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 $ template make_parser()
  19
  20 # State of the parser automata as stored in the parser stack.
  21 private class State
  22         # The internal state number
  23         readable writable var _state: Int
  24
  25         # The node stored with the state in the stack
  26         readable writable var _nodes: nullable Object
  27
  28         init(state: Int, nodes: nullable Object)
  29         do
  30                 _state = state
  31                 _nodes = nodes
  32         end
  33 end
  34
  35 class Parser
  36 special ParserTable
  37         # Associated lexer
  38         var _lexer: Lexer
  39
  40         # Stack of pushed states and productions
  41         var _stack: Array[State]
  42
  43         # Position in the stack
  44         var _stack_pos: Int
  45
  46         # Create a new parser based on a given lexer
  47         init(lexer: Lexer)
  48         do
  49                 _lexer = lexer
  50                 _stack = new Array[State]
  51                 _stack_pos = -1
  52                 build_goto_table
  53                 build_action_table
  54                 build_reduce_table
  55         end
  56
  57         # Do a transition in the automata
  58         private fun go_to(index: Int): Int
  59         do
  60                 var state = state
  61                 var table = _goto_table[index]
  62                 var low = 1
  63                 var high = table.length/2 - 1
  64
  65                 while low <= high do
  66                         var middle = (low + high) / 2
  67                         var subindex = middle * 2
  68
  69                         if state < table[subindex] then
  70                                 high = middle - 1
  71                         else if state > table[subindex] then
  72                                 low = middle + 1
  73                         else
  74                                 return table[subindex + 1]
  75                         end
  76                 end
  77
  78                 return table[1] # Default value
  79         end
  80
  81         # Push someting in the state stack
  82         private fun push(numstate: Int, list_node: nullable Object)
  83         do
  84                 var pos = _stack_pos + 1
  85                 _stack_pos = pos
  86                 if pos < _stack.length then
  87                         var state = _stack[pos]
  88                         state.state = numstate
  89                         state.nodes = list_node
  90                 else
  91                         _stack.push(new State(numstate, list_node))
  92                 end
  93         end
  94
  95         # The current state
  96         private fun state: Int
  97         do
  98                 return _stack[_stack_pos].state
  99         end
 100
 101         # Pop something from the stack state
 102         private fun pop: nullable Object
 103         do
 104                 var res = _stack[_stack_pos].nodes
 105                 _stack_pos = _stack_pos -1
 106                 return res
 107         end
 108
 109         # Build and return a full AST.
 110         fun parse: Start
 111         do
 112                 push(0, null)
 113
 114                 var lexer = _lexer
 115                 while true do
 116                         var token = lexer.peek
 117                         var last_pos = token.location.column_start
 118                         var last_line = token.location.line_start
 119
 120                         if token isa PError then
 121                                 return new Start(null, token)
 122                         end
 123
 124                         var index = token.parser_index
 125                         var table = _action_table[state]
 126                         var action_type = table[1]
 127                         var action_value = table[2]
 128
 129                         var low = 1
 130                         var high = table.length/3 - 1
 131
 132                         while low <= high do
 133                                 var middle = (low + high) / 2
 134                                 var subindex = middle * 3
 135
 136                                 if index < table[subindex] then
 137                                         high = middle - 1
 138                                 else if index > table[subindex] then
 139                                         low = middle + 1
 140                                 else
 141                                         action_type = table[subindex + 1]
 142                                         action_value = table[subindex + 2]
 143                                         high = low -1 # break
 144                                 end
 145                         end
 146
 147                         if action_type == 0 then # SHIFT
 148                                 push(action_value, lexer.next)
 149                         else if action_type == 1 then # REDUCE
 150                                 _reduce_table[action_value].action(self)
 151                         else if action_type == 2 then # ACCEPT
 152                                 var node2 = lexer.next
 153                                 assert node2 isa EOF
 154                                 var node1 = pop
 155                                 assert node1 isa ${/parser/prods/prod/@ename}
 156                                 var node = new Start(node1, node2)
 157                                 (new ComputeProdLocationVisitor).enter_visit(node)
 158                                 return node
 159                         else if action_type == 3 then # ERROR
 160                                 var location = new Location(lexer.filename, last_line, last_line, last_pos, last_pos)
 161                                 var node2 = new PError.init_error(error_messages[errors[action_value]],location)
