tools/erlparse.lua

   1 -- Prosody IM
   2 -- Copyright (C) 2008-2010 Matthew Wild
   3 -- Copyright (C) 2008-2010 Waqas Hussain
   4 --
   5 -- This project is MIT/X11 licensed. Please see the
   6 -- COPYING file in the source package for more information.
   7 --
   8
   9 local string_byte, string_char = string.byte, string.char;
  10 local t_concat, t_insert = table.concat, table.insert;
  11 local type, tonumber, tostring = type, tonumber, tostring;
  12
  13 local file = nil;
  14 local last = nil;
  15 local line = 1;
  16 local function read(expected)
  17         local ch;
  18         if last then
  19                 ch = last; last = nil;
  20         else
  21                 ch = file:read(1);
  22                 if ch == "\n" then line = line + 1; end
  23         end
  24         if expected and ch ~= expected then error("expected: "..expected.."; got: "..(ch or "nil").." on line "..line); end
  25         return ch;
  26 end
  27 local function pushback(ch)
  28         if last then error(); end
  29         last = ch;
  30 end
  31 local function peek()
  32         if not last then last = read(); end
  33         return last;
  34 end
  35
  36 local _A, _a, _Z, _z, _0, _9, __, _at, _space, _minus = string_byte("AaZz09@_ -", 1, 10);
  37 local function isLowerAlpha(ch)
  38         ch = string_byte(ch) or 0;
  39         return (ch >= _a and ch <= _z);
  40 end
  41 local function isNumeric(ch)
  42         ch = string_byte(ch) or 0;
  43         return (ch >= _0 and ch <= _9) or ch == _minus;
  44 end
  45 local function isAtom(ch)
  46         ch = string_byte(ch) or 0;
  47         return (ch >= _A and ch <= _Z) or (ch >= _a and ch <= _z) or (ch >= _0 and ch <= _9) or ch == __ or ch == _at;
  48 end
  49 local function isSpace(ch)
  50         ch = string_byte(ch) or "x";
  51         return ch <= _space;
  52 end
  53
  54 local escapes = {["\\b"]="\b", ["\\d"]="\127", ["\\e"]="\27", ["\\f"]="\f", ["\\n"]="\n", ["\\r"]="\r", ["\\s"]=" ", ["\\t"]="\t", ["\\v"]="\v", ["\\\""]="\"", ["\\'"]="'", ["\\\\"]="\\"};
  55 local function readString()
  56         read("\""); -- skip quote
  57         local slash = nil;
  58         local str = {};
  59         while true do
  60                 local ch = read();
  61                 if slash then
  62                         slash = slash..ch;
  63                         if not escapes[slash] then error("Unknown escape sequence: "..slash); end
  64                         str[#str+1] = escapes[slash];
  65                         slash = nil;
  66                 elseif ch == "\"" then
  67                         break;
  68                 elseif ch == "\\" then
  69                         slash = ch;
  70                 else
  71                         str[#str+1] = ch;
  72                 end
  73         end
  74         return t_concat(str);
  75 end
  76 local function readAtom1()
  77         local var = { read() };
  78         while isAtom(peek()) do
  79                 var[#var+1] = read();
  80         end
  81         return t_concat(var);
  82 end
  83 local function readAtom2()
  84         local str = { read("'") };
  85         local slash = nil;
  86         while true do
  87                 local ch = read();
  88                 str[#str+1] = ch;
  89                 if ch == "'" and not slash then break; end
  90         end
  91         return t_concat(str);
  92 end
  93 local function readNumber()
  94         local num = { read() };
  95         while isNumeric(peek()) do
  96                 num[#num+1] = read();
  97         end
  98         if peek() == "." then
  99                 num[#num+1] = read();
 100                 while isNumeric(peek()) do
 101                         num[#num+1] = read();
 102                 end
 103         end
 104         return tonumber(t_concat(num));
 105 end
 106 local readItem = nil;
 107 local function readTuple()
 108         local t = {};
 109         local s = {}; -- string representation
 110         read(); -- read {, or [, or <
 111         while true do
 112                 local item = readItem();
 113                 if not item then break; end
 114                 if type(item) ~= "number" or item > 255 then
 115                         s = nil;
 116                 elseif s then
 117                         s[#s+1] = string_char(item);
 118                 end
 119                 t_insert(t, item);
 120         end
 121         read(); -- read }, or ], or >
 122         if s and #s > 0  then
 123                 return t_concat(s)
 124         else
 125                 return t
 126         end;
 127 end
 128 local function readBinary()
 129         read("<"); -- read <
 130         -- Discard PIDs
 131         if isNumeric(peek()) then
 132                 while peek() ~= ">" do read(); end
 133                 read(">");
 134                 return {};
 135         end
 136         local t = readTuple();
 137         read(">") -- read >
 138         local ch = peek();
 139         if type(t) == "string" then
 140                 -- binary is a list of integers
 141                 return t;
 142         elseif type(t) == "table" then
 143                 if t[1] then
 144                         -- binary contains string
 145                         return t[1];
 146                 else
 147                         -- binary is empty
 148                         return "";
 149                 end;
 150         else
 151                 error();
 152         end
 153 end
 154 readItem = function()
 155         local ch = peek();
 156         if ch == nil then return nil end
 157         if ch == "{" or ch == "[" then
 158                 return readTuple();
 159         elseif isLowerAlpha(ch) then
 160                 return readAtom1();
 161         elseif ch == "'" then
 162                 return readAtom2();
 163         elseif isNumeric(ch) then
 164                 return readNumber();
 165         elseif ch == "\"" then
 166                 return readString();
 167         elseif ch == "<" then
 168                 return readBinary();
 169         elseif isSpace(ch) or ch == "," or ch == "|" then
 170                 read();
 171                 return readItem();
 172         else
 173                 --print("Unknown char: "..ch);
 174                 return nil;
 175         end
 176 end
 177 local function readChunk()
 178         local x = readItem();
 179         if x then read("."); end
 180         return x;
 181 end
 182 local function readFile(filename)
 183         file = io.open(filename);
 184         if not file then error("File not found: "..filename); os.exit(0); end
 185         return function()
 186                 local x = readChunk();
 187                 if not x and peek() then error("Invalid char: "..peek()); end
 188                 return x;
 189         end;
 190 end
 191
 192 local _M = {};
 193
 194 function _M.parseFile(file)
 195         return readFile(file);
 196 end
 197
 198 return _M;