new/xml.lua

   1 #!/usr/bin/lua
   2
   3 local parseargs, collect
   4
   5 local escapes = {
   6         gt = '>',
   7         lt = '<',
   8 }
   9
  10 local strip_escapes = function (s)
  11         s = string.gsub(s, '&(%a+);', escapes)
  12         return s
  13 end
  14
  15
  16 function parseargs(s)
  17         local arg = {}
  18         string.gsub(s, "([%w_]+)=([\"'])(.-)%2", function (w, _, a)
  19                 arg[strip_escapes(w)] = strip_escapes(a)
  20         end)
  21         return arg
  22 end
  23
  24 function collect(s)
  25         local stack = {}
  26         local index = {}
  27         local top = {}
  28         table.insert(stack, top)
  29         local ni,c,label,xarg, empty
  30         local i, j = 1, 1
  31         while true do
  32                 ni,j,c,label,xarg, empty = string.find(s, "<(%/?)(%w+)(.-)(%/?)>", j)
  33                 if not ni then break end
  34                 local text = string.sub(s, i, ni-1)
  35                 if not string.find(text, "^%s*$") then
  36                         table.insert(top, strip_escapes(text))
  37                 end
  38                 if empty == "/" then  -- empty element tag
  39                         table.insert(top, {label=label, xarg=parseargs(xarg), empty=1})
  40                 elseif c == "" then   -- start tag
  41                         top = {label=label, xarg=parseargs(xarg)}
  42                         table.insert(stack, top)   -- new level
  43                 else  -- end tag
  44                         local toclose = table.remove(stack)  -- remove top
  45                         top = stack[#stack]
  46                         if #stack < 1 then
  47                                 error("nothing to close with "..label)
  48                         end
  49                         if toclose.label ~= label then
  50                                 error("trying to close "..toclose.label.." with "..label)
  51                         end
  52                         table.insert(top, toclose)
  53                         toclose.parent = top
  54                         toclose.index = #top
  55                         index[toclose] = true
  56                 end
  57                 i = j+1
  58         end
  59         local text = string.sub(s, i)
  60         if not string.find(text, "^%s*$") then
  61                 table.insert(stack[#stack], text)
  62         end
  63         if #stack > 1 then
  64                 error("unclosed "..stack[stack.n].label)
  65         end
  66         return stack[1], index
  67 end
  68
  69 return collect