I'd suggest diving into discrete probability for a few hours. That should be enough to understand this post. Huffman coding is a little advanced but you can understand the algorithm (not the theory) with just the probability.

I have an alpha version of a text about lossless compression here (http://danbcstupidideas.neocities.org/compression.html)

It's very rough and I'm not a good writer. It's probably horribly patronising. I welcome any corrections and suggestions and etc.

this is 100 pages into MacKay's text, which is well paced

http://www.inference.phy.cam.ac.uk/itprnn/book.html