Getter:peek () is weird! about luamacro

[stevedonovan]

All I can say is yes, peek is a weird method. Let me have a look this
weekend.

On Tue, Jan 6, 2015 at 3:52 AM, Gary V. Vaughan [email protected]
wrote:

Hi Steve,

It makes sense to me that get:peek (n, true) counts spaces and comment
tokens, and that since you can't predict what space and comment tokens
might be in the input, why that is not the default.

But, with only one argument, peek is behaving weird… apparently space
tokens are not filled in correctly, containing a copy of a neighbouring
token, but taking up space in the 'skipped' token stream anyway. Is this a
bug, or am I not understanding how it works?

local macro = require "macro"

macro.define ('target', function (get)
local i = 0
repeat
i = i + 1
local tk, v = get:peek (i, true)
local tk_, v_ = get:peek (i)
print (i, tk, v. tk_, v_)
until v == " EOS"
return nil, trueend)

macro.substitute [[ target foo = "bar", function (t) -- comment print "test" end]]

What I'd like to do is check whether the 2nd non-space. non-comment token
is an "=" before expanding the macro, but that token is always 3 peek
positions forward, irrespective of dont_skip mode (unless of course the
unexpanded text has different spacing to my example).it seems like
1-argument peek ought to be able to do this?

Ideally, I'd then like to preserve line-breaks as I copy the remaining
tokens to the output, injecting some other missing lua syntax as I go. peek
with true as the second argument seems like it will let me detect and copy
whitespace once I get past the first issue?

Cheers!

—
Reply to this email directly or view it on GitHub
#5.

[gvvaughan]

The code above, and the macro I am writing are working with #6.

Unless I'm grossly misunderstanding how peek is designed to be used, this seems like the right approach to me.