Compare commits

...
Sign in to create a new pull request.

11 commits

Author SHA1 Message Date
7d20913d56 player: preserve Deobfuscation error class through map_url wrapping
Followup to audit CRIT-1. The original audit added Deobfuscation to
the switch_client whitelist but missed that map_url re-wraps the
cipher_to_url_params error into ExtractionError::InvalidData on line
~727 — InvalidData is NOT switchable, so when a cipher stream appears
on Mobile/Desktop (where our sig fn is unavailable), the player chain
still died instead of falling through to the next client.

Caught by the full integration suite: get_player_from_client::case_2_mobile
panicked with InvalidData wrapping our "sig fn unavailable" error.
Now stays as Deobfuscation through the wrap so switch_client trips
and iOS / Tv handle the request instead.
2026-05-24 12:36:17 -07:00
1211f601df deobfuscate: silence dead_code on Deobfuscator::has_sig/has_nsig
Public API methods exposed for downstream consumers (e.g. straw can
call deobf.has_sig() to skip cipher streams without observing an Err).
The internal short-circuit uses the struct field directly so the
methods register as unused at compile time.
2026-05-24 12:21:55 -07:00
8126cc0da5 audit-fix sprint: all 13 findings (CRIT/HIGH/MED/LOW)
CRIT-1: ExtractionError::Deobfuscation is now switchable.
        Deobfuscator gains has_sig()/has_nsig() — deobfuscate_sig/_nsig
        short-circuit with a recognisable error class so cipher streams
        on the wrong client fall through to the next client in the chain
        instead of killing the whole call.

CRIT-2: Soft-failed DeobfData now caches with a 1-hour retry instead of
        living for 24h. Re-extraction kicks in automatically once YT
        rotates back to a player.js shape we recognise — no more
        wall-clock-day-of-poisoned-cache.

HIGH-1: Reporter now emits a Level::WRN `extract_deobf_soft_fail` report
        on partial extraction. straw / torttube get an artefact when
        sig/nsig regex starts missing.

HIGH-2: player_client_order branches on opts.auth. With botguard
        + authed-cookie users, Desktop is now position 2 (where their
        cookie maps to an OAuth session) instead of position 4.

HIGH-3: Android dropped from the default order. needs_po_token doesn't
        flag Android, so requests were firing unsigned and tripping
        YT's bot-check rejection — which is also not switchable.
        Re-add when a real po_token strategy lands.

MED-1: Comment in needs_deobf softened — the iOS/Android-no-deobf
        property is a current YT behaviour, not a permanent protocol.

MED-2: Cargo.toml workspace pin bumped 0.11.4 → 0.11.5 so it matches
        the package version (avoids future 0.12.x bump surprises).

MED-3: Smoke test fixture uses an isolated per-process scratch dir
        instead of the repo root, avoiding cache-race with
        tests/youtube.rs (which uses CARGO_MANIFEST_DIR and could
        wipe OAuth tokens).

LOW-1: Misleading "dead-code fallback" comment in extract_fns replaced
        with the actual behaviour description.

LOW-2: get_deobf_data uses read-then-write — concurrent player calls
        on warm cache no longer serialise on the write lock.

LOW-3: Smoke test catches IpBan via exact UnavailabilityReason match
        instead of substring "Sign in/IpBan/bot" — a real regression
        won't silently pass anymore.

LOW-4: TV smoke test now asserts !audio_streams.is_empty() too,
        matching iOS / default-order tests.

LOW-5: needs_deobf comment notes YT's historical n= experiments on
        Android — sets expectation for future review passes.
2026-05-24 12:20:14 -07:00
8d7f6b4455 release: 0.11.5 — semver-compatible version bump for fork
Use 0.11.5 instead of 0.11.4-sulkta.1 so the in-workspace
rustypipe-downloader / rustypipe-cli crates (which require
`rustypipe = ^0.11.4`) keep resolving. The original upstream rev
on codeberg is at 0.11.4; we tag this internal release as 0.11.5
to keep cargo happy without needing to bump dependents.
2026-05-24 11:58:30 -07:00
84bb666bb2 release: 0.11.4-sulkta.1 — soft-fail sig + iOS-first default order
Smoke-tested against current YT player c2f7551f (May 2026):

  test ios_player_returns_streams ........... ok
  test default_client_order_returns_streams . ok  (audio Range-GET 206 Partial Content, 1024 bytes)
  test tv_player_returns_streams ............ ok (or env-skipped on IP-banned egress)

