improve category tags

Cargo.toml

@@ -1,13 +1,13 @@
 [package]
 name = "disbahn"
 version = "0.1.0"
-edition = "2021"
+edition = "2024"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
 rss = "2.0.12"
-reqwest = {version = "0.12.22", default-features = false, features = ["rustls-tls"]}
+reqwest = "0.13.4"
 tokio = {version = "1.47.0", features = ["rt-multi-thread", "signal"]}
 lazy-regex = "3.4.1"
 chrono = "0.4.41"
@@ -19,6 +19,7 @@ dotenvy = "0.15.7"
 diesel = { version = "2.2.12", features = ["sqlite", "chrono"] }
 diesel_migrations = "2.2.0"
 getset = "0.1.6"
+itertools = "0.15.0"
 
 [dependencies.serenity]
 version = "0.12.4"

migrations/2026-07-28-124138-0000_create_post_versions/down.sql

@@ -0,0 +1 @@
+DROP TABLE post_versions;

migrations/2026-07-28-124138-0000_create_post_versions/up.sql

@@ -0,0 +1,6 @@
+CREATE TABLE post_versions (
+    announcement_id TEXT NOT NULL,
+    publish_date TIMESTAMP NOT NULL,
+    title TEXT NOT NULL,
+    PRIMARY KEY (announcement_id, publish_date, title)
+);

src/database/models.rs

@@ -49,3 +49,24 @@ impl<'a> NewPost<'a> {
         }
     }
 }
+
+#[derive(Queryable, Insertable, Getters)]
+#[diesel(table_name = post_versions)]
+pub struct PostVersion {
+    #[getset(get = "pub")]
+    announcement_id: String,
+    #[getset(get = "pub")]
+    publish_date: NaiveDateTime,
+    #[getset(get = "pub")]
+    title: String,
+}
+
+impl PostVersion {
+    pub fn new(announcement_id: String, publish_date: NaiveDateTime, title: String) -> Self {
+        Self {
+            announcement_id,
+            publish_date,
+            title,
+        }
+    }
+}

src/database/schema.rs

@@ -1,5 +1,13 @@
 // @generated automatically by Diesel CLI.
 
+diesel::table! {
+    post_versions (announcement_id, publish_date, title) {
+        announcement_id -> Text,
+        publish_date -> Timestamp,
+        title -> Text,
+    }
+}
+
 diesel::table! {
     posts (announcement_id, webhook_id) {
         announcement_id -> Text,
@@ -8,3 +16,5 @@ diesel::table! {
         last_updated -> Timestamp,
     }
 }
+
+diesel::allow_tables_to_appear_in_same_query!(post_versions, posts,);

src/hint.txt

@@ -1 +0,0 @@
--# **Hinweis:** Die aus dem RSS-Feed von zuginfo.nrw stammenden Beginn- und Endzeiten scheinen oft inakkurat zu sein und sind daher mit Vorsicht zu genießen.

src/lib.rs

@@ -1,18 +1,185 @@
 pub mod database;
 
-use crate::database::models::{NewPost, Post};
-use crate::database::schema::posts::dsl::posts;
+use std::borrow::Cow;
+use std::fmt::Write as _;
+
 use crate::database::Database;
-use anyhow::{anyhow, Context};
-use chrono::{DateTime, NaiveDateTime, TimeZone};
+use crate::database::models::{NewPost, Post, PostVersion};
+use crate::database::schema::post_versions;
+use anyhow::{Context, Ok, anyhow};
+use chrono::{DateTime, NaiveDateTime, TimeZone, Utc};
 use diesel::{ExpressionMethods, OptionalExtension, QueryDsl, RunQueryDsl};
-use lazy_regex::{regex, Lazy, Regex};
+use itertools::Itertools;
+use lazy_regex::{Lazy, Regex, regex};
 use log::{debug, error, info};
 use reqwest::IntoUrl;
-use serenity::all::{CreateEmbed, CreateEmbedFooter, EditWebhookMessage, ExecuteWebhook};
+use serenity::all::{
+    CreateEmbed, CreateEmbedAuthor, CreateEmbedFooter, EditWebhookMessage, ExecuteWebhook,
+};
 use serenity::http::Http;
 use serenity::model::webhook::Webhook;
 
