fix(services/january): remove image if video present and hence fix logic error

refactor(services/january): throw an error if embed fails to generate

crates/core/files/src/lib.rs

@@ -124,13 +124,24 @@ pub fn image_size(f: &NamedTempFile) -> Option<(usize, usize)> {
 }
 
 /// Determine size of image with buffer
-pub fn image_size_vec(v: &[u8]) -> Option<(usize, usize)> {
+pub fn image_size_vec(v: &[u8], mime: &str) -> Option<(usize, usize)> {
-    if let Ok(size) = imagesize::blob_size(v)
+    match mime {
-        .inspect_err(|err| tracing::error!("Failed to generate image size! {err:?}"))
+        "image/svg+xml" => {
-    {
+            let tree =
-        Some((size.width, size.height))
+                report_internal_error!(usvg::Tree::from_data(v, &Default::default())).ok()?;
-    } else {
+
-        None
+            let size = tree.size();
+            Some((size.width() as usize, size.height() as usize))
+        }
+        _ => {
+            if let Ok(size) = imagesize::blob_size(v)
+                .inspect_err(|err| tracing::error!("Failed to generate image size! {err:?}"))
+            {
+                Some((size.width, size.height))
+            } else {
+                None
+            }
+        }
     }
 }
 
@@ -167,7 +178,7 @@ pub fn decode_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> Res
                                 jxl_image.height(),
                                 frame.image().buf().to_vec(),
                             )
-                            .ok_or_else(|| create_error!(LabelMe))?,
+                            .ok_or_else(|| create_error!(ImageProcessingFailed))?,
                         )
                         .to_rgb8(),
                     )),
@@ -178,14 +189,14 @@ pub fn decode_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> Res
                                 jxl_image.height(),
                                 frame.image().buf().to_vec(),
                             )
-                            .ok_or_else(|| create_error!(LabelMe))?,
+                            .ok_or_else(|| create_error!(ImageProcessingFailed))?,
                         )
                         .to_rgba8(),
                     )),
-                    _ => Err(create_error!(LabelMe)),
+                    _ => Err(create_error!(ImageProcessingFailed)),
                 }
             } else {
-                Err(create_error!(LabelMe))
+                Err(create_error!(ImageProcessingFailed))
             }
         }
         // Read image using resvg
@@ -197,7 +208,7 @@ pub fn decode_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> Res
             let tree = report_internal_error!(usvg::Tree::from_data(&buf, &Default::default()))?;
             let size = tree.size();
             let mut pixmap = Pixmap::new(size.width() as u32, size.height() as u32)
-                .ok_or_else(|| create_error!(LabelMe))?;
+                .ok_or_else(|| create_error!(ImageProcessingFailed))?;
 
             let mut pixmap_mut = pixmap.as_mut();
             resvg::render(&tree, Default::default(), &mut pixmap_mut);
@@ -208,7 +219,7 @@ pub fn decode_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> Res
                     size.height() as u32,
                     pixmap.data().to_vec(),
                 )
-                .ok_or_else(|| create_error!(LabelMe))?,
+                .ok_or_else(|| create_error!(ImageProcessingFailed))?,
             ))
         }
         // Check if we can read using image-rs crate

crates/core/result/src/axum.rs

@@ -1,4 +1,5 @@
 use axum::{http::StatusCode, response::IntoResponse, Json};
+use rocket::http::Status;
 
 use crate::{Error, ErrorType};
 
@@ -78,6 +79,8 @@ impl IntoResponse for Error {
             ErrorType::FileTooSmall => StatusCode::UNPROCESSABLE_ENTITY,
             ErrorType::FileTooLarge { .. } => StatusCode::UNPROCESSABLE_ENTITY,
             ErrorType::FileTypeNotAllowed => StatusCode::BAD_REQUEST,
+            ErrorType::ImageProcessingFailed => StatusCode::INTERNAL_SERVER_ERROR,
+            ErrorType::NoEmbedData => StatusCode::BAD_REQUEST,
         };
 
         (status, Json(&self)).into_response()

crates/core/result/src/lib.rs

@@ -162,6 +162,8 @@ pub enum ErrorType {
         max: usize,
     },
     FileTypeNotAllowed,
+    ImageProcessingFailed,
+    NoEmbedData,
 
     // ? Legacy errors
     VosoUnavailable,

crates/core/result/src/rocket.rs

@@ -84,6 +84,8 @@ impl<'r> Responder<'r, 'static> for Error {
             ErrorType::FileTooSmall => Status::UnprocessableEntity,
             ErrorType::FileTooLarge { .. } => Status::UnprocessableEntity,
             ErrorType::FileTypeNotAllowed => Status::BadRequest,
+            ErrorType::ImageProcessingFailed => Status::InternalServerError,
+            ErrorType::NoEmbedData => Status::BadRequest,
         };
 
         // Serialize the error data structure into JSON.

