Massive parser clamp-down

6 files changed, 175 insertions, 47 deletions

diff --git a/Makefile b/Makefile
index 3fdfcd2..8c7b6eb 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,4 @@
-CFLAGS += -std=c99 -D_XOPEN_SOURCE=500 -Wall -Wextra
+CFLAGS += -std=c99 -D_XOPEN_SOURCE=500 -D_POSIX_C_SOURCE=200809L -Wall -Wextra
 
 all: simulator
 
diff --git a/gate.c b/gate.c
index c196b91..f745143 100644
--- a/gate.c
+++ b/gate.c
@@ -151,6 +151,7 @@ void gate_update(void) {
 
 	for (i = 0; i < gate_count; i++) {
 		g = &gates[i];
+		emit_error("Name: '%s'\n", g->name);
 		in1 = g->in1->output;
 		in2 = g->in2->output;
 		g->output = (g->operation)(in1, in2);
diff --git a/parser.c b/parser.c
index 822441c..b756eae 100644
--- a/parser.c
+++ b/parser.c
@@ -1,5 +1,6 @@
 #include <ctype.h>
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 
 #include "common.h"
@@ -41,16 +42,21 @@ static struct tok_lookup tok_handlers[] = {
 };
 
 int
-expect(const char *expect, char *actual) {
-	int min_len = 0;
+rtrim(char *string) {
+	char *old_end = string + strlen(string);
+	char *w = NULL;
 
-	min_len = MIN(strlen(expect), strlen(actual));
+	string[strcspn(string, "\r\n")] = '\0';
 
-	if (strncmp(expect, actual, min_len) != 0) {
-		emit_error("Expected '%s' at start of '%s'\n", expect, actual);
-		return 1;
+	/* find first member of right whitespace */
+	for (w = old_end; w >= string; w--) {
+		if (strchr("\t ", *w) == NULL) {
+			break;
+		}
+		*w = '\0';
 	}
-	return 0;
+
+	return old_end - w;
 }
 
 char *
@@ -61,131 +67,238 @@ eat_whitespace(char *string) {
 	return string;
 }
 
+char *
+get_token_special(char *string, char **rest, char *delims) {
+	size_t tok_len = 0;
+	char *ret = NULL;
+	string = eat_whitespace(string);
+	tok_len = strcspn(string, delims);
+	if (tok_len == 0) {
+		emit_error("unexpected end of string");
+		return NULL;
+	}
+	if (NULL == (ret = strndup(string, tok_len))) {
+		emit_error("get_token: ");
+		perror("strndup");
+		return NULL;
+	}
+	*rest = string + tok_len;
+	return ret;
+}
+
+char *
+get_token(char *string, char **rest) {
+	return get_token_special(string, rest, " \t");
+}
+
+int
+expect(const char *expect, char *actual) {
+	int min_len = 0;
+
+	min_len = MIN(strlen(expect), strlen(actual));
+
+	if (strncmp(expect, actual, min_len) != 0) {
+		emit_error("Expected '%s' at start of '%s'\n", expect, actual);
+		return 1;
+	}
+	return 0;
+}
+
 int parse_uop(char *str, char *name, enum BINARY (*handler)(enum BINARY, enum BINARY)) {
-	char *gate_name = str;
+	char *source_name = NULL;
+	char *next = NULL;
 
-	strtok(gate_name, " ");
+	if (NULL == (source_name = get_token(str, &next))) {
+		return 1;
+	}
 
 	/* FIXME allow input from other gates, not just inputs */
-	struct gate *in = gate_get_input_by_name(gate_name);
+	struct gate *in = gate_get_input_by_name(source_name);
+	free(source_name);
 
-	gate_add(name, handler, in, in);
-	return 0;
+	return gate_add(name, handler, in, in);
 