 162                                 var node = new Start(null, node2)
 163                                 return node
 164                         end
 165                 end
 166                 abort
 167         end
 168
 169         var _reduce_table: Array[ReduceAction]
 170         private fun build_reduce_table
 171         do
 172                 _reduce_table = new Array[ReduceAction].with_items(
 173 $ foreach {rules/rule}
 174                         new ReduceAction@index[-sep ','-]
 175 $ end foreach
 176                 )
 177         end
 178 end
 179
 180 redef class Prod
 181         # Location on the first token after the start of a production
 182         # So outside the production for epilon production
 183         var _first_location: nullable Location
 184
 185         # Location of the last token before the end of a production
 186         # So outside the production for epilon production
 187         var _last_location: nullable Location
 188 end
 189
 190 # Find location of production nodes
 191 # Uses existing token locations to infer location of productions.
 192 private class ComputeProdLocationVisitor
 193 special Visitor
 194         # Currenlty visited productions that need a first token
 195         var _need_first_prods: Array[Prod] = new Array[Prod]
 196
 197         # Already visited epsilon productions that waits something after them
 198         var _need_after_epsilons: Array[Prod] = new Array[Prod]
 199
 200         # Already visited epsilon production that waits something before them
 201         var _need_before_epsilons: Array[Prod] = new Array[Prod]
 202
 203         # Location of the last visited token in the current production
 204         var _last_location: nullable Location = null
 205
 206         redef fun visit(n: nullable PNode)
 207         do
 208                 if n == null then
 209                         return
 210                 else if n isa Token then
 211                         var loc = n.location
 212                         _last_location = loc
 213
 214                         # Add a first token to productions that need one
 215                         for no in _need_first_prods do
 216                                 no._first_location = loc
 217                         end
 218                         _need_first_prods.clear
 219
 220                         # Find location for already visited epsilon production that need one
 221                         for no in _need_after_epsilons do
 222                                 # Epsilon production that is in the middle of a non-epsilon production
 223                                 # The epsilon production has both a token before and after it
 224                                 var endl = loc
 225                                 var startl = no._last_location
 226                                 no.location = new Location(endl.file, startl.line_end, endl.line_start, startl.column_end, endl.column_start)
 227                         end
 228                         _need_after_epsilons.clear
 229                 else
 230                         assert n isa Prod
 231                         _need_first_prods.add(n)
 232
 233                         var old_last = _last_location
 234                         _last_location = null
 235                         n.visit_all(self)
 236                         var endl = _last_location
 237                         if endl == null then _last_location = old_last
 238
 239                         n._last_location = endl
 240                         var startl = n._first_location
 241                         if startl != null then
 242                                 # Non-epsilon production
 243                                 assert endl != null
 244
 245                                 n.location = new Location(startl.file, startl.line_start, endl.line_end, startl.column_start, endl.column_end)
 246
 247                                 for no in _need_before_epsilons do
 248                                         # Epsilon production that starts the current non-epsilon production
 249                                         #var startl = n.location
 250                                         no.location = new Location(startl.file, startl.line_start, startl.line_start, startl.column_start, startl.column_start)
 251                                 end
 252                                 _need_before_epsilons.clear
 253
 254                                 for no in _need_after_epsilons do
 255                                         # Epsilon production that finishes the current non-epsilon production
 256                                         #var endl = n.location
 257                                         no.location = new Location(endl.file, endl.line_end, endl.line_end, endl.column_end, endl.column_end)
 258                                 end
 259                                 _need_after_epsilons.clear
 260                         else
 261                                 # No first token means epsilon production (or "throw all my tokens" production)
 262                                 # So, it must be located it later
 263                                 if endl == null then
 264                                         # Epsilon production that starts a parent non-epsilon production
 265                                         _need_before_epsilons.add(n)
 266                                 else
 267                                         # Epsilon production in the middle or that finishes a parent non-epsilon production
 268                                         _need_after_epsilons.add(n)
 269                                 end
 270                         end
 271                 end
 272         end
 273
 274         init do end
 275 end
 276
 277 # Each reduca action has its own class, this one is the root of the hierarchy.
