extract EscapedURI into subcrate

Author: syphar

Cargo.lock

@@ -22,9 +22,11 @@ exclude = [
 
 [workspace.dependencies]
 anyhow = { version = "1.0.42", features = ["backtrace"]}
+askama = "0.14.0"
 bincode = "2.0.1"
 chrono = { version = "0.4.11", default-features = false, features = ["clock", "serde"] }
 derive_more = { version = "2.0.0", features = ["display", "deref", "from", "into", "from_str"] }
+http = "1.0.0"
 opentelemetry = "0.31.0"
 opentelemetry-otlp = { version = "0.31.0", features = ["grpc-tonic",  "metrics"] }
 opentelemetry-resource-detectors = "0.10.0"
@@ -44,6 +46,7 @@ docs_rs_env_vars = { path = "crates/lib/docs_rs_env_vars" }
 docs_rs_logging = { path = "crates/lib/docs_rs_logging" }
 docs_rs_opentelemetry = { path = "crates/lib/docs_rs_opentelemetry" }
 docs_rs_types = { path = "crates/lib/docs_rs_types" }
+docs_rs_uri = { path = "crates/lib/docs_rs_uri" }
 docs_rs_utils = { path = "crates/lib/docs_rs_utils" }
 sentry = { workspace = true }
 log = "0.4"
@@ -98,7 +101,7 @@ async-stream = "0.3.5"
 aws-config = { version = "1.0.0", default-features = false, features = ["rt-tokio", "default-https-client"] }
 aws-sdk-s3 = "1.3.0"
 aws-smithy-types-convert = { version = "0.60.0", features = ["convert-chrono"] }
-http = "1.0.0"
+http = { workspace = true }
 
 # Data serialization and deserialization
 serde = { workspace = true }
@@ -112,13 +115,12 @@ axum-extra = { version = "0.12.0", features = ["typed-header", "routing", "middl
 tower = "0.5.1"
 tower-http = { version = "0.6.0", features = ["fs", "trace", "timeout", "catch-panic"] }
 mime = "0.3.16"
-percent-encoding = "2.2.0"
 
 tempfile = "3.1.0"
 fn-error-context = "0.2.0"
 
 # Templating
-askama = "0.14.0"
+askama = { workspace = true }
 walkdir = "2"
 phf = "0.13.1"
 

Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "docs_rs_uri"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+anyhow = { workspace = true }
+askama = { workspace = true }
+bincode = { workspace = true }
+http = { workspace = true }
+percent-encoding = "2.2.0"
+url = { workspace = true }
+
+[dev-dependencies]
+test-case = { workspace = true }

crates/lib/docs_rs_uri/Cargo.toml

@@ -0,0 +1,29 @@
+use anyhow::Result;
+use percent_encoding::{AsciiSet, CONTROLS, utf8_percent_encode};
+use std::borrow::Cow;
+
+// from https://github.com/servo/rust-url/blob/master/url/src/parser.rs
+// and https://github.com/tokio-rs/axum/blob/main/axum-extra/src/lib.rs
+const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ').add(b'"').add(b'<').add(b'>').add(b'`');
+const PATH: &AsciiSet = &FRAGMENT.add(b'#').add(b'?').add(b'{').add(b'}');
+
+pub fn encode_url_path(path: &str) -> String {
+    utf8_percent_encode(path, PATH).to_string()
+}
+
+pub fn url_decode<'a>(input: &'a str) -> Result<Cow<'a, str>> {
+    Ok(percent_encoding::percent_decode(input.as_bytes()).decode_utf8()?)
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use test_case::test_case;
+
+    #[test_case("/something/", "/something/")] // already valid path
+    #[test_case("/something>", "/something%3E")] // something to encode
+    #[test_case("/something%3E", "/something%3E")] // re-running doesn't change anything
+    fn test_encode_url_path(input: &str, expected: &str) {
+        assert_eq!(encode_url_path(input), expected);
+    }
+}

crates/lib/docs_rs_uri/src/encode.rs

@@ -1,4 +1,4 @@
-use crate::web::{encode_url_path, url_decode};
+use crate::encode::{encode_url_path, url_decode};
 use askama::filters::HtmlSafe;
 use http::{Uri, uri::PathAndQuery};
 use std::{borrow::Borrow, fmt::Display, iter, str::FromStr};
@@ -200,7 +200,7 @@ impl EscapedURI {
         self.uri
     }
 
-    pub(crate) fn with_fragment(mut self, fragment: impl AsRef<str>) -> Self {
+    pub fn with_fragment(mut self, fragment: impl AsRef<str>) -> Self {
         self.fragment = Some(encode_url_path(fragment.as_ref()));
         self
     }
@@ -298,8 +298,6 @@ impl PartialEq<str> for EscapedURI {
 #[cfg(test)]
 mod tests {
     use super::EscapedURI;
-    use crate::web::{cache::CachePolicy, error::AxumNope};
-    use axum::response::IntoResponse as _;
     use http::Uri;
     use test_case::test_case;
 
@@ -309,18 +307,6 @@ mod tests {
         assert_eq!(s.parse::<EscapedURI>().unwrap(), *input);
     }
 
