From 234f93ee56725c9b459a284e626340af6d08646d Mon Sep 17 00:00:00 2001
From: pagedmov <kylerclay@proton.me>
Date: Thu, 5 Mar 2026 01:26:42 -0500
Subject: [PATCH] fixed comments breaking the parser

---
 examples/surround_plugin.sh | 140 ++++++++++++++++++++++++++++++++++++
 src/parse/lex.rs            |   9 ++-
 src/parse/mod.rs            |   2 +-
 test_scripts/cmdsub.sh      |   1 -
 test_scripts/hello.sh       |   1 -
 test_scripts/multiline.sh   |   4 --
 6 files changed, 148 insertions(+), 9 deletions(-)
 create mode 100644 examples/surround_plugin.sh
 delete mode 100755 test_scripts/cmdsub.sh
 delete mode 100755 test_scripts/hello.sh
 delete mode 100644 test_scripts/multiline.sh

diff --git a/examples/surround_plugin.sh b/examples/surround_plugin.sh
new file mode 100644
index 0000000..82d751b
--- /dev/null
+++ b/examples/surround_plugin.sh
@@ -0,0 +1,140 @@
+_get_surround_target() {
+	read_key -v _s_ch
+	case "$_s_ch" in
+	  \(|\)) _sl='('; _sr=')' ;;
+	  \[|\]) _sl='['; _sr=']' ;;
+	  \{|\}) _sl='{'; _sr='}' ;;
+	  \<|\>) _sl='<'; _sr='>' ;;
+	  *) _sl="$_s_ch"; _sr="$_s_ch" ;;
+	esac
+
+}
+_read_obj() {
+	_obj=""
+	while read_key -v key; do
+	  if [[ "${#_obj}" -ge 3 ]]; then return 1; fi
+	  case "$key" in
+	    i|a)
+	      if [ -n "$_obj" ]; then return 1; fi
+	      _obj="$key"
+	      ;;
+	    b|e)
+	      if [ -n "$_obj" ]; then return 1; fi
+	      _obj="$key"
+	      break
+	      ;;
+	    w|W)
+	      _obj="$_obj$key"
+	      break
+	      ;;
+	    f|F)
+	      read_key -v char
+	      _obj="$key$char"
+	      break
+	    ;;
+	    \(|\)|\[|\]|\{|\}|\"|\')
+	      if [ -z "$_obj" ]; then return 1; fi
+	      _obj="$_obj$key"
+	      break
+	      ;;
+	  esac
+	done
+
+}
+_scan_left() {
+	local needle="$1"
+	local haystack="$2"
+	local i=$((${#haystack} - 1))
+
+
+	while [ "$i" -ge 0 ]; do
+	  ch="${haystack:$i:1}"
+	  if [ "$ch" = "$needle" ]; then
+	    left=$i
+	    return 0
+	  fi
+	  i=$((i - 1))
+	done
+
+	return 1
+
+}
+_scan_right() {
+	local needle="$1"
+	local haystack="$2"
+	local i=0
+
+
+	while [ "$i" -lt "${#haystack}" ]; do
+	  ch="${haystack:$i:1}"
+	  if [ "$ch" = "$needle" ]; then
+	    right="$i"
+	    return 0
+	  fi
+	  i=$((i + 1))
+	done
+
+	return 1
+
+}
+_surround_1() {
+	local _obj
+	_read_obj
+	_get_surround_target
+	_KEYS="v$_obj"
+
+}
+_surround_2() {
+	local start
+	local end
+	if [ "$_ANCHOR" -lt "$_CURSOR" ]; then
+	  start=$_ANCHOR
+	  end=$_CURSOR
+	else
+	  start=$_CURSOR
+	  end=$_ANCHOR
+	fi
+	end=$((end + 1))
+
+	delta=$((end - start))
+
+	left="${_BUFFER:0:$start}"
+	mid="${_BUFFER:$start:$delta}"
+	right="${_BUFFER:$end}"
+	_BUFFER="$left$_sl$mid$_sr$right"
+	_CURSOR=$start
+
+}
+_surround_del() {
+	_get_surround_target
+	local left_buf="${_BUFFER:0:$_CURSOR}"
+	local right_buf="${_BUFFER:$left}"
+	local left=""
+	local right=""
+	_scan_left $_sl "$left_buf"
+
+	if [ "$?" -ne 0 ]; then
+	  _scan_right $_sl "$right_buf"
+
+	  [ "$?" -ne 0 ] && return 1
+	  left=$right
+	fi
+
+	mid_start=$((left + 1))
+	right=""
+	left_buf="${_BUFFER:0:$left}"
+	right_buf="${_BUFFER:$mid_start}"
+	_scan_right $_sr "$right_buf"
+
+	[ "$?" -ne 0 ] && return 1
+
+	mid_end=$((mid_start + right))
+	right_start=$((mid_end + 1))
+	new_left_buf="${_BUFFER:0:$left}"
+	new_mid_buf="${_BUFFER:$mid_start:$right}"
+	new_right_buf="${_BUFFER:$right_start}"
+
+
+	_BUFFER="$new_left_buf$new_mid_buf$new_right_buf"
+
+}
diff --git a/src/parse/lex.rs b/src/parse/lex.rs
index 0b54a47..262b938 100644
--- a/src/parse/lex.rs
+++ b/src/parse/lex.rs
@@ -868,13 +868,18 @@ impl Iterator for LexStream {
         self.cursor += 1;
 
