From 8a59d19e5c402256db416d9847906286f8b0fa8f Mon Sep 17 00:00:00 2001
From: Robin Jadoul <robin.jadoul@gmail.com>
Date: Tue, 24 May 2016 18:09:38 +0200
Subject: [PATCH] Speed up mssc

---
 include/Lexesis/automata.h |  3 +++
 src/automata.cpp           | 28 ++++++++++++++++++++--------
 2 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/include/Lexesis/automata.h b/include/Lexesis/automata.h
index 67d4680..f54a36d 100644
--- a/include/Lexesis/automata.h
+++ b/include/Lexesis/automata.h
@@ -34,6 +34,9 @@ namespace lxs {
     struct ENFA : public NFA {
         std::map<State, std::set<State> > epsilonTransitions;
         virtual std::set<State> eClose(State) const;
+
+        private:
+            mutable std::map<State, std::set<State> > eCloseCache;
     };
 
     std::string toDot(const DFA& d);
diff --git a/src/automata.cpp b/src/automata.cpp
index c55ee61..3c1ef24 100644
--- a/src/automata.cpp
+++ b/src/automata.cpp
@@ -98,6 +98,10 @@ namespace lxs {
     }
 
     std::set<State> ENFA::eClose(State s) const {
+        auto cachedIt = eCloseCache.find(s);
+        if (cachedIt != eCloseCache.end()) {
+            return cachedIt->second;
+        }
         std::set<State> states;
         std::queue<State> statequeue;
         statequeue.push(s);
@@ -114,7 +118,7 @@ namespace lxs {
                 } 
             }
         } 
-        return states;
+        return eCloseCache[s] = states;
     }
 
     std::set<State> NFA::eClose(State s) const {
@@ -282,9 +286,9 @@ namespace lxs {
     
     namespace { // Utility function for mssc
         
-        std::set<State> getNextState(std::set<State> oldstate, char symbol, const NFA& e) {
+        std::set<State> getNextState(const std::set<State>& oldstate, char symbol, const NFA& e) {
             std::set<State> states;
-            for(auto &state: oldstate) {
+            for(const auto &state: oldstate) {
                  auto a = e.delta.find(state);
                  if(a != e.delta.end()) {
                      auto newStates = a->second.find(symbol);
@@ -306,26 +310,34 @@ namespace lxs {
     
     DFA mssc(const NFA& e) {
         //Temporary, improper dfa structure
-        std::map<std::set<State>,std::map<char,std::set<State> > > dfa;
+        std::map<std::set<State>, std::map<char,std::set<State> > > dfa;
         std::map<char, std::set<State> > trans;
         for (int c = 0; c < 256; c++) {
             trans[c] = {};
         }
         dfa[{}] = trans;
-        
+
+        std::set<char> usedChars;
+        for (const auto& stateTrans : e.delta) {
+            for (const auto& tr : stateTrans.second) {
+                usedChars.insert(tr.first);
+            }
+        }
+
         //Lazy evaluation, on a still implicit DFA
         std::queue<std::set<State> > tocheck;
         tocheck.push(e.eClose(e.starting));
         while(!tocheck.empty()) {
-            auto state = tocheck.front();
+            auto state = std::move(tocheck.front());
             tocheck.pop();
             std::map<char, std::set<State> > trans;
-            for (int c = 0; c < 256; c++) {
+            for (char c : usedChars) {
                 auto nextstate = getNextState(state,c,e);
-                trans[c] = nextstate;
                 if(dfa.find(nextstate) == dfa.end()) {
+                    dfa[nextstate] = {};
                     tocheck.push(nextstate);
                 }
+                trans[c] = std::move(nextstate);
             } 
             dfa[state] = trans;
         }