Question

我正在创建一个npm模块，使用chunk upload在facebook api上传视频。我想知道是否应该使用节点流目前我这样做：
1.请求api获得块的大小。
2.使用unix split cmd
分割原始视频 3.按块发送块 4.删除所有块

我使用流进行一些测试，输出块看起来很好。

var readable = fs.createReadStream('video.mp4');
var i = 0;
readable.on('readable', () => {
  var chunk;
  while(null !== (chunk = readable.read(1048576))) {
    var fd = fs.openSync('chunk' + i++, 'w');
    fs.writeSync(fd, chunk, 0, chunk.length);
  }
});

我应该使用Stream进行重构吗？这是指向当前git repo的链接谢谢你的时间。

Answer 1

更大的问题是“为什么”？你肯定可以简单地传播这个，但是你应该吗？如果您的代码现在令人满意，也许您不应该触摸它。

另一方面，如果您使用request-promise，则请求已支持流式传输 - 与require('fs').readFileStream().pipe(request(options))中一样;也许您可以简单地整体播放视频。

流式传输文件可以减少整个模块的内存占用，在某些情况下甚至可能更快地完成上传，但实际上取决于多个方面。试着看看它是怎么回事。

顺便说一句，您的示例使用writeSync。为何同步？我想只是为了测试它是否做得对吗？

Answer 2

以下是我一年前使用Facebook Graph API在Node Js上传视频的代码：

// It publishes the video in chunks to Facebook until it finishes
var publishVideo = function(access_token, text, file) {

  var stats = fs.statSync(file.path);

  var formData = {
    access_token: access_token,
    upload_phase: 'start',
    file_size: stats.size
  };

  // First step, we send the video size
  request.post({ url: 'https://graph-video.facebook.com/v2.3/me/videos', form: formData, json: true },
  function(err, response, body) {
    if (!err) {
      // With the response, we start making the video transfer
      transferProcess(undefined, file, access_token, body, function(err, currentUploadSession) {
        if (!err) {

          var formData = {
            access_token: access_token,
            upload_phase: 'finish',
            upload_session_id: currentUploadSession,
            description: text
          };

          // Once the video transfer ended, we publish the video
          request.post({ url: 'https://graph-video.facebook.com/v2.3/me/videos', form: formData, json: true });
        }
      });
    }
  });
};

// It processes each part of the video until it finishes
var transferProcess = function(uploadSession, file, access_token, body, callback) {

  // First we generate a copy of the file in order to be independent to the original file
  // because it can have problems when opening it at the same time from other file
  var copyFileName = file.path + '-facebook';
  fse.copySync(file.path, copyFileName);

  // Once we have the copy, we open it
  var fd = fs.openSync(copyFileName, 'r');

  var bytesRead, data, bufferLength = 1000000000;
  var buffer = new Buffer(bufferLength);

  var length = body.end_offset - body.start_offset;

  // We read the amount of bytes specified from body.start_offset until length
  bytesRead = fs.readSync(fd, buffer, body.start_offset, length, null);
  data = bytesRead < bufferLength ? buffer.slice(0, bytesRead) : buffer;

  // We generate a file with the recently read data, and with a name of copyFileName-chunked-12313123
  var chunkFileName = copyFileName + '-chunked-' + body.start_offset;

  // We create the file so then we can read it and send it
  fs.writeFile(chunkFileName, data, function(err) {
    if (err) {
      callback(err);
    }
    else {

      var currentUploadSession = uploadSession ? uploadSession : body.upload_session_id;
      var startOffset = parseInt(body.start_offset);

      var formData = {
        upload_phase: 'transfer',
        start_offset: startOffset,
        upload_session_id: currentUploadSession,
        access_token: access_token
      };

      formData.video_file_chunk = fs.createReadStream(chunkFileName);

      // Once we have the file written, we upload it
      request.post({ url: 'https://graph-video.facebook.com/v2.3/me/videos',
      formData: formData, json: true }, function (err, response, body) {
        // If there was an error, we return it
        if (err || body.error) {
          callback(err ? err : body.error, null);
        }
        // If the lecture of the file has ended, facebook will send us the body.start_offset and the body.end_offset,
        // if they are the same, it means that we have ended uploading the video, so we return
        else if (body.start_offset === body.end_offset) {
          callback(err, currentUploadSession);
        }
        // Else, we keep reading the file
        else {
          transferProcess(currentUploadSession, file, access_token, body, callback);
        }
      });
    }
  });
};

希望你能用它。

乔尔

以chunks api call

2 个答案: