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         super TablesCapable
  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_reduce_table
  53         end
  54
  55         # Do a transition in the automata
  56         private fun go_to(index: Int): Int
  57         do
  58                 var state = state
  59                 var low = 1
  60                 var high = parser_goto(index, 0) - 1
  61
  62                 while low <= high do
  63                         var middle = (low + high) / 2
  64                         var subindex = middle * 2 + 1 # +1 because parser_goto(index, 0) is the length
  65
  66                         var goal = parser_goto(index, subindex)
  67                         if state < goal then
  68                                 high = middle - 1
  69                         else if state > goal then
  70                                 low = middle + 1
  71                         else
  72                                 return parser_goto(index, subindex+1)
  73                         end
  74                 end
  75
  76                 return parser_goto(index, 2) # Default value
  77         end
  78
  79         # Push someting in the state stack
  80         private fun push(numstate: Int, list_node: nullable Object)
  81         do
  82                 var pos = _stack_pos + 1
  83                 _stack_pos = pos
  84                 if pos < _stack.length then
  85                         var state = _stack[pos]
  86                         state.state = numstate
  87                         state.nodes = list_node
  88                 else
  89                         _stack.push(new State(numstate, list_node))
  90                 end
  91         end
  92
  93         # The current state
  94         private fun state: Int
  95         do
  96                 return _stack[_stack_pos].state
  97         end
  98
  99         # Pop something from the stack state
 100         private fun pop: nullable Object
 101         do
 102                 var res = _stack[_stack_pos].nodes
 103                 _stack_pos = _stack_pos -1
 104                 return res
 105         end
 106
 107         # Build and return a full AST.
 108         fun parse: Start
 109         do
 110                 push(0, null)
 111
 112                 var lexer = _lexer
 113                 loop
 114                         var token = lexer.peek
 115                         if token isa PError then
 116                                 return new Start(null, token)
 117                         end
 118
 119                         var state = self.state
 120                         var index = token.parser_index
 121                         var action_type = parser_action(state, 2)
 122                         var action_value = parser_action(state, 3)
 123
 124                         var low = 1
 125                         var high = parser_action(state, 0) - 1
 126
 127                         while low <= high do
 128                                 var middle = (low + high) / 2
 129                                 var subindex = middle * 3 + 1 # +1 because parser_action(state, 0) is the length
 130
 131                                 var goal = parser_action(state, subindex)
 132                                 if index < goal then
 133                                         high = middle - 1
 134                                 else if index > goal then
 135                                         low = middle + 1
 136                                 else
 137                                         action_type = parser_action(state, subindex+1)
 138                                         action_value = parser_action(state, subindex+2)
 139                                         break
 140                                 end
 141                         end
 142
 143                         if action_type == 0 then # SHIFT
 144                                 push(action_value, lexer.next)
 145                         else if action_type == 1 then # REDUCE
 146                                 _reduce_table[action_value].action(self)
 147                         else if action_type == 2 then # ACCEPT
 148                                 var node2 = lexer.next
 149                                 assert node2 isa EOF
 150                                 var node1 = pop
 151                                 assert node1 isa ${/parser/prods/prod/@ename}
 152                                 var node = new Start(node1, node2)
 153                                 (new ComputeProdLocationVisitor).enter_visit(node)
 154                                 return node
 155                         else if action_type == 3 then # ERROR
 156                                 var node2 = new PParserError.init_parser_error("Syntax error: unexpected {token}.", token.location, token)
 157                                 var node = new Start(null, node2)
 158                                 return node
 159                         end
 160                 end
 161         end
 162
 163         var _reduce_table: Array[ReduceAction]
 164         private fun build_reduce_table
 165         do
 166                 _reduce_table = new Array[ReduceAction].with_items(
 167 $ foreach {rules/rule}
 168                         new ReduceAction@index(@leftside)[-sep ','-]
 169 $ end foreach
 170                 )
 171         end
 172 end
 173
 174 redef class Prod
 175         # Location on the first token after the start of a production
 176         # So outside the production for epilon production
 177         var _first_location: nullable Location
 178
 179         # Location of the last token before the end of a production
 180         # So outside the production for epilon production
 181         var _last_location: nullable Location
 182 end
 183
 184 # Find location of production nodes
 185 # Uses existing token locations to infer location of productions.
