.sync_tags_json,s.sync_from,s.sync_to,s.sync_attachment_mode,s.attachment_storage_mode,s.attachment_cache_limit_bytes,s.attachment_cache_error,s.remote_display_name,s.remote_avatar_url,s.last_connection_at,s.last_connection_error FROM sources s JOIN source_members sm ON sm.source_id=s.id WHERE sm.user_id=? ORDER BY s.id DESC").all(user.id) as Source[]; + const sourceRows = db.prepare("SELECT s.id,s.name,s.base_url,s.integration_type,s.rss_feed_url,s.sync_status,s.last_synced_at,s.last_error,s.webhook_secret_hash,s.webhook_mode,s.last_webhook_at,s.user_id AS owner_id,s.is_enabled,s.disabled_at,s.sync_tags_json,s.sync_from,s.sync_to,s.sync_attachment_mode,s.sync_batch_size,s.sync_max_posts,s.sync_cursor,s.sync_imported_count,s.attachment_storage_mode,s.attachment_cache_limit_bytes,s.attachment_cache_error,s.remote_display_name,s.remote_avatar_url,s.last_connection_at,s.last_connection_error FROM sources s JOIN source_members sm ON sm.source_id=s.id WHERE sm.user_id=? ORDER BY s.id DESC").all(user.id) as Source[]; const sources = sourceRows.map((source) => ({ ...source, syncTags: (() => { try { return JSON.parse(source.sync_tags_json) as string[]; } catch { return []; } })(), members: db.prepare("SELECT u.username,u.id,sm.role FROM source_members sm JOIN users u ON u.id=sm.user_id WHERE sm.source_id=? ORDER BY sm.role DESC,u.username").all(source.id) as { username: string; id: number; role: string }[], jobs: db.prepare("SELECT id,kind,trigger,status,attempts,last_error,created_at,finished_at FROM sync_jobs WHERE source_id=? ORDER BY id DESC LIMIT 5").all(source.id) as Job[] })); const publishSources = sources.filter((source) => source.is_enabled && source.integration_type === "memos"); return <> @@ -24,11 +24,11 @@ export default async function Dashboard({ searchParams }: { searchParams: Promis

連接 RSS

RSS 為唯讀來源，不需要 API Key，也不會回寫原網站。同步時會更新 Feed 內公開的項目。

已連接來源

{sources.map((source) =>
{source.name}{source.is_enabled ? source.sync_status : "disabled"}
-

來源 ID：{source.id}
{source.integration_type === "rss" ? source.rss_feed_url : source.base_url}{source.remote_display_name && <>
Memos 帳號：{source.remote_avatar_url && } {source.remote_display_name}}
成員：{source.members.map((member) => `${member.username}${member.role === "owner" ? "（建立者）" : ""}`).join("、")}
上次同步：{source.last_synced_at || "尚未完成"}
{source.integration_type === "memos" && <>附件保存：{source.attachment_storage_mode === "remote" ? "遠端連結" : source.attachment_storage_mode === "images" ? "只快取圖片" : "完整備份"}（配額 {Math.round(source.attachment_cache_limit_bytes / 1024 / 1024)} MiB）
連線：{source.last_connection_at ? `最近成功：${new Date(source.last_connection_at + "Z").toLocaleString("zh-TW")}` : "尚未測試"}
Webhook：{source.webhook_secret_hash ? (source.last_webhook_at ? (Date.now() - new Date(source.last_webhook_at + "Z").getTime() > 7 * 24 * 60 * 60 * 1000 ? `警示：超過 7 天未收到（最近：${new Date(source.last_webhook_at + "Z").toLocaleString("zh-TW")}）` : `健康（最近收到：${new Date(source.last_webhook_at + "Z").toLocaleString("zh-TW")}）`) : "已建立 URL，尚未收到呼叫") : "尚未建立 URL"}}{!source.is_enabled && <>
已停用：{source.disabled_at ? new Date(source.disabled_at + "Z").toLocaleString("zh-TW") : "是"}}{source.last_error && <>
同步：{source.last_error}}{source.last_connection_error && <>
連線：{source.last_connection_error}}{source.attachment_cache_error && <>
附件快取：{source.attachment_cache_error}}

+

來源 ID：{source.id}
{source.integration_type === "rss" ? source.rss_feed_url : source.base_url}{source.remote_display_name && <>
Memos 帳號：{source.remote_avatar_url && } {source.remote_display_name}}
成員：{source.members.map((member) => `${member.username}${member.role === "owner" ? "（建立者）" : ""}`).join("、")}