Fork changes since upstream v0.11.4:
- client::ClientType::needs_deobf: skip player.js deobf for Android too
- client::player::player_client_order: prefer iOS first (no botguard),
  iOS/Android/Tv/Desktop (with botguard)
- deobfuscate::DeobfData::extract_fns: soft-fail sig_fn/nsig_fn extraction
  so Tv/Desktop callers keep working when YT rotates player.js to a shape
  our regex doesn't recognise — only the load-bearing sig_timestamp is
  required for the request payload
- tests/sulkta_smoke.rs: end-to-end sanity covering iOS, Tv, default-order
  and a Range-GET probe to confirm YT actually serves the audio bytes
2026-05-24 11:57:46 -07:00
947f67834a tests: smoke — switch HEAD to Range GET + iOS UA
YouTube googlevideo CDN 403s HEAD requests + 403s requests with a
non-client User-Agent. Use the iOS client UA on the probe so the CDN
treats it as the same client that requested the URL.
2026-05-24 11:56:41 -07:00
b50f04d565 tests: sulkta smoke — iOS / TV / default-order player_from_client + HEAD probe
Exercises the patched default client_order + soft-fail DeobfData
end-to-end against current YouTube. Verifies:
1. iOS player_from_client returns streams (no deobf path).
2. TV player_from_client returns streams (deobf path with soft-fail).
3. default-clients player() picks iOS primary and a returned audio
   URL HEADs to a 2xx/3xx (i.e. YouTube CDN accepts it).

Lives alongside the upstream tests/youtube.rs so we don't fork their
big snapshot-based test suite, but stays standalone so a single
`cargo test --test sulkta_smoke` exercises just the load-bearing
playback path for our consumers (straw, future torttube).
2026-05-24 11:54:18 -07:00
bda0fea193 deobfuscate: soft-fail sig_fn/nsig_fn extraction
When YouTube rotates player.js to a shape our six sig/nsig regex
patterns don't recognise (eg. c2f7551f, May 2026), the whole player
path used to die at extract_fns even for clients that don't need the
sig fn at all (iOS, Android, Tv all get pre-signed stream URLs).

Now sig_fn / nsig_fn extraction is best-effort. Only the signature
timestamp is required — every `needs_deobf` client needs sts in
the request payload, but the actual deobfuscation functions are only
consumed by map_url when a stream URL carries `&s=` or `&n=`.
On failure we log a warning and store an empty string; Deobfuscator
then skips the JS eval, and any deobfuscate_sig/deobfuscate_nsig
call will fail loudly with "sig fn unavailable" instead of crashing
the player.

Keeps the Tv fallback alive even when sig deobf regex breaks.
2026-05-24 11:53:12 -07:00
a6df2ff7f4 client: prefer iOS over Android in default order
Android-only path requires Google device attestation (po_token /
botguard signing). iOS path has neither attestation nor sig deobf
requirements, so it's the cleanest "just works" default. Keep
Android in the rotation only when botguard is wired.
2026-05-24 11:50:56 -07:00
765a90e808 client: skip player.js deobf for Android + prefer Android-first client order
YouTube's Android InnerTube path returns pre-signed stream URLs (no `s=`
cipher param, no `n=` throttling param) just like the iOS path. Mark
Android as deobf-exempt and put it first in the default player client
order so the typical playback path stops fetching player.js entirely.
Avoids the `could not extract sig fn name` failure on YouTube's newer
player.js shapes (eg. c2f7551f).

Desktop stays in the rotation behind botguard for completeness; it
will still try to deobf and may fail, but it's only consulted as a
fallback for botguard-signed sessions now.
2026-05-24 11:50:15 -07:00
a25907b494 docs: porting plan — NPE sig/nsig pipeline + globalVar indirection (M1) 2026-05-24 11:43:20 -07:00
7 changed files with 587 additions and 77 deletions

View file

@ -1,6 +1,6 @@
[package]
name = "rustypipe"
version = "0.11.4"
version = "0.11.5"
rust-version = "1.67.1"
edition.workspace = true
authors.workspace = true
@ -74,7 +74,7 @@ path_macro = "1.0.0"
tracing-test = "0.2.5"
# Included crates
rustypipe = { path = ".", version = "0.11.4", default-features = false }
rustypipe = { path = ".", version = "0.11.5", default-features = false }
rustypipe-downloader = { path = "./downloader", version = "0.3.1", default-features = false, features = [
"indicatif",
"audiotag",

View file

@ -0,0 +1,123 @@
# Porting NPE's player-JS pipeline into rustypipe
**Branch:** `kayos/m1-sig-port`
**Goal:** Replace `src/deobfuscate.rs`'s narrow regex approach with
NewPipeExtractor's full pipeline so the fork keeps working as YouTube
rotates its `player_ias.vflset/.../base.js`.
## The diagnosis
Upstream rustypipe 0.11.4 (June 2025) extracts the signature
deobfuscation function with six regex patterns aimed at the call site
(`var&&(var=SIGFN(decodeURIComponent(var)))`). On current YouTube player
`c2f7551f` (May 2026) all six miss. NewPipeExtractor master's six
patterns also miss on the same file — and NPE-master's nsig (throttling)
pipeline is openly broken (`TeamNewPipe/NewPipeExtractor#1339`, open
since 2026-02-03; the dev branch has had no sig/nsig commits in 60
days). The reason NPE *appears* to work in apps is that the
Innertube paths for Android / iOS / TV clients return stream URLs that
don't carry an obfuscated `s=` signature for most videos — sig deobf
is a fallback the typical playback path never reaches.
Two structural changes have happened since rustypipe was last cut:
1. **The sig fn call site now sometimes takes a numeric prefix arg.**
New shape: `var&&(var=SIGFN(123,decodeURIComponent(var)))`. NPE's
regex set has one pattern for this; rustypipe doesn't.
2. **YT routes literal token references through a global string array.**
Near the top of every recent `player.js`:
```js
var e="startsWith{redirector.googlevideo.com{split{...{decodeURIComponent{...".split("{")
```
Calls then reference `e[N]` instead of the literal symbol. So an
anchor like `decodeURIComponent` is no longer present at the sig-fn
call site as text — it's `e[37]` (or whatever the index is).
NPE's pipeline handles (1) but not (2). To make the fork robust we
do both.
## What we're porting
| NPE file | Rust target | Notes |
|---|---|---|
| `YoutubeSignatureUtils.java` | `src/deobfuscate.rs` (rewritten) | Sig fn name + body + helper-obj + global-var assembly |
| `YoutubeThrottlingParameterUtils.java` | new `src/deobfuscate/throttling.rs` module | nsig fn name + body + early-return fixup |
| `utils/jsextractor/JavaScriptExtractor.matchToClosingBrace` | new `src/deobfuscate/jslexer.rs` | Find a `name=function` site, walk braces until balanced |
| `YoutubeJavaScriptPlayerManager.java` | already covered by rustypipe's `cache.rs` | We keep rustypipe's cache shape but extend the cached payload to include nsig fn + global var |
## Pipeline (the desired flow)
```
player.js (string)
├── extract_sig_fn_name // 6+ regex patterns, w/ globalVar[N] retry
│ │
│ └── fall back to: // globalVar[N] indirection
│ 1. extract_global_string_array_indices()
│ 2. find N where arr[N] == "decodeURIComponent"
│ 3. re-run patterns with `(?:decodeURIComponent|globalVar\[N\])`
├── extract_sig_fn_body // lexer brace-walk, regex fallback
├── extract_global_var // var X="...".split("{") (verbatim)
├── extract_helper_obj_name // from inside fn body: [;,]NAME[..
├── extract_helper_obj_body // var NAME={...};
└── assemble:
globalVar + ";" + helperObj + ";" + deobfFn + ";" + callerFn
── eval in rquickjs ──→ deobf_sig(input) ⇒ deobf(input)
player.js (string)
├── extract_nsig_fn_name // 7 NPE patterns including arr-index variants
│ │
│ └── if array variant: resolve var NAME=[fn1,fn2,fnN]
├── extract_nsig_fn_body // lexer brace-walk
├── fixup_early_return // strip `if(typeof X==="undefined")return arg;`
└── eval in rquickjs ──→ deobf_nsig(input) ⇒ deobf(input)
```
## Milestones
| ID | Subject | Effort | Gate |
|---|---|---|---|
| M1.1 | Port `matchToClosingBrace` (clean brace walker) to `src/deobfuscate/jslexer.rs` | S | Standalone unit test against a tiny `var Wka=function(d){return /,/}/` fixture |
| M1.2 | Replace `get_sig_fn_name` with NPE's 6 patterns (including `(\d+,)decodeURIComponent`) | S | T-1 fixture is the prior-working `9216d1f7` player + new fixture `c2f7551f.js` |
| M1.3 | Add `extract_global_string_array` returning `(var_name, Vec<String>)` | S | unit test for the `var e="…".split("{")` shape |
| M1.4 | Add `extract_helper_obj_name` from fn body + `extract_helper_obj_body` | S | unit test against the `qB={w8:..,EC:..,Np:..}` style fixture |
| M1.5 | Assemble globalVar + helperObj + sigFn + caller; round-trip via rquickjs | M | the existing `t_deobfuscate_sig` test fixture passes via new code path |
| M1.6 | Add globalVar[N] indirection retry to sig fn name extraction | M | new test: a fixture where the call site uses `e[N]` instead of `decodeURIComponent` |
| M1.7 | Port nsig pipeline (`YoutubeThrottlingParameterUtils`) — 7 patterns + array-resolution + early-return fixup | M | port + run NPE's `nsig_tests` table in `tests/sig_tests.rs` |
| M1.8 | Add live integration test downloading current `player.js` and asserting round-trip end-to-end | S | `cargo test --features live -- t_update` |
| M1.9 | Bump `Cargo.toml` to `0.12.0-sulkta.1`, tag, push to `Sulkta-Coop/rustypipe` `kayos/m1-sig-port` | S | clean release |
## Not in M1 (parking lot)
- Deno / external-JS-runtime swap (yt-dlp's path; we revisit if M1
doesn't hold).
- Caching the assembled deobf code across processes (cookie-jar style
on Android).
- N-tier fallback against multiple geo `player.js` variants if YT ever
splits them.
## Why this is safe-ish to ship
NPE's pipeline is what straw v0.1.0-X currently relies on for the rare
videos that hit the sig path. Porting it 1:1 to Rust gives us a
behavioural baseline equivalent to what NPE provides — no regression
from the Java side. The globalVar[N] indirection added in M1.6 is the
forward-looking piece that handles current `c2f7551f`-style
obfuscation NPE doesn't yet handle. If M1.6 turns out unnecessary
(e.g. NPE-dev lands its own fix first), we can pull the patterns into
parity but keep our generalised resolution layer.
## Tracking
Workspace task IDs:
- `#226` parent — fork + ship the patched fork
- `#230` audit + port the sig pipeline (this milestone)
- `#231` build pipeline + crafting-table integration
When M1 lands, U-2..U-5 revival becomes a `Cargo.toml` dep flip in
`rust/strawcore/` + cherry-pick of the parked commits
(`7ff5ac79e..a13896f5e` on `Sulkta-Coop/straw`).

View file

@ -100,7 +100,17 @@ impl ClientType {
}
fn needs_deobf(self) -> bool {
!matches!(self, ClientType::Ios)
// As of YT iOS client v19.x and Android InnerTube v19.x (Dec 2024+),
// both paths return pre-signed stream URLs (no &s= cipher param, no
// &n= throttling param), so they don't need player.js deobfuscation.
// YT has experimented with serving cipher streams to iOS and n= params
// to Android in past A/B tests — if that returns, this matcher needs
// to be revisited along with the po_token strategy and a switchable
// fallback to a Tv/Desktop client.
//
// Skipping the deobf fetch here keeps the player path alive even when
// YouTube rotates player.js to a shape our extractor doesn't recognise.
!matches!(self, ClientType::Ios | ClientType::Android)
}
fn needs_po_token(self) -> bool {
@ -1272,55 +1282,87 @@ impl RustyPipe {
/// Get deobfuscation data (either from cache or extracted from YouTube's JavaScript code)
async fn get_deobf_data(&self) -> Result<DeobfData, Error> {
// Cheap read-path first: avoid serialising concurrent player calls behind
// the write lock when the cache is already fresh (the common case after
// the first request). Only escalate to a write lock on cache miss.
// (Sulkta fork audit LOW-2.)
{
let read_guard = self.inner.cache.deobf.read().await;
if let Some(data) = read_guard.get() {
return Ok(data.clone());
}
}
// Write lock here to prevent concurrent tasks from fetching the same data
let mut deobf_data = self.inner.cache.deobf.write().await;
match deobf_data.get() {
Some(deobf_data) => Ok(deobf_data.clone()),
None => {
// Only attempt to fetch deobf data every 24 hours to avoid a flood of error reports
// if the client JS cannot be parsed
if deobf_data.should_retry() {
tracing::debug!("getting deobf data");
// Recheck under the write lock — another writer may have filled the cache
// between the read drop and the write acquire.
if let Some(data) = deobf_data.get() {
return Ok(data.clone());
}
match DeobfData::extract(&self.inner.http, self.inner.reporter.as_deref()).await
{
Ok(new_data) => {
// Write new data to the cache
*deobf_data = CacheEntry::from(new_data.clone());
drop(deobf_data);
self.store_cache().await;
Ok(new_data)
}
Err(e) => {
// Try to fall back to expired cache data if available, otherwise return error
deobf_data.retry_later(24);
let res = match deobf_data.get_expired() {
Some(d) => {
tracing::warn!("could not get new deobf data ({e}), falling back to expired cache");
Ok(d.clone())
}
None => Err(e),
};
drop(deobf_data);
self.store_cache().await;
res
}
// Only attempt to fetch deobf data every 24 hours to avoid a flood of error reports
// if the client JS cannot be parsed
if deobf_data.should_retry() {
tracing::debug!("getting deobf data");
match DeobfData::extract(&self.inner.http, self.inner.reporter.as_deref()).await
{
Ok(new_data) => {
// Sulkta fork (audit CRIT-2): soft-failed DeobfData (missing
// sig_fn or nsig_fn) is cached with a much shorter freshness
// window so the next player call retries extraction soon, in
// case YouTube has rotated player.js back to a shape we
// recognise. Otherwise a single rotation could poison the
// cache for the full 24h freshness window even if YT fixed
// things minutes later.
let mut entry = CacheEntry::from(new_data.clone());
if !new_data.is_complete() {
// Force re-extraction on the next request after ~1h
// (so AddrLane workers and short-lived processes get a
// chance to recover without restarting), AND stamp
// failed_version so a library version bump also triggers
// re-extraction immediately.
entry.retry_later(1);
tracing::warn!(
"deobf data partial (has_sig={}, has_nsig={}); caching with 1h retry",
new_data.has_sig(),
new_data.has_nsig()
);
}
} else {
match deobf_data.get_expired() {
*deobf_data = entry;
drop(deobf_data);
self.store_cache().await;
Ok(new_data)
}
Err(e) => {
// Try to fall back to expired cache data if available, otherwise return error
deobf_data.retry_later(24);
let res = match deobf_data.get_expired() {
Some(d) => {
tracing::warn!(
"could not get new deobf data, falling back to expired cache"
);
tracing::warn!("could not get new deobf data ({e}), falling back to expired cache");
Ok(d.clone())
}
None => Err(Error::Extraction(ExtractionError::Deobfuscation(
"could not get deobf data".into(),
))),
}
None => Err(e),
};
drop(deobf_data);
self.store_cache().await;
res
}
}
} else {
match deobf_data.get_expired() {
Some(d) => {
tracing::warn!(
"could not get new deobf data, falling back to expired cache"
);
Ok(d.clone())
}
None => Err(Error::Extraction(ExtractionError::Deobfuscation(
"could not get deobf data".into(),
))),
}
}
}

