fix: escape control characters in filenames (#3400)

Author: sxyazi

Cargo.lock

@@ -33,7 +33,7 @@ futures             = "0.3.31"
 globset             = "0.4.18"
 hashbrown           = { version = "0.16.1", features = [ "serde" ] }
 indexmap            = { version = "2.12.1", features = [ "serde" ] }
-libc                = "0.2.177"
+libc                = "0.2.178"
 lru                 = "0.16.2"
 mlua                = { version = "0.11.5", features = [ "anyhow", "async", "error-send", "lua54", "macros", "serde" ] }
 objc2               = "0.6.3"
@@ -44,7 +44,7 @@ percent-encoding    = "2.3.2"
 rand                = { version = "0.9.2", default-features = false, features = [ "os_rng", "small_rng", "std" ] }
 ratatui             = { version = "0.29.0", features = [ "unstable-rendered-line-info", "unstable-widget-ref" ] }
 regex               = "1.12.2"
-russh               = { version = "0.54.6", default-features = false, features = [ "ring", "rsa" ] }
+russh               = { version = "0.55.0", default-features = false, features = [ "ring", "rsa" ] }
 scopeguard          = "1.2.0"
 serde               = { version = "1.0.228", features = [ "derive" ] }
 serde_json          = "1.0.145"

Cargo.toml

@@ -1,11 +1,11 @@
-use std::{mem, path::MAIN_SEPARATOR_STR};
+use std::mem;
 
 use anyhow::Result;
-use yazi_fs::{CWD, path::expand_url, provider::{DirReader, FileHolder}};
+use yazi_fs::{CWD, path::clean_url, provider::{DirReader, FileHolder}};
 use yazi_macro::{act, render, succ};
 use yazi_parser::cmp::{CmpItem, ShowOpt, TriggerOpt};
 use yazi_proxy::CmpProxy;
-use yazi_shared::{AnyAsciiChar, data::Data, natsort, path::{PathBufDyn, PathDyn, PathLike}, scheme::{SchemeCow, SchemeLike}, strand::{AsStrand, StrandLike}, url::{UrlBuf, UrlCow, UrlLike}};
+use yazi_shared::{AnyAsciiChar, data::Data, natsort, path::{AsPath, PathBufDyn, PathCow, PathDyn, PathLike}, scheme::{SchemeCow, SchemeLike}, strand::{AsStrand, StrandLike}, url::{UrlBuf, UrlCow, UrlLike}};
 use yazi_vfs::provider;
 
 use crate::{Actor, Ctx};
@@ -74,24 +74,25 @@ impl Trigger {
 			return None; // We don't autocomplete a `~`, but `~/`
 		}
 
+		let cwd = CWD.load();
+		let abs = if !path.is_absolute() && cwd.scheme().covariant(&scheme) {
+			cwd.loc().try_join(&path).ok()?.into()
+		} else {
+			PathCow::from(&path)
+		};
+
 		let sep = if cfg!(windows) {
 			AnyAsciiChar::new(b"/\\").unwrap()
 		} else {
 			AnyAsciiChar::new(b"/").unwrap()
 		};
 
-		Some(match path.rsplit_pred(sep) {
-			Some((p, c)) if p.is_empty() => {
-				let root = PathDyn::with_str(scheme.kind(), MAIN_SEPARATOR_STR);
-				(UrlCow::try_from((scheme, root)).ok()?.into_owned(), c.into())
-			}
-			Some((p, c)) => (expand_url(UrlCow::try_from((scheme, p)).ok()?), c.into()),
-			None if CWD.load().scheme().covariant(&scheme) => (CWD.load().as_ref().clone(), path.into()),
-			None => {
-				let empty = PathDyn::with_str(scheme.kind(), "");
-				(UrlCow::try_from((scheme, empty)).ok()?.into_owned(), path.into())
-			}
-		})
+		let child = path.rsplit_pred(sep).map_or(path.as_path(), |(_, c)| c);
+		let parent =
+			PathDyn::with(scheme.kind(), abs.encoded_bytes().strip_suffix(child.encoded_bytes())?)
+				.ok()?;
+
+		Some((clean_url(UrlCow::try_from((scheme, parent)).ok()?), child.into()))
 	}
 }
 
@@ -130,11 +131,11 @@ mod tests {
 		compare("/foo/bar", "/foo/", "bar");
 		compare("///foo/bar", "/foo/", "bar");
 