上次同步：{source.last_synced_at || "尚未完成"}{source.sync_cursor && <>（批次匯入中：{source.sync_imported_count} 篇）}
{source.integration_type === "memos" && <>附件保存：{source.attachment_storage_mode === "remote" ? "遠端連結" : source.attachment_storage_mode === "images" ? "只快取圖片" : "完整備份"}（配額 {Math.round(source.attachment_cache_limit_bytes / 1024 / 1024)} MiB）
連線：{source.last_connection_at ? `最近成功：${new Date(source.last_connection_at + "Z").toLocaleString("zh-TW")}` : "尚未測試"}
Webhook：{source.webhook_secret_hash ? (source.last_webhook_at ? (Date.now() - new Date(source.last_webhook_at + "Z").getTime() > 7 * 24 * 60 * 60 * 1000 ? `警示：超過 7 天未收到（最近：${new Date(source.last_webhook_at + "Z").toLocaleString("zh-TW")}）` : `健康（最近收到：${new Date(source.last_webhook_at + "Z").toLocaleString("zh-TW")}）`) : "已建立 URL，尚未收到呼叫") : "尚未建立 URL"}}{!source.is_enabled && <>
已停用：{source.disabled_at ? new Date(source.disabled_at + "Z").toLocaleString("zh-TW") : "是"}}{source.last_error && <>
同步：{source.last_error}}{source.last_connection_error && <>
連線：{source.last_connection_error}}{source.attachment_cache_error && <>
附件快取：{source.attachment_cache_error}}

{source.owner_id === user.id ? <> {source.integration_type === "memos" && }
-
來源管理
{source.members.length > 1 &&
}
+
來源管理
{source.members.length > 1 &&
}
:
}
最近同步工作{source.jobs.length ?
    {source.jobs.map((job) =>
  • {job.kind} · {job.trigger || "legacy"} · {job.status} · 嘗試 {job.attempts} 次
    建立：{new Date(job.created_at + "Z").toLocaleString("zh-TW")}{job.finished_at && `；完成：${new Date(job.finished_at + "Z").toLocaleString("zh-TW")}`}{job.last_error && <>
    {job.last_error}}
    • )}
:

尚無同步工作。

}
diff --git a/lib/db.ts b/lib/db.ts index e0f925c..25e05d1 100644 --- a/lib/db.ts +++ b/lib/db.ts @@ -130,6 +130,12 @@ applyColumnMigration(25, "sources", "webhook_remote_name", "ALTER TABLE sources applyColumnMigration(26, "sources", "webhook_signing_secret_encrypted", "ALTER TABLE sources ADD COLUMN webhook_signing_secret_encrypted TEXT"); applyColumnMigration(27, "sources", "integration_type", "ALTER TABLE sources ADD COLUMN integration_type TEXT NOT NULL DEFAULT 'memos'"); applyColumnMigration(28, "sources", "rss_feed_url", "ALTER TABLE sources ADD COLUMN rss_feed_url TEXT"); +applyColumnMigration(29, "sources", "sync_batch_size", "ALTER TABLE sources ADD COLUMN sync_batch_size INTEGER NOT NULL DEFAULT 100"); +applyColumnMigration(30, "sources", "sync_max_posts", "ALTER TABLE sources ADD COLUMN sync_max_posts INTEGER"); +applyColumnMigration(31, "sources", "sync_cursor", "ALTER TABLE sources ADD COLUMN sync_cursor TEXT"); +applyColumnMigration(32, "sources", "sync_imported_count", "ALTER TABLE sources ADD COLUMN sync_imported_count INTEGER NOT NULL DEFAULT 0"); +applyColumnMigration(33, "sources", "sync_run_id", "ALTER TABLE sources ADD COLUMN sync_run_id TEXT"); +applyColumnMigration(34, "posts", "last_seen_sync_run", "ALTER TABLE posts ADD COLUMN last_seen_sync_run TEXT"); const admin = process.env.ADMIN_USERNAME; const adminPassword = process.env.ADMIN_PASSWORD; diff --git a/lib/memos.ts b/lib/memos.ts index 8308a5c..343c32f 100644 --- a/lib/memos.ts +++ b/lib/memos.ts @@ -19,26 +19,28 @@ export async function getMemosIdentity(baseUrl: string, token: string) { return user; } export function memoUrl(baseUrl: string, memoName: string) { const id = memoName.split("/").at(-1); return id ? `${baseUrl.replace(/\/$/, "")}/m/${encodeURIComponent(id)}` : null; } -export async function listMemos(baseUrl: string, token: string, rules: MemosSyncRules = {}) { - const all: MemosMemo[] = []; let pageToken = ""; +export type MemosPage = { memos: MemosMemo[]; nextPageToken: string }; +export async function listMemos(baseUrl: string, token: string, rules: MemosSyncRules = {}, page: { pageToken?: string; pageSize?: number } = {}): Promise { + let pageToken = page.pageToken || ""; const tags = rules.tags?.filter(Boolean) || []; const mode = rules.attachmentMode || "all"; const quote = (value: string) => JSON.stringify(value); const filters = [rules.creator && `creator == ${quote(rules.creator)}`, `visibility == "PUBLIC"`, ...tags.map((tag) => `tags.exists(t, t == ${quote(tag)})`)].filter(Boolean).join(" && "); let useServerFilter = Boolean(filters); - do { const params = new URLSearchParams({ pageSize: "100" }); if (pageToken) params.set("pageToken", pageToken); if (useServerFilter && filters) params.set("filter", filters); let data: any; + do { const params = new URLSearchParams({ pageSize: String(Math.max(10, Math.min(100, page.pageSize || 100))) }); if (pageToken) params.set("pageToken", pageToken); if (useServerFilter && filters) params.set("filter", filters); let data: any; try { data = await (await request(`${base(baseUrl)}/memos?${params}`, token)).json(); } catch (error) { if (!pageToken && useServerFilter && error instanceof Error && error.message.startsWith("Memos API 400")) { useServerFilter = false; continue; } throw error; } - all.push(...(data.memos || [])); pageToken = data.nextPageToken || ""; } while (pageToken); - return all.filter((memo) => { + const memos = (data.memos || []).filter((memo: MemosMemo) => { if (memo.visibility !== "PUBLIC") return false; if (rules.creator && memo.creator !== rules.creator) return false; if (tags.length && !tags.every((tag) => memo.tags?.includes(tag))) return false; const created = memo.createTime?.slice(0, 10); if (rules.from && (!created || created < rules.from)) return false; if (rules.to && (!created || created > rules.to)) return false; return true; - }).map((memo) => { + }).map((memo: MemosMemo) => { if (mode === "all") return memo; const onlyImages = (items: T[] | undefined) => mode === "none" ? [] : (items || []).filter((item) => item.type?.startsWith("image/")); return { ...memo, attachments: onlyImages(memo.attachments), resources: onlyImages(memo.resources) }; - }); + }); + return { memos, nextPageToken: data.nextPageToken || "" }; + } while (true); } export async function createMemo(baseUrl: string, token: string, memo: Pick & { attachments?: unknown[]; resources?: unknown[] }) { return (await request(`${base(baseUrl)}/memos`, token, { method: "POST", body: JSON.stringify({ state: "NORMAL", ...memo }) })).json() as Promise; diff --git a/lib/sync.ts b/lib/sync.ts index d31b25c..2faf33c 100644 --- a/lib/sync.ts +++ b/lib/sync.ts @@ -1,6 +1,6 @@ import { db } from "@/lib/db"; -export type SyncTrigger = "manual" | "webhook" | "scheduled" | "source-created"; +export type SyncTrigger = "manual" | "webhook" | "scheduled" | "source-created" | "batch"; /** Queue one pull per source at a time. Returns true only when a new job was created. */ export function queuePull(sourceId: number, trigger: SyncTrigger, payload: unknown = {}) { diff --git a/tests/sync.test.ts b/tests/sync.test.ts index 41a47f6..fe7cb55 100644 --- a/tests/sync.test.ts +++ b/tests/sync.test.ts @@ -16,7 +16,7 @@ test("applies tracked migrations and deduplicates active pull jobs", async () => const { queuePull } = await import("../lib/sync"); const { notify } = await import("../lib/notifications"); const migrations = db.prepare("SELECT version FROM schema_migrations ORDER BY version").all() as { version: number }[]; - assert.deepEqual(migrations.map((item) => item.version), Array.from({ length: 28 }, (_, index) => index + 1)); + assert.deepEqual(migrations.map((item) => item.version), Array.from({ length: 34 }, (_, index) => index + 1)); const