Remove emoji from clip titles

Author: MinnDevelopment

Cargo.lock

@@ -13,6 +13,8 @@ lru = "0.8"
 tracing = "0.1"
 tracing-subscriber = "0.3"
 async-trait = "0.1"
+emojis = "0.6"
+fxhash = "0.2"
 
 [dependencies.once_cell]
 version = "1"

Cargo.toml

@@ -1,3 +1,5 @@
+use fxhash::FxHashSet;
+use once_cell::sync::Lazy;
 use serde::{Deserialize, Serialize};
 use std::{num::NonZeroU64, ops::Add};
 
@@ -33,3 +35,37 @@ impl Add<u64> for Timestamp {
         unsafe { Self(NonZeroU64::new_unchecked(self.0.get() + rhs)) }
     }
 }
+
+const EMOJI_ARRAY: [u32; 9280] = include!("../emojis.txt");
+
+static EMOJI_SET: Lazy<FxHashSet<u32>> = Lazy::new(|| EMOJI_ARRAY.iter().copied().collect());
+
+pub fn strip_emoji<'a>(text: &'a str) -> String {
+    text.chars()
+        .filter(|&c| !EMOJI_SET.contains(&(c as u32)))
+        .fold(String::with_capacity(text.len()), |mut acc, c| {
+            if c.is_whitespace() || c.is_control() {
+                acc.push(' ');
+            } else {
+                acc.push(c);
+            }
+            acc
+        })
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_strip_emoji() {
+        assert_eq!(strip_emoji("no emoji"), "no emoji");
+        assert_eq!(strip_emoji("simple smiley 😄"), "simple smiley ");
+        assert_eq!(strip_emoji("with skin tone 👍🏽"), "with skin tone ");
+        assert_eq!(strip_emoji("basic flag 🇩🇪 "), "basic flag  ");
+        assert_eq!(
+            strip_emoji("simple | text - with $ special & chars %"),
+            "simple | text - with $ special & chars %"
+        );
+    }
+}

emojis.txt

@@ -12,7 +12,7 @@ use crate::{
     discord::WebhookClient,
     error::{AsyncError as Error, RequestError},
     twitch::{Game, Stream, TwitchClient, VideoDuration},
-    util::Timestamp,
+    util::{Timestamp, strip_emoji},
 };
 
 const fn split_duration(secs: u32) -> (u8, u8, u8) {
@@ -385,7 +385,7 @@ impl StreamWatcher {
                     format!(
                         "`{}.` [**{} \u{1F855}**]({}) \u{2022} **{}**\u{00A0}views\n",
                         i + 1,
-                        title,
+                        strip_emoji(&title),
                         c.url,
                         c.view_count
                     )