Organize lexer functions

1 files changed, 64 insertions, 44 deletions

diff --git a/parsing/lexer.sh b/parsing/lexer.sh
index a3ab762..292da43 100644
--- a/parsing/lexer.sh
+++ b/parsing/lexer.sh
@@ -5,12 +5,64 @@ c=
 tok=
 tokens=
 
+#
+# Error handling (used by scanning and interface functions)
+#
+
+error()
+{
+	local fmt="${1}"
+	shift 1
+
+	case "${fname}" in
+		'-')
+			printf "stdin:%d: ${fmt}\n" ${lineno} "${@}" >&2
+			;;
+		*)
+			printf "%s:%d: ${fmt}\n" "${fname}" ${lineno} "${@}" >&2
+			;;
+	esac
+
+	# The parser and lexer run in a subshell, so this just returns up to the
+	# caller like an exception.
+	exit 1
+}
+
+synexp()
+{
+	local t="${1}"
+	shift 1
+
+	if [ "x${t}" = 'x' ]; then
+		synerr '%s unexpected' "$(tokname "${tok}")"
+	else
+		synerr '%s unexpected (expecting %s)' "$(tokname "${tok}")" \
+			"$(tokname "${t}")"
+	fi
+}
+
+synerr()
+{
+	local fmt="${1}"
+	shift 1
+
+	error "Syntax error: ${fmt}" "${@}"
+}
+
+#
+# Input reading
+#
+
 pgetc()
 {
 	c="$(dd bs=1 count=1 2>/dev/null; printf '.')"
 	c="${c%.}"
 }
 
+#
+# Token recognition
+#
+
 next()
 {
 	while :; do
@@ -177,6 +229,10 @@ next_word()
 	tok="T_WORD${US}${word}"
 }
 
+#
+# Token scanning
+#
+
 scan_word()
 {
 	local in_param="${1}"
@@ -266,7 +322,7 @@ scan_word()
 		pgetc
 	done
 
-	printf "%d${RS}%s${RS}%s" ${lines} "${c}" "${word}"
+	printf "%d${RS}%c${RS}%s" ${lines} "${c}" "${word}"
 }
 
 scan_wordexp()
@@ -428,7 +484,7 @@ scan_wordexp()
 			;;
 	esac
 
-	printf "%d${RS}%s${RS}%s" ${ln_off} "${c}" "${wordexp}"
+	printf "%d${RS}%c${RS}%s" ${ln_off} "${c}" "${wordexp}"
 	return 0
 }
 
@@ -474,10 +530,14 @@ scan_param()
 			;;
 	esac
 
-	printf "%d${RS}%s${RS}%s" 0 "${c}" "${param}"
+	printf "%d${RS}%c${RS}%s" 0 "${c}" "${param}"
 	return 0
 }
 
+#
+# Interface
+#
+
 # Check the current token.  If it matches, add it to the syntax array.
 accept()
 {
@@ -586,29 +646,10 @@ expect()
 		return 0
 	else
 		synexp "${t}"
-		return 1
 	fi
 }
 
-error()
-{
-	local fmt="${1}"
-	shift 1
-
-	case "${fname}" in
-		'-')
-			printf "stdin:%d: ${fmt}\n" ${lineno} "${@}" >&2
-			;;
-		*)
-			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
-}
-
+# Called by the lexer, not the parser
 run_sublexer()
 {
 	local fn="${1}"
@@ -668,24 +709,3 @@ run_lexer()
 
 	return 0
 }
-
-synexp()
-{
-	local t="${1}"
-	shift 1
-
-	if [ "x${t}" = 'x' ]; then
-		synerr '%s unexpected' "$(tokname "${tok}")"
-	else
-		synerr '%s unexpected (expecting %s)' "$(tokname "${tok}")" \
-			"$(tokname "${t}")"
-	fi
-}
-
-synerr()
-{
-	local fmt="${1}"
-	shift 1
-
-	error "Syntax error: ${fmt}" "${@}"
-}