userId = Number(db.prepare("INSERT INTO users(username,password_hash) VALUES('sync-test','hash')").run().lastInsertRowid); const sourceId = Number(db.prepare("INSERT INTO sources(user_id,name,base_url,token_encrypted,is_enabled) VALUES(?,?,?,?,1)").run(userId, "Test", "https://example.test", "encrypted").lastInsertRowid); assert.equal(queuePull(sourceId, "manual"), true); diff --git a/worker/index.ts b/worker/index.ts index ec3a917..a4a5735 100644 --- a/worker/index.ts +++ b/worker/index.ts @@ -1,13 +1,13 @@ import { readFile, mkdir, readdir, unlink, writeFile } from "node:fs/promises"; import { extname, join } from "node:path"; -import { createHash } from "node:crypto"; +import { createHash, randomUUID } from "node:crypto"; import { db } from "../lib/db"; import { decrypt } from "../lib/crypto"; import { createMemo, createRemoteFile, getMemosIdentity, listMemos, memoUrl, setMemoAttachments } from "../lib/memos"; import { recordError } from "../lib/observability"; import { fetchRss } from "../lib/rss"; -type Source = { id: number; user_id: number; base_url: string; token_encrypted: string; remote_user: string | null; integration_type: "memos" | "rss"; rss_feed_url: string | null; is_enabled: number; sync_tags_json: string; sync_from: string | null; sync_to: string | null; sync_attachment_mode: "all" | "images" | "none"; attachment_storage_mode: "remote" | "images" | "all"; attachment_cache_limit_bytes: number }; +type Source = { id: number; user_id: number; base_url: string; token_encrypted: string; remote_user: string | null; integration_type: "memos" | "rss"; rss_feed_url: string | null; is_enabled: number; sync_tags_json: string; sync_from: string | null; sync_to: string | null; sync_attachment_mode: "all" | "images" | "none"; attachment_storage_mode: "remote" | "images" | "all"; attachment_cache_limit_bytes: number; sync_batch_size: number; sync_max_posts: number | null; sync_cursor: string | null; sync_imported_count: number; sync_run_id: string | null }; type Job = { id: number; source_id: number; kind: "pull" | "push"; payload_json: string | null; attempts: number }; function remoteAttachmentUrl(attachment: any, baseUrl: string) { @@ -43,25 +43,27 @@ async function cleanupCache(source: Source) { await Promise.all(names.filter((name) => !used.has(name)).map((name) => unlink(join(directory, name)).catch(() => undefined))); } -async function upsertRemote(source: Source, memo: any) { +async function upsertRemote(source: Source, memo: any, syncRun: string) { const cached = await cacheAttachments(source, memo.attachments || memo.resources || []); const tags = JSON.stringify(memo.tags || []), attachments = JSON.stringify(cached.items); - db.prepare(`INSERT INTO posts(source_id,author_id,remote_memo_name,content,visibility,tags_json,attachments_json,origin,remote_created_at,remote_updated_at,sync_status,hidden,remote_url) VALUES(?,?,?,?,?,?,?,?,?,?, 'synced',0,?) ON CONFLICT(source_id,remote_memo_name) DO UPDATE SET content=excluded.content,visibility=excluded.visibility,tags_json=excluded.tags_json,attachments_json=excluded.attachments_json,remote_updated_at=excluded.remote_updated_at,remote_url=excluded.remote_url,hidden=0,updated_at=CURRENT_TIMESTAMP`).run(source.id, source.user_id, memo.name, memo.content, memo.visibility, tags, attachments, "memos", memo.createTime || null, memo.updateTime || null, memoUrl(source.base_url, memo.name)); + db.prepare(`INSERT INTO posts(source_id,author_id,remote_memo_name,content,visibility,tags_json,attachments_json,origin,remote_created_at,remote_updated_at,sync_status,hidden,remote_url,last_seen_sync_run) VALUES(?,?,?,?,?,?,?,?,?,?, 'synced',0,?,?) ON CONFLICT(source_id,remote_memo_name) DO