feat: ya emit and ya emit-to support invalid UTF-8 as command argument values (#3290)

Author: sxyazi

CHANGELOG.md

@@ -19,19 +19,19 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 - Shell formatting ([#3232])
 - Multi-entry support for plugin system ([#3154])
 - Zoom in or out of the preview image ([#2864])
-- Improve the UX of the pick and input component ([#2906], [#2935])
+- Improve the UX of the pick and input components ([#2906], [#2935])
 - Show progress of each task in task manager ([#3121], [#3131], [#3134])
 - New `bulk_rename` command always renames files with the editor ([#2984])
 - `key-*` DDS events to allow changing or canceling user key events ([#3005], [#3037])
-- New `--bg` specifying image background color for the preset `svg` and `magick` previewers ([#3189])
+- New `--bg` specifying image background color in the preset SVG and ImageMagick previewers ([#3189])
 - `filter` by full path (prefix + filename) in search view instead of just filename ([#2915])
 - New `casefy` command for case conversion of the input content ([#3235])
 - Allow dynamic adjustment of layout ratio via `rt.mgr.ratio` ([#2964])
 - Support `.deb` packages ([#2807], [#3128], [#3209])
 - Port several widespread GUI keys to the input component ([#2849])
-- Support invalid UTF-8 paths throughout the codebase ([#2884], [#2889], [#2890], [#2895], [#3023])
+- Support invalid UTF-8 paths throughout the codebase ([#2884], [#2889], [#2890], [#2895], [#3023], [#3290])
 - Allow upgrading only specific packages with `ya pkg` ([#2841])
-- Respect the user's `image_filter` setting for the preset `magick` previewer ([#3286])
+- Respect the user's `image_filter` setting in the preset ImageMagick previewer ([#3286])
 - Allow custom mouse click behavior for individual files ([#2925])
 - Display newlines in input as spaces to improve readability ([#2932])
 - Fill in error messages if preview fails ([#2917])
@@ -49,7 +49,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 
 ### Deprecated
 
-- Deprecate `$n`, `$@` (Unix-like) and `%n`, `%*` (Windows) in `shell` command and opener rules in favor of new shell formatting ([#3232])
+- Deprecate `$n`, `$@` (\*nix) and `%n`, `%*` (Windows) in `shell` command and opener rules in favor of new shell formatting ([#3232])
 - Deprecate `ya.hide`, `ya.render`, and `ya.truncate` in favor of `ui.hide`, `ui.render`, and `ui.truncate` ([#2939])
 - Deprecate `position` property of `ya.input()` in favor of `pos` to align with `ya.confirm()` and its type `ui.Pos` ([#2921])
 - Deprecate `cx.tasks.progress` in favor of `cx.tasks.summary` ([#3131])
@@ -84,7 +84,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 - Zero-copy `UrlBuf` to `Url` conversion ([#3117])
 - String interning to reduce memory usage of mimetype and URL domain ([#3084], [#3091])
 - Do not pre-allocate memory for Lua tables ([#2879])
-- Copy-on-write on command data & avoid converting primitive types to strings thereby allocating memory ([#2862])
+- Copy-on-write on command data, and avoid converting primitive types to strings thereby allocating memory ([#2862])
 - Use `AnyUserData::type_id()` to reduce stack pushes ([#2834])
 - App data instead of Lua registry to reduce stack pushes ([#2880])
 
@@ -140,7 +140,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 
 ### Fixed
 
-- Respect the user's `max_width` setting for the built-in video previewer ([#2560])
+- Respect the user's `max_width` setting in the preset video previewer ([#2560])
 - Reverse the mixing order of theme and flavor configuration ([#2594])
 - No title is set when starts the first time ([#2700])
 - `ya pub-to 0` checks if any peer is able to receive the message ([#2697])
@@ -227,9 +227,9 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 - New `tbl_col` and `tbl_cell` in theme system for spotter table styling ([#2391])
 - Allow different separators to be applied individually to the left and right sides of the status bar ([#2313])
 - `ripgrep-all` support for the `search` command ([#2383])
-- Respect the user's `max_width` setting for the built-in PDF preloader ([#2331])
-- Respect the user's `wrap` setting for the built-in JSON previewer ([#2337])
-- Respect the user's `image_alloc` setting for the built-in ImageMagick previewer ([#2403])
+- Respect the user's `max_width` setting in the preset PDF preloader ([#2331])
+- Respect the user's `wrap` setting in the preset JSON previewer ([#2337])
+- Respect the user's `image_alloc` setting in the preset ImageMagick previewer ([#2403])
 - New `external` and `removable` fields in the `fs.partitions()` API ([#2343])
 - CSI-based Vim and Neovim built-in terminal detection for better accuracy ([#2327])
 
@@ -1532,3 +1532,4 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/):
 [#3268]: https://github.com/sxyazi/yazi/pull/3268
 [#3271]: https://github.com/sxyazi/yazi/pull/3271
 [#3286]: https://github.com/sxyazi/yazi/pull/3286
+[#3290]: https://github.com/sxyazi/yazi/pull/3290

yazi-cli/src/args.rs

@@ -1,8 +1,8 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, ffi::OsString};
 
 use anyhow::{Result, bail};
 use clap::{Parser, Subcommand};
-use yazi_shared::{Id, event::Cmd};
+use yazi_shared::{Either, Id};
 
 #[derive(Parser)]
 #[command(name = "Ya", about, long_about = None)]
@@ -38,7 +38,7 @@ pub(super) struct CommandEmit {
 	pub(super) name: String,
 	/// Arguments of the command.
 	#[arg(allow_hyphen_values = true, trailing_var_arg = true)]
-	pub(super) args: Vec<String>,
+	pub(super) args: Vec<OsString>,
 }
 
 #[derive(clap::Args)]
@@ -49,7 +49,7 @@ pub(super) struct CommandEmitTo {
 	pub(super) name:     String,
 	/// Arguments of the command.
 	#[arg(allow_hyphen_values = true, trailing_var_arg = true)]
-	pub(super) args:     Vec<String>,
+	pub(super) args:     Vec<OsString>,
 }
 
 #[derive(Subcommand)]
@@ -139,10 +139,11 @@ macro_rules! impl_emit_body {
 		impl $name {
 			#[allow(dead_code)]
 			pub(super) fn body(self) -> Result<String> {
-				Ok(serde_json::to_string(&(
-					self.name,
-					Cmd::parse_args(self.args.into_iter(), None, false)?,
-				))?)
+				let cmd: Vec<_> = [Either::Left(self.name)]
+					.into_iter()
+					.chain(self.args.into_iter().map(|s| Either::Right(s.into_encoded_bytes())))
+					.collect();
+				Ok(serde_json::to_string(&cmd)?)
 			}
 		}
 	};

yazi-parser/src/app/plugin.rs

@@ -30,7 +30,7 @@ impl TryFrom<CmdCow> for PluginOpt {
 
 		let args = if let Ok(s) = c.second() {
 			let (words, last) = yazi_shared::shell::split_unix(s, true)?;
-			Cmd::parse_args(words.into_iter(), last, true)?
+			Cmd::parse_args(words, last)?
 		} else {
 			Default::default()
 		};

yazi-plugin/preset/plugins/dds.lua

@@ -1,7 +1,21 @@
 local M = {}
 
 function M:setup()
-	ps.sub_remote("dds-emit", function(cmd) ya.emit(cmd[1], cmd[2]) end)
+	ps.sub_remote("dds-emit", function(cmd)
+		local i, args = 1, {}
+		for j = 2, #cmd do
+			local word = string.char(table.unpack(cmd[j]))
+			local key = word:match("^%-%-([^=]+)")
+			if not key then
+				i, args[i] = i + 1, word
+			elseif #key + 2 == #word then
+				args[key] = true
+			else
+				args[key] = word:sub(#key + 4)
+			end
+		end
+		ya.emit(cmd[1], args)
+	end)
 end
 
 return M

yazi-shared/src/either.rs

@@ -1,3 +1,5 @@
+use serde::Serialize;
+
 #[derive(Clone, Copy, Debug)]
 pub enum Either<L, R> {
 	Left(L),
@@ -69,3 +71,19 @@ impl<L, R> Either<L, R> {
 		}
 	}
 }
+
+impl<L, R> Serialize for Either<L, R>
+where
+	L: Serialize,
+	R: Serialize,
+{
+	fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
+	where
+		S: serde::Serializer,
+	{
+		match self {
+			Self::Left(l) => l.serialize(serializer),
+			Self::Right(r) => r.serialize(serializer),
+		}
+	}
+}

yazi-shared/src/event/cmd.rs

@@ -178,34 +178,26 @@ impl Cmd {
 	}
 
 	// Parse
-	pub fn parse_args(
-		words: impl Iterator<Item = String>,
-		last: Option<String>,
-		obase: bool,
-	) -> Result<HashMap<DataKey, Data>> {
+	pub fn parse_args<I>(words: I, last: Option<String>) -> Result<HashMap<DataKey, Data>>
+	where
+		I: IntoIterator<Item = String>,
+	{
 		let mut i = 0i64;
 		words
 			.into_iter()
 			.map(|s| (s, true))
 			.chain(last.into_iter().map(|s| (s, false)))
 			.map(|(word, normal)| {
-				let Some(arg) = word.strip_prefix("--").filter(|_| normal) else {
+				let Some(arg) = word.strip_prefix("--").filter(|&s| normal && !s.is_empty()) else {
 					i += 1;
-					return Ok((DataKey::Integer(i - obase as i64), Data::String(word.into())));
+					return Ok((DataKey::Integer(i - 1), word.into()));
 				};
 
 				let mut parts = arg.splitn(2, '=');
-				let Some(key) = parts.next().map(|s| s.to_owned()) else {
-					bail!("invalid argument: {arg}");
-				};
-
-				let val = if let Some(val) = parts.next() {
-					Data::String(val.to_owned().into())
-				} else {
-					Data::Boolean(true)
-				};
+				let key = parts.next().expect("at least one part");
+				let val = parts.next().map_or(Data::Boolean(true), Data::from);
 
-				Ok((DataKey::String(Cow::Owned(key)), val))
+				Ok((DataKey::from(key.to_owned()), val))
 			})
 			.collect()
 	}
@@ -245,7 +237,7 @@ impl FromStr for Cmd {
 		}
 
 		let mut me = Self::new(mem::take(&mut words[0]), Default::default(), Some(Default::default()))?;
-		me.args = Self::parse_args(words.into_iter().skip(1), last, true)?;
+		me.args = Self::parse_args(words.into_iter().skip(1), last)?;
 		Ok(me)
 	}
 }