Parsing the Link HTTP header with lpeg

link_header.lua

local lpeg = require 'lpeg'
local P, C, R, S, Cs, Cc, Ct, Cf, Cg, V, Cmt = lpeg.P, lpeg.C, lpeg.R, lpeg.S, lpeg.Cs, lpeg.Cc, lpeg.Ct, lpeg.Cf, lpeg.Cg, lpeg.V, lpeg.Cmt
local lpegmatch, lpegpatterns, replacer = lpeg.match, lpeg.patterns, lpeg.replacer
local inspect = require 'inspect'

local link_header = '<https://api.github.com/organizations/XXXXXXX/members?per_page=10&page=2>; rel="next", <https://api.github.com/organizations/XXXXXX/members?per_page=10&page=6>; rel="last"'

-- This is our placeholder for matches
local links = links or {}

-- As we get a table match, we add a new entry to the links table in the format of:
-- { reltype = "target" }
local function build_pagination(t1)
  links[t1.rel] = t1.url
  return t1
end

-- link headers look like so:
-- <someurl>; rel="sometype"
-- multiple links can be joined together with commas:
-- <someurl1>; rel="foo",<someurl2>; rel="bar"
--
local openbracket = P("<")
local closebracket = P(">")
local semicolon = P(";")
local equal = P("=")
local comma = P(",")
local quote = P('"')
local endofstring = P(-1)
local nothing = Cc("")
local whitespace = P(" ")^1
-- target is the url in in between <>
local target = Cg(((P(1) - closebracket)^-0), "url")
-- reltype is the value for rel
local reltype = Cg(((P(1) - quote)^-0), "rel")
-- Make a capture table of one match
-- { rel = "next", url = "https://....." }
local match_one = Ct((whitespace)^0 * openbracket * target * closebracket * semicolon * (whitespace)^0 * "rel=" * quote * reltype * quote * (whitespace)^0 * (comma)^0)
-- apply function to one match
local linkwfunc = (match_one / build_pagination)
-- make a capture table of one or more matches
local match_any = Ct((linkwfunc)^1)
-- do the match
match_links = lpegmatch(match_any, link_header)
print(inspect(links))
-- returns
-- {
--   last = "https://api.github.com/organizations/XXXXXX/members?per_page=10&page=6",
--   next = "https://api.github.com/organizations/XXXXXXX/members?per_page=10&page=2"
-- }