feat: 添加 OCR 词汇提取功能

新增 OCR 页面，用户可上传教材词汇表截图，使用 GLM-4.6V 视觉模型
提取单词-释义对并保存到指定文件夹。

- AI 管道: src/lib/bigmodel/ocr/ (orchestrator, types)
- 后端模块: src/modules/ocr/ (action-service-repository 架构)
- 前端页面: src/app/(features)/ocr/ (拖拽上传、folder 选择)
- i18n: 8 种语言翻译支持

src/modules/ocr/ocr-action-dto.ts

@@ -0,0 +1,20 @@
+import { z } from "zod";
+
+export const schemaActionInputProcessOCR = z.object({
+  imageBase64: z.string().min(1, "Image is required"),
+  folderId: z.number().int().positive("Folder ID must be positive"),
+  sourceLanguage: z.string().optional(),
+  targetLanguage: z.string().optional(),
+});
+
+export type ActionInputProcessOCR = z.infer<typeof schemaActionInputProcessOCR>;
+
+export interface ActionOutputProcessOCR {
+  success: boolean;
+  message: string;
+  data?: {
+    pairsCreated: number;
+    sourceLanguage?: string;
+    targetLanguage?: string;
+  };
+}

src/modules/ocr/ocr-action.ts

@@ -0,0 +1,25 @@
+"use server";
+
+import { validate } from "@/utils/validate";
+import { ValidateError } from "@/lib/errors";
+import { createLogger } from "@/lib/logger";
+import { serviceProcessOCR } from "./ocr-service";
+import { schemaActionInputProcessOCR } from "./ocr-action-dto";
+import type { ActionOutputProcessOCR } from "./ocr-action-dto";
+
+const log = createLogger("ocr-action");
+
+export async function actionProcessOCR(
+  input: unknown
+): Promise<ActionOutputProcessOCR> {
+  try {
+    const validatedInput = validate(input, schemaActionInputProcessOCR);
+    return serviceProcessOCR(validatedInput);
+  } catch (e) {
+    if (e instanceof ValidateError) {
+      return { success: false, message: e.message };
+    }
+    log.error("OCR action failed", { error: e });
+    return { success: false, message: "Unknown error occurred." };
+  }
+}

src/modules/ocr/ocr-repository-dto.ts

@@ -0,0 +1 @@
+export type { RepoInputCreatePair } from "@/modules/folder/folder-repository-dto";

src/modules/ocr/ocr-repository.ts

@@ -0,0 +1,5 @@
+import { repoCreatePair, repoGetUserIdByFolderId } from "@/modules/folder/folder-repository";
+import type { RepoInputCreatePair } from "./ocr-repository-dto";
+
+export { repoCreatePair, repoGetUserIdByFolderId };
+export type { RepoInputCreatePair };

src/modules/ocr/ocr-service-dto.ts

@@ -0,0 +1,20 @@
+import { z } from "zod";
+
+export const schemaServiceInputProcessOCR = z.object({
+  imageBase64: z.string().min(1, "Image is required"),
+  folderId: z.number().int().positive("Folder ID must be positive"),
+  sourceLanguage: z.string().optional(),
+  targetLanguage: z.string().optional(),
+});
+
+export type ServiceInputProcessOCR = z.infer<typeof schemaServiceInputProcessOCR>;
+
+export interface ServiceOutputProcessOCR {
+  success: boolean;
+  message: string;
+  data?: {
+    pairsCreated: number;
+    sourceLanguage?: string;
+    targetLanguage?: string;
+  };
+}

src/modules/ocr/ocr-service.ts

@@ -0,0 +1,96 @@
+"use server";
+
+import { executeOCR } from "@/lib/bigmodel/ocr/orchestrator";
+import { repoCreatePair, repoGetUserIdByFolderId } from "@/modules/folder/folder-repository";
+import { auth } from "@/auth";
+import { headers } from "next/headers";
+import { createLogger } from "@/lib/logger";
+import type { ServiceInputProcessOCR, ServiceOutputProcessOCR } from "./ocr-service-dto";
+
+const log = createLogger("ocr-service");
+
+export async function serviceProcessOCR(
+  input: ServiceInputProcessOCR
+): Promise<ServiceOutputProcessOCR> {
+  log.info("Processing OCR request", { folderId: input.folderId });
+
+  const session = await auth.api.getSession({ headers: await headers() });
+  if (!session?.user?.id) {
+    log.warn("Unauthorized OCR attempt");
+    return { success: false, message: "Unauthorized" };
+  }
+
+  const folderOwner = await repoGetUserIdByFolderId(input.folderId);
+  if (folderOwner !== session.user.id) {
+    log.warn("Folder ownership mismatch", { 
+      folderId: input.folderId, 
+      userId: session.user.id 
+    });
+    return { 
+      success: false, 
+      message: "You don't have permission to modify this folder" 
+    };
+  }
+
+  let ocrResult;
+  try {
+    log.debug("Calling OCR pipeline");
+    ocrResult = await executeOCR({
+      imageBase64: input.imageBase64,
+      sourceLanguage: input.sourceLanguage,
+      targetLanguage: input.targetLanguage,
+    });
+  } catch (error) {
+    log.error("OCR pipeline failed", { error });
+    return { 
+      success: false, 
+      message: "Failed to process image. Please try again." 
+    };
+  }
+
+  if (!ocrResult.pairs || ocrResult.pairs.length === 0) {
+    log.info("No vocabulary pairs extracted from image");
+    return { 
+      success: false, 
+      message: "No vocabulary pairs could be extracted from the image" 
+    };
+  }
+
+  const sourceLanguage = ocrResult.detectedSourceLanguage || input.sourceLanguage || "Unknown";
+  const targetLanguage = ocrResult.detectedTargetLanguage || input.targetLanguage || "Unknown";
+
+  let pairsCreated = 0;
+  for (const pair of ocrResult.pairs) {
+    try {
+      await repoCreatePair({
+        folderId: input.folderId,
+        language1: sourceLanguage,
+        language2: targetLanguage,
+        text1: pair.word,
+        text2: pair.definition,
+      });
+      pairsCreated++;
+    } catch (error) {
+      log.error("Failed to create pair", { 
+        word: pair.word, 
+        error 
+      });
+    }
+  }
+
+  log.info("OCR processing complete", { 
+    pairsCreated, 
+    sourceLanguage, 
+    targetLanguage 
+  });
+
+  return {
+    success: true,
+    message: `Successfully created ${pairsCreated} vocabulary pairs`,
+    data: {
+      pairsCreated,
+      sourceLanguage,
+      targetLanguage,
+    },
+  };
+}