refactor: remove unicode dep and simplify UntaggedEither impl

Cargo.lock

@@ -2783,7 +2783,6 @@ dependencies = [
  "toml 0.8.13",
  "tracing",
  "tracing-subscriber",
- "unicode-ellipsis",
 ]
 
 [[package]]
@@ -3290,16 +3289,6 @@ version = "2.0.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7eec5d1121208364f6793f7d2e222bf75a915c19557537745b195b253dd64217"
 
-[[package]]
-name = "unicode-ellipsis"
-version = "0.2.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "924ab5b8c3fed9966b8cde2dc7169146331cba3dacba97cbd0e8866e7cfd4dff"
-dependencies = [
- "unicode-segmentation",
- "unicode-width",
-]
-
 [[package]]
 name = "unicode-ident"
 version = "1.0.12"
@@ -3315,12 +3304,6 @@ dependencies = [
  "tinyvec",
 ]
 
-[[package]]
-name = "unicode-segmentation"
-version = "1.11.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d4c87d22b6e3f4a18d4d40ef354e97c90fcb14dd91d7dc0aa9d8a1172ebf7202"
-
 [[package]]
 name = "unicode-width"
 version = "0.1.12"

Cargo.toml

@@ -19,7 +19,7 @@ cadence = "1"
 crates-index = { version = "2", default-features = false, features = ["git"] }
 derive_more = "0.99"
 dotenvy = "0.15"
-either = "1.12.0"
+either = "1.12"
 font-awesome-as-a-crate = "0.3"
 futures-util = { version = "0.3", default-features = false, features = ["std"] }
 hyper = { version = "0.14.10", features = ["full"] }
@@ -34,15 +34,14 @@ relative-path = { version = "1", features = ["serde"] }
 reqwest = { version = "0.12", features = ["json"] }
 route-recognizer = "0.3"
 rustsec = "0.29"
-semver = { version = "1.0", features = ["serde"] }
+semver = { version = "1", features = ["serde"] }
 serde = { version = "1", features = ["derive"] }
 serde_urlencoded = "0.7"
-serde_with = "3.8.1"
+serde_with = "3"
 tokio = { version = "1.24.2", features = ["rt-multi-thread", "macros", "sync", "time"] }
 toml = "0.8"
 tracing = "0.1.30"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
-unicode-ellipsis = "0.2.0"
 
 [target.'cfg(any())'.dependencies]
 gix = { version = "0.63", default-features = false, features = ["blocking-http-transport-reqwest-rust-tls"] }

src/server/mod.rs

@@ -10,7 +10,6 @@ use once_cell::sync::Lazy;
 use route_recognizer::{Params, Router};
 use semver::VersionReq;
 use serde::Deserialize;
-use unicode_ellipsis::truncate_str;
 
 mod assets;
 mod views;
@@ -25,9 +24,11 @@ use crate::{
         repo::RepoPath,
         SubjectPath,
     },
-    utils::common::{UntaggedEither, WrappedBool},
+    utils::common::{safe_truncate, UntaggedEither, WrappedBool},
 };
 
+const MAX_SUBJECT_WIDTH: usize = 100;
+
 #[derive(Debug, Clone, Copy, PartialEq)]
 enum StatusFormat {
     Html,
@@ -430,10 +431,13 @@ static SELF_BASE_URL: Lazy<String> =
 pub struct ExtraConfig {
     /// Badge style to show
     style: BadgeStyle,
+
     /// Whether the inscription _"dependencies"_ should be abbreviated as _"deps"_ in the badge.
     compact: bool,
+
     /// Custom text on the left (it's the same concept as `label` in shields.io).
     subject: Option<String>,
+
     /// Path in which the crate resides within the repository
     path: Option<String>,
 }
@@ -447,7 +451,7 @@ impl ExtraConfig {
 
         impl<T> QueryParam<T> {
             fn opt(self) -> Option<T> {
-                self.0.into_either().left()
+                either::Either::from(self.0).left()
             }
         }
 
@@ -459,7 +463,6 @@ impl ExtraConfig {
             path: Option<String>,
         }
 
-        const MAX_WIDTH: usize = 100;
         let extra_config = qs
             .and_then(|qs| serde_urlencoded::from_str::<ExtraConfigPartial>(qs).ok())
             .unwrap_or_default();
@@ -477,8 +480,21 @@ impl ExtraConfig {
             subject: extra_config
                 .subject
                 .filter(|t| !t.is_empty())
-                .map(|t| truncate_str(&t, MAX_WIDTH).into()),
+                .map(|subject| safe_truncate(&subject, MAX_SUBJECT_WIDTH).to_owned()),
             path: extra_config.path,
         }
     }
+
+    /// Returns subject for badge.
+    ///
+    /// Returns `subject` if set, or "dependencies" / "deps" depending on value of `compact`.
+    pub(crate) fn subject(&self) -> &str {
+        if let Some(subject) = &self.subject {
+            subject
+        } else if self.compact {
+            "deps"
+        } else {
+            "dependencies"
+        }
+    }
 }

src/server/views/badge.rs

@@ -7,13 +7,7 @@ pub fn badge(
     analysis_outcome: Option<&AnalyzeDependenciesOutcome>,
     badge_knobs: ExtraConfig,
 ) -> Badge {
-    let subject = if let Some(subject) = badge_knobs.subject {
+    let subject = badge_knobs.subject().to_owned();
-        subject
-    } else if badge_knobs.compact {
-        "deps".into()
-    } else {
-        "dependencies".into()
-    };
 
     let opts = match analysis_outcome {
         Some(outcome) => {

src/utils/common.rs

@@ -41,31 +41,19 @@ impl<L, R> From<Either<L, R>> for UntaggedEither<L, R> {
     }
 }
 
