Make Git remote URL parsing more robust (#19924)

Marshall Bowers created 1 year ago

This PR improves the parsing of Git remote URLs in order to make
features that depend on them more robust.

Previously we were just treating these as plain strings and doing
one-off shotgun parsing to massage them into the right format. This
meant that we weren't accounting for edge cases in URL structure.

One of these cases was HTTPS Git URLs containing a username, which can
arise when using GitHub Enterprise (see
https://github.com/zed-industries/zed/issues/11160).

We now have a `RemoteUrl` typed to represent a parsed Git remote URL and
use the `Url` parser to parse it.

Release Notes:

- Improved the parsing of Git remote URLs to support additional
scenarios.

Change summary

crates/git/src/git.rs                                   | 16 +-
crates/git/src/hosting_provider.rs                      | 10 
crates/git/src/remote.rs                                | 85 +++++++++++
crates/git_hosting_providers/src/providers/bitbucket.rs | 53 ++++--
crates/git_hosting_providers/src/providers/codeberg.rs  | 49 +++--
crates/git_hosting_providers/src/providers/gitee.rs     | 54 ++++--
crates/git_hosting_providers/src/providers/github.rs    | 69 +++++---
crates/git_hosting_providers/src/providers/gitlab.rs    | 64 ++++---
crates/git_hosting_providers/src/providers/sourcehut.rs | 65 ++++---
9 files changed, 304 insertions(+), 161 deletions(-)

Detailed changes

crates/git/src/git.rs

@@ -1,4 +1,10 @@
+pub mod blame;
+pub mod commit;
+pub mod diff;
 mod hosting_provider;
+mod remote;
+pub mod repository;
+pub mod status;
 
 use anyhow::{anyhow, Context, Result};
 use serde::{Deserialize, Serialize};
@@ -7,15 +13,9 @@ use std::fmt;
 use std::str::FromStr;
 use std::sync::LazyLock;
 
-pub use git2 as libgit;
-
 pub use crate::hosting_provider::*;
-
-pub mod blame;
-pub mod commit;
-pub mod diff;
-pub mod repository;
-pub mod status;
+pub use crate::remote::*;
+pub use git2 as libgit;
 
 pub static DOT_GIT: LazyLock<&'static OsStr> = LazyLock::new(|| OsStr::new(".git"));
 pub static COOKIES: LazyLock<&'static OsStr> = LazyLock::new(|| OsStr::new("cookies"));

crates/git/src/hosting_provider.rs

@@ -69,7 +69,7 @@ pub trait GitHostingProvider {
     /// Returns a formatted range of line numbers to be placed in a permalink URL.
     fn format_line_numbers(&self, start_line: u32, end_line: u32) -> String;
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>>;
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote>;
 
     fn extract_pull_request(
         &self,
@@ -159,10 +159,10 @@ impl GitHostingProviderRegistry {
     }
 }
 
-#[derive(Debug)]
-pub struct ParsedGitRemote<'a> {
-    pub owner: &'a str,
-    pub repo: &'a str,
+#[derive(Debug, PartialEq)]
+pub struct ParsedGitRemote {
+    pub owner: Arc<str>,
+    pub repo: Arc<str>,
 }
 
 pub fn parse_git_remote_url(

crates/git/src/remote.rs

@@ -0,0 +1,85 @@
+use derive_more::Deref;
+use url::Url;
+
+/// The URL to a Git remote.
+#[derive(Debug, PartialEq, Eq, Clone, Deref)]
+pub struct RemoteUrl(Url);
+
+impl std::str::FromStr for RemoteUrl {
+    type Err = url::ParseError;
+
+    fn from_str(input: &str) -> Result<Self, Self::Err> {
+        if input.starts_with("git@") {
+            // Rewrite remote URLs like `[e-mail]:user/repo.git` to `ssh://[e-mail]/user/repo.git`
+            let ssh_url = input.replacen(':', "/", 1).replace("git@", "ssh://git@");
+            Ok(RemoteUrl(Url::parse(&ssh_url)?))
+        } else {
+            Ok(RemoteUrl(Url::parse(input)?))
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use pretty_assertions::assert_eq;
+
+    use super::*;
+
+    #[test]
+    fn test_parsing_valid_remote_urls() {
+        let valid_urls = vec![
+            (
+                "https://github.com/octocat/zed.git",
+                "https",
+                "github.com",
+                "/octocat/zed.git",
+            ),
+            (
+                "[e-mail]:octocat/zed.git",
+                "ssh",
+                "github.com",
+                "/octocat/zed.git",
+            ),
+            (
+                "ssh://[e-mail]/octocat/zed.git",
+                "ssh",
+                "github.com",
+                "/octocat/zed.git",
+            ),
+            (
+                "file:///path/to/local/zed",
+                "file",
+                "",
+                "/path/to/local/zed",
+            ),
+        ];
+
+        for (input, expected_scheme, expected_host, expected_path) in valid_urls {
+            let parsed = input.parse::<RemoteUrl>().expect("failed to parse URL");
+            let url = parsed.0;
+            assert_eq!(
+                url.scheme(),
+                expected_scheme,
+                "unexpected scheme for {input:?}",
+            );
+            assert_eq!(
+                url.host_str().unwrap_or(""),
+                expected_host,
+                "unexpected host for {input:?}",
+            );
+            assert_eq!(url.path(), expected_path, "unexpected path for {input:?}");
+        }
+    }
+
+    #[test]
+    fn test_parsing_invalid_remote_urls() {
+        let invalid_urls = vec!["not_a_url", "http://"];
+
+        for url in invalid_urls {
+            assert!(
+                url.parse::<RemoteUrl>().is_err(),
+                "expected \"{url}\" to not parse as a Git remote URL",
+            );
+        }
+    }
+}