-    #[test]
-    fn test_redirect_error_encodes_url_path() {
-        let response = AxumNope::Redirect(
-            EscapedURI::from_path("/something>"),
-            CachePolicy::ForeverInCdnAndBrowser,
-        )
-        .into_response();
-
-        assert_eq!(response.status(), 302);
-        assert_eq!(response.headers().get("Location").unwrap(), "/something%3E");
-    }
-
     #[test_case("/something" => "/something")]
     #[test_case("/something>" => "/something%3E")]
     fn test_escaped_uri_encodes_from_path(input: &str) -> String {

src/web/escaped_uri.rs crates/lib/docs_rs_uri/src/escaped_uri.rssrc/web/escaped_uri.rs renamed to crates/lib/docs_rs_uri/src/escaped_uri.rs

@@ -0,0 +1,5 @@
+mod encode;
+mod escaped_uri;
+
+pub use encode::{encode_url_path, url_decode};
+pub use escaped_uri::EscapedURI;

crates/lib/docs_rs_uri/src/lib.rs

@@ -1,14 +1,15 @@
 use crate::{
     db::PoolError,
     storage::PathNotFoundError,
-    web::{AxumErrorPage, cache::CachePolicy, escaped_uri::EscapedURI, releases::Search},
+    web::{AxumErrorPage, cache::CachePolicy, releases::Search},
 };
 use anyhow::{Result, anyhow};
 use axum::{
     Json,
     http::StatusCode,
     response::{IntoResponse, Response as AxumResponse},
 };
+use docs_rs_uri::EscapedURI;
 use std::borrow::Cow;
 
 #[derive(Debug, thiserror::Error)]

src/web/error.rs

@@ -2,10 +2,7 @@
 
 use crate::{
     storage::CompressionAlgorithm,
-    web::{
-        MatchedRelease, MetaData, error::AxumNope, escaped_uri::EscapedURI, extractors::Path,
-        url_decode,
-    },
+    web::{MatchedRelease, MetaData, error::AxumNope, extractors::Path},
 };
 use anyhow::Result;
 use axum::{
@@ -14,6 +11,7 @@ use axum::{
     http::{Uri, request::Parts},
 };
 use docs_rs_types::{BuildId, KrateName, ReqVersion};
+use docs_rs_uri::{EscapedURI, url_decode};
 use itertools::Itertools as _;
 use serde::Deserialize;
 

src/web/extractors/rustdoc.rs

@@ -1,8 +1,8 @@
-use crate::web::escaped_uri::EscapedURI;
 use anyhow::Result;
 use askama::filters::HtmlSafe;
 use axum::http::uri::Uri;
 use axum_extra::headers::{Header, HeaderName, HeaderValue};
+use docs_rs_uri::EscapedURI;
 use serde::Serialize;
 use std::{fmt, ops::Deref};
 

src/web/headers/canonical_url.rs

@@ -6,7 +6,6 @@ pub(crate) mod cache;
 pub(crate) mod crate_details;
 mod csp;
 pub(crate) mod error;
-mod escaped_uri;
 pub(crate) mod extractors;
 mod features;
 mod file;
@@ -48,7 +47,6 @@ use docs_rs_types::{BuildStatus, CrateId, KrateName, ReqVersion, Version, Versio
 use docs_rs_utils::rustc_version::parse_rustc_date;
 use error::AxumNope;
 use page::TemplateData;
-use percent_encoding::{AsciiSet, CONTROLS, utf8_percent_encode};
 use sentry::integrations::tower as sentry_tower;
 use serde::Serialize;
 use serde_json::Value;
@@ -61,19 +59,6 @@ use tower::ServiceBuilder;
 use tower_http::{catch_panic::CatchPanicLayer, timeout::TimeoutLayer, trace::TraceLayer};
 use tracing::{info, instrument};
 
-// from https://github.com/servo/rust-url/blob/master/url/src/parser.rs
-// and https://github.com/tokio-rs/axum/blob/main/axum-extra/src/lib.rs
-const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ').add(b'"').add(b'<').add(b'>').add(b'`');
-const PATH: &AsciiSet = &FRAGMENT.add(b'#').add(b'?').add(b'{').add(b'}');
-
-pub(crate) fn encode_url_path(path: &str) -> String {
-    utf8_percent_encode(path, PATH).to_string()
-}
-
-pub(crate) fn url_decode<'a>(input: &'a str) -> Result<Cow<'a, str>> {
-    Ok(percent_encoding::percent_decode(input.as_bytes()).decode_utf8()?)
-}
-
 /// Picks the correct "rustdoc.css" static file depending on which rustdoc version was used to
 /// generate this version of this crate.
 pub fn get_correct_docsrs_style_file(version: &str) -> Result<String> {
@@ -641,8 +626,6 @@ impl_axum_webpage! {
 
 #[cfg(test)]
 mod test {
-    use std::str::FromStr as _;
-
     use super::*;
     use crate::docbuilder::DocCoverage;
     use crate::test::{
@@ -653,6 +636,7 @@ mod test {
     use kuchikiki::traits::TendrilSink;
     use pretty_assertions::assert_eq;
     use serde_json::json;
+    use std::str::FromStr as _;
     use test_case::test_case;
 
     async fn release(version: &str, env: &TestEnvironment) -> ReleaseId {
@@ -1236,11 +1220,4 @@ mod test {
         assert!(axum_redirect(path).is_err());
         assert!(axum_cached_redirect(path, cache::CachePolicy::NoCaching).is_err());
     }
-
-    #[test_case("/something/", "/something/")] // already valid path
-    #[test_case("/something>", "/something%3E")] // something to encode
-    #[test_case("/something%3E", "/something%3E")] // re-running doesn't change anything
-    fn test_encode_url_path(input: &str, expected: &str) {
-        assert_eq!(encode_url_path(input), expected);
-    }
 }