+struct Item<'a> {
+    guid: &'a str,
+    title: &'a str,
+    link: &'a str,
+    validity_begin: DateTime<Utc>,
+    validity_end: DateTime<Utc>,
+    icon: &'a str,
+    description: &'a str,
+    pub_datetime: DateTime<Utc>,
+}
+
+impl Item<'_> {
+    fn parse_validity_time(input: &str) -> anyhow::Result<DateTime<Utc>> {
+        let naive = NaiveDateTime::parse_from_str(input, "%Y-%m-%d %H:%M:%S")?;
+        let datetime = chrono_tz::Europe::Berlin
+            .from_local_datetime(&naive)
+            .unwrap()
+            .to_utc();
+        Ok(datetime)
+    }
+
+    fn html_to_discord_markdown(input: &str) -> String {
+        static RE_TIMES: &Lazy<Regex> = regex!(r#"^.*?<br\s*/>\s*<br\s*/>"#i);
+        static RE_BOLD: &Lazy<Regex> = regex!(r#"<b\s*>((.|\n)*?)</b\s*>"#i);
+        static RE_ITALIC: &Lazy<Regex> = regex!(r#"<i\s*>((.|\n)*?)</i\s*>"#i);
+        static RE_STRIKETHROUGH: &Lazy<Regex> = regex!(r#"<s\s*>((.|\n)*?)</s\s*>"#i);
+        static RE_NEWLINE: &Lazy<Regex> = regex!(r#"<br\s*/?>"#i);
+
+        let input = RE_TIMES.replace(input, "");
+        let input = RE_BOLD.replace_all(&input, "**$1**");
+        let input = RE_ITALIC.replace_all(&input, "*$1*");
+        let input = RE_STRIKETHROUGH.replace_all(&input, "~~$1~~");
+        let input = RE_NEWLINE.replace_all(&input, "\n");
+        input.to_string()
+    }
+
+    fn tag(&self) -> Cow<'static, str> {
+        match self.icon {
+            "HIM0" => "\u{2139}\u{FE0F} Information".into(),
+            "HIM1" => "\u{1F6A7} Bauarbeiten".into(),
+            "HIM2" => "\u{26A0}\u{FE0F} Störung".into(),
+            "HIM3" => "\u{270A} Streik".into(),
+            other => format!("\u{1F937} Unbekannte Kategorie {other:?}").into(),
+        }
+    }
+
+    fn colour(&self) -> u32 {
+        match self.icon {
+            "HIM0" => 0x154889,
+            "HIM1" => 0xf5c211,
+            "HIM2" => 0xc1121c,
+            "HIM3" => 0xff6a00,
+            _ => 0xffffff,
+        }
+    }
+
+    fn to_embed(&self, mut previous_versions: &[PostVersion]) -> anyhow::Result<CreateEmbed> {
+        const FOOTER_ICON_URL: &str = "https://www.zuginfo.nrw/img/customer/apple-touch-icon.png";
+
+        let mut embed = CreateEmbed::new()
+            .title(self.title)
+            .url(self.link)
+            .author(CreateEmbedAuthor::new(self.tag()))
+            .colour(self.colour())
+            .description(Self::html_to_discord_markdown(self.description))
+            .field(
+                "Beginn:",
+                format!("<t:{}:f>", self.validity_begin.timestamp()),
+                true,
+            )
+            .field(
+                "Ende:",
+                format!("<t:{}:f>", self.validity_end.timestamp()),
+                true,
+            )
+            .timestamp(self.pub_datetime)
+            .footer(CreateEmbedFooter::new("zuginfo.nrw").icon_url(FOOTER_ICON_URL));
+
+        // Remove trailing titles matching the current title
+        while let Some(last) = previous_versions.last()
+            && last.title() == self.title
+        {
+            previous_versions = &previous_versions[..previous_versions.len() - 1]
+        }
+
+        let mut previous_titles_str = String::new();
+        for version in previous_versions
+            .iter()
+            .dedup_by(|&x, &y| x.title() == y.title())
+        {
+            writeln!(
+                &mut previous_titles_str,
+                "-# * <t:{}:s> {}",
+                version.publish_date().and_utc().timestamp(),
+                version.title()
+            )
+            .unwrap();
+        }
+
+        if !previous_titles_str.is_empty() {
+            embed = embed.field("Vorherige Titel:", previous_titles_str, false);
+        }
+
+        Ok(embed)
+    }
+}
+
+impl<'a> TryFrom<&'a rss::Item> for Item<'a> {
+    type Error = anyhow::Error;
+    fn try_from(rss_item: &'a rss::Item) -> Result<Self, Self::Error> {
+        let categories = rss_item.categories();
+
+        let guid = rss_item.guid().ok_or(anyhow!("Missing GUID"))?.value();
+
+        let title = rss_item.title().ok_or(anyhow!("Missing title"))?;
+        let link = rss_item.link().ok_or(anyhow!("Missing link"))?;
+
+        let validity_begin = &categories
+            .iter()
+            .find(|c| c.domain() == Some("validityBegin"))
+            .ok_or(anyhow!("Missing validityBegin category"))?
+            .name;
+        let validity_begin = Self::parse_validity_time(validity_begin)?;
+
+        let validity_end = &categories
+            .iter()
+            .find(|c| c.domain() == Some("validityEnd"))
+            .ok_or(anyhow!("Missing validityEnd category"))?
+            .name;
+        let validity_end = Self::parse_validity_time(validity_end)?;
+
+        let icon = categories
+            .iter()
+            .find(|c| c.domain() == Some("icon"))
+            .map_or("", |c| c.name());
+
+        let description = rss_item
+            .description()
+            .ok_or(anyhow!("Missing description"))?;
+
+        let pub_date_str = rss_item
+            .pub_date()
+            .ok_or(anyhow!("Missing publication date"))?;
+        let pub_datetime = DateTime::parse_from_rfc2822(pub_date_str)
+            .with_context(|| format!("Unable to parse publication date string {pub_date_str:?}"))?
+            .naive_utc()
+            .and_utc();
+
+        Ok(Self {
+            guid,
+            title,
+            link,
+            validity_begin,
+            validity_end,
+            icon,
+            description,
+            pub_datetime,
+        })
+    }
+}
+
 pub struct DisbahnClient {
     database: Database,
     webhook: Webhook,
@@ -36,103 +203,16 @@ impl DisbahnClient {
         Ok(channel)
     }
 
