feat: invalid UTF-8 support to Command:arg() and Command:env() APIs (#2895)

yazi-binding/src/url.rs

@@ -139,7 +139,7 @@ impl UserData for Url {
 		});
 
 		methods.add_function_mut("into_search", |_, (ud, frag): (AnyUserData, mlua::String)| {
-			Ok(Self::new(ud.take::<Self>()?.inner.into_search(frag.as_bytes().as_ref().into_os_str()?)))
+			Ok(Self::new(ud.take::<Self>()?.inner.into_search(frag.as_bytes().into_os_str()?)))
 		});
 
 		methods.add_meta_method(MetaMethod::Eq, |_, me, other: UrlRef| Ok(me.inner == other.inner));

yazi-core/src/mgr/commands/update_mimes.rs

@@ -1,20 +1,20 @@
-use std::{borrow::Cow, collections::HashMap};
+use std::collections::HashMap;
 
 use tracing::error;
 use yazi_macro::render;
-use yazi_shared::{event::CmdCow, url::Url};
+use yazi_shared::event::{CmdCow, Data, DataKey};
 
 use crate::{mgr::{LINKED, Mgr}, tasks::Tasks};
 
 pub struct Opt {
-	updates: HashMap<Cow<'static, str>, Cow<'static, str>>,
+	updates: HashMap<DataKey, Data>,
 }
 
 impl TryFrom<CmdCow> for Opt {
 	type Error = ();
 
 	fn try_from(mut c: CmdCow) -> Result<Self, Self::Error> {
-		Ok(Self { updates: c.try_take("updates").ok_or(())?.into_dict_string() })
+		Ok(Self { updates: c.try_take("updates").and_then(Data::into_dict).ok_or(())? })
 	}
 }
 
@@ -28,7 +28,7 @@ impl Mgr {
 		let updates = opt
 			.updates
 			.into_iter()
-			.flat_map(|(url, mime)| Url::try_from(url.as_ref()).map(|u| (u, mime)))
+			.flat_map(|(key, value)| key.into_url().zip(value.into_string()))
 			.filter(|(url, mime)| self.mimetype.by_url(url) != Some(mime))
 			.fold(HashMap::new(), |mut map, (u, m)| {
 				for u in linked.from_file(&u) {

yazi-plugin/src/process/command.rs

@@ -3,6 +3,7 @@ use std::{any::TypeId, io, process::Stdio};
 use mlua::{AnyUserData, ExternalError, IntoLuaMulti, Lua, MetaMethod, Table, UserData, Value};
 use tokio::process::{ChildStderr, ChildStdin, ChildStdout};
 use yazi_binding::Error;
+use yazi_shared::IntoOsStr;
 
 use super::{Child, output::Output};
 use crate::process::Status;
@@ -143,11 +144,11 @@ impl UserData for Command {
 				let mut me = ud.borrow_mut::<Self>()?;
 				match arg {
 					Value::String(s) => {
-						me.inner.arg(String::from_utf8_lossy(&s.as_bytes()).as_ref());
+						me.inner.arg(s.as_bytes().into_os_str()?);
 					}
 					Value::Table(t) => {
 						for s in t.sequence_values::<mlua::String>() {
-							me.inner.arg(String::from_utf8_lossy(&s?.as_bytes()).as_ref());
+							me.inner.arg(s?.as_bytes().into_os_str()?);
 						}
 					}
 					_ => return Err("arg must be a string or table of strings".into_lua_err()),
@@ -162,10 +163,9 @@ impl UserData for Command {
 		methods.add_function_mut(
 			"env",
 			|_, (ud, key, value): (AnyUserData, mlua::String, mlua::String)| {
-				ud.borrow_mut::<Self>()?.inner.env(
-					String::from_utf8_lossy(&key.as_bytes()).as_ref(),
-					String::from_utf8_lossy(&value.as_bytes()).as_ref(),
-				);
+				ud.borrow_mut::<Self>()?
+					.inner
+					.env(key.as_bytes().into_os_str()?, value.as_bytes().into_os_str()?);
 				Ok(ud)
 			},
 		);

