Move setup of mergeable_symbols to constructor, use set throughout

spec/compiler/build_tables/distinctive_tokens_spec.cc

@@ -27,7 +27,7 @@ describe("recovery_tokens(rule)", []() {
       })),
     };
 
-    AssertThat(recovery_tokens(grammar), Equals<vector<Symbol>>({
+    AssertThat(recovery_tokens(grammar), Equals<set<Symbol>>({
       Symbol(1, true),
     }));
   });

src/compiler/build_tables/build_parse_table.cc

@@ -75,6 +75,8 @@ class ParseTableBuilder {
         for (const auto &pair2 : state.entries)
           parse_table.symbols[pair1.first].compatible_symbols.insert(pair2.first);
 
+    parse_table.mergeable_symbols = recovery_tokens(lexical_grammar);
+
     build_error_parse_state();
 
     allow_any_conflict = true;
@@ -112,11 +114,7 @@ class ParseTableBuilder {
   void build_error_parse_state() {
     ParseState error_state;
 
-    auto recovery_symbols = recovery_tokens(lexical_grammar);
-
-    parse_table.mergeable_symbols.insert(recovery_symbols.begin(), recovery_symbols.end());
-
-    for (const Symbol &symbol : recovery_symbols)
+    for (auto &symbol : parse_table.mergeable_symbols)
       add_out_of_context_parse_state(&error_state, symbol);
 
     for (const Symbol &symbol : grammar.extra_tokens)

src/compiler/build_tables/recovery_tokens.cc

@@ -11,7 +11,7 @@ namespace tree_sitter {
 namespace build_tables {
 
 using rules::Symbol;
-using std::vector;
+using std::set;
 
 template <bool left, bool right>
 class CharacterAggregator : public rules::RuleFn<void> {
@@ -47,8 +47,8 @@ class FirstCharacters : public CharacterAggregator<true, false> {};
 class LastCharacters : public CharacterAggregator<false, true> {};
 class AllCharacters : public CharacterAggregator<true, true> {};
 
-vector<Symbol> recovery_tokens(const LexicalGrammar &grammar) {
-  vector<Symbol> result;
+set<Symbol> recovery_tokens(const LexicalGrammar &grammar) {
+  set<Symbol> result;
 
   AllCharacters all_separator_characters;
   for (const rule_ptr &separator : grammar.separators)
@@ -79,7 +79,7 @@ vector<Symbol> recovery_tokens(const LexicalGrammar &grammar) {
       !all_characters.result.intersects(all_separator_characters.result);
 
     if ((has_distinct_start && has_distinct_end) || has_no_separators)
-      result.push_back(Symbol(i, true));
+      result.insert(Symbol(i, true));
   }
 
   return result;

src/compiler/build_tables/recovery_tokens.h

@@ -3,7 +3,7 @@
 
 #include "compiler/rule.h"
 #include "compiler/rules/symbol.h"
-#include <vector>
+#include <set>
 
 namespace tree_sitter {
 
@@ -11,7 +11,7 @@ struct LexicalGrammar;
 
 namespace build_tables {
 
-std::vector<rules::Symbol> recovery_tokens(const LexicalGrammar &);
+std::set<rules::Symbol> recovery_tokens(const LexicalGrammar &);
 
 }  // namespace build_tables
 }  // namespace tree_sitter