From c37f9dcf625b487276c6d29c51a1376ec109fb1b Mon Sep 17 00:00:00 2001
From: noelorin <noelorin@outlook.com>
Date: Fri, 25 Sep 2026 02:56:23 +0800
Subject: [PATCH] fix(trpc): bound scrape concurrency and surface DB errors

---
 packages/trpc/src/services/library.service.ts | 38 +++++++++++--------
 packages/trpc/src/services/scrape.service.ts  | 37 +++++++++---------
 2 files changed, 42 insertions(+), 33 deletions(-)

diff --git a/packages/trpc/src/services/library.service.ts b/packages/trpc/src/services/library.service.ts
index 09a481a..da4c1c3 100644
--- a/packages/trpc/src/services/library.service.ts
+++ b/packages/trpc/src/services/library.service.ts
@@ -5,6 +5,10 @@ import { createWebdav, isVideoFilename, joinWebdavPath, listDirectory } from "./
 import { scrapeService } from "./scrape.service.js";
 
 const MAX_SCAN_FILES = 2000;
+/** 同时执行的 upsert+刮削 job 上限，避免最坏 2000 并发打 Bangumi 被限流 */
+const SCRAPE_CONCURRENCY = 4;
+
+type ScanJob = () => Promise<void>;
 
 async function walkVideos(
     userId: string,
@@ -12,7 +16,7 @@ async function walkVideos(
     root: string,
     rel: string,
     budget: { left: number },
-    out: Promise<void>[],
+    out: ScanJob[],
 ): Promise<void> {
     if (budget.left <= 0) return;
     const mount = await mountDao.getByIdForUser(mountId, userId);
@@ -38,19 +42,18 @@ async function walkVideos(
         if (!isVideoFilename(entry.basename)) continue;
         budget.left -= 1;
         const title = entry.basename.replace(/\.[a-z0-9]+$/i, "");
-        out.push(
-            (async () => {
-                const item = await mediaItemDao.upsertFromScan({
-                    userId,
-                    mountId,
-                    path: childRel,
-                    rawName: entry.basename,
-                    title,
-                    size: entry.size,
-                });
-                await scrapeService.autoScrape(userId, item.id);
-            })(),
-        );
+        // 延迟执行：扫描阶段只收集 job，随后分批限流
+        out.push(async () => {
+            const item = await mediaItemDao.upsertFromScan({
+                userId,
+                mountId,
+                path: childRel,
+                rawName: entry.basename,
+                title,
+                size: entry.size,
+            });
+            await scrapeService.autoScrape(userId, item.id);
+        });
     }
 }
 
@@ -78,9 +81,12 @@ export const libraryService = {
         if (!mount) throw new TRPCError({ code: "NOT_FOUND", message: "挂载不存在" });
         if (!mount.enabled) throw new TRPCError({ code: "BAD_REQUEST", message: "挂载已禁用" });
         const budget = { left: MAX_SCAN_FILES };
-        const jobs: Promise<void>[] = [];
+        const jobs: ScanJob[] = [];
         await walkVideos(userId, mountId, mount.rootPath || "/", "/", budget, jobs);
-        await Promise.allSettled(jobs);
+        for (let i = 0; i < jobs.length; i += SCRAPE_CONCURRENCY) {
+            const batch = jobs.slice(i, i + SCRAPE_CONCURRENCY).map((job) => job());
+            await Promise.allSettled(batch);
+        }
         return { scanned: MAX_SCAN_FILES - budget.left, message: "扫描完成" };
     },
 };
diff --git a/packages/trpc/src/services/scrape.service.ts b/packages/trpc/src/services/scrape.service.ts
index fbc2676..eb8e6bf 100644
--- a/packages/trpc/src/services/scrape.service.ts
+++ b/packages/trpc/src/services/scrape.service.ts
@@ -104,29 +104,32 @@ export const scrapeService = {
             });
             return;
         }
+        // 只包网络搜索：DB 写失败必须向外抛，避免被误标 failed 后吞掉
+        let hits: BangumiSearchHit[];
         try {
-            const hits = await bangumiSearch(q);
-            const hit = hits[0];
-            if (!hit) {
-                await mediaItemDao.updateScrape(mediaItemId, userId, {
-                    scrapeStatus: "unmatched",
-                    scrapedAt: new Date(),
-                });
-                return;
-            }
-            await mediaItemDao.updateScrape(mediaItemId, userId, {
-                bangumiId: hit.id,
-                scrapeStatus: "ok",
-                scrapedAt: new Date(),
-                posterUrl: hit.image || null,
-                title: hit.nameCn || hit.name,
-                epNumber: parseEpisodeFromFilename(row.rawName),
-            });
+            hits = await bangumiSearch(q);
         } catch {
             await mediaItemDao.updateScrape(mediaItemId, userId, {
                 scrapeStatus: "failed",
                 scrapedAt: new Date(),
             });
+            return;
         }
+        const hit = hits[0];
+        if (!hit) {
+            await mediaItemDao.updateScrape(mediaItemId, userId, {
+                scrapeStatus: "unmatched",
+                scrapedAt: new Date(),
+            });
+            return;
+        }
+        await mediaItemDao.updateScrape(mediaItemId, userId, {
+            bangumiId: hit.id,
+            scrapeStatus: "ok",
+            scrapedAt: new Date(),
+            posterUrl: hit.image || null,
+            title: hit.nameCn || hit.name,
+            epNumber: parseEpisodeFromFilename(row.rawName),
+        });
     },
 };