new/xml.lua

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