From afd8c906bab23a1967a2f4eed2ca54cc5ad0b5dc Mon Sep 17 00:00:00 2001
From: Paul Makles <me@insrt.uk>
Date: Wed, 2 Oct 2024 11:31:55 +0100
Subject: [PATCH] fix(services/january): YouTube fallback

---
 crates/services/january/src/main.rs          |  2 +
 crates/services/january/src/requests.rs      | 11 ++++-
 crates/services/january/src/website_embed.rs | 45 ++++++++++++++------
 3 files changed, 44 insertions(+), 14 deletions(-)

diff --git a/crates/services/january/src/main.rs b/crates/services/january/src/main.rs
index 2175ca0a..b905b87d 100644
--- a/crates/services/january/src/main.rs
+++ b/crates/services/january/src/main.rs
@@ -66,6 +66,8 @@ async fn main() -> Result<(), std::io::Error> {
         .nest("/", api::router().await);
 
     // Configure TCP listener and bind
+    tracing::info!("Listening on 0.0.0.0:14705");
+    tracing::info!("Play around with the API: http://localhost:14705/scalar");
     let address = SocketAddr::from((Ipv4Addr::UNSPECIFIED, 14705));
     let listener = TcpListener::bind(&address).await?;
     axum::serve(listener, app.into_make_service()).await
diff --git a/crates/services/january/src/requests.rs b/crates/services/january/src/requests.rs
index c8e540de..44218f1f 100644
--- a/crates/services/january/src/requests.rs
+++ b/crates/services/january/src/requests.rs
@@ -156,7 +156,7 @@ impl Request {
             let response = if let Some(Request { response, .. }) = request {
                 response
             } else {
-                let Request { response, mime } = Request::new(&url).await?;
+                let Request { response, mime } = Request::new(url).await?;
                 if matches!(mime.type_(), mime::VIDEO) {
                     response
                 } else {
@@ -262,4 +262,13 @@ impl Request {
 
         Ok(Request { response, mime })
     }
+
+    /// Check if something exists
+    pub async fn exists(url: &str) -> bool {
+        if let Ok(response) = CLIENT.head(url).send().await {
+            response.status().is_success()
+        } else {
+            false
+        }
+    }
 }
diff --git a/crates/services/january/src/website_embed.rs b/crates/services/january/src/website_embed.rs
index 794fe9dc..b52f01c4 100644
--- a/crates/services/january/src/website_embed.rs
+++ b/crates/services/january/src/website_embed.rs
@@ -207,24 +207,43 @@ pub fn populate_special(original_url: String, metadata: &mut WebsiteMetadata) {
             id: captures[1].to_string(),
         })
     } else if let Some(captures) = RE_YOUTUBE.captures_iter(url).next() {
+        let id = captures[1].to_string();
+
         lazy_static! {
             static ref RE_TIMESTAMP: Regex = Regex::new("(?:\\?|&)(?:t|start)=([\\w]+)").unwrap();
         }
 
-        if let Some(video) = &metadata.video {
-            if let Some(timestamp_captures) = RE_TIMESTAMP.captures_iter(&video.url).next() {
-                Some(Special::YouTube {
-                    id: captures[1].to_string(),
-                    timestamp: Some(timestamp_captures[1].to_string()),
-                })
-            } else {
-                Some(Special::YouTube {
-                    id: captures[1].to_string(),
-                    timestamp: None,
-                })
-            }
+        // YouTube now blocks datacentre IPs from fetching information
+        // This is a fallback to prevent the embed from looking weird
+        if metadata.video.is_none() {
+            metadata.title.replace("YouTube".to_owned());
+            metadata.description.take();
+            metadata.colour.take();
+            metadata.icon_url.take();
+            metadata.site_name.take();
+
+            // Verify the video exists
+            // TODO: breaks axum :(
+            // if !crate::requests::Request::exists(&format!(
+            //     "http://img.youtube.com/vi/{}/sddefault.jpg",
+            //     id
+            // ))
+            // .await
+            // {
+            //     return;
+            // }
+        }
+
+        if let Some(timestamp_captures) = RE_TIMESTAMP.captures_iter(url).next() {
+            Some(Special::YouTube {
+                id,
+                timestamp: Some(timestamp_captures[1].to_string()),
+            })
         } else {
-            None
+            Some(Special::YouTube {
+                id,
+                timestamp: None,
+            })
         }
     } else if let Some(captures) = RE_LIGHTSPEED.captures_iter(url).next() {
         Some(Special::Lightspeed {