Add error handling to lexer

2 files changed, 152 insertions, 64 deletions

diff --git a/parsing/lexer.sh b/parsing/lexer.sh
index 3081a79..1f40138 100644
--- a/parsing/lexer.sh
+++ b/parsing/lexer.sh
@@ -13,7 +13,7 @@ pgetc()
 next()
 {
 	while :; do
-		echo "parsing char '$c'" >&2
+		dbg "parsing char '$c' at lineno $lineno"
 		case "${c}" in
 			'')
 				pgetc
@@ -143,7 +143,9 @@ next_word()
 	local lineno_offset=
 	local word=
 
-	res="$(scan_word false)"
+	if ! res="$(scan_word false)"; then
+		exit 1
+	fi
 	lineno_offset=${res%%${RS}*}
 	res="${res#*${RS}}"
 	c="${res%%${RS}*}"
@@ -169,7 +171,7 @@ scan_word()
 	quoted=false
 	lines=0
 	while :; do
-		echo "parsing word char '$c'" >&2
+		dbg "parsing word char '$c' at lineno $lineno"
 		case "${c}" in
 			'')
 				break
@@ -190,7 +192,9 @@ scan_word()
 				;;
 			'$')
 				pgetc
-				res=$(scan_wordexp)
+				if ! res=$(scan_wordexp); then
+					exit 1
+				fi
 				lineno_offset=${res%%${RS}*}
 				res="${res#*${RS}}"
 				c="${res%%${RS}*}"
@@ -205,6 +209,11 @@ scan_word()
 				# below.
 				continue
 				;;
+			'`')
+				synerr 'Backquoted (old-style) %s' \
+					'command substitution not supported'
+				break
+				;;
 			\')
 				word="${word}${c}"
 				while :; do
@@ -243,16 +252,18 @@ scan_word()
 scan_wordexp()
 {
 	local wordexp=
-	local lineno_offset=
+	local ln_off=
 	local mod=
 	local res=
+	local param=
 	local word=
 
 	wordexp=''
-	lineno_offset=0
+	ln_off=0
 	case "${c}" in
 		'{')
 			# Parameter expansion brace
+			mod=true
 			pgetc
 			case "${c}" in
 				'#')
@@ -261,7 +272,16 @@ scan_wordexp()
 						[@*#?$!A-Za-z0-9_-])
 							# String length
 							# expansion
-							next_param
+							res="$(scan_param)"
+							ln_off=${res%%${RS}*}
+							res="${res#*${RS}}"
+							c="${res%%${RS}*}"
+							res="${res#*${RS}}"
+							param="${res%%${RS}*}"
+							lineno=$((${lineno} + \
+								${ln_off}))
+							# Disable modifications.
+							mod=false
 							;;
 						*)
 							# Special parameter "#"
@@ -270,62 +290,72 @@ scan_wordexp()
 					esac
 					;;
 				*)
-					next_param
+					if ! res="$(scan_param)"; then
+						exit 1
+					fi
+					ln_off=${res%%${RS}*}
+					res="${res#*${RS}}"
+					c="${res%%${RS}*}"
+					res="${res#*${RS}}"
+					param="${res%%${RS}*}"
+					lineno=$((${lineno} + ${ln_off}))
 					;;
 			esac
 			wordexp="\${${param}"
-			# Check for modifications
-			mod=false
-			case "${c}" in
-				':')
-					mod=true
-					wordexp="${wordexp}${c}"
-					pgetc
-					case "${c}" in '-'|'='|'?'|'+')
+			if ${mod}; then
+				# Check for modifications
+				mod=false
+				case "${c}" in
+					':')
+						mod=true
 						wordexp="${wordexp}${c}"
 						pgetc
+						case "${c}" in '-'|'='|'?'|'+')
+							wordexp="${wordexp}${c}"
+							pgetc
 						;;
-					esac
-					;;
-				'-'|'='|'?'|'+')
-					mod=true
-					wordexp="${wordexp}${c}"
-					pgetc
-					;;
-				'%')
-					mod=true
-					wordexp="${wordexp}${c}"
-					pgetc
-					case "${c}" in '%')
+						esac
+						;;
+					'-'|'='|'?'|'+')
+						mod=true
 						wordexp="${wordexp}${c}"
 						pgetc
 						;;
-					esac
-					;;
-				'#')
-					mod=true
-					wordexp="${wordexp}${c}"
-					pgetc
-					case "${c}" in '#')
+					'%')
+						mod=true
 						wordexp="${wordexp}${c}"
 						pgetc
+						case "${c}" in '%')
+							wordexp="${wordexp}${c}"
+							pgetc
+							;;
+						esac
 						;;
-					esac
-					;;
-			esac
+					'#')
+						mod=true
+						wordexp="${wordexp}${c}"
+						pgetc
+						case "${c}" in '#')
+							wordexp="${wordexp}${c}"
+							pgetc
+							;;
+						esac
+						;;
+				esac
+			fi
 			if ${mod}; then
 				# Get word.
 				res="$(scan_word true)"
-				lineno_offset=${res%%${RS}*}
+				ln_off=${res%%${RS}*}
 				res="${res#*${RS}}"
 				c="${res%%${RS}*}"
 				res="${res#*${RS}}"
 				word="${res%%${RS}*}"
 				# We must advance lineno because scan_word() was
 				# run in a subshell.
-				lineno=$((${lineno} + ${lineno_offset}))
+				lineno=$((${lineno} + ${ln_off}))
 				wordexp="${wordexp}${word}"
-				echo "param mod word: '$word'" >&2
+				dbg "param mod word: '$word'"
 			fi
 			# Check for right brace.
 			case "${c}" in
@@ -339,19 +369,28 @@ scan_wordexp()
 			esac
 			;;
 		'(')
+			# Arithmetic expansion or command substitution
 			;;
 		[@*#?$!A-Za-z0-9_-])
-			next_param
+			res="$(scan_param)"
+			ln_off=${res%%${RS}*}
+			res="${res#*${RS}}"
+			c="${res%%${RS}*}"
+			res="${res#*${RS}}"
+			param="${res%%${RS}*}"
+			lineno=$((${lineno} + ${ln_off}))
 			wordexp="\$${param}"
 			;;
 	esac
 
