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                 loop
 116                         var token = lexer.peek
 117                         if token isa PError then
 118                                 return new Start(null, token)
 119                         end
 120
 121                         var index = token.parser_index
 122                         var table = _action_table[state]
 123                         var action_type = table[1]
 124                         var action_value = table[2]
 125
 126                         var low = 1
 127                         var high = table.length/3 - 1
 128
 129                         while low <= high do
 130                                 var middle = (low + high) / 2
 131                                 var subindex = middle * 3
 132
 133                                 if index < table[subindex] then
 134                                         high = middle - 1
 135                                 else if index > table[subindex] then
 136                                         low = middle + 1
 137                                 else
 138                                         action_type = table[subindex + 1]
 139                                         action_value = table[subindex + 2]
 140                                         high = low -1 # break
 141                                 end
 142                         end
 143
 144                         if action_type == 0 then # SHIFT
 145                                 push(action_value, lexer.next)
 146                         else if action_type == 1 then # REDUCE
 147                                 _reduce_table[action_value].action(self)
 148                         else if action_type == 2 then # ACCEPT
 149                                 var node2 = lexer.next
 150                                 assert node2 isa EOF
 151                                 var node1 = pop
 152                                 assert node1 isa ${/parser/prods/prod/@ename}
 153                                 var node = new Start(node1, node2)
 154                                 (new ComputeProdLocationVisitor).enter_visit(node)
 155                                 return node
 156                         else if action_type == 3 then # ERROR
 157                                 var node2 = new PError.init_error("Syntax error: unexpected token.", token.location)
 158                                 var node = new Start(null, node2)
 159                                 return node
 160                         end
 161                         if false then break # FIXME remove once unreach loop exits are in c_src
 162                 end
 163                 abort # FIXME remove once unreach loop exits are in c_src
 164         end
 165
 166         var _reduce_table: Array[ReduceAction]
 167         private fun build_reduce_table
 168         do
 169                 _reduce_table = new Array[ReduceAction].with_items(
 170 $ foreach {rules/rule}
 171                         new ReduceAction@index[-sep ','-]
 172 $ end foreach
 173                 )
 174         end
 175 end
 176
 177 redef class Prod
 178         # Location on the first token after the start of a production
 179         # So outside the production for epilon production
 180         var _first_location: nullable Location
 181
 182         # Location of the last token before the end of a production
 183         # So outside the production for epilon production
 184         var _last_location: nullable Location
 185 end
 186
 187 # Find location of production nodes
 188 # Uses existing token locations to infer location of productions.
 189 private class ComputeProdLocationVisitor
 190 special Visitor
 191         # Currenlty visited productions that need a first token
 192         var _need_first_prods: Array[Prod] = new Array[Prod]
 193
 194         # Already visited epsilon productions that waits something after them
 195         var _need_after_epsilons: Array[Prod] = new Array[Prod]
 196
 197         # Already visited epsilon production that waits something before them
 198         var _need_before_epsilons: Array[Prod] = new Array[Prod]
 199
 200         # Location of the last visited token in the current production
 201         var _last_location: nullable Location = null
 202
 203         redef fun visit(n: nullable PNode)
 204         do
 205                 if n == null then
 206                         return
 207                 else if n isa Token then
 208                         var loc = n.location
 209                         _last_location = loc
 210
 211                         # Add a first token to productions that need one
 212                         for no in _need_first_prods do
 213                                 no._first_location = loc
 214                         end
 215                         _need_first_prods.clear
 216
 217                         # Find location for already visited epsilon production that need one
 218                         for no in _need_after_epsilons do
 219                                 # Epsilon production that is in the middle of a non-epsilon production
 220                                 # The epsilon production has both a token before and after it
 221                                 var endl = loc
 222                                 var startl = no._last_location
 223                                 no.location = new Location(endl.file, startl.line_end, endl.line_start, startl.column_end, endl.column_start)
 224                         end
 225                         _need_after_epsilons.clear
 226                 else
 227                         assert n isa Prod
 228                         _need_first_prods.add(n)
 229
 230                         var old_last = _last_location
 231                         _last_location = null
 232                         n.visit_all(self)
 233                         var endl = _last_location
 234                         if endl == null then _last_location = old_last
 235
 236                         n._last_location = endl
 237                         var startl = n._first_location
 238                         if startl != null then