crates/git_hosting_providers/src/providers/bitbucket.rs

@@ -1,6 +1,11 @@
+use std::str::FromStr;
+
 use url::Url;
 
-use git::{BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote};
+use git::{
+    BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote,
+    RemoteUrl,
+};
 
 pub struct Bitbucket;
 
@@ -25,18 +30,22 @@ impl GitHostingProvider for Bitbucket {
         format!("lines-{start_line}:{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        if url.contains("bitbucket.org") {
-            let (_, repo_with_owner) = url.trim_end_matches(".git").split_once("bitbucket.org")?;
-            let (owner, repo) = repo_with_owner
-                .trim_start_matches('/')
-                .trim_start_matches(':')
-                .split_once('/')?;
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != "bitbucket.org" {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        let repo = path_segments.next()?.trim_end_matches(".git");
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -88,8 +97,8 @@ mod tests {
         let url = "https://[e-mail]/thorstenzed/testingrepo.git";
         let (provider, parsed) = parse_git_remote_url(provider_registry, url).unwrap();
         assert_eq!(provider.name(), "Bitbucket");
-        assert_eq!(parsed.owner, "thorstenzed");
-        assert_eq!(parsed.repo, "testingrepo");
+        assert_eq!(parsed.owner.as_ref(), "thorstenzed");
+        assert_eq!(parsed.repo.as_ref(), "testingrepo");
     }
 
     #[test]
@@ -99,8 +108,8 @@ mod tests {
         let url = "https://bitbucket.org/thorstenzed/testingrepo.git";
         let (provider, parsed) = parse_git_remote_url(provider_registry, url).unwrap();
         assert_eq!(provider.name(), "Bitbucket");
-        assert_eq!(parsed.owner, "thorstenzed");
-        assert_eq!(parsed.repo, "testingrepo");
+        assert_eq!(parsed.owner.as_ref(), "thorstenzed");
+        assert_eq!(parsed.repo.as_ref(), "testingrepo");
     }
 
     #[test]
@@ -110,15 +119,15 @@ mod tests {
         let url = "[e-mail]:thorstenzed/testingrepo.git";
         let (provider, parsed) = parse_git_remote_url(provider_registry, url).unwrap();
         assert_eq!(provider.name(), "Bitbucket");
-        assert_eq!(parsed.owner, "thorstenzed");
-        assert_eq!(parsed.repo, "testingrepo");
+        assert_eq!(parsed.owner.as_ref(), "thorstenzed");
+        assert_eq!(parsed.repo.as_ref(), "testingrepo");
     }
 
     #[test]
     fn test_build_bitbucket_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "thorstenzed",
-            repo: "testingrepo",
+            owner: "thorstenzed".into(),
+            repo: "testingrepo".into(),
         };
         let permalink = Bitbucket.build_permalink(
             remote,
@@ -136,8 +145,8 @@ mod tests {
     #[test]
     fn test_build_bitbucket_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "thorstenzed",
-            repo: "testingrepo",
+            owner: "thorstenzed".into(),
+            repo: "testingrepo".into(),
         };
         let permalink = Bitbucket.build_permalink(
             remote,
@@ -156,8 +165,8 @@ mod tests {
     #[test]
     fn test_build_bitbucket_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "thorstenzed",
-            repo: "testingrepo",
+            owner: "thorstenzed".into(),
+            repo: "testingrepo".into(),
         };
         let permalink = Bitbucket.build_permalink(
             remote,

crates/git_hosting_providers/src/providers/codeberg.rs

@@ -1,3 +1,4 @@
+use std::str::FromStr;
 use std::sync::Arc;
 
 use anyhow::{bail, Context, Result};
@@ -9,6 +10,7 @@ use url::Url;
 
 use git::{
     BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, Oid, ParsedGitRemote,
+    RemoteUrl,
 };
 
