hawkins / druid Goto Github PK

All plans are subject to change as optimizations or re-structuring is needed. As a basic first step, the druid preprocessor should be able to mimic the "include" capability of the C preprocessor.

For now, this will take the form of building an ordered memory model of ASCII characters consisting of the contents of the include files, and being able to output this char stream to stdout.

We're working on a variety of platforms (windows, mac, *nix...), we may want to try a CMake setup for consistent build products.

This issue was generated by todo based on a TODO comment in e7213ee. It's been assigned to @assemblyDruid because they committed the code.

We need a tool to use that will:

• take some input filename
• read in that file and pass it through the pipeline

Effectively this CLI will pass fork to several child processes, the first of which will call the preprocessor on the file, then pipe to lexer, then pipe to parser, then pipe to compiler.

For no reason beyond a Druid casting magic on the file to make it do things, pretend this CLI is called cast:

$ cast hello.druid --verbose
* preprocessing hello.druid
  * summoned ./echo.druid
    * summoned :math
  * summoned :io
* parse check successful
* compiled to web assembly 
* done in 0.27 sec

• Come up with a list of token types
• Create token struct
  • type, text, length
• Tokenize input stream
  • From file? From stdin?
• Write output stream
  • To file? To Stdout? Where?

• Handle #summon in apply
  • Read file in parameter and write in-place into fp
• Write output to stdout
  • For use in feeding downstream, to lexical analyzer.
• Read files into memory instead of continuously reading/writing to disk

For the next step in the pipeline, the parser, we'll need a proper grammar. I think these are typically done in EBNF (or see these lecture notes).