aiquery2/api/app/controllers/tools.js

const { spawn } = require('child_process');
const fsPromises = require('fs').promises;
const fs = require('fs');
const path = require('path');
const os = require('os');
const { convertPdfToImages } = require('../utils/pdfToImages');

const soffice = process.env.SOFFICE_PATH || 'soffice';

module.exports.docToDocx = async (ctx, next) => {
   let tempDir = null;
   let outputPath = null;

   try {
      if (!ctx.file) {
         ctx.status = 400;
         ctx.body = { message: '请上传一个 .doc 文件' };
         return;
      }

      const file = ctx.file; // { fieldname, originalname, encoding, mimetype, destination, filename, path, size }

      // 校验扩展名
      if (!file.originalname.toLowerCase().endsWith('.doc')) {
         ctx.status = 400;
         ctx.body = { message: '仅支持 .doc 格式的文件' };
         return;
      }

      // 创建专属临时目录（避免并发冲突）
      tempDir = await fsPromises.mkdtemp(path.join(os.tmpdir(), 'doc2docx-'));
      const inputPath = path.join(tempDir, file.originalname);
      const outputDir = tempDir;

      // 移动 multer 生成的临时文件到我们的目录（可选，也可直接用 file.path）
      await fsPromises.rename(file.path, inputPath);

      // 调用 LibreOffice 转换
      const args = ['--headless', '--convert-to', 'docx', '--outdir', outputDir, inputPath];
      await new Promise((resolve, reject) => {
         const child = spawn(soffice, args, { stdio: 'ignore', windowsHide: true });
         child.on('error', reject);
         child.on('close', (code) => (code === 0 ? resolve() : reject(new Error(`Exit code ${code}`))));
      });

      // 构造输出路径
      const baseName = path.basename(file.originalname, '.doc');
      outputPath = path.join(outputDir, `${baseName}.docx`);
      await fsPromises.access(outputPath); // 确保文件存在

      // 返回文件流
      ctx.set('Content-Type', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document');
      ctx.set('Content-Disposition', `attachment; filename="${encodeURIComponent(baseName)}.docx"`);
      ctx.body = fs.createReadStream(outputPath);

      // 清理
      ctx.res.on('finish', () => fsPromises.rm(tempDir, { recursive: true, force: true }).catch(() => { }));

   } catch (error) {
      ctx.logger?.error('docToDocx error:', error);
      ctx.status = 400;
      ctx.body = { message: error.message || '文件转换失败' };

      // 出错时立即清理
      if (tempDir) fsPromises.rm(tempDir, { recursive: true, force: true }).catch(() => { });
   }
};


// ========== 新增：检测 pdftoppm 是否可用 ==========
function checkPopplerAvailable() {
   return new Promise((resolve) => {
      const { spawn } = require('child_process');
      let found = false;
      const child = spawn('pdftoppm', ['-h'], { timeout: 2000 });

      child.on('error', (err) => {
         console.warn('[PDF2IMG] pdftoppm not found or failed to start:', err.message);
         resolve(false);
      });

      child.on('close', (code) => {
         if (code === 0) {
            console.log('[PDF2IMG] ✅ Poppler (pdftoppm) is available');
            resolve(true);
         } else {
            console.warn(`[PDF2IMG] pdftoppm exited with code ${code}`);
            resolve(false);
         }
      });
   });
}

module.exports.pdfToImage = async (ctx, next) => {
   let tempDir = null;

   // 记录请求开始
   const requestId = Math.random().toString(36).substring(2, 10);
   const logPrefix = `[REQ-${requestId}]`;

   try {
      console.log(`${logPrefix} 📥 Received PDF upload request`);

      if (!ctx.file) {
         ctx.status = 400;
         ctx.body = { message: '请上传一个 .pdf 文件' };
         return;
      }

      const file = ctx.file;
      console.log(`${logPrefix} Original filename: ${file.originalname}`);
      console.log(`${logPrefix} Temp file path: ${file.path}`);
      console.log(`${logPrefix} File size: ${file.size} bytes`);

      if (!file.originalname.toLowerCase().endsWith('.pdf')) {
         ctx.status = 400;
         ctx.body = { message: '仅支持 .pdf 格式的文件' };
         return;
      }

      // 创建临时目录
      tempDir = await fsPromises.mkdtemp(path.join(os.tmpdir(), 'pdf2image-'));
      console.log(`${logPrefix} Using temp dir: ${tempDir}`);

      const safeFilename = `input_${Date.now()}.pdf`;
      const inputPath = path.join(tempDir, safeFilename);
      const outputDir = tempDir;

      // 复制文件（关键修复）
      console.log(`${logPrefix} Copying file to: ${inputPath}`);
      await fsPromises.copyFile(file.path, inputPath);
      await fsPromises.unlink(file.path); // 清理 multer 临时文件
      console.log(`${logPrefix} Multer temp file deleted`);

      // 可选：调试时保留临时文件（通过环境变量控制）
      const keepTemp = process.env.KEEP_TEMP_FILES === '1';
      if (keepTemp) {
         console.warn(`${logPrefix} ⚠️ KEEP_TEMP_FILES=1: Will NOT clean up ${tempDir}`);
      }

      // 检查 Poppler（仅一次，可选）
      // await checkPopplerAvailable(); // 取消注释可验证

      // 转换
      const images = await convertPdfToImages(inputPath, outputDir, 150); // 先用 150 DPI 调试

      ctx.status = 200;
      ctx.body = { message: '文件转换成功', data: { images } };

   } catch (error) {
      // 统一日志
      const errMsg = error.message || 'Unknown error';
      ctx.logger?.error(`${logPrefix} pdfToImage failed: ${errMsg}`, { error });
      console.error(`${logPrefix} 🚨 HANDLED ERROR:`, errMsg);

      ctx.status = 400;
      ctx.body = { message: errMsg };

   } finally {
      // 清理临时目录（除非调试）
      //  if (tempDir && process.env.KEEP_TEMP_FILES !== '1') {
      //    fsPromises.rm(tempDir, { recursive: true, force: true })
      //      .then(() => console.log(`${logPrefix} 🧹 Cleaned up temp dir: ${tempDir}`))
      //      .catch(err => console.warn(`${logPrefix} Failed to clean temp dir:`, err.message));
      //  }
   }
};