 #[derive(Debug, Deserialize)]
@@ -103,19 +105,22 @@ impl GitHostingProvider for Codeberg {
         format!("L{start_line}-L{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        if url.starts_with("[e-mail]:") || url.starts_with("https://codeberg.org/") {
-            let repo_with_owner = url
-                .trim_start_matches("[e-mail]:")
-                .trim_start_matches("https://codeberg.org/")
-                .trim_end_matches(".git");
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            let (owner, repo) = repo_with_owner.split_once('/')?;
-
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != "codeberg.org" {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        let repo = path_segments.next()?.trim_end_matches(".git");
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -175,8 +180,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,
@@ -194,8 +199,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,
@@ -213,8 +218,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,
@@ -232,8 +237,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,
@@ -251,8 +256,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_https_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,
@@ -270,8 +275,8 @@ mod tests {
     #[test]
     fn test_build_codeberg_permalink_from_https_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Codeberg.build_permalink(
             remote,

crates/git_hosting_providers/src/providers/gitee.rs

@@ -1,6 +1,11 @@
+use std::str::FromStr;
+
 use url::Url;
 
-use git::{BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote};
+use git::{
+    BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote,
+    RemoteUrl,
+};
 
 pub struct Gitee;
 
@@ -25,19 +30,22 @@ impl GitHostingProvider for Gitee {
         format!("L{start_line}-{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        if url.starts_with("[e-mail]:") || url.starts_with("https://gitee.com/") {
-            let repo_with_owner = url
-                .trim_start_matches("[e-mail]:")
-                .trim_start_matches("https://gitee.com/")
-                .trim_end_matches(".git");
-
-            let (owner, repo) = repo_with_owner.split_once('/')?;
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != "gitee.com" {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        let repo = path_segments.next()?.trim_end_matches(".git");
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -81,8 +89,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,
@@ -100,8 +108,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,
@@ -119,8 +127,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,
@@ -138,8 +146,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,
@@ -157,8 +165,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_https_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,
@@ -176,8 +184,8 @@ mod tests {
     #[test]
     fn test_build_gitee_permalink_from_https_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "libkitten",
-            repo: "zed",
+            owner: "libkitten".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitee.build_permalink(
             remote,

crates/git_hosting_providers/src/providers/github.rs

@@ -1,3 +1,4 @@
+use std::str::FromStr;
 use std::sync::{Arc, OnceLock};
 
 use anyhow::{bail, Context, Result};
@@ -10,7 +11,7 @@ use url::Url;
 
 use git::{
     BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, Oid, ParsedGitRemote,
-    PullRequest,
+    PullRequest, RemoteUrl,
 };
 
 fn pull_request_number_regex() -> &'static Regex {
@@ -107,19 +108,22 @@ impl GitHostingProvider for Github {
         format!("L{start_line}-L{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        if url.starts_with("[e-mail]:") || url.starts_with("https://github.com/") {
-            let repo_with_owner = url
-                .trim_start_matches("[e-mail]:")
-                .trim_start_matches("https://github.com/")
-                .trim_end_matches(".git");
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            let (owner, repo) = repo_with_owner.split_once('/')?;
-
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != "github.com" {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        let repo = path_segments.next()?.trim_end_matches(".git");
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -198,11 +202,26 @@ mod tests {
 
     use super::*;
 
+    #[test]
+    fn test_parse_remote_url_given_https_url_with_username() {
+        let parsed_remote = Github
+            .parse_remote_url("https://[e-mail]/some-org/some-repo.git")
+            .unwrap();
+
+        assert_eq!(
+            parsed_remote,
+            ParsedGitRemote {
+                owner: "some-org".into(),
+                repo: "some-repo".into(),
+            }
+        );
+    }
+
     #[test]
     fn test_build_github_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -220,8 +239,8 @@ mod tests {
     #[test]
     fn test_build_github_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -239,8 +258,8 @@ mod tests {
     #[test]
     fn test_build_github_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -258,8 +277,8 @@ mod tests {
     #[test]
     fn test_build_github_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -277,8 +296,8 @@ mod tests {
     #[test]
     fn test_build_github_permalink_from_https_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -296,8 +315,8 @@ mod tests {
     #[test]
     fn test_build_github_permalink_from_https_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Github.build_permalink(
             remote,
@@ -315,8 +334,8 @@ mod tests {
     #[test]
     fn test_github_pull_requests() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
 
         let message = "This does not contain a pull request";

crates/git_hosting_providers/src/providers/gitlab.rs

@@ -1,8 +1,13 @@
+use std::str::FromStr;
+
 use anyhow::{anyhow, bail, Result};
 use url::Url;
 use util::maybe;
 
