From b0a6fb6b2414dde90f7bf10e2e30afab78bf567a Mon Sep 17 00:00:00 2001
From: Charles Baker <charles.baker@sweetsoftware.co.nz>
Date: Fri, 2 Jun 2023 21:33:29 +1200
Subject: [PATCH] Track and iterate only over set range in GrammarSymbolSet

FIX: Iterate over all elements when combining sets
---
 src/lalr/GrammarGenerator.cpp |  4 +++-
 src/lalr/GrammarSymbolSet.cpp | 38 ++++++++++++++++++++++++++++-------
 src/lalr/GrammarSymbolSet.hpp |  5 ++++-
 3 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/src/lalr/GrammarGenerator.cpp b/src/lalr/GrammarGenerator.cpp
index 9e64370..b730ade 100644
--- a/src/lalr/GrammarGenerator.cpp
+++ b/src/lalr/GrammarGenerator.cpp
@@ -956,7 +956,9 @@ void GrammarGenerator::generate_reduce_transitions()
             if ( item->position() >= production->length() )
             {
                 const GrammarSymbolSet& lookaheads = lookaheads_[item->index()].lookaheads();
-                for ( int index = 0; index < int(symbols_.size()); ++index )
+                int start = lookaheads.minimum_index();
+                int finish = lookaheads.maximum_index();
+                for ( int index = start; index < finish; ++index )
                 {
                     if ( lookaheads.contains(index) )
                     {
diff --git a/src/lalr/GrammarSymbolSet.cpp b/src/lalr/GrammarSymbolSet.cpp
index 90e83bf..d4a4cc3 100644
--- a/src/lalr/GrammarSymbolSet.cpp
+++ b/src/lalr/GrammarSymbolSet.cpp
@@ -6,8 +6,13 @@
 #include "GrammarSymbolSet.hpp"
 #include "GrammarSymbol.hpp"
 #include "assert.hpp"
+#include <algorithm>
+#include <limits>
 
+using std::min;
+using std::max;
 using std::vector;
+using std::numeric_limits;
 using namespace lalr;
 
 static const size_t ONE = 1;
@@ -15,17 +20,23 @@ static const size_t BITS_PER_ELEMENT = sizeof(size_t) * 8;
 
 GrammarSymbolSet::GrammarSymbolSet( size_t symbols )
 : set_()
+, minimum_(numeric_limits<size_t>::max())
+, maximum_(numeric_limits<size_t>::min())
 {
     set_.resize( symbols / BITS_PER_ELEMENT );
 }
 
 GrammarSymbolSet::GrammarSymbolSet( GrammarSymbolSet&& set )
 : set_( std::move(set.set_) )
+, minimum_(numeric_limits<size_t>::max())
+, maximum_(numeric_limits<size_t>::min())
 {
 }
 
 GrammarSymbolSet::GrammarSymbolSet( const GrammarSymbolSet& set )
 : set_( set.set_ )
+, minimum_(numeric_limits<size_t>::max())
+, maximum_(numeric_limits<size_t>::min())
 {
 }
 
@@ -34,6 +45,8 @@ GrammarSymbolSet& GrammarSymbolSet::operator=( GrammarSymbolSet&& set )
     if ( this != &set )
     {
         std::swap( set_, set.set_ );
+        std::swap( minimum_, set.minimum_ );
+        std::swap( maximum_, set.maximum_ );
     }
     return *this;
 }
@@ -43,14 +56,26 @@ GrammarSymbolSet& GrammarSymbolSet::operator=( const GrammarSymbolSet& set )
     if ( this != &set )
     {
         set_ = set.set_;
+        minimum_ = set.minimum_;
+        maximum_ = set.maximum_;
     }
     return *this;    
 }
 
+int GrammarSymbolSet::minimum_index() const
+{
+    return int(minimum_ * BITS_PER_ELEMENT);
+}
+
+int GrammarSymbolSet::maximum_index() const
+{
+    return int((maximum_ + 1) * BITS_PER_ELEMENT);
+}
+
 bool GrammarSymbolSet::contains( int symbol_index ) const
 {
     size_t index = symbol_index / BITS_PER_ELEMENT;
-    if ( index < set_.size() )
+    if ( index >= minimum_ && index <= maximum_ )
     {
         size_t mask = ONE << (symbol_index % BITS_PER_ELEMENT);
         return (set_[index] & mask) != 0;
@@ -71,6 +96,8 @@ bool GrammarSymbolSet::insert( const GrammarSymbol* symbol )
         if ( !(set_[index] & mask) )
         {
             set_[index] |= mask;
+            minimum_ = min( minimum_, index );
+            maximum_ = max( maximum_, index );
             return true;
         }
     }
@@ -85,7 +112,7 @@ int GrammarSymbolSet::insert( const GrammarSymbolSet& set )
     }
 
     int added = 0;
-    for ( size_t i = 0; i < set.set_.size(); ++i )
+    for ( size_t i = set.minimum_; i < set.maximum_ + 1; ++i )
     {
         size_t mask = set_[i];
         size_t new_mask = mask | set.set_[i];
@@ -95,10 +122,7 @@ int GrammarSymbolSet::insert( const GrammarSymbolSet& set )
             ++added;
         }
     }
+    minimum_ = min( minimum_, set.minimum_ );
+    maximum_ = min( maximum_, set.maximum_ );
     return added;
 }
-
-void GrammarSymbolSet::swap( GrammarSymbolSet&& set )
-{
-    std::swap( set_, set.set_ );
-}
diff --git a/src/lalr/GrammarSymbolSet.hpp b/src/lalr/GrammarSymbolSet.hpp
index 098dc0d..03f2b48 100644
--- a/src/lalr/GrammarSymbolSet.hpp
+++ b/src/lalr/GrammarSymbolSet.hpp
@@ -12,6 +12,8 @@ class GrammarSymbol;
 class GrammarSymbolSet
 {
     std::vector<size_t> set_;
+    size_t minimum_;
+    size_t maximum_;
 
 public:
     GrammarSymbolSet( size_t symbols );
@@ -20,10 +22,11 @@ public:
     GrammarSymbolSet& operator=( GrammarSymbolSet&& set );
     GrammarSymbolSet& operator=( const GrammarSymbolSet& set );
 
+    int minimum_index() const;
+    int maximum_index() const;
     bool contains( int symbol_index ) const;
     bool insert( const GrammarSymbol* symbol );
     int insert( const GrammarSymbolSet& set );
-    void swap( GrammarSymbolSet&& set );
 };
 
 }