Re: Ada bench : count words

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On Tue, 22 Mar 2005 12:47:51 +0000, Marius Amado Alves wrote:

>>>> Is Text_IO that bad?
>>>
>>> No, if you can solve The Get_Line puzzle :-)
>>
>> What about Get (Item : out Character)?
> 
> I tried and was too slow. Anyway I think I cracked the Get_Line puzzle 
> (review welcome). So now it reads from standard input as required. But 
> it's still 3 to 4 times slower than the C version.
> 
> -- Count words in Ada for the language shootout
> -- by Marius Amado Alves
> 
> with Ada.Characters.Handling;
> with Ada.Characters.Latin_1;
> with Ada.Text_IO;
> 
> procedure Count_Words is
> 
>     use Ada.Characters.Handling;
>     use Ada.Characters.Latin_1;
>     use Ada.Text_IO;
> 
>     Buffer : String (1 .. 4096);
>     Lines : Natural := 0;
>     Words : Natural := 0;
>     Total : Natural := 0;
>     In_Word : Boolean := False;
>     N : Natural;
> 
>     function Is_Separator (C : Character) return Boolean is
>     begin
>        return Is_Control (C) or C = ' ';
>     end;
> 
>     procedure Begin_Word is
>     begin
>        In_Word := True;
>     end;
> 
>     procedure End_Word is
>     begin
>        if In_Word then
>           Words := Words + 1;
>           In_Word := False;
>        end if;
>     end;
> 
>     procedure End_Line is
>     begin
>        Lines := Lines + 1;
>        Total := Total + 1;
>        End_Word;
>     end;
> 
>     procedure Count_Words (S : in String) is
>     begin
>        Total := Total + S'Length;
>        for I in S'Range loop
>           if Is_Separator (S (I)) then
>              if In_Word then End_Word; end if;
>           else
>              if not In_Word then Begin_Word; end if;
>           end if;
>        end loop;
>     end;
> 
> begin
>     while not End_Of_File loop

Replace End_Of_File with End_Error handling.

>        Get_Line (Buffer, N);

Get_Line does one extra line scan. So it will be inherently slower. Then it
would not take any advantage of having Buffer if lines are shorter than 4K.
Once Count_Words is inlined the buffer size does not matter.

BTW, you can safely declare Buffer either 1 or 1G bytes, because hidden
buffering happens anyway in Text_IO. (You only save calls to Get_Line.) Who
knows how large are buffers there? This probably disqualifies Text_IO, as
well as C's getc! It should be raw "read".

>        Count_Words (Buffer (1 .. N));

Wouldn't it count buffer ends as word separators for lines longer than 4K?

>        if N < Buffer'Length then
>           End_Line;
>        end if;
>     end loop;
> 
>     Ada.Text_IO.Put_Line
>       (Natural'Image (Lines) &
>        Natural'Image (Words) &
>        Natural'Image (Total));
> end;

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de