I have a web service that accepts really huge files. Usually in the range of 10 - 15 GB (not MB).
However upload using a browser is only possible using Chrome on Linux. All 3 major browsers have different flaws trying to upload such a file:
Internet Explorer stops after exactly 4GB.
Firefox does not start at all.
Chrome (on Windows) transfers the whole file but fails to send the closing bondary (send 0xff instead).
Now we are searching for a way to get uploads to work. Preferably using HTML/JS only, but I see no way to make that happen. Second try would be flash, but FileReference seems to break for files > 4GB. Last way would be Java but that is not what we are looking for in the browser client.
Note that this is about the client. I know that the server side code works, as I can upload a 12GB file using standard HTML-Upload with Chrome on Linux. It is the only browser/os combination that works so far, but therefor I am sure, the server coode is fine.
Does anyone know any way to get huge file uploads to work?
Regards,
Steffen
There is a fairly young JS/HTML5 API which might cover your user case:
https://developer.mozilla.org/en-US/docs/Using_files_from_web_applications
I can't speak to its suitability though.
If you're using IIS, the default max file upload is 4GB. You need to change this in your script or your server settings.
See: Increasing Max Upload File Size on IIS7/Win7 Pro
Normally you would break and upload such files in chunks using stream upload. If you take a limited amount data of the file, upload that part to the server, server appends data to the file. Repeat till complete file is uploaded. You can read a file in parts using FileStream (update: Adobe AIR only) or with javascript using the experimental Blob/FileReader. I guess this could be a workaround for the problem.
I think this solution could help:
AS3 fileStream appears to read the file into memory
Let me know if this works out, its an interesting problem.
Related
I hope you can help me.
I am writing a desktop program that will run in a web browser (in HTML/CSS/javascript in case that wasn't clear). It will be entirely disconnected from the internet and obtain files and data using only FILE protocol. My question is: how can you obtain a listing of the contents of a directory referenced this way?
I've been searching for months and really turned up almost nothing! Maybe I just don't know how to search but there doesn't seem to be much information about how browsers actually deal with File protocol.
For example, when you open a directory in Chrome, it gives you a nice table with hyperlinks of all the parent directory's children. However, when you look at the source code, it's as if Chrome just magically knew exactly what files were in the directory. I feel that if I could understand how it knew that, maybe I'd be able to get somewhere...
Also, I am open to other ideas about how to get a directory listing. I've read about being able to do it with php but it requires running a server. Does anyone know if it is possible to run php code with File protocol rather than HTTP?
Thanks for reading this far and truely any information that could remove me from this standstill is appreciated!
Web Apps do not have access to the user's file system so you will not be able to do what the chrome file browser does with a web app. I believe chrome is using some sort of native code to do this.
I would recommend trying something a little more on the native-side. A chrome app will let you use html, css, javascript while also allowing you access to the file system. https://developer.chrome.com/apps/app_storage#filesystem
Another alternative is you could write some sort of native java application. That would allow you to read/write all the files you want.
I have a 6MB binary file in my AppEngine backed app. I need to get this file to my GWT client to do many computations. I'm trying to avoid doing the computations on the server to save my instance hours quota. But this means I have to get the file to my client somehow.
I also don't want to download it to the app each time the app is opened, that would hurt my appengine outgoing bandwidth quota.
It would be nice to download it once, then have the client somehow store it for use the next time the app is opened.
How to best do this?
A ClientBundle? Use HTML5 Storage? Is this sort of thing just not done, and should I just do the computations on the server and send the result to the client?
The most easiest way to permanently cache the file is using application cache.
Another ways is HTML5 Storage (localStorage), but there is 2.5 MB limit. To use more than that amount, you have use IndexedDB (Chrome, Firefox, IE10) and WebSQL (Safari, Opera). A wrapper library like YDN-DB help in this case.
You can use DataResource:
https://developers.google.com/web-toolkit/doc/latest/DevGuideClientBundle#DataResource
The only problem you face is the file size. Different browsers have different limits - I've seen 5MB and 8MB numbers. If you can compress your file to be under 5MB, you should be fine with most browsers. (By the way, it's an enormous amount of data if it can't be compressed any further.)
Also, you can detect a mobile browser (or offer a different URL for mobile version), and do computations on the server for mobile users.
Can I upload file more than 2gb size in 32bit browser using HTML5 or not ?
It's very unlikely that you can upload a file that big via HTTP. Not only is unlikely that you can, it is also a terrible idea because before the server saves the file to disk, it will read the entire content in memory and chances are that something else will happen before it gets to write to disk (it'll run out of memory, for example).
If you want to upload a file like this, use sftp.
always uploading the total file isn't the good idea, so upload large file as multipart and at the same time we need high performance, for that use webworkers to send and get requests from server. I hope this should work.
I have an application that people use through Remote Desktop/Terminal Server. The application supports digital signatures. Well, the digital signature pad is on the client, but the program runs on the server. The signature pad also does not support being shared as a device through Remote Desktop(not listed with "Supported Plug And Play Devices" in local resources).