         while let Some(ch) = get_char(&self.source, self.cursor) {
-          self.cursor += 1;
+          self.cursor += ch.len_utf8();
           if ch == '\n' {
             break;
           }
         }
 
-        self.get_token(ch_idx..self.cursor, TkRule::Comment)
+        if self.flags.contains(LexFlags::LEX_UNFINISHED) {
+          self.get_token(ch_idx..self.cursor, TkRule::Comment)
+        } else {
+					// After consuming the comment, we call next() recursively. This effectively filters out comment tokens.
+          return self.next();
+        }
       }
       '|' => {
         let ch_idx = self.cursor;
diff --git a/src/parse/mod.rs b/src/parse/mod.rs
index 05d9b29..797181e 100644
--- a/src/parse/mod.rs
+++ b/src/parse/mod.rs
@@ -87,6 +87,7 @@ impl ParsedSrc {
         Err(error) => return Err(vec![error]),
       }
     }
+		log::debug!("Tokens: {:#?}", tokens);
 
     let mut errors = vec![];
     let mut nodes = vec![];
@@ -1595,7 +1596,6 @@ impl ParseStream {
             redirs.push(redir);
           }
         }
-        TkRule::Comment => { /* Skip comments in command position */ }
         _ => unimplemented!("Unexpected token rule `{:?}` in parse_cmd()", tk.class),
       }
     }
diff --git a/test_scripts/cmdsub.sh b/test_scripts/cmdsub.sh
deleted file mode 100755
index 5fcae70..0000000
--- a/test_scripts/cmdsub.sh
+++ /dev/null
@@ -1 +0,0 @@
-echo "foo $(echo "$(echo "$(echo "$(echo Hello)")")") bar"
diff --git a/test_scripts/hello.sh b/test_scripts/hello.sh
deleted file mode 100755
index 8fbf7fa..0000000
--- a/test_scripts/hello.sh
+++ /dev/null
@@ -1 +0,0 @@
-echo Hello, World!
diff --git a/test_scripts/multiline.sh b/test_scripts/multiline.sh
deleted file mode 100644
index 0e88b53..0000000
--- a/test_scripts/multiline.sh
+++ /dev/null
@@ -1,4 +0,0 @@
-echo foo
-echo bar
-echo biz
-echo buzz