How to read from buffer?

[markb-trustifi]

I'm using solution from issue #121:

let pdfReader = hummus.createReader(new inputStream(bufpdf));

But receive an error: Unable to start parsing PDF file The PDF buffer is correct and I can save it on a disk as a pdf file.

[galkahana]

what's this inputStream object? you need to send either a path or an object if this type - https://github.com/galkahana/HummusJS/wiki/Custom-streams#reading-streams

[markb-trustifi]

it's PDFRStreamForBuffer from #121 issue.

[galkahana]

donnow. wanna share the script and buffer?

[markb-trustifi]

This code works:

let filestream = new hummus.PDFRStreamForFile('somefile.pdf');
let pdfReader = hummus.createReader(filestream);
let isEncrypted = pdfReader.isEncrypted();

This code doesn't work:

let buff = fs.readFileSync('somefile.pdf');
let pdfReader = hummus.createReader(new inputStream(buff));
let isEncrypted = pdfReader.isEncrypted();

Where inputStream is:

class inputStream {
    constructor(buffer) {
        this.innerBuffer = buffer;
        this.position = 0;
        this.fileSize = buffer.byteLength;
    }

    read(inAmount) {
        const previousPosition = this.position;
        this.position += inAmount;
        return [].concat(this.innerBuffer.slice(previousPosition, this.position));
    }

    notEnded() {
        return this.position < this.fileSize;
    }

    setPosition(inPosition) {
        this.position = inPosition;
    }

    setPositionFromEnd(inPosition) {
        this.position = this.fileSize - inPosition;
    }

    skip(inAmount) {
        this.position += inAmount;
    }

    getCurrentPosition() {
        return this.position;
    }
}

[galkahana]

i say there's a bug in input stream. wanna give this a try? :

function PDFRStreamForBuffer(buffer){
    this.innerArray = Array.prototype.slice.call(buffer, 0)
    this.rposition = 0;
    this.fileSize = buffer.byteLength;
}

PDFRStreamForBuffer.prototype.read = function(inAmount){
    var arr = this.innerArray.slice(this.rposition,this.rposition+inAmount);
    this.rposition += inAmount;
    return arr;
}

PDFRStreamForBuffer.prototype.notEnded = function(){
    return this.rposition < this.fileSize;
}

PDFRStreamForBuffer.prototype.setPosition = function(inPosition){
    this.rposition = inPosition;
}

PDFRStreamForBuffer.prototype.setPositionFromEnd = function(inPosition){
    this.rposition = this.fileSize-inPosition;
}

PDFRStreamForBuffer.prototype.skip = function(inAmount){
    this.rposition += inAmount;
}

PDFRStreamForBuffer.prototype.getCurrentPosition = function(){
    return this.rposition;
}

module.exports = PDFRStreamForBuffer;

[markb-trustifi]

Fixed, thanks

[ChrisGrigg]

This worked for me I was using the other version of this I found elsewhere that was in class format and that didn't work correctly.

[Frank683]

This is a working solution but unfortunately using the approach with PDFRStreamForBuffer instead of passing a local filepath to hummus.createReader() takes approx. 10 seconds vs. 0,5 seconds when reading and parsing a 16 MB PDF file with hundreds of form fields.

// fileContent is a buffer with 16 MB PDF data in it.
let rstream = new hummus.PDFRStreamForBuffer(fileContent);  // Takes 1 second
let pdfParser = hummus.createReader(rstream);
let digitalForm = new PDFDigitalForm(pdfParser);  // Takes 9 seconds

Since hummus is only a wrapper of the native c++ library I can't debug into the details but as far as I can see the function PDFRStreamForBuffer.prototype.read is called very often with inAmount = 1. So it seems slicing lots of small pieces out of a huge array takes quite some time.

As a workaround I just dump the buffer content into a temporary file and pass that to hummus. Using this approach it takes only a fraction of a second to parse the 16 MB pdf file.

This is just an additional information for anyone who came here through google like I did to find out how to pass a buffer of PDF data to hummus.