Add alt text to clip titles (#67)

Author: MinnDevelopment

Cargo.lock

@@ -4,5 +4,11 @@ version = "1.2.5"
 edition.workspace = true
 
 [dependencies.serde]
-version = "1.0"
-features = ["rc", "derive"]
+workspace = true
+features = ["rc", "derive"]
+
+[dependencies.regex]
+workspace = true
+
+[dependencies.once_cell]
+workspace = true

commons/Cargo.toml

@@ -1,3 +1,5 @@
+use once_cell::sync::Lazy;
+use regex::Regex;
 use serde::{Deserialize, Serialize};
 use std::{num::NonZeroU64, ops::Add};
 
@@ -34,3 +36,33 @@ impl Add<u64> for Timestamp {
         unsafe { Self(NonZeroU64::new_unchecked(self.0.get() + rhs)) }
     }
 }
+
+static ALT_TEXT_WHITELIST: Lazy<Regex> = Lazy::new(|| Regex::new(r"\s*(?:[_*`]+|~~+|\|\|+)\s*|(\s+|^)\w+://").unwrap());
+
+pub fn sanitize_markdown(text: &str) -> String {
+    ALT_TEXT_WHITELIST.replace_all(text, " ").trim().to_owned()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_sanitize_markdown() {
+        assert_eq!(
+            sanitize_markdown("simple | text - with $ special & chars %"),
+            "simple | text - with $ special & chars %"
+        );
+        assert_eq!(
+            sanitize_markdown("hello world https://example.com"),
+            "hello world example.com"
+        );
+        assert_eq!(
+            sanitize_markdown("https://example.com hello world"),
+            "example.com hello world"
+        );
+        assert_eq!(sanitize_markdown("~~hello world~~"), "hello world");
+        assert_eq!(sanitize_markdown("*hello world*"), "hello world");
+        assert_eq!(sanitize_markdown("hello ||world||"), "hello world");
+    }
+}

commons/src/util.rs

@@ -1,6 +1,6 @@
 use std::sync::Arc;
 
-use commons::util::Timestamp;
+use commons::util::{sanitize_markdown, Timestamp};
 use discord_api::{config::EventName, WebhookClient};
 use eos::DateTime;
 use serde::{Deserialize, Serialize};
@@ -367,8 +367,9 @@ impl StreamWatcher {
                 .iter()
                 .enumerate()
                 .map(|(i, c)| {
+                    let sanitized_title = sanitize_markdown(&c.title);
                     let mut title = String::with_capacity(30);
-                    for (i, c) in c.title.chars().enumerate() {
+                    for (i, c) in sanitized_title.chars().enumerate() {
                         if i == 26 {
                             title.push_str("...");
                             break;
@@ -381,10 +382,11 @@ impl StreamWatcher {
                         });
                     }
                     format!(
-                        "`{}.` [**{} \u{1F855}**]({}) \u{2022} **{}**\u{00A0}views\n",
+                        "`{}.` [**{} \u{1F855}**]({} '{}') \u{2022} **{}**\u{00A0}views\n",
                         i + 1,
                         title,
                         c.url,
+                        sanitized_title,
                         c.view_count
                     )
                 })