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 SearchTokensVisitor).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 # Find first and last tokens of production nodes
 181 private class SearchTokensVisitor
 182 special Visitor
 183         var _untokenned_nodes: Array[Prod]
 184         var _last_token: nullable Token = null
 185         redef fun visit(n: nullable PNode)
 186         do
 187                 if n == null then
 188                         return
 189                 else if n isa Token then
 190                         _last_token = n
 191                         for no in _untokenned_nodes do
 192                                 no.first_token = n
 193                         end
 194                         _untokenned_nodes.clear
 195                 else
 196                         assert n isa Prod
 197                         _untokenned_nodes.add(n)
 198                         n.visit_all(self)
 199                         n.last_token = _last_token
 200
 201                         if n.first_token != null then
 202                                 var start_location = n.first_token.location
 203                                 var end_location = _last_token.location
 204
 205                                 if start_location != null and end_location != null then
 206                                         var file = end_location.file
 207                                         var line_start = start_location.line_start
 208                                         var line_end = end_location.line_end
 209                                         var column_start = start_location.column_start
 210                                         var column_end = end_location.column_end
 211                                         n.location = new Location(file, line_start, line_end, column_start, column_end)
 212                                 end
 213                         end
 214                 end
 215         end
 216         init
 217         do
 218                 _untokenned_nodes = new Array[Prod]
 219         end
 220 end
 221
 222 # Each reduca action has its own class, this one is the root of the hierarchy.
 223 private abstract class ReduceAction
 224         fun action(p: Parser) is abstract
 225 end
 226
 227 $ foreach {rules/rule}
 228 private class ReduceAction@index
 229 special ReduceAction
 230         redef fun action(p: Parser)
 231         do
 232                                         var node_list: nullable Object = null
 233 $   foreach {action}
 234 $   choose
 235 $     when {@cmd='POP'}
 236                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = p.pop
 237 $     end
 238 $     when {@cmd='FETCHLIST'}
 239                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 240                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa Array[Object]
 241 $     end
 242 $     when {@cmd='FETCHNODE'}
 243                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@from,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 244                                         assert ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} isa nullable @etype
 245 $     end
 246 $     when {@cmd='ADDNODE'}
 247                                         if ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 248                                                 ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.add(${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 249                                         end
 250 $     end
 251 $     when {@cmd='ADDLIST'}
 252 #                                       if ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} != null then
 253                                                 if ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.is_empty then
 254                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = ${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 255                                                 else
 256                                                         ${translate(@tolist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}.append(${translate(@fromlist,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")})
 257                                                 end
 258 #                                       end
 259 $     end
 260 $     when {@cmd='MAKELIST'}
 261                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")} = new Array[Object]
 262 $     end
 263 $     when {@cmd='MAKENODE'}
 264                                         var ${translate(@result,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}: nullable @etype = new @etype.init_${translate(@etype,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}(
 265 $       foreach {arg}
 266 $           if @null
 267                                                 null[-sep ','-]
 268 $           else
 269                                                 ${translate(.,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}[-sep ','-]
 270 $           end
 271 $       end foreach
 272                                         )
 273 $     end
 274 $     when {@cmd='RETURNNODE'}
 275 $       if @null
 276                                         node_list = null
 277 $       else
 278                                         node_list = ${translate(@node,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 279 $       end
 280 $     end
 281 $     when {@cmd='RETURNLIST'}
 282                                         node_list = ${translate(@list,"ABCDEFGHIJKLMNOPQRSTUVWXYZ","abcdefghijklmnopqrstuvwxyz")}
 283 $     end
 284 $   end choose
 285 $   end foreach
 286                                         p.push(p.go_to(@leftside), node_list)
 287         end
 288 init do end
 289 end
 290 $ end foreach
 291 $ end template
 292
 293 $ template make_parser_tables()
 294 # Parser that build a full AST
 295 abstract class ParserTable
 296         var _action_table: Array[Array[Int]]
 297         private fun build_action_table
 298         do
 299                 _action_table = once [
 300 $ foreach {parser_data/action_table/row}
 301                         action_table_row${position()}[-sep ','-]
 302 $ end foreach
 303                 ]
 304         end
 305
 306 $ foreach {parser_data/action_table/row}
 307         private fun action_table_row${position()}: Array[Int]
 308         do
 309                 return [
 310 $   foreach {action}
 311                                 @from, @action, @to[-sep ','-]
 312 $   end foreach
 313                         ]
 314         end
 315 $ end foreach
 316
 317         var _goto_table: Array[Array[Int]]
 318         private fun build_goto_table
 319         do
 320                 _goto_table = once [
 321 $ foreach {parser_data/goto_table/row}
 322                         [
 323 $   foreach {goto}
 324                                 @from, @to[-sep ','-]
 325 $   end foreach
 326                         ][-sep ','-]
 327 $ end foreach
 328                 ]
 329         end
 330
 331         private fun error_messages: Array[String]
 332         do
 333                 return once [
 334 $ foreach {parser_data/error_messages/msg}
 335                         "${sablecc:string2escaped_unicode(.)}"[-sep ','-]
 336 $ end
 337                 ]
 338         end
 339
 340         private fun errors: Array[Int]
 341         do
 342                 return once [
 343                         [-foreach {parser_data/errors/i}-]${.}[-sep ','-][-end-]
 344                 ]
 345         end
 346
 347         init do end
 348 end
 349 $ end template