View file

@ -245,8 +245,31 @@ impl RustyPipeQuery {
/// The order may change in the future in case YouTube applies changes to their
/// platform that disable a client or make it less reliable.
pub fn player_client_order(&self) -> &'static [ClientType] {
// iOS first — it skips player.js deobfuscation entirely (pre-signed
// stream URLs) AND doesn't require device attestation the way Android
// does. Tv is the secondary fallback (needs sig_timestamp in the
// request payload, but the soft-fail extraction keeps that piece alive
// even when sig_fn/nsig_fn extraction breaks).
//
// Android is intentionally NOT in the order: `needs_po_token` doesn't
// flag Android, so requests would fire unsigned and increasingly trip
// YouTube's "Sign in to confirm you're not a bot" — and that mapping
// becomes Unavailable{Captcha} which is not switchable. Re-add when
// a real po_token strategy for Android lands.
//
// Desktop is only consulted when botguard is wired (po_token signing
// available). For authenticated-via-cookie users on botguard
// sessions we put Desktop second so they don't walk through three
// wrong clients before reaching the one their cookie works on.
if self.client.inner.botguard.is_some() {
&[ClientType::Desktop, ClientType::Ios, ClientType::Tv]
if self.opts.auth == Some(true) {
// Authed-cookie users: prefer Desktop second (where the cookie
// actually maps to an OAuth session), Tv third (OAuth token),
// iOS first as a quick anonymous path.
&[ClientType::Ios, ClientType::Desktop, ClientType::Tv]
} else {
&[ClientType::Ios, ClientType::Tv, ClientType::Desktop]
}
} else {
&[ClientType::Ios, ClientType::Tv]
}
@ -702,10 +725,19 @@ impl<'a> StreamsMapper<'a> {
None => match signature_cipher {
Some(signature_cipher) => {
self.cipher_to_url_params(signature_cipher).map_err(|e| {
ExtractionError::InvalidData(
format!("Could not deobfuscate signatureCipher `{signature_cipher}`: {e}")
.into(),
)
// Audit follow-up to CRIT-1: keep the Deobfuscation
// error class through the wrapper so `switch_client`
// still trips when a cipher stream surfaces on a
// client whose deobf is unavailable. The previous
// `InvalidData(...)` wrapping silently demoted this
// to a non-switchable error and killed the whole
// player_from_clients chain on Mobile/Desktop sig paths.
ExtractionError::Deobfuscation(
format!(
"signatureCipher `{signature_cipher}`: {e}"
)
.into(),
)
})
}
None => Err(ExtractionError::InvalidData(

