Rudolf / huffman

This saves more than huffman encoding. Note the header is still stored as usual.

This is simpler and less error prone. Just compare right->ch and if it doesn't
match, check if next left is the last left.

This allows the last byte to be interpreted correctly, by not allowing the
cursor to move further than the max bits to read.

This allows custom use by moving the start of bit reading.

This somehow worked, but with upcoming first-3 bits as throwaways, it no longer
does. FIXIT

No need to allocate +1 now.

This might otherwise confuse with regular '0'.

As I heard today, the compressor should do binary files.

NB! This EOF solution works perfectly fine, as 0xff is interpreted as ' '
(not space). Normal text files never have this character in them.

This value should never be less than 0

Also remove variables as I had something in mind I am not planning to do.

Also adjust some spaces and whatnot.

Use same style on all files and add guards to headers, where missing.

The 0 char was encoded as just 0 (as the left->ch == 0 when it is a branch).
Fix is to first compare right index, and if it doesn't match then use last left.

Our special 0xff as EOF does not allow some files.

If the file has been padded with extra 0's (since file is written in total
bytes), there is no way to determine what should be the last value. Implement
special 0xFF for this case.

This also changes char to unsigned to avoid comparison failure.

While stdin/stdout is all nice and dandy for piping, we want to show useful
information about compression, so remove this feature.

Not really needed but for consistency's sake

Horizontal tree print fills all my needs

This is useful to read in a full byte. Note this cannot be done using simple
fread, as we need to consider the bit cursor.

This is useful to write out a full byte. Note this cannot be done using simple
fwrite, as we need to consider the bit cursor.

https://gist.github.com/hackedbellini/366ccf1cead1ba14049f