From 40211df2192e91a80fbdbab3f0e7e938ff914f47 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E4=B8=89=E5=92=B2=E9=9B=85=20=C2=B7=20Misaki=20Masa?=
 <sxyazi@gmail.com>
Date: Mon, 30 Sep 2024 00:37:33 +0800
Subject: [PATCH] fix: replace control characters to printable characters in
 plain text preview (#1704)

---
 yazi-plugin/src/external/highlighter.rs | 17 ++++++++-----
 yazi-shared/src/chars.rs                | 33 ++++++++++++++++++++++---
 2 files changed, 40 insertions(+), 10 deletions(-)

diff --git a/yazi-plugin/src/external/highlighter.rs b/yazi-plugin/src/external/highlighter.rs
index 3f6831c38..6ecf3ebda 100644
--- a/yazi-plugin/src/external/highlighter.rs
+++ b/yazi-plugin/src/external/highlighter.rs
@@ -5,7 +5,7 @@ use ratatui::{layout::Rect, text::{Line, Span, Text}};
 use syntect::{LoadingError, dumps, easy::HighlightLines, highlighting::{self, Theme, ThemeSet}, parsing::{SyntaxReference, SyntaxSet}};
 use tokio::{fs::File, io::{AsyncBufReadExt, BufReader}, sync::OnceCell};
 use yazi_config::{PREVIEW, THEME, preview::PreviewWrap};
-use yazi_shared::PeekError;
+use yazi_shared::{PeekError, replace_to_printable};
 
 static INCR: AtomicUsize = AtomicUsize::new(0);
 static SYNTECT: OnceCell<(Theme, SyntaxSet)> = OnceCell::const_new();
@@ -58,7 +58,7 @@ impl Highlighter {
 				return Err("Binary file".into());
 			}
 
-			if !plain && (buf.len() > 5000 || buf.contains(&0x1b)) {
+			if !plain && (buf.len() > 5000 || Self::contains_control_chars(&buf)) {
 				plain = true;
 				drop(mem::take(&mut before));
 			}
@@ -93,7 +93,7 @@ impl Highlighter {
 		}
 
 		Ok(if plain {
-			Text::from(after.join("").replace('\x1b', "^[").replace('\t', &PREVIEW.indent()))
+			Text::from(replace_to_printable(&after.join(""), PREVIEW.tab_size))
 		} else {
 			Self::highlight_with(before, after, syntax.unwrap()).await?
 		})
@@ -186,9 +186,14 @@ impl Highlighter {
 	}
 
 	#[inline(always)]
-	fn carriage_return_to_line_feed(c: &mut u8) {
-		if *c == b'\r' {
-			*c = b'\n';
+	fn contains_control_chars(buf: &[u8]) -> bool {
+		buf.iter().any(|&b| b.is_ascii_control() && !matches!(b, b'\t' | b'\n' | b'\r'))
+	}
+
+	#[inline(always)]
+	fn carriage_return_to_line_feed(b: &mut u8) {
+		if *b == b'\r' {
+			*b = b'\n';
 		}
 	}
 }
diff --git a/yazi-shared/src/chars.rs b/yazi-shared/src/chars.rs
index 33f55ea9c..5d0035e27 100644
--- a/yazi-shared/src/chars.rs
+++ b/yazi-shared/src/chars.rs
@@ -1,3 +1,5 @@
+use core::str;
+
 pub const MIME_DIR: &str = "inode/directory";
 
 #[derive(Clone, Copy, PartialEq, Eq)]
@@ -20,11 +22,34 @@ impl CharKind {
 }
 
 pub fn strip_trailing_newline(mut s: String) -> String {
-	if s.ends_with('\n') {
-		s.pop();
-	}
-	if s.ends_with('\r') {
+	while s.ends_with('\n') || s.ends_with('\r') {
 		s.pop();
 	}
 	s
 }
+
+pub fn replace_to_printable(s: &str, tab_size: u8) -> String {
+	let mut buf = Vec::new();
+	buf.try_reserve_exact(s.len() | 15).unwrap_or_else(|_| panic!());
+
+	for &b in s.as_bytes() {
+		match b {
+			b'\n' => buf.push(b'\n'),
+			b'\t' => {
+				for _ in 0..tab_size {
+					buf.push(b' ');
+				}
+			}
+			b'\0'..=b'\x1F' => {
+				buf.push(b'^');
+				buf.push(b + b'@');
+			}
+			0x7f => {
+				buf.push(b'^');
+				buf.push(b'?');
+			}
+			_ => buf.push(b),
+		}
+	}
+	unsafe { String::from_utf8_unchecked(buf) }
+}