Natools

   package Dict_Sources is
      type Enum is
        (S_Expression,
         Text_List,
         Unoptimized_Text_List);
   end Dict_Sources;

   package Methods is
      type Enum is (Encoded, Frequency, Gain);
   end Methods;

   package Options is
      type Id is
        (Output_Ada_Dict,
         Dictionary_Input,
         Decode,
         Encode,

         Max_Sub_Size,
         Stat_Output,
         No_Stat_Output,
         Text_List_Input,
         Fast_Text_Input,
         Max_Word_Size,
         Sx_Output,
         No_Sx_Output,
         Score_Method);
   end Options;

   package Getopt is new Natools.Getopt_Long (Options.Id);

   type Callback is new Getopt.Handlers.Callback with record
      Display_Help : Boolean := False;
      Need_Dictionary : Boolean := False;
      Stat_Output : Boolean := False;
      Sx_Output : Boolean := False;
      Sx_Dict_Output : Boolean := False;
      Min_Sub_Size : Positive := 1;
      Max_Sub_Size : Positive := 3;
      Max_Word_Size : Positive := 10;
      Job_Count : Natural := 0;
      Filter_Threshold : Natools.Smaz.Tools.String_Count := 0;
      Score_Method : Methods.Enum := Methods.Encoded;
      Action : Actions.Enum := Actions.Nothing;
      Ada_Dictionary : Ada.Strings.Unbounded.Unbounded_String;
      Hash_Package : Ada.Strings.Unbounded.Unbounded_String;
      Dict_Source : Dict_Sources.Enum := Dict_Sources.S_Expression;
   end record;

   overriding procedure Option

      Counts : out Natools.Smaz.Tools.Dictionary_Counts);
      --  Dispatch to parallel or non-parallel version of Evaluate_Dictionary
      --  depending on Job_Count.

   function Getopt_Config return Getopt.Configuration;
      --  Build the configuration object












   function Length
     (Dictionary : in Natools.Smaz.Dictionary;
      E : in Ada.Streams.Stream_Element)
     return Score_Value
     is (Natools.Smaz.Dict_Entry (Dictionary, E)'Length);
      --  Length of a dictionary entry

   procedure Optimization_Round
     (Dict : in out Holders.Holder;
      Score : in out Ada.Streams.Stream_Element_Count;
      Counts : in out Natools.Smaz.Tools.Dictionary_Counts;
      Pending_Words : in out Natools.Smaz.Tools.String_Lists.List;
      Input_Texts : in Natools.Smaz.Tools.String_Lists.List;
      Job_Count : in Natural;
      Method : in Methods.Enum;
      Updated : out Boolean);
      --  Try to improve on Dict by replacing a single entry from it with
      --  one of the substring in Pending_Words.

   function Optimize_Dictionary
     (Base : in Natools.Smaz.Dictionary;
      Pending_Words : in Natools.Smaz.Tools.String_Lists.List;
      Input_Texts : in Natools.Smaz.Tools.String_Lists.List;
      Job_Count : in Natural;
      Method : in Methods.Enum)
     return Natools.Smaz.Dictionary;
      --  Optimize the dictionary on Input_Texts, starting with Base and
      --  adding substrings from Pending_Words.

   procedure Parallel_Evaluate_Dictionary
     (Job_Count : in Positive;
      Dict : in Natools.Smaz.Dictionary;

     (Dictionary : in Natools.Smaz.Dictionary;
      Counts : in Natools.Smaz.Tools.Dictionary_Counts;
      E : Ada.Streams.Stream_Element)
     return Score_Value
     is (Score_Value (Counts (E)) * (Length (Dictionary, E) - 1));
      --  Score value using the number of bytes saved using E

   function Score
     (Dictionary : in Natools.Smaz.Dictionary;
      Counts : in Natools.Smaz.Tools.Dictionary_Counts;
      E : in Ada.Streams.Stream_Element;
      Method : in Methods.Enum)
     return Score_Value
     is (case Method is
         when Methods.Encoded => Score_Encoded (Dictionary, Counts, E),
         when Methods.Frequency => Score_Frequency (Dictionary, Counts, E),
         when Methods.Gain => Score_Gain (Dictionary, Counts, E));
      --  Scare value with dynamically chosen method

   function To_Dictionary
     (Handler : in Callback'Class;
      Input : in Natools.Smaz.Tools.String_Lists.List)
     return Natools.Smaz.Dictionary;
      --  Convert the input into a dictionary given the option in Handler

   function Worst_Index
     (Dict : in Natools.Smaz.Dictionary;
      Counts : in Natools.Smaz.Tools.Dictionary_Counts;
      Method : in Methods.Enum)
     return Ada.Streams.Stream_Element;
      --  Remove the worstly-scored item from Dict


   overriding procedure Option
     (Handler  : in out Callback;
      Id       : in Options.Id;

         when Options.Job_Count =>
            Handler.Job_Count := Natural'Value (Argument);

         when Options.Filter_Threshold =>
            Handler.Filter_Threshold
              := Natools.Smaz.Tools.String_Count'Value (Argument);

         when Options.Score_Method =>
            Handler.Score_Method := Methods.Enum'Value (Argument);
      end case;
   end Option;


   procedure Evaluate_Dictionary
     (Job_Count : in Natural;
      Dict : in Natools.Smaz.Dictionary;

