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