first-update

easy-dataset-main/app/api/projects/[projectId]/distill/questions/route.js

@@ -0,0 +1,101 @@
+import { NextResponse } from 'next/server';
+import { distillQuestionsPrompt } from '@/lib/llm/prompts/distillQuestions';
+import { db } from '@/lib/db';
+
+const LLMClient = require('@/lib/llm/core');
+
+/**
+ * 生成问题接口：根据某个标签链路构造指定数量的问题
+ */
+export async function POST(request, { params }) {
+  try {
+    const { projectId } = params;
+
+    // 验证项目ID
+    if (!projectId) {
+      return NextResponse.json({ error: '项目ID不能为空' }, { status: 400 });
+    }
+
+    const { tagPath, currentTag, tagId, count = 5, model, language = 'zh' } = await request.json();
+
+    if (!currentTag || !tagPath) {
+      const errorMsg = language === 'en' ? 'Tag information cannot be empty' : '标签信息不能为空';
+      return NextResponse.json({ error: errorMsg }, { status: 400 });
+    }
+
+    // 首先获取或创建蒸馏文本块
+    let distillChunk = await db.chunks.findFirst({
+      where: {
+        projectId,
+        name: 'Distilled Content'
+      }
+    });
+
+    if (!distillChunk) {
+      // 创建一个特殊的蒸馏文本块
+      distillChunk = await db.chunks.create({
+        data: {
+          name: 'Distilled Content',
+          projectId,
+          fileId: 'distilled',
+          fileName: 'distilled.md',
+          content:
+            'This text block is used to store questions generated through data distillation and is not related to actual literature.',
+          summary: 'Questions generated through data distillation',
+          size: 0
+        }
+      });
+    }
+
+    // 获取已有的问题，避免重复
+    const existingQuestions = await db.questions.findMany({
+      where: {
+        projectId,
+        label: currentTag,
+        chunkId: distillChunk.id // 使用蒸馏文本块的 ID
+      },
+      select: { question: true }
+    });
+
+    const existingQuestionTexts = existingQuestions.map(q => q.question);
+
+    const llmClient = new LLMClient(model);
+    const prompt = await distillQuestionsPrompt(
+      language,
+      { tagPath, currentTag, count, existingQuestionTexts },
+      projectId
+    );
+    const { answer } = await llmClient.getResponseWithCOT(prompt);
+
+    let questions = [];
+    try {
+      questions = JSON.parse(answer);
+    } catch (error) {
+      console.error('解析问题JSON失败:', String(error));
+      // 尝试使用正则表达式提取问题
+      const matches = answer.match(/"([^"]+)"/g);
+      if (matches) {
+        questions = matches.map(match => match.replace(/"/g, ''));
+      }
+    }
+
+    // 保存问题到数据库
+    const savedQuestions = [];
+    for (const questionText of questions) {
+      const question = await db.questions.create({
+        data: {
+          question: questionText,
+          projectId,
+          label: currentTag,
+          chunkId: distillChunk.id
+        }
+      });
+      savedQuestions.push(question);
+    }
+
+    return NextResponse.json(savedQuestions);
+  } catch (error) {
+    console.error('生成问题失败:', String(error));
+    return NextResponse.json({ error: error.message || '生成问题失败' }, { status: 500 });
+  }
+}