fix conflict resolution with charset productions

src/backends/lalr.c

@@ -208,6 +208,46 @@ static bool match_production(HLREnhGrammar *eg, HCFChoice **p,
           && state == endstate);
 }
 
+// variant of match_production where the production lhs is a charset
+// [..x..] -> x
+static bool match_charset_production(const HLRTable *table, HLREnhGrammar *eg,
+                                     const HCFChoice *lhs, HCFChoice *rhs,
+                                     size_t endstate)
+{
+  assert(lhs->type == HCF_CHARSET);
+  assert(rhs->type == HCF_CHAR);
+
+  if(!charset_isset(lhs->charset, rhs->chr))
+    return false;
+
+  // determine the enhanced-grammar right-hand side and check end state
+  HLRTransition *t = h_hashtable_get(eg->smap, lhs);
+  assert(t != NULL);
+  return (follow_transition(table, t->from, rhs) == endstate);
+}
+
+// check wether any production for sym (enhanced-grammar) matches the given
+// (original-grammar) rhs and terminates in the given end state.
+static bool match_any_production(const HLRTable *table, HLREnhGrammar *eg,
+                                 const HCFChoice *sym, HCFChoice **rhs,
+                                 size_t endstate)
+{
+  assert(sym->type == HCF_CHOICE || sym->type == HCF_CHARSET);
+
+  if(sym->type == HCF_CHOICE) {
+    for(HCFSequence **p=sym->seq; *p; p++) {
+      if(match_production(eg, (*p)->items, rhs, endstate))
+        return true;
+    }
+  } else {  // HCF_CHARSET
+    assert(rhs[0] != NULL);
+    assert(rhs[1] == NULL);
+    return match_charset_production(table, eg, sym, rhs[0], endstate);
+  }
+
+  return false;
+}
+
 // desugar parser with a fresh start symbol
 // this guarantees that the start symbol will not occur in any productions
 HCFChoice *h_desugar_augmented(HAllocator *mm__, HParser *parser)
@@ -286,28 +326,7 @@ int h_lalr_compile(HAllocator* mm__, HParser* parser, const void* params)
         HHashSet *lhss = h_hashtable_get(eg->corr, item->lhs);
         assert(lhss != NULL);
         H_FOREACH_KEY(lhss, HCFChoice *lhs)
-          assert(lhs->type == HCF_CHOICE || lhs->type == HCF_CHARSET);
-
-          bool match = false;
-          if(lhs->type == HCF_CHOICE) {
-            for(HCFSequence **p=lhs->seq; *p; p++) {
-              HCFChoice **rhs = (*p)->items;
-              if(match_production(eg, rhs, item->rhs, state)) {
-                match = true;
-                break;
-              }
-            }
-          } else {  // HCF_CHARSET
-            assert(item->rhs[0] != NULL);
-            assert(item->rhs[1] == NULL);
-            assert(item->rhs[0]->type == HCF_CHAR);
-            HLRTransition *t = h_hashtable_get(eg->smap, lhs);
-            assert(t != NULL);
-            match = (t->to == state
-                     && charset_isset(lhs->charset, item->rhs[0]->chr));
-          }
-
-          if(match) {
+          if(match_any_production(table, eg, lhs, item->rhs, state)) {
             // the left-hand symbol's follow set is this production's
             // contribution to the lookahead
             const HStringMap *fs = h_follow(1, eg->grammar, lhs);

src/t_regression.c

@@ -121,13 +121,24 @@ static void test_llk_zero_end(void) {
 static void test_lalr_charset_lhs(void) {
     HParserBackend be = PB_LALR;
 
-    HParser *p = h_choice(h_ch('A'), h_uint8(), NULL);
+    HParser *p = h_many(h_choice(h_sequence(h_ch('A'), h_ch('B'), NULL),
+                                 h_in((uint8_t*)"AB",2), NULL));
 
-    // the above would fail to compile because of an unhandled case in trying
-    // to resolve a conflict where an item's left-hand-side was an HCF_CHARSET.
+    // the above would abort because of an unhandled case in trying to resolve
+    // a conflict where an item's left-hand-side was an HCF_CHARSET.
+    // however, the compile should fail - the conflict cannot be resolved.
 
-    g_check_parse_match(p, be, "A",1, "u0x41");
-    g_check_parse_match(p, be, "B",1, "u0x42");
+    if(h_compile(p, be, NULL) == 0) {
+        g_test_message("LALR compile didn't detect ambiguous grammar");
+
+        // it says it compiled it - well, then it should parse it!
+        // (this helps us see what it thinks it should be doing.)
+        g_check_parse_match(p, be, "AA",2, "(u0x41 u0x41)");
+        g_check_parse_match(p, be, "AB",2, "((u0x41 u0x42))");
+
+        g_test_fail();
+        return;
+    }
 }
 
 static void test_cfg_many_seq(void) {