-		CWD.set(&"sftp://test/".parse::<UrlBuf>().unwrap(), || {});
-		compare("sftp://test/a", "sftp://test/", "a");
-		compare("sftp://test//a", "sftp://test:0//", "a");
-		compare("sftp://test2/a", "sftp://test2/", "a");
-		compare("sftp://test2//a", "sftp://test2:0//", "a");
+		CWD.set(&"sftp://test".parse::<UrlBuf>().unwrap(), || {});
+		compare("sftp://test/a", "sftp://test/.", "a");
+		compare("sftp://test//a", "sftp://test//", "a");
+		compare("sftp://test2/a", "sftp://test2/.", "a");
+		compare("sftp://test2//a", "sftp://test2//", "a");
 	}
 
 	#[cfg(windows)]

yazi-actor/src/cmp/trigger.rs

@@ -5,7 +5,7 @@ use tokio::pin;
 use tokio_stream::{StreamExt, wrappers::UnboundedReceiverStream};
 use yazi_config::popup::InputCfg;
 use yazi_dds::Pubsub;
-use yazi_fs::{File, FilesOp, path::expand_url};
+use yazi_fs::{File, FilesOp, path::{clean_url, expand_url}};
 use yazi_macro::{act, err, render, succ};
 use yazi_parser::mgr::CdOpt;
 use yazi_proxy::{CmpProxy, InputProxy, MgrProxy};
@@ -71,6 +71,7 @@ impl Cd {
 					Ok(s) => {
 						let Ok(url) = UrlBuf::try_from(s).map(expand_url) else { return };
 						let Ok(url) = provider::absolute(&url).await else { return };
+						let url = clean_url(url);
 
 						let Ok(file) = File::new(&url).await else { return };
 						if file.is_dir() {

yazi-actor/src/mgr/cd.rs

@@ -21,7 +21,7 @@ fn clean_path_impl(path: PathDyn, base: usize, trail: usize) -> (PathBufDyn, usi
 
 	macro_rules! push {
 		($i:ident, $c:ident) => {{
-			out.push($c);
+			out.push(($i, $c));
 			if $i >= base {
 				uri_count += 1;
 			}
@@ -31,12 +31,25 @@ fn clean_path_impl(path: PathDyn, base: usize, trail: usize) -> (PathBufDyn, usi
 		}};
 	}
 
+	macro_rules! pop {
+		() => {{
+			if let Some((i, _)) = out.pop() {
+				if i >= base {
+					uri_count -= 1;
+				}
+				if i >= trail {
+					urn_count -= 1;
+				}
+			}
+		}};
+	}
+
 	for (i, c) in path.components().enumerate() {
 		match c {
 			CurDir => {}
-			ParentDir => match out.last() {
+			ParentDir => match out.last().map(|(_, c)| c) {
 				Some(RootDir) => {}
-				Some(Normal(_)) => _ = out.pop(),
+				Some(Normal(_)) => pop!(),
 				None | Some(CurDir) | Some(ParentDir) | Some(Prefix(_)) => push!(i, c),
 			},
 			c => push!(i, c),
@@ -47,7 +60,8 @@ fn clean_path_impl(path: PathDyn, base: usize, trail: usize) -> (PathBufDyn, usi
 	let path = if out.is_empty() {
 		PathBufDyn::with_str(kind, ".")
 	} else {
-		PathBufDyn::from_components(kind, out).expect("components with same kind")
+		PathBufDyn::from_components(kind, out.into_iter().map(|(_, c)| c))
+			.expect("components with same kind")
 	};
 
 	(path, uri_count, urn_count)
@@ -70,10 +84,11 @@ mod tests {
 			("archive://:3:2//../../tmp/test.zip/foo/bar", "archive://:3:2//tmp/test.zip/foo/bar"),
 			("archive://:3:2//tmp/../../test.zip/foo/bar", "archive://:3:2//test.zip/foo/bar"),
 			("archive://:4:2//tmp/test.zip/../../foo/bar", "archive://:2:2//foo/bar"),
-			("archive://:5:2//tmp/test.zip/../../foo/bar", "archive://:3:2//foo/bar"),
-			("archive://:4:4//tmp/test.zip/foo/bar/../../", "archive://:1:1//tmp/test.zip"),
-			("archive://:5:4//tmp/test.zip/foo/bar/../../", "archive://:2:1//tmp/test.zip"),
+			("archive://:5:2//tmp/test.zip/../../foo/bar", "archive://:2:2//foo/bar"),
+			("archive://:4:4//tmp/test.zip/foo/bar/../../", "archive:////tmp/test.zip"),
+			("archive://:5:4//tmp/test.zip/foo/bar/../../", "archive://:1//tmp/test.zip"),
 			("archive://:4:4//tmp/test.zip/foo/bar/../../../", "archive:////tmp"),
+			("sftp://test//root/.config/yazi/../../Downloads", "sftp://test//root/Downloads"),
 		];
 
