kig / datastream.js Goto Github PK

When using Chrome, I encountered range limits when using readString on large binary objects.

In order to get around these errors I modified the readString method as follows. I haven't 100% tested this but I figured I would share in case anyone else encounters these issues:

DataStream.prototype.readString = function(length, encoding) {
    if (encoding == null || encoding == "ASCII") {
        try{
            var str = "";
            var arr =  this.mapUint8Array(length == null ? this.byteLength-this.position : length);
            var length = arr.length;
            for (var i=0; i < length; i++){
                str += String.fromCharCode(arr[i])
            }
            //return String.fromCharCode.apply(null,arr);
            return str
        }
        catch(err){
            console.error("ERROR " + err.message)
        }
    } else {
        return (new TextDecoder(encoding)).decode(this.mapUint8Array(length));
    }
};

Hi,

In CSP evaluation throws error()

regards,
@gsivaprabu

Hello kig,

First of all, I would like to say that DataStream is a fundamental piece of software, a must ! Many, many thanks to you and for all your works (HTML5 Rocks !). From our point of view, we should say
the approach you have choosen is the best one.
We plan to use DataStream for validating upload of files in a production environment, and, maybe, for creating efficient storage of datas. But, we nead a little help.

1. testTypeArray routine : line 232 throws an error :
   var arr = new this[t+"Array"](ds.byteLength/elen);
   should be (perhaps)
   var arr = new this[t+"Array"]( Math.floor(ds.byteLength/elen) );
2. Map tests fails (never executed)
   in line 136 :
   if (ds.byteOffset & rarr.BYTES_PER_ELEMENT === 0) {...}
   the testing value is always false and inside routines is never executed.

Furthermore, I can't understand why you perform a this._realloc(len * N);
at the beginning of all the map functions : in my eyes they are not really necessary.

Besides 'seek' and 'isEo' api's, you could add at no charge :

• a 'skip' function (saut){return this.seek(this.position+saut);},
  and
• a 'where' function (){return this.position;} ("where".length=5 chars ! - sorry)

Best regards and thank you for the time that you will devote ourselves

for some strange reason functions incorrectly assume that arr.byteOffset is 0.

Currently, it's unclear what licence this repo is under.

Not sure how I would do this with the DataStream object.
In python / C I would do something like

class CCSDS_PriHdr_t_StreamId (my_endian):
    """Some comments"""
    _pack_ = 1
    _fields_ = [ ("packet_version_number"    , ctypes.c_ushort ,3),
                 ("packet_type"              , ctypes.c_ushort ,1),
                 ("secondary_header_present" , ctypes.c_ushort ,1),
                 ("apid"                     , ctypes.c_ushort ,11)]

This would create a 16 bit ctype structure with 4 fields of bits 3,1,1,11.
Trying to stream data from an embedded device and then decode it in a browser.

Hi,

Opera 12 is little buggy with typed arrays, BYTES_PER_ELEMENT is not defined in prototypes, so some functions in ds.js are not working. Simplest workaround (possibly the only and most simple bugfix) is to define it in prototypes like this:

if (Uint8Array.prototype.BYTES_PER_ELEMENT == undefined) {
    Uint8Array.prototype.BYTES_PER_ELEMENT = Uint8Array.BYTES_PER_ELEMENT; 
    Int8Array.prototype.BYTES_PER_ELEMENT = Int8Array.BYTES_PER_ELEMENT; 
    Uint8ClampedArray.prototype.BYTES_PER_ELEMENT = Uint8ClampedArray.BYTES_PER_ELEMENT; 
    Uint16Array.prototype.BYTES_PER_ELEMENT = Uint16Array.BYTES_PER_ELEMENT; 
    Int16Array.prototype.BYTES_PER_ELEMENT = Int16Array.BYTES_PER_ELEMENT; 
    Uint32Array.prototype.BYTES_PER_ELEMENT = Uint32Array.BYTES_PER_ELEMENT; 
    Int32Array.prototype.BYTES_PER_ELEMENT = Int32Array.BYTES_PER_ELEMENT; 
    Float64Array.prototype.BYTES_PER_ELEMENT = Float64Array.BYTES_PER_ELEMENT; 
}

This is a great utility! I'd like to parse a series of variable-size items into an array using the readStruct method. The number of elements isn't known; only the total size of the segment containing the items is known. It should read as many items as possible without exceeding the segment size. Is this possible with the current DataStream.js implementation? Also, can you add licensing information? Thanks!

Setup:
install node js
install requirejs module (sudo npm install requirejs)

Steps:
Create a file (example.js) with the following code in the same dir as your DataStream.js file:

(function(requirejs){
requirejs(['./DataStream'],function(){
var ds = new DataStream();
});
})(require('requirejs'));

run the file: node example.js

Result:
Crashes with the following error.
RangeError: byteOffset out of range.
at Object.defineProperty.set (... /DataStream.js:108:24)
at Object.DataStream (../DataStream.js:19:17)
at example.js:8:18

It seems that node cannot handle declaring a 0 byte ArrayBuffer

Workaround/Fix:
In line 19 of DataStream.js
set the minimum ArrayBuffer size to 1
Change
this.buffer = new ArrayBuffer(arrayBuffer || 0);
to
this.buffer = new ArrayBuffer(arrayBuffer || 1);