feat: URL preview support

from upstream MR https://gitlab.com/famedly/conduit/-/merge_requests/347
with the following changes (so far):
- remove hardcoded list of allowed hosts (strongly disagree with this,
even if it is desired, it should not be harcoded)
- add more allow config options for granularity via URL contains,
host contains, and domain is (explicit match) for security
- warn if a user is allowing all URLs to be previewed for security reasons
- replace an expect with proper error handling
- bump webpage to 2.0
- improved code style a tad

Co-authored-by: rooot <hey@rooot.gay>
Signed-off-by: rooot <hey@rooot.gay>
Signed-off-by: strawberry <strawberry@puppygock.gay>

src/service/globals/mod.rs

@@ -390,6 +390,18 @@ impl Service<'_> {
         &self.config.emergency_password
     }
 
+    pub fn url_preview_domain_contains_allowlist(&self) -> &Vec<String> {
+        &self.config.url_preview_domain_contains_allowlist
+    }
+
+    pub fn url_preview_domain_explicit_allowlist(&self) -> &Vec<String> {
+        &self.config.url_preview_domain_explicit_allowlist
+    }
+
+    pub fn url_preview_url_contains_allowlist(&self) -> &Vec<String> {
+        &self.config.url_preview_url_contains_allowlist
+    }
+
     pub fn forbidden_room_names(&self) -> &RegexSet {
         &self.config.forbidden_room_names
     }

src/service/media/data.rs

@@ -17,4 +17,15 @@ pub trait Data: Send + Sync {
         width: u32,
         height: u32,
     ) -> Result<(Option<String>, Option<String>, Vec<u8>)>;
+
+    fn remove_url_preview(&self, url: &str) -> Result<()>;
+
+    fn set_url_preview(
+        &self,
+        url: &str,
+        data: &super::UrlPreviewData,
+        timestamp: std::time::Duration,
+    ) -> Result<()>;
+
+    fn get_url_preview(&self, url: &str) -> Option<super::UrlPreviewData>;
 }

src/service/media/mod.rs

@@ -1,7 +1,13 @@
 mod data;
-use std::io::Cursor;
+use std::{
+    collections::HashMap,
+    io::Cursor,
+    sync::{Arc, RwLock},
+    time::SystemTime,
+};
 
 pub(crate) use data::Data;
+use serde::Serialize;
 
 use crate::{services, Result};
 use image::imageops::FilterType;
@@ -9,6 +15,7 @@ use image::imageops::FilterType;
 use tokio::{
     fs::File,
     io::{AsyncReadExt, AsyncWriteExt, BufReader},
+    sync::Mutex,
 };
 
 pub struct FileMeta {
@@ -17,8 +24,43 @@ pub struct FileMeta {
     pub file: Vec<u8>,
 }
 
+#[derive(Serialize, Default)]
+pub struct UrlPreviewData {
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "og:title")
+    )]
+    pub title: Option<String>,
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "og:description")
+    )]
+    pub description: Option<String>,
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "og:image")
+    )]
+    pub image: Option<String>,
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "matrix:image:size")
+    )]
+    pub image_size: Option<usize>,
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "og:image:width")
+    )]
+    pub image_width: Option<u32>,
+    #[serde(
+        skip_serializing_if = "Option::is_none",
+        rename(serialize = "og:image:height")
+    )]
+    pub image_height: Option<u32>,
+}
+
 pub struct Service {
     pub db: &'static dyn Data,
+    pub url_preview_mutex: RwLock<HashMap<String, Arc<Mutex<()>>>>,
 }
 
 impl Service {
@@ -260,6 +302,22 @@ impl Service {
             Ok(None)
         }
     }
+
+    pub async fn get_url_preview(&self, url: &str) -> Option<UrlPreviewData> {
+        self.db.get_url_preview(url)
+    }
+
+    pub async fn remove_url_preview(&self, url: &str) -> Result<()> {
+        // TODO: also remove the downloaded image
+        self.db.remove_url_preview(url)
+    }
+
+    pub async fn set_url_preview(&self, url: &str, data: &UrlPreviewData) -> Result<()> {
+        let now = SystemTime::now()
+            .duration_since(SystemTime::UNIX_EPOCH)
+            .expect("valid system time");
+        self.db.set_url_preview(url, data, now)
+    }
 }
 
 #[cfg(test)]

src/service/mod.rs

@@ -1,6 +1,6 @@
 use std::{
     collections::{BTreeMap, HashMap},
-    sync::{Arc, Mutex},
+    sync::{Arc, Mutex, RwLock},
 };
 
 use lru_cache::LruCache;
@@ -114,7 +114,10 @@ impl Services<'_> {
             account_data: account_data::Service { db },
             admin: admin::Service::build(),
             key_backups: key_backups::Service { db },
-            media: media::Service { db },
+            media: media::Service {
+                db,
+                url_preview_mutex: RwLock::new(HashMap::new()),
+            },
             sending: sending::Service::build(db, &config),
 
             globals: globals::Service::load(db, config)?,