View file

@ -15,6 +15,28 @@ use crate::{
pub struct Deobfuscator {
ctx: Context,
has_sig: bool,
has_nsig: bool,
}
impl DeobfData {
/// True when both sig_fn and nsig_fn were extracted from the player.js.
/// Used by the cache layer to stamp partial extractions with a shorter
/// retry window than fully-good ones (see `get_deobf_data`).
pub fn is_complete(&self) -> bool {
!self.sig_fn.is_empty() && !self.nsig_fn.is_empty()
}
/// True when the signature deobfuscation fn was extracted successfully.
pub fn has_sig(&self) -> bool {
!self.sig_fn.is_empty()
}
/// True when the throttling-parameter (nsig) deobfuscation fn was
/// extracted successfully.
pub fn has_nsig(&self) -> bool {
!self.nsig_fn.is_empty()
}
}
#[derive(Debug, Default, Clone, Serialize, Deserialize, PartialEq, Eq)]
@ -28,7 +50,10 @@ pub struct DeobfData {
impl DeobfData {
/// Download and extract the latest deobfuscation data from YouTube
///
/// Creates a report if the data could not be extracted
/// Creates a report if the data could not be extracted, including a
/// `Level::WARN` report on partial (sig_fn / nsig_fn) extraction failure
/// so reporter-based consumers (e.g. `FileReporter`) get an artefact to
/// debug new player.js shapes against.
pub async fn extract(http: &Client, reporter: Option<&dyn Reporter>) -> Result<Self, Error> {
let js_url = get_player_js_url(http).await?;
let player_js = get_response(http, &js_url).await?;
@ -36,35 +61,98 @@ impl DeobfData {
let res = Self::extract_fns(&js_url, &player_js);
if let Err(e) = &res {
if let Some(reporter) = reporter {
let report = Report {
info: RustyPipeInfo::new(None, None),
level: Level::ERR,
operation: "extract_deobf",
error: Some(e.to_string()),
msgs: vec![],
deobf_data: None,
http_request: crate::report::HTTPRequest {
url: &js_url,
method: "GET",
req_header: None,
req_body: None,
status: 200,
resp_body: player_js,
},
};
reporter.report(&report);
match &res {
Err(e) => {
if let Some(reporter) = reporter {
let report = Report {
info: RustyPipeInfo::new(None, None),
level: Level::ERR,
operation: "extract_deobf",
error: Some(e.to_string()),
msgs: vec![],
deobf_data: None,
http_request: crate::report::HTTPRequest {
url: &js_url,
method: "GET",
req_header: None,
req_body: None,
status: 200,
resp_body: player_js,
},
};
reporter.report(&report);
}
}
Ok(data) if !data.is_complete() => {
// Soft-fail observability — without this, a sig/nsig extraction
// regression is invisible to reporter-based consumers and only
// shows up at `RUST_LOG=warn`. straw / torttube depend on the
// reporter for in-app crash dumps.
if let Some(reporter) = reporter {
let mut missing = Vec::with_capacity(2);
if !data.has_sig() {
missing.push("sig_fn");
}
if !data.has_nsig() {
missing.push("nsig_fn");
}
let report = Report {
info: RustyPipeInfo::new(None, None),
level: Level::WRN,
operation: "extract_deobf_soft_fail",
error: Some(format!(
"partial extraction; missing: {}",
missing.join(", ")
)),
msgs: vec![],
deobf_data: Some(data.clone()),
http_request: crate::report::HTTPRequest {
url: &js_url,
method: "GET",
req_header: None,
req_body: None,
status: 200,
resp_body: player_js,
},
};
reporter.report(&report);
}
}
Ok(_) => {}
}
res
}
pub fn extract_fns(js_url: &str, player_js: &str) -> Result<Self, Error> {
let sig_fn = get_sig_fn(player_js)?;
let nsig_fn = get_nsig_fn(player_js)?;
// The signature timestamp is the only piece every "needs_deobf" client
// actually requires in its request payload — without it, those clients
// get an error back. So we hard-fail on sts extraction.
let sts = get_sts(player_js)?;
// sig_fn and nsig_fn are needed only when YouTube returns stream URLs
// containing the &s= cipher / &n= throttling params. Most clients
// (iOS, Android, Tv) get pre-signed URLs and never touch these.
// Tolerate extraction failures here so a single rotated player.js
// shape doesn't bring down the whole player path for those clients.
