Fix file urls with dos driver letter on unix

I think the previous behavior was actually the correct one for Windows
systems, but because my main dev machine is Linux, it is easier to make
sure everything works correctly on those platforms first. I will fix
Windows next.

Author: niklaswimmer

gix-url/src/parse/mod.rs

@@ -163,19 +163,35 @@ fn parse_scp(input: &BStr, colon: usize) -> Result<crate::Url, Error> {
 }
 
 fn parse_file_url(input: &BStr, protocol_colon: usize) -> Result<crate::Url, Error> {
-    let (input, url) = input_to_utf8_and_url(input, UrlKind::Url)?;
+    let input = input_to_utf8(input, UrlKind::Url)?;
+    let input_after_protocol = &input[protocol_colon + 3..];
 
-    if !input[protocol_colon + 3..].contains('/') {
+    let Some(first_slash) = input_after_protocol.find('/') else {
         return Err(Error::MissingRepositoryPath {
             url: input.to_owned().into(),
             kind: UrlKind::Url,
         });
-    }
+    };
+
+    // We can not use the url crate to parse host and path because it special cases Windows
+    // driver letters. With the url crate an input of `file://x:/path/to/git` is parsed as empty
+    // host and with `x:/path/to/git` as path. This behavior is wrong for Git which parses the
+    // `x:` as the host.
+    // TODO: this behavior is most likely different on Windows
+    let host = if first_slash == 0 {
+        // file:///path/to/git
+        None
+    } else {
+        // file://host/path/to/git
+        Some(&input_after_protocol[..first_slash])
+    };
+    // path includes the slash character
+    let path = &input_after_protocol[first_slash..];
 
     Ok(crate::Url {
         serialize_alternative_form: false,
-        host: url.host_str().map(Into::into),
-        ..parse_local(url.path().into())?
+        host: host.map(Into::into),
+        ..parse_local(path.into())?
     })
 }
 

gix-url/tests/parse/file.rs

@@ -154,11 +154,12 @@ fn url_from_relative_path_with_colon_in_name() -> crate::Result {
     Ok(())
 }
 
+#[cfg(windows)]
 mod windows {
+    use crate::parse::{assert_url, assert_url_roundtrip, url, url_alternate};
     use gix_url::Scheme;
 
     #[test]
-    #[cfg(windows)]
     fn url_from_absolute_path() -> crate::Result {
         assert_url(
             url::Url::from_directory_path("c:\\users\\1")
@@ -181,8 +182,6 @@ mod windows {
         Ok(())
     }
 
-    use crate::parse::{assert_url, assert_url_roundtrip, url, url_alternate};
-
     #[test]
     fn file_path_without_protocol() -> crate::Result {
         let url = assert_url(
@@ -213,3 +212,39 @@ mod windows {
         )
     }
 }
+
+#[cfg(not(windows))]
+mod unix {
+    use crate::parse::{assert_url, assert_url_roundtrip, url, url_alternate};
+    use gix_url::Scheme;
+
+    #[test]
+    fn file_path_without_protocol() -> crate::Result {
+        let url = assert_url(
+            "x:/path/to/git",
+            url_alternate(Scheme::Ssh, None, "x", None, b"/path/to/git"),
+        )?
+        .to_bstring();
+        assert_eq!(url, "x:/path/to/git");
+        Ok(())
+    }
+
+    #[test]
+    fn file_path_with_backslashes_without_protocol() -> crate::Result {
+        let url = assert_url(
+            "x:\\path\\to\\git",
+            url_alternate(Scheme::Ssh, None, "x", None, b"\\path\\to\\git"),
+        )?
+        .to_bstring();
+        assert_eq!(url, "x:\\path\\to\\git");
+        Ok(())
+    }
+
+    #[test]
+    fn file_path_with_protocol() -> crate::Result {
+        assert_url_roundtrip(
+            "file://x:/path/to/git",
+            url(Scheme::File, None, "x:", None, b"/path/to/git"),
+        )
+    }
+}