julia/reader.jl

   1 module reader
   2
   3 export read_str
   4
   5 import types
   6
   7 type Reader
   8     tokens
   9     position::Int64
  10 end
  11
  12 function next(rdr::Reader)
  13     if rdr.position > length(rdr.tokens)
  14         return nothing
  15     end
  16     rdr.position += 1
  17     rdr.tokens[rdr.position-1]
  18 end
  19
  20 function peek(rdr::Reader)
  21     if rdr.position > length(rdr.tokens)
  22         return nothing
  23     end
  24     rdr.tokens[rdr.position]
  25 end
  26
  27
  28 function tokenize(str)
  29     re = r"[\s,]*(~@|[\[\]{}()'`~^@]|\"(?:\\.|[^\\\"])*\"|;.*|[^\s\[\]{}('\"`,;)]*)"
  30     tokens = map((m) -> m.captures[1], eachmatch(re, str))
  31     filter((t) -> t != "" && t[1] != ';', tokens)
  32 end
  33
  34 function read_atom(rdr)
  35     token = next(rdr)
  36     if ismatch(r"^-?[0-9]+$", token)
  37         parse(Int,token)
  38     elseif ismatch(r"^-?[0-9][0-9.]*$", token)
  39         float(token)
  40     elseif ismatch(r"^\".*\"$", token)
  41         replace(
  42             replace(
  43                 replace(token[2:end-1],
  44                         "\\\"", "\""),
  45                 "\\n", "\n"),
  46             "\\\\", "\\")
  47     elseif token[1] == ':'
  48         "\u029e$(token[2:end])"
  49     elseif token == "nil"
  50         nothing
  51     elseif token == "true"
  52         true
  53     elseif token == "false"
  54         false
  55     else
  56         symbol(token)
  57     end
  58 end
  59
  60 function read_list(rdr, start="(", last=")")
  61     ast = Any[]
  62     token = next(rdr)
  63     if (token != start)
  64         error("expected '$(start)'")
  65     end
  66     while ((token = peek(rdr)) != last)
  67         if token == nothing
  68             error("expected '$(last)', got EOF")
  69         end
  70         push!(ast, read_form(rdr))
  71     end
  72     next(rdr)
  73     ast
  74 end
  75
  76 function read_vector(rdr)
  77     lst = read_list(rdr, "[", "]")
  78     tuple(lst...)
  79 end
  80
  81 function read_hash_map(rdr)
  82     lst = read_list(rdr, "{", "}")
  83     types.hash_map(lst...)
  84 end
  85
  86 function read_form(rdr)
  87     token = peek(rdr)
  88     if token == "'"
  89         next(rdr)
  90         [[:quote]; Any[read_form(rdr)]]
  91     elseif token == "`"
  92         next(rdr)
  93         [[:quasiquote]; Any[read_form(rdr)]]
  94     elseif token == "~"
  95         next(rdr)
  96         [[:unquote]; Any[read_form(rdr)]]
  97     elseif token == "~@"
  98         next(rdr)
  99         [[symbol("splice-unquote")]; Any[read_form(rdr)]]
 100     elseif token == "^"
 101         next(rdr)
 102         meta = read_form(rdr)
 103         [[symbol("with-meta")]; Any[read_form(rdr)]; Any[meta]]
 104     elseif token == "@"
 105         next(rdr)
 106         [[symbol("deref")]; Any[read_form(rdr)]]
 107
 108     elseif token == ")"
 109         error("unexpected ')'")
 110     elseif token == "("
 111         read_list(rdr)
 112     elseif token == "]"
 113         error("unexpected ']'")
 114     elseif token == "["
 115         read_vector(rdr)
 116     elseif token == "}"
 117         error("unexpected '}'")
 118     elseif token == "{"
 119         read_hash_map(rdr)
 120     else
 121         read_atom(rdr)
 122     end
 123 end
 124
 125 function read_str(str)
 126     tokens = tokenize(str)
 127     if length(tokens) == 0
 128         return nothing
 129     end
 130     read_form(Reader(tokens, 1))
 131 end
 132
 133 end