 239                                 # Non-epsilon production
 240                                 assert endl != null
 241
 242                                 n.location = new Location(startl.file, startl.line_start, endl.line_end, startl.column_start, endl.column_end)
 243
 244                                 for no in _need_before_epsilons do
 245                                         # Epsilon production that starts the current non-epsilon production
 246                                         #var startl = n.location
 247                                         no.location = new Location(startl.file, startl.line_start, startl.line_start, startl.column_start, startl.column_start)
 248                                 end
 249                                 _need_before_epsilons.clear
 250
 251                                 for no in _need_after_epsilons do
 252                                         # Epsilon production that finishes the current non-epsilon production
 253                                         #var endl = n.location
 254                                         no.location = new Location(endl.file, endl.line_end, endl.line_end, endl.column_end, endl.column_end)
 255                                 end
 256                                 _need_after_epsilons.clear
 257                         else
 258                                 # No first token means epsilon production (or "throw all my tokens" production)
 259                                 # So, it must be located it later
 260                                 if endl == null then
 261                                         # Epsilon production that starts a parent non-epsilon production
 262                                         _need_before_epsilons.add(n)
 263                                 else
 264                                         # Epsilon production in the middle or that finishes a parent non-epsilon production
 265                                         _need_after_epsilons.add(n)
 266                                 end
 267                         end
 268                 end
 269         end
 270
 271         init do end
 272 end
 273
 274 # Each reduca action has its own class, this one is the root of the hierarchy.
 275 private abstract class ReduceAction
 276         fun action(p: Parser) is abstract
 277         fun concat(l1, l2 : Array[Object]): Array[Object]
 278         do
 279                 if l1.is_empty then return l2
 280                 l1.append(l2)
 281                 return l1
 282         end
 283 end
 284
 285 $ foreach {rules/rule}
 286 private class ReduceAction@index
 287 special ReduceAction
 288         redef fun action(p: Parser)
 289         do
 290                                         var node_list: nullable Object = null
 291 $   foreach {action}
 292 $   choose
 293 $     when {@cmd='POP'}
 294                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 295 $     end
 296 $     when {@cmd='FETCHLIST'}
 297                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 298                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 299 $     end
 300 $     when {@cmd='FETCHNODE'}
 301                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 302                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa nullable @etype
 303 $     end
 304 $     when {@cmd='ADDNODE'}
 305                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 306                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 307                                         end
 308 $     end
 309 $     when {@cmd='ADDLIST'}
 310                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = concat(${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}, ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 311 $     end
 312 $     when {@cmd='MAKELIST'}
 313                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 314 $     end
 315 $     when {@cmd='MAKENODE'}
 316                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 317 $       foreach {arg}
 318 $           if @null
 319                                                 null[-sep ','-]
 320 $           else
 321                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 322 $           end
 323 $       end foreach
 324                                         )
 325 $     end
 326 $     when {@cmd='RETURNNODE'}
 327 $       if @null
 328                                         node_list = null
 329 $       else
 330                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 331 $       end
 332 $     end
 333 $     when {@cmd='RETURNLIST'}
 334                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 335 $     end
 336 $   end choose
 337 $   end foreach
 338                                         p.push(p.go_to(@leftside), node_list)
 339         end
 340 init do end
 341 end
 342 $ end foreach
 343 $ end template
 344
 345 $ template make_parser_tables()
 346 # Parser that build a full AST
 347 abstract class ParserTable
 348         var _action_table: Array[Array[Int]]
 349         private fun build_action_table
 350         do
 351                 _action_table = once [
 352 $ foreach {parser_data/action_table/row}
 353                         action_table_row${position()}[-sep ','-]
 354 $ end foreach
 355                 ]
 356         end
 357
 358 $ foreach {parser_data/action_table/row}
 359         private fun action_table_row${position()}: Array[Int]
 360         do
 361                 return [
 362 $   foreach {action}
 363                                 @from, @action, @to[-sep ','-]
 364 $   end foreach
 365                         ]
 366         end
 367 $ end foreach
 368
 369         var _goto_table: Array[Array[Int]]
 370         private fun build_goto_table
 371         do
 372                 _goto_table = once [
 373 $ foreach {parser_data/goto_table/row}
 374                         [
 375 $   foreach {goto}
 376                                 @from, @to[-sep ','-]
 377 $   end foreach
 378                         ][-sep ','-]
 379 $ end foreach
 380                 ]
 381         end
 382
 383         init do end
 384 end
 385 $ end template