From 550062f46bf4cfe694b80ae0abc3abe3242a8f30 Mon Sep 17 00:00:00 2001
From: Ben_Kosytorz <benkosytorz@mail.me>
Date: Sat, 10 Oct 2026 23:52:47 +0200
Subject: [PATCH] feat(rust): extract .tgz archives (Rust-only short suffix)

---
 version-rust/src/extract.rs | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

diff --git a/version-rust/src/extract.rs b/version-rust/src/extract.rs
index 91b4ed3..92bda17 100644
--- a/version-rust/src/extract.rs
+++ b/version-rust/src/extract.rs
@@ -4,7 +4,7 @@
 //!
 //! Extraction backends: `sevenz-rust2` for .7z, `zip` for .zip, `unrar` for
 //! .rar — pure Rust, no external binaries. Tarballs and bare compressed
-//! files (.tar/.tar.gz/.tar.bz2/.tar.xz/.gz/.bz2/.xz) are handled by the
+//! files (.tar/.tar.gz/.tgz/.tar.bz2/.tar.xz/.gz/.bz2/.xz) are handled by the
 //! system `tar`/`gzip`/`bzip2`/`xz` binaries, exactly like the python
 //! version shelled out to `7z`: merged output capture, exit-code parity
 //! (0 ok, 1 warnings), and the 4h per-archive timeout. The tar/flate2/
@@ -37,8 +37,11 @@ use std::time::{Duration, Instant};
 #[cfg(unix)]
 use std::os::unix::fs::PermissionsExt;
 
-/// Archive suffixes recognized by the extractors (python `ARCHIVE_SUFFIXES`).
-pub const ARCHIVE_SUFFIXES: [&str; 7] = [".zip", ".7z", ".rar", ".tar", ".gz", ".bz2", ".xz"];
+/// Archive suffixes recognized by the extractors (python `ARCHIVE_SUFFIXES`,
+/// plus `.tgz` — a Rust-only extension; the python version never extracted
+/// those). `.tgz` is a gzip-compressed tar, so it maps onto the tarball path.
+pub const ARCHIVE_SUFFIXES: [&str; 8] =
+    [".zip", ".7z", ".rar", ".tar", ".gz", ".bz2", ".xz", ".tgz"];
 
 /// Stop extracting below this free space (python `MIN_FREE_BYTES`).
 pub const MIN_FREE_BYTES: u64 = 20 * 1024 * 1024 * 1024;
@@ -388,6 +391,9 @@ impl Extractor {
                 }
             }
             Some(".tar") => extract_tar_cmd(archive, &actual, None),
+            // .tgz is a gzip-compressed tar: same as .tar.gz, but the tar flag
+            // cannot be inferred from the name (Rust-only extension).
+            Some(".tgz") => extract_tar_cmd(archive, &actual, Some(".gz")),
             _ => Err(format!("no extractor for {name}")),
         };
         match res {
@@ -957,6 +963,10 @@ mod tests {
         let targz = gzip_stored(&make_ustar("delta.txt", b"tarred payload"));
         fs::write(s.join("logs.tar.gz"), targz).unwrap();
 
+        // a .tgz (gzip-compressed tar under the short suffix — Rust-only)
+        let tgz = gzip_stored(&make_ustar("epsilon.txt", b"tgz payload"));
+        fs::write(s.join("bundle.tgz"), tgz).unwrap();
+
         // garbage zip -> must land in .broken
         fs::write(s.join("bad.zip"), b"this is not a zip archive at all").unwrap();
 
@@ -1081,6 +1091,10 @@ mod tests {
             fs::read_to_string(o.join("src1/logs/delta.txt")).unwrap(),
             "tarred payload"
         );
+        assert_eq!(
+            fs::read_to_string(o.join("src1/bundle/epsilon.txt")).unwrap(),
+            "tgz payload"
+        );
 
         // no archives left in the extracted tree (outside .broken)
         for p in walk_files_sorted(o) {
@@ -1164,6 +1178,7 @@ mod tests {
                 "The Twin Lance",
             ),
             ("src1/logs.tar.gz", "src1/logs", "logs"),
+            ("src1/bundle.tgz", "src1/bundle", "bundle"),
         ]
         .iter()
         .map(|(a, d, u)| (a.to_string(), d.to_string(), u.to_string()))
@@ -1255,6 +1270,10 @@ mod tests {
         );
         assert_eq!(display_unit("Pack.zip"), "Pack");
         assert!(is_archive("c.7z.7z"));
+        // .tgz: Rust-only short suffix for a gzip-compressed tar
+        assert_eq!(archive_suffixes("x.tgz"), vec![".tgz"]);
+        assert_eq!(display_unit("x.tgz"), "x");
+        assert!(is_archive("X.TGZ"));
     }
 
     #[test]