      R.Add_Option ("stats",         's', No_Argument,       Stat_Output);
      R.Add_Option ("no-stats",      'S', No_Argument,       No_Stat_Output);
      R.Add_Option ("text-list",     't', No_Argument,       Text_List_Input);
      R.Add_Option ("fast-text-list", 'T', No_Argument,       Fast_Text_Input);
      R.Add_Option ("max-word-len",  'W', Required_Argument, Max_Word_Size);
      R.Add_Option ("s-expr",        'x', No_Argument,       Sx_Output);
      R.Add_Option ("no-s-expr",     'X', No_Argument,       No_Sx_Output);
      R.Add_Option ("score-method",       Required_Argument, Score_Method);

      return R;
   end Getopt_Config;


   procedure Optimization_Round
     (Dict : in out Holders.Holder;
      Score : in out Ada.Streams.Stream_Element_Count;
      Counts : in out Natools.Smaz.Tools.Dictionary_Counts;
      Pending_Words : in out Natools.Smaz.Tools.String_Lists.List;
      Input_Texts : in Natools.Smaz.Tools.String_Lists.List;
      Job_Count : in Natural;
      Method : in Methods.Enum;
      Updated : out Boolean)
   is
      use type Ada.Streams.Stream_Element_Offset;

      New_Value : Ada.Strings.Unbounded.Unbounded_String;
      New_Position : Natools.Smaz.Tools.String_Lists.Cursor;
      Worst_Index : constant Ada.Streams.Stream_Element
        := Smaz.Worst_Index (Dict.Element, Counts, Method);
      Worst_Value : constant String
        := Natools.Smaz.Dict_Entry (Dict.Element, Worst_Index);
      Worst_Count : constant Natools.Smaz.Tools.String_Count
        := Counts (Worst_Index);
      Base : constant Natools.Smaz.Dictionary
        := Natools.Smaz.Tools.Remove_Element (Dict.Element, Worst_Index);
      Old_Score : constant Ada.Streams.Stream_Element_Count := Score;

   end Optimization_Round;


   function Optimize_Dictionary
     (Base : in Natools.Smaz.Dictionary;
      Pending_Words : in Natools.Smaz.Tools.String_Lists.List;
      Input_Texts : in Natools.Smaz.Tools.String_Lists.List;
      Job_Count : in Natural;
      Method : in Methods.Enum)
     return Natools.Smaz.Dictionary
   is
      Holder : Holders.Holder := Holders.To_Holder (Base);
      Pending : Natools.Smaz.Tools.String_Lists.List := Pending_Words;
      Score : Ada.Streams.Stream_Element_Count;
      Counts : Natools.Smaz.Tools.Dictionary_Counts;
      Running : Boolean := True;
   begin
      Evaluate_Dictionary (Job_Count, Base, Input_Texts, Score, Counts);

      while Running loop
         Optimization_Round
           (Holder,
            Score,
            Counts,
            Pending,
            Input_Texts,
            Job_Count,
            Method,
            Running);
      end loop;

      return Holder.Element;
   end Optimize_Dictionary;

            when Options.Filter_Threshold =>
               Put_Line (Output, " <threshold>");
               Put_Line (Output, Indent & Indent
                 & "Before building a dictionary from substrings, remove");
               Put_Line (Output, Indent & Indent
                 & "substrings whose count is below the threshold.");

            when Options.Score_Method =>
               Put_Line (Output, " <method>");
               Put_Line (Output, Indent & Indent
                 & "Select heuristic method to replace dictionary items"
                 & " during optimization");
         end case;
      end loop;
   end Print_Help;


   function To_Dictionary
     (Handler : in Callback'Class;

                     Natools.Smaz.Tools.Simple_Dictionary_And_Pending
                       (Counter, 254, Selected, Pending);

                     return Optimize_Dictionary
                       (Natools.Smaz.Tools.To_Dictionary (Selected, True),
                        Pending,
                        Input,
                        Handler.Job_Count,
                        Handler.Score_Method);
                  end;
               else
                  return Natools.Smaz.Tools.To_Dictionary
                    (Natools.Smaz.Tools.Simple_Dictionary (Counter, 254),
                     True);
               end if;
            end;
      end case;
   end To_Dictionary;


   function Worst_Index
     (Dict : in Natools.Smaz.Dictionary;
      Counts : in Natools.Smaz.Tools.Dictionary_Counts;
      Method : in Methods.Enum)
     return Ada.Streams.Stream_Element
   is
      Result : Ada.Streams.Stream_Element := 0;
      Worst_Score : Score_Value := Score_Encoded (Dict, Counts, 0);
      S : Score_Value;
   begin
      for I in 1 .. Dict.Dict_Last loop
         S := Score (Dict, Counts, I, Method);

         if S < Worst_Score then
            Result := I;
            Worst_Score := S;
         end if;
      end loop;