Re: Luma errors

[Tony Finch <dot@...>]

Roberto Ierusalimschy <roberto@inf.puc-rio.br> wrote:

> > > Would you mind explaining how these changes make writing and composing
> > > grammars *much* more convenient?
> >
> > The second change lets the grammar writer refactor a non-terminal to a
> > definition without adding % to all of its uses.
>
> Can't you simply keep the non-terminal as a non-terminal, adding one new
> rule like this?
>
>   non-terminal -> %new-definition

I usually use re to define stand-alone rules, since it has nicer syntax
but there are things I can do with stand-alone rules that I can't do with
re, such as folds and argument captures. I can't use a re grammar to
define my low-level non-terminals in one go, since re grammars are
immediately fixed and I can't extract non-terminal definitions from a
compiled parser.

So I use the missing % patch to make it easier to compose re patterns.

The usual way I do this is with this code:

    local parse = (function()
        -- environment for non-terminals
        local env = {}
        local function p(pat)
                return re.compile(pat, env)
        end
        setfenv(1,env)

        -- then define various non-terminals

        CRLF = lpeg.S"\r\n"

        name = p[[ {[a-z]+} ]]
        num  = p[[ {[0-9]+} ]] / tonumber
        hex  = p[[ {[0-9a-fA-F]+} ]] /
                function (n) return tonumber(n,16) end

        ipv4 = p[[ num "." num "." num "." num ]] /
            function (a,b,c,d)
                return a * 16777216 + b * 65536 + c * 256 + d
            end

	-- etc.

        return env
    end)()

    parse.ipv4:match(str)

A couple of examples:

http://www-uxsup.csx.cam.ac.uk/~fanf2/hermes/conf/exim/sbin/bogons
http://www-uxsup.csx.cam.ac.uk/~fanf2/misc/geoip/geoip2lua.lua

Tony.
-- 
f.anthony.n.finch  <dot@dotat.at>  http://dotat.at/
Hebrides, Bailey, Fair Isle, Faeroes: West backing southwest 5 to 7. Rough or
very rough, occasionally high. Squally showers, rain later. Moderate or good.