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 attr _state: Int
  25
  26         # The node stored with the state in the stack
  27         readable writable attr _nodes: Object
  28
  29         init(state: Int, nodes: Object)
  30         do
  31                 _state = state
  32                 _nodes = nodes
  33         end
  34 end
  35
  36 class Parser
  37 special ParserTable
  38         # Associated lexer
  39         attr _lexer: Lexer
  40
  41         # Stack of pushed states and productions
  42         attr _stack: Array[State]
  43
  44         # Position in the stack
  45         attr _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 meth 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 meth push(numstate: Int, list_node: 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 meth state: Int
  98         do
  99                 return _stack[_stack_pos].state
 100         end
 101
 102         # Pop something from the stack state
 103         private meth pop: 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         meth parse: Start
 112         do
 113                 push(0, null)
 114
 115                 var ign: List[Token] = null
 116                 var lexer = _lexer
 117                 while true do
 118                         var token = lexer.peek
 119                         var last_pos = token.pos
 120                         var last_line = token.line
 121
 122                         if token isa PError then
 123                                 assert token isa PError
 124                                 return new Start(null, token)
 125                         end
 126
 127                         var index = token.parser_index
 128                         var table = _action_table[state]
 129                         var action_type = table[1]
 130                         var action_value = table[2]
 131
 132                         var low = 1
 133                         var high = table.length/3 - 1
 134
 135                         while low <= high do
 136                                 var middle = (low + high) / 2
 137                                 var subindex = middle * 3
 138
 139                                 if index < table[subindex] then
 140                                         high = middle - 1
 141                                 else if index > table[subindex] then
 142                                         low = middle + 1
 143                                 else
 144                                         action_type = table[subindex + 1]
 145                                         action_value = table[subindex + 2]
 146                                         high = low -1 # break
 147                                 end
 148                         end
 149
 150                         if action_type == 0 then # SHIFT
 151                                 push(action_value, lexer.next)
 152                         else if action_type == 1 then # REDUCE
 153                                 _reduce_table[action_value].action(self)
 154                         else if action_type == 2 then # ACCEPT
 155                                 var node2 = lexer.next
 156                                 assert node2 isa EOF
 157                                 var node1 = pop
 158                                 assert node1 isa ${/parser/prods/prod/@ename}
 159                                 var node = new Start(node1, node2)
 160                                 (new SearchTokensVisitor).visit(node)
 161                                 return node
 162                         else if action_type == 3 then # ERROR
 163                                 var node2 = new PError.init_error(lexer.filename, last_line, last_pos, error_messages[errors[action_value]])
 164                                 var node = new Start(null, node2)
 165                                 return node
 166                         end
 167                 end
 168                 return null
 169         end
 170
 171         attr _reduce_table: Array[ReduceAction]
 172         private meth build_reduce_table
 173         do
 174                 _reduce_table = new Array[ReduceAction].with(
 175 $ foreach {rules/rule}
 176                         new ReduceAction@index[-sep ','-]
 177 $ end foreach
 178                 )
 179         end
 180 end
 181
 182 # Find first and last tokens of production nodes
 183 private class SearchTokensVisitor
 184 special Visitor
 185         attr _untokenned_nodes: Array[Prod]
 186         attr _last_token: Token
 187         redef meth visit(n: PNode)
 188         do
 189                 if n isa Token then
 190                         assert n isa Token
 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                 end
 202         end
 203         init
 204         do
 205                 _untokenned_nodes = new Array[Prod]
 206         end
 207 end
 208
 209 # Each reduca action has its own class, this one is the root of the hierarchy.
 210 private abstract class ReduceAction
 211         meth action(p: Parser) is abstract
 212 end
 213
 214 $ foreach {rules/rule}
 215 private class ReduceAction@index
 216 special ReduceAction
 217         redef meth action(p: Parser)
 218         do
 219                                         var node_list: Object = null
 220 $   foreach {action}
 221 $   choose
 222 $     when {@cmd='POP'}
 223                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 224 $     end
 225 $     when {@cmd='FETCHLIST'}
 226                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 227                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 228 $     end
 229 $     when {@cmd='FETCHNODE'}
 230                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 231                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa @etype
 232 $     end
 233 $     when {@cmd='ADDNODE'}
 234                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 235                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 236                                         end
 237 $     end
 238 $     when {@cmd='ADDLIST'}
 239                                         if ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 240                                                 if ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.is_empty then
 241                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 242                                                 else
 243                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.append(${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 244                                                 end
 245                                         end
 246 $     end
 247 $     when {@cmd='MAKELIST'}
 248                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 249 $     end
 250 $     when {@cmd='MAKENODE'}
 251                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 252 $       foreach {arg}
 253 $           if @null
 254                                                 null[-sep ','-]
 255 $           else
 256                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 257 $           end
 258 $       end foreach
 259                                         )
 260 $     end
 261 $     when {@cmd='RETURNNODE'}
 262 $       if @null
 263                                         node_list = null
 264 $       else
 265                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 266 $       end
 267 $     end
 268 $     when {@cmd='RETURNLIST'}
 269                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 270 $     end
 271 $   end choose
 272 $   end foreach
 273                                         p.push(p.go_to(@leftside), node_list)
 274         end
 275 init do end
 276 end
 277 $ end foreach
 278 $ end template
 279
 280 $ template make_parser_tables()
 281 # Parser that build a full AST
 282 abstract class ParserTable
 283         attr _action_table: Array[Array[Int]] = null
 284         private meth build_action_table
 285         do
 286                 _action_table = once [
 287 $ foreach {parser_data/action_table/row}
 288                         action_table_row${position()}[-sep ','-]
 289 $ end foreach
 290                 ]
 291         end
 292
 293 $ foreach {parser_data/action_table/row}
 294         private meth action_table_row${position()}: Array[Int]
 295         do
 296                 return [
 297 $   foreach {action}
 298                                 @from, @action, @to [-sep ','-]
 299 $   end foreach
 300                         ]
 301         end
 302 $ end foreach
 303
 304         attr _goto_table: Array[Array[Int]] = null
 305         private meth build_goto_table
 306         do
 307                 _goto_table = once [
 308 $ foreach {parser_data/goto_table/row}
 309                         [
 310 $   foreach {goto}
 311                                 @from, @to [-sep ','-]
 312 $   end foreach
 313                         ] [-sep ','-]
 314 $ end foreach
 315                 ]
 316         end
 317
 318         private meth error_messages: Array[String]
 319         do
 320                 return once [
 321 $ foreach {parser_data/error_messages/msg}
 322                         "${sablecc:string2escaped_unicode(.)}" [-sep ','-]
 323 $ end
 324                 ]
 325         end
 326
 327         private meth errors: Array[Int]
 328         do
 329                 return once [
 330                         [-foreach {parser_data/errors/i}-]${.} [-sep ','-] [-end-]
 331                 ]
 332         end
 333 end
 334 $ end template