From 7f132c1517835c00ceb61fdd06b689583e09a5bc Mon Sep 17 00:00:00 2001
From: Ashley <iamashley@duck.com>
Date: Sun, 10 Dec 2023 10:27:07 +0000
Subject: [PATCH] inital source for january

---
 january/src/routes/embed.rs | 98 +++++++++++++++++++++++++++++++++++++
 january/src/routes/info.rs  | 14 ++++++
 january/src/routes/mod.rs   |  3 ++
 january/src/routes/proxy.rs | 43 ++++++++++++++++
 4 files changed, 158 insertions(+)
 create mode 100644 january/src/routes/embed.rs
 create mode 100644 january/src/routes/info.rs
 create mode 100644 january/src/routes/mod.rs
 create mode 100644 january/src/routes/proxy.rs

diff --git a/january/src/routes/embed.rs b/january/src/routes/embed.rs
new file mode 100644
index 00000000..911ff83b
--- /dev/null
+++ b/january/src/routes/embed.rs
@@ -0,0 +1,98 @@
+use std::time::Duration;
+
+use actix_web::{
+    web::{self, Query},
+    Responder,
+};
+use regex::Regex;
+use serde::Deserialize;
+
+use crate::structs::metadata::Metadata;
+use crate::structs::{embed::Embed, media::Video};
+use crate::util::request::fetch;
+use crate::{
+    structs::media::{Image, ImageSize},
+    util::{request::consume_size, result::Error},
+};
+
+lazy_static! {
+    static ref CACHE: moka::future::Cache<String, Result<Embed, Error>> =
+        moka::future::Cache::builder()
+            .max_capacity(1_000)
+            .time_to_live(Duration::from_secs(60))
+            .build();
+}
+
+#[derive(Deserialize)]
+pub struct Parameters {
+    url: String,
+}
+
+async fn embed(mut url: String) -> Result<Embed, Error> {
+    // Twitter is a piece of shit and does not
+    // provide metadata in an easily consumable format.
+    //
+    // So... we just redirect everything to Nitter.
+    //
+    // Fun bonus: Twitter denied our developer application
+    // which would've been the only way to pull properly
+    // formatted Tweet data out and what's worse is that this
+    // also prevents us adding those "connections" that other
+    // platforms have.
+    //
+    // In any case, because Twitter, they
+    // do not provide OpenGraph data.
+    lazy_static! {
+        static ref RE_TWITTER: Regex =
+            Regex::new("^(?:https?://)?(?:www\\.)?twitter\\.com").unwrap();
+    }
+
+    if RE_TWITTER.is_match(&url) {
+        url = RE_TWITTER.replace(&url, "https://nitter.net").into();
+    }
+
+    // Fetch URL
+    let (resp, mime) = fetch(&url).await?;
+
+    // Match appropriate MIME type to process
+    match (mime.type_(), mime.subtype()) {
+        (_, mime::HTML) => {
+            let mut metadata = Metadata::from(resp, url).await?;
+            metadata.resolve_external().await;
+
+            if metadata.is_none() {
+                return Ok(Embed::None);
+            }
+
+            Ok(Embed::Website(metadata))
+        }
+        (mime::IMAGE, _) => {
+            if let Ok((width, height)) = consume_size(resp, mime).await {
+                Ok(Embed::Image(Image {
+                    url,
+                    width,
+                    height,
+                    size: ImageSize::Large,
+                }))
+            } else {
+                Ok(Embed::None)
+            }
+        }
+        (mime::VIDEO, _) => {
+            if let Ok((width, height)) = consume_size(resp, mime).await {
+                Ok(Embed::Video(Video { url, width, height }))
+            } else {
+                Ok(Embed::None)
+            }
+        }
+        _ => Ok(Embed::None),
+    }
+}
+
+pub async fn get(Query(info): Query<Parameters>) -> Result<impl Responder, Error> {
+    let url = info.url;
+    let result = CACHE
+        .get_with(url.clone(), async { embed(url).await })
+        .await;
+    result.map(web::Json)
+}
diff --git a/january/src/routes/info.rs b/january/src/routes/info.rs
new file mode 100644
index 00000000..e76d9e91
--- /dev/null
+++ b/january/src/routes/info.rs
@@ -0,0 +1,14 @@
+use actix_web::web;
+use actix_web::Responder;
+use serde::Serialize;
+
+#[derive(Debug, Serialize)]
+pub struct Info {
+    january: &'static str,
+}
+
+pub async fn get() -> impl Responder {
+    web::Json(Info {
+        january: env!("CARGO_PKG_VERSION"),
+    })
+}
diff --git a/january/src/routes/mod.rs b/january/src/routes/mod.rs
new file mode 100644
index 00000000..c4aa207d
--- /dev/null
+++ b/january/src/routes/mod.rs
@@ -0,0 +1,3 @@
+pub mod embed;
+pub mod info;
+pub mod proxy;
diff --git a/january/src/routes/proxy.rs b/january/src/routes/proxy.rs
new file mode 100644
index 00000000..f53e1634
--- /dev/null
+++ b/january/src/routes/proxy.rs
@@ -0,0 +1,43 @@
+use std::time::Duration;
+
+use actix_web::web::Bytes;
+use actix_web::{web::Query, HttpResponse, Responder};
+use serde::Deserialize;
+
+use crate::util::request::{fetch, get_bytes};
+use crate::util::result::Error;
+
+lazy_static! {
+    static ref CACHE: moka::future::Cache<String, Result<Bytes, Error>> =
+        moka::future::Cache::builder()
+            .weigher(|_key, value: &Result<Bytes, Error>| {
+                value.as_ref().map(|bytes| bytes.len() as u32).unwrap_or(1)
+            })
+            .max_capacity(1024 * 1024 * 1024)
+            .time_to_live(Duration::from_secs(60))
+            .build();
+}
+
+#[derive(Deserialize)]
+pub struct Parameters {
+    url: String,
+}
+
+async fn proxy(url: String) -> Result<Bytes, Error> {
+    let (mut resp, mime) = fetch(&url).await?;
+
+    if matches!(mime.type_(), mime::IMAGE | mime::VIDEO) {
+        let bytes = get_bytes(&mut resp).await?;
+        Ok(bytes)
+    } else {
+        Err(Error::NotAllowedToProxy)
+    }
+}
+
+pub async fn get(Query(info): Query<Parameters>) -> Result<impl Responder, Error> {
+    let url = info.url;
+    let result = CACHE
+        .get_with(url.clone(), async { proxy(url).await })
+        .await;
+    result.map(|b| HttpResponse::Ok().body(b))
+}