-use git::{BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote};
+use git::{
+    BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote,
+    RemoteUrl,
+};
 
 #[derive(Debug)]
 pub struct Gitlab {
@@ -64,21 +69,22 @@ impl GitHostingProvider for Gitlab {
         format!("L{start_line}-{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        let host = self.base_url.host_str()?;
-
-        if url.starts_with(&format!("git@{host}")) || url.starts_with(&format!("https://{host}/")) {
-            let repo_with_owner = url
-                .trim_start_matches(&format!("git@{host}:"))
-                .trim_start_matches(&format!("https://{host}/"))
-                .trim_end_matches(".git");
-
-            let (owner, repo) = repo_with_owner.split_once('/')?;
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != self.base_url.host_str()? {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        let repo = path_segments.next()?.trim_end_matches(".git");
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -127,8 +133,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -146,8 +152,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -165,8 +171,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -184,8 +190,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -203,8 +209,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_https_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -222,8 +228,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_permalink_from_https_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let permalink = Gitlab::new().build_permalink(
             remote,
@@ -241,8 +247,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_self_hosted_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let gitlab =
             Gitlab::from_remote_url("[e-mail]:zed-industries/zed.git")
@@ -263,8 +269,8 @@ mod tests {
     #[test]
     fn test_build_gitlab_self_hosted_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "zed-industries",
-            repo: "zed",
+            owner: "zed-industries".into(),
+            repo: "zed".into(),
         };
         let gitlab =
             Gitlab::from_remote_url("https://gitlab-instance.big-co.com/zed-industries/zed.git")

crates/git_hosting_providers/src/providers/sourcehut.rs

@@ -1,6 +1,11 @@
+use std::str::FromStr;
+
 use url::Url;
 
-use git::{BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote};
+use git::{
+    BuildCommitPermalinkParams, BuildPermalinkParams, GitHostingProvider, ParsedGitRemote,
+    RemoteUrl,
+};
 
 pub struct Sourcehut;
 
@@ -25,21 +30,27 @@ impl GitHostingProvider for Sourcehut {
         format!("L{start_line}-{end_line}")
     }
 
-    fn parse_remote_url<'a>(&self, url: &'a str) -> Option<ParsedGitRemote<'a>> {
-        if url.starts_with("[e-mail]:") || url.starts_with("https://git.sr.ht/") {
-            // sourcehut indicates a repo with '.git' suffix as a separate repo.
-            // For example, "[e-mail]:~username/repo" and "[e-mail]:~username/repo.git"
-            // are two distinct repositories.
-            let repo_with_owner = url
-                .trim_start_matches("[e-mail]:~")
-                .trim_start_matches("https://git.sr.ht/~");
-
-            let (owner, repo) = repo_with_owner.split_once('/')?;
+    fn parse_remote_url(&self, url: &str) -> Option<ParsedGitRemote> {
+        let url = RemoteUrl::from_str(url).ok()?;
 
-            return Some(ParsedGitRemote { owner, repo });
+        let host = url.host_str()?;
+        if host != "git.sr.ht" {
+            return None;
         }
 
-        None
+        let mut path_segments = url.path_segments()?;
+        let owner = path_segments.next()?;
+        // We don't trim the `.git` suffix here like we do elsewhere, as
+        // sourcehut treats a repo with `.git` suffix as a separate repo.
+        //
+        // For example, `[e-mail]:~username/repo` and `[e-mail]:~username/repo.git`
+        // are two distinct repositories.
+        let repo = path_segments.next()?;
+
+        Some(ParsedGitRemote {
+            owner: owner.into(),
+            repo: repo.into(),
+        })
     }
 
     fn build_commit_permalink(
@@ -83,8 +94,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_ssh_url() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -102,8 +113,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_ssh_url_with_git_prefix() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed.git",
+            owner: "rajveermalviya".into(),
+            repo: "zed.git".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -121,8 +132,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_ssh_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -140,8 +151,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_ssh_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -159,8 +170,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_https_url() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -178,8 +189,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_https_url_single_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,
@@ -197,8 +208,8 @@ mod tests {
     #[test]
     fn test_build_sourcehut_permalink_from_https_url_multi_line_selection() {
         let remote = ParsedGitRemote {
-            owner: "rajveermalviya",
-            repo: "zed",
+            owner: "rajveermalviya".into(),
+            repo: "zed".into(),
         };
         let permalink = Sourcehut.build_permalink(
             remote,