fix truncate on unicode

fuck unicode all my homies stan ascii

3 files changed, 20 insertions, 10 deletions

diff --git a/Cargo.lock b/Cargo.lock
index ef303f4..f892f6e 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -228,6 +228,7 @@ dependencies = [
  "tracing",
  "tracing-futures",
  "tracing-subscriber",
+ "unicode-segmentation",
  "urlparse",
 ]
 
@@ -2132,6 +2133,12 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "5a5f39404a5da50712a4c1eecf25e90dd62b613502b7e925fd4e4d19b5c96512"
 
 [[package]]
+name = "unicode-segmentation"
+version = "1.12.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f6ccf251212114b54433ec949fd6a7841275f9ada20dddd2f29e9ceea4501493"
+
+[[package]]
 name = "untrusted"
 version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
diff --git a/Cargo.toml b/Cargo.toml
index 8bbfb15..4886700 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -54,6 +54,8 @@ async-trait = "0.1"
 
 bitflags = "2"
 
+unicode-segmentation = "1"
+
 [dev-dependencies]
 mockito = "1"
 
diff --git a/src/util/formatting/truncate.rs b/src/util/formatting/truncate.rs
index 80caa23..9528492 100644
--- a/src/util/formatting/truncate.rs
+++ b/src/util/formatting/truncate.rs
@@ -1,3 +1,5 @@
+use unicode_segmentation::UnicodeSegmentation;
+
 /// Truncates a string after a certain number of characters.
 ///
 /// Function always tries to truncate on a word boundary.
@@ -6,15 +8,7 @@ pub fn truncate(text: &str, len: usize) -> String {
     if text.len() <= len {
         return text.to_string();
     }
-    format!(
-        "{}…",
-        text[..len]
-            .rsplitn(2, " ")
-            .collect::<Vec<&str>>()
-            .last()
-            .copied()
-            .expect("This can never happen >inb4 it happens")
-    )
+    format!("{}…", text.graphemes(true).take(len).collect::<String>())
 }
 
 #[cfg(test)]
@@ -39,7 +33,7 @@ mod tests {
     fn test_truncate_with_input_of_greater_length_than_limit() {
         let input = "some longer text";
         let result = truncate(input, input.len() - 1);
-        assert_eq!("some longer…", result)
+        assert_eq!("some longer tex…", result)
     }
 
     #[test]
@@ -48,4 +42,11 @@ mod tests {
         let result = truncate(input, input.len() - 1);
         assert_eq!("somelongertex…", result)
     }
+
+    #[test]
+    fn test_truncate_with_unicode() {
+        let input = "short ° text";
+        let result = truncate(input, 7);
+        assert_eq!(result, "short °…")
+    }
 }