apps are likely to be published on G Play first, then to 3rd party markets.

If an app is exclusively published on 3rd pary market, then it's likely distributed directly through apk , there's no way to identify them using a centralized way.

> then it's likely distributed directly through apk , there's no way to identify them using a centralized way.

You can identify them via md5 hash of the apk, the "versionName" attribute and the "versionCode" attribute.