user69260 user69260 - 4 months ago 15
Javascript Question

Unzipping files

I want to display OpenOffice files, .odt and .odp at client side using a web browser.

These files are zipped files. Using Ajax, I can get these files from server but these are zipped files. I have to unzip them using JavaScript, I have tried using inflate.js, http://www.onicos.com/staff/iz/amuse/javascript/expert/inflate.txt, but without success.

How can I do this?

Answer Source

I wrote an unzipper in Javascript. It works.

It relies on Andy G.P. Na's binary file reader and some RFC1951 inflate logic from notmasteryet. I added the ZipFile class.

working example:
http://cheeso.members.winisp.net/Unzip-Example.htm (dead link)

The source:
http://cheeso.members.winisp.net/srcview.aspx?dir=js-unzip (dead link)

NB: the links are dead; I'll find a new host soon.

Included in the source is a ZipFile.htm demonstration page, and 3 distinct scripts, one for the zipfile class, one for the inflate class, and one for a binary file reader class. The demo also depends on jQuery and jQuery UI. If you just download the js-zip.zip file, all of the necessary source is there.


Here's what the application code looks like in Javascript:

// In my demo, this gets attached to a click event.
// it instantiates a ZipFile, and provides a callback that is
// invoked when the zip is read.  This can take a few seconds on a
// large zip file, so it's asynchronous. 
var readFile = function(){
    $("#status").html("<br/>");
    var url= $("#urlToLoad").val();
    var doneReading = function(zip){
        extractEntries(zip);
    };

    var zipFile = new ZipFile(url, doneReading);
};


// this function extracts the entries from an instantiated zip
function extractEntries(zip){
    $('#report').accordion('destroy');

    // clear
    $("#report").html('');

    var extractCb = function(id) {
        // this callback is invoked with the entry name, and entry text
        // in my demo, the text is just injected into an accordion panel.
        return (function(entryName, entryText){
            var content = entryText.replace(new RegExp( "\\n", "g" ), "<br/>");
            $("#"+id).html(content);
            $("#status").append("extract cb, entry(" + entryName + ")  id(" + id + ")<br/>");
            $('#report').accordion('destroy');
            $('#report').accordion({collapsible:true, active:false});
        });
    }

    // for each entry in the zip, extract it. 
    for (var i=0; i<zip.entries.length;  i++) {
        var entry = zip.entries[i];

        var entryInfo = "<h4><a>" + entry.name + "</a></h4>\n<div>";

        // contrive an id for the entry, make it unique
        var randomId = "id-"+ Math.floor((Math.random() * 1000000000));

        entryInfo += "<span class='inputDiv'><h4>Content:</h4><span id='" + randomId +
            "'></span></span></div>\n";

        // insert the info for one entry as the last child within the report div
        $("#report").append(entryInfo);

        // extract asynchronously
        entry.extract(extractCb(randomId));
    }
}

The demo works in a couple of steps: The readFile fn is triggered by a click, and instantiates a ZipFile object, which reads the zip file. There's an asynchronous callback for when the read completes (usually happens in less than a second for reasonably sized zips) - in this demo the callback is held in the doneReading local variable, which simply calls extractEntries, which just blindly unzips all the content of the provided zip file. In a real app you would probably choose some of the entries to extract (allow the user to select, or choose one or more entries programmatically, etc).

The extractEntries fn iterates over all entries, and calls extract() on each one, passing a callback. Decompression of an entry takes time, maybe 1s or more for each entry in the zipfile, which means asynchrony is appropriate. The extract callback simply adds the extracted content to an jQuery accordion on the page. If the content is binary, then it gets formatted as such (not shown).


It works, but I think that the utility is somewhat limited.

For one thing: It's very slow. Takes ~4 seconds to unzip the 140k AppNote.txt file from PKWare. The same uncompress can be done in less than .5s in a .NET program. EDIT: The Javascript ZipFile unpacks considerably faster than this now, in IE9 and in Chrome. It is still slower than a compiled program, but it is plenty fast for normal browser usage.

For another: it does not do streaming. It basically slurps in the entire contents of the zipfile into memory. In a "real" programming environment you could read in only the metadata of a zip file (say, 64 bytes per entry) and then read and decompress the other data as desired. There's no way to do IO like that in javascript, as far as I know, therefore the only option is to read the entire zip into memory and do random access in it. This means it will place unreasonable demands on system memory for large zip files. Not so much a problem for a smaller zip file.

Also: It doesn't handle the "general case" zip file - there are lots of zip options that I didn't bother to implement in the unzipper - like ZIP encryption, WinZip encryption, zip64, UTF-8 encoded filenames, and so on. (EDIT - it handles UTF-8 encoded filenames now). The ZipFile class handles the basics, though. Some of these things would not be hard to implement. I have an AES encryption class in Javascript; that could be integrated to support encryption. Supporting Zip64 would probably useless for most users of Javascript, as it is intended to support >4gb zipfiles - don't need to extract those in a browser.

I also did not test the case for unzipping binary content. Right now it unzips text. If you have a zipped binary file, you'd need to edit the ZipFile class to handle it properly. I didn't figure out how to do that cleanly. It does binary files now, too.


EDIT - I updated the JS unzip library and demo. It now does binary files, in addition to text. I've made it more resilient and more general - you can now specify the encoding to use when reading text files. Also the demo is expanded - it shows unzipping an XLSX file in the browser, among other things.

So, while I think it is of limited utility and interest, it works. I guess it would work in Node.js.