I nearly hate this problem :) Two hours for coding and two days for troubleshooting - and, as usual, AC! I use the best Pascal compiler ever been - Delphi 7.0, but piracy-haters and opensource-admirers from the jury use FreePascal I dislike very much :( So during two days I have been solving a very interesting problem - "How to find the differences between Delphi and FreePascal without FreePascal itself?" ;) So some hints for Pascal-programmers:

1. Store the compressed text in such way:

const
   TEXT: AnsiString = 'compressedtext';

If you write

TEXT = 'compressedtext';

or

TEXT: String = 'compressedtext';

you'll get WA(1).

2. You should divide your compressed text into the parts of 255 symbols each:

TEXT: AnsiString = 'part1'+'part2'+...+'partN';

3. Your should replace the following symbols

#0,#10,#13,#26,#39

with their codes to avoid compilation error of your output and getting WA(1). BTW, I get WA(1) several times because of #26. The example of correct string:

TEXT: AnsiString = 'a'#0#39'bcdef'#26;

And be careful while replacing and dividing - this technical problem is not difficult, but pay attention to it.

P.S. As for my solution, I've used Shannon-Fano algo, and the compression rate was at least 50% - all the tests larger than 5 Kb were passed with flying colors.

Edited by author 27.07.2004 12:38

Do you think piracy is good?

I think you should use {$H+} directive in your source and archive and you get Delphi-like strings. (I wrote about this directive earlier)

To guess about this feature you should try to compile your archive on FreePascal just once! (Don't you do this after first WA(1)?)

P.S. This is wonderful problem, isn't it?

>Do you think piracy is good?

Why should something be good or bad? Is a rain good? Or maybe bad? The sun? The Universe? The piracy is just inevitable natural phenomenon, the reflexion and the offspring of reality. I - and I bet you too - cannot deny it was piracy that curved our way to computers. Thanks to the piracy I am the Master of Computer Science. No offence, no holy wars :)

BTW, the thing was I believe Delphi is better than FreePascal...

>I think you should use {$H+} directive in your source and archive and you get Delphi-like strings. (I wrote about this directive earlier)

I know it works but $LONGSTRINGS (equal to $H) is ON by default in Delphi and I cannot understand why is it OFF in FreePascal.

>To guess about this feature you should try to compile your archive on FreePascal just once! (Don't you do this after first WA(1)?)

No, as I have said I haven't got FreePascal, so I couldn't compile my text with it. And in Delphi #26 is ok.

>P.S. This is wonderful problem, isn't it?

Yeah, it is. The most original problem in all the archive. But I think easy a bit. It would have been better to impose restrictions on the compression rate - I think 80% is enough to block bitshift-based solutions.

Many thanks for your helps. They have very much helped me.
But what for to use standard algorithms of compression (they not so difficult, but tiresome)? I am simple on each symbol spent on 7 bit. (Word is English, i.e. their codes from 1 up to 128).


Большое спасибо за твои подсказки. Они мне очень помогли.
Но зачем пользоваться стандартными алгоритмами сжатия (они не очень сложные, но нудные)? Я просто на каждый символ тратил по 7 бит. (Слова-та английские, т.е. их коды от 0 до 128).

How did You store them in the output file, hah?

[quote]
2. You should divide your compressed text into the parts of 255 symbols each:

TEXT: AnsiString = 'part1'+'part2'+...+'partN';
[/quote]

How does it work? Delphi will compile such long string, but Free Pascal (2.0.4) doesn't (so we have to use arrays); I used C++ for archive, it can compile code with long strings; so how you aviod it (if avoid)?