Change code base to improve performance of PDFJS (see the issue https://github.com/mozilla/pdf.js/issues/6493)

All,

According to the issue https://github.com/mozilla/pdf.js/issues/6493, there is a problem that relating to performance when rendering a special scanning file (CCITTFaxStream).

I used the latest version of pdf js (v1.1.469). I deployed to IIS Manager and tested it on localhost but pdf.js renders slowly. Here is detail information:

• File for testing: https://drive.google.com/open?id=0BzoKdQwBinAnXzVJcmdVZjBmbkE
• System detail: my computer’s hardware: Window 7x 64, RAM: 12 GB, Processors: Intel Core i5. I use Chrome browser and its version is 46.0.2490.80 m

I have logged time at render function of PDF.JS as follows:

render: function PDFPageProxy_render(params) {
      var stats = this.stats;
      stats.time('Overall');
      ....
    function complete(error) {
       ...
        stats.timeEnd('Rendering');
        stats.timeEnd('Overall');

        console.log(stats.toString());
      }
}

This is logging time: Page Request: 20ms Rendering: 3284ms => Overall: 3304ms

I would like to change the code base to improve performance. I have done the following things:

• I found that pdf.js takes long time at the function putBinaryImageData
• I have checked the function putBinaryImageData . This function renders an image into canvas from bytes array. Because bytes array is too big (16,843,200 bytes), this function takes about 3s to render. Thus, in order to reduce the processing time, I think the solution is to scale down the input bytes array before rendering. But it is unable to do with JavaScript.

Could you possibly advise how to improve in this case?

Regards, Huy

[timvandermeij]

The pull request above is closed; refer to https://github.com/mozilla/pdf.js/pull/7047#issuecomment-454581807 for the reasons. We’ll keep this open for tracking the issue and we can reopen/revisit the PR if the other PR does not already have enough effect for this.

[Snuffleupagus]

File for testing: https://drive.google.com/open?id=0BzoKdQwBinAnXzVJcmdVZjBmbkE

That file is no longer available, but given that https://github.com/mozilla/pdf.js/issues/6575#issue-114027248 mentions “CCITTFaxStream” let’s mark this as duplicate of #6741.