Iv recently been working on a compiler, so far i have a fairly decent working lexical analyzer and token scanner. So i can break down the code into a series of tokens and analyze them for correct strucutre.

I also understand how to generate x86 instructions. However im lost somewhere in between linking these two. The book i have describes the first and last step but seems to omit how i get from:

Have a series of tokens that are correct eg :

import java.whatever;
Keyword | package Name | seperator | class name | seperator

It does not explain how i go from having a series of tokens that are known to be in the correct structure of the language to outputing binary or even assembly language from these tokens. Must i scan through the tokens again and generate binary or an assembly statement for each line? Or should it be implemented in a higher level of abstraction and add to the total value's of an executeable class?

I know its kind of a broad question, but any help on the general logic and filling the gap between correct tokens to outputing binary would be good :) its not something im grasping from the book.

What text book(s) are you using? There are some good ones out there that deal with this stuff. Generally, you tokenize the code, and then create a set of tree structures or finite-state-machine to represent the program. As you might infer from this, there is more than one approach that can work for you.

Two books that I have used in the past are Allen Holub's "Compiler Design in C", and Fischer/LeBlanc's "Crafting a Compiler". Both are on my shelf. FWIW, between the two of them, they take up almost 4 inches of shelf space... :-)

Very helpful thanks, the book im using is compilers tools and techniques by Alfred V.aho, Ravi sethi, jeffrey D Ullman. It has been good for my overall understanding but there are parts i must admit that iv found guilty of going too deep before you fully understand the concept. Hence not understanding exactly what had to be done.

Thanks for the insight!