src/parser/xss/parser.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
  19 $ template make_parser()
  20
  21 # State of the parser automata as stored in the parser stack.
  22 private class State
  23         # The internal state number
  24         readable writable var _state: Int
  25
  26         # The node stored with the state in the stack
  27         readable writable var _nodes: nullable Object
  28
  29         init(state: Int, nodes: nullable Object)
  30         do
  31                 _state = state
  32                 _nodes = nodes
  33         end
  34 end
  35
  36 class Parser
  37 special ParserTable
  38         # Associated lexer
  39         var _lexer: Lexer
  40
  41         # Stack of pushed states and productions
  42         var _stack: Array[State]
  43
  44         # Position in the stack
  45         var _stack_pos: Int
  46
  47         # Create a new parser based on a given lexer
  48         init(lexer: Lexer)
  49         do
  50                 _lexer = lexer
  51                 _stack = new Array[State]
  52                 _stack_pos = -1
  53                 build_goto_table
  54                 build_action_table
  55                 build_reduce_table
  56         end
  57
  58         # Do a transition in the automata
  59         private fun go_to(index: Int): Int
  60         do
  61                 var state = state
  62                 var table = _goto_table[index]
  63                 var low = 1
  64                 var high = table.length/2 - 1
  65
  66                 while low <= high do
  67                         var middle = (low + high) / 2
  68                         var subindex = middle * 2
  69
  70                         if state < table[subindex] then
  71                                 high = middle - 1
  72                         else if state > table[subindex] then
  73                                 low = middle + 1
  74                         else
  75                                 return table[subindex + 1]
  76                         end
  77                 end
  78
  79                 return table[1] # Default value
  80         end
  81
  82         # Push someting in the state stack
  83         private fun push(numstate: Int, list_node: nullable Object)
  84         do
  85                 var pos = _stack_pos + 1
  86                 _stack_pos = pos
  87                 if pos < _stack.length then
  88                         var state = _stack[pos]
  89                         state.state = numstate
  90                         state.nodes = list_node
  91                 else
  92                         _stack.push(new State(numstate, list_node))
  93                 end
  94         end
  95
  96         # The current state
  97         private fun state: Int
  98         do
  99                 return _stack[_stack_pos].state
 100         end
 101
 102         # Pop something from the stack state
 103         private fun pop: nullable Object
 104         do
 105                 var res = _stack[_stack_pos].nodes
 106                 _stack_pos = _stack_pos -1
 107                 return res
 108         end
 109
 110         # Build and return a full AST.
 111         fun parse: Start
 112         do
 113                 push(0, null)
 114
 115                 var lexer = _lexer
 116                 while true do
 117                         var token = lexer.peek
 118                         var last_pos = token.location.column_start
 119                         var last_line = token.location.line_start
 120
 121                         if token isa PError then
 122                                 return new Start(null, token)
 123                         end
 124
 125                         var index = token.parser_index
 126                         var table = _action_table[state]
 127                         var action_type = table[1]
 128                         var action_value = table[2]
 129
 130                         var low = 1
 131                         var high = table.length/3 - 1
 132
 133                         while low <= high do
 134                                 var middle = (low + high) / 2
 135                                 var subindex = middle * 3
 136
 137                                 if index < table[subindex] then
 138                                         high = middle - 1
 139                                 else if index > table[subindex] then
 140                                         low = middle + 1
 141                                 else
 142                                         action_type = table[subindex + 1]
 143                                         action_value = table[subindex + 2]
 144                                         high = low -1 # break
 145                                 end
 146                         end
 147
 148                         if action_type == 0 then # SHIFT
 149                                 push(action_value, lexer.next)
 150                         else if action_type == 1 then # REDUCE
 151                                 _reduce_table[action_value].action(self)
 152                         else if action_type == 2 then # ACCEPT
 153                                 var node2 = lexer.next
 154                                 assert node2 isa EOF
 155                                 var node1 = pop
 156                                 assert node1 isa ${/parser/prods/prod/@ename}
 157                                 var node = new Start(node1, node2)
 158                                 (new SearchTokensVisitor).enter_visit(node)
 159                                 return node
 160                         else if action_type == 3 then # ERROR
 161                                 var location = new Location(lexer.filename, last_line, last_line, last_pos, last_pos)
 162                                 var node2 = new PError.init_error(error_messages[errors[action_value]],location)
 163                                 var node = new Start(null, node2)
 164                                 return node
 165                         end
 166                 end
 167                 abort
 168         end
 169
 170         var _reduce_table: Array[ReduceAction]
 171         private fun build_reduce_table
 172         do
 173                 _reduce_table = new Array[ReduceAction].with_items(
 174 $ foreach {rules/rule}
 175                         new ReduceAction@index[-sep ','-]
 176 $ end foreach
 177                 )
 178         end
 179 end
 180
 181 # Find first and last tokens of production nodes
 182 private class SearchTokensVisitor
 183 special Visitor
 184         var _untokenned_nodes: Array[Prod]
 185         var _last_token: nullable Token = null
 186         redef fun visit(n: nullable PNode)
 187         do
 188                 if n == null then
 189                         return
 190                 else if n isa Token then
 191                         _last_token = n
 192                         for no in _untokenned_nodes do
 193                                 no.first_token = n
 194                         end
 195                         _untokenned_nodes.clear
 196                 else
 197                         assert n isa Prod
 198                         _untokenned_nodes.add(n)
 199                         n.visit_all(self)
 200                         n.last_token = _last_token
 201
 202                         if n.first_token != null then
 203                                 var start_location = n.first_token.location
 204                                 var end_location = _last_token.location
 205
 206                                 if start_location != null and end_location != null then
 207                                         var file = end_location.file
 208                                         var line_start = start_location.line_start
 209                                         var line_end = end_location.line_end
 210                                         var column_start = start_location.column_start
 211                                         var column_end = end_location.column_end
 212                                         n.location = new Location(file, line_start, line_end, column_start, column_end)
 213                                 end
 214                         end
 215                 end
 216         end
 217         init
 218         do
 219                 _untokenned_nodes = new Array[Prod]
 220         end
 221 end
 222
 223 # Each reduca action has its own class, this one is the root of the hierarchy.
