mdurrant / php-binary-reader Goto Github PK

Hi,

I'm not sure but I think the method bitReader in the class Int32 is wrong.
My fix is on the masks. I think it should be :

        $hiMask = $bitmask->getMask($br->getCurrentBit(), BitMask::MASK_LO) ^ 0xFF;
        $loMask = $bitmask->getMask($br->getCurrentBit(), BitMask::MASK_HI) ^ 0xFF;

Case :

Buffer :
FF 07 00 00 00

Position : 1 (2nd byte)
Current bit : 2
Nest Byte : FF

Try to read int32.

The masks are :
hiMask : 00111111
loMask : 00000011

But should be :
hiMask : 11111100
loMask : 11000000

The result will be 1058799616 (111111000111000000000000000000) and should be 4229693440 (11111100000111000000000000000000).

Great project 👍

Can the latest commits from june be tagged into a release? Need the floating point reader but don't want to depend on dev-master

Thanks

Most of the functions specify return type int. While I believe the ones reading from $this->byteReader are returning string. $this->byteReader->read uses fread which returns a string (as mentioned in the comments for $this->byteReader->read).

Conversely, getUInt16 has return type string in the comments. This should probably be int.

I think some other functions have similar problems.

I've been using the library to get contiguous bits from a binary file where the bits are not byte-aligned. I have not been getting the values I expect.

After much head-scratching, tweaking and testing I came to realise that the Bit reader wasn't working the way I expected.

It seems to be a little naive in that it assumes byte grouping and order is always based on standard approaches and so it doesn't return the anticipated values when I read bits in sequence from a file that doesn't follow convention.

For example, I have a data file with a series of 40-bit frames. Byte order becomes insignificant because the bits are sequential and variable-length, e.g. a two 14-bit integers, followed by a 5-bit integer, two individual bits, a 3-bit integer and finally 2 bits for padding to a full byte.

The problem appears to be that, for non-standard length frames like this (that potentially cross traditional byte boundaries), this library doesn't handle the bits in the strict order I'm asking for because it breaks the bits down into multi-byte chunks.

Starting at Type/Bit.php#37 for example, it explicitly tries to break down the length of bits requested into a number of whole bytes to read using the 'standard' methods (Int32, Int16 etc).

I've adapted this for my use case and I'm conscious I might not be the only person experiencing this issue, so I'm keen to integrate my changes.

While I'd like to merge my changes in, I first wanted to open up discussion for the best way to handle this as I'm sure the current Bit reader is still useful the way it is and I don't want to disrupt that.