refactor: migrate web server to Actix Web

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Cargo.toml

@@ -18,9 +18,10 @@ actix-web = "4"
 actix-web-lab = "0.20"
 anyhow = "1"
 cadence = "1"
-crates-index = { version = "2", default-features = false, features = ["git"] }
+crates-index = { version = "3", default-features = false, features = ["git"] }
-derive_more = "0.99"
+derive_more = { version = "1", features = ["display", "error", "from"] }
 dotenvy = "0.15"
+either = "1.12"
 font-awesome-as-a-crate = "0.3"
 futures-util = { version = "0.3", default-features = false, features = ["std"] }
 error_reporter = "1"
@@ -30,13 +31,14 @@ maud = "0.26"
 mime = "0.3"
 once_cell = "1"
 parking_lot = "0.12"
-pulldown-cmark = "0.11"
+pulldown-cmark = "0.12"
 relative-path = { version = "1", features = ["serde"] }
 reqwest = { version = "0.12", features = ["json"] }
 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"
 tokio = { version = "1.24.2", features = ["rt", "macros", "sync", "time"] }
 toml = "0.8"
 tracing = "0.1.30"

README.md

@@ -19,10 +19,15 @@ To analyze the state of your dependencies you can use the following URLs:
 
 On the analysis page, you will also find the markdown code to include a fancy badge in your project README so visitors (and you) can see at a glance if your dependencies are still up to date!
 
-Badges have a few style options, specified with query parameters, that match the styles from `shields.io`:
+Badges have a few options, specified with query parameters:
-- `?style=flat` (default)
+- `style`: which matches the styles from `shields.io`:
-- `?style=flat-square`
+  - `?style=flat` (default)
-- `?style=for-the-badge`
+  - `?style=flat-square`
+  - `?style=for-the-badge`
+- `subject`: customize the text on the left (which is the same concept as `label` in `shields.io`, and [URL-Encoding](https://developer.mozilla.org/en-US/docs/Glossary/percent-encoding) is needed for spaces or special characters!). e.g.:
+  - `?subject=yourdeps`
+  - `?subject=git%20deps`
+  - `?subject=deps%3Acore`
 
 ## Contributing
 

src/server/error.rs

@@ -9,22 +9,22 @@ use crate::server::views::html::error::{render, render_404};
 
 #[derive(Debug, Display)]
 pub(crate) enum ServerError {
-    #[display(fmt = "Could not retrieve popular items")]
+    #[display("Could not retrieve popular items")]
     PopularItemsFailed,
 
-    #[display(fmt = "Crate not found")]
+    #[display("Crate not found")]
     CrateNotFound,
 
-    #[display(fmt = "Could not parse crate path")]
+    #[display("Could not parse crate path")]
     BadCratePath,
 
-    #[display(fmt = "Could not fetch crate information")]
+    #[display("Could not fetch crate information")]
     CrateFetchFailed,
 
-    #[display(fmt = "Could not parse repository path")]
+    #[display("Could not parse repository path")]
     BadRepoPath,
 
-    #[display(fmt = "Crate/repo analysis failed")]
+    #[display("Crate/repo analysis failed")]
     AnalysisFailed(Markup),
 }
 

src/server/mod.rs

@@ -38,8 +38,11 @@ use crate::{
         repo::RepoPath,
         SubjectPath,
     },
+    utils::common::{safe_truncate, UntaggedEither, WrappedBool},
 };
 
+const MAX_SUBJECT_WIDTH: usize = 100;
+
 #[derive(Debug, Clone, Copy, PartialEq)]
 enum StatusFormat {
     Html,
@@ -310,18 +313,35 @@ 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>,
 }
 
 impl ExtraConfig {
     fn from_query_string(qs: Option<&str>) -> Self {
+        /// This wrapper can make the deserialization process infallible.
+        #[derive(Debug, Clone, Deserialize)]
+        #[serde(transparent)]
+        struct QueryParam<T>(UntaggedEither<T, String>);
+
+        impl<T> QueryParam<T> {
+            fn opt(self) -> Option<T> {
+                either::Either::from(self.0).left()
+            }
+        }
+
         #[derive(Debug, Clone, Default, Deserialize)]
         struct ExtraConfigPartial {
-            style: Option<BadgeStyle>,
+            style: Option<QueryParam<BadgeStyle>>,
-            compact: Option<bool>,
+            compact: Option<QueryParam<WrappedBool>>,
+            subject: Option<String>,
             path: Option<String>,
         }
 
