0.3.5 共享功能修复

wolai-frontend/convex/mediaAssets.ts

@@ -1,7 +1,7 @@
 import { mutation, query } from "./_generated/server";
 import { v } from "convex/values";
 import { nowIso } from "./_utils/time";
-import { enqueueIngestMediaAssetJob } from "./_utils/ingestJobs";
+import { enqueueExtractMediaAssetTextJob, enqueueIngestMediaAssetJob } from "./_utils/ingestJobs";
 import type { MutationCtx, QueryCtx } from "./_generated/server";
 
 async function assertWorkspaceMember(ctx: QueryCtx | MutationCtx, userId: string, workspaceId: string) {
@@ -15,6 +15,26 @@ async function assertWorkspaceMember(ctx: QueryCtx | MutationCtx, userId: string
   return membership;
 }
 
+function shouldExtractAttachmentText(args: {
+  assetType?: string | null;
+  mimeType?: string | null;
+  fileName?: string | null;
+}): boolean {
+  if (String(args.assetType ?? "") !== "file") return false;
+  const mime = String(args.mimeType ?? "").toLowerCase().trim();
+  if (mime === "application/pdf") return true;
+  if (mime === "application/vnd.openxmlformats-officedocument.wordprocessingml.document") return true;
+  if (mime === "application/vnd.openxmlformats-officedocument.presentationml.presentation") return true;
+  if (mime === "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet") return true;
+
+  const name = String(args.fileName ?? "").toLowerCase().trim();
+  if (name.endsWith(".pdf")) return true;
+  if (name.endsWith(".docx")) return true;
+  if (name.endsWith(".pptx")) return true;
+  if (name.endsWith(".xlsx")) return true;
+  return false;
+}
+
 export const getById = query({
   args: { userId: v.string(), id: v.string() },
   handler: async (ctx, args) => {
@@ -76,6 +96,44 @@ export const listByWorkspace = query({
   },
 });
 
+export const listSearchDataByWorkspace = query({
+  args: {
+    userId: v.string(),
+    workspaceId: v.string(),
+    includeDeleted: v.optional(v.boolean()),
+    limit: v.optional(v.number()),
+  },
+  handler: async (ctx, args) => {
+    await assertWorkspaceMember(ctx, args.userId, args.workspaceId);
+
+    const includeDeleted = Boolean(args.includeDeleted);
+    const limit = Math.max(1, Math.min(5000, Math.floor(args.limit ?? 5000)));
+
+    const rows = await ctx.db
+      .query("media_assets")
+      .withIndex("by_workspace", (q) => q.eq("workspace_id", args.workspaceId))
+      .order("desc")
+      .take(limit * 2);
+
+    return rows
+      .filter((r) => (includeDeleted ? true : !r.deleted_at))
+      .filter((r) => !r.purged_at)
+      .slice(0, limit)
+      .map((r) => ({
+        id: r.id,
+        workspace_id: r.workspace_id,
+        document_id: r.document_id,
+        asset_type: r.asset_type,
+        file_name: r.file_name ?? null,
+        mime_type: r.mime_type ?? null,
+        ocr_text: r.ocr_text ?? null,
+        ocr_status: r.ocr_status ?? null,
+        created_at: r.created_at ?? null,
+        updated_at: r.updated_at ?? null,
+      }));
+  },
+});
+
 export const listByDocument = query({
   args: {
     userId: v.string(),
@@ -162,7 +220,13 @@ export const create = mutation({
       ...args.asset,
       storage_id: args.asset.storage_id ?? null,
       ocr_text: null,
-      ocr_status: null,
+      ocr_status: shouldExtractAttachmentText({
+        assetType: args.asset.asset_type,
+        mimeType: args.asset.mime_type,
+        fileName: args.asset.file_name,
+      })
+        ? "queued"
+        : null,
       deleted_at: null,
       deleted_by: null,
       purged_at: null,
@@ -171,6 +235,16 @@ export const create = mutation({
       updated_at: ts,
     });
 
+    if (
+      shouldExtractAttachmentText({
+        assetType: args.asset.asset_type,
+        mimeType: args.asset.mime_type,
+        fileName: args.asset.file_name,
+      })
+    ) {
+      await enqueueExtractMediaAssetTextJob(ctx, { userId: args.userId, assetId: args.asset.id, debounceMs: 800 });
+    }
+
     return args.asset;
   },
 });
@@ -222,7 +296,13 @@ export const createWithStorage = mutation({
       mime_type: args.asset.mime_type,
 
       ocr_text: null,
-      ocr_status: null,
+      ocr_status: shouldExtractAttachmentText({
+        assetType: args.asset.asset_type,
+        mimeType: args.asset.mime_type,
+        fileName: args.asset.file_name,
+      })
+        ? "queued"
+        : null,
       deleted_at: null,
       deleted_by: null,
       purged_at: null,
@@ -232,6 +312,16 @@ export const createWithStorage = mutation({
     };
 
     await ctx.db.insert("media_assets", row);
+
+    if (
+      shouldExtractAttachmentText({
+        assetType: args.asset.asset_type,
+        mimeType: args.asset.mime_type,
+        fileName: args.asset.file_name,
+      })
+    ) {
+      await enqueueExtractMediaAssetTextJob(ctx, { userId: args.userId, assetId: args.asset.id, debounceMs: 800 });
+    }
     return row;
   },
 });
@@ -312,13 +402,57 @@ export const replaceStorageFromUpload = mutation({
       thumbnail_url: url,
       bucket: null,
       storage_path: null,
+      ocr_text: null,
+      ocr_status: shouldExtractAttachmentText({
+        assetType: row.asset_type,
+        mimeType: row.mime_type ?? null,
+        fileName: row.file_name ?? null,
+      })
+        ? "queued"
+        : row.ocr_status ?? null,
       updated_at: nowIso(),
     });
 
+    if (
+      shouldExtractAttachmentText({
+        assetType: row.asset_type,
+        mimeType: row.mime_type ?? null,
+        fileName: row.file_name ?? null,
+      })
+    ) {
+      await enqueueExtractMediaAssetTextJob(ctx, { userId: args.userId, assetId: args.id, debounceMs: 800 });
+    }
+
     return { ok: true, fileUrl: url };
   },
 });
 
+export const enqueueExtractText = mutation({
+  args: { userId: v.string(), id: v.string() },
+  handler: async (ctx, args) => {
+    const row = await ctx.db
+      .query("media_assets")
+      .withIndex("by_asset_id", (q) => q.eq("id", args.id))
+      .first();
+    if (!row) throw new Error("资源不存在");
+    await assertWorkspaceMember(ctx, args.userId, row.workspace_id);
+
+    if (
+      !shouldExtractAttachmentText({
+        assetType: row.asset_type,
+        mimeType: row.mime_type ?? null,
+        fileName: row.file_name ?? null,
+      })
+    ) {
+      return { ok: false, reason: "unsupported" as const };
+    }
+
+    await ctx.db.patch(row._id, { ocr_status: "queued", updated_at: nowIso() });
+    await enqueueExtractMediaAssetTextJob(ctx, { userId: args.userId, assetId: args.id, debounceMs: 0 });
+    return { ok: true };
+  },
+});
+
 export const purgeById = mutation({
   args: { userId: v.string(), id: v.string(), expiredDeletedAt: v.string() },
   handler: async (ctx, args) => {