 
 }
 
 int parse_bop(char *str, char *name, enum BINARY (*handler)(enum BINARY, enum BINARY)) {
-	char *gate_name1 = str;
-	char *gate_name2 = NULL;
+	char *source_name1 = NULL;
+	char *source_name2 = NULL;
+	char *next = NULL;
+	struct gate *in1 = NULL;
+	struct gate *in2 = NULL;
+
+	if (NULL == (source_name1 = get_token(str, &next))) {
+		return 1;
+	}
 
-	strtok(gate_name1, " ");
-	gate_name2 = gate_name1 + strlen(gate_name1) + 1;
-	strtok(gate_name2, " ");
+	if (NULL == (source_name2 = get_token(next, &next))) {
+		free(source_name1);
+		return 1;
+	}
+
+	if (*next != '\0') {
+		emit_error("superfluous text \"%s\"\n", next);
+		free(source_name1);
+		free(source_name2);
+		return 1;
+	}
 
 	/* FIXME allow input from other gates, not just inputs */
-	struct gate *in1 = gate_get_input_by_name(gate_name1);
-	struct gate *in2 = gate_get_input_by_name(gate_name2);
+	in1 = gate_get_input_by_name(source_name1);
+	in2 = gate_get_input_by_name(source_name2);
 
-	gate_add(name, handler, in1, in2);
-	return 0;
+	free(source_name1);
+	free(source_name2);
+
+	if (in1 == NULL || in2 == NULL) {
+		emit_error("Undeclared identifier in source(s) for expression '%s'\n",
+		           name);
+		return 1;
+	}
+
+	return gate_add(name, handler, in1, in2);
 }
 
 int
 parse_op(char *str, char *name) {
+	char *tok = NULL;
+	char *next = NULL;
 	int match = 0;
 	size_t i = 0;
 
-	strtok(str, " ");
+	if (NULL == (tok = get_token(str, &next))) {
+		return 1;
+	}
+
 	match = 0;
 	for (i = 0; i < sizeof(bop_handlers)/sizeof(bop_handlers[0]); i++) {
-		if (strcmp(bop_handlers[i].str, str) == 0) {
+		if (strcmp(bop_handlers[i].str, tok) == 0) {
 			match = 1;
-			if (parse_bop(str+strlen(str)+1, name, bop_handlers[i].handler)) {
+			if (parse_bop(next, name, bop_handlers[i].handler)) {
+				free(tok);
 				return 1;
 			}
 		}
 	}
 	for (i = 0; i < sizeof(uop_handlers)/sizeof(uop_handlers[0]); i++) {
-		if (strcmp(uop_handlers[i].str, str) == 0) {
+		if (strcmp(uop_handlers[i].str, tok) == 0) {
 			match = 1;
-			if (parse_uop(str+strlen(str)+1, name, uop_handlers[i].handler)) {
+			if (parse_uop(next, name, uop_handlers[i].handler)) {
+				free(tok);
 				return 1;
 			}
 		}
 	}
 	if (match == 0) {
-		emit_error("Invalid operator \"%s\"\n", str);
+		emit_error("Invalid operator \"%s\"\n", tok);
+		free(tok);
 		return 1;
 	}
+	free(tok);
 	return 0;
 }
 
 int
 parse_expr(char *str) {
 	const char *expr_sep = ": ";
-	size_t ident_len = 0;
 	char *ident = NULL;
+	char *next = NULL;
 	char *op = NULL;
 
-	/* FIXME magic constant string */
-	ident_len = strspn(str, "abcdefghijklmnopqrstuvwxyz0123456789");
+	if (NULL == (ident = get_token_special(str, &next, ": \t"))) {
+		return 1;
+	}
 
-	if (expect(expr_sep, str + ident_len)) {
-		return -1;
+	if (expect(expr_sep, next)) {
+		free(ident);
+		return 1;
 	}
 
