aiquery2/api/app/utils/pdfToImages.js

const fs = require('fs');
const path = require('path');
const { execFile } = require('child_process');
const { promisify } = require('util');

const execFileAsync = promisify(execFile);

// ========== 用 pdfinfo 获取页数（保持不变） ==========
async function getPDFPageCount(pdfPath) {
  try {
    const { stdout } = await execFileAsync('pdfinfo', [pdfPath], {
      timeout: 10000,
      maxBuffer: 1024 * 1024
    });
    const match = stdout.match(/Pages:\s*(\d+)/i);
    return match ? parseInt(match[1], 10) : 1;
  } catch (error) {
    console.error(`[PDF2IMG] pdfinfo failed:`, error.message);
    return 1;
  }
}

// ========== 直接调用 pdftoppm 转换单页 ==========
async function convertPdfPageToPng(pdfPath, outputDir, pageNum, dpi = 150) {
  // 通过 -singlefile + 带页码的基础名，确保输出文件名稳定为 page-<pageNum>.png
  const expectedOutputPath = path.join(outputDir, `page-${pageNum}.png`);
  const outputBaseName = path.join(outputDir, `page-${pageNum}`); // 不包含扩展名

  try {
    await execFileAsync('pdftoppm', [
      '-png',
      '-singlefile',
      '-r', dpi.toString(),
      '-f', pageNum.toString(),
      '-l', pageNum.toString(),
      pdfPath,
      outputBaseName // 只提供基础名称
    ], {
      timeout: 30000, // 30秒超时
      windowsHide: true
    });

    if (fs.existsSync(expectedOutputPath)) {
      const size = fs.statSync(expectedOutputPath).size;
      console.log(`[PDF2IMG] Page ${pageNum} saved (${size} bytes): ${expectedOutputPath}`);
      return expectedOutputPath;
    } else {
      throw new Error(`Expected output file not found: ${expectedOutputPath}`);
    }
  } catch (error) {
    console.error(`[PDF2IMG] pdftoppm failed for page ${pageNum}:`, error.message);
    throw error;
  }
}

// ========== 主转换函数 ==========
async function convertPdfToImages(
  pdfPath,
  outputDir,
  dpi = 150,
  startPage = 1,
  endPage = null
) {
  const startTime = Date.now();

  if (!fs.existsSync(outputDir)) {
    fs.mkdirSync(outputDir, { recursive: true });
  }

  const stats = fs.statSync(pdfPath);
  console.log(`[PDF2IMG] Input PDF: ${pdfPath} (${(stats.size / 1024 / 1024).toFixed(2)} MB)`);
  
  const totalPages = await getPDFPageCount(pdfPath);
  console.log(`[PDF2IMG] Detected ${totalPages} pages`);

  const normalizedStartPage = Math.max(
    1,
    Math.min(totalPages, Number(startPage) || 1)
  );
  const normalizedEndPage = Math.max(
    normalizedStartPage,
    Math.min(totalPages, Number(endPage) || totalPages)
  );

  const imagePaths = [];
  console.log(
    `[PDF2IMG] Converting pages ${normalizedStartPage}-${normalizedEndPage} with DPI=${dpi}...`
  );

  for (let pageNum = normalizedStartPage; pageNum <= normalizedEndPage; pageNum++) {
    try {
      const imagePath = await convertPdfPageToPng(pdfPath, outputDir, pageNum, dpi);
      imagePaths.push(imagePath);
    } catch (err) {
      console.warn(`[PDF2IMG] Skipping page ${pageNum} due to error:`, err.message);
    }
  }

  const duration = Date.now() - startTime;
  const plannedPages = normalizedEndPage - normalizedStartPage + 1;
  console.log(`[PDF2IMG] ✅ Converted ${imagePaths.length}/${plannedPages} pages in ${duration}ms`);
  return imagePaths;
}

module.exports = { convertPdfToImages };