FEATURE: Ability to prevent resources from being loaded.

[morganrallen]

morganrallen@gmail.com commented:

The desire here is to be abel to cancel a resource from being loaded during page.onResourceRequested something like

page.onResourceRequested = function(request) {
if(request.url.indexOf("run.js") > -1) {
return false; // or request.cancel() or request.abort();
}
}

Disclaimer:
This issue was migrated on 2013-03-15 from the project's former issue tracker on Google Code, Issue #230.
🌟   37 people had starred this issue at the time of migration.

[ariya]

ariya.hi...@gmail.com commented:

 

 
Metadata Updates

• Label(s) removed:
  • Type-Defect
• Label(s) added:
  • Type-Enhancement

[xbeumala]

xavi.beu...@gmail.com commented:

this feature would be very helpful to minimize load times and start performing tasks earlier (i.e. avoiding Ads loading times, certaing .js , etc...)

It would be a fantastic feature

[ariya]

ariya.hi...@gmail.com commented:

Issue 339 has been merged into this issue.

[iwek]

iwokadzi...@gmail.com commented:

Hoping this gets added soon.

[ariya]

olife...@gmail.com commented:

This would be very helpful!

[ariya]

blademcc...@gmail.com commented:

whats the workaround for this?

[fritzfs]

frit...@gmail.com commented:

At this moment there's only one way, modify the createRequest function in networkaccessmanager.cpp and rebuild phantomjs. Take a look at this example, http://stackoverflow.com/questions/4575245/how-to-tell-qwebpage-not-to-load-specific-type-of-resources.
Note: I'm not a phantomjs guru so perhaps someone else can comment this.

[ariya]

blademcc...@gmail.com commented:

... what kind of donation to the project could get this as a high priority item? i might be able to arrange something.

[Rebolon]

richard....@gmail.com commented:

I also think that it's a really good feature. Actually event like onResourceRequested is only usefull for logging (or other features). But it would be interesting ot disable some request (javascript that load advertising and that may break the testsuite...).

[ariya]

nicefive...@gmail.com commented:

waiting for this feature!

[ariya]

jonas.ge...@googlemail.com commented:

Yes, that would be nice.

I have created an small patch against PhantomJS for me and added an option --ignore-host='(google.com|facebook.com)'

So PhantomJS will not load any resources from hosts that matches the above regexp.

[ariya]

seve...@gmail.com commented:

I would love this feature too. J

[ariya]

jonas.ge...@googlemail.com commented:

My hacky patch for that:

https://github.com/hggh/phantomjs/compare/ignore-host.patch

If you apply that patch, you can use this:

--ignore-host='(google.com|twitter.com)'

That's perfect for Nagios/Icinga checks with https://github.com/hggh/phantomjs-nagios to ignore social media stuff or tracking.

[ariya]

ariya.hi...@gmail.com commented:

Issue 395 has been merged into this issue.

[JamesMGreene]

james.m....@gmail.com commented:

Issue 288 has been merged into this issue.

[ariya]

ariya.hi...@gmail.com commented:

Allow to abort network requests.
09e929d599

 
Metadata Updates

• Label(s) added:
  • Component-Logic
  • Module-WebPage
• Milestone updated: Release1.9 (was: ---)
• Status updated: Fixed

[ariya]

ariya.hi...@gmail.com commented:

For completeness, there is a new example loadurlwithoutcss.js which demonstrates this.

The gist:

page.onResourceRequested = function(requestData, request) {
if ((/http://.+?.css/gi).test(requestData['url']) || requestData['Content-Type'] == 'text/css') {
console.log('The url of the request is matching. Aborting: ' + requestData['url']);
request.abort();
}
};

[JamesMGreene]

james.m....@gmail.com commented:

 

 
Metadata Updates

• CC(s) added:
  • vitaliy....@gmail.com
• Owner updated: ariya.hi...@gmail.com

[JamesMGreene]

james.m....@gmail.com commented:

For posterity, related discussion of new API: https://groups.google.com/d/topic/phantomjs/0OnJPkDAqSk/discussion

[JamesMGreene]

james.m....@gmail.com commented:

Vitaliy:
Is it possible to CHANGE the URL at that point rather than just abort the request? I know that is desired, too, e.g. Issue #539 (read through the initial comment).

I also just looked at the first example for Qt 5.0 WebKit and saw it is doing just that (but with the QWebView class, which is higher level than what PhantomJS does): http://qt-project.org/doc/qt-5.0/qtwebkit/qtwebkit-index.html

[morganrallen]

morganrallen@gmail.com commented:

This was my original desired end result also but I was thinking along the
lines of aborting the original request and adding a new script tag for the
new resource wanted.

http://ithoughtyouweretherobot.com
Metal and Wire

http://nolonelyguineapigs.com/
Wandering and Rambling

http://morglog.org
Old and Neglected.

[ariya]

ariya.hi...@gmail.com commented:

I don't see anything in http://qt-project.org/doc/qt-5.0/qtwebkit/qtwebkit-index.html which says something about changing the URL. You may have misunderstood the "delegate" term there, it's not about redirecting the URL but it's rather handling the functionality differently.

[JamesMGreene]

james.m....@gmail.com commented:

Ah, yeah, that was indeed how I interpreted it. Drat....

In reality, though, in Qt, we can create and connect arbitrary signals, right? If not, it seems like we could easily generate a signal before we call [or at the start inside of?] NetworkAccessManager#createRequest and connect another on____ signal handler for that in the WebPage object. Am I totally off-base here? I recall Vitaliy mentioned something about not being able to appropriately stop/abort a request before it is issued... if that's the case, then we'd probably have to short-circuit such requests BEFORE the createRequest method is invoked.

[JamesMGreene]

james.m....@gmail.com commented:

What if we implemented a custom QNetworkReply and set it to a terminal status in the createRequest method e.g. the first code sample on this page:
https://blogs.kde.org/2010/08/28/implementing-reusable-custom-qnetworkreply

I know an HTTP error status is not the same as aborting the request (reply) but I'd probably still prefer it for network performance over the current solution. Unless the reply object can be aborted AND returned in the createRequest method...?

[JamesMGreene]

james.m....@gmail.com commented:

Vitaliy/Ariya, any thoughts on that?

[vitallium]

vitaliy....@gmail.com commented:

I'm still looking for solution, but no luck.

[JamesMGreene]

james.m....@gmail.com commented:

Otherwise, seems like the method in the SO answer linked to earlier by Fritz (Comment #7) would work for both "aborting" (with an empty QUrl) or redirecting:
http://stackoverflow.com/questions/4575245/how-to-tell-qwebpage-not-to-load-specific-type-of-resources/4588113#4588113

[arlsr]

Usage

PhantomJS >= 1.9

// Abort all requests for URLs starting 'http://example.com/'

page.onResourceRequested = function(request, net) {
    if (request.url.indexOf('http://example.com/') === 0)
        net.abort();
};

https://github.com/ariya/phantomjs/wiki/API-Reference-WebPage