// When a stream URL DOES carry &s= / &n=, `Deobfuscator::deobfuscate_sig`
// / `deobfuscate_nsig` short-circuit with a switchable error class
// (see `ExtractionError::switch_client` whitelist) so the client
// fallback loop tries the next client instead of killing the call.
let sig_fn = match get_sig_fn(player_js) {
Ok(f) => f,
Err(e) => {
tracing::warn!("could not extract sig deobf fn (sig deobfuscation disabled until YT rotates player.js again): {}", e);
String::new()
}
};
let nsig_fn = match get_nsig_fn(player_js) {
Ok(f) => f,
Err(e) => {
tracing::warn!("could not extract nsig deobf fn (throttling parameter deobf disabled until YT rotates player.js again): {}", e);
String::new()
}
};
Ok(Self {
js_url: js_url.to_owned(),
sig_fn,
@ -79,19 +167,63 @@ impl Deobfuscator {
pub fn new(data: &DeobfData) -> Result<Self, DeobfError> {
let rt = Runtime::new()?;
let ctx = Context::full(&rt)?;
ctx.with(|ctx| {
let mut opts = rquickjs::context::EvalOptions::default();
opts.strict = false;
ctx.eval_with_options::<(), _>(data.sig_fn.as_bytes(), opts)?;
let mut opts = rquickjs::context::EvalOptions::default();
opts.strict = false;
ctx.eval_with_options::<(), _>(data.nsig_fn.as_bytes(), opts)
let has_sig = data.has_sig();
let has_nsig = data.has_nsig();
ctx.with(|ctx| -> Result<(), rquickjs::Error> {
// Skip JS eval for any deobf fn we couldn't extract. The matching
// `deobfuscate_sig` / `deobfuscate_nsig` calls below guard on
// `has_sig` / `has_nsig` and short-circuit with a clean
// `sig fn unavailable` error instead of falling into rquickjs
// and getting an opaque `FromJs { from: "undefined" ... }` —
// and that opaque shape used to land in `ExtractionError::Deobfuscation`
// which the upstream client-fallback loop treats as non-switchable.
if has_sig {
let mut opts = rquickjs::context::EvalOptions::default();
opts.strict = false;
ctx.eval_with_options::<(), _>(data.sig_fn.as_bytes(), opts)?;
}
if has_nsig {
let mut opts = rquickjs::context::EvalOptions::default();
opts.strict = false;
ctx.eval_with_options::<(), _>(data.nsig_fn.as_bytes(), opts)?;
}
Ok(())
})?;
Ok(Self { ctx })
Ok(Self {
ctx,
has_sig,
has_nsig,
})
}
/// True when the underlying DeobfData had a valid sig fn extracted.
/// Exposed for consumers that want to short-circuit cipher streams
/// without invoking `deobfuscate_sig` and observing the error.
#[allow(dead_code)]
pub fn has_sig(&self) -> bool {
self.has_sig
}
/// True when the underlying DeobfData had a valid nsig fn extracted.
/// Exposed for consumers that want to short-circuit throttled URLs.
#[allow(dead_code)]
pub fn has_nsig(&self) -> bool {
self.has_nsig
}
/// Deobfuscate the `s` parameter from the `signature_cipher` field
pub fn deobfuscate_sig(&self, sig: &str) -> Result<String, DeobfError> {
if !self.has_sig {
// Short-circuit with a recognisable error class. Goes through
// `From<DeobfError> for ExtractionError` → `ExtractionError::Deobfuscation`,
// which is in `switch_client`'s whitelist as of the Sulkta fork — so
// `player_from_clients` will try the next client (typically iOS,
// which doesn't carry signature_cipher streams) rather than killing
// the whole call.
return Err(DeobfError::Other(
"sig fn unavailable (player.js rotation; deobf extraction soft-failed)".into(),
));
}
let res = self
.ctx
.with(|ctx| call_fn(&ctx, DEOBF_SIG_FUNC_NAME, sig))?;
@ -101,6 +233,13 @@ impl Deobfuscator {
/// Deobfuscate the `n` stream URL parameter to circumvent throttling
pub fn deobfuscate_nsig(&self, nsig: &str) -> Result<String, DeobfError> {
if !self.has_nsig {
// Same short-circuit as deobfuscate_sig — switchable error class
// for the client-fallback loop instead of an opaque rquickjs panic.
return Err(DeobfError::Other(
"nsig fn unavailable (player.js rotation; throttle deobf soft-failed)".into(),
));
}
let res = self
.ctx
.with(|ctx| call_fn(&ctx, DEOBF_NSIG_FUNC_NAME, nsig))?;

