Continue building DFA

lib/imbecile/regex/dfa.rb

@@ -4,6 +4,29 @@ module Imbecile
     class DFA
 
       class State
+
+        class Transition
+
+          attr_reader :code_point_range
+          attr_reader :destination
+
+          def initialize(code_point_range, destination)
+            @code_point_range = code_point_range
+          end
+
+        end
+
+        attr_accessor :accepts
+        attr_reader :transitions
+
+        def initialize
+          @transitions = []
+        end
+
+        def add_transition(code_point_range, destination)
+          @transitions << Transition.new(code_point_range, destination)
+        end
+
       end
 
       def initialize(nfas)
@@ -11,34 +34,49 @@ module Imbecile
         nfas.each do |nfa|
           start_nfa.start_state.add_transition(nil, nfa.start_state)
         end
-        @states = {}
+        @nfa_state_sets = {}
+        @states = []
         @to_process = Set.new
         nil_transition_states = start_nfa.start_state.nil_transition_states
-        @states[nil_transition_states] = 0
+        register_nfa_state_set(nil_transition_states)
-        process_nfa_state_set(nil_transition_states)
+        while @to_process.size > 0
+          state_set = @to_process.first
+          @to_process.delete(state_set)
+          process_nfa_state_set(state_set)
+        end
       end
 
       private
 
+      def register_nfa_state_set(nfa_state_set)
+        unless @nfa_state_sets.include?(nfa_state_set)
+          state_id = @states.size
+          @nfa_state_sets[nfa_state_set] = state_id
+          @states << State.new
+          @to_process << nfa_state_set
+        end
+      end
+
       def process_nfa_state_set(nfa_state_set)
+        state = @states[@nfa_state_sets[nfa_state_set]]
         transitions = transitions_for(nfa_state_set)
         while transitions.size > 0
           subrange = CodePointRange.first_subrange(transitions.map(&:code_point_range))
           dest_nfa_states = transitions.reduce(Set.new) do |result, transition|
             if transition.code_point_range.include?(subrange)
-              result << transition.last
+              result << transition.destination
             end
             result
           end
-          unless @states.include?(dest_nfa_states)
+          register_nfa_state_set(dest_nfa_states)
-            @to_process << dest_nfa_states
+          dest_state = @states[@nfa_state_sets[dest_nfa_states]]
-          end
+          state.add_transition(subrange, dest_state)
           transitions.delete_if do |transition|
             transition.code_point_range.last <= subrange.last
           end
           transitions.map! do |transition|
             if transition.code_point_range.first <= subrange.last
-              Transition.new(CodePointRange.new(subrange.last + 1, transition.code_point_range.last), transition.destination)
+              NFA::State::Transition.new(CodePointRange.new(subrange.last + 1, transition.code_point_range.last), transition.destination)
             else
               transition
             end
@@ -46,8 +84,8 @@ module Imbecile
         end
       end
 
-      def transitions_for(states)
+      def transitions_for(nfa_state_set)
-        states.reduce([]) do |result, state|
+        nfa_state_set.reduce([]) do |result, state|
           result + state.cp_transitions
         end
       end

lib/imbecile/regex/nfa.rb

@@ -12,6 +12,7 @@ module Imbecile
 
           def initialize(code_point_range, destination)
             @code_point_range = code_point_range
+            @destination = destination
           end
 
           def nil?