use viewer in atom package

I’m developing an Atom package including a PDF viewer, and I’m trying to do that based on pdf.js. My package inserts an iframe into its view, and loads viewer.html into it.

As a first test, I loaded https://mozilla.github.io/pdf.js/web/viewer.html into the iframe, and that worked perfectly, both with the hardlinked pdf and with local files.

But I don’t want to depend on Internet access, so I made my package depend on pdfjs-dist.

First problem: Unlike pdf.js, pdfjs-dist does not contain viewer.html under web/. I’ve read the remark

However, we do ask if you plan to embed the viewer in your own site, that it not just be an unmodified version. Please re-skin it or build upon it.

and that that is the reason not to include it, but I honestly intend to reskin the viewer, that in itself is not a reason not to include it. Moreover, if that’s the reason, then the viewer JavaScript and CSS should not be included either.

My solution was to download viewer.html from pdf.js. I had to adjust some paths in the HTML, and add a dependency on systemjs, and download systemjs.config.js.

Unfortunately, now I’ve run into a second problem I don’t know how to fix:

pdfjs.js:23 Uncaught ReferenceError: require is not defined
    at Object.defineProperty.value (pdfjs.js:23)
    at __w_pdfjs_require__ (bootstrap:19)
    at Object.pdfjsLib (annotation_layer_builder.js:1)
    at __w_pdfjs_require__ (bootstrap:19)
    at Object.defineProperty.value (pdf_viewer.component.js:17)
    at __w_pdfjs_require__ (bootstrap:19)
    at Object.defineProperty.value (bootstrap:83)
    at bootstrap:83
    at webpackUniversalModuleDefinition (universalModuleDefinition:9)
    at universalModuleDefinition:10

Can you tell me how to fix this?

Or, is there something fundamentally wrong in how I approach this?

Configuration:

• Atom 1.41.0
• Debian GNU/Linux 10.2
• PDF.js version: pdfjs-dist 2.3.200
• Is a browser extension: no

[allefeld]

building of a custom viewer

I’m afraid building a custom viewer is a bit beyond my abilities, at least for the moment. Moreover, examples only get you so far if the API exposed by viewer.js is basically undocumented.

Is there a problem with me using viewer.html in my own project, as long as I re-skin it?

without actually building it first

As far as I understand, pdfjs-dist contains a pre-built version.

I got it working for now by a combination of scraping from https://mozilla.github.io/pdf.js/web/viewer.html and downloading select files from the official repository pdf.js. (Btw., does that mean that pdfjs-dist is unofficial?)

I ended up with the files

web/viewer.css
web/viewer.html
web/viewer.js
web/images/*
web/locale/locale.properties
web/locale/en-US/viewer.properties
build/pdf.worker.js
build/pdf.js

Is this actually everything I (could possibly) need? The two directories contain many more files…

It would be great if you could tell me which files exactly I would have to include from the repository in my code to be sure everything works.

[allefeld]

Thanks, the added text looks good!