Fix unpacking of filenames with contains UTF-8 characters

Change from the C to C.UTF-8 locale, allowing libarchive to handle
filenames in UTF-8. We restrict to change LC_CTYPE only, since
libarchive only needs the charset set.

See on libarchive Website for a more complete description of the issue:

  libarchive/libarchive#587
  https://github.com/libarchive/libarchive/wiki/Filenames

Once we complete the uncompress operation, we restore the original
LC_CTYPE after extraction to avoid side effects.

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>

Cargo.toml

@@ -33,6 +33,7 @@ futures-executor = { version = "0.3.5", optional = true }
 blocking = { version = "1.0.0", optional = true }
 tokio = { version = "1.0.0", features = ["rt-multi-thread", "macros", "fs", "net"], optional = true }
 tokio-util = { version = "0.6.0", features = ["compat"], optional = true }
+libc = "0.2.86"
 
 [features]
 async_support = ["async-trait", "futures-channel", "futures-core", "futures-io", "futures-util", "futures-executor"]

src/lib.rs

@@ -365,7 +365,11 @@ where
                 archive_reader,
             )?;
 
-            f(archive_reader, archive_writer, archive_entry)
+            let old_locale = set_utf8_locale();
+            let r = f(archive_reader, archive_writer, archive_entry);
+            restore_locale(old_locale);
+
+            r
         })();
 
         archive_result(ffi::archive_read_close(archive_reader), archive_reader)?;
@@ -421,7 +425,11 @@ where
                 archive_reader,
             )?;
 
-            f(archive_reader, archive_writer, archive_entry)
+            let old_locale = set_utf8_locale();
+            let r = f(archive_reader, archive_writer, archive_entry);
+            restore_locale(old_locale);
+
+            r
         })();
 
         archive_result(ffi::archive_read_close(archive_reader), archive_reader)?;
@@ -547,3 +555,48 @@ unsafe extern "C" fn libarchive_read_callback(
         }
     }
 }
+
+// Change from the C to C.UTF-8 locale, allowing libarchive to
+// handle filenames in UTF-8. We restrict to change LC_CTYPE only,
+// since libarchive only needs the charset set.
+//
+// See on libarchive Website for a more complete description
+// of the issue:
+//
+//   https://github.com/libarchive/libarchive/issues/587
+//   https://github.com/libarchive/libarchive/wiki/Filenames
+#[cfg(unix)]
+unsafe fn set_utf8_locale() -> libc::locale_t {
+    #[cfg(target_os = "linux")]
+    let locale = b"C.UTF-8\0";
+
+    #[cfg(target_os = "macos")]
+    let locale = b"UTF-8\0";
+
+    let utf8_locale = libc::newlocale(
+        libc::LC_CTYPE_MASK,
+        std::ffi::CStr::from_bytes_with_nul_unchecked(locale).as_ptr(),
+        std::ptr::null_mut(),
+    );
+
+    libc::uselocale(utf8_locale)
+}
+
+// Restore the original LC_CTYPE after extraction to avoid side effects.
+#[cfg(unix)]
+unsafe fn restore_locale(old_locale: libc::locale_t) {
+    libc::uselocale(old_locale);
+}
+
+#[cfg(windows)]
+unsafe fn set_utf8_locale() -> *mut i8 {
+    let locale = b".UTF-8\0";
+
+    libc::setlocale(
+        libc::LC_CTYPE,
+        std::ffi::CStr::from_bytes_with_nul_unchecked(locale).as_ptr(),
+    )
+}
+
+#[cfg(windows)]
+unsafe fn restore_locale(old_locale: libc::locale_t) {}

tests/integration_test.rs

@@ -309,6 +309,21 @@ fn uncompress_to_dir_not_preserve_owner() {
     );
 }
 
+#[test]
+fn uncompress_to_dir_with_utf8_pathname() {
+    let dir = tempfile::TempDir::new().expect("Failed to create the tmp directory");
+    let mut source = std::fs::File::open("tests/fixtures/utf8.tar").unwrap();
+
+    uncompress_archive(&mut source, dir.path(), Ownership::Ignore)
+        .expect("Failed to uncompress the file");
+
+    assert_eq!(
+        dir.path().join("utf-8-file-name-őúíá").exists(),
+        true,
+        "the path doesn't exist"
+    );
+}
+
 #[test]
 fn uncompress_same_file_not_preserve_owner() {
     uncompress_archive(