Read integers from binary <8 bits in R - r

Question

In R, I am trying to convert binary data to integer values, but instead of 1 value being stored in 1 byte, multiple values are stored within and across bytes.
I know there are 12 integer values stored across 64 bits (8 bytes). The 12 integers have the following bit count: 5,6,5,5,4,7,5,6,5,5,4,7
After the following code:
time <- readBin(fid,integer(),size=1,n=8,signed='FALSE')
The return is:
[1] 25 156 113 63 214 158 113 63
The correct data should be:
25 32 19 17 11 31 22 54 19 17 11 31
I have tried using bitAnd and bitShiftL (package bitops), but have had no real success. And help would be greatly appreciated.

Answer 1

Note that the operation on each 4-byte integer is the same (the pattern is repeated twice). Thus, it suffices to solve the problem for a 4-byte integer, and loop over the 4-byte integers in the file (retrieved via readBin). This is much simpler than considering the problem byte-by-byte.
# length(x) should be 1
bitint <- function(x, bitlens) {
result <- integer(length(bitlens))
for (i in seq_along(bitlens)) {
result[i] <- bitwAnd(x, (2^bitlens[i])-1)
x <- bitwShiftR(x, bitlens[i])
}
return(result)
}
bitlens <- c(5,6,5,5,4,7)
x <- c(1064410137L, 1064410838L)
c(sapply(x, function(i) bitint(i, bitlens)))
## [1] 25 32 19 17 11 31 22 54 19 17 11 31

Answer 2

I don't know of a clean elegant way to do this with the standard data reading base functions (function like redBin seems to prefer to no less than a byte at a time). So i've created a function that does some of the messy calculations to extract bits from bytes and turn them into numbers. I did end up using the bitwise operators in base R (see ?bitwAnd) Here's the function
bitints <- function(bytes, bitlengths) {
stopifnot(sum(bitlengths) <= 8*length(bytes))
stopifnot(all(bitlengths <= 8))
bytebits <- rep.int(8, length(bytes))
masks <- c(1L,3L,7L,15L,31L,63L,127L, 255L)
outs <- numeric(length(bitlengths))
for(i in seq_along(bitlengths)) {
need <- bitlengths[i]
got <- 0
r <- 0
while(need>0) {
j <- which(bytebits>0)[1]
bitget <- min(need, bytebits[j])
r <- r + bitwShiftL(bitwAnd(bytes[j],masks[bitget]), got)
bytebits[j] = bytebits[j]-bitget
bytes[j] = bitwShiftR(bytes[j], bitget)
need <- need - bitget
got <- got + bitget
}
outs[i] <- r
}
outs
}
You just pass in your array of byte values and your array of bit sizes to get the values you need. Here's an example using your data.
bytes <- c(25L, 156L, 113L, 63L, 214L, 158L, 113L, 63L)
bitlens <- c(5,6,5,5,4,7,5,6,5,5,4,7)
bitints( bytes, c(5,6,5,5,4,7,5,6,5,5,4,7) )
# [1] 25 32 19 17 11 31 22 54 19 17 11 31
Note that I had to change around some of your bit lengths to get the values you were expecting. You might want to double check that you either had the expected output correct or that your bit lengths were correct.