-    fn validity_time_to_timestamp(input: &str) -> anyhow::Result<i64> {
-        let naive = NaiveDateTime::parse_from_str(input, "%Y-%m-%d %H:%M:%S")?;
-        let timestamp = chrono_tz::Europe::Berlin
-            .from_local_datetime(&naive)
-            .unwrap()
-            .timestamp();
-        Ok(timestamp)
-    }
-
-    fn html_to_discord_markdown(input: &str) -> String {
-        static RE_TIMES: &Lazy<Regex> = regex!(r#"^.*<br\s*/>\s*<br\s*/>"#i);
-        static RE_BOLD: &Lazy<Regex> = regex!(r#"<b\s*>((.|\n)*?)</b\s*>"#i);
-        static RE_ITALIC: &Lazy<Regex> = regex!(r#"<i\s*>((.|\n)*?)</i\s*>"#i);
-        static RE_STRIKETHROUGH: &Lazy<Regex> = regex!(r#"<s\s*>((.|\n)*?)</s\s*>"#i);
-        static RE_NEWLINE: &Lazy<Regex> = regex!(r#"<br\s*/?>"#i);
-
-        let input = RE_TIMES.replace(input, "");
-        let input = RE_BOLD.replace_all(&input, "**$1**");
-        let input = RE_ITALIC.replace_all(&input, "*$1*");
-        let input = RE_STRIKETHROUGH.replace_all(&input, "~~$1~~");
-        let input = RE_NEWLINE.replace_all(&input, "\n");
-        input.to_string()
-    }
-
-    fn icon_name_to_url(name: &str) -> &str {
-        match name {
-            "HIM1" => "https://upload.wikimedia.org/wikipedia/commons/thumb/a/a9/Zeichen_123_-_Arbeitsstelle%2C_StVO_2013.svg/273px-Zeichen_123_-_Arbeitsstelle%2C_StVO_2013.svg.png",
-            "HIM2" => "https://upload.wikimedia.org/wikipedia/commons/thumb/0/02/Zeichen_101_-_Gefahrstelle%2C_StVO_1970.svg/273px-Zeichen_101_-_Gefahrstelle%2C_StVO_1970.svg.png",
-            "HIM3" => "https://upload.wikimedia.org/wikipedia/commons/thumb/6/6c/Strike_worker.svg/314px-Strike_worker.svg.png",
-            _ => "https://upload.wikimedia.org/wikipedia/commons/thumb/5/56/Zeichen_365-61_-_Informationsstelle%2C_StVO_2013.svg/240px-Zeichen_365-61_-_Informationsstelle%2C_StVO_2013.svg.png",
-        }
-    }
-
-    fn icon_name_to_colour(name: &str) -> u32 {
-        match name {
-            "HIM1" => 0xf5c211,
-            "HIM2" | "HIM3" => 0xc1121c,
-            _ => 0x154889,
-        }
-    }
-
-    fn item_to_embed(item: &rss::Item) -> anyhow::Result<CreateEmbed> {
-        const FOOTER_ICON_URL: &str = "https://www.zuginfo.nrw/img/customer/apple-touch-icon.png";
-
-        let categories = item.categories();
-
-        let title = item.title().ok_or(anyhow!("Missing title"))?;
-        let link = item.link().ok_or(anyhow!("Missing link"))?;
-
-        let validity_begin = &categories
-            .iter()
-            .find(|c| c.domain() == Some("validityBegin"))
-            .ok_or(anyhow!("Missing validityBegin category"))?
-            .name;
-        let validity_begin = Self::validity_time_to_timestamp(validity_begin)?;
-
-        let validity_end = &categories
-            .iter()
-            .find(|c| c.domain() == Some("validityEnd"))
-            .ok_or(anyhow!("Missing validityEnd category"))?
-            .name;
-        let validity_end = Self::validity_time_to_timestamp(validity_end)?;
-
-        let icon = categories
-            .iter()
-            .find(|c| c.domain() == Some("icon"))
-            .map_or("", |c| c.name());
-        let icon_url = Self::icon_name_to_url(icon);
-
-        let description = Self::html_to_discord_markdown(
-            item.description().ok_or(anyhow!("Missing description"))?,
-        );
-
-        let pub_date_str = item.pub_date().ok_or(anyhow!("Missing publication date"))?;
-        let pub_datetime = DateTime::parse_from_rfc2822(pub_date_str)
-            .with_context(|| format!("Unable to parse publication date string {pub_date_str:?}"))?
-            .naive_utc()
-            .and_utc();
-
-        let embed = CreateEmbed::new()
-            .title(title)
-            .url(link)
-            .thumbnail(icon_url)
-            .colour(Self::icon_name_to_colour(icon))
-            .description(description)
-            .field("Beginn:", format!("<t:{validity_begin}:F>"), true)
-            .field("Ende:", format!("<t:{validity_end}:F>"), true)
-            .field("", include_str!("hint.txt"), false)
-            .timestamp(pub_datetime)
-            .footer(
-                CreateEmbedFooter::new(
-                    "Quelle: https://zuginfo.nrw/ \u{2013} Alle Angaben ohne Gewehr \u{1F52B}",
-                )
-                .icon_url(FOOTER_ICON_URL),
-            );
-
-        Ok(embed)
+    async fn add_post_version_entry(&mut self, item: &Item<'_>) -> anyhow::Result<()> {
+        diesel::insert_into(post_versions::table)
+            .values(PostVersion::new(
+                item.guid.to_string(),
+                item.pub_datetime.naive_utc(),
+                item.title.to_string(),
+            ))
+            .execute(self.database.conn())
+            .with_context(|| "Error inserting new post version into database")?;
+        Ok(())
     }
 
