An LPEG grammar for the IRC protocol

irc-lpeg.lua

require("luarocks.require")
require("lpeg")

local v = lpeg.V
local c = lpeg.C
local grammar = {
    "message",      -- Initial rule
    message     = (v"COLON" * c(v"prefix"))^-1 * c(v"command") * c(v"params") * v"CRLF",
    params      = v"SPACE" * ((v"COLON" * v"trailing") + (v"middle" * v"params"))^-1,
    nospcrlfcl  = lpeg.R("\001\009", "\011\012", "\014\031", "\033\057", "\059\255"),
    middle      = v"nospcrlfcl" * (v"COLON" + v"nospcrlfcl")^0,
    trailing    = (v"COLON" + v"SPACE" + v"nospcrlfcl")^0,
    prefix      = (c(v"servername") * v"SPACE") + ((c(v"nickname") * (v"BANG" * c(v"user"))^-1 * (v"AT" * c(v"host"))^-1) * v"SPACE"),
    command     = (v"LETTER" ^ 1) + (v"DIGIT" * v"DIGIT" * v"DIGIT"),
    triple      = v"DIGIT" * (v"DIGIT"^-1) * (v"DIGIT"^-1),
    servername  = v"hostname",
    host        = v"hostname" + v"hostaddr",
    hostname    = v"shortname" * (v"DOT" * v"shortname")^0,
    shortname   = (v"LETTER" + v"DIGIT") * (v"LETTER" + v"DIGIT" + v"DASH")^0 * (v"LETTER" + v"DIGIT")^0,
    hostaddr    = v"ip4addr",
    ip4addr     = v"triple" * v"DOT" * v"triple" * v"DOT" * v"triple" * v"DOT" * v"triple",
    nickname    = (v"LETTER" + v"SPECIAL") * (v"LETTER" + v"DIGIT" + v"SPECIAL" + v"DASH")^-8,
    user        = lpeg.R("\001\009", "\011\012", "\014\031", "\033\063", "\065\255")^1,

    CRLF        = lpeg.P("\r\n"),
    DASH        = lpeg.P("-"),
    USCORE      = lpeg.P("_"),
    DOT         = lpeg.P("."),
    SPACE       = lpeg.P(" "),
    COLON       = lpeg.P(":"),
    BANG        = lpeg.P("!"),
    AT          = lpeg.P("@"),
    LETTER      = lpeg.R("az", "AZ"),
    DIGIT       = lpeg.R("09"),
    SPECIAL     = lpeg.S(";[]\\`_^{|}"),
}

parser = lpeg.P(grammar)

local function test(entry, subject, expected)
    subject = subject
    local nt = {}
    for k, v in pairs(grammar) do
        nt[k] = v
    end
    nt[1] = entry
    local pattern, err = pcall(lpeg.P, nt)
    if not pattern then
        error("Failed to compile test for entry: " .. entry .. "(" .. tostring(err) .. ")")
    end
    local match = lpeg.match(err, subject)
    assert(match, "Failed test for " .. entry .. " with '" .. subject .. "'")
    if match ~= expected then
        error("Failed a complete match for " .. entry .. " got " .. match .. " for " .. #subject .. " string " ..  subject)
    end
    io.stdout:write(".")
end

local function fail(entry, subject)
    local succ,err = pcall(test, entry, subject)
    assert(not succ, "Did not fail when expected: " .. entry .. " with " .. subject)
end

do
    return
end

io.stdout:write("testing")
test("CRLF", "\r\n", 3)
test("COLON", ":", 2)
test("BANG", "!", 2)
test("AT", "@", 2)
test("LETTER", "Monkey", 2)
test("DIGIT", 31337, 2)
test("prefix", "MoNkEyS ", 9)
test("command", "PRIVMSG", 8)
test("nickname", "Cladhaire", 10)
test("nickname", "{clad|foo}", 10)
test("hostname", "wowforge.com", 13)
test("host", "wowforge.com", 13)
test("hostaddr", "128.256.232.122", 16)
test("command", "PING", 5)
test("params", " :anthony.freenode.net", 23)
test("message", "PING :anthony.freenode.net\r\n", 29)
test("nickname", "Aelo", 5)
test("user", "~Aeolbin", 9)
test("host", "81.170.1.150", 13)
test("prefix", "anthony.freenode.net ", 22)
test("prefix", "Aelo!~Aelobin@81.170.1.150 ", 28)
test("COLON", ":", 2)
test("SPACE", " ", 2)
test("nickname", "Aelo", 5)
test("BANG", "!", 2)
test("user", "~Aelobin", 9)
test("AT", "@", 2)
test("host", "81.170.1.150", 13)
test("command", "PRIVMSG", 8)
test("middle", "#wowui", 7)
test("params", " #wowui :rawr", 14)
test("message", ":Aelo!~Aelobin@81.170.1.150 PRIVMSG #wowui :rawr\r\n", 51)

-- Sanity
fail("LETTER", "1")
print(" complete!")