fix(services/january): reddit embeds by spoofing as discord / mapping to old reddit

closes #360

.github/workflows/docker.yaml

@@ -61,7 +61,7 @@ jobs:
     if: github.event_name != 'pull_request'
     strategy:
       matrix:
-        project: [delta, bonfire, autumn]
+        project: [delta, bonfire, autumn, january]
     name: Build ${{ matrix.project }} image
     steps:
       # Configure build environment
@@ -102,6 +102,10 @@ jobs:
               "autumn": {
                 "path": "crates/services/autumn",
                 "tag": "${{ github.repository_owner }}/autumn"
+              },
+              "january": {
+                "path": "crates/services/january",
+                "tag": "${{ github.repository_owner }}/january"
               }
             }
           export_to: output

.github/workflows/rust.yaml

@@ -41,11 +41,7 @@ jobs:
 
       - name: Run services in background
         run: |
-          docker compose -f docker-compose.db.yml up -d
-
-      - name: Copy .env.example
-        run: |
-          cp .env.example .env
+          docker compose -f compose.yml up -d
 
       - name: Run cargo test
         env:

crates/bindings/node/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-nodejs-bindings"
-version = "0.7.16"
+version = "0.7.18"
 description = "Node.js bindings for the Revolt software"
 authors = ["Paul Makles <me@insrt.uk>"]
 license = "MIT"
@@ -20,6 +20,6 @@ serde = { version = "1", features = ["derive"] }
 
 async-std = "1.12.0"
 
-revolt-config = { version = "0.7.16", path = "../../core/config" }
-revolt-result = { version = "0.7.16", path = "../../core/result" }
-revolt-database = { version = "0.7.16", path = "../../core/database" }
+revolt-config = { version = "0.7.18", path = "../../core/config" }
+revolt-result = { version = "0.7.18", path = "../../core/result" }
+revolt-database = { version = "0.7.18", path = "../../core/database" }

crates/bonfire/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-bonfire"
-version = "0.7.16"
+version = "0.7.18"
 license = "AGPL-3.0-or-later"
 edition = "2021"
 
@@ -41,7 +41,7 @@ revolt-result = { path = "../core/result" }
 revolt-models = { path = "../core/models" }
 revolt-config = { path = "../core/config" }
 revolt-database = { path = "../core/database" }
-revolt-permissions = { version = "0.7.16", path = "../core/permissions" }
+revolt-permissions = { version = "0.7.18", path = "../core/permissions" }
 revolt-presence = { path = "../core/presence", features = ["redis-is-patched"] }
 
 # redis

crates/core/config/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-config"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "MIT"
 authors = ["Paul Makles <me@insrt.uk>"]
@@ -35,4 +35,4 @@ pretty_env_logger = "0.4.0"
 sentry = "0.31.5"
 
 # Core
-revolt-result = { version = "0.7.16", path = "../result", optional = true }
+revolt-result = { version = "0.7.18", path = "../result", optional = true }

crates/core/database/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-database"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "AGPL-3.0-or-later"
 authors = ["Paul Makles <me@insrt.uk>"]
@@ -24,15 +24,15 @@ default = ["mongodb", "async-std-runtime", "tasks"]
 
 [dependencies]
 # Core
-revolt-config = { version = "0.7.16", path = "../config", features = [
+revolt-config = { version = "0.7.18", path = "../config", features = [
     "report-macros",
 ] }
-revolt-result = { version = "0.7.16", path = "../result" }
-revolt-models = { version = "0.7.16", path = "../models", features = [
+revolt-result = { version = "0.7.18", path = "../result" }
+revolt-models = { version = "0.7.18", path = "../models", features = [
     "validator",
 ] }
-revolt-presence = { version = "0.7.16", path = "../presence" }
-revolt-permissions = { version = "0.7.16", path = "../permissions", features = [
+revolt-presence = { version = "0.7.18", path = "../presence" }
+revolt-permissions = { version = "0.7.18", path = "../permissions", features = [
     "serde",
     "bson",
 ] }

crates/core/files/Cargo.toml

@@ -1,12 +1,18 @@
 [package]
 name = "revolt-files"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "AGPL-3.0-or-later"
 authors = ["Paul Makles <me@insrt.uk>"]
 description = "Revolt Backend: S3 and encryption subroutines"
 
 [dependencies]
+tracing = "0.1"
+
+ffprobe = "0.4.0"
+imagesize = "0.13.0"
+tempfile = "3.12.0"
+
 base64 = "0.22.1"
 aes-gcm = "0.10.3"
 typenum = "1.17.0"
@@ -14,7 +20,19 @@ typenum = "1.17.0"
 aws-config = "1.5.5"
 aws-sdk-s3 = { version = "1.46.0", features = ["behavior-version-latest"] }
 
-revolt-config = { version = "0.7.16", path = "../config", features = [
+revolt-config = { version = "0.7.18", path = "../config", features = [
     "report-macros",
 ] }
-revolt-result = { version = "0.7.16", path = "../result" }
+revolt-result = { version = "0.7.18", path = "../result" }
+
+# image processing
+jxl-oxide = "0.8.1"
+image = { version = "0.25.2" }
+
+# svg rendering
+usvg = "0.44.0"
+resvg = "0.44.0"
+tiny-skia = "0.11.4"
+
+# encoding
+webp = "0.3.0"

crates/core/files/src/lib.rs

@@ -1,9 +1,10 @@
-use std::io::Write;
+use std::io::{BufRead, Read, Seek, Write};
 
 use aes_gcm::{
     aead::{AeadCore, AeadMutInPlace, OsRng},
     Aes256Gcm, Key, KeyInit, Nonce,
 };
+use image::{DynamicImage, ImageBuffer};
 use revolt_config::{config, report_internal_error, FilesS3};
 use revolt_result::{create_error, Result};
 
@@ -13,6 +14,8 @@ use aws_sdk_s3::{
 };
 
 use base64::prelude::*;
+use tempfile::NamedTempFile;
+use tiny_skia::Pixmap;
 
 /// Size of the authentication tag in the buffer
 pub const AUTHENTICATION_TAG_SIZE_BYTES: usize = 16;
@@ -62,6 +65,11 @@ pub async fn fetch_from_s3(bucket_id: &str, path: &str, nonce: &str) -> Result<V
         // we just want the Vec<u8>
     }
 
+    // File is not encrypted
+    if nonce.is_empty() {
+        return Ok(buf);
+    }
+
     // Recover nonce as bytes
     let nonce = &BASE64_STANDARD.decode(nonce).unwrap()[..];
     let nonce: &Nonce<typenum::consts::U12> = nonce.into();
@@ -103,3 +111,151 @@ pub async fn upload_to_s3(bucket_id: &str, path: &str, buf: &[u8]) -> Result<Str
 
     Ok(BASE64_STANDARD.encode(nonce))
 }
+
+/// Determine size of image at temp file
+pub fn image_size(f: &NamedTempFile) -> Option<(usize, usize)> {
+    if let Ok(size) = imagesize::size(f.path())
+        .inspect_err(|err| tracing::error!("Failed to generate image size! {err:?}"))
+    {
+        Some((size.width, size.height))
+    } else {
+        None
+    }
+}
+
+/// Determine size of image with buffer
+pub fn image_size_vec(v: &[u8]) -> Option<(usize, 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
+    }
+}
+
+/// Determine size of video at temp file
+pub fn video_size(f: &NamedTempFile) -> Option<(i64, i64)> {
+    if let Ok(data) = ffprobe::ffprobe(f.path())
+        .inspect_err(|err| tracing::error!("Failed to ffprobe file! {err:?}"))
+    {
+        // Use first valid stream
+        for stream in data.streams {
+            if let (Some(w), Some(h)) = (stream.width, stream.height) {
+                return Some((w, h));
+            }
+        }
+
+        None
+    } else {
+        None
+    }
+}
+
+/// Decode image from reader
+pub fn decode_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> Result<DynamicImage> {
+    match mime {
+        // Read image using jxl-oxide crate
+        "image/jxl" => {
+            let jxl_image = report_internal_error!(jxl_oxide::JxlImage::builder().read(reader))?;
+            if let Ok(frame) = jxl_image.render_frame(0) {
+                match frame.color_channels().len() {
+                    3 => Ok(DynamicImage::ImageRgb8(
+                        DynamicImage::ImageRgb32F(
+                            ImageBuffer::from_vec(
+                                jxl_image.width(),
+                                jxl_image.height(),
+                                frame.image().buf().to_vec(),
+                            )
+                            .ok_or_else(|| create_error!(LabelMe))?,
+                        )
+                        .to_rgb8(),
+                    )),
+                    4 => Ok(DynamicImage::ImageRgba8(
+                        DynamicImage::ImageRgba32F(
+                            ImageBuffer::from_vec(
+                                jxl_image.width(),
+                                jxl_image.height(),
+                                frame.image().buf().to_vec(),
+                            )
+                            .ok_or_else(|| create_error!(LabelMe))?,
+                        )
+                        .to_rgba8(),
+                    )),
+                    _ => Err(create_error!(LabelMe)),
+                }
+            } else {
+                Err(create_error!(LabelMe))
+            }
+        }
+        // Read image using resvg
+        "image/svg+xml" => {
+            // usvg doesn't support Read trait so copy to buffer
+            let mut buf = Vec::new();
+            report_internal_error!(reader.read_to_end(&mut buf))?;
+
+            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))?;
+
+            let mut pixmap_mut = pixmap.as_mut();
+            resvg::render(&tree, Default::default(), &mut pixmap_mut);
+
+            Ok(DynamicImage::ImageRgba8(
+                ImageBuffer::from_vec(
+                    size.width() as u32,
+                    size.height() as u32,
+                    pixmap.data().to_vec(),
+                )
+                .ok_or_else(|| create_error!(LabelMe))?,
+            ))
+        }
+        // Check if we can read using image-rs crate
+        _ => report_internal_error!(report_internal_error!(
+            image::ImageReader::new(reader).with_guessed_format()
+        )?
+        .decode()),
+    }
+}
+
+/// Check whether given reader has a valid image
+pub fn is_valid_image<R: Read + BufRead + Seek>(reader: &mut R, mime: &str) -> bool {
+    match mime {
+        // Check if we can read using jxl-oxide crate
+        "image/jxl" => jxl_oxide::JxlImage::builder()
+            .read(reader)
+            .inspect_err(|err| tracing::error!("Failed to read JXL! {err:?}"))
+            .is_ok(),
+        // Check if we can read using image-rs crate
+        _ => !matches!(
+            image::ImageReader::new(reader)
+                .with_guessed_format()
+                .inspect_err(|err| tracing::error!("Failed to read image! {err:?}"))
+                .map(|f| f.decode()),
+            Err(_) | Ok(Err(_))
+        ),
+    }
+}
+
+/// Create thumbnail from given image
+pub async fn create_thumbnail(image: DynamicImage, tag: &str) -> Vec<u8> {
+    // Load configuration
+    let config = config().await;
+    let [w, h] = config.files.preview.get(tag).unwrap();
+
+    // Create thumbnail
+    //.resize(width as u32, height as u32, image::imageops::FilterType::Gaussian)
+    // resize is about 2.5x slower,
+    // thumbnail doesn't have terrible quality
+    // so we use thumbnail
+    let image = image.thumbnail(image.width().min(*w as u32), image.height().min(*h as u32));
+
+    // Encode it into WEBP
+    let encoder = webp::Encoder::from_image(&image).expect("Could not create encoder.");
+    if config.files.webp_quality != 100.0 {
+        encoder.encode(config.files.webp_quality).to_vec()
+    } else {
+        encoder.encode_lossless().to_vec()
+    }
+}

crates/core/models/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-models"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "MIT"
 authors = ["Paul Makles <me@insrt.uk>"]
@@ -20,8 +20,8 @@ default = ["serde", "partials", "rocket"]
 
 [dependencies]
 # Core
-revolt-config = { version = "0.7.16", path = "../config" }
-revolt-permissions = { version = "0.7.16", path = "../permissions" }
+revolt-config = { version = "0.7.18", path = "../config" }
+revolt-permissions = { version = "0.7.18", path = "../permissions" }
 
 # Utility
 regex = "1"

crates/core/models/src/v0/embeds.rs

@@ -14,9 +14,9 @@ auto_derived!(
         /// URL to the original image
         pub url: String,
         /// Width of the image
-        pub width: isize,
+        pub width: usize,
         /// Height of the image
-        pub height: isize,
+        pub height: usize,
         /// Positioning and size
         pub size: ImageSize,
     }
@@ -26,9 +26,9 @@ auto_derived!(
         /// URL to the original video
         pub url: String,
         /// Width of the video
-        pub width: isize,
+        pub width: usize,
         /// Height of the video
-        pub height: isize,
+        pub height: usize,
     }
 
     /// Type of remote Twitch content
@@ -86,7 +86,7 @@ auto_derived!(
         },
         AppleMusic {
             album_id: String,
-            
+
             #[serde(skip_serializing_if = "Option::is_none")]
             track_id: Option<String>,
         },
@@ -119,8 +119,6 @@ auto_derived!(
         #[serde(skip_serializing_if = "Option::is_none")]
         pub video: Option<Video>,
 
-        // #[serde(skip_serializing_if = "Option::is_none")]
-        // opengraph_type: Option<String>,
         /// Site name
         #[serde(skip_serializing_if = "Option::is_none")]
         pub site_name: Option<String>,
@@ -156,11 +154,56 @@ auto_derived!(
 
     /// Embed
     #[serde(tag = "type")]
+    #[derive(Default)]
     pub enum Embed {
         Website(WebsiteMetadata),
         Image(Image),
         Video(Video),
         Text(Text),
+        #[default]
         None,
     }
 );
+
+impl WebsiteMetadata {
+    /// Truncate strings in metadata
+    pub fn truncate(&mut self) {
+        if let Some(s) = self.url.as_mut() {
+            s.truncate(256);
+        }
+
+        if let Some(s) = self.original_url.as_mut() {
+            s.truncate(256);
+        }
+
+        if let Some(s) = self.title.as_mut() {
+            s.truncate(100);
+        }
+
+        if let Some(s) = self.description.as_mut() {
+            s.truncate(1000);
+        }
+
+        if let Some(s) = self.site_name.as_mut() {
+            s.truncate(32);
+        }
+
+        if let Some(s) = self.icon_url.as_mut() {
+            s.truncate(256);
+        }
+
+        if let Some(s) = self.colour.as_mut() {
+            s.truncate(32);
+        }
+    }
+
+    /// Check if this is considered "empty"
+    pub fn is_empty(&self) -> bool {
+        (self.title.is_none() || self.title.as_ref().is_some_and(|f| f.is_empty()))
+            && (self.description.is_none()
+                || self.description.as_ref().is_some_and(|f| f.is_empty()))
+            && self.special.is_none()
+            && self.video.is_none()
+            && self.image.is_none()
+    }
+}

crates/core/permissions/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-permissions"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "MIT"
 authors = ["Paul Makles <me@insrt.uk>"]
@@ -21,7 +21,7 @@ async-std = { version = "1.8.0", features = ["attributes"] }
 
 [dependencies]
 # Core
-revolt-result = { version = "0.7.16", path = "../result" }
+revolt-result = { version = "0.7.18", path = "../result" }
 
 # Utility
 auto_ops = "0.3.0"

crates/core/presence/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-presence"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "AGPL-3.0-or-later"
 authors = ["Paul Makles <me@insrt.uk>"]

crates/core/result/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-result"
-version = "0.7.16"
+version = "0.7.18"
 edition = "2021"
 license = "MIT"
 authors = ["Paul Makles <me@insrt.uk>"]

crates/delta/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-delta"
-version = "0.7.16"
+version = "0.7.18"
 license = "AGPL-3.0-or-later"
 authors = ["Paul Makles <paulmakles@gmail.com>"]
 edition = "2018"

crates/services/autumn/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "revolt-autumn"
-version = "0.7.14"
+version = "0.7.18"
 edition = "2021"
 
 [dependencies]
@@ -42,12 +42,12 @@ tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 
 # Core crates
-revolt-files = { version = "0.7.16", path = "../../core/files" }
-revolt-config = { version = "0.7.16", path = "../../core/config" }
-revolt-database = { version = "0.7.16", path = "../../core/database", features = [
+revolt-files = { version = "0.7.18", path = "../../core/files" }
+revolt-config = { version = "0.7.18", path = "../../core/config" }
+revolt-database = { version = "0.7.18", path = "../../core/database", features = [
     "axum-impl",
 ] }
-revolt-result = { version = "0.7.16", path = "../../core/result", features = [
+revolt-result = { version = "0.7.18", path = "../../core/result", features = [
     "utoipa",
     "axum",
 ] }

crates/services/autumn/src/api.rs

@@ -11,11 +11,12 @@ use axum::{
     Json, Router,
 };
 use axum_typed_multipart::{FieldData, TryFromMultipart, TypedMultipart};
-use image::ImageReader;
 use lazy_static::lazy_static;
 use revolt_config::{config, report_internal_error};
 use revolt_database::{iso8601_timestamp::Timestamp, Database, FileHash, Metadata, User};
-use revolt_files::{fetch_from_s3, upload_to_s3, AUTHENTICATION_TAG_SIZE_BYTES};
+use revolt_files::{
+    create_thumbnail, decode_image, fetch_from_s3, upload_to_s3, AUTHENTICATION_TAG_SIZE_BYTES,
+};
 use revolt_result::{create_error, Result};
 use serde::{Deserialize, Serialize};
 use sha2::Digest;
@@ -352,30 +353,12 @@ async fn fetch_preview(
     // Original image data
     let data = retrieve_file_by_hash(&hash).await?;
 
-    // Dimensions we need to resize to
-    let config = config().await;
-    let [w, h] = config.files.preview.get(tag).unwrap();
-
-    // Read the image and resize it
-    // TODO: use jxl_oxide to process image/jxl files
-    let image = report_internal_error!(report_internal_error!(ImageReader::new(Cursor::new(
-        data
-    ))
-    .with_guessed_format())?
-    .decode())?
-    //.resize(width as u32, height as u32, image::imageops::FilterType::Gaussian)
-    // resize is about 2.5x slower,
-    // thumbnail doesn't have terrible quality
-    // so we use thumbnail
-    .thumbnail(*w as u32, *h as u32);
-
-    // Encode it into WEBP
-    let encoder = webp::Encoder::from_image(&image).expect("Could not create encoder.");
-    let data = if config.files.webp_quality != 100.0 {
-        encoder.encode(config.files.webp_quality).to_vec()
-    } else {
-        encoder.encode_lossless().to_vec()
-    };
+    // Read image and create thumbnail
+    let data = create_thumbnail(
+        decode_image(&mut Cursor::new(data), &file.content_type)?,
+        tag,
+    )
+    .await;
 
     Ok((
         [

crates/services/autumn/src/metadata.rs

@@ -1,6 +1,7 @@
 use std::io::Cursor;
 
 use revolt_database::Metadata;
+use revolt_files::{image_size, video_size};
 use tempfile::NamedTempFile;
 
 /// Intersection of what infer can detect and what image-rs supports
@@ -21,32 +22,19 @@ static SUPPORTED_IMAGE_MIME: [&str; 9] = [
 /// Generate metadata from file, using mime type as a hint
 pub fn generate_metadata(f: &NamedTempFile, mime_type: &str) -> Metadata {
     if SUPPORTED_IMAGE_MIME.contains(&mime_type) {
-        if let Ok(size) = imagesize::size(f.path())
-            .inspect_err(|err| tracing::error!("Failed to generate image size! {err:?}"))
-        {
-            if let (Ok(width), Ok(height)) = (size.width.try_into(), size.height.try_into()) {
-                return Metadata::Image { width, height };
-            }
-        }
-
-        Metadata::File
+        image_size(f)
+            .map(|(width, height)| Metadata::Image {
+                width: width as isize,
+                height: height as isize,
+            })
+            .unwrap_or_default()
     } else if mime_type.starts_with("video/") {
-        if let Ok(data) = ffprobe::ffprobe(f.path())
-            .inspect_err(|err| tracing::error!("Failed to ffprobe file! {err:?}"))
-        {
-            // Use first valid stream
-            for stream in data.streams {
-                if let (Some(w), Some(h)) = (stream.width, stream.height) {
-                    if let (Ok(width), Ok(height)) = (w.try_into(), h.try_into()) {
-                        return Metadata::Video { width, height };
-                    }
-                }
-            }
-
-            Metadata::File
-        } else {
-            Metadata::File
-        }
+        video_size(f)
+            .map(|(width, height)| Metadata::Video {
+                width: width as isize,
+                height: height as isize,
+            })
+            .unwrap_or_default()
     } else if mime_type.starts_with("audio/") {
         Metadata::Audio
     } else if mime_type == "plain/text" {

crates/services/autumn/src/mime_type.rs

@@ -19,7 +19,11 @@ pub fn determine_mime_type(f: &mut NamedTempFile, buf: &[u8], file_name: &str) -
 
     // See if the file is actually just plain Unicode/ASCII text
     if mime_type == "application/octet-stream" && simdutf8::basic::from_utf8(buf).is_ok() {
-        return "plain/text";
+        if file_name.to_lowercase().ends_with(".svg") {
+            return "image/svg+xml";
+        } else {
+            return "plain/text";
+        }
     }
 
     mime_type

crates/services/january/Cargo.toml

@@ -1,19 +1,26 @@
 [package]
 name = "revolt-january"
-version = "0.7.14"
+version = "0.7.18"
 edition = "2021"
 
 [dependencies]
 # Utility
 mime = "0.3.17"
+regex = "1.11.0"
+tempfile = "3.13.0"
 lazy_static = "1.5.0"
 moka = { version = "0.12.8", features = ["future"] }
 
+# Web scraping
+scraper = "0.20.0"
+encoding_rs = "0.8.34"
+
 # Serialisation
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0.68"
 
 # Async runtime
+async-recursion = "1.1.1"
 tokio = { version = "1.0", features = ["full"] }
 
 # Web requests
@@ -24,12 +31,13 @@ tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 
 # Core crates
-revolt-config = { version = "0.7.16", path = "../../core/config" }
-revolt-models = { version = "0.7.16", path = "../../core/models" }
-revolt-result = { version = "0.7.16", path = "../../core/result", features = [
+revolt-config = { version = "0.7.18", path = "../../core/config" }
+revolt-models = { version = "0.7.18", path = "../../core/models" }
+revolt-result = { version = "0.7.18", path = "../../core/result", features = [
     "utoipa",
     "axum",
 ] }
+revolt-files = { version = "0.7.18", path = "../../core/files" }
 
 # Axum / web server
 axum = { version = "0.7.5" }

crates/services/january/Dockerfile

@@ -0,0 +1,12 @@
+# Build Stage
+FROM ghcr.io/revoltchat/base:latest AS builder
+
+# Bundle Stage
+FROM gcr.io/distroless/cc-debian12:nonroot
+COPY --from=builder /home/rust/src/target/release/revolt-january ./
+COPY --from=mwader/static-ffmpeg:7.0.2 /ffmpeg /usr/local/bin/
+COPY --from=mwader/static-ffmpeg:7.0.2 /ffprobe /usr/local/bin/
+
+EXPOSE 14705
+USER nonroot
+CMD ["./revolt-january"]

crates/services/january/src/api.rs

@@ -1,16 +1,13 @@
-use axum::{body::Bytes, extract::Query, routing::get, Json, Router};
-use revolt_models::v0::Embed;
+use axum::{extract::Query, response::IntoResponse, routing::get, Json, Router};
+use reqwest::header;
 use revolt_result::Result;
 use serde::{Deserialize, Serialize};
 use utoipa::ToSchema;
 
-use axum_extra::{
-    headers::{authorization::Bearer, Authorization},
-    TypedHeader,
-};
-
 use crate::requests::Request;
 
+pub static CACHE_CONTROL: &str = "public, max-age=600, immutable";
+
 pub async fn router() -> Router {
     Router::new()
         .route("/", get(root))
@@ -59,8 +56,17 @@ struct UrlQuery {
         ("url" = String, Query, description = "URL to fetch")
     ),
 )]
-async fn proxy(Query(UrlQuery { url }): Query<UrlQuery>) -> Result<Bytes> {
-    Request::proxy_file(&url).await
+async fn proxy(Query(UrlQuery { url }): Query<UrlQuery>) -> Result<impl IntoResponse> {
+    Request::proxy_file(&url).await.map(|(content_type, data)| {
+        (
+            [
+                (header::CONTENT_TYPE, content_type),
+                (header::CONTENT_DISPOSITION, "inline".to_owned()),
+                (header::CACHE_CONTROL, CACHE_CONTROL.to_owned()),
+            ],
+            data,
+        )
+    })
 }
 
 /// Generate embed for a given URL
@@ -79,7 +85,7 @@ async fn proxy(Query(UrlQuery { url }): Query<UrlQuery>) -> Result<Bytes> {
 )]
 async fn embed(
     Query(UrlQuery { url }): Query<UrlQuery>,
-    TypedHeader(Authorization(_bearer)): TypedHeader<Authorization<Bearer>>,
-) -> Result<Json<Embed>> {
-    Request::generate_embed(&url).await.map(Json)
+    // TypedHeader(Authorization(_bearer)): TypedHeader<Authorization<Bearer>>,
+) -> Result<impl IntoResponse> {
+    Request::generate_embed(url).await.map(Json)
 }

crates/services/january/src/main.rs

@@ -11,6 +11,7 @@ use utoipa_scalar::{Scalar, Servable as ScalarServable};
 
 mod api;
 pub mod requests;
+pub mod website_embed;
 
 #[tokio::main]
 async fn main() -> Result<(), std::io::Error> {
@@ -65,6 +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

crates/services/january/src/requests.rs

@@ -1,15 +1,23 @@
-use std::time::Duration;
-
-use axum::body::Bytes;
+use encoding_rs::{Encoding, UTF_8_INIT};
 use lazy_static::lazy_static;
 use mime::Mime;
-use reqwest::{header::CONTENT_TYPE, redirect, Client, Response};
-use revolt_models::v0::Embed;
+use regex::Regex;
+use reqwest::{
+    header::{self, CONTENT_TYPE},
+    redirect, Client, Response,
+};
+use revolt_config::report_internal_error;
+use revolt_files::{create_thumbnail, decode_image, image_size_vec, is_valid_image, video_size};
+use revolt_models::v0::{Embed, Image, ImageSize, Video};
 use revolt_result::{create_error, Result};
+use std::{
+    io::{Cursor, Write},
+    time::Duration,
+};
 
 lazy_static! {
+    /// Request client
     static ref CLIENT: Client = reqwest::Client::builder()
-        .user_agent("Mozilla/5.0 (compatible; January/2.0; +https://github.com/revoltchat/backend)")
         .timeout(Duration::from_secs(10)) // TODO config
         .connect_timeout(Duration::from_secs(5)) // TODO config
         .redirect(redirect::Policy::custom(|attempt| {
@@ -24,14 +32,20 @@ lazy_static! {
         .build()
         .expect("reqwest Client");
 
+    /// Spoof User Agent as Discord
+    static ref RE_USER_AGENT_SPOOFING_AS_DISCORD: Regex = Regex::new("^(?:(?:https?:)?//)?(?:(?:vx|fx)?twitter|(?:fixv|fixup)?x|(?:old\\.|new\\.|www\\.)reddit).com").expect("valid regex");
+
+    /// Regex for matching new Reddit URLs
+    static ref RE_URL_NEW_REDDIT: Regex = Regex::new("^(?:(?:https?:)?//)?(?:(?:new\\.|www\\.)?reddit).com").expect("valid regex");
+
     /// Cache for proxy results
-    static ref PROXY_CACHE: moka::future::Cache<String, Result<Bytes>> = moka::future::Cache::builder()
+    static ref PROXY_CACHE: moka::future::Cache<String, Result<(String, Vec<u8>)>> = moka::future::Cache::builder()
         .max_capacity(10_000) // TODO config
         .time_to_live(Duration::from_secs(60)) // TODO config
         .build();
 
     /// Cache for embed results
-    static ref EMBED_CACHE: moka::future::Cache<String, Result<Embed>> = moka::future::Cache::builder()
+    static ref EMBED_CACHE: moka::future::Cache<String, Embed> = moka::future::Cache::builder()
         .max_capacity(1_000) // TODO config
         .time_to_live(Duration::from_secs(60)) // TODO config
         .build();
@@ -45,20 +59,187 @@ pub struct Request {
 
 impl Request {
     /// Proxy a given URL
-    pub async fn proxy_file(url: &str) -> Result<Bytes> {
+    pub async fn proxy_file(url: &str) -> Result<(String, Vec<u8>)> {
         if let Some(hit) = PROXY_CACHE.get(url).await {
             hit
         } else {
-            todo!()
+            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 result = match bytes {
+                    Ok(bytes) => {
+                        if matches!(mime.type_(), mime::IMAGE) {
+                            let reader = &mut Cursor::new(&bytes);
+
+                            if matches!(mime.subtype(), mime::GIF) {
+                                if is_valid_image(reader, "image/gif") {
+                                    Ok(("image/gif".to_owned(), bytes.to_vec()))
+                                } else {
+                                    Err(create_error!(LabelMe))
+                                }
+                            } else {
+                                Ok((
+                                    "image/webp".to_owned(),
+                                    create_thumbnail(
+                                        decode_image(reader, mime.as_ref())?,
+                                        "attachments",
+                                    )
+                                    .await,
+                                ))
+                            }
+                        } else {
+                            let mut file = report_internal_error!(tempfile::NamedTempFile::new())?;
+                            report_internal_error!(file.write_all(&bytes))?;
+
+                            if video_size(&file).is_some() {
+                                Ok((mime.to_string(), bytes.to_vec()))
+                            } else {
+                                Err(create_error!(LabelMe))
+                            }
+                        }
+                    }
+                    Err(err) => Err(err),
+                };
+
+                PROXY_CACHE.insert(url.to_owned(), result.clone()).await;
+                result
+            } else {
+                Err(create_error!(LabelMe))
+            }
+        }
+    }
+
+    /// Fetch metadata for an image
+    pub async fn fetch_image_metadata(
+        url: &str,
+        request: Option<Request>,
+    ) -> Result<Option<Image>> {
+        if let Some(hit) = EMBED_CACHE.get(url).await {
+            match hit {
+                Embed::Image(img) => Ok(Some(img)),
+                _ => Ok(None),
+            }
+        } else {
+            let response = if let Some(Request { response, .. }) = request {
+                response
+            } else {
+                let Request { response, mime } = Request::new(url).await?;
+                if matches!(mime.type_(), mime::IMAGE) {
+                    response
+                } else {
+                    return Err(create_error!(LabelMe));
+                }
+            };
+
+            if let Some((width, height)) =
+                image_size_vec(&report_internal_error!(response.bytes().await)?)
+            {
+                Ok(Some(Image {
+                    url: url.to_owned(),
+                    width,
+                    height,
+                    size: ImageSize::Large,
+                }))
+            } else {
+                Ok(None)
+            }
+        }
+    }
+
+    /// Fetch metadata for an video
+    pub async fn fetch_video_metadata(
+        url: &str,
+        request: Option<Request>,
+    ) -> Result<Option<Video>> {
+        if let Some(hit) = EMBED_CACHE.get(url).await {
+            match hit {
+                Embed::Video(vid) => Ok(Some(vid)),
+                _ => Ok(None),
+            }
+        } else {
+            let response = if let Some(Request { response, .. }) = request {
+                response
+            } else {
+                let Request { response, mime } = Request::new(url).await?;
+                if matches!(mime.type_(), mime::VIDEO) {
+                    response
+                } else {
+                    return Err(create_error!(LabelMe));
+                }
+            };
+
+            let mut file = report_internal_error!(tempfile::NamedTempFile::new())?;
+            report_internal_error!(
+                file.write_all(&report_internal_error!(response.bytes().await)?)
+            )?;
+
+            if let Some((width, height)) = video_size(&file) {
+                Ok(Some(Video {
+                    url: url.to_owned(),
+                    width: width as usize,
+                    height: height as usize,
+                }))
+            } else {
+                Ok(None)
+            }
         }
     }
 
     /// Generate embed for a given URL
-    pub async fn generate_embed(url: &str) -> Result<Embed> {
-        if let Some(hit) = EMBED_CACHE.get(url).await {
-            hit
+    pub async fn generate_embed(mut url: String) -> Result<Embed> {
+        // Re-map certain links for better metadata generation
+        if RE_URL_NEW_REDDIT.is_match(&url) {
+            url = RE_URL_NEW_REDDIT
+                // Reddit has a bunch of clickbait-y marketing on the new URLs, so we use the old site instead
+                .replace(&url, "https://old.reddit.com")
+                .to_string();
+        }
+
+        // Generate the actual embed
+        if let Some(hit) = EMBED_CACHE.get(&url).await {
+            Ok(hit)
         } else {
-            todo!()
+            let request = Request::new(&url).await?;
+            let embed = match (request.mime.type_(), request.mime.subtype()) {
+                (_, mime::HTML) => {
+                    let content_type = request
+                        .response
+                        .headers()
+                        .get(header::CONTENT_TYPE)
+                        .and_then(|value| value.to_str().ok())
+                        .and_then(|value| value.parse::<Mime>().ok());
+
+                    let encoding_name = content_type
+                        .as_ref()
+                        .and_then(|mime| mime.get_param("charset").map(|charset| charset.as_str()))
+                        .unwrap_or("utf-8");
+
+                    let encoding =
+                        Encoding::for_label(encoding_name.as_bytes()).unwrap_or(&UTF_8_INIT);
+
+                    let bytes = report_internal_error!(request.response.bytes().await)?;
+                    let (text, _, _) = encoding.decode(&bytes);
+
+                    crate::website_embed::create_website_embed(&url, &text)
+                        .await
+                        .map(Embed::Website)
+                        .unwrap_or_default()
+                }
+                (mime::IMAGE, _) => Request::fetch_image_metadata(&url, Some(request))
+                    .await
+                    .map(|res| res.map(Embed::Image).unwrap_or_default())
+                    .unwrap_or_default(),
+                (mime::VIDEO, _) => Request::fetch_video_metadata(&url, Some(request))
+                    .await
+                    .map(|res| res.map(Embed::Video).unwrap_or_default())
+                    .unwrap_or_default(),
+                _ => Embed::None,
+            };
+
+            EMBED_CACHE.insert(url.to_owned(), embed.clone()).await;
+            Ok(embed)
         }
     }
 
@@ -66,6 +247,15 @@ impl Request {
     pub async fn new(url: &str) -> Result<Request> {
         let response = CLIENT
             .get(url)
+            .header(
+                "User-Agent",
+                if RE_USER_AGENT_SPOOFING_AS_DISCORD.is_match(url) {
+                    "Mozilla/5.0 (compatible; Discordbot/2.0; +https://discordapp.com)"
+                } else {
+                    "Mozilla/5.0 (compatible; January/2.0; +https://github.com/revoltchat/backend)"
+                },
+            )
+            .header("Accept-Language", "en-US,en;q=0.5")
             .send()
             .await
             .map_err(|_| create_error!(ProxyError))?;
@@ -88,4 +278,13 @@ 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
+        }
+    }
 }

crates/services/january/src/website_embed.rs

@@ -0,0 +1,323 @@
+use std::collections::HashMap;
+
+use lazy_static::lazy_static;
+use regex::Regex;
+use revolt_models::v0::{
+    BandcampType, Image, ImageSize, LightspeedType, Special, TwitchType, Video, WebsiteMetadata,
+};
+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);
+
+    // create selectors
+    let meta_selector = Selector::parse("meta").ok()?;
+    let link_selector = Selector::parse("link").ok()?;
+
+    // extract meta tags
+    let mut meta = HashMap::new();
+    for el in document.select(&meta_selector) {
+        let node = el.value();
+
+        if let (Some(property), Some(content)) = (
+            node.attr("property").or_else(|| node.attr("name")),
+            node.attr("content"),
+        ) {
+            meta.insert(property.to_string(), content.to_string());
+        }
+    }
+
+    // extract rel links
+    let mut link = HashMap::new();
+    for el in document.select(&link_selector) {
+        let node = el.value();
+
+        if let (Some(property), Some(content)) = (node.attr("rel"), node.attr("href")) {
+            link.insert(property.to_string(), content.to_string());
+        }
+    }
+
+    // build metadata
+    let mut metadata = WebsiteMetadata {
+        title: meta
+            .remove("og:title")
+            .or_else(|| meta.remove("twitter:title"))
+            .or_else(|| meta.remove("title"))
+            .map(|s| s.trim().to_owned()),
+        description: meta
+            .remove("og:description")
+            .or_else(|| meta.remove("twitter:description"))
+            .or_else(|| meta.remove("description"))
+            .map(|s| s.trim().to_owned()),
+        image: meta
+            .remove("og:image")
+            .or_else(|| meta.remove("og:image:secure_url"))
+            .or_else(|| meta.remove("twitter:image"))
+            .or_else(|| meta.remove("twitter:image:src"))
+            .map(|s| s.trim().to_owned())
+            .map(|mut url| {
+                // If relative URL, prepend root URL. Also if root URL ends with a slash, remove it.
+                if let Some(ch) = url.chars().next() {
+                    if ch == '/' {
+                        url = format!("{}{}", &original_url.trim_end_matches('/'), url);
+                    }
+                }
+
+                let mut size = ImageSize::Preview;
+                if let Some(card) = meta.remove("twitter:card") {
+                    if &card == "summary_large_image" {
+                        size = ImageSize::Large;
+                    }
+                }
+
+                Image {
+                    url: url.to_owned(),
+                    width: meta
+                        .remove("og:image:width")
+                        .unwrap_or_default()
+                        .parse()
+                        .unwrap_or(0),
+                    height: meta
+                        .remove("og:image:height")
+                        .unwrap_or_default()
+                        .parse()
+                        .unwrap_or(0),
+                    size,
+                }
+            }),
+        video: meta
+            .remove("og:video")
+            .or_else(|| meta.remove("og:video:url"))
+            .or_else(|| meta.remove("og:video:secure_url"))
+            .map(|s| s.trim().to_owned())
+            .map(|mut url| {
+                // If relative URL, prepend root URL. Also if root URL ends with a slash, remove it.
+                if let Some(ch) = url.chars().next() {
+                    if ch == '/' {
+                        url = format!("{}{}", &original_url.trim_end_matches('/'), url);
+                    }
+                }
+
+                Video {
+                    url: url.to_owned(),
+                    width: meta
+                        .remove("og:video:width")
+                        .unwrap_or_default()
+                        .parse()
+                        .unwrap_or(0),
+                    height: meta
+                        .remove("og:video:height")
+                        .unwrap_or_default()
+                        .parse()
+                        .unwrap_or(0),
+                }
+            }),
+        icon_url: link
+            .remove("apple-touch-icon")
+            .or_else(|| link.remove("icon"))
+            .map(|s| s.trim().to_owned())
+            .map(|mut v| {
+                // If relative URL, prepend root URL.
+                if let Some(ch) = v.chars().next() {
+                    if ch == '/' {
+                        v = format!("{}{}", &original_url.trim_end_matches('/'), v);
+                    }
+                }
+
+                v
+            }),
+        colour: meta.remove("theme-color").map(|s| s.trim().to_owned()),
+        site_name: meta.remove("og:site_name").map(|s| s.trim().to_owned()),
+        url: meta
+            .remove("og:url")
+            .or_else(|| Some(original_url.to_owned())),
+        original_url: Some(original_url.to_owned()),
+        special: None,
+    };
+
+    // populate extra metadata for popular websites
+    populate_special(original_url.to_owned(), &mut metadata);
+
+    // TODO: these do not work because compiler is tripping:
+
+    // fetch video size if missing
+    // if let Some(Video { width, height, url }) = &metadata.video {
+    //     if width == &0 || height == &0 {
+    //         metadata.video = match crate::requests::Request::fetch_video_metadata(url, None).await {
+    //             Ok(Some(video)) => Some(video),
+    //             _ => None,
+    //         }
+    //     }
+    // }
+
+    // fetch image size if missing
+    // if let Some(Image {
+    //     width, height, url, ..
+    // }) = &metadata.image
+    // {
+    //     if width == &0 || height == &0 {
+    //         metadata.image = match crate::requests::Request::fetch_image_metadata(url, None).await {
+    //             Ok(Some(image)) => Some(image),
+    //             _ => None,
+    //         }
+    //     }
+    // }
+
+    // truncate data
+    metadata.truncate();
+
+    // if it's empty, don't return anything
+    if metadata.is_empty() {
+        None
+    } else {
+        Some(metadata)
+    }
+}
+
+pub 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+)?$").unwrap();
+
+        static ref RE_LIGHTSPEED: Regex = Regex::new("^(?:https?://)?(?:[\\w]+\\.)?lightspeed\\.tv/([a-z0-9_]{4,25})").unwrap();
+
+        static ref RE_TWITCH: Regex = Regex::new("^(?:https?://)?(?:www\\.|go\\.)?twitch\\.tv/([a-z0-9_]+)($|\\?)").unwrap();
+        static ref RE_TWITCH_VOD: Regex = Regex::new("^(?:https?://)?(?:www\\.|go\\.)?twitch\\.tv/videos/([0-9]+)($|\\?)").unwrap();
+        static ref RE_TWITCH_CLIP: Regex = Regex::new("^(?:https?://)?(?:www\\.|go\\.)?twitch\\.tv/(?:[a-z0-9_]+)/clip/([A-z0-9_-]+)($|\\?)").unwrap();
+
+        static ref RE_SPOTIFY: Regex = Regex::new("^(?:https?://)?open.spotify.com/(track|user|artist|album|playlist)/([A-z0-9]+)").unwrap();
+        static ref RE_SOUNDCLOUD: Regex = Regex::new("^(?:https?://)?soundcloud.com/([a-zA-Z0-9-]+)/([A-z0-9-]+)").unwrap();
+        static ref RE_BANDCAMP: Regex = Regex::new("^(?:https?://)?(?:[A-z0-9_-]+).bandcamp.com/(track|album)/([A-z0-9_-]+)").unwrap();
+        static ref RE_APPLE_MUSIC: Regex = Regex::new("^(?:https?://)?music\\.apple\\.com/(?:[a-z]{2}/)?album/(?:[a-zA-Z0-9-]+)/(\\d+)(?:\\?i=(\\d+))?").unwrap();
+
+        static ref RE_STREAMABLE: Regex = Regex::new("^(?:https?://)?(?:www\\.)?streamable\\.com/([\\w\\d-]+)").unwrap();
+
+        static ref RE_GIF: Regex = Regex::new("^(?:https?://)?(www\\.)?(gifbox\\.me/view|yiffbox\\.me/view|tenor\\.com/view|giphy\\.com/gifs|gfycat\\.com|redgifs\\.com/watch)/[\\w\\d-]+").unwrap();
+    }
+
+    let url = metadata
+        .url
+        .as_ref()
+        .or(metadata.original_url.as_ref())
+        .unwrap_or(&original_url)
+        .as_str();
+
+    metadata.special = if let Some(captures) = RE_STREAMABLE.captures_iter(url).next() {
+        Some(Special::Streamable {
+            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();
+        }
+
+        // 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 {
+            Some(Special::YouTube {
+                id,
+                timestamp: None,
+            })
+        }
+    } else if let Some(captures) = RE_LIGHTSPEED.captures_iter(url).next() {
+        Some(Special::Lightspeed {
+            id: captures[1].to_string(),
+            content_type: LightspeedType::Channel,
+        })
+    } else if let Some(captures) = RE_TWITCH.captures_iter(url).next() {
+        Some(Special::Twitch {
+            id: captures[1].to_string(),
+            content_type: TwitchType::Channel,
+        })
+    } else if let Some(captures) = RE_TWITCH_VOD.captures_iter(url).next() {
+        Some(Special::Twitch {
+            id: captures[1].to_string(),
+            content_type: TwitchType::Video,
+        })
+    } else if let Some(captures) = RE_TWITCH_CLIP.captures_iter(url).next() {
+        Some(Special::Twitch {
+            id: captures[1].to_string(),
+            content_type: TwitchType::Clip,
+        })
+    } else if let Some(captures) = RE_SPOTIFY.captures_iter(url).next() {
+        Some(Special::Spotify {
+            content_type: captures[1].to_string(),
+            id: captures[2].to_string(),
+        })
+    } else if RE_SOUNDCLOUD.is_match(url) {
+        Some(Special::Soundcloud)
+    } else if RE_BANDCAMP.is_match(url) {
+        lazy_static! {
+            static ref RE_TRACK: Regex = Regex::new("track=(\\d+)").unwrap();
+            static ref RE_ALBUM: Regex = Regex::new("album=(\\d+)").unwrap();
+        }
+
+        if let Some(video) = &metadata.video {
+            if let Some(captures) = RE_TRACK.captures_iter(&video.url).next() {
+                Some(Special::Bandcamp {
+                    content_type: BandcampType::Track,
+                    id: captures[1].to_string(),
+                })
+            } else {
+                RE_ALBUM
+                    .captures_iter(&video.url)
+                    .next()
+                    .map(|captures| Special::Bandcamp {
+                        content_type: BandcampType::Album,
+                        id: captures[1].to_string(),
+                    })
+            }
+        } else {
+            None
+        }
+    } else if RE_GIF.is_match(url) {
+        Some(Special::GIF)
+    } else {
+        RE_APPLE_MUSIC
+            .captures_iter(url)
+            .next()
+            .map(|captures| Special::AppleMusic {
+                album_id: captures[1].to_string(),
+                track_id: captures.get(2).map(|m| m.as_str().to_string()),
+            })
+    };
+
+    // add colours for popular websites
+    if let Some(special) = &metadata.special {
+        match special {
+            Special::YouTube { .. } => metadata.colour = Some("#FF424F".to_string()),
+            Special::Twitch { .. } => metadata.colour = Some("#7B68EE".to_string()),
+            Special::Lightspeed { .. } => metadata.colour = Some("#7445D9".to_string()),
+            Special::Spotify { .. } => metadata.colour = Some("#1ABC9C".to_string()),
+            Special::Soundcloud { .. } => metadata.colour = Some("#FF7F50".to_string()),
+            Special::AppleMusic { .. } => metadata.colour = Some("#FA233B".to_string()),
+            _ => {}
+        }
+    }
+}