1173132 - "Failed to load article from page" on Breitbart.com

Status: RESOLVED WORKSFORME

(Toolkit :: Reader Mode, defect)

Description

[Mark Capella [:capella]]

On (all?) Breitbart pages (ex: [0]) switching to ReaderView fails.

Failure also occurs on desktop. It was working before I pulled and rebuilt from m-c just now, so this must be a fairly recent regression.

Interestingly, on Android, if I swipe away the app while displaying the error page, then restart FF with tabs=restore always, the page is displayed properly in ReaderView.


[0] http://www.breitbart.com/big-government/2015/06/09/exclusive-obamatrade-support-falling-apart-in-house-despite-leadership-push-123-against-and-counting/

Comment 1

[Mark Capella [:capella]]

Correction, desktop toggling in and out of ReaderView is intermittent fail.

Comment 2

[Mark Capella [:capella]]

Few more minutes to dig, and found LOGCAT:

I/Gecko   (19612): *************************
I/Gecko   (19612): A coding exception was thrown and uncaught in a Task.
I/Gecko   (19612): 
I/Gecko   (19612): Full message: TypeError: this._doc.documentElement is undefined
I/Gecko   (19612): Full stack: Readability.prototype.parse@resource://gre/modules/reader/Readability.js:1770:9
I/Gecko   (19612): Agent.parseDocument@resource://gre/modules/reader/ReaderWorker.js:48:12
I/Gecko   (19612): worker.dispatch@resource://gre/modules/reader/ReaderWorker.js:21:24
I/Gecko   (19612): anonymous/AbstractWorker.prototype.handleMessage@resource://gre/modules/workers/PromiseWorker.js:122:16
I/Gecko   (19612): @resource://gre/modules/reader/ReaderWorker.js:35:41
I/Gecko   (19612): 
I/Gecko   (19612): *************************
E/GeckoConsole(19612): [JavaScript Error: "Error in ReaderWorker: TypeError: this._doc.documentElement is undefined" {file: "resource://gre/modules/ReaderMode.jsm" line: 357}]

Comment 3

[:Margaret Leibovic]

(In reply to Mark Capella [:capella] from comment #0)

> Interestingly, on Android, if I swipe away the app while displaying the
> error page, then restart FF with tabs=restore always, the page is displayed
> properly in ReaderView.

This inconsistency is due to the fact that during session restore, we use an XHR to download the raw article content, rather than trying to parse the DOM that's already loaded.

Comment 4

[:Gijs (he/him)]

(In reply to Mark Capella [:capella] from comment #2)
> Few more minutes to dig, and found LOGCAT:
> 
> I/Gecko   (19612): *************************
> I/Gecko   (19612): A coding exception was thrown and uncaught in a Task.
> I/Gecko   (19612): 
> I/Gecko   (19612): Full message: TypeError: this._doc.documentElement is
> undefined
> I/Gecko   (19612): Full stack:
> Readability.prototype.parse@resource://gre/modules/reader/Readability.js:
> 1770:9
> I/Gecko   (19612):
> Agent.parseDocument@resource://gre/modules/reader/ReaderWorker.js:48:12
> I/Gecko   (19612):
> worker.dispatch@resource://gre/modules/reader/ReaderWorker.js:21:24
> I/Gecko   (19612):
> anonymous/AbstractWorker.prototype.handleMessage@resource://gre/modules/
> workers/PromiseWorker.js:122:16
> I/Gecko   (19612): @resource://gre/modules/reader/ReaderWorker.js:35:41
> I/Gecko   (19612): 
> I/Gecko   (19612): *************************
> E/GeckoConsole(19612): [JavaScript Error: "Error in ReaderWorker: TypeError:
> this._doc.documentElement is undefined" {file:
> "resource://gre/modules/ReaderMode.jsm" line: 357}]

This means the parsing of the document with JSDOMParser failed.

Comment 5

[Mark Capella [:capella]]

This seems to be working now \o/ The mercurial log suggests bug 1182778 might be the reason.

(I can't review the bugzilla for some reason I get: |You are not authorized to access bug 1182778.|)

But a quick read of the changeset looks promising:
https://hg.mozilla.org/mozilla-central/rev/9fc0f0ba0cca

Comment 6

[:Gijs (he/him)]

(In reply to Mark Capella [:capella] from comment #5)
> This seems to be working now \o/ The mercurial log suggests bug 1182778
> might be the reason.
> 
> (I can't review the bugzilla for some reason I get: |You are not authorized
> to access bug 1182778.|)

It's a security-sensitive bug.

> But a quick read of the changeset looks promising:
> https://hg.mozilla.org/mozilla-central/rev/9fc0f0ba0cca

This works on 40 for me... I suspect the page markup has just changed, but I can't tell for sure.

Comment 7

[Mark Capella [:capella]]

oic ... well, as the issue no longer seems to exist, I'll close out my original report "wfm"