UPDATE SET content=excluded.content,visibility=excluded.visibility,tags_json=excluded.tags_json,attachments_json=excluded.attachments_json,remote_updated_at=excluded.remote_updated_at,remote_url=excluded.remote_url,last_seen_sync_run=excluded.last_seen_sync_run,hidden=0,updated_at=CURRENT_TIMESTAMP`).run(source.id, source.user_id, memo.name, memo.content, memo.visibility, tags, attachments, "memos", memo.createTime || null, memo.updateTime || null, memoUrl(source.base_url, memo.name), syncRun); return cached.errors; } async function pull(source: Source) { - if (source.integration_type === "rss") { const items = await fetchRss(source.rss_feed_url || ""); for (const item of items) db.prepare("INSERT INTO posts(source_id,author_id,remote_memo_name,content,visibility,tags_json,attachments_json,origin,remote_created_at,remote_updated_at,sync_status,hidden,remote_url) VALUES(?,?,?,?,?,'[]','[]','rss',?,?, 'synced',0,?) ON CONFLICT(source_id,remote_memo_name) DO UPDATE SET content=excluded.content,remote_created_at=excluded.remote_created_at,remote_url=excluded.remote_url,hidden=0,updated_at=CURRENT_TIMESTAMP").run(source.id,source.user_id,`rss:${item.id}`,item.content,"PUBLIC",item.publishedAt,item.publishedAt,item.link); const names=items.map((item)=>`rss:${item.id}`); if(names.length) db.prepare(`UPDATE posts SET hidden=1 WHERE source_id=? AND remote_memo_name NOT IN (${names.map(()=>"?").join(",")})`).run(source.id,...names); db.prepare("UPDATE sources SET sync_status='synced',last_synced_at=CURRENT_TIMESTAMP,last_error=NULL WHERE id=?").run(source.id); return; } + if (source.integration_type === "rss") { const items = await fetchRss(source.rss_feed_url || ""); for (const item of items) db.prepare("INSERT INTO posts(source_id,author_id,remote_memo_name,content,visibility,tags_json,attachments_json,origin,remote_created_at,remote_updated_at,sync_status,hidden,remote_url) VALUES(?,?,?,?,?,'[]','[]','rss',?,?, 'synced',0,?) ON CONFLICT(source_id,remote_memo_name) DO UPDATE SET content=excluded.content,remote_created_at=excluded.remote_created_at,remote_url=excluded.remote_url,hidden=0,updated_at=CURRENT_TIMESTAMP").run(source.id,source.user_id,`rss:${item.id}`,item.content,"PUBLIC",item.publishedAt,item.publishedAt,item.link); const names=items.map((item)=>`rss:${item.id}`); if(names.length) db.prepare(`UPDATE posts SET hidden=1 WHERE source_id=? AND remote_memo_name NOT IN (${names.map(()=>"?").join(",")})`).run(source.id,...names); db.prepare("UPDATE sources SET sync_status='synced',last_synced_at=CURRENT_TIMESTAMP,last_error=NULL WHERE id=?").run(source.id); return false; } const token = decrypt(source.token_encrypted); const identity = await getMemosIdentity(source.base_url, token); const creator = identity.name; const rules = { creator, tags: JSON.parse(source.sync_tags_json || "[]") as string[], from: source.sync_from, to: source.sync_to, attachmentMode: source.sync_attachment_mode }; - const memos = await listMemos(source.base_url, token, rules); - const cacheErrors: string[] = []; for (const memo of memos) cacheErrors.push(...await upsertRemote(source, memo)); - const names = memos.map((memo) => memo.name); - if (names.length) { const placeholders = names.map(() => "?").join(","); db.prepare(`UPDATE posts SET hidden=1,updated_at=CURRENT_TIMESTAMP WHERE source_id=? AND remote_memo_name IS NOT NULL AND remote_memo_name NOT IN (${placeholders})`).run(source.id, ...names); } - else db.prepare("UPDATE posts SET hidden=1,updated_at=CURRENT_TIMESTAMP WHERE source_id=? AND remote_memo_name IS NOT NULL").run(source.id); + const remaining = source.sync_max_posts ? Math.max(0, source.sync_max_posts - source.sync_imported_count) : source.sync_batch_size; + if (remaining === 0) { db.prepare("UPDATE sources SET sync_cursor=NULL,sync_status='synced',last_synced_at=CURRENT_TIMESTAMP WHERE id=?").run(source.id); return false; } + const page = await listMemos(source.base_url, token, rules, { pageToken: source.sync_cursor || "", pageSize: Math.min(source.sync_batch_size, remaining) }); const memos = page.memos; const syncRun = source.sync_run_id || randomUUID(); + const cacheErrors: string[] = []; for (const memo of memos) cacheErrors.push(...await upsertRemote(source, memo, syncRun)); + const imported = source.sync_imported_count + memos.length; const capped = Boolean(source.sync_max_posts && imported >= source.sync_max_posts); const complete = !page.nextPageToken && !capped; const more = Boolean(page.nextPageToken) && !capped; + if (complete) db.prepare("UPDATE posts SET hidden=1,updated_at=CURRENT_TIMESTAMP WHERE source_id=? AND origin='memos' AND COALESCE(last_seen_sync_run,'')<>?").run(source.id, syncRun); await cleanupCache(source); const avatar = identity.avatarUrl || identity.avatar || null; const avatarUrl = avatar?.startsWith("/") ? `${source.base_url.replace(/\/$/, "")}${avatar}` : avatar; const name = identity.displayName || identity.nickname || identity.username || identity.name; - db.prepare("UPDATE sources SET name=?,remote_user=?,sync_status='synced',last_synced_at=CURRENT_TIMESTAMP,last_error=NULL,last_connection_at=CURRENT_TIMESTAMP,last_connection_error=NULL,attachment_cache_error=?,remote_display_name=?,remote_avatar_url=? WHERE id=?").run(name, creator, cacheErrors.length ? `${cacheErrors.length} 個附件快取失敗；可按「立即同步」重試。` : null, name, avatarUrl, source.id); + db.prepare("UPDATE sources SET name=?,remote_user=?,sync_status=?,last_synced_at=CASE WHEN ? THEN NULL ELSE CURRENT_TIMESTAMP END,last_error=NULL,last_connection_at=CURRENT_TIMESTAMP,last_connection_error=NULL,attachment_cache_error=?,remote_display_name=?,remote_avatar_url=?,sync_cursor=?,sync_imported_count=?,sync_run_id=? WHERE id=?").run(name, creator, more ? 'importing' : 'synced', more ? 1 : 0, cacheErrors.length ? `${cacheErrors.length} 個附件快取失敗；可按「立即同步」重試。` : null, name, avatarUrl, more ? page.nextPageToken : null, more ? imported : 0, more ? syncRun : null, source.id); + return more; } async function push(source: Source, payload: any) { @@ -86,9 +88,9 @@ async function run() { const source = db.prepare("SELECT * FROM sources WHERE id=?").get(job.source_id) as Source | undefined; if (!source || !source.is_enabled) { db.prepare("UPDATE sync_jobs SET status='cancelled',last_error='Source is disabled or deleted',finished_at=CURRENT_TIMESTAMP WHERE id=?").run(job.id); return; } try { - if (job.kind === "pull") await pull(source); else if (job.kind === "push") await push(source, JSON.parse(job.payload_json || "{}")); + const more = job.kind === "pull" ? await pull(source) : false; if (job.kind === "push") await push(source, JSON.parse(job.payload_json || "{}")); db.prepare("UPDATE sync_jobs SET status='done',finished_at=CURRENT_TIMESTAMP WHERE id=?").run(job.id); - db.prepare("UPDATE sources SET sync_status='synced',last_error=NULL,last_synced_at=CURRENT_TIMESTAMP WHERE id=?").run(source.id); + if (more) db.prepare("INSERT INTO sync_jobs(source_id,kind,trigger) VALUES(?,'pull','batch')").run(source.id); else db.prepare("UPDATE sources SET sync_status='synced',last_error=NULL,last_synced_at=CURRENT_TIMESTAMP WHERE id=?").run(source.id); } catch (error) { const message = error instanceof Error ? error.message : "Sync failure"; const exhausted = job.attempts + 1 >= 5; recordError("sync", error, { sourceId: source.id, jobId: job.id, kind: job.kind, attempts: job.attempts });