Solution 1:

2019 modern browsers update

This is the approach I'd now recommend with a few caveats:

  • A relatively modern browser is required
  • If the file is expected to be very large you should likely do something similar to the original approach (iframe and cookie) because some of the below operations could likely consume system memory at least as large as the file being downloaded and/or other interesting CPU side effects.

fetch('https://jsonplaceholder.typicode.com/todos/1')
  .then(resp => resp.blob())
  .then(blob => {
    const url = window.URL.createObjectURL(blob);
    const a = document.createElement('a');
    a.style.display = 'none';
    a.href = url;
    // the filename you want
    a.download = 'todo-1.json';
    document.body.appendChild(a);
    a.click();
    window.URL.revokeObjectURL(url);
    alert('your file has downloaded!'); // or you know, something with better UX...
  })
  .catch(() => alert('oh no!'));

2012 Original jQuery/iframe/Cookie based approach

Bluish is completely right about this, you can't do it through Ajax because JavaScript cannot save files directly to a user's computer (out of security concerns). Unfortunately pointing the main window's URL at your file download means you have little control over what the user experience is when a file download occurs.

I created jQuery File Download which allows for an "Ajax like" experience with file downloads complete with OnSuccess and OnFailure callbacks to provide for a better user experience. Take a look at my blog post on the common problem that the plugin solves and some ways to use it and also a demo of jQuery File Download in action. Here is the source

Here is a simple use case demo using the plugin source with promises. The demo page includes many other, 'better UX' examples as well.

$.fileDownload('some/file.pdf')
    .done(function () { alert('File download a success!'); })
    .fail(function () { alert('File download failed!'); });

Depending on what browsers you need to support you may be able to use https://github.com/eligrey/FileSaver.js/ which allows more explicit control than the IFRAME method jQuery File Download uses.

Solution 2:

Noone posted this @Pekka's solution... so I'll post it. It can help someone.

You don't need to do this through Ajax. Just use

window.location="download.action?para1=value1...."

Solution 3:

You can with HTML5

NB: The file data returned MUST be base64 encoded because you cannot JSON encode binary data

In my AJAX response I have a data structure that looks like this:

{
    result: 'OK',
    download: {
        mimetype: string(mimetype in the form 'major/minor'),
        filename: string(the name of the file to download),
        data: base64(the binary data as base64 to download)
    }
}

That means that I can do the following to save a file via AJAX

var a = document.createElement('a');
if (window.URL && window.Blob && ('download' in a) && window.atob) {
    // Do it the HTML5 compliant way
    var blob = base64ToBlob(result.download.data, result.download.mimetype);
    var url = window.URL.createObjectURL(blob);
    a.href = url;
    a.download = result.download.filename;
    a.click();
    window.URL.revokeObjectURL(url);
}

The function base64ToBlob was taken from here and must be used in compliance with this function

function base64ToBlob(base64, mimetype, slicesize) {
    if (!window.atob || !window.Uint8Array) {
        // The current browser doesn't have the atob function. Cannot continue
        return null;
    }
    mimetype = mimetype || '';
    slicesize = slicesize || 512;
    var bytechars = atob(base64);
    var bytearrays = [];
    for (var offset = 0; offset < bytechars.length; offset += slicesize) {
        var slice = bytechars.slice(offset, offset + slicesize);
        var bytenums = new Array(slice.length);
        for (var i = 0; i < slice.length; i++) {
            bytenums[i] = slice.charCodeAt(i);
        }
        var bytearray = new Uint8Array(bytenums);
        bytearrays[bytearrays.length] = bytearray;
    }
    return new Blob(bytearrays, {type: mimetype});
};

This is good if your server is dumping filedata to be saved. However, I've not quite worked out how one would implement a HTML4 fallback

Solution 4:

The simple way to make the browser downloads a file is to make the request like that:

 function downloadFile(urlToSend) {
     var req = new XMLHttpRequest();
     req.open("GET", urlToSend, true);
     req.responseType = "blob";
     req.onload = function (event) {
         var blob = req.response;
         var fileName = req.getResponseHeader("fileName") //if you have the fileName header available
         var link=document.createElement('a');
         link.href=window.URL.createObjectURL(blob);
         link.download=fileName;
         link.click();
     };

     req.send();
 }

This opens the browser download pop up.

Solution 5:

1. Framework agnostic: Servlet downloading file as attachment

<!-- with JS -->
<a href="javascript:window.location='downloadServlet?param1=value1'">
    download
</a>

<!-- without JS -->
<a href="downloadServlet?param1=value1" >download</a>

2. Struts2 Framework: Action downloading file as attachment

<!-- with JS -->
<a href="javascript:window.location='downloadAction.action?param1=value1'">
    download
</a>

<!-- without JS -->
<a href="downloadAction.action?param1=value1" >download</a>

It would be better to use <s:a> tag pointing with OGNL to an URL created with <s:url> tag:

<!-- without JS, with Struts tags: THE RIGHT WAY -->    
<s:url action="downloadAction.action" var="url">
    <s:param name="param1">value1</s:param>
</s:ulr>
<s:a href="%{url}" >download</s:a>

In the above cases, you need to write the Content-Disposition header to the response, specifying that the file needs to be downloaded (attachment) and not opened by the browser (inline). You need to specify the Content Type too, and you may want to add the file name and length (to help the browser drawing a realistic progressbar).

For example, when downloading a ZIP:

response.setContentType("application/zip");
response.addHeader("Content-Disposition", 
                   "attachment; filename=\"name of my file.zip\"");
response.setHeader("Content-Length", myFile.length()); // or myByte[].length...

With Struts2 (unless you are using the Action as a Servlet, an hack for direct streaming, for example), you don't need to directly write anything to the response; simply using the Stream result type and configuring it in struts.xml will work: EXAMPLE

<result name="success" type="stream">
   <param name="contentType">application/zip</param>
   <param name="contentDisposition">attachment;filename="${fileName}"</param>
   <param name="contentLength">${fileLength}</param>
</result>

3. Framework agnostic (/ Struts2 framework): Servlet(/Action) opening file inside the browser

If you want to open the file inside the browser, instead of downloading it, the Content-disposition must be set to inline, but the target can't be the current window location; you must target a new window created by javascript, an <iframe> in the page, or a new window created on-the-fly with the "discussed" target="_blank":

<!-- From a parent page into an IFrame without javascript -->   
<a href="downloadServlet?param1=value1" target="iFrameName">
    download
</a>

<!-- In a new window without javascript --> 
<a href="downloadServlet?param1=value1" target="_blank">
    download
</a>

<!-- In a new window with javascript -->    
<a href="javascript:window.open('downloadServlet?param1=value1');" >
    download
</a>