crates/services/january/src/api.rs

@@ -1,6 +1,7 @@
 use axum::{extract::Query, response::IntoResponse, routing::get, Json, Router};
 use reqwest::header;
-use revolt_result::Result;
+use revolt_models::v0::Embed;
+use revolt_result::{create_error, Result};
 use serde::{Deserialize, Serialize};
 use utoipa::ToSchema;
 
@@ -87,5 +88,9 @@ async fn embed(
     Query(UrlQuery { url }): Query<UrlQuery>,
     // TypedHeader(Authorization(_bearer)): TypedHeader<Authorization<Bearer>>,
 ) -> Result<impl IntoResponse> {
-    Request::generate_embed(url).await.map(Json)
+    match Request::generate_embed(url).await {
+        Ok(Embed::None) => Err(create_error!(NoEmbedData)),
+        result => result,
+    }
+    .map(Json)
 }

crates/services/january/src/requests.rs

@@ -66,7 +66,7 @@ impl Request {
             let Request { response, mime } = Request::new(url).await?;
 
             if matches!(mime.type_(), mime::IMAGE | mime::VIDEO) {
-                let bytes = response.bytes().await.map_err(|_| create_error!(LabelMe));
+                let bytes = report_internal_error!(response.bytes().await);
 
                 let result = match bytes {
                     Ok(bytes) => {
@@ -77,7 +77,7 @@ impl Request {
                                 if is_valid_image(reader, "image/gif") {
                                     Ok(("image/gif".to_owned(), bytes.to_vec()))
                                 } else {
-                                    Err(create_error!(LabelMe))
+                                    Err(create_error!(FileTypeNotAllowed))
                                 }
                             } else {
                                 Ok((
@@ -96,7 +96,7 @@ impl Request {
                             if video_size(&file).is_some() {
                                 Ok((mime.to_string(), bytes.to_vec()))
                             } else {
-                                Err(create_error!(LabelMe))
+                                Err(create_error!(FileTypeNotAllowed))
                             }
                         }
                     }
@@ -106,7 +106,7 @@ impl Request {
                 PROXY_CACHE.insert(url.to_owned(), result.clone()).await;
                 result
             } else {
-                Err(create_error!(LabelMe))
+                Err(create_error!(FileTypeNotAllowed))
             }
         }
     }
@@ -122,20 +122,21 @@ impl Request {
                 _ => Ok(None),
             }
         } else {
-            let response = if let Some(Request { response, .. }) = request {
+            let request = if let Some(request) = request {
-                response
+                request
             } else {
-                let Request { response, mime } = Request::new(url).await?;
+                let request = Request::new(url).await?;
-                if matches!(mime.type_(), mime::IMAGE) {
+                if matches!(request.mime.type_(), mime::IMAGE) {
-                    response
+                    request
                 } else {
-                    return Err(create_error!(LabelMe));
+                    return Err(create_error!(FileTypeNotAllowed));
                 }
             };
 
-            if let Some((width, height)) =
+            if let Some((width, height)) = image_size_vec(
-                image_size_vec(&report_internal_error!(response.bytes().await)?)
+                &report_internal_error!(request.response.bytes().await)?,
-            {
+                request.mime.as_ref(),
+            ) {
                 Ok(Some(Image {
                     url: url.to_owned(),
                     width,
@@ -166,7 +167,7 @@ impl Request {
                 if matches!(mime.type_(), mime::VIDEO) {
                     response
                 } else {
-                    return Err(create_error!(LabelMe));
+                    return Err(create_error!(FileTypeNotAllowed));
                 }
             };
 

crates/services/january/src/website_embed.rs

@@ -9,34 +9,38 @@ use scraper::{Html, Selector};
 
 /// Create website metadata from URL and document
 pub async fn create_website_embed(original_url: &str, document: &str) -> Option<WebsiteMetadata> {
-    let document = Html::parse_document(document);
+    let (mut meta, mut link) = {
+        let document = Html::parse_document(document);
 
-    // create selectors
+        // create selectors
-    let meta_selector = Selector::parse("meta").ok()?;
+        let meta_selector = Selector::parse("meta").ok()?;
-    let link_selector = Selector::parse("link").ok()?;
+        let link_selector = Selector::parse("link").ok()?;
 
-    // extract meta tags
+        // extract meta tags
-    let mut meta = HashMap::new();
+        let mut meta = HashMap::new();
-    for el in document.select(&meta_selector) {
+        for el in document.select(&meta_selector) {
-        let node = el.value();
+            let node = el.value();
 
-        if let (Some(property), Some(content)) = (
+            if let (Some(property), Some(content)) = (
-            node.attr("property").or_else(|| node.attr("name")),
+                node.attr("property").or_else(|| node.attr("name")),
-            node.attr("content"),
+                node.attr("content"),
-        ) {
+            ) {
-            meta.insert(property.to_string(), content.to_string());
+                meta.insert(property.to_string(), content.to_string());
+            }
         }
-    }
 
-    // extract rel links
+        // extract rel links
-    let mut link = HashMap::new();
+        let mut link = HashMap::new();
-    for el in document.select(&link_selector) {
+        for el in document.select(&link_selector) {
-        let node = el.value();
+            let node = el.value();
 
