From 506a0723291b4abc4b707a6353a840ef51f95ee8 Mon Sep 17 00:00:00 2001
From: P. J. McDermott <pj@pehjota.net>
Date: Sun, 21 Feb 2016 12:12:20 -0500
Subject: eshtrans/frontend: Fix handling of words beginning with "\"

---
(limited to 'eshtrans/frontend/lexer.esh')

diff --git a/eshtrans/frontend/lexer.esh b/eshtrans/frontend/lexer.esh
index 0991239..d904aa0 100644
--- a/eshtrans/frontend/lexer.esh
+++ b/eshtrans/frontend/lexer.esh
@@ -361,14 +361,14 @@ next_word()
 	local res=
 	local word=
 
-	if ! res="$(scan_word false)"; then
+	if ! res="$(scan_word false "${prev_c}")"; then
 		exit 1
 	fi
 	ln_off=${res%%${RS}*}
 	res="${res#*${RS}}"
 	c="${res%%${RS}*}"
 	res="${res#*${RS}}"
-	word="${prev_c}${res%%${RS}*}"
+	word="${res%%${RS}*}"
 
 	# We must advance lineno because scan_word() was run in a subshell.
 	lineno=$((${lineno} + ${ln_off}))
@@ -396,16 +396,28 @@ next_word()
 scan_word()
 {
 	local in_param="${1}"
-	shift 1
-	local res=
+	local prev_c="${2}"
+	shift 2
+	local lines=
 	local word=
 	local quoted=
-	local lines=
+	local tmp_c=
+	local res=
 	local wordexp=
 
+	lines=0
 	word=''
 	quoted=false
-	lines=0
+
+	# Sort of a localized ungetc().
+	case "${prev_c}" in
+		'') ;;
+		*)
+			tmp_c="${c}"
+			c="${prev_c}"
+			;;
+	esac
+
 	while :; do
 		dbg "parsing word char '$c' at lineno $lineno"
 		case "${c}" in
@@ -436,7 +448,10 @@ scan_word()
 							'delimiters'
 					fi
 				esac
-				lgetc
+				case "${prev_c}" in
+					'') lgetc;;
+					*) c="${tmp_c}"; prev_c='';;
+				esac
 				if ! res=$(scan_wordexp); then
 					exit 1
 				fi
@@ -460,8 +475,13 @@ scan_word()
 				break
 				;;
 			\\)
+				dbg 'first backslash in word'
 				word="${word}${c}"
-				lgetc
+				case "${prev_c}" in
+					'') lgetc;;
+					*) c="${tmp_c}"; prev_c='';;
+				esac
+				dbg "next char: '$c'"
 				case "${c}" in '')
 					# Bash, ksh93, mksh, and zsh ignore a
 					# backslash at the end of a file, but
@@ -479,7 +499,10 @@ scan_word()
 			\')
 				word="${word}${c}"
 				while :; do
-					lgetc
+					case "${prev_c}" in
+						'') lgetc;;
+						*) c="${tmp_c}"; prev_c='';;
+					esac
 					word="${word}${c}"
 					case "${c}" in
 						'')
@@ -511,7 +534,10 @@ scan_word()
 				word="${word}${c}"
 				;;
 		esac
-		lgetc
+		case "${prev_c}" in
+			'') lgetc;;
+			*) c="${tmp_c}"; prev_c='';;
+		esac
 	done
 
 	if ${quoted}; then
@@ -673,7 +699,7 @@ scan_wordexp_param_brace()
 	# If a modification was found
 	if ${mod}; then
 		# Get word.
-		if ! res="$(scan_word true)"; then
+		if ! res="$(scan_word true '')"; then
 			exit 1
 		fi
 		ln_off=${res%%${RS}*}
--
cgit v0.9.1