-	printf "%d${RS}%s${RS}%s" ${lineno_offset} "${c}" "${wordexp}"
-
+	printf "%d${RS}%s${RS}%s" ${ln_off} "${c}" "${wordexp}"
+	return 0
 }
 
-next_param()
+scan_param()
 {
+	local param=
+
 	param=''
 	case "${c}" in
 		[@*#?$!0-])
@@ -385,7 +424,13 @@ next_param()
 				pgetc
 			done
 			;;
+		*)
+			synerr 'Bad parameter name'
+			;;
 	esac
+
+	printf "%d${RS}%s${RS}%s" 0 "${c}" "${param}"
+	return 0
 }
 
 # Check the current token.  If it matches, add it to the syntax array.
@@ -394,7 +439,7 @@ accept()
 	local t="${1}"
 
 	if [ "x${tok%%${US}*}" = "x${t}" ]; then
-		echo "accept $t" >&2
+		dbg "accept $t"
 		tokens="${tokens}${tok}${RS}"
 		next
 		return 0
@@ -402,6 +447,18 @@ accept()
 	return 1
 }
 
+expect()
+{
+	local t="${1}"
+
+	if accept "${t}"; then
+		return 0
+	else
+		synexp "${t}"
+		return 1
+	fi
+}
+
 error()
 {
 	local fmt="${1}"
@@ -415,6 +472,10 @@ error()
 			printf "%s:%d: ${fmt}\n" "${fname}" ${lineno} "${@}" >&2
 			;;
 	esac
+
+	# The parser runs in a subshell, so this just returns up to the caller
+	# like an exception.
+	exit 1
 }
 
 init_lexer()
diff --git a/parsing/parse.sh b/parsing/parse.sh
index 2ac4215..d85a42e 100644
--- a/parsing/parse.sh
+++ b/parsing/parse.sh
@@ -3,6 +3,15 @@ LF="$(printf '\n.')";   LF="${LF%.}"
 RS="$(printf '\036.')"; RS="${RS%.}"
 US="$(printf '\037.')"; US="${US%.}"
 
+dbg()
+{
+	dbg=true
+	dbg=false
+	if ${dbg}; then
+		printf 'DEBUG: %s\n' "${@}" >&2
+	fi
+}
+
 . ./tokens.sh
 . ./lexer.sh
 
@@ -109,25 +118,43 @@ parse()
 	shift 1
 
 	init_lexer "${fn}"
+
+	# If this returns (does not exit), there are no errors.
 	while complete_command; do :; done
-	if :; then  # TODO: Test for EOF or errors
-		get_tokens
-		return 0
+
+	get_tokens
+
+	return 0
+}
+
+try()
+{
+	local tokens=
+	local t=
+
+	printf 'Trying script:\n'
+	printf '\t%s\n' "${@}"
+	if tokens="$(printf '%s\n' "${@}" | parse -)"; then
+		IFS="${RS}"
+		for t in ${tokens}; do
+			printf 'Token: %s\n' "$(tokname "${t}")"
+			case "${t%${US}*}" in T_WORD)
+				printf '       "%s"\n' "${t#T_WORD${US}}"
+				;;
+			esac
+		done
+		unset IFS
+	else
+		printf 'FAIL\n'
 	fi
-	return 1
+	printf '\n\n'
 }
 
-if tokens="$(printf '%s\n' '"foo bar" && $baz || qux' '${quux%uux } quuux' | \
-		parse -)"; then
-	IFS="${RS}"
-	for t in ${tokens}; do
-		printf 'Token: %s\n' "$(tokname "${t}")"
-		case "${t%${US}*}" in T_WORD)
-			printf '       "%s"\n' "${t#T_WORD${US}}"
-			;;
-		esac
-	done
-	unset IFS
-else
-	echo FAIL
-fi
+try '"foo bar" && $baz || qux' '${quux%uux  quuux'
+try '"foo bar" && $baz || qux' '${quux%uux } quuux'
+try 'foo ${bar}'
+try 'foo ${#bar}'
+try 'foo ${bar#baz}'
+try 'foo ${#bar#}'
+try 'foo ${^}'
+try 'foo `bar`'