C++ function with rounding (cross-platform)

Question

I have the following function that I just can't get right. I am working on it again and again, and somehow something always turns out wrong.

I have encoded some audio data. The audio data is encoded in "frames" of 3 bytes each.

For example like this:

Frame #1
0
1
2

Frame #2
3
4
5

Frame #3
6
7
8

Now when I want to decode some audio (let's say from byte position 4 with the length of 3 bytes, I first have to calculate in which frame this audio would be found and how many frames I have to decode and what the offset in the decoded frame is.

In this case, I would have to read frame #2 and #3, and the offset would be 1.

I have tried to set up the following void:

int g_iByteSize1FrameDecoded = 3;

void CalcFrames(unsigned long uByteStart,unsigned long uByteCount,unsigned long &uStartFrame,unsigned long &uFramesToRead,unsigned long& uOffset)
{
    ////calculate in which decoded frame the byte from uByteStart would be found in
    uStartFrame = ((uByteStart) / g_iByteSize1FrameDecoded) + 1;
    unsigned long iEndFrame = ((uByteStart + uByteCount) / g_iByteSize1FrameDecoded) + 1;

    uFramesToRead = (iEndFrame - uStartFrame + 1);

    uOffset = (uByteStart) % g_iByteSize1FrameDecoded;
}

But it just doesn't work, something ALWAYS goes wrong... rounding, the math itself...

Could somebody with some math skills perhaps have a look where my error(s) could be?

Answer 1

Byte i is in frame i / 3 at offset i % 3:

void CalcFrames(unsigned long uByteStart,unsigned long uByteCount,unsigned long &uStartFrame,unsigned long &uFramesToRead,unsigned long& uOffset) {
    uStartFrame = uByteStart / g_iByteSize1FrameDecoded + 1;
    uOffset = uByteStart % g_iByteSize1FrameDecoded;
    unsigned long lastFrame = (uByteStart + uByteCount - 1) / g_iByteSize1FrameDecoded + 1;
    uFramesToRead = lastFrame - uStartFrame + 1;
}

Answer 2

I'm dubious about your decision to count frames and bytes starting from 1 rather than 0, but perhaps it works better for the rest of your code - I don't know.

As far as CalcFrames goes, I can only see an issue with uOffset . The calculation should be:

uOffset = (uByteStart % g_iByteSize1FrameDecoded) - 1;

For your example, the offset of byte 5 from the start of the frame is 1 byte. Your original calculation simply does 5 % 3 which is 2, hence the need to additionally subtract 1.