 224 private abstract class ReduceAction
 225         fun action(p: Parser) is abstract
 226 end
 227
 228 $ foreach {rules/rule}
 229 private class ReduceAction@index
 230 special ReduceAction
 231         redef fun action(p: Parser)
 232         do
 233                                         var node_list: nullable Object = null
 234 $   foreach {action}
 235 $   choose
 236 $     when {@cmd='POP'}
 237                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 238 $     end
 239 $     when {@cmd='FETCHLIST'}
 240                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 241                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 242 $     end
 243 $     when {@cmd='FETCHNODE'}
 244                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 245                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa nullable @etype
 246 $     end
 247 $     when {@cmd='ADDNODE'}
 248                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 249                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 250                                         end
 251 $     end
 252 $     when {@cmd='ADDLIST'}
 253 #                                       if ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 254                                                 if ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.is_empty then
 255                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 256                                                 else
 257                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.append(${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 258                                                 end
 259 #                                       end
 260 $     end
 261 $     when {@cmd='MAKELIST'}
 262                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 263 $     end
 264 $     when {@cmd='MAKENODE'}
 265                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 266 $       foreach {arg}
 267 $           if @null
 268                                                 null[-sep ','-]
 269 $           else
 270                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 271 $           end
 272 $       end foreach
 273                                         )
 274 $     end
 275 $     when {@cmd='RETURNNODE'}
 276 $       if @null
 277                                         node_list = null
 278 $       else
 279                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 280 $       end
 281 $     end
 282 $     when {@cmd='RETURNLIST'}
 283                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 284 $     end
 285 $   end choose
 286 $   end foreach
 287                                         p.push(p.go_to(@leftside), node_list)
 288         end
 289 init do end
 290 end
 291 $ end foreach
 292 $ end template
 293
 294 $ template make_parser_tables()
 295 # Parser that build a full AST
 296 abstract class ParserTable
 297         var _action_table: Array[Array[Int]]
 298         private fun build_action_table
 299         do
 300                 _action_table = once [
 301 $ foreach {parser_data/action_table/row}
 302                         action_table_row${position()}[-sep ','-]
 303 $ end foreach
 304                 ]
 305         end
 306
 307 $ foreach {parser_data/action_table/row}
 308         private fun action_table_row${position()}: Array[Int]
 309         do
 310                 return [
 311 $   foreach {action}
 312                                 @from, @action, @to[-sep ','-]
 313 $   end foreach
 314                         ]
 315         end
 316 $ end foreach
 317
 318         var _goto_table: Array[Array[Int]]
 319         private fun build_goto_table
 320         do
 321                 _goto_table = once [
 322 $ foreach {parser_data/goto_table/row}
 323                         [
 324 $   foreach {goto}
 325                                 @from, @to[-sep ','-]
 326 $   end foreach
 327                         ][-sep ','-]
 328 $ end foreach
 329                 ]
 330         end
 331
 332         private fun error_messages: Array[String]
 333         do
 334                 return once [
 335 $ foreach {parser_data/error_messages/msg}
 336                         "${sablecc:string2escaped_unicode(.)}"[-sep ','-]
 337 $ end
 338                 ]
 339         end
 340
 341         private fun errors: Array[Int]
 342         do
 343                 return once [
 344                         [-foreach {parser_data/errors/i}-]${.}[-sep ','-][-end-]
 345                 ]
 346         end
 347
 348         init do end
 349 end
 350 $ end template