Improve backslash and quote handling in words

2 files changed, 30 insertions, 5 deletions

diff --git a/parsing/lexer.sh b/parsing/lexer.sh
index c856d99..53af926 100644
--- a/parsing/lexer.sh
+++ b/parsing/lexer.sh
@@ -429,6 +429,23 @@ scan_word()
 					'command substitution not supported'
 				break
 				;;
+			\\)
+				word="${word}${c}"
+				lgetc
+				case "${c}" in '')
+					# Bash, ksh93, mksh, and zsh ignore a
+					# backslash at the end of a file, but
+					# dash and BusyBox ash include it in the
+					# word.  To help with script
+					# portability, we'll throw an error
+					# (which is a reasonable thing to do
+					# anyway).
+					synerr 'Unexpected end of file %s' \
+						'after "\"'
+					;;
+				esac
+				word="${word}${c}"
+				;;
 			\')
 				word="${word}${c}"
 				while :; do
@@ -461,6 +478,10 @@ scan_word()
 		lgetc
 	done
 
+	if ${quoted}; then
+		synerr 'Unterminated quoted string'
+	fi
+
 	printf "%d${RS}%c${RS}%s" ${lines} "${c}" "${word}"
 }
 
diff --git a/parsing/parse.sh b/parsing/parse.sh
index 600a247..c9e8000 100644
--- a/parsing/parse.sh
+++ b/parsing/parse.sh
@@ -659,9 +659,13 @@ try()
 #try 'foo <<EOF' 'bar' 'EOF'
 #try 'foo <<-EOF' "${HT}bar" "${HT}EOF"
 #try 'foo <<EOF' '$(bar)' 'EOF'
-try 'foo <<E"O"F' '$(bar)' 'EOF'
-try 'foo <<"EOF"' '$(bar)' 'EOF'
-try 'foo <<E\OF' '$(bar)' 'EOF'
-try 'foo <<\EOF' '$(bar)' 'EOF'
+#try 'foo <<E"O"F' '$(bar)' 'EOF'
+#try 'foo <<"EOF"' '$(bar)' 'EOF'
+#try 'foo <<E\OF' '$(bar)' 'EOF'
+#try 'foo <<\EOF' '$(bar)' 'EOF'
 #try 'foo <<EOF1; bar <<EOF2' 'baz' 'EOF1' 'qux' 'EOF2'
-try '\foo'
+#try '\foo'
+try '"foo bar" baz'
+try '"foo'
+try 'foo\" bar'
+try 'foo\'