1 files changed, 40 insertions, 18 deletions

diff --git a/libparser-generate.c b/libparser-generate.c
index a8761f8..0a91835 100644
--- a/libparser-generate.c
+++ b/libparser-generate.c
@@ -38,6 +38,7 @@ struct node {
 	struct node *next; /* next element in list */
 	struct node *data; /* beginning of subsentence */
 	struct node **head; /* end of subsentence */
+	int nondeterministic;
 };
 
 
@@ -65,15 +66,6 @@ emalloc(size_t n)
 }
 
 static void *
-ecalloc(size_t n, size_t m)
-{
-	void *ret = calloc(n, m);
-	if (!ret)
-		eprintf("%s: calloc %zu %zu: %s\n", argv0, n, m, strerror(errno));
-	return ret;
-}
-
-static void *
 erealloc(void *ptr, size_t n)
 {
 	void *ret = realloc(ptr, n);
@@ -104,6 +96,20 @@ estrdup(char *s)
 }
 
 
+static struct node *
+new_node(void)
+{
+	struct node *ret = emalloc(sizeof(*ret));
+	ret->token = NULL;
+	ret->parent = NULL;
+	ret->next = NULL;
+	ret->data = NULL;
+	ret->head = NULL;
+	ret->nondeterministic = 0;
+	return ret;
+}
+
+
 static int
 strpcmp(const void *av, const void *bv)
 {
@@ -120,6 +126,7 @@ isidentifier(char c)
 }
 
 
+
 static int
 check_utf8(char *buf, size_t *ip, size_t len)
 {
@@ -379,10 +386,11 @@ emit_and_free_sentence(struct node *node, size_t *indexp)
 	case '!': type = "REJECTION"; unary:
 		emit_and_free_sentence(node->data, indexp);
 		printf("static union libparser_sentence sentence_%zu_%zu = {.unary = {"
-		           ".type = LIBPARSER_SENTENCE_TYPE_%s, .sentence = &sentence_%zu_%zu"
+		           ".type = LIBPARSER_SENTENCE_TYPE_%s%s, .sentence = &sentence_%zu_%zu"
 		       "}};\n",
 		       nrule_names, index,
-		       type, nrule_names, index + 1u);
+		       node->nondeterministic ? "ND_" : "", type,
+		       nrule_names, index + 1u);
 		break;
 
 	case '<':
@@ -412,11 +420,11 @@ emit_and_free_sentence(struct node *node, size_t *indexp)
 		left = *indexp;
 		emit_and_free_sentence(node->data, indexp);
 		printf("static union libparser_sentence sentence_%zu_%zu = {.binary = {"
-		           ".type = LIBPARSER_SENTENCE_TYPE_%s, "
+		           ".type = LIBPARSER_SENTENCE_TYPE_%s%s, "
 		           ".left = &sentence_%zu_%zu, .right = &sentence_%zu_%zu"
 		       "}};\n",
 		       nrule_names, index,
-		       type,
+		       node->nondeterministic ? "ND_" : "", type,
 		       nrule_names, left, nrule_names, right);
 		break;
 
@@ -587,7 +595,7 @@ main(int argc, char *argv[])
 	struct node *stack = NULL, *parent_node, *node;
 	char *data;
 	struct token **tokens;
-	size_t i, j;
+	size_t i, j, nondeterministic;
 	int cmp, err, val;
 
 	if (argc) {
@@ -613,6 +621,7 @@ main(int argc, char *argv[])
 
 	printf("#include <libparser.h>\n");
 
+	nondeterministic = SIZE_MAX;
 	i = 0;
 again:
 	for (; tokens[i]; i++) {
@@ -635,7 +644,7 @@ again:
 		}
 
 		/* Also remove any whitespace (the tokeniser
-		 * simple and does not recognise mulltisymbol
+		 * simple and does not recognise multisymbol
 		 * tokens (that is apart form strings and
 		 * identifiers) so it cannot ignore whitespace. */
 		if (isspace(tokens[i]->s[0])) {
@@ -643,6 +652,16 @@ again:
 			continue;
 		}
 
+		/* Check if next token is non-deterministic */
+		if (tokens[i]->s[0] == '?') {
+			if (tokens[i + 1u]->s[0] == '[' ||
+			    tokens[i + 1u]->s[0] == '{' ||
+			    tokens[i + 1u]->s[0] == '|') {
+				nondeterministic = i + 1u;
+				continue;
+			}
+		}
+
 		/* For the sake of code readability, identify
 		 * the token type */
 		if (tokens[i]->s[0] == '"') {
@@ -661,7 +680,8 @@ again:
 				eprintf("%s: expected an identifier on line %zu at column %zu (character %zu)\n",
 				        argv0, tokens[i]->lineno, tokens[i]->column, tokens[i]->character);
 			}
-			stack = calloc(1, sizeof(*stack));
+			stack = new_node();
+			stack->nondeterministic = i == nondeterministic;
 			stack->token = tokens[i];
 			stack->head = &stack->data;
 			/* and then we expect an equals sign */
@@ -715,7 +735,8 @@ again:
 					state = EXPECT_RANGE_LOW;
 				push_stack:
 					parent_node = stack;
-					stack = ecalloc(1, sizeof(*stack));
+					stack = new_node();
+					stack->nondeterministic = i == nondeterministic;
 					stack->parent = parent_node;
 					stack->token = tokens[i];
 					stack->head = &stack->data;
@@ -756,7 +777,8 @@ again:
 				 * token to be an operand */
 				state = EXPECT_OPERAND;
 			add_singleton:
-				node = calloc(1u, sizeof(*node));
+				node = new_node();
+				node->nondeterministic = i == nondeterministic;
 				node->token = tokens[i];
 				*stack->head = node;
 				stack->head = &node->next;