yazi-shared/src/event/data.rs

@@ -46,6 +46,32 @@ impl Data {
 		}
 	}
 
+	#[inline]
+	pub fn into_string(self) -> Option<Cow<'static, str>> {
+		match self {
+			Self::String(s) => Some(s),
+			_ => None,
+		}
+	}
+
+	#[inline]
+	pub fn into_dict(self) -> Option<HashMap<DataKey, Data>> {
+		match self {
+			Self::Dict(d) => Some(d),
+			_ => None,
+		}
+	}
+
+	#[inline]
+	pub fn into_url(self) -> Option<Url> {
+		match self {
+			Self::String(s) => Url::try_from(s.as_ref()).ok(),
+			Self::Url(u) => Some(u),
+			Self::Bytes(b) => Url::try_from(b.as_slice()).ok(),
+			_ => None,
+		}
+	}
+
 	#[inline]
 	pub fn into_any<T: 'static>(self) -> Option<T> {
 		match self {
@@ -54,34 +80,12 @@ impl Data {
 		}
 	}
 
-	#[inline]
-	pub fn into_url(self) -> Option<Url> {
-		match self {
-			Data::String(s) => Url::try_from(s.as_ref()).ok(),
-			Data::Url(u) => Some(u),
-			_ => None,
-		}
-	}
-
-	pub fn into_dict_string(self) -> HashMap<Cow<'static, str>, Cow<'static, str>> {
-		let Self::Dict(dict) = self else {
-			return Default::default();
-		};
-
-		let mut map = HashMap::with_capacity(dict.len());
-		for pair in dict {
-			if let (DataKey::String(k), Self::String(v)) = pair {
-				map.insert(k, v);
-			}
-		}
-		map
-	}
-
 	#[inline]
 	pub fn to_url(&self) -> Option<Url> {
 		match self {
 			Self::String(s) => Url::try_from(s.as_ref()).ok(),
 			Self::Url(u) => Some(u.clone()),
+			Self::Bytes(b) => Url::try_from(b.as_slice()).ok(),
 			_ => None,
 		}
 	}
@@ -158,6 +162,16 @@ impl DataKey {
 		}
 	}
 
+	#[inline]
+	pub fn into_url(self) -> Option<Url> {
+		match self {
+			Self::String(s) => Url::try_from(s.as_ref()).ok(),
+			Self::Url(u) => Some(u),
+			Self::Bytes(b) => Url::try_from(b.as_slice()).ok(),
+			_ => None,
+		}
+	}
+
 	fn deserialize_integer<'de, D>(deserializer: D) -> Result<i64, D::Error>
 	where
 		D: de::Deserializer<'de>,
@@ -204,9 +218,9 @@ macro_rules! impl_as_integer {
 			#[inline]
 			pub fn $name(&self) -> Option<$t> {
 				match self {
-					Data::Integer(i) => <$t>::try_from(*i).ok(),
-					Data::String(s) => s.parse().ok(),
-					Data::Id(i) => <$t>::try_from(i.get()).ok(),
+					Self::Integer(i) => <$t>::try_from(*i).ok(),
+					Self::String(s) => s.parse().ok(),
+					Self::Id(i) => <$t>::try_from(i.get()).ok(),
 					_ => None,
 				}
 			}
@@ -220,10 +234,10 @@ macro_rules! impl_as_number {
 			#[inline]
 			pub fn $name(&self) -> Option<$t> {
 				match self {
-					Data::Integer(i) if *i == (*i as $t as _) => Some(*i as $t),
-					Data::Number(n) => <$t>::try_from(*n).ok(),
-					Data::String(s) => s.parse().ok(),
-					Data::Id(i) if i.0 == (i.0 as $t as _) => Some(i.0 as $t),
+					Self::Integer(i) if *i == (*i as $t as _) => Some(*i as $t),
+					Self::Number(n) => <$t>::try_from(*n).ok(),
+					Self::String(s) => s.parse().ok(),
+					Self::Id(i) if i.0 == (i.0 as $t as _) => Some(i.0 as $t),
 					_ => None,
 				}
 			}