Fix LALR infinite loop

include/Parsodus/lrtables/LALR1Itemset.h

@@ -13,7 +13,7 @@ public:
     LALR1Itemset(std::shared_ptr<Rule> start);
 
     bool canMerge(const LALR1Itemset& rhs) const;
-    void merge(const LALR1Itemset& rhs);
+    bool merge(const LALR1Itemset& rhs);
 
 private:
 };

include/Parsodus/lrtables/LR0ItemsetBase.h

@@ -21,7 +21,7 @@ public:
     Itemset succ(std::string sym) const;
     bool operator==(const Itemset& rhs) const;
     bool canMerge(const Itemset& rhs) const;
-    void merge(const Itemset& rhs);
+    bool merge(const Itemset& rhs);
     bool empty() const;
 
 protected:
@@ -89,8 +89,9 @@ bool LR0ItemsetBase<Itemset>::canMerge(const Itemset&) const {
 }
 
 template <typename Itemset>
-void LR0ItemsetBase<Itemset>::merge(const Itemset&) {
+bool LR0ItemsetBase<Itemset>::merge(const Itemset&) {
     //NO-OP
+    return false;
 }
 
 template <typename Itemset>

include/Parsodus/lrtables/LR1Itemset.h

@@ -13,7 +13,7 @@ public:
     LR1Itemset(std::shared_ptr<Rule> start);
 
     bool canMerge(const LR1Itemset& rhs) const;
-    void merge(const LR1Itemset& rhs);
+    bool merge(const LR1Itemset& rhs);
 
 private:
 };

include/Parsodus/lrtables/generator.h

@@ -24,7 +24,7 @@ namespace lr {
  * - Itemset succ(std::string) const; compute the successor of this set, over the given symbol
  * - bool operator==(const Itemset&); are these two Itemsets equal
  * - bool canMerge(const Itemset&) const; Can the given Itemset be merged into this one
- * - void merge(const Itemset&); Merge the given Itemset into this one
+ * - bool merge(const Itemset&); Merge the given Itemset into this one, return whether there was a change
  * - bool empty() const; is this Itemset empty (== not useful)
  * - std::set<std::size_t> getReduces(const Grammar&, std::string) const; get all Rule indices where a reduce should happen with given lookahead (not necessarily a set, but iterable)
  * - static bool needsFollow() const; does this type of Itemset need Follow sets to work, if so the first and follow unique_ptr's of the grammar passed will be initialized
@@ -99,8 +99,9 @@ LRTable Generator<Itemset>::generate() {
                 if (itemsets[idx] == s) {
                     break;
                 } else if (itemsets[idx].canMerge(s)) {
-                    itemsets[idx].merge(s);
+                    if (itemsets[idx].merge(s)) {
-                    q.emplace(idx, std::move(s));
+                        q.emplace(idx, std::move(s));
+                    }
                     break;
                 }
             }

parser.example.pds

@@ -1,8 +1,12 @@
-parser: lalr(1) 
+parser: SLR(1) 
 lexesis: lexer.lxs
 terminals:
-    A B
+    TERMINAL
 start: s
 grammar:
-    s -> x x;
+    s -> a s
-    x -> A x | B;
+           | b
+           ;
+
+    a → TERMINAL;
+    b -> a;

src/lrtables/LALR1Itemset.cpp

@@ -31,18 +31,25 @@ bool LALR1Itemset::canMerge(const LALR1Itemset& rhs) const {
     return true;
 }
 
-void LALR1Itemset::merge(const LALR1Itemset& rhs) {
+bool LALR1Itemset::merge(const LALR1Itemset& rhs) {
+    bool change = false;
     for (auto& toMerge : rhs.m_items) {
         bool found = false;
         for (auto& it : m_items) {
             if (it.rule == toMerge.rule && it.dotIdx == toMerge.dotIdx) {
                 found = true;
-                it.lookaheads.insert(toMerge.lookaheads.begin(), toMerge.lookaheads.end());
+                for (auto& la : toMerge.lookaheads) {
+                    if (!it.lookaheads.count(la)) {
+                        change = true;
+                        it.lookaheads.insert(la);
+                    }
+                }
                 break;
             }
         }
         assert(found);
     }
+    return change;
 }
 
 } /* lr  */ 

src/lrtables/LR1Itemset.cpp

@@ -13,8 +13,9 @@ bool LR1Itemset::canMerge(const LR1Itemset&) const {
     return false;
 }
 
-void LR1Itemset::merge(const LR1Itemset&) {
+bool LR1Itemset::merge(const LR1Itemset&) {
     //NO-OP
+    return false;
 }
 
 } /* lr */