fix: extract visitor data from html page

Reported issue: https://github.com/rust-lang/regex/issues/1060

Cargo.toml

@@ -31,7 +31,7 @@ quick-js-dtp = { version = "0.4.1", default-features = false, features = [
     "patch-dateparser",
 ] }
 once_cell = "1.12.0"
-regex = "1.6.0"
+regex = ">=1.6.0, <1.9.0"
 fancy-regex = "0.11.0"
 thiserror = "1.0.36"
 url = "2.2.2"

src/client/mod.rs

@@ -209,8 +209,10 @@ const ANDROID_API_KEY: &str = "AIzaSyA8eiZmM1FaDVjRy-df2KTyQ_vz_yYM39w";
 const IOS_API_KEY: &str = "AIzaSyB-63vPrdThhKuerbB2N_l7Kwwcxj6yUAc";
 const IOS_DEVICE_MODEL: &str = "iPhone14,5";
 
-static CLIENT_VERSION_REGEXES: Lazy<[Regex; 1]> =
+static CLIENT_VERSION_REGEX: Lazy<Regex> =
-    Lazy::new(|| [Regex::new(r#"INNERTUBE_CONTEXT_CLIENT_VERSION":"([\w\d\._-]+?)""#).unwrap()]);
+    Lazy::new(|| Regex::new(r#""INNERTUBE_CONTEXT_CLIENT_VERSION":"([\w\d\._-]+?)""#).unwrap());
+static VISITOR_DATA_REGEX: Lazy<Regex> =
+    Lazy::new(|| Regex::new(r#""visitorData":"([\w\d_\-%]+?)""#).unwrap());
 
 /// The RustyPipe client used to access YouTube's API
 ///
@@ -814,11 +816,11 @@ impl RustyPipe {
                 )
                 .await?;
 
-            util::get_cg_from_regexes(CLIENT_VERSION_REGEXES.iter(), &swjs, 1).ok_or(
+            util::get_cg_from_regex(&CLIENT_VERSION_REGEX, &swjs, 1).ok_or(Error::Extraction(
-                Error::Extraction(ExtractionError::InvalidData(Cow::Borrowed(
+                ExtractionError::InvalidData(Cow::Borrowed(
                     "Could not find client version in sw.js",
-                ))),
+                )),
-            )
+            ))
         });
 
         let from_html = async {
@@ -829,11 +831,11 @@ impl RustyPipe {
 
             let html = self.http_request_txt(&builder.build().unwrap()).await?;
 
-            util::get_cg_from_regexes(CLIENT_VERSION_REGEXES.iter(), &html, 1).ok_or(
+            util::get_cg_from_regex(&CLIENT_VERSION_REGEX, &html, 1).ok_or(Error::Extraction(
-                Error::Extraction(ExtractionError::InvalidData(Cow::Borrowed(
+                ExtractionError::InvalidData(Cow::Borrowed(
                     "Could not find client version on html page",
-                ))),
+                )),
-            )
+            ))
         };
 
         if let Some(from_swjs) = from_swjs {
@@ -965,11 +967,15 @@ impl RustyPipe {
     ///
     /// Since the cookie is shared between YT and YTM and the YTM page loads faster,
     /// we request that.
+    ///
+    /// Sometimes YouTube does not set the `__Secure-YEC` cookie. In this case, the
+    /// visitor data is extracted from the html page.
     async fn get_visitor_data(&self) -> Result<String, Error> {
         log::debug!("getting YT visitor data");
         let resp = self.inner.http.get(YOUTUBE_MUSIC_HOME_URL).send().await?;
 
-        resp.headers()
+        let vdata = resp
+            .headers()
             .get_all(header::SET_COOKIE)
             .iter()
             .find_map(|c| {
@@ -979,12 +985,29 @@ impl RustyPipe {
                     }
                 }
                 None
-            })
+            });
-            .ok_or(Error::Extraction(ExtractionError::InvalidData(
+
-                Cow::Borrowed("could not get YTM cookies"),
+        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(Cow::Borrowed(
+                            "Could not find visitor data on html page",
+                        )),
+                    ))
+                } else {
+                    Err(Error::Extraction(ExtractionError::InvalidData(
+                        format!("Could not get visitor data, status: {}", resp.status()).into(),
                     )))
                 }
             }
+        }
+    }
+}
 
 impl RustyPipeQuery {
     /// Set the language parameter used when accessing the YouTube API

src/util/mod.rs

@@ -41,14 +41,11 @@ pub const ARTIST_DISCOGRAPHY_PREFIX: &str = "MPAD";
 const CONTENT_PLAYBACK_NONCE_ALPHABET: &[u8; 64] =
     b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_";
 
-/// Return the given capture group that matches first in a list of regexes
+/// Return the given capture group that matches the regex
-pub fn get_cg_from_regexes<'a, I>(mut regexes: I, text: &str, cg: usize) -> Option<String>
+pub fn get_cg_from_regex(regex: &Regex, text: &str, cg: usize) -> Option<String> {
-where
+    regex
-    I: Iterator<Item = &'a Regex>,
+        .captures(text)
-{
+        .and_then(|c| c.get(cg).map(|c| c.as_str().to_owned()))
-    regexes
-        .find_map(|pattern| pattern.captures(text))
-        .map(|c| c.get(cg).unwrap().as_str().to_owned())
 }
 
 /// Return the given capture group that matches first in a list of fancy regexes
@@ -58,7 +55,7 @@ where
 {
     regexes
         .find_map(|pattern| pattern.captures(text).ok().flatten())
-        .map(|c| c.get(cg).unwrap().as_str().to_owned())
+        .and_then(|c| c.get(cg).map(|c| c.as_str().to_owned()))
 }
 
 /// Generate a random string with given length and byte charset.