     pub async fn refresh(&mut self) -> anyhow::Result<()> {
@@ -154,27 +234,31 @@ impl DisbahnClient {
         Ok(())
     }
 
-    async fn refresh_item(&mut self, item: &rss::Item) -> anyhow::Result<()> {
-        use crate::database::schema::posts::{self, dsl};
+    async fn refresh_item(&mut self, rss_item: &rss::Item) -> anyhow::Result<()> {
+        use crate::database::schema::{post_versions, posts};
 
-        let guid = item.guid().ok_or(anyhow!("Missing GUID"))?.value();
-        let pub_date_str = item.pub_date().ok_or(anyhow!("Missing publication date"))?;
-        let pub_datetime = DateTime::parse_from_rfc2822(pub_date_str)
-            .with_context(|| format!("Unable to parse publication date string {pub_date_str:?}"))?
-            .naive_utc()
-            .and_utc();
+        let item: Item = rss_item.try_into()?;
 
-        let existing_post: Option<Post> = posts
-            .filter(dsl::webhook_id.eq(i64::from_le_bytes(self.webhook.id.get().to_le_bytes())))
-            .filter(dsl::announcement_id.eq(guid))
+        let existing_post: Option<Post> = posts::dsl::posts
+            .filter(
+                posts::dsl::webhook_id.eq(i64::from_le_bytes(self.webhook.id.get().to_le_bytes())),
+            )
+            .filter(posts::dsl::announcement_id.eq(item.guid))
             .first(self.database.conn())
             .optional()
             .with_context(|| "Error loading posts from database")?;
 
         if let Some(existing_post) = existing_post {
-            if existing_post.last_updated().and_utc() < pub_datetime {
-                info!("Updated item: {guid}");
-                let embed = Self::item_to_embed(item)?;
+            if existing_post.last_updated().and_utc() < item.pub_datetime {
+                info!("Updated item with id {}", item.guid);
+
+                let previous_posts = post_versions::dsl::post_versions
+                    .filter(post_versions::dsl::announcement_id.eq(item.guid))
+                    .order_by(post_versions::publish_date)
+                    .load::<PostVersion>(self.database.conn())
+                    .with_context(|| "Error loading previous posts from database")?;
+
+                let embed = item.to_embed(&previous_posts)?;
                 self.webhook
                     .edit_message(
                         &self.http,
@@ -184,17 +268,19 @@ impl DisbahnClient {
                     .await
                     .with_context(|| "Failed to edit message")?;
 
-                diesel::update(posts.find((
-                    guid,
+                diesel::update(posts::dsl::posts.find((
+                    item.guid,
                     i64::from_le_bytes(self.webhook.id.get().to_le_bytes()),
                 )))
-                .set(dsl::last_updated.eq(pub_datetime.naive_utc()))
+                .set(posts::dsl::last_updated.eq(item.pub_datetime.naive_utc()))
                 .execute(self.database.conn())
                 .with_context(|| "Error updating post in database")?;
+
+                self.add_post_version_entry(&item).await?;
             }
         } else {
-            info!("New item: {guid}");
-            let embed = Self::item_to_embed(item)?;
+            info!("New item with id {}", item.guid);
+            let embed = item.to_embed(&[])?;
             let message = self
                 .webhook
                 .execute(&self.http, true, ExecuteWebhook::new().embed(embed))
@@ -204,13 +290,15 @@ impl DisbahnClient {
 
             diesel::insert_into(posts::table)
                 .values(NewPost::new(
-                    guid,
+                    item.guid,
                     self.webhook.id,
                     message.id,
-                    pub_datetime.naive_utc(),
+                    item.pub_datetime.naive_utc(),
                 ))
                 .execute(self.database.conn())
                 .with_context(|| "Error inserting new post into database")?;
+
+            self.add_post_version_entry(&item).await?;
         }
         Ok(())
     }

src/main.rs

@@ -1,6 +1,6 @@
-use anyhow::{anyhow, Context};
-use disbahn::database::Database;
+use anyhow::{Context, anyhow};
 use disbahn::DisbahnClient;
+use disbahn::database::Database;
 use log::error;
 use serenity::http::Http;
 use std::env;
@@ -15,11 +15,11 @@ fn env_var(name: &str) -> anyhow::Result<String> {
 
 #[tokio::main]
 async fn main() -> anyhow::Result<()> {
-    if let Err(err) = dotenvy::dotenv() {
-        if !err.not_found() {
+    if let Err(err) = dotenvy::dotenv()
+        && !err.not_found()
+    {
         return Err(err).context("Unable to load .env file");
     }
-    }
 
     env_logger::builder()
         .filter_module(module_path!(), log::LevelFilter::Debug)
@@ -31,7 +31,7 @@ async fn main() -> anyhow::Result<()> {
         Some(s) => {
             return Err(anyhow!(format!(
                 "invalid argument `{s}`; the only allowed argument is `daemon`"
-            )))
+            )));
         }
         None => false,
     };