WIP

Add UTF8LocaleGuard struct and use coarse grained locale setting

Missing:
- setlocale is not threadsafe, and hance on windows tests fails due to
  wrong locale beeing restaured

Signed-off-by: Jonathas-Conceicao <jonathas.conceicao@ossystems.com.br>

src/ffi/locale.rs

@@ -0,0 +1,86 @@
+// Copyright (C) 2021 O.S. Systems Software LTDA
+//
+// SPDX-License-Identifier: MIT OR Apache-2.0
+
+// Change from the C to system 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
+pub(crate) struct UTF8LocaleGuard {
+    #[cfg(unix)]
+    save: libc::locale_t,
+
+    #[cfg(windows)]
+    save: Option<std::ffi::CString>,
+}
+
+#[cfg(unix)]
+impl UTF8LocaleGuard {
+    pub(crate) fn new() -> Self {
+        #[cfg(target_os = "linux")]
+        let locale = b"\0";
+
+        #[cfg(target_os = "macos")]
+        let locale = b"UTF-8\0";
+
+        let utf8_locale = unsafe {
+            libc::newlocale(
+                libc::LC_CTYPE_MASK,
+                locale.as_ptr() as *const libc::c_char,
+                std::ptr::null_mut(),
+            )
+        };
+
+        let save = unsafe { libc::uselocale(utf8_locale) };
+
+        Self { save }
+    }
+}
+
+#[cfg(unix)]
+impl Drop for UTF8LocaleGuard {
+    fn drop(&mut self) {
+        unsafe { libc::uselocale(self.save) };
+    }
+}
+
+#[cfg(windows)]
+impl UTF8LocaleGuard {
+    pub(crate) fn new() -> Self {
+        let locale = b".UTF-8\0";
+
+        let save = {
+            let old_locale = unsafe { libc::setlocale(libc::LC_CTYPE, std::ptr::null()) };
+            if old_locale.is_null() {
+                None
+            } else {
+                Some(unsafe { std::ffi::CStr::from_ptr(old_locale) }.to_owned())
+            }
+        };
+
+        unsafe {
+            libc::setlocale(
+                libc::LC_CTYPE,
+                std::ffi::CStr::from_bytes_with_nul_unchecked(locale).as_ptr(),
+            )
+        };
+
+        Self { save }
+    }
+}
+
+#[cfg(windows)]
+impl Drop for UTF8LocaleGuard {
+    fn drop(&mut self) {
+        if let Some(locale) = &self.save {
+            println!("Restauring to: {}", locale.to_str().unwrap());
+            unsafe { libc::setlocale(libc::LC_CTYPE, locale.as_ptr()) };
+        } else {
+            println!("No locale to restaure to");
+        }
+    }
+}

src/ffi/mod.rs

@@ -3,76 +3,7 @@
 // SPDX-License-Identifier: MIT OR Apache-2.0
 
 mod generated;
+mod locale;
 
 pub(crate) use crate::ffi::generated::*;
-
-pub(crate) unsafe fn archive_read_next_header(
-    archive: *mut archive,
-    entry: *mut *mut archive_entry,
-) -> ::std::os::raw::c_int {
-    let old_locale = set_utf8_locale();
-    let ret = generated::archive_read_next_header(archive, entry);
-    restore_locale(old_locale);
-
-    ret
-}
-
-// Change from the C to the system 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"\0";
-
-    #[cfg(target_os = "macos")]
-    let locale = b"UTF-8\0";
-
-    let utf8_locale = libc::newlocale(
-        libc::LC_CTYPE_MASK,
-        locale.as_ptr() as *const libc::c_char,
-        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() -> Option<ffi::CString> {
-    let locale = b".UTF-8\0";
-
-    let old_locale = {
-        let old_locale = libc::setlocale(libc::LC_CTYPE, ptr::null());
-        if old_locale.is_null() {
-            None
-        } else {
-            Some(ffi::CStr::from_ptr(old_locale).to_owned())
-        }
-    };
-
-    libc::setlocale(
-        libc::LC_CTYPE,
-        ffi::CStr::from_bytes_with_nul_unchecked(locale).as_ptr(),
-    );
-
-    old_locale
-}
-
-#[cfg(windows)]
-unsafe fn restore_locale(old_locale: Option<ffi::CString>) {
-    if let Some(old_locale) = old_locale {
-        libc::setlocale(libc::LC_CTYPE, old_locale.as_ptr());
-    }
-}
+pub(crate) use locale::UTF8LocaleGuard;

src/lib.rs

@@ -116,6 +116,7 @@ pub fn list_archive_files<R>(source: R) -> Result<Vec<String>>
 where
     R: Read + Seek,
 {
+    let _utf8_guard = ffi::UTF8LocaleGuard::new();
     run_with_seekable_archive(source, |archive_reader, _, mut entry| unsafe {
         let mut file_list = Vec::new();
         #[allow(clippy::vec_init_then_push)]
@@ -171,6 +172,7 @@ where
     R: Read,
     W: Write,
 {
+    let _utf8_guard = ffi::UTF8LocaleGuard::new();
     run_with_archive(
         WriteMode::Buffer,
         source,
@@ -206,6 +208,7 @@ pub fn uncompress_archive<R>(source: R, dest: &Path, ownership: Ownership) -> Re
 where
     R: Read + Seek,
 {
+    let _utf8_guard = ffi::UTF8LocaleGuard::new();
     run_with_archive(
         WriteMode::Disk { ownership },
         source,
@@ -273,6 +276,7 @@ where
     R: Read + Seek,
     W: Write,
 {
+    let _utf8_guard = ffi::UTF8LocaleGuard::new();
     run_with_seekable_archive(source, |archive_reader, _, mut entry| unsafe {
         loop {
             match ffi::archive_read_next_header(archive_reader, &mut entry) {
@@ -301,6 +305,7 @@ where
     F: FnOnce(*mut ffi::archive, *mut ffi::archive, *mut ffi::archive_entry) -> Result<T>,
     R: Read,
 {
+    let _utf8_guard = ffi::UTF8LocaleGuard::new();
     unsafe {
         let archive_entry: *mut ffi::archive_entry = std::ptr::null_mut();
         let archive_reader = ffi::archive_read_new();