Fwd: Rest - Ajax request - build file

Okay, I dropped the idea of cut the file.

One solution (not tested yet):  ZnConstants maximumEntitySize: 104857600 (100 * 1024 * 1024) in order to increase the size of incoming entities to 100 M for example.

But, if we find one solution for the other case, we could upload file faster.

Happy smalltalk ;)

2015-03-12 15:21 GMT-10:00 Sebastien Audier <[hidden email]>:

Hi all,

I work on pharo 2.0 and seaside-REST with ZincServer.

I need to rebuild a file which is uploaded by POST requests through Seaside-REST but I don't really use WASession. 

On javascript layer, I cut the file with this function:

document.querySelector('input[type="file"]').addEventListener('change', function(e) {
  var name = this.files.name;
  var blob = this.files[0];

  const BYTES_PER_CHUNK = 1024 * 1024; // 1MB chunk sizes.
  const SIZE = blob.size;

  var start = 0;
  var end = BYTES_PER_CHUNK;

  while(start < SIZE) {
    upload(name, blob.slice(start, end));

    start = end;
    end = start + BYTES_PER_CHUNK;
  }
}, false);

And I send datas with this function:

function upload(aFileName, files) {
  var formData = new FormData();

  for (var i = 0, file; file = files[i]; ++i) {
    formData.append(aFileName, file);
  }

  var xhr = new XMLHttpRequest();
  xhr.open('POST', '/myPath', true);
  xhr.onload = function(e) { ... };

  xhr.send(formData);  // multipart/form-data
}

I have a POST method on my handler, but I receive multiple requests.
On this way, I don't have access to the requestContext in order to get the WAFile in request. 

If I don't cut the file, it works and I can store on disk. But not in the other case.

Of course, my goal is to store the file on disk correctly.

So,
How to get file content of each request ?
How  to rebuild and store the file correctly if the requests aren't ordored ?

Any help ?

Thank's a lot.

--

Sébastien AUDIER

--

Sébastien AUDIER

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Hi Bob, Thanks for your answer.

Right, but before this problem, I failed to recover the file content through the request. 

If I don't slice the file, I can recover a WAFile instance by:

| file |
file := self requestContext request postFields values first...
And I stream on disk after that.

If I slice the file on web client, I can't recover the WAFile, because: 

self requestContext  -> return an error. 

What's wrong in my process ?

Thanks again,

2015-03-19 15:02 GMT-10:00 Bob Arning <[hidden email]>:

Not really sure I understand the question, but it sounds like you would need to include some sort of position information in each POST. That way the receiving end knows where in the file to put the data.

On 3/19/15 8:42 PM, Sebastien Audier wrote:

Okay, I dropped the idea of cut the file.

One solution (not tested yet):  ZnConstants maximumEntitySize: 104857600 (100 * 1024 * 1024) in order to increase the size of incoming entities to 100 M for example.

But, if we find one solution for the other case, we could upload file faster.

Happy smalltalk ;)

2015-03-12 15:21 GMT-10:00 Sebastien Audier <[hidden email]>:

Hi all,

I work on pharo 2.0 and seaside-REST with ZincServer.

I need to rebuild a file which is uploaded by POST requests through Seaside-REST but I don't really use WASession. 

On javascript layer, I cut the file with this function:

document.querySelector('input[type="file"]').addEventListener('change', function(e) {
  var name = this.files.name;
  var blob = this.files[0];

  const BYTES_PER_CHUNK = 1024 * 1024; // 1MB chunk sizes.
  const SIZE = blob.size;

  var start = 0;
  var end = BYTES_PER_CHUNK;

  while(start < SIZE) {
    upload(name, blob.slice(start, end));

    start = end;
    end = start + BYTES_PER_CHUNK;
  }
}, false);

And I send datas with this function:

function upload(aFileName, files) {
  var formData = new FormData();

  for (var i = 0, file; file = files[i]; ++i) {
    formData.append(aFileName, file);
  }

  var xhr = new XMLHttpRequest();
  xhr.open('POST', '/myPath', true);
  xhr.onload = function(e) { ... };

  xhr.send(formData);  // multipart/form-data
}

I have a POST method on my handler, but I receive multiple requests.
On this way, I don't have access to the requestContext in order to get the WAFile in request. 

If I don't cut the file, it works and I can store on disk. But not in the other case.

Of course, my goal is to store the file on disk correctly.

So,
How to get file content of each request ?
How  to rebuild and store the file correctly if the requests aren't ordored ?

Any help ?

Thank's a lot.

--

Sébastien AUDIER

--

Sébastien AUDIER

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

--

Sébastien AUDIER

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

2015-03-19 15:46 GMT-10:00 Bob Arning <[hidden email]>:

sounds like the sliced and non-sliced versions are posting in somewhat different ways.

Yes, sounds like the sliced version is posting in different format. And may be, Zinc doesn't handle correctly.

 

does the sample code below work ok if the data is less than BYTES_PER_CHUNK? IOW, is the code ok for a single chunk POST, but failing for multiple chunks? 

Not only multiple chunks. Even if the file is smaller than the BYTES_PER_CHUNK limit, it doesn't work.  It doesn't work with just one request. 

 

 

What does the receiving code look like? Is it different for the sliced and non-sliced versions?

No it's the same method.

 

Can you say more about what "return an error" is? Is there a debugger stack you could include?

Okay, this is my post method on the handler.

test

<post>

<produces: 'application/json; charset=utf-8'>

| uploadedFile disk file | 

uploadedFile := self requestContext request postFields values first.

        "here, uploadedFile is a WAFile if we don't slice the file"

[disk := (FileSystem disk workingDirectory / 'ressources' / 'upload').

disk ensureDirectory. 

file := ((disk /  (uploadedFile fileName)) ensureFile).

file writeStreamDo: [:str |

str

nextPutAll: (uploadedFile rawContents)]] fork.

^''

when I slice the file, the self requestContext return a signal does not undestood.

Did you need more informations ?

Thanks again,

--

Sébastien AUDIER

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside