feat(webapi): impl atom feed and use tag URI for feed id

Cargo.lock

@@ -317,6 +317,7 @@ dependencies = [
  "expect-test",
  "futures-util",
  "hex",
+ "html-escape",
  "http-body-util",
  "humantime",
  "nix",

blahd/Cargo.toml

@@ -11,6 +11,7 @@ clap = { version = "4", features = ["derive"] }
 ed25519-dalek = "2"
 futures-util = "0.3"
 hex = { version = "0.4", features = ["serde"] }
+html-escape = "0.2"
 http-body-util = "0.1"
 humantime = "2"
 parking_lot = "0.12" # Maybe no better performance, just that we hate poisoning. ¯\_(ツ)_/¯

blahd/src/feed.rs

@@ -1,15 +1,19 @@
 //! Room feed generation.
+use std::fmt;
 use std::num::NonZero;
 use std::time::{Duration, SystemTime};
 
 use axum::http::header;
 use axum::response::{IntoResponse, Response};
 use axum::Json;
-use blah_types::{SignedChatMsg, WithMsgId};
+use blah_types::{Id, SignedChatMsg, WithMsgId};
 use serde::{Deserialize, Serialize};
 use url::Url;
 
+use crate::id::timestamp_of_id;
+
 const JSON_FEED_MIME: &str = "application/feed+json";
+const ATOM_FEED_MIME: &str = "application/atom+xml";
 
 #[derive(Debug, Clone, Deserialize)]
 #[serde(default)]
@@ -25,10 +29,44 @@ impl Default for Config {
     }
 }
 
+pub struct FeedData {
+    pub rid: Id,
+    pub title: String,
+    pub msgs: Vec<WithMsgId<SignedChatMsg>>,
+    pub self_url: Url,
+    pub next_url: Option<Url>,
+}
+
+pub trait FeedType {
+    fn to_feed_response(data: FeedData) -> Response;
+}
+
+fn timestamp_to_rfc3339(timestamp: u64) -> impl fmt::Display {
+    humantime::format_rfc3339(SystemTime::UNIX_EPOCH + Duration::from_secs(timestamp))
+}
+
+/// See:
+/// - <https://taguri.org/>
+/// - <https://www.rfc-editor.org/rfc/rfc4151>
+/// - <https://en.wikipedia.org/wiki/Tag_URI_scheme>
+#[derive(Clone, Copy)]
+struct TagUri<'a>(&'a Url, &'a str, Id);
+
+impl fmt::Display for TagUri<'_> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let Self(url, typ, id) = *self;
+        let domain = url.domain().expect("base_url must have domain");
+        let id_time = timestamp_to_rfc3339(timestamp_of_id(id)).to_string();
+        // Because github.com also only uses year.
+        let id_year = &id_time[..4];
+        write!(f, "tag:{domain},{id_year}:blah/{typ}/{id}")
+    }
+}
+
 /// Ref: <https://www.jsonfeed.org/version/1.1/>
 #[derive(Debug, Serialize)]
 #[serde(tag = "version", rename = "https://jsonfeed.org/version/1.1")]
-struct JsonFeed {
+pub struct JsonFeed {
     title: String,
     feed_url: Url,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -52,35 +90,125 @@ struct JsonFeedAuthor {
     name: String,
 }
 
-pub fn to_json_feed(
-    title: String,
-    msgs: Vec<WithMsgId<SignedChatMsg>>,
-    self_url: Url,
-    next_url: Option<Url>,
-) -> Response {
-    let items = msgs
-        .into_iter()
-        .map(|WithMsgId { cid, msg }| {
-            let time = SystemTime::UNIX_EPOCH + Duration::from_secs(msg.signee.timestamp);
-            let author = JsonFeedAuthor {
-                // TODO: Retrieve id_url as name.
-                name: msg.signee.user.id_key.to_string(),
-            };
-            JsonFeedItem {
-                id: cid.to_string(),
-                content_html: msg.signee.payload.rich_text.html().to_string(),
-                date_published: humantime::format_rfc3339(time).to_string(),
-                authors: (author,),
-            }
-        })
-        .collect::<Vec<_>>();
+impl FeedType for JsonFeed {
+    fn to_feed_response(data: FeedData) -> Response {
+        let FeedData {
+            title,
+            msgs,
+            self_url,
+            next_url,
+            ..
+        } = data;
+        let items = msgs
+            .into_iter()
+            .map(|WithMsgId { cid, msg }| {
+                let author = JsonFeedAuthor {
+                    // TODO: Retrieve id_url as name.
+                    name: msg.signee.user.id_key.to_string(),
+                };
+                JsonFeedItem {
+                    id: TagUri(&self_url, "msg", cid).to_string(),
+                    content_html: msg.signee.payload.rich_text.html().to_string(),
+                    date_published: timestamp_to_rfc3339(msg.signee.timestamp).to_string(),
+                    authors: (author,),
+                }
+            })
+            .collect::<Vec<_>>();
 
-    let feed = JsonFeed {
-        title,
-        items,
-        feed_url: self_url,
-        next_url,
-    };
+        let feed = JsonFeed {
+            title,
+            items,
+            feed_url: self_url,
+            next_url,
+        };
 
-    ([(header::CONTENT_TYPE, JSON_FEED_MIME)], Json(feed)).into_response()
+        ([(header::CONTENT_TYPE, JSON_FEED_MIME)], Json(feed)).into_response()
+    }
+}
+
+pub struct AtomFeed(FeedData);
+
+// We write XML manually here, because existing crates (`feed-rs` and `atom_syndication`)
+// pull in heavy `quick_xml` and `chrono` which overdoes too much.
+//
+// Ref: <https://validator.w3.org/feed/docs/atom.html>
+impl fmt::Display for AtomFeed {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        use html_escape::{encode_quoted_attribute, encode_text};
+
+        let FeedData {
+            rid,
+            title,
+            msgs,
+            self_url,
+            next_url,
+        } = &self.0;
+
+        let room_id = TagUri(self_url, "room", *rid);
+        let esc_room_title = encode_text(title.trim());
+        // TODO: This should track the latest msg even for non-first page.
+        let feed_timestamp =
+            timestamp_to_rfc3339(msgs.first().map_or(0, |msg| msg.msg.signee.timestamp));
+        let esc_self_url = encode_quoted_attribute(self_url.as_str());
+
+        write!(
+            f,
+            // NB. XML requires no-newline at start.
+            r#"<?xml version="1.0" encoding="utf-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom">
+  <id>{room_id}</id>
+  <title>{esc_room_title}</title>
+  <updated>{feed_timestamp}</updated>
+  <link rel="self" type="application/atom+xml" href="{esc_self_url}"/>
+"#
+        )?;
+        if let Some(next_url) = &next_url {
+            let esc_next_url = encode_quoted_attribute(next_url.as_str());
+            // <https://www.rfc-editor.org/rfc/rfc5005>
+            writeln!(
+                f,
+                r#"  <link rel="next" type="application/atom+xml" href="{esc_next_url}"/>"#
+            )?;
+        }
+        // TODO: HTML links.
+
+        for msg in msgs {
+            let content = &msg.msg.signee.payload.rich_text;
+            let msg_id = TagUri(self_url, "msg", msg.cid);
+            let plain_text = content.plain_text().to_string();
+            let esc_msg_title =
+                encode_text(plain_text.lines().next().unwrap_or("(untitled)").trim());
+            let msg_timestamp = timestamp_to_rfc3339(msg.msg.signee.timestamp);
+            let author = msg.msg.signee.user.id_key.to_string();
+            let content = content.html().to_string();
+            let esc_content = encode_text(&content);
+            write!(
+                f,
+                r#"
+  <entry>
+    <id>{msg_id}</id>
+    <title type="text">{esc_msg_title}</title>
+    <published>{msg_timestamp}</published>
+    <updated>{msg_timestamp}</updated>
+    <author><name>{author}</name></author>
+    <content type="html">{esc_content}</content>
+  </entry>
+    "#
+            )?;
+        }
+
+        write!(
+            f,
+            r#"
+</feed>
+"#
+        )
+    }
+}
+
+impl FeedType for AtomFeed {
+    fn to_feed_response(data: FeedData) -> Response {
+        let body = AtomFeed(data).to_string();
+        ([(header::CONTENT_TYPE, ATOM_FEED_MIME)], body).into_response()
+    }
 }

blahd/src/id.rs

@@ -5,6 +5,10 @@ use std::time::SystemTime;
 
 use blah_types::Id;
 
+pub fn timestamp_of_id(id: Id) -> u64 {
+    (id.0 as u64 >> 16) / 1000
+}
+
 pub trait IdExt {
     fn gen() -> Self;
     fn gen_peer_chat_rid() -> Self;

blahd/src/lib.rs

@@ -17,6 +17,7 @@ use blah_types::{
     X_BLAH_NONCE,
 };
 use database::{Transaction, TransactionOps};
+use feed::FeedData;
 use id::IdExt;
 use middleware::{Auth, MaybeAuth, ResultExt as _, SignedJson};
 use parking_lot::Mutex;
@@ -69,6 +70,9 @@ fn de_base_url<'de, D: Deserializer<'de>>(de: D) -> Result<Url, D::Error> {
             "base_url must be able to be a base",
         ));
     }
+    if url.domain().is_none() {
+        return Err(serde::de::Error::custom("base_url must have a domain"));
+    }
     Ok(url)
 }
 
@@ -137,7 +141,8 @@ pub fn router(st: Arc<AppState>) -> Router {
         .route("/room", get(room_list))
         .route("/room/create", post(room_create))
         .route("/room/:rid", get(room_get_metadata).delete(room_delete))
-        .route("/room/:rid/feed.json", get(room_get_feed))
+        .route("/room/:rid/feed.json", get(room_get_feed::<feed::JsonFeed>))
+        .route("/room/:rid/feed.atom", get(room_get_feed::<feed::AtomFeed>))
         .route("/room/:rid/msg", get(room_msg_list).post(room_msg_post))
         .route("/room/:rid/msg/:cid/seen", post(room_msg_mark_seen))
         .route("/room/:rid/admin", post(room_admin))
@@ -424,7 +429,7 @@ async fn room_get_metadata(
     }))
 }
 
-async fn room_get_feed(
+async fn room_get_feed<FT: feed::FeedType>(
     st: ArcState,
     R(OriginalUri(req_uri), _): RE<OriginalUri>,
     R(Path(rid), _): RE<Path<Id>>,
@@ -470,7 +475,13 @@ async fn room_get_feed(
         next_url
     });
 
-    Ok(feed::to_json_feed(title, msgs, self_url, next_url))
+    Ok(FT::to_feed_response(FeedData {
+        rid,
+        title,
+        msgs,
+        self_url,
+        next_url,
+    }))
 }
 
 /// Get room messages with pagination parameters,

blahd/tests/webapi.rs

@@ -726,15 +726,17 @@ async fn room_chat_post_read(server: Server) {
 }
 
 #[rstest]
+#[case::json("json")]
+#[case::atom("atom")]
 #[tokio::test]
-async fn room_feed(server: Server) {
+async fn room_feed(server: Server, #[case] typ: &'static str) {
     // Only public readable rooms provides feed. Not even for public joinable ones.
     let rid_need_join = server
         .create_room(&ALICE, RoomAttrs::PUBLIC_JOINABLE, "not so public")
         .await
         .unwrap();
     server
-        .get::<NoContent>(&format!("/room/{rid_need_join}/feed.json"), None)
+        .get::<NoContent>(&format!("/room/{rid_need_join}/feed.{typ}"), None)
         .await
         .expect_api_err(StatusCode::NOT_FOUND, "room_not_found");
 
@@ -754,30 +756,46 @@ async fn room_feed(server: Server) {
     let cid2 = server.post_chat(rid, &BOB, "b1").await.unwrap().cid;
     server.post_chat(rid, &BOB, "b2").await.unwrap();
 
-    let feed = server
-        .get::<serde_json::Value>(&format!("/room/{rid}/feed.json"), None)
-        .await
-        .unwrap();
-    // TODO: Ideally we should assert on the result, but it contains time and random id currently.
-    assert_eq!(feed["title"].as_str().unwrap(), "public");
-    assert_eq!(feed["items"].as_array().unwrap().len(), 2);
-    let feed_url = format!("{BASE_URL}/_blah/room/{rid}/feed.json");
-    assert_eq!(feed["feed_url"].as_str().unwrap(), feed_url,);
-    assert_eq!(
-        feed["next_url"].as_str().unwrap(),
-        format!("{feed_url}?skipToken={cid2}&top=2"),
-    );
+    if typ == "json" {
+        let feed = server
+            .get::<serde_json::Value>(&format!("/room/{rid}/feed.json"), None)
+            .await
+            .unwrap();
+        // TODO: Ideally we should assert on the result, but it contains time and random id currently.
+        assert_eq!(feed["title"].as_str().unwrap(), "public");
+        assert_eq!(feed["items"].as_array().unwrap().len(), 2);
+        let feed_url = format!("{BASE_URL}/_blah/room/{rid}/feed.json");
+        assert_eq!(feed["feed_url"].as_str().unwrap(), feed_url,);
+        assert_eq!(
+            feed["next_url"].as_str().unwrap(),
+            format!("{feed_url}?skipToken={cid2}&top=2"),
+        );
 
-    let feed2 = server
-        .get::<serde_json::Value>(
-            &format!("/room/{rid}/feed.json?skipToken={cid2}&top=2"),
-            None,
-        )
-        .await
-        .unwrap();
-    let items = feed2["items"].as_array().unwrap();
-    assert_eq!(items.len(), 1);
-    assert_eq!(items[0]["content_html"].as_str().unwrap(), "a");
+        let feed2 = server
+            .get::<serde_json::Value>(
+                &format!("/room/{rid}/feed.json?skipToken={cid2}&top=2"),
+                None,
+            )
+            .await
+            .unwrap();
+        let items = feed2["items"].as_array().unwrap();
+        assert_eq!(items.len(), 1);
+        assert_eq!(items[0]["content_html"].as_str().unwrap(), "a");
+    } else {
+        let resp = server
+            .client
+            .get(server.url(format!("/room/{rid}/feed.atom")))
+            .send()
+            .await
+            .unwrap()
+            .error_for_status()
+            .unwrap()
+            .text()
+            .await
+            .unwrap();
+        assert!(resp.starts_with(r#"<?xml version="1.0" encoding="utf-8"?>"#));
+        assert_eq!(resp.matches("<entry>").count(), 2);
+    }
 }
 
 #[rstest]