"commonObj" and "opList" hash difference between Node.js and Chromium environments

The goal:

I’m trying to get the same hash of the commonObjs and opList objects ( collections ) in both the Chromium and Node.js environment. These “hashes” I’m trying to get are Sha256 hashes of the JSON stringified commonObjs and opList objects.

The issue:

The issue is that I’m getting a different hash of the opList and commonObjs objects in the two different environments ( Node.js and the Browser ). My assumption is that the structure of the objects is different.

I wish that this can be amended by a simple configuration option, but it could be something tied to the environment that can’t be helped.

Configuration:

Web browser: Chromium Version 89.0.4389.114 (Developer Build) built on Debian 10.9, running on Debian 10.9 (64-bit) OS: Debian 10.9 PDF.js version: 2.7.570 Node.js version: 10.24.1 The gulp configuration used for building the libraries in both environments: gulp minified-legacy

How to reproduce:

Try out the code in this repository: https://github.com/m-majetic/pdfjs-environment-differences

[Snuffleupagus]

The issue is that I’m getting a different hash of the opList and commonObjs objects in the two different environments ( Node.js and the Browser ). […] This is what I’m trying to do with the opList and commonObjs objects after I extract them from the page:

var opListHash = sha256(JSON.stringify(opList))
console.log("This is the opList hash", opListHash);
var commonObjHash = sha256(JSON.stringify(page.commonObjs))
console.log("This is the commonObj hash", commonObjHash);

Thank you; that’s a lot clearer and it also saves me some time (by not having to clone/run your code).

Unfortunately that’s not really a use-case that’s relevant for e.g. the viewer, and as such we can’t/won’t guarantee that it behaves identically when comparing e.g. browsers and Node.js environments. Please note that the PDF.js library is/was written primarily for usage in browsers, with support for Node.js essentially tacked on afterwards (which unfortunately shows), and we thus by default modify some options to prevent bugs specifically in Node.js environments; see https://github.com/mozilla/pdf.js/blob/master/src/display/api_compatibility.js

Most likely, the differing default options in browsers/Node.js should explain what you’re seeing. You can obviously set the options manually, when calling getDocument, such that you get the same behaviour in all environments.

[m-majetic]

Thank you !

I will try out different options and see if they make any difference.

I’d consider this issue closed.