find equal ref (from, to) in members of relation

relations.lua

#!/usr/bin/env lua

--
-- find equal ref (from, to) in members of relation
--
-- @author  leite <[email protected]>
-- @license GPL version 3
--

assert(arg[1], 'argument should be a file path')

local io, string, table    = require 'io', require 'string', require 'table'
local i, last, lines, dump = 1, 1, {}, ''
local handler              = io.open(arg[1], 'r')

function parse_chunk (chunk)
  local _last

  while true do
    local index, offset, context = string.find(chunk, '<relation(.-)</relation>', _last)

    if not index then
      break
    end

    _last = offset

    local index, __, id, ref0, role0, ref1, role1 = string.find(
      context, " id='([^']+)' .- ref='([^']+)' role='([fromt]+)'.- ref='([^']+)' role='([fromt]+)'"
    )

    if index then
      id, ref0, ref1 = tonumber(id), tonumber(ref0), tonumber(ref1)

      if ref0 == ref1 then
        io.write(string.format("relation %d duplicated\n", id))
      end
    end
  end

  return _last or #chunk - 500
end



for line in handler:lines() do
  table.insert(lines, line)
  if i > 500 then
    dump           = string.sub(dump, last) .. table.concat(lines, '')
    i, lines, last = 1, {}, parse_chunk(dump)
    collectgarbage()
  end
  i = i + 1
end

handler:close()

parse_chunk(string.sub(dump, last) .. table.concat(lines, ''))

luajit profiler ...

real 0m0.592s
user 0m0.584s
sys 0m0.008s

lua -jp=a relations.lua /home/leite/Downloads/caray.osm
[...]
====== relations.lua ======
@@ 19 @@
      |   while true do
      |     local index, offset, context = string.find(chunk, '<relation(.-)</relation>', _last)
      | 
  59% |     if not index then
      |       break
      |     end
      | 
@@ 29 @@
      |       context, " id='([^']+)' .- ref='([^']+)' role='([fromt]+)'.- ref='([^']+)' role='([fromt]+)'"
      |     )
      | 
  10% |     if index then
      |       id, ref0, ref1 = tonumber(id), tonumber(ref0), tonumber(ref1)
      | 
      |       if ref0 == ref1 then
@@ 43 @@
      | 
      | 
      | 
  17% | for line in handler:lines() do
      |   table.insert(lines, line)
      |   if i > 500 then
      |     dump           = string.sub(dump, last) .. table.concat(lines, '')
      |     i, lines, last = 1, {}, parse_chunk(dump)
      |     collectgarbage()
      |   end
   8% |   i = i + 1
      | end
      | 
      | handler:close()