From 05212749e37aa8f5853ba1493ca279a7b09182b9 Mon Sep 17 00:00:00 2001
From: pagedmov <kylerclay@proton.me>
Date: Sat, 15 Mar 2025 23:52:58 -0400
Subject: [PATCH] Properly implemented word splitting

---
 src/expand.rs          | 41 ++++++++++++++++++++++++++++++++++-------
 src/parse/lex.rs       | 26 +++++++++++++++++---------
 src/prompt/readline.rs | 16 +++++++++++++++-
 3 files changed, 66 insertions(+), 17 deletions(-)

diff --git a/src/expand.rs b/src/expand.rs
index 56bb0e8..56b7473 100644
--- a/src/expand.rs
+++ b/src/expand.rs
@@ -1,7 +1,11 @@
-use crate::{parse::lex::{is_hard_sep, LexFlags, LexStream, Tk, Span, TkFlags, TkRule}, state::read_vars};
+use crate::{prelude::*, parse::lex::{is_field_sep, is_hard_sep, LexFlags, LexStream, Span, Tk, TkFlags, TkRule}, state::read_vars};
 
 /// Variable substitution marker
 pub const VAR_SUB: char = '\u{fdd0}';
+/// Double quote '"' marker
+pub const DUB_QUOTE: char = '\u{fdd1}';
+/// Single quote '\\'' marker
+pub const SNG_QUOTE: char = '\u{fdd2}';
 
 impl<'t> Tk<'t> {
 	/// Create a new expanded token
@@ -34,12 +38,33 @@ impl<'t> Expander {
 	}
 	pub fn expand(&'t mut self) -> Vec<String> {
 		self.raw = self.expand_raw();
-		// Unwrap here is safe because LexFlags::RAW has no error states
-		let tokens: Vec<_> = LexStream::new(&self.raw, LexFlags::RAW)
-			.filter(|tk| !matches!(tk.as_ref().unwrap().class, TkRule::EOI | TkRule::SOI))
-			.map(|tk| tk.unwrap().to_string())
-			.collect();
-		tokens
+		self.split_words()
+	}
+	pub fn split_words(&mut self) -> Vec<String> {
+		let mut words = vec![];
+		let mut chars = self.raw.chars();
+		let mut cur_word = String::new();
+
+		'outer: while let Some(ch) = chars.next() {
+			match ch {
+				DUB_QUOTE | SNG_QUOTE => {
+					while let Some(q_ch) = chars.next() {
+						match q_ch {
+							_ if q_ch == ch => continue 'outer, // Isn't rust cool
+							_ => cur_word.push(q_ch)
+						}
+					}
+				}
+				_ if is_field_sep(ch) => {
+					words.push(mem::take(&mut cur_word));
+				}
+				_ => cur_word.push(ch)
+			}
+		}
+		if !cur_word.is_empty() {
+			words.push(cur_word);
+		}
+		words
 	}
 	pub fn expand_raw(&self) -> String {
 		let mut chars = self.raw.chars();
@@ -97,6 +122,8 @@ pub fn unescape_str(raw: &str) -> String {
 					result.push(next_ch)
 				}
 			}
+			'"' => result.push(DUB_QUOTE),
+			'\'' => result.push(SNG_QUOTE),
 			'$' => result.push(VAR_SUB),
 			_ => result.push(ch)
 		}
diff --git a/src/parse/lex.rs b/src/parse/lex.rs
index a118765..90c2323 100644
--- a/src/parse/lex.rs
+++ b/src/parse/lex.rs
@@ -286,6 +286,19 @@ impl<'t> LexStream<'t> {
 		let mut quote_pos = None;
 		while let Some(ch) = chars.next() {
 			match ch {
+				_ if self.flags.contains(LexFlags::RAW) => {
+					if ch.is_whitespace() {
+						break;
+					} else {
+						pos += ch.len_utf8()
+					}
+				}
+				'\\' => {
+					pos += 1;
+					if chars.next().is_some() {
+						pos += 1;
+					}
+				}
 				'"' | '\'' => {
 					self.in_quote = true;
 					quote_pos = Some(pos);
@@ -293,8 +306,10 @@ impl<'t> LexStream<'t> {
 					while let Some(q_ch) = chars.next() {
 						match q_ch {
 							'\\' => {
-								pos += 2;
-								chars.next();
+								pos += 1;
+								if chars.next().is_some() {
+									pos += 1;
+								}
 							}
 							_ if q_ch == ch => {
 								pos += 1;
@@ -309,13 +324,6 @@ impl<'t> LexStream<'t> {
 						}
 					}
 				}
-				_ if self.flags.contains(LexFlags::RAW) => {
-					if ch.is_whitespace() {
-						break;
-					} else {
-						pos += ch.len_utf8()
-					}
-				}
 				_ if !self.in_quote && is_op(ch) => break,
 				_ if is_hard_sep(ch) => break,
 				_ => pos += ch.len_utf8()
diff --git a/src/prompt/readline.rs b/src/prompt/readline.rs
index 4db8b36..eab13c8 100644
--- a/src/prompt/readline.rs
+++ b/src/prompt/readline.rs
@@ -2,7 +2,7 @@ use std::borrow::Cow;
 
 use rustyline::{completion::Completer, highlight::Highlighter, hint::{Hint, Hinter}, validate::{ValidationResult, Validator}, Helper};
 
-use crate::libsh::term::{Style, Styled};
+use crate::{libsh::term::{Style, Styled}, parse::{lex::{LexFlags, LexStream}, ParseStream}};
 
 pub struct FernReadline {
 }
@@ -65,6 +65,20 @@ impl Highlighter for FernReadline {
 
 impl Validator for FernReadline {
 	fn validate(&self, ctx: &mut rustyline::validate::ValidationContext) -> rustyline::Result<rustyline::validate::ValidationResult> {
+		let mut tokens = vec![];
+		let tk_stream = LexStream::new(ctx.input(), LexFlags::empty());
+		for tk in tk_stream {
+			if tk.is_err() {
+				return Ok(ValidationResult::Incomplete)
+			}
+			tokens.push(tk.unwrap());
+		}
+		let nd_stream = ParseStream::new(tokens);
+		for nd in nd_stream {
+			if nd.is_err() {
+				return Ok(ValidationResult::Incomplete)
+			}
+		}
 		Ok(ValidationResult::Valid(None))
 	}
 }