 		for (input, expected) in cases {

yazi-fs/src/path/clean.rs

@@ -31,22 +31,22 @@ function Entity:prefix()
 end
 
 function Entity:highlights()
-	local name = self._file.name:gsub("\r", "?", 1)
+	local name, p = self._file.name, ui.printable
 	local highlights = self._file:highlights()
 	if not highlights or #highlights == 0 then
-		return name
+		return p(name)
 	end
 
 	local spans, last = {}, 0
 	for _, h in ipairs(highlights) do
 		if h[1] > last then
-			spans[#spans + 1] = name:sub(last + 1, h[1])
+			spans[#spans + 1] = p(name:sub(last + 1, h[1]))
 		end
-		spans[#spans + 1] = ui.Span(name:sub(h[1] + 1, h[2])):style(th.mgr.find_keyword)
+		spans[#spans + 1] = ui.Span(p(name:sub(h[1] + 1, h[2]))):style(th.mgr.find_keyword)
 		last = h[2]
 	end
 	if last < #name then
-		spans[#spans + 1] = name:sub(last + 1)
+		spans[#spans + 1] = p(name:sub(last + 1))
 	end
 	return ui.Line(spans)
 end

yazi-plugin/preset/components/entity.lua

@@ -64,7 +64,7 @@ function Status:name()
 		return ""
 	end
 
-	return " " .. h.name:gsub("\r", "?", 1)
+	return " " .. ui.printable(h.name)
 end
 
 function Status:perm()

yazi-plugin/preset/components/status.lua

@@ -1,19 +1,23 @@
+use std::borrow::Cow;
+
 use mlua::{AnyUserData, ExternalError, IntoLua, Lua, ObjectLike, Table, Value};
 use tracing::error;
 use unicode_width::{UnicodeWidthChar, UnicodeWidthStr};
 use yazi_binding::{Composer, ComposerGet, ComposerSet, Permit, PermitRef, elements::{Line, Rect, Span}};
 use yazi_config::LAYOUT;
 use yazi_proxy::{AppProxy, HIDER};
+use yazi_shared::replace_to_printable;
 
 pub fn compose() -> Composer<ComposerGet, ComposerSet> {
 	fn get(lua: &Lua, key: &[u8]) -> mlua::Result<Value> {
 		match key {
 			b"area" => area(lua)?,
 			b"hide" => hide(lua)?,
-			b"width" => width(lua)?,
+			b"printable" => printable(lua)?,
 			b"redraw" => redraw(lua)?,
 			b"render" => render(lua)?,
 			b"truncate" => truncate(lua)?,
+			b"width" => width(lua)?,
 			_ => return Ok(Value::Nil),
 		}
 		.into_lua(lua)
@@ -54,28 +58,12 @@ pub(super) fn hide(lua: &Lua) -> mlua::Result<Value> {
 	f.into_lua(lua)
 }
 
-pub(super) fn width(lua: &Lua) -> mlua::Result<Value> {
-	let f = lua.create_function(|_, v: Value| match v {
-		Value::String(s) => {
-			let (mut acc, b) = (0, s.as_bytes());
-			for c in b.utf8_chunks() {
-				acc += c.valid().width();
-				if !c.invalid().is_empty() {
-					acc += 1;
-				}
-			}
-			Ok(acc)
-		}
-		Value::UserData(ud) => {
-			if let Ok(line) = ud.borrow::<Line>() {
-				Ok(line.width())
-			} else if let Ok(span) = ud.borrow::<Span>() {
-				Ok(span.width())
-			} else {
-				Err("expected a string, Line, or Span".into_lua_err())?
-			}
-		}
-		_ => Err("expected a string, Line, or Span".into_lua_err())?,
+pub(super) fn printable(lua: &Lua) -> mlua::Result<Value> {
+	let f = lua.create_function(|lua, s: mlua::String| {
+		Ok(match replace_to_printable(&*s.as_bytes(), false, 1, true) {
+			Cow::Borrowed(_) => s,
+			Cow::Owned(new) => lua.create_string(&new)?,
+		})
 	})?;
 
 	f.into_lua(lua)
@@ -176,6 +164,33 @@ pub(super) fn truncate(lua: &Lua) -> mlua::Result<Value> {
 	f.into_lua(lua)
 }
 
