获取Word、PPT、Excel、PDF文件页数及加密校验

低保和光头哪个先来

于 2024-08-29 18:06:56 发布

阅读量70

点赞数

文章标签：前端 javascript 数据库

本文链接：https://blog.csdn.net/XZZXBC/article/details/141685135

版权

想要获取一个pdf文件的页数，有多种实现方式。可以利用pdfjs，也可以利用PDFDocument：

// 方法一：利用文件的arrayBuffer
let arrayBuffer = await file.arrayBuffer();
const pdfDoc = await PDFDocument.load(arrayBuffer, { ignoreEncryption: true });
const pageCount =  pdfDoc.getPageCount();
 
// 方法二：利用文件生成blob对象
const pdfObj = await PDFDocument.load(blob, { ignoreEncryption: true });
const pageCount = pdfObj.getPageCount();
 
// 方法三：利用文件的在线预览地址
const byteString = atob(file.base64.split(',')[1]);
const mimeString = file.base64.split(',')[0].split(':')[1].split(';')[0];
const ab = new ArrayBuffer(byteString.length);
const ia = new Uint8Array(ab);
for (let i = 0; i < byteString.length; i++) {
    ia[i] = byteString.charCodeAt(i);
}
       
const loadingTask = pdfjsLib.getDocument({
    data: ab,
    password: file.code || '',
});
 
const pdfDocument = await loadingTask.promise;
const pageCount = pdfDocument.numPages;

如果是加密文件，想要使用pdfjs方式获取文件页数，需要提前给用户弹出密码框进行密码输入及验证，如何进行密码验证呢

（office文件暂时没有密码正确性验证的方式，因此只能校验是否是加密文件，而pdf文件可以校验出是否其他权限有无加密）

import { PDFDocument  } from 'pdf-lib';
import '@/assets/pdfjs/build/pdf.js'
import '@/assets/pdfjs/build/pdf.worker.js'
import * as XLSX from 'xlsx';
import JSZip from 'jszip';
import { gaEvent } from '@/utils/gtag'
 
// 判断是否已加密
const checkIfFileEncrypted = async (file: any, isCloud: boolean) => {
  try {
    let arrayBuffer;
    if (isCloud) {
      const link = await getAccessLink(file.cloudId || file.file_id, file.name || file.file_name);
      const response = await fetch(link);
      if (!response.ok) throw new Error('Failed to fetch file');
      arrayBuffer = await response.arrayBuffer();
    } else {
      if(!file.base64) {
        arrayBuffer = await file.arrayBuffer();
      } else {
        const byteString = atob(file.base64.split(',')[1]);
        const mimeString = file.base64.split(',')[0].split(':')[1].split(';')[0];
        const ab = new ArrayBuffer(byteString.length);
        const ia = new Uint8Array(ab);
        for (let i = 0; i < byteString.length; i++) {
          ia[i] = byteString.charCodeAt(i);
        }
 
        // 创建 Blob
        const blob = new Blob([ab], { type: mimeString });
        const fileObj = blobToFile(blob, file.name);
        arrayBuffer = await fileObj.arrayBuffer();
      }
    }
 
    if (file.type === 'application/pdf' || file.file_type === 'pdf' || file.name.endsWith('.pdf')) {
      // PDF 文件的加密检查
      const result = await checkPDFFileEncryption(arrayBuffer);
      return result.isEncrypted;
    } else if (
      file.type === 'application/vnd.openxmlformats-officedocument.wordprocessingml.document' ||
      file.type === 'application/msword' ||
      file.file_type === 'docx' ||
      file.file_type === 'doc'
    ) {
      // Word 文件的加密检查
      return checkWordFileEncryption(arrayBuffer);
    } else if (
      file.type === 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet' ||
      file.type === 'application/vnd.ms-excel' ||
      file.file_type === 'xlsx' ||
      file.file_type === 'xls'
    ) {
      // Excel 文件的加密检查
      return checkExcelFileEncryption(arrayBuffer);
    } else if (
      file.type === 'application/vnd.openxmlformats-officedocument.presentationml.presentation' ||
      file.type === 'application/vnd.ms-powerpoint' ||
      file.file_type === 'pptx' ||
      file.file_type === 'ppt'
    ) {
      // PPT 文件的加密检查
      return checkPPTFileEncryption(arrayBuffer);
    } else {
      return false; // 默认不加密
    }
  } catch (error) {
    return false;
  }
};
 