-impl<L, R> UntaggedEither<L, R> {
-    pub fn into_either(self) -> Either<L, R> {
-        self.into()
-    }
-}
-
 /// A generic newtype which serialized using `Display` and deserialized using `FromStr`.
 #[derive(Default, Clone, DeserializeFromStr, SerializeDisplay)]
 pub struct SerdeDisplayFromStr<T>(pub T);
 
-impl<T> From<T> for SerdeDisplayFromStr<T> {
-    fn from(value: T) -> Self {
-        Self(value)
-    }
-}
-
 impl<T: Debug> Debug for SerdeDisplayFromStr<T> {
     fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
-        self.0.fmt(f)
+        Debug::fmt(&self.0, f)
     }
 }
 
 impl<T: Display> Display for SerdeDisplayFromStr<T> {
     fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
-        self.0.fmt(f)
+        Display::fmt(&self.0, f)
     }
 }
 
@@ -73,7 +61,7 @@ impl<T: FromStr> FromStr for SerdeDisplayFromStr<T> {
     type Err = T::Err;
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        Ok(s.parse::<T>()?.into())
+        s.parse::<T>().map(Self)
     }
 }
 
@@ -83,3 +71,80 @@ impl<T: FromStr> FromStr for SerdeDisplayFromStr<T> {
 /// are used. The Wrap type here forces the deserialization process to
 /// be delegated to `FromStr`.
 pub type WrappedBool = SerdeDisplayFromStr<bool>;
+
+/// Returns truncated string accounting for multi-byte characters.
+pub(crate) fn safe_truncate(s: &str, len: usize) -> &str {
+    if len == 0 {
+        return "";
+    }
+
+    if s.len() <= len {
+        return s;
+    }
+
+    if s.is_char_boundary(len) {
+        return &s[0..len];
+    }
+
+    // Only 3 cases possible: 1, 2, or 3 bytes need to be removed for a new,
+    // valid UTF-8 string to appear when truncated, just enumerate them,
+    // Underflow is not possible since position 0 is always a valid boundary.
+
+    if let Some((slice, _rest)) = s.split_at_checked(len - 1) {
+        return slice;
+    }
+
+    if let Some((slice, _rest)) = s.split_at_checked(len - 2) {
+        return slice;
+    }
+
+    if let Some((slice, _rest)) = s.split_at_checked(len - 3) {
+        return slice;
+    }
+
+    unreachable!("all branches covered");
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn safe_truncation() {
+        assert_eq!(safe_truncate("", 0), "");
+        assert_eq!(safe_truncate("", 1), "");
+        assert_eq!(safe_truncate("", 9), "");
+
+        assert_eq!(safe_truncate("a", 0), "");
+        assert_eq!(safe_truncate("a", 1), "a");
+        assert_eq!(safe_truncate("a", 9), "a");
+
+        assert_eq!(safe_truncate("lorem\nipsum", 0), "");
+        assert_eq!(safe_truncate("lorem\nipsum", 5), "lorem");
+        assert_eq!(safe_truncate("lorem\nipsum", usize::MAX), "lorem\nipsum");
+
+        assert_eq!(safe_truncate("café", 1), "c");
+        assert_eq!(safe_truncate("café", 2), "ca");
+        assert_eq!(safe_truncate("café", 3), "caf");
+        assert_eq!(safe_truncate("café", 4), "caf");
+        assert_eq!(safe_truncate("café", 5), "café");
+
+        // 2-byte char
+        assert_eq!(safe_truncate("é", 0), "");
+        assert_eq!(safe_truncate("é", 1), "");
+        assert_eq!(safe_truncate("é", 2), "é");
+
+        // 3-byte char
+        assert_eq!(safe_truncate("⊕", 0), "");
+        assert_eq!(safe_truncate("⊕", 1), "");
+        assert_eq!(safe_truncate("⊕", 2), "");
+        assert_eq!(safe_truncate("⊕", 3), "⊕");
+
+        // 4-byte char
+        assert_eq!(safe_truncate("🦊", 0), "");
+        assert_eq!(safe_truncate("🦊", 1), "");
+        assert_eq!(safe_truncate("🦊", 2), "");
+        assert_eq!(safe_truncate("🦊", 3), "");
+        assert_eq!(safe_truncate("🦊", 4), "🦊");
+    }
+}