src/parser/xss/lexer.xss

   1 $ // This file is part of NIT ( http://www.nitlanguage.org ).
   2 $ //
   3 $ // Copyright 2008 Jean Privat <jean@pryen.org>
   4 $ // Based on algorithms developped for ( http://www.sablecc.org/ ).
   5 $ //
   6 $ // Licensed under the Apache License, Version 2.0 (the "License");
   7 $ // you may not use this file except in compliance with the License.
   8 $ // You may obtain a copy of the License at
   9 $ //
  10 $ //     http://www.apache.org/licenses/LICENSE-2.0
  11 $ //
  12 $ // Unless required by applicable law or agreed to in writing, software
  13 $ // distributed under the License is distributed on an "AS IS" BASIS,
  14 $ // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  15 $ // See the License for the specific language governing permissions and
  16 $ // limitations under the License.
  17
  18 $ template make_lexer()
  19
  20 # The lexer extract NIT tokens from an input stream.
  21 # It is better user with the Parser
  22 class Lexer
  23         # Last peeked token
  24         var _token: nullable Token
  25
  26         # Lexer current state
  27         var _state: Int = 0
  28
  29         # Name of the stream (as given to tokens)
  30         readable var _filename: String
  31
  32         # Input stream where character are read
  33         var _stream: IStream
  34
  35         # Pushback buffer to store unread character
  36         var _stream_buf: Buffer
  37
  38         # Number of character stored in the pushback buffer
  39         var _stream_pos: Int
  40
  41         # Current line number in the input stream
  42         var _line: Int = 0
  43
  44         # Current column in the input stream
  45         var _pos: Int = 0
  46
  47         # Was the last character a cariage-return?
  48         var _cr: Bool = false
  49
  50         # If the end of stream?
  51         var _eof: Bool = false
  52
  53         # Current working text read from the input stream
  54         var _text: Buffer
  55
  56 $ foreach {lexer_data/state}
  57         # Constante state values
  58         private fun state_${translate(@name,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: Int do return @id end
  59 $ end foreach
  60
  61         # Create a new lexer for a stream (and a name)
  62         init(stream: IStream, fname: String)
  63         do
  64                 _filename = fname
  65                 _text = new Buffer
  66                 _stream = stream
  67                 _stream_pos = -1
  68                 _stream_buf = new Buffer
  69                 build_goto_table
  70                 build_accept_table
  71         end
  72
  73         # Give the next token (but do not consume it)
  74         fun peek: Token
  75         do
  76                 while _token == null do
  77                         _token = get_token
  78                 end
  79                 return _token.as(not null)
  80         end
  81
  82         # Give and consume the next token
  83         fun next: Token
  84         do
  85                 var result = _token
  86                 while result == null do
  87                         result = get_token
  88                 end
  89                 _token = null
  90                 return result.as(not null)
  91         end
  92
  93         # Get a token, or null if it is discarded
  94         private fun get_token: nullable Token
  95         do
  96                 var dfa_state = 0
  97
  98                 var start_pos = _pos
  99                 var start_line = _line
 100
 101                 var accept_state = -1
 102                 var accept_token = -1
 103                 var accept_length = -1
 104                 var accept_pos = -1
 105                 var accept_line = -1
 106
 107                 var goto_table = _goto_table[_state]
 108                 var accept = _accept_table[_state]
 109                 var text = _text
 110                 text.clear
 111
 112                 loop
 113                         var c = get_char
 114
 115                         if c != -1 then
 116                                 var cr = _cr
 117                                 var line = _line
 118                                 var pos = _pos
 119                                 if c == 10 then
 120                                         if cr then
 121                                                 cr = false
 122                                         else
 123                                                 line = line + 1
 124                                                 pos = 0
 125                                         end
 126                                 else if c == 13 then
 127                                         line = line + 1
 128                                         pos = 0
 129                                         cr = true
 130                                 else
 131                                         pos = pos + 1
 132                                         cr = false
 133                                 end
 134
 135                                 text.add(c.ascii)
 136
 137                                 loop
 138                                         var old_state = dfa_state
 139                                         if dfa_state < -1 then
 140                                                 old_state = -2 - dfa_state
 141                                         end
 142
 143                                         dfa_state = -1
 144
 145                                         var tmp0 = goto_table[old_state]
 146                                         var low = 0
 147                                         var high = tmp0.length - 1
 148
 149                                         if high >= 0 then
 150                                                 var tmp1 = tmp0.intern_items
 151                                                 while low <= high do
 152                                                         var middle = (low + high) / 2
 153                                                         var tmp2 = tmp1[middle].intern_items
 154
 155                                                         if c < tmp2[0] then
 156                                                                 high = middle - 1
 157                                                         else if c > tmp2[1] then
 158                                                                 low = middle + 1
 159                                                         else
 160                                                                 dfa_state = tmp2[2]
