grammars: more test cases

common/grammar-parser.cpp

@@ -172,9 +172,10 @@ namespace grammar_parser {
             // S{m,} -->  S'     ::= Scopy Scopy Scopy (m times) Sstar
             //            Scopy  ::= S
             //            Sstar  ::= Scopy Sstar |
+            // And if S is a reference to a rule, then we skip the S_copy indirection
 
             uint32_t content_rule_id = 0;
-            if (last_sym_start >= 0 && last_sym_start == out_elements.size() - 1 && out_elements[last_sym_start].type == LLAMA_GRETYPE_RULE_REF) {
+            if (last_sym_start == out_elements.size() - 1 && out_elements[last_sym_start].type == LLAMA_GRETYPE_RULE_REF) {
                 // The repeated content is already a rule ref, no need to copy it
                 content_rule_id = out_elements[last_sym_start].value;
             } else {

tests/test-grammar-parser.cpp

@@ -173,6 +173,32 @@ int main()
         {LLAMA_GRETYPE_END, 0},
     });
 
+    verify_parsing(R"""(
+        root  ::= a+
+        a     ::= "a"
+    )""", {
+        {"a", 1},
+        {"root", 0},
+        {"root_2", 2},
+        {"root_star_3", 3},
+    }, {
+        // root (index 0)
+        {LLAMA_GRETYPE_RULE_REF, /* root_2 */ 2},
+        {LLAMA_GRETYPE_END, 0},
+        // a (index 1)
+        {LLAMA_GRETYPE_CHAR, 'a'},
+        {LLAMA_GRETYPE_END, 0},
+        // root_2 (index 2)
+        {LLAMA_GRETYPE_RULE_REF, /* a */ 1},
+        {LLAMA_GRETYPE_RULE_REF, /* root_star_3 */ 3},
+        {LLAMA_GRETYPE_END, 0},
+        // root_star_3 (index 3)
+        {LLAMA_GRETYPE_RULE_REF, /* a */ 1},
+        {LLAMA_GRETYPE_RULE_REF, /* root_star_3 */ 3},
+        {LLAMA_GRETYPE_ALT, 0},
+        {LLAMA_GRETYPE_END, 0},
+    });
+
     verify_parsing(R"""(
         root  ::= "a"+
     )""", {
@@ -198,6 +224,30 @@ int main()
         {LLAMA_GRETYPE_END, 0},
     });
 
+    verify_parsing(R"""(
+        root  ::= a?
+        a     ::= "a"
+    )""", {
+        {"a", 1},
+        {"root", 0},
+        {"root_1_3", 3},
+        {"root_2", 2},
+    }, {
+        // root (index 0)
+        {LLAMA_GRETYPE_RULE_REF, /* root_2 */ 2},
+        {LLAMA_GRETYPE_END, 0},
+        // a (index 1)
+        {LLAMA_GRETYPE_CHAR, 'a'},
+        {LLAMA_GRETYPE_END, 0},
+        // root_2 (index 2)
+        {LLAMA_GRETYPE_RULE_REF, /* root_1_3 */ 3},
+        {LLAMA_GRETYPE_END, 0},
+        // root_1_3 (index 3)
+        {LLAMA_GRETYPE_RULE_REF, /* a */ 1},
+        {LLAMA_GRETYPE_ALT, 0},
+        {LLAMA_GRETYPE_END, 0},
+    });
+
     verify_parsing(R"""(
         root  ::= "a"?
     )""", {
@@ -221,6 +271,31 @@ int main()
         {LLAMA_GRETYPE_END, 0},
     });
 
+    verify_parsing(R"""(
+        root  ::= a*
+        a     ::= "a"
+    )""", {
+        {"a", 1},
+        {"root", 0},
+        {"root_2", 2},
+        {"root_star_3", 3},
+    }, {
+        // root (index 0)
+        {LLAMA_GRETYPE_RULE_REF, /* root_2 */ 2},
+        {LLAMA_GRETYPE_END, 0},
+        // a (index 1)
+        {LLAMA_GRETYPE_CHAR, 'a'},
+        {LLAMA_GRETYPE_END, 0},
+        // root_2 (index 2)
+        {LLAMA_GRETYPE_RULE_REF, /* root_star_3 */ 3},
+        {LLAMA_GRETYPE_END, 0},
+        // root_star_3 (index 3)
+        {LLAMA_GRETYPE_RULE_REF, /* a */ 1},
+        {LLAMA_GRETYPE_RULE_REF, /* root_star_3 */ 3},
+        {LLAMA_GRETYPE_ALT, 0},
+        {LLAMA_GRETYPE_END, 0},
+    });
+
     verify_parsing(R"""(
         root  ::= "a"*
     )""", {