fallbacks/lxp.lua

   1
   2 local coroutine = coroutine;
   3 local tonumber = tonumber;
   4 local string = string;
   5 local setmetatable, getmetatable = setmetatable, getmetatable;
   6 local pairs = pairs;
   7
   8 local deadroutine = coroutine.create(function() end);
   9 coroutine.resume(deadroutine);
  10
  11 module("lxp")
  12
  13 local entity_map = setmetatable({
  14         ["amp"] = "&";
  15         ["gt"] = ">";
  16         ["lt"] = "<";
  17         ["apos"] = "'";
  18         ["quot"] = "\"";
  19 }, {__index = function(_, s)
  20                 if s:sub(1,1) == "#" then
  21                         if s:sub(2,2) == "x" then
  22                                 return string.char(tonumber(s:sub(3), 16));
  23                         else
  24                                 return string.char(tonumber(s:sub(2)));
  25                         end
  26                 end
  27         end
  28 });
  29 local function xml_unescape(str)
  30         return (str:gsub("&(.-);", entity_map));
  31 end
  32 local function parse_tag(s)
  33         local name,sattr=(s):gmatch("([^%s]+)(.*)")();
  34         local attr = {};
  35         for a,b in (sattr):gmatch("([^=%s]+)=['\"]([^'\"]*)['\"]") do attr[a] = xml_unescape(b); end
  36         return name, attr;
  37 end
  38
  39 local function parser(data, handlers, ns_separator)
  40         local function read_until(str)
  41                 local pos = data:find(str, nil, true);
  42                 while not pos do
  43                         data = data..coroutine.yield();
  44                         pos = data:find(str, nil, true);
  45                 end
  46                 local r = data:sub(1, pos);
  47                 data = data:sub(pos+1);
  48                 return r;
  49         end
  50         local function read_before(str)
  51                 local pos = data:find(str, nil, true);
  52                 while not pos do
  53                         data = data..coroutine.yield();
  54                         pos = data:find(str, nil, true);
  55                 end
  56                 local r = data:sub(1, pos-1);
  57                 data = data:sub(pos);
  58                 return r;
  59         end
  60         local function peek()
  61                 while #data == 0 do data = coroutine.yield(); end
  62                 return data:sub(1,1);
  63         end
  64
  65         local ns = { xml = "http://www.w3.org/XML/1998/namespace" };
  66         ns.__index = ns;
  67         local function apply_ns(name, dodefault)
  68                 local prefix,n = name:match("^([^:]*):(.*)$");
  69                 if prefix and ns[prefix] then
  70                         return ns[prefix]..ns_separator..n;
  71                 end
  72                 if dodefault and ns[""] then
  73                         return ns[""]..ns_separator..name;
  74                 end
  75                 return name;
  76         end
  77         local function push(tag, attr)
  78                 ns = setmetatable({}, ns);
  79                 for k,v in pairs(attr) do
  80                         local xmlns = k == "xmlns" and "" or k:match("^xmlns:(.*)$");
  81                         if xmlns then
  82                                 ns[xmlns] = v;
  83                                 attr[k] = nil;
  84                         end
  85                 end
  86                 local newattr, n = {}, 0;
  87                 for k,v in pairs(attr) do
  88                         n = n+1;
  89                         k = apply_ns(k);
  90                         newattr[n] = k;
  91                         newattr[k] = v;
  92                 end
  93                 tag = apply_ns(tag, true);
  94                 ns[0] = tag;
  95                 ns.__index = ns;
  96                 return tag, newattr;
  97         end
  98         local function pop()
  99                 local tag = ns[0];
 100                 ns = getmetatable(ns);
 101                 return tag;
 102         end
 103
 104         while true do
 105                 if peek() == "<" then
 106                         local elem = read_until(">"):sub(2,-2);
 107                         if elem:sub(1,1) == "!" or elem:sub(1,1) == "?" then -- neglect comments and processing-instructions
 108                         elseif elem:sub(1,1) == "/" then -- end tag
 109                                 elem = elem:sub(2);
 110                                 local name = pop();
 111                                 handlers:EndElement(name); -- TODO check for start-end tag name match
 112                         elseif elem:sub(-1,-1) == "/" then -- empty tag
 113                                 elem = elem:sub(1,-2);
 114                                 local name,attr = parse_tag(elem);
 115                                 name,attr = push(name,attr);
 116                                 handlers:StartElement(name,attr);
 117                                 name = pop();
 118                                 handlers:EndElement(name);
 119                         else -- start tag
 120                                 local name,attr = parse_tag(elem);
 121                                 name,attr = push(name,attr);
 122                                 handlers:StartElement(name,attr);
 123                         end
 124                 else
 125                         local text = read_before("<");
 126                         handlers:CharacterData(xml_unescape(text));
 127                 end
 128         end
 129 end
 130
 131 function new(handlers, ns_separator)
 132         local co = coroutine.create(parser);
 133         return {
 134                 parse = function(self, data)
 135                         if not data then
 136                                 co = deadroutine;
 137                                 return true; -- eof
 138                         end
 139                         local success, result = coroutine.resume(co, data, handlers, ns_separator);
 140                         if result then
 141                                 co = deadroutine;
 142                                 return nil, result; -- error
 143                         end
 144                         return true; -- success
 145                 end;
 146         };
 147 end
 148
 149 return _M;