 161                                                                 break
 162                                                         end
 163                                                 end
 164                                         end
 165                                         if dfa_state > -2 then break
 166                                 end
 167
 168                                 _cr = cr
 169                                 _line = line
 170                                 _pos = pos
 171                         else
 172                                 dfa_state = -1
 173                         end
 174
 175                         if dfa_state >= 0 then
 176                                 if accept[dfa_state] != -1 then
 177                                         accept_state = dfa_state
 178                                         accept_token = accept[dfa_state]
 179                                         accept_length = text.length
 180                                         accept_pos = _pos
 181                                         accept_line = _line
 182                                 end
 183                         else
 184                                 if accept_state != -1 then
 185                                         var location = new Location(_filename, start_line + 1, accept_line + 1, start_pos + 1, accept_pos)
 186                                         _pos = accept_pos
 187                                         _line = accept_line
 188                                         push_back(accept_length)
 189 $ foreach {//token}
 190                                         if accept_token == ${position()-1} then
 191 $    if {count(transition[@from!=@to])!=0}
 192                                                 var state_id = _state
 193 $        foreach transition in {transition[@from!=@to]}
 194                                                 if state_id == ${/parser/lexer_data/state[@name=$transition/@from]/@id} then
 195                                                         _state = state_${translate(@to,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 196                                                 end
 197 $        end
 198 $    end if
 199 $    if {@parser_index}
 200 $        if {not(@text)}
 201                                                 var token_text = text.substring(0, accept_length)
 202                                                 return new @ename.init_tk(token_text, location)
 203 $        else
 204                                                 return new @ename.init_tk(location)
 205 $        end
 206 $    else
 207                                                 return null
 208 $    end
 209                                         end
 210 $ end foreach
 211                                 else
 212                                         var location = new Location(_filename, start_line + 1, start_line + 1, start_pos + 1, start_pos + 1)
 213                                         if text.length > 0 then
 214                                                 var token = new PError.init_error("Syntax error: unknown token {text}.", location)
 215                                                 return token
 216                                         else
 217                                                 var token = new EOF(location)
 218                                                 return token
 219                                         end
 220                                 end
 221                         end
 222                         if false then break # FIXME remove once unreach loop exits are in c_src
 223                 end
 224                 return null # FIXME remove once unreach loop exits are in c_src
 225         end
 226
 227         # Read the next character.
 228         # The character is read from the stream of from the pushback buffer.
 229         private fun get_char: Int
 230         do
 231                 if _eof then
 232                         return -1
 233                 end
 234
 235                 var result: Int
 236
 237                 var sp = _stream_pos
 238                 if sp >= 0 then
 239                         var res = _stream_buf[_stream_pos]
 240                         _stream_pos = sp - 1
 241                         result = res.ascii
 242                 else
 243                         result = _stream.read_char
 244                 end
 245
 246                 if result == -1 then
 247                         _eof = true
 248                 end
 249
 250                 return result
 251         end
 252
 253         # Unread some characters.
 254         # Unread characters are stored in the pushback buffer.
 255         private fun push_back(accept_length: Int)
 256         do
 257                 var length = _text.length
 258                 var i = length - 1
 259                 while i >= accept_length do
 260                         _eof = false
 261                         _stream_pos = _stream_pos + 1
 262                         _stream_buf[_stream_pos] = _text[i]
 263                         i = i - 1
 264                 end
 265         end
 266
 267         var _goto_table: Array[Array[Array[Array[Int]]]]
 268         private fun build_goto_table
 269         do
 270                 _goto_table = once [
 271 $ foreach {lexer_data/goto_table/state}
 272                         [
 273 $     foreach {row}
 274 $         if {count(goto)!=0}
 275                                 [
 276 $             foreach {goto}
 277                                         [@low, @high, @state][-sep ','-]
 278 $             end foreach
 279                                 ][-sep ','-]
 280 $         else
 281                                 nil_array[-sep ','-]
 282 $         end
 283 $     end foreach
 284                         ][-sep ','-]
 285 $ end foreach
 286                 ]
 287         end
 288
 289         private fun nil_array: Array[Array[Int]]
 290         do
 291                 return once new Array[Array[Int]]
 292         end
 293
 294         var _accept_table: Array[Array[Int]]
 295         private fun build_accept_table do
 296                 _accept_table = once [
 297 $ foreach {lexer_data/accept_table/state}
 298                         [
 299                                 [-foreach {i}-]${.}[-sep ','-][-end foreach-]
 300
 301                         ][-sep ','-]
 302 $ end foreach
 303                 ]
 304         end
 305 end
 306
 307 $ end template