@@ -330,9 +350,33 @@ impl ExtraConfig {
             .unwrap_or_default();
 
         Self {
-            style: extra_config.style.unwrap_or_default(),
+            style: extra_config
-            compact: extra_config.compact.unwrap_or_default(),
+                .style
+                .and_then(|qp| qp.opt())
+                .unwrap_or_default(),
+            compact: extra_config
+                .compact
+                .and_then(|qp| qp.opt())
+                .unwrap_or_default()
+                .0,
+            subject: extra_config
+                .subject
+                .filter(|t| !t.is_empty())
+                .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

@@ -7,12 +7,7 @@ pub fn badge(
     analysis_outcome: Option<&AnalyzeDependenciesOutcome>,
     badge_knobs: ExtraConfig,
 ) -> Badge {
-    let subject = if badge_knobs.compact {
+    let subject = badge_knobs.subject().to_owned();
-        "deps"
-    } else {
-        "dependencies"
-    }
-    .to_owned();
 
     let opts = match analysis_outcome {
         Some(outcome) => {

src/utils/common.rs

@@ -0,0 +1,150 @@
+use either::Either;
+use serde::{Deserialize, Serialize};
+use serde_with::{DeserializeFromStr, SerializeDisplay};
+use std::{
+    fmt::{self, Debug, Display, Formatter},
+    str::FromStr,
+};
+
+/// An `untagged` version of `Either`.
+///
+/// The reason this structure is needed is that `either::Either` is
+/// by default an `Externally Tagged` enum, and it is possible to
+/// implement `untagged` via `#[serde(with = "either::serde_untagged_optional")]`
+/// as well. But this approach can cause problems with deserialization,
+/// resulting in having to manually add the `#[serde(default)]` tag,
+/// and this leads to less readable as well as less flexible code.
+/// So it would be better if we manually implement this `UntaggedEither` here,
+/// while providing a two-way conversion to `either::Either`.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(untagged)]
+pub enum UntaggedEither<L, R> {
+    Left(L),
+    Right(R),
+}
+
+impl<L, R> From<UntaggedEither<L, R>> for Either<L, R> {
+    fn from(value: UntaggedEither<L, R>) -> Self {
+        match value {
+            UntaggedEither::Left(l) => Self::Left(l),
+            UntaggedEither::Right(r) => Self::Right(r),
+        }
+    }
+}
+
+impl<L, R> From<Either<L, R>> for UntaggedEither<L, R> {
+    fn from(value: Either<L, R>) -> Self {
+        match value {
+            Either::Left(l) => UntaggedEither::Left(l),
+            Either::Right(r) => UntaggedEither::Right(r),
+        }
+    }
+}
+
+/// A generic newtype which serialized using `Display` and deserialized using `FromStr`.
+#[derive(Default, Clone, DeserializeFromStr, SerializeDisplay)]
+pub struct SerdeDisplayFromStr<T>(pub T);
+
+impl<T: Debug> Debug for SerdeDisplayFromStr<T> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        Debug::fmt(&self.0, f)
+    }
+}
+
+impl<T: Display> Display for SerdeDisplayFromStr<T> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        Display::fmt(&self.0, f)
+    }
+}
+
+impl<T: FromStr> FromStr for SerdeDisplayFromStr<T> {
+    type Err = T::Err;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        s.parse::<T>().map(Self)
+    }
+}
+
+/// The reason it's needed here is that using `Deserialize` generated
+/// by default by `serde` will cause deserialization to fail if
+/// both untyped formats (such as `urlencoded`) and `untagged enum`
+/// 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), "🦊");
+    }
+}

src/utils/mod.rs

@@ -1,2 +1,3 @@
 pub mod cache;
+pub mod common;
 pub mod index;