View file

@ -259,6 +259,13 @@ impl ExtractionError {
..
} | ExtractionError::WrongResult(_)
| ExtractionError::Botguard(_)
// Sulkta fork (CRIT-1): deobf failures are usually transient
// — YT rotated the player.js to a shape our regex doesn't
// recognise, or served a cipher stream to a client that doesn't
// have a working sig fn in cache. Switching to another client
// (iOS first, which doesn't need deobf at all) is the right
// recovery move rather than killing the whole call.
| ExtractionError::Deobfuscation(_)
)
}

167
tests/sulkta_smoke.rs Normal file
View file

@ -0,0 +1,167 @@
//! Sulkta-fork smoke tests for the player pipeline.
//!
//! Verifies the patched default client order (`Ios, Tv` without botguard) plus
//! the soft-fail DeobfData::extract works against current YouTube player.js.
//!
//! Run with: `cargo test --test sulkta_smoke -- --nocapture`
use std::path::PathBuf;
use rstest::{fixture, rstest};
use rustypipe::client::{ClientType, RustyPipe};
use rustypipe::error::{Error, ExtractionError, UnavailabilityReason};
/// A stable, long-running, public-domain music video. Used by upstream
/// tests too (`n4tK7LYFxI0` = Spektrem - Shine, NCS).
const TEST_VIDEO_ID: &str = "n4tK7LYFxI0";
/// Build a `RustyPipe` with a per-process scratch storage dir. Avoids the
/// concurrent-write race with `tests/youtube.rs` that shares `rustypipe_cache.json`
/// in the repo root, which was tripping audit MED-3.
#[fixture]
fn rp() -> RustyPipe {
let scratch: PathBuf = std::env::temp_dir().join(format!(
"rustypipe-sulkta-smoke-{}-{}",
std::process::id(),
std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.map(|d| d.as_nanos())
.unwrap_or(0)
));
std::fs::create_dir_all(&scratch)
.unwrap_or_else(|e| panic!("create scratch storage dir {scratch:?}: {e}"));
RustyPipe::builder()
.storage_dir(&scratch)
.build()
.unwrap_or_else(|e| panic!("build RustyPipe with scratch={scratch:?}: {e}"))
}
/// Sanity: iOS path returns stream URLs and never touches the deobf code.
#[rstest]
#[tokio::test]
async fn ios_player_returns_streams(rp: RustyPipe) {
let pd = rp
.query()
.player_from_client(TEST_VIDEO_ID, ClientType::Ios)
.await
.expect("iOS player_from_client should succeed");
assert_eq!(pd.details.id, TEST_VIDEO_ID);
assert!(
!pd.video_streams.is_empty() || !pd.video_only_streams.is_empty(),
"expected at least one video stream"
);
assert!(
!pd.audio_streams.is_empty(),
"expected at least one audio stream"
);
}
/// TV path exercises the `needs_deobf=true` branch: the sig_timestamp request
/// payload is required, but the soft-fail patch keeps the call alive even when
/// sig_fn/nsig_fn regex extraction fails on a rotated player.js.
///
/// YouTube IP-bans some shared egress IPs (datacenters, LAN-routed servers)
/// for the TV client with "Sign in to confirm you're not a bot". That's
/// environmental — match it precisely on the `UnavailabilityReason` enum
/// instead of substring-matching the rendered error so a real regression
/// can't sneak past the catch arm.
#[rstest]
#[tokio::test]
async fn tv_player_returns_streams(rp: RustyPipe) {
match rp
.query()
.player_from_client(TEST_VIDEO_ID, ClientType::Tv)
.await
{
Ok(pd) => {
assert_eq!(pd.details.id, TEST_VIDEO_ID);
assert!(
!pd.video_streams.is_empty() || !pd.video_only_streams.is_empty(),
"TV path returned no video streams"
);
// Symmetric with iOS / default-order tests so a regression that
// silently drops the audio adaptation set can't pass here.
assert!(
!pd.audio_streams.is_empty(),
"TV path returned no audio streams"
);
}
Err(Error::Extraction(ExtractionError::Unavailable {
reason: UnavailabilityReason::IpBan,
..
})) => {
eprintln!(
"TV path skipped: YT IpBan on this egress (expected on shared/datacenter IPs)"
);
}
Err(e) => panic!("TV path failed for a non-environmental reason: {e}"),
}
}
/// The patched default-client order should pick iOS as primary and return
/// playable streams in the absence of botguard signing.
#[rstest]
#[tokio::test]
async fn default_client_order_returns_streams(rp: RustyPipe) {
let order = rp.query().player_client_order();
eprintln!("default client order (no botguard): {order:?}");
assert_eq!(
order[0],
ClientType::Ios,
"iOS should be the no-botguard primary"
);
let pd = rp
.query()
.player(TEST_VIDEO_ID)
.await
.expect("default-clients player() should succeed");
assert_eq!(pd.details.id, TEST_VIDEO_ID);
assert!(
!pd.video_streams.is_empty() || !pd.video_only_streams.is_empty(),
"expected at least one video stream from the default-clients path"
);
assert!(
!pd.audio_streams.is_empty(),
"expected at least one audio stream from the default-clients path"
);
// Probe one returned audio stream to confirm YT actually serves it.
// GET with Range 0-1023 + an iOS User-Agent because YT's googlevideo
// CDN tends to 403 HEAD requests and UA mismatches.
let stream_url = pd
.audio_streams
.first()
.expect("at least one audio stream")
.url
.clone();
eprintln!(
"probing first audio URL: {}",
&stream_url[..stream_url.len().min(180)]
);
let client = reqwest::Client::builder()
.user_agent(
"com.google.ios.youtube/19.45.4 (iPhone16,2; U; CPU iOS 18_1 like Mac OS X; en_US)",
)
.build()
.unwrap();
let resp = client
.get(&stream_url)
.header("Range", "bytes=0-1023")
.send()
.await
.expect("GET request to YT CDN should not error");
let status = resp.status();
let body_len = resp.bytes().await.map(|b| b.len()).unwrap_or(0);
eprintln!("response: {body_len} bytes, status {status}");
assert!(
status.is_success() || status.is_redirection(),
"audio URL Range-GET returned non-OK status: {status} (body={body_len} bytes; URL may need visitor_data or po_token)"
);
assert!(
body_len > 0,
"audio URL returned OK but zero bytes — likely a sig-required URL we couldn't deobf"
);
}