Sorry your browser is not supported!

I am using SWF files in my DBPro game, as cutscenes/minigames. I already have a plugin that lets me use the flash files in my game (thanx www.accode.com). However, now I need to read the width & height of the movie from the file.

I already have obtained the file format spec from macromedia, read through the important parts etc. The problem lies in reading strange datatypes - you will see if you look at the attached screenshot from the SWF specification what I mean.

I need to be able to read an unsigned 5 bit number, and four signed Nbit numbers. If you understand, please help - otherwise dont try.

I too tried to understand the flash format. As TCA explained to me - they're just integers with different ranges. I think for this level of detail you may need to write it in C++ and access it via a TPC.

well, the C++ book and compiler are coming out... wish me luck

Hi DJ,

This is easier in C++ or assembler, but you can pull it off in most flavors of Basic. Here are some thoughts in case you're still stuck.

I think UB[5] in the flash docs means you want the rightmost 5 bits of an unsigned byte. In the example they gave where they encode the value 11 decimal in 5 bits, you want to mask the bit positions with X's here (remembering bit positions are numbered 7 to 0, starting from the left and working right):

x x x 0 1 0 1 1 (binary) = 11 (decimal)

In assembler you could do this with boolean AND or you could do arithmetic shift lefts on the value, which is equiv to repeatedly multiplying by two but throwing away the overflow bits. In Basic, you could fake this, I think, like this:

read value_from_file

for e = 7 to 5 step -1
test = 2 ** e
if value_from_file > test
value_from file = value_from_file - test
endif
next e

What I'm proposing, given the example of a 5-bit integer encoded into an 8-bit field, is treating the input as an 8-bit integer and then "masking" off the leftmost 3 bits by comparing the input value to what you'd have if you took 2 raised to the power of the bit position you wish to mask.

If you look at just the values you'd have if bit 7 was set, or only bit 6 was set, or only bit 5 was set, you'd see:

Binary Dec
--------------- -------------
1 0 0 0 0 0 0 0 = 128 (2 ^ 7)
0 1 0 0 0 0 0 0 = 64 (2 ^ 6)
0 0 1 0 0 0 0 0 = 32 (2 ^ 5)

So if you read an 8-bit value and saw it was > 128, then you know bit 7 is set. You don't want bit 7 in your result, so you subtract 128 from the value you read. And so on. for example:

If value_from_file = 11101011 = 235 decimal

In the pseudo code I wrote above, the loop would execute 3 times giving you these values:

e = 7, test = 128, value_from_file = 235 - 128 = 107
e = 6, test = 64, value_from_file = 107 - 64 = 43
e = 5, test = 32, value_from_file = 43 - 32 = 11

You'd want to alter the hard-coded values in the loop "for e = 7 to 5 skip -1" of course, after you learned your NBits value was 11 (that is, 11 bits were used in the reset of the rectangle definition to define the window). In the shockwave example of NBits=11, you'd go "e = 15 to 12 skip -1" to mask the bits NOT needed in an 11 bit integer.

Hmmm, i understand what you're saying Clueless. I did think of what you are saying above, but I don't know if the 5bit number is within a byte - it appears not to be. And an 11 bit number would not fit inside a byte (which has 8 bits), so I cannot read a byte... I have started to work on a plugin (installing C++ from that disk I found under my PC, and reading the dbpro sdk), so it shouldnt take me that much time anyhow.

Hi DJ,



Very true. I meant only that they're telling you (in the SWF docs) that they're storing the NBITS value as the lower 5 bits of a byte, when they refer to UB[5], so before you can do anything else you've got to decode the lower 5 bits of the first byte. From there, you could theoretically be faced with the need to decode longwords ("11111" = 31 bits maximum), not just the 8-bit bytes or 16-bit words.

Generally though, any time you're faced with the need to read variable-length data and the data itself contains a field length value, there's always a predefined starting point that everybody can agree upon. From the page in your example, the 5 bit length always appears as the lower 5 bits of the first byte of data in the RECT structure.

From there, you're guaranteed to have your xmin,xmax,ymin,ymax values actually span multiple bytes of data, so you'll have to slice bits and reassemble them. Xmin is going to always start at the 6th bit (immediately after the last bit of the NBits) but could end within the same byte as the NBITS field or stretch almost all the way across the following longword.

Anyway, you'll solve those mysteries one way or the other as you develop your plugin I think you're heading in the right direction doing it in C++. Let us know what you find out. You've shown me something I didn't know already, which was the ability to play flash movies via a plugin. Thankee.

It plays Flash games too....

And I may not develop a plugin, if I can do it in DBPro, as I will end up with some functions instead