// PDF 文件加密检测
const checkPDFFileEncryption = async (arrayBuffer: any) => {
  try {
    const loadingTask = pdfjsLib.getDocument({ data: arrayBuffer });
    const pdfDoc = await loadingTask.promise;
    const permissions = await pdfDoc.getPermissions();
    if (permissions) {
      return {
        isEncrypted: false,
        hasOwnerPassword: true
      };
    }
    return {
      isEncrypted: false,
      hasOwnerPassword: false
    };
  } catch (error: any) {
    if (error.name === 'PasswordException') {
      return {
        isEncrypted: true,
        hasOwnerPassword: false
      };
    }
    return {
      isEncrypted: false,
      hasOwnerPassword: false
    };
  }
};
 
// 验证 PDF 密码的函数
const verifyPDFPassword = async (file: any, password: string, isCode: boolean) => {
  try {
    let arrayBuffer;
    if (isCode) { // 云盘文件
      const link = await getAccessLink(file.cloudId || file.file_id, file.name || file.file_name);
      const response = await fetch(link);
      if (!response.ok) throw new Error('Failed to fetch file');
      arrayBuffer = await response.arrayBuffer();
    } else {
      arrayBuffer = await file.arrayBuffer();
    }
 
    const loadingTask = pdfjsLib.getDocument({
      data: arrayBuffer,
      password: password,
    });
    await loadingTask.promise;
    return true; // 密码正确
  } catch (error: any) {
    if (error.name === 'PasswordException') {
      return false; // 密码错误
    }
    return false; // 其他错误
  }
};
 
// 提示用户输入密码的函数
const promptUserForPassword = async (file: any) => {
  return new Promise((resolve) => {
    EventBus.$emit('show-password-prompt', file, (action: any, password: string) => {
      resolve({ action, password });
    });
  });
};
 
// Word 文件加密检测
const checkWordFileEncryption = async (arrayBuffer: any) => {
  try {
    const zip = await JSZip.loadAsync(arrayBuffer);
    if (zip.files['word/document.xml']) {
      return false; // 文件解压成功，说明未加密
    }
  } catch (error: any) {
    if (
      error.message.includes('End of data reached') ||
      error.message.includes('Corrupted zip') ||
      error.message.includes('Can\'t find end of central directory')
    ) {
      return true; // 文件加密或损坏
    }
  }
  return false;
};
 
// Excel 文件加密检测
const checkExcelFileEncryption = async (arrayBuffer: any) => {
  try {
    const workbook = XLSX.read(arrayBuffer, { type: 'array' });
    if (workbook.SheetNames.length > 0) {
      return false; // 文件解压成功，说明未加密
    }
  } catch (error: any) {
    if (
      error.message.includes('Encrypted') ||
      error.message.includes('File is password-protected')
    ) {
      return true; // 文件加密
    }
  }
  return false;
};
 
// PPT 文件加密检测
const checkPPTFileEncryption = async (arrayBuffer: any) => {
  try {
    const zip = await JSZip.loadAsync(arrayBuffer);
    if (zip.files['ppt/presentation.xml']) {
      return false; // 文件解压成功，说明未加密
    }
  } catch (error: any) {
    if (
      error.message.includes('Encrypted') ||
      error.message.includes('Can\'t find end of central directory')
    ) {
      return true; // 文件加密
    }
  }
  return false;
};

通过以上方式即可成功拿到进行加密校验，下面是word、PPT、Excel格式文件获取文件页数的方式：

// 获取PDF文件页数
const getPDFPageCount = async (arrayBuffer, password) => {
  const loadingTask = pdfjsLib.getDocument({ data: arrayBuffer, password });
  const pdfDocument = await loadingTask.promise;
  return pdfDocument.numPages;
};
 
// 获取Word文件页数
const getWordPageCount = async (arrayBuffer) => {
  try {
    const zip = await JSZip.loadAsync(arrayBuffer);
    if (zip.files['word/document.xml']) {
      // 解析 document.xml 获取页数（此处仅为示例，具体解析需要更复杂的处理）
      return 1; // Word文件页数暂时返回1
    }
  } catch (error) {
    console.error('Error processing Word file:', error);
  }
  return 1;
};
 
// 获取Excel文件页数
const getExcelPageCount = async (arrayBuffer) => {
  try {
    const workbook = XLSX.read(arrayBuffer, { type: 'array' });
    // 以工作表的数量作为页数
    return workbook.SheetNames.length;
  } catch (error) {
    console.error('Error processing Excel file:', error);
  }
  return 1;
};
 
// 获取PPT文件页数
const getPPTPageCount = async (arrayBuffer) => {
  try {
    const zip = await JSZip.loadAsync(arrayBuffer);
    if (zip.files['ppt/slides/']) {
      // 以幻灯片数量作为页数
      return Object.keys(zip.files['ppt/slides/']).length;
    }
  } catch (error) {
    console.error('Error processing PPT file:', error);
  }
  return 1;
};