Recent players failing on all variants

[coletdjnz]

All variants of ecc3e9a7 are failing sig and n challenge solving.

yt-dlp/yt-dlp#16118

https://www.youtube.com/s/player/ecc3e9a7/player_ias.vflset/en_US/base.js
https://www.youtube.com/s/player/ecc3e9a7/tv-player-ias.vflset/tv-player-ias.js

[bashonly]

The next player (99f55c01) has the same issue:

https://www.youtube.com/s/player/99f55c01/player_ias.vflset/en_US/base.js
https://www.youtube.com/s/player/99f55c01/tv-player-ias.vflset/tv-player-ias.js
https://www.youtube.com/s/player/99f55c01/player_es6.vflset/en_US/base.js

[Vangelis66]

(Next) Players 56211dc2 and 6c5cb4f4 similarly affected...

https://www.youtube.com/s/player/56211dc2/player_es6.vflset/en_US/base.js

https://www.youtube.com/s/player/6c5cb4f4/player_es6.vflset/en_US/base.js

[Figim]

The next player (99f55c01) has the same issue:

https://www.youtube.com/s/player/99f55c01/player_ias.vflset/en_US/base.js https://www.youtube.com/s/player/99f55c01/tv-player-ias.vflset/tv-player-ias.js https://www.youtube.com/s/player/99f55c01/player_es6.vflset/en_US/base.js

The hash "9f4cc5e4" also no longer works.

[azihassan]

I did some digging for player some of the new players (99f55c01 and 6c5cb4f4). The descrambling code entry point can be traced with the .\.url=...(.\.url) regex, ... being the name of the wrapper function. It creates an object called g.eD and passes it the video URL, somehow this object descrambles N, and returns it when we call .get("n") on it.

BRX = function(D) {
    try {
        var X = (new g.eD(D, !0)).get("n");
        console.log("X = ", X); //added this line to dump the descrambled N value
        if (X) {
            var B = D.match(/\/n\/([^/]+)/);
            if (B && B[1] && B[1] !== X) return D.replace("/n/" + B[1], "/n/" + X)
        }
    } catch (C) {
        g.SN(C)
    }
    return D
};

The thing is that the entry point accepts an entire URL and expects the N param to be in the form of .../n/XXX/ instead of &n=XXX, but if someone finds a regex way to reliably match the g.eD call then we can just pass it the entire URL anyway and get the real N value with .get("n"). More details here though I'm still experimenting with it: azihassan/youtube-d#115 (comment)

[alphaleadership]

6c5cb4f4-tv with bun also

[solarizeddev]

According to Claude:

What changed in the new YouTube players (6c5cb4f4, 56211dc2):
YouTube has restructured how the n-function and sig-function are exposed. Previously they were standalone functions that matched the ejs patterns:

Old n-pattern: var X = [nFuncName] (single-element array) or a 1-param function with try-catch
Old sig-pattern: 3-param function with specific decodeURIComponent usage

Now both functions are embedded inside multi-param XOR-dispatched functions (like Dp with 10 params in 6c5cb4f4, or sO with 8 params in 56211dc2). Different first-param values (r/c) activate different behaviors via bitwise checks like (r&90)==r, (r|80)==r, etc.
The n-solver logic is inside a branch of Dp/sO and is called with specific XOR parameters, e.g.:

sO(1,8444,Q) — extract sig/n from format object
sO(5,8440,K) — same for live formats
Dp(b^8574,b^9872,W,Date) — the actual n-param transform

This is a fundamental architecture change by YouTube — not just a pattern tweak. The ejs extractors can't find these because they look for top-level standalone function patterns.