-	str[ident_len] = '\0';
-	ident = str;
-	op = str + ident_len + strlen(expr_sep);
+	op = next + strlen(expr_sep);
 
 	return parse_op(op, ident);
 }
 
 int
 parse_input(char *str) {
-	strtok(str, " ");
-	str = eat_whitespace(str);
-	if (strlen(str) == 0) {
+	char *ident = NULL;
+	char *next = NULL;
+	int ret = 0;
+
+	if (NULL == (ident = get_token(str, &next))) {
+		return 1;
+	}
+
+	if (strlen(ident) == 0) {
 		emit_error("input label must not be empty");
+		free(ident);
 		return 1;
 	}
-	emit_info("Add input '%s'\n", str);
-	return gate_input_add(str);
+
+	if (*next != '\0') {
+		emit_error("superfluous text after \"%s\" identifier: \"%s\"\n",
+		           ident, next);
+		free(ident);
+		return 1;
+	}
+
+	emit_info("Add input '%s'\n", ident);
+	ret = gate_input_add(ident);
+	free(ident);
+	return ret;
 }
 
 int
 parse_module(char *str) {
-	printf("FIXME module name is \"%s\" but modules are not implemented\n", str);
+	char *mod_name = NULL;
+	char *next = NULL;
+	if (NULL == (mod_name = get_token(str, &next))) {
+		return 1;
+	}
+
+	if (*next != '\0') {
+		emit_error("superfluous text following module name");
+		free(mod_name);
+		return 1;
+	}
+
+	printf("FIXME module name is \"%s\" but modules are not implemented\n", mod_name);
+	free(mod_name);
 	return 0;
 }
 
 int
 parse_line(char *line) {
 	size_t i = 0;
-	char *tok = line;
-	char *delim = " ";
+	char *next = NULL;
+	char *tok = NULL;
 	int match = 0;
 
-	line[strcspn(line, "\r\n")] = '\0';
+	rtrim(line);
 
 	if (   strlen(line) == 0
 	    || (strlen(line) == 1 && isspace(line[0]))) {
 		return 0;
 	}
 
-	tok = strtok(line, delim);
+	if (NULL == (tok = get_token(line, &next))) {
+		return 1;
+	}
 
 	match = 0;
 	for (i = 0; i < sizeof(tok_handlers)/sizeof(tok_handlers[0]); i++) {
 		if (strcmp(tok_handlers[i].str, tok) == 0) {
 			match = 1;
-			if ((tok_handlers[i].handler)(&tok[strlen(tok)+1])) {
+			if ((tok_handlers[i].handler)(next)) {
+				free(tok);
 				return 1;
 			}
 		}
 	}
 
+	free(tok);
+
 	if (match == 0) {
 		emit_error("invalid token \"%s\"", tok);
 		return 1;
diff --git a/test/test-duplicate-input b/test/test-duplicate-input
index 236c6fc..141d6aa 100755
--- a/test/test-duplicate-input
+++ b/test/test-duplicate-input
diff --git a/test/test-short-keyword b/test/test-short-keyword
index 39f58dc..17b4858 100755
--- a/test/test-short-keyword
+++ b/test/test-short-keyword
diff --git a/test/test-short-keyword.c b/test/test-short-keyword.c
index 50d44d3..226c097 100644
--- a/test/test-short-keyword.c
+++ b/test/test-short-keyword.c
@@ -7,13 +7,25 @@
 #include "parser.h"
 
 char *failures[] = {
+	"input"
 	"input\n",
 	"input \n",
+	"input aa aa\n",
+	"module",
+	" module ",
+	"\tmodule     ",
+	"  moudle  \t\n ",
 };
 
 char *passes[] = {
+	"input aa",
 	"input a\n",
 	"input aaaaaaaa\n",
+	"module test",
+	"module   test",
+	"module   test   ",
+	"module test    ",
+	" \tmodule \ttest  \t",
 	"\n",
 	"\n\n",
 	"\r",
@@ -22,6 +34,8 @@ char *passes[] = {
 	"\n\r",
 	"\n\n\r",
 	"\r\n\r",
+	"",
+	"\t"
 };
 
 int main(void) {