-        if let (Some(property), Some(content)) = (node.attr("rel"), node.attr("href")) {
+            if let (Some(property), Some(content)) = (node.attr("rel"), node.attr("href")) {
-            link.insert(property.to_string(), content.to_string());
+                link.insert(property.to_string(), content.to_string());
+            }
         }
-    }
+
+        (meta, link)
+    };
 
     // build metadata
     let mut metadata = WebsiteMetadata {
@@ -137,32 +141,41 @@ pub async fn create_website_embed(original_url: &str, document: &str) -> Option<
     };
 
     // populate extra metadata for popular websites
-    populate_special(original_url.to_owned(), &mut metadata);
+    populate_special(original_url.to_owned(), &mut metadata).await;
-
-    // TODO: these do not work because compiler is tripping:
 
     // fetch video size if missing
-    // if let Some(Video { width, height, url }) = &metadata.video {
+    if metadata.special.is_none() {
-    //     if width == &0 || height == &0 {
+        if let Some(Video { width, height, url }) = &metadata.video {
-    //         metadata.video = match crate::requests::Request::fetch_video_metadata(url, None).await {
+            if width == &0 || height == &0 {
-    //             Ok(Some(video)) => Some(video),
+                metadata.video =
-    //             _ => None,
+                    match crate::requests::Request::fetch_video_metadata(url, None).await {
-    //         }
+                        Ok(Some(video)) => Some(video),
-    //     }
+                        _ => None,
-    // }
+                    }
+            }
+        }
+    }
+
+    // remove image if video exists
+    if metadata.video.is_some() {
+        metadata.image.take();
+    }
 
     // fetch image size if missing
-    // if let Some(Image {
+    if metadata.special.is_none() {
-    //     width, height, url, ..
+        if let Some(Image {
-    // }) = &metadata.image
+            width, height, url, ..
-    // {
+        }) = &metadata.image
-    //     if width == &0 || height == &0 {
+        {
-    //         metadata.image = match crate::requests::Request::fetch_image_metadata(url, None).await {
+            if width == &0 || height == &0 {
-    //             Ok(Some(image)) => Some(image),
+                metadata.image =
-    //             _ => None,
+                    match crate::requests::Request::fetch_image_metadata(url, None).await {
-    //         }
+                        Ok(Some(image)) => Some(image),
-    //     }
+                        _ => None,
-    // }
+                    }
+            }
+        }
+    }
 
     // truncate data
     metadata.truncate();
@@ -175,7 +188,7 @@ pub async fn create_website_embed(original_url: &str, document: &str) -> Option<
     }
 }
 
-pub fn populate_special(original_url: String, metadata: &mut WebsiteMetadata) {
+pub async fn populate_special(original_url: String, metadata: &mut WebsiteMetadata) {
     lazy_static! {
         static ref RE_YOUTUBE: Regex = Regex::new("^(?:(?:https?:)?//)?(?:(?:www|m)\\.)?(?:(?:youtube\\.com|youtu.be))(?:/(?:[\\w\\-]+\\?v=|embed/|v/)?)([\\w\\-]+)(?:\\S+)?$").unwrap();
 
@@ -223,15 +236,14 @@ pub fn populate_special(original_url: String, metadata: &mut WebsiteMetadata) {
             metadata.site_name.take();
 
             // Verify the video exists
-            // TODO: breaks axum :(
+            if !crate::requests::Request::exists(&format!(
-            // if !crate::requests::Request::exists(&format!(
+                "http://img.youtube.com/vi/{}/sddefault.jpg",
-            //     "http://img.youtube.com/vi/{}/sddefault.jpg",
+                id
-            //     id
+            ))
-            // ))
+            .await
-            // .await
+            {
-            // {
+                return;
-            //     return;
+            }
-            // }
         }
 
         if let Some(timestamp_captures) = RE_TIMESTAMP.captures_iter(url).next() {

scripts/publish-debug-image.sh

@@ -24,6 +24,7 @@ docker build -t ghcr.io/revoltchat/base:latest -f Dockerfile.useCurrentArch .
 docker build -t ghcr.io/revoltchat/server:$TAG - < crates/delta/Dockerfile
 docker build -t ghcr.io/revoltchat/bonfire:$TAG - < crates/bonfire/Dockerfile
 docker build -t ghcr.io/revoltchat/autumn:$TAG - < crates/services/autumn/Dockerfile
+docker build -t ghcr.io/revoltchat/january:$TAG - < crates/services/january/Dockerfile
 
 if [ "$DEBUG" = "true" ]; then
   git restore Cargo.toml
@@ -32,3 +33,4 @@ fi
 docker push ghcr.io/revoltchat/server:$TAG
 docker push ghcr.io/revoltchat/bonfire:$TAG
 docker push ghcr.io/revoltchat/autumn:$TAG
+docker push ghcr.io/revoltchat/january:$TAG