 186 private class ComputeProdLocationVisitor
 187         super Visitor
 188         # Currenlty visited productions that need a first token
 189         var _need_first_prods: Array[Prod] = new Array[Prod]
 190
 191         # Already visited epsilon productions that waits something after them
 192         var _need_after_epsilons: Array[Prod] = new Array[Prod]
 193
 194         # Already visited epsilon production that waits something before them
 195         var _need_before_epsilons: Array[Prod] = new Array[Prod]
 196
 197         # Location of the last visited token in the current production
 198         var _last_location: nullable Location = null
 199
 200         redef fun visit(n: PNode)
 201         do
 202                 if n isa Token then
 203                         var loc = n.location
 204                         _last_location = loc
 205
 206                         # Add a first token to productions that need one
 207                         if not _need_first_prods.is_empty then
 208                                 for no in _need_first_prods do
 209                                         no._first_location = loc
 210                                 end
 211                                 _need_first_prods.clear
 212                         end
 213
 214                         # Find location for already visited epsilon production that need one
 215                         if not _need_after_epsilons.is_empty then
 216                                 for no in _need_after_epsilons do
 217                                         # Epsilon production that is in the middle of a non-epsilon production
 218                                         # The epsilon production has both a token before and after it
 219                                         var endl = loc
 220                                         var startl = no._last_location
 221                                         no.location = new Location(endl.file, startl.line_end, endl.line_start, startl.column_end, endl.column_start)
 222                                 end
 223                                 _need_after_epsilons.clear
 224                         end
 225                 else
 226                         assert n isa Prod
 227                         _need_first_prods.add(n)
 228
 229                         var old_last = _last_location
 230                         _last_location = null
 231                         n.visit_all(self)
 232                         var endl = _last_location
 233                         if endl == null then _last_location = old_last
 234
 235                         n._last_location = endl
 236                         var startl = n._first_location
 237                         if startl != null then
 238                                 # Non-epsilon production
 239                                 assert endl != null
 240
 241                                 n.location = new Location(startl.file, startl.line_start, endl.line_end, startl.column_start, endl.column_end)
 242
 243                                 if not _need_before_epsilons.is_empty then
 244                                         var loc = new Location(startl.file, startl.line_start, startl.line_start, startl.column_start, startl.column_start)
 245                                         for no in _need_before_epsilons do
 246                                                 # Epsilon production that starts the current non-epsilon production
 247                                                 no.location = loc
 248                                         end
 249                                         _need_before_epsilons.clear
 250                                 end
 251
 252                                 if not _need_after_epsilons.is_empty then
 253                                         var loc = new Location(endl.file, endl.line_end, endl.line_end, endl.column_end, endl.column_end)
 254                                         for no in _need_after_epsilons do
 255                                                 # Epsilon production that finishes the current non-epsilon production
 256                                                 no.location = loc
 257                                         end
 258                                         _need_after_epsilons.clear
 259                                 end
 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         fun concat(l1, l2 : Array[Object]): Array[Object]
 281         do
 282                 if l1.is_empty then return l2
 283                 l1.append(l2)
 284                 return l1
 285         end
 286         var _goto: Int
 287         init(g: Int) do _goto = g
 288 end
 289
 290 $ foreach {rules/rule}
 291 private class ReduceAction@index
 292         super ReduceAction
 293         redef fun action(p: Parser)
 294         do
 295                                         var node_list: nullable Object = null
 296 $   foreach {action}
 297 $   choose
 298 $     when {@cmd='POP'}
 299                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 300 $     end
 301 $     when {@cmd='FETCHLIST'}
 302                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 303                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 304 $     end
 305 $     when {@cmd='FETCHNODE'}
 306                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 307                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa nullable @etype
 308 $     end
 309 $     when {@cmd='ADDNODE'}
 310                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 311                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 312                                         end
 313 $     end
 314 $     when {@cmd='ADDLIST'}
 315                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = concat(${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}, ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 316 $     end
 317 $     when {@cmd='MAKELIST'}
 318                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 319 $     end
 320 $     when {@cmd='MAKENODE'}
 321 $      if {count(arg)!=0}
 322                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 323 $       foreach {arg}
 324 $           if @null
 325                                                 null[-sep ','-]
 326 $           else
 327                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 328 $           end
 329 $       end foreach
 330                                         )
 331 $      else
 332                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 333 $      end
 334 $     end
 335 $     when {@cmd='RETURNNODE'}
 336 $       if @null
 337                                         node_list = null
 338 $       else
 339                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 340 $       end
 341 $     end
 342 $     when {@cmd='RETURNLIST'}
 343                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 344 $     end
 345 $   end choose
 346 $   end foreach
 347                                         p.push(p.go_to(_goto), node_list)
 348         end
 349 end
 350 $ end foreach
 351 $ end template
 352
 353 $ template make_parser_table()
 354 $ foreach {parser_data/action_table/row}
 355 static int parser_action_row${position()}[] = {
 356         ${count(action)},
 357 $   foreach {action}
 358         @from, @action, @to[-sep ','-]
 359 $   end foreach
 360 };
 361 $ end foreach
 362
 363 const int* const parser_action_table[] = {
 364 $ foreach {parser_data/action_table/row}
 365         parser_action_row${position()}[-sep ','-]
 366 $ end foreach
 367 };
 368
 369 $ foreach {parser_data/goto_table/row}
 370 static int parser_goto_row${position()}[] = {
 371         ${count(goto)},
 372 $   foreach {goto}
 373         @from, @to[-sep ','-]
 374 $   end foreach
 375 };
 376 $ end foreach
 377
 378 const int* const parser_goto_table[] = {
 379 $ foreach {parser_data/goto_table/row}
 380         parser_goto_row${position()}[-sep ','-]
 381 $ end foreach
 382 };
 383 $ end template