This is what is commonly refered to as "the dragon book". http://en.wikipedia.org/wiki/Compilers:_Principles,_Techniqu...

But that may be too massive for what you are looking for. I found that Crockfords essay http://javascript.crockford.com/tdop/tdop.html really helps if you know javascript. And reading something like tinypy http://www.philhassey.com/blog/2008/02/28/tinypy-10-mit-lice... may help.

Also not necessarily a compiler book, but The Little Schemer does have you building a basic scheme interpreter at the end of the book.

The dragon book is Aho, Sethi, & Ullman. If you look through modern string search algorithm implementation, for example, the number of hits for Aho is up there with Knuth. Not that string searching has an awful lot to do with compilation, more lexing and parsing, but still. Aho and Ullman are pillars and the book is sort of the bible.

http://en.wikipedia.org/wiki/Compilers:_Principles,_Techniqu...

It's been updated to cover parallelization and other topics that were considered specialties when it was originally written. I have an older copy and am now wondering whether it mightn't be fun to trade up. We'll see.

Have you played with ANTLR? That's a neat tool. The Ragel state machine compiler is a fun toy, as well.

Cooper & Torczon does cover scanners and parsers.