Apparently YouTube can detect randomly generated visitor data and prevents both the iOS and TV player from being fetched (Error: Sign in to confirm you’re not a bot). Therefore the visitor data generation code was removed and replaced with a cache that randomly chooses from a selection of real visitor data.
176 lines
5.7 KiB
Rust
176 lines
5.7 KiB
Rust
use std::sync::{atomic::AtomicU32, Arc, RwLock};
|
|
|
|
use once_cell::sync::Lazy;
|
|
use rand::Rng;
|
|
use regex::Regex;
|
|
use reqwest::{header, Client};
|
|
|
|
use crate::{
|
|
client::YOUTUBE_MUSIC_HOME_URL,
|
|
error::{Error, ExtractionError},
|
|
util,
|
|
};
|
|
|
|
/// To increase privacy and possibly circumvent rate limits, RustyPipe uses multiple
|
|
/// visitor data IDs. These are held in this cache object.
|
|
///
|
|
/// On instantiation, the cache is empty, so for the first requests new visitor data IDs
|
|
/// have to be requested. For subsequent requests a random ID from the cache is picked.
|
|
/// After req_limit requests, a new token is requested asynchronously and added to the cache
|
|
/// to prevent the IDs from being overused.
|
|
///
|
|
/// The cache holds a maximum of 100 visitor data IDs. If more are added, the oldest ones
|
|
/// are evicted.
|
|
#[derive(Clone)]
|
|
pub struct VisitorDataCache {
|
|
inner: Arc<VisitorDataCacheRef>,
|
|
}
|
|
|
|
struct VisitorDataCacheRef {
|
|
req_counter: AtomicU32,
|
|
visitor_data: RwLock<Vec<String>>,
|
|
http: Client,
|
|
}
|
|
|
|
static VISITOR_DATA_REGEX: Lazy<Regex> =
|
|
Lazy::new(|| Regex::new(r#""visitorData":"([\w\d_\-%]+?)""#).unwrap());
|
|
/// Number of requests after which a new token is requested
|
|
const REQ_LIMIT: u32 = 10;
|
|
/// Maximum size of the cache (-1)
|
|
const MAX_SIZE: usize = 99;
|
|
|
|
impl VisitorDataCache {
|
|
pub fn new(http: Client) -> Self {
|
|
Self {
|
|
inner: VisitorDataCacheRef {
|
|
req_counter: Default::default(),
|
|
visitor_data: Default::default(),
|
|
http,
|
|
}
|
|
.into(),
|
|
}
|
|
}
|
|
|
|
async fn get_visitor_data(&self) -> Result<String, Error> {
|
|
tracing::debug!("getting YT visitor data");
|
|
let resp = self
|
|
.inner
|
|
.http
|
|
.get(YOUTUBE_MUSIC_HOME_URL)
|
|
.header(header::ORIGIN, YOUTUBE_MUSIC_HOME_URL)
|
|
.header(header::REFERER, YOUTUBE_MUSIC_HOME_URL)
|
|
.send()
|
|
.await?;
|
|
|
|
let vdata = resp
|
|
.headers()
|
|
.get_all(header::SET_COOKIE)
|
|
.iter()
|
|
.find_map(|c| {
|
|
if let Ok(cookie) = c.to_str() {
|
|
if let Some(after) = cookie.strip_prefix("__Secure-YEC=") {
|
|
return after
|
|
.split_once(';')
|
|
.map(|s| s.0.to_owned())
|
|
.filter(|s| !s.is_empty());
|
|
}
|
|
}
|
|
None
|
|
});
|
|
|
|
match vdata {
|
|
Some(vdata) => Ok(vdata),
|
|
None => {
|
|
if resp.status().is_success() {
|
|
// Extract visitor data from html
|
|
let html = resp.text().await?;
|
|
|
|
util::get_cg_from_regex(&VISITOR_DATA_REGEX, &html, 1).ok_or(Error::Extraction(
|
|
ExtractionError::InvalidData(
|
|
"Could not find visitor data on html page".into(),
|
|
),
|
|
))
|
|
} else {
|
|
Err(Error::Extraction(ExtractionError::InvalidData(
|
|
format!("Could not get visitor data, status: {}", resp.status()).into(),
|
|
)))
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
pub async fn new_visitor_data(&self) -> Result<String, Error> {
|
|
self.inner
|
|
.req_counter
|
|
.store(0, std::sync::atomic::Ordering::SeqCst);
|
|
let vd = self.get_visitor_data().await.unwrap();
|
|
let mut vds = self.inner.visitor_data.write().unwrap();
|
|
for _ in 0..(vds.len().saturating_sub(MAX_SIZE)) {
|
|
let rem = vds.remove(0);
|
|
tracing::debug!("visitor data {rem} removed from cache");
|
|
}
|
|
vds.push(vd.to_owned());
|
|
tracing::debug!("visitor data {} added to cache ({} ids)", vd, vds.len());
|
|
Ok(vd)
|
|
}
|
|
|
|
pub async fn get(&self) -> Result<String, Error> {
|
|
// Request new visitor data in the background every 10 requests
|
|
if self
|
|
.inner
|
|
.req_counter
|
|
.fetch_add(1, std::sync::atomic::Ordering::SeqCst)
|
|
>= REQ_LIMIT
|
|
{
|
|
let nc = self.clone();
|
|
tokio::spawn(async move { nc.new_visitor_data().await });
|
|
}
|
|
|
|
{
|
|
let vds = self.inner.visitor_data.read().unwrap();
|
|
if !vds.is_empty() {
|
|
let mut rng = rand::thread_rng();
|
|
let vd = vds[rng.gen_range(0..vds.len())].to_owned();
|
|
tracing::debug!("visitor data {vd} picked from cache");
|
|
return Ok(vd);
|
|
}
|
|
}
|
|
// Fetch new visitor data if the cache is empty
|
|
self.new_visitor_data().await
|
|
}
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use std::time::Duration;
|
|
|
|
use crate::client::DEFAULT_UA;
|
|
|
|
use super::*;
|
|
|
|
use tracing_test::traced_test;
|
|
|
|
#[tokio::test]
|
|
#[traced_test]
|
|
async fn get_visitor_data() {
|
|
let cache =
|
|
VisitorDataCache::new(Client::builder().user_agent(DEFAULT_UA).build().unwrap());
|
|
// Get initial visitor data
|
|
let v1 = cache.get().await.unwrap();
|
|
|
|
// Run as many request as necessary to fetch second visitor data
|
|
for _ in 0..=REQ_LIMIT {
|
|
let got = cache.get().await.unwrap();
|
|
assert_eq!(got, v1);
|
|
}
|
|
|
|
// Second visitor data does not arrive instantly, request immediately after returns the first data
|
|
let vds_len = cache.inner.visitor_data.read().unwrap().len();
|
|
assert_eq!(vds_len, 1);
|
|
|
|
// Wait for the second visitor data to arrive
|
|
tokio::time::sleep(Duration::from_millis(1000)).await;
|
|
let vds_len = cache.inner.visitor_data.read().unwrap().len();
|
|
assert_eq!(vds_len, 2);
|
|
}
|
|
}
|