Promizer v2.0 format
Created by MC68000 / Masque / TRSI (Frank Hlsmann)
Analysed by Asle / ReDoX (asle@mygale.org)

NOTE: It's more or less the same structure as PM18a. The 'only' great
     difference is the repacking of the Notes (no more like ptk). thus,
     all the crap (!) about the finetuned value in PM18a are no longer in
     this version. Periods are replaced by relative note numbers. Far
     easier to handle when converting :).

Source:
 - Two ripped PM20 from various cracktro ..
 - "PM20.Drd" sent by Gryzor (Thx !). Bubble Bobble rules :)


Offset    size (byte)    Comment
------    -----------    -------

 0        5094 (13E6h)   Replay routine and, certainly, near the end of this
                         block, some unidentified bytes. Since I could manage
                         without these to convert, I guess they are not that
                         very important.
5094        72 (48h)     all the 36 note pitch (similar to ptk's). Each is
(13E6h)                  saved as a word. It's not much use to know this,
                         but there are so many unclear bytes in this format,
                         that I welcome every small "victory" like this :).

5198           2         number of pattern (REAL !)
(144Eh)

5200           2         "used" size of the pattern table
(1450h)                  /2 to get pattern list size ...

5202          128        64 word pattern table.
(1452h)

      ***************************************
      * the following is repeated 31 times  *
      * with 8 bytes description for 1 smp  *
******************************************************
                                                     *
5458 (1552h)   2         Sample Size / 2             *
5460 (1554h)   1         Finetune (0 -> 1E) (*2 !)   *
5461 (1555h)   1         Volume (0 - 40h)            *
5462 (1556h)   2         Loop Start / 2              *
5464 (1558h)   2         Loop Size / 2               *
                                                     *
******************************************************


5706           4         Address of sample data + $144E
(164Ah)                  ($144E is the number of pattern address)

5710           4         Address of the "reference table" + $144E
(164Eh)

   NOTE: These two dword give us the size of the "reference table" when
        substracting. ($164A - $164E)

5714                     pattern data. Each pattern is 512 bytes long. Though,
(1652h)                  that's the theory because some patterns may NOT be
                         that long. In fact, you'll have to check if the note
                         isn't a PATTERN BREAK command or a PATTERN JUMP. If
                         so, the pattern ends and another begins ... therefore,
                         its size is less than the 512 'standard bytes'.
                         Now, whatever the pattern size is, its meaning does
                         not vary :). In fact, those data are some values which
                         refer to a "reference table" (description is right
                         below).
                         RMQ: first value is 00h.

$144E          ?         the "reference table" which contains the standard
+$164E                   PTK pattern data. Its size is easy to calculate since
                         it consists of blocks of 4 bytes, knowing that there
                         are 'n' blocks ('n' is the highest value in the
                         pattern data).
                         The other way to get the size is described upward.


      * Here's a quick description of these four bytes

       0000-0000  0000-0000  0000-0000  0000-0000
       ||    |||  |       |  |  | |  |  |       |
       | \  / ||   \     /   |  |  \/    \     /
       |sample||   relative   \/ effect  effect
       |      ||   note nbr unused       value
    unused   unused 

 
relative note number (PTK):

C-1 to B-1 : 856,808,762,720,678,640,604,570,538,508,480,453
C-2 to B-2 : 428,404,381,360,339,320,302,285,269,254,240,226
C-3 to B-3 : 214,202,190,180,170,160,151,143,135,127,120,113

becomes (PM20):

C-1 to B-1 : 02,04,06,08,0a,0c,0e,10,12,14,16,18
C-2 to B-2 : 1a,1c,1e,20,22,24,26,28,2a,2c,2e,30
C-3 to B-3 : 32,34,36,38,3a,3c,3e,40,42,44,46,48

 
/*******************/

?              4         Four empty bytes as in PM18a. Still not figured
                         out their meaning !.

$144E          ?         Sample datas.
+$164A


(NOTE: It seems ProWizard adds four empty bytes here when ripping ... .
The original packer, promizer 2.0, doesn't. ).