ltokenp is now available at
	http://www.tecgraf.puc-rio.br/~lhf/ftp/lua/5.3/ltokenp.tar.gz

It should work for Lua 5.1, 5.2, and 5.3.

The README is attached. Enjoy. All feedback welcome.
--lhf

ltokenp is a token processor for Lua: it allows you to process the
stream of tokens coming from the Lua lexer.

Potential uses include:
- Compressing Lua programs by removing comments and whitespace (strip.lua)
- Removing assertions (assert.lua)
- Adding new syntax sugar (self.lua)
- Experimenting with new syntax without hacking the Lua source (reserved.lua)

ltokenp accepts Lua scripts to run and files to process.
The scripts and files are executed and processed as seen.
Each script appears as a separate argument after '-s', one '-s' per script.

Scripts typically output the contents of the files with some modifications.
Unfortunately, all comments and whitespace are eaten by the lexer and never
reach the token stream.

A typical usage is
	ltokenp -s script.lua [file.lua ...]
but you can also do
	ltokenp -s s1.lua f1.lua -s s2.lua f2.lua

A global function named FILTER will be called once for each token seen
in the files. So, scripts typically define FILTER. If no scripts are
given, ltokenp just dumps the token stream with this:

	function FILTER(line,token,text,value)
		print(line,token,text,value)
	end

As can been above, the FILTER function receives 4 arguments from the lexer:
- the line number where the token appears
- the token as a number
- the token as text
- the value of names, numbers, and strings;
  for other tokens, the value is the same as the text.

To try ltokenp, just edit Makefile to reflect your installation of Lua
and then run make. This will build ltokenp and run a simple test. Note
that ltokenp needs inside information from the Lua private headers and
so it needs to be built using a Lua build directory, not just the usual
installation directories.

This code is hereby placed in the public domain.
Please send comments, suggestions, and bug reports to lhf@tecgraf.puc-rio.br .