Performance issue getTextContent() method

Configuration:

• Web browser and its version: Google Chrome 78.0.3904.108 / Opera 64.0.3417.92
• Operating system and its version: macOS Majove 10.14.6
• PDF.js version: 2.0.550: No bugs 2.2.228: Bug 2.3.200: Bug
• Is a browser extension: No

PDF used 216 pages http://www.montefiore.ulg.ac.be/~boigelot/cours/oop/slides/oop.pdf

Steps to reproduce the problem: I am using PDFJS (pdfjs-dist) to render all pages in an iframe. I loop sequentially on pages:

• I use the render() method to render the current page
• I use the getTextContent() method to apply custom text highlights on the text layer

This process was quite fast using version 2.0.550.

However, since version 2.2.228, the getTextContent() method has performance issues:

• For the first 50 pages of the PDF, the method is executed at a normal speed (~100ms)
• For the following pages, the method is executed 10 times slower (~1s-3s)

I measure the execution time as follows:

let startTime;
...
.then(() =>
{
  startTime = Date.now();
  return page.getTextContent();
})
.then((textContent) =>
{
  console.log(Date.now() - startTime, pageNumber);
  ...
})

Do you have any idea what caused this performance problem on Google Chrome/Opera?

[Snuffleupagus]

I created a Fiddle where I render the pages of my PDF one at a time then add them in a div.

As mentioned in the Wiki, see https://github.com/mozilla/pdf.js/wiki/Frequently-Asked-Questions#allthepages, rendering all of the pages at once is simply never a good idea!

[timvandermeij]

Closing since this is answered. Rendering all pages is a heavy operation that, depending on the resolution, most browsers won’t handle well. This is a browser limitation; perhaps the experimental SVG back-end could provide a solution for this, but it’s not production-ready yet. Refer to https://github.com/mozilla/pdf.js/projects/2 for the tracking of that feature.