In the latter case, the lookup table has to have 256*256=65536 entries. In the former case, the lookup table has to have 256 entries. Either the lookup table maps individual ascii hex characters to half bytes or it maps two hex characters to a full byte. I see two possible ways to solve this with a lookup table. As the question was how to convert files, our implementation should be fast on large data. Here is how I create a header with a character array 'testdata' of 104857600 bytes, roughly 105 MB. Firstly, we want to have the whole test data in memory so that we avoid disk I/O influencing the test. Strangly enough, none of the answers above implement a purely arithmetic solution and some answers even assume that 'converting to binary' means converting to a ascii string of characters '0' and '1'.
I wanted to find a solution which only uses a lookup table and benchmark that against a solution that uses arithmetics instead. There are many ways to solve this question that use some arithmetics to convert from ascii character ranges 0-9 and a-f (or A-F) to binary.