What is the best way of being able to send the signature to the server from the client machine? Preferably with having the least amount of setup for the users(there are a lot of clients and a fair amount of servers this must be done for)
My best idea so far is sharing the clipboard and using it to send messages from server to client(with the client application "polling" the clipboard for a special clipboard format) I feel like this may not be very fast or stable though as I don't think Remote Desktop was designed for it.
Also, we are open to [reasonable] language choices like C/C++, C#, Delphi(the application is written in this), etc. Also, the signature pad is a Topaz TS460(connects by USB).
Can anyone give me ideas on how this can be done or if the clipboard idea of mine is probably the best?
tl;dr: What is the best way of sending an image from a client to a server through remote desktop?
Update:
Well, I've done a bit of testing with plain ASCII text(I can't get files to transfer) and it seems that there is problems copying large amounts of text. I tried copying 43M of text and after a long period of waiting I just got an empty clipboard(Like it did a paste, but there was no text pasted) I was able to transfer about 2M of data though (at decent speeds) between server and client, so this may be feasible for signature images(which will be either jpeg or png compressed)
Have you looked into using Remote Desktop Virtual Channels? http://msdn.microsoft.com/en-us/library/aa383509(VS.85).aspx
for topaz signature pad and credit card swiper you will need the serial type. It will work, already tried it. but I guess this question is too old for me to answer. Does IPAD as well as well as other tablets work on terminal and citrix setup?
I have not tried with Remote Desktop, but one thing that comes to mind is installing a good macro tool on the client. AutoHotKey ( http://www.autohotkey.com/ ) is a free tool that lets you create runable scripts that can do things like open applications and send key strokes to them.
I'm not sure how well it would work with remote desktop, but I know for certain that you can easily setup a script that would launch an application, send it "key strokes" to generate data, copy the data to the clipboard, switch to another application and then paste in the data.
When AutoHotKey is installed, you have the option of associating the file types of the scripts with the app so that end users could just double click your scripts desktop icon to run it. No command line messyness for them.
If all you need to do is transfer an amount of data (a file) from the client to the server it is fairly easy. Polling for a file seems also more logical as polling via the clipboard.
When you connect the client should enable sharing a harddisk (at least one). You can specify the options every time you connect, or you can send the client a .RDP file that is preconfigured.
If you can get the user to put the file on a fixed position, you can access the file C:\Shared \File.jpg using a path like \tsclient\c\Shared\File.jpg.
Here's an explanation (with nice screenshot) how to copy files with Remote Desktop:
http://www.jakeludington.com/ask_jake/20051218_copying_files_with_remote_desktop.html
I wasn't sure if your question rules already out this approach or not.
I'm writing a flash app using the open source tools. I would like to load a data file in to the app and capture a screenshot of the stage on the server. The only part that seems mysterious is running the app on the server. In fact, I don't even care if it's the same app running on the server and in the browser--if I can use the flash stage and drawing routines to produce an image server-side, I'm happy. If I have to delve in to flex, fine. Right now I'm having problems finding any starting point at all.
I gather Adobe has some commercial products that may fit the bill, but I'd like to stick with open source, apache, and linux. I know this is probably possible with haxe/neko, but I'd like to use more mainstream tools if possible. Am I asking too much?
EDIT/CLARIFICATION: Many thanks for the responses so far, but I think I've been a bit muddy in my description. I've already written the actual stage-grabbing stuff using the same PNGEncoder class as was suggested. The problem is in actually running the swf on the server side. I don't want to let the client take the screen shot itself, because this opens up the possibility of the client maliciously submitting a screenshot which does not correspond to what is on the stage, that is, I don't want users uploading porn. If I could run the the actionscript code on the server, then I could generate the screenshot from my data files and be sure that the screenshot matches the data, but I have no idea how to run the actionscript or swf on the server.
Swfs run on a client computer, not on the server. The only way it would run on the server would be if you set up a special environment on your server so that it ran a web browser, opened up the page and ran the swf. But even then it would have no correlation to what an external user was doing.
You'll need to run it client side. As far as your security concerns, the best way to get rid of those is to have the php writing the actual image only accept an encrypted form of the image file, which the flash can encrypt. That way they can't simply use the PHP file to upload whatever image they want unless they happened to encrypt it the exact same that your swf did. Next encrypt the swf itself (I recommend SWF Shield) so that a potential hacker cannot read the code to know how to encrypt the image.
We just completed a similar project where we rendered JPGs from SWFs that loaded dynamic data, we used IECapt
Did you try actionscript print commands?
Try and look at this:
http://www.phpclasses.org/browse/package/4312.html
I know this question is long dead, but I had a similar problem and ended up writing a script using applescript + ui scripting to grab the inside area of the preview window of the standalone flash player in OS X. You can grab it off github here.
How about swfdec-thumbnailer from the swfdec-gnome package? It's used to create thumbnails of SWF files but can accept arbitrarily large resolutions with the -s argument.
EDIT: swfdec-gnome has been deprecated in Ubuntu 10.10 in favour of Gnash. Here is a guide on taking screenshots with Gnash (note that certain features like gradients are not yet properly supported).