[openload] re-open #15118

[360up]

Fix 620ee87 worked for a couple of days. Now broken again.

Please re-open #15118

[IDerr]

PR incoming : #15161
seems to be related but not sure :D

[bagud]

Looks like span element holding stream URL changed ID from "streamurl" to "streamurj".

[yan12125]

Thanks @bagud. They are really boring LOL

[steebchen]

Why not simply check #video[src]? It should always work, regardless of what funny ids they think of, right?

[bagud]

Aaargh. I'm working on PHP/PhantomJS implementation of openload link extractor, but phantomjs returns completely useless streamurl, like: "Ø�ﾋV�QRS®ﾹ�ð´￘SX�ﾋA?õﾙÑyl￨eZ0� Ó¡'�±�￟�DÔ-". What could I be doing wrong? I understand this is not the place to ask this but any help would be really appreciated.

[yan12125]

@bagud: Possibly phantomjs is too old. Version 2.1.1 is known to work.

[bagud]

@yan12125 Thanks for replying. It's not that, I'm using the latest version 2.1.1.

It's frustrating because you can easily see the url with standard browser using developer tools, but when the page is rendered with phantomjs the url gets obfuscated?

I'm not python guy, but as much as I can see in your openload extractor code you don't use any additional decoding of the stream url, you just output it as plain text, right?

Also I'm setting only user-agent header, no additonal headers are required in request?

I'll be really grateful for your answer on this or anybody who would be kind enough to help out. :)

[Hrxn]

Maybe @Tithen-Firion knows the answer?

[yan12125]

Also I'm setting only user-agent header, no additonal headers are required in request?

A point. openload extraction is sensitive to user-agent. It might be better to avoid another user-agent than the default one.

Anyway pasting full verbose log helps debugging.

[Tithen-Firion]

@steebchen no because it's not in video's src until you click on specific element. Their JS code is encoded with aaencode and some obfuscator. I've got some dirty-sketch of deobfuscator but even if we use it they will change it to something else.

@bagud take a look at this. Keep in mind that video urls work only on the same IP that accessed the site.

It's the same thing over again. In past they changed how the video url is encoded. Now they change where that url is stored. Of course you can scan all elements searching for specific pattern but it's easy to add dummy ones. What we need is to simulate click event. Tried that with PhantomJS but it failed.

[steebchen]

@Tithen-Firion In my browser, it's directly filled in without clicking anything. And even if, maybe you could just click on the play button and wait for it to be filled in. (EDIT: yep, that's what you said, didn't even read your comment till the end)

[Tithen-Firion]

Tested https://openload.co/f/*/ and https://openload.co/embed/*/ in Firefox, Chrome and PhantomJS. Src not set.

As for the click part: read my previous comment till the end.

[bagud]

Thank you all for your suggestions. I managed to solve my problem checking into @Tithen-Firion code.

page.onInitialized = function() { page.evaluate(function() { delete window._phantom; delete window.callPhantom; }); };
I was missing this part in my code.

I naively assumed setting up custom user agent header would be enough to hide PhantomJS browser, but it's not. You have to manually delete those two references also (which I didn't know existed), otherwise you can be still easily identified as PhantomJS and denied, what openload obviously does check.

@Tithen-Firion I have tried to simulate click event and it fires, eg. I can see element class changes, but there must be something else preventing generation of src attribute, but I assume that you already know that :). If I come up with something I'll post it.