From 05d83a3726d718f19b645c8f1b5de67ea9be8ef8 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Jo=C3=A3o=20M=2E=20Bezerra?= <marcospb19@hotmail.com>
Date: Wed, 10 Nov 2021 19:59:36 -0300
Subject: [PATCH] Utf8 checks, using references and avoid allocating

And add docs
---
 src/archive/zip.rs |  5 ++---
 src/utils/mod.rs   | 22 +++++++---------------
 2 files changed, 9 insertions(+), 18 deletions(-)

diff --git a/src/archive/zip.rs b/src/archive/zip.rs
index 0cd6031..6a46b31 100644
--- a/src/archive/zip.rs
+++ b/src/archive/zip.rs
@@ -10,14 +10,13 @@ use fs_err as fs;
 use walkdir::WalkDir;
 use zip::{self, read::ZipFile, ZipArchive};
 
-use self::utf8::get_invalid_utf8_paths;
 use crate::{
     error::FinalError,
     info,
     list::FileInArchive,
     utils::{
-        cd_into_same_dir_as, concatenate_os_str_list, dir_is_empty, strip_cur_dir, to_utf, user_wants_to_overwrite,
-        Bytes,
+        cd_into_same_dir_as, concatenate_os_str_list, dir_is_empty, get_invalid_utf8_paths, strip_cur_dir, to_utf,
+        user_wants_to_overwrite, Bytes,
     },
     QuestionPolicy,
 };
diff --git a/src/utils/mod.rs b/src/utils/mod.rs
index 3d0001b..7e948b7 100644
--- a/src/utils/mod.rs
+++ b/src/utils/mod.rs
@@ -16,23 +16,15 @@ pub use question::{
 pub use utf8::{get_invalid_utf8_paths, is_invalid_utf8};
 
 mod utf8 {
-    use std::path::{Path, PathBuf};
+    use std::{ffi::OsStr, path::PathBuf};
 
-    pub fn is_invalid_utf8(path: &Path) -> bool {
-        #[cfg(unix)]
-        {
-            use std::{os::unix::prelude::OsStrExt, str};
-
-            let bytes = path.as_os_str().as_bytes();
-            str::from_utf8(bytes).is_err()
-        }
-        #[cfg(not(unix))]
-        {
-            path.to_str().is_none()
-        }
+    /// Check, without allocating, if os_str can be converted into &str
+    pub fn is_invalid_utf8(os_str: impl AsRef<OsStr>) -> bool {
+        os_str.as_ref().to_str().is_none()
     }
 
-    pub fn get_invalid_utf8_paths(paths: &[PathBuf]) -> Vec<PathBuf> {
-        paths.iter().filter_map(|path| is_invalid_utf8(&path).then(|| path.clone())).collect()
+    /// Filter out list of paths that are not utf8 valid
+    pub fn get_invalid_utf8_paths(paths: &[PathBuf]) -> Vec<&PathBuf> {
+        paths.iter().filter_map(|path| is_invalid_utf8(path).then(|| path)).collect()
     }
 }
-- 
2.11.4.GIT