 278 private abstract class ReduceAction
 279         fun action(p: Parser) is abstract
 280 end
 281
 282 $ foreach {rules/rule}
 283 private class ReduceAction@index
 284 special ReduceAction
 285         redef fun action(p: Parser)
 286         do
 287                                         var node_list: nullable Object = null
 288 $   foreach {action}
 289 $   choose
 290 $     when {@cmd='POP'}
 291                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 292 $     end
 293 $     when {@cmd='FETCHLIST'}
 294                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 295                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 296 $     end
 297 $     when {@cmd='FETCHNODE'}
 298                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 299                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa nullable @etype
 300 $     end
 301 $     when {@cmd='ADDNODE'}
 302                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 303                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 304                                         end
 305 $     end
 306 $     when {@cmd='ADDLIST'}
 307 #                                       if ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 308                                                 if ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.is_empty then
 309                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 310                                                 else
 311                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.append(${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 312                                                 end
 313 #                                       end
 314 $     end
 315 $     when {@cmd='MAKELIST'}
 316                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 317 $     end
 318 $     when {@cmd='MAKENODE'}
 319                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 320 $       foreach {arg}
 321 $           if @null
 322                                                 null[-sep ','-]
 323 $           else
 324                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 325 $           end
 326 $       end foreach
 327                                         )
 328 $     end
 329 $     when {@cmd='RETURNNODE'}
 330 $       if @null
 331                                         node_list = null
 332 $       else
 333                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 334 $       end
 335 $     end
 336 $     when {@cmd='RETURNLIST'}
 337                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 338 $     end
 339 $   end choose
 340 $   end foreach
 341                                         p.push(p.go_to(@leftside), node_list)
 342         end
 343 init do end
 344 end
 345 $ end foreach
 346 $ end template
 347
 348 $ template make_parser_tables()
 349 # Parser that build a full AST
 350 abstract class ParserTable
 351         var _action_table: Array[Array[Int]]
 352         private fun build_action_table
 353         do
 354                 _action_table = once [
 355 $ foreach {parser_data/action_table/row}
 356                         action_table_row${position()}[-sep ','-]
 357 $ end foreach
 358                 ]
 359         end
 360
 361 $ foreach {parser_data/action_table/row}
 362         private fun action_table_row${position()}: Array[Int]
 363         do
 364                 return [
 365 $   foreach {action}
 366                                 @from, @action, @to[-sep ','-]
 367 $   end foreach
 368                         ]
 369         end
 370 $ end foreach
 371
 372         var _goto_table: Array[Array[Int]]
 373         private fun build_goto_table
 374         do
 375                 _goto_table = once [
 376 $ foreach {parser_data/goto_table/row}
 377                         [
 378 $   foreach {goto}
 379                                 @from, @to[-sep ','-]
 380 $   end foreach
 381                         ][-sep ','-]
 382 $ end foreach
 383                 ]
 384         end
 385
 386         private fun error_messages: Array[String]
 387         do
 388                 return once [
 389 $ foreach {parser_data/error_messages/msg}
 390                         "${sablecc:string2escaped_unicode(.)}"[-sep ','-]
 391 $ end
 392                 ]
 393         end
 394
 395         private fun errors: Array[Int]
 396         do
 397                 return once [
 398                         [-foreach {parser_data/errors/i}-]${.}[-sep ','-][-end-]
 399                 ]
 400         end
 401
 402         init do end
 403 end
 404 $ end template