Fix v2.0.1 for Toshiba dTV #605

jozefchutka · 2016-11-24T10:29:00Z

Making shaka player working with Toshiba dTV

Mozilla/5.0 (Linux; U; Linux; ja-jp; DTV; TSBNetTV/T3A01C0.0203.0DD) AppleWebKit/536(KHTML, like Gecko) NX/3.0 (DTV; HTML; R1.0;) DTVNetBrowser/2.2 (000039;T3A01C0;0203;0DD) InettvBrowser/2.2 (000039;T3A01C0;0203;0DD) TOSHIBA dTV REGZA BROWSER 43Z700X

Toshiba has unprefixed HTMLMediaElement apis (generateKeyRequest etc.) and events (needkey etc.) while it has prefixed window.performance.webkitNow()

googlebot · 2016-11-24T10:29:04Z

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed, please reply here (e.g. I signed it!) and we'll verify. Thanks.

If you've already signed a CLA, it's possible we don't have your GitHub username or you're using a different email address. Check your existing CLA data and verify that your email is set on your git commits.
If you signed the CLA as a corporation, please let us know the company's name.

jozefchutka · 2016-11-24T10:30:20Z

I signed it!

googlebot · 2016-11-24T10:30:22Z

CLAs look good, thanks!

joeyparrish · 2016-11-28T18:32:21Z

Hi Jozef,

I see a lot of duplication here and I think we can find a better way to solve this. Let me start by clarifying some points.

The platform implements EME v0.1b, but without the webkit prefix. Correct?

The platform implements performance.now with a webkit prefix on the method. Correct?

Thanks,
Joey

jozefchutka · 2016-11-29T19:59:33Z

hi Joey,

just like you say, these apis like generateKeyRequest exists without webkit prefix. I cant say its "EME v0.1b" or something else as the available doc for the device doesnt state such information.
on the platform window.performance.webkitNow() is available while window.performance.now() is not.
also there was a third thing with FakeEvent which I had to replace by CustomEvent see around code event2.initCustomEvent('encrypted'....

The way I worked is I copied PatchedMediaKeysWebkit into PatchedMediaKeys so if you compare these files you will see they are almost identical.

joeyparrish · 2016-11-29T20:44:53Z

Yes, the fact that they are almost identical makes me think there should be a solution that avoids all this code duplication. And based on the API, it looks like the very old EME v0.1b API, which is what we polyfill on top of in patchedmediakeys_webkit.js.

I think you should create separate polyfills for each of these things.

For performance.now, it should be very simple to use performance.now = performance.now || performance.webkitNow. You can follow lib/polyfill/fullscreen.js as a template, but your performance polyfill will be much simpler than the fullscreen polyfill.

For EME, let's refactor the existing v0.1b polyfill so that it doesn't hard-code the prefix. If shaka.polyfill.PatchedMediaKeysWebkit.install took a prefix parameter, then the main MediaKeys polyfill could do this:

  if (navigator.requestMediaKeySystemAccess &&
      MediaKeySystemAccess.prototype.getConfiguration) {
    shaka.log.info('Using native EME as-is.');


  } else if (HTMLMediaElement.prototype.webkitGenerateKeyRequest) {
    shaka.log.info('Using webkit-prefixed EME v0.1b');
    shaka.polyfill.PatchedMediaKeysWebkit.install('webkit');

  } else if (HTMLMediaElement.prototype.generateKeyRequest) {
    shaka.log.info('Using unprefixed EME v0.1b');
    shaka.polyfill.PatchedMediaKeysWebkit.install('');


  } else if (window.MSMediaKeys) {
    shaka.log.info('Using ms-prefixed EME v20140218');
    shaka.polyfill.PatchedMediaKeysMs.install();
  } else {
    shaka.log.info('EME not available.');
    shaka.polyfill.PatchedMediaKeysNop.install();
  }

Something in this direction would be acceptable, but I don't want to duplicate the entire file. That will lead to bugs in the long run.

Since we do not have a Toshiba TV for development, would you be willing to work on this and revise your PR?

jozefchutka · 2016-11-30T13:10:27Z

@joey it makes sense to parametrize the existing PatchedMediaKeys. Do you also have any idea how to handle the FakeEvent problem for 'encrypted' event? I can also try to polyfill performance.now, last time there was some issue with that.
However, I have no idea of the timeline and when the update would be ready, hopefully within next few weeks...

joeyparrish · 2016-11-30T20:04:05Z

@jozefchutka, I don't really know what the issue is what FakeEvent. Can you provide some details? What's the error?

jozefchutka · 2016-11-30T21:40:45Z

using FakeEvent it throws UNSPECIFIED_EVENT_TYPE_ERR: DOM Events Exception 0: The Event's type was not specified by initializing the event before the method was called. so I replaced it with CustomEvent (copied from PatchedMediaKeysMs)

joeyparrish · 2016-12-06T23:02:56Z

Ah, I see. In onWebkitNeedKey_, we're dispatching FakeEvent on a real EventTarget, as opposed to a FakeEventTarget. That seems like a bug. I think that one instance should definitely be replaced with CustomEvent.

Let us know how it's going refactoring the polyfill, and if we can do anything else to help.

…player/pull/605/files

jozefchutka · 2016-12-12T15:42:56Z

@joeyparrish updated. see the latest PR and changes. There are some issues with polyfilling performance, it seems that it only lives for a thread or something, lets say if I do

window.performance.now = window.performance.webkitNow

it works in the next line, but it doesnt work after a second or so...

joeyparrish

This looks really good overall. If polyfilling performance.now doesn't work on your platform, then it's really not a big deal. Falling back to Date.now() is good enough. I recommended a few small changes, but otherwise I like this very much. Thank you for contributing!

joeyparrish · 2016-12-13T20:25:23Z