+pub(super) fn width(lua: &Lua) -> mlua::Result<Value> {
+	let f = lua.create_function(|_, v: Value| match v {
+		Value::String(s) => {
+			let (mut acc, b) = (0, s.as_bytes());
+			for c in b.utf8_chunks() {
+				acc += c.valid().width();
+				if !c.invalid().is_empty() {
+					acc += 1;
+				}
+			}
+			Ok(acc)
+		}
+		Value::UserData(ud) => {
+			if let Ok(line) = ud.borrow::<Line>() {
+				Ok(line.width())
+			} else if let Ok(span) = ud.borrow::<Span>() {
+				Ok(span.width())
+			} else {
+				Err("expected a string, Line, or Span".into_lua_err())?
+			}
+		}
+		_ => Err("expected a string, Line, or Span".into_lua_err())?,
+	})?;
+
+	f.into_lua(lua)
+}
+
 #[cfg(test)]
 mod tests {
 	use mlua::{Lua, chunk};

yazi-plugin/src/elements/elements.rs

@@ -6,7 +6,7 @@ use syntect::{LoadingError, dumps, easy::HighlightLines, highlighting::{self, Th
 use tokio::io::{AsyncBufReadExt, AsyncSeekExt, BufReader};
 use yazi_config::{THEME, YAZI, preview::PreviewWrap};
 use yazi_fs::provider::{Provider, local::Local};
-use yazi_shared::{Ids, errors::PeekError, replace_to_printable};
+use yazi_shared::{Ids, errors::PeekError, push_printable_char};
 
 static INCR: Ids = Ids::new();
 static SYNTECT: OnceLock<(Theme, SyntaxSet)> = OnceLock::new();
@@ -95,7 +95,7 @@ impl Highlighter {
 		}
 
 		Ok(if plain {
-			Text::from(replace_to_printable(&after, YAZI.preview.tab_size))
+			Text::from(Self::merge_highlight_lines(&after, YAZI.preview.tab_size))
 		} else {
 			Self::highlight_with(before, after, syntax.unwrap()).await?
 		})
@@ -203,6 +203,16 @@ impl Highlighter {
 			*b = b'\n';
 		}
 	}
+
+	fn merge_highlight_lines(s: &[String], tab_size: u8) -> String {
+		let mut buf = Vec::new();
+		buf.reserve_exact(s.iter().map(|s| s.len()).sum::<usize>() | 15);
+
+		for &b in s.iter().flat_map(|s| s.as_bytes()) {
+			push_printable_char(&mut buf, b, true, tab_size, false);
+		}
+		unsafe { String::from_utf8_unchecked(buf) }
+	}
 }
 
 impl Highlighter {

yazi-plugin/src/external/highlighter.rs

@@ -94,26 +94,48 @@ pub fn replace_vec_cow<'a>(v: &'a [u8], from: &[u8], to: &[u8]) -> Cow<'a, [u8]>
 	Cow::Owned(out)
 }
 
-pub fn replace_to_printable(s: &[String], tab_size: u8) -> String {
-	let mut buf = Vec::new();
-	buf.try_reserve_exact(s.iter().map(|s| s.len()).sum::<usize>() | 15).unwrap_or_else(|_| panic!());
-
-	for &b in s.iter().flat_map(|s| s.as_bytes()) {
-		match b {
-			b'\n' => buf.push(b'\n'),
-			b'\t' => {
-				buf.extend((0..tab_size).map(|_| b' '));
-			}
-			b'\0'..=b'\x1F' => {
+pub fn replace_to_printable(b: &[u8], lf: bool, tab_size: u8, replacement: bool) -> Cow<'_, [u8]> {
+	// Fast path to skip over printable chars at the beginning of the string
+	let printable_len = b.iter().take_while(|&&c| !c.is_ascii_control()).count();
+	if printable_len >= b.len() {
+		return Cow::Borrowed(b);
+	}
+
+	let (printable, rest) = b.split_at(printable_len);
+
+	let mut out = Vec::new();
+	out.reserve_exact(b.len() | 15);
+	out.extend_from_slice(printable);
+
+	for &c in rest {
+		push_printable_char(&mut out, c, lf, tab_size, replacement);
+	}
+	Cow::Owned(out)
+}
+
+#[inline]
+pub fn push_printable_char(buf: &mut Vec<u8>, c: u8, lf: bool, tab_size: u8, replacement: bool) {
+	match c {
+		b'\n' if lf => buf.push(b'\n'),
+		b'\t' => {
+			buf.extend((0..tab_size).map(|_| b' '));
+		}
+		b'\0'..=b'\x1F' => {
+			if replacement {
+				buf.extend_from_slice(&[0xef, 0xbf, 0xbd]);
+			} else {
 				buf.push(b'^');
-				buf.push(b + b'@');
+				buf.push(c + b'@');
 			}
-			0x7f => {
+		}
+		0x7f => {
+			if replacement {
+				buf.extend_from_slice(&[0xef, 0xbf, 0xbd]);
+			} else {
 				buf.push(b'^');
 				buf.push(b'?');
 			}
-			_ => buf.push(b),
 		}
+		_ => buf.push(c),
 	}
-	unsafe { String::from_utf8_unchecked(buf) }
 }