router: don't decode %25 to '%' (#357)

actix-router/CHANGES.md

@@ -1,6 +1,9 @@
 # Changes
 
 ## Unreleased - 2021-xx-xx
+* When matching URL parameters, `%25` is kept in the percent-encoded form - no longer decoded to `%`. [#357]
+
+[#357]: https://github.com/actix/actix-net/pull/357
 
 
 ## 0.2.7 - 2021-02-06

actix-router/src/url.rs

@@ -31,7 +31,7 @@ fn set_bit(array: &mut [u8], ch: u8) {
 }
 
 thread_local! {
-    static DEFAULT_QUOTER: Quoter = Quoter::new(b"@:", b"/+");
+    static DEFAULT_QUOTER: Quoter = Quoter::new(b"@:", b"%/+");
 }
 
 #[derive(Default, Clone, Debug)]
@@ -204,24 +204,59 @@ mod tests {
     use super::*;
     use crate::{Path, ResourceDef};
 
+    const PROTECTED: &[u8] = b"%/+";
+
+    fn match_url(pattern: &'static str, url: impl AsRef<str>) -> Path<Url> {
+        let re = ResourceDef::new(pattern);
+        let uri = Uri::try_from(url.as_ref()).unwrap();
+        let mut path = Path::new(Url::new(uri));
+        assert!(re.match_path(&mut path));
+        path
+    }
+
+    fn percent_encode(data: &[u8]) -> String {
+        data.into_iter().map(|c| format!("%{:02X}", c)).collect()
+    }
+
     #[test]
     fn test_parse_url() {
-        let re = ResourceDef::new("/user/{id}/test");
+        let re = "/user/{id}/test";
 
-        let url = Uri::try_from("/user/2345/test").unwrap();
+        let path = match_url(re, "/user/2345/test");
-        let mut path = Path::new(Url::new(url));
-        assert!(re.match_path(&mut path));
         assert_eq!(path.get("id").unwrap(), "2345");
 
-        let url = Uri::try_from("/user/qwe%25/test").unwrap();
+        // "%25" should never be decoded into '%' to gurantee the output is a valid
-        let mut path = Path::new(Url::new(url));
+        // percent-encoded format
-        assert!(re.match_path(&mut path));
+        let path = match_url(re, "/user/qwe%25/test");
-        assert_eq!(path.get("id").unwrap(), "qwe%");
+        assert_eq!(path.get("id").unwrap(), "qwe%25");
 
-        let url = Uri::try_from("/user/qwe%25rty/test").unwrap();
+        let path = match_url(re, "/user/qwe%25rty/test");
-        let mut path = Path::new(Url::new(url));
+        assert_eq!(path.get("id").unwrap(), "qwe%25rty");
-        assert!(re.match_path(&mut path));
+    }
-        assert_eq!(path.get("id").unwrap(), "qwe%rty");
+
+    #[test]
+    fn test_protected_chars() {
+        let encoded = percent_encode(PROTECTED);
+        let path = match_url("/user/{id}/test", format!("/user/{}/test", encoded));
+        assert_eq!(path.get("id").unwrap(), &encoded);
+    }
+
+    #[test]
+    fn test_non_protecteed_ascii() {
+        let nonprotected_ascii = ('\u{0}'..='\u{7F}')
+            .filter(|&c| c.is_ascii() && !PROTECTED.contains(&(c as u8)))
+            .collect::<String>();
+        let encoded = percent_encode(nonprotected_ascii.as_bytes());
+        let path = match_url("/user/{id}/test", format!("/user/{}/test", encoded));
+        assert_eq!(path.get("id").unwrap(), &nonprotected_ascii);
+    }
+
+    #[test]
+    fn test_valid_utf8_multibyte() {
+        let test = ('\u{FF00}'..='\u{FFFF}').collect::<String>();
+        let encoded = percent_encode(test.as_bytes());
+        let path = match_url("/a/{id}/b", format!("/a/{}/b", &encoded));
+        assert_eq!(path.get("id").unwrap(), &test);
     }
 
     #[test]