parsing_cocci/.#lexer_cocci.mll.1.84

   1 (*
   2 * Copyright 2005-2009, Ecole des Mines de Nantes, University of Copenhagen
   3 * Yoann Padioleau, Julia Lawall, Rene Rydhof Hansen, Henrik Stuart, Gilles Muller
   4 * This file is part of Coccinelle.
   5 *
   6 * Coccinelle is free software: you can redistribute it and/or modify
   7 * it under the terms of the GNU General Public License as published by
   8 * the Free Software Foundation, according to version 2 of the License.
   9 *
  10 * Coccinelle is distributed in the hope that it will be useful,
  11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 * GNU General Public License for more details.
  14 *
  15 * You should have received a copy of the GNU General Public License
  16 * along with Coccinelle.  If not, see <http://www.gnu.org/licenses/>.
  17 *
  18 * The authors reserve the right to distribute this or future versions of
  19 * Coccinelle under other licenses.
  20 *)
  21
  22
  23 {
  24 open Parser_cocci_menhir
  25 module D = Data
  26 module Ast = Ast_cocci
  27 module Ast0 = Ast0_cocci
  28 module P = Parse_aux
  29 exception Lexical of string
  30 let tok = Lexing.lexeme
  31
  32 let line = ref 1
  33 let logical_line = ref 0
  34
  35 (* ---------------------------------------------------------------------- *)
  36 (* control codes *)
  37
  38 (* Defined in data.ml
  39 type line_type = MINUS | OPTMINUS | UNIQUEMINUS | PLUS | CONTEXT | UNIQUE | OPT
  40 *)
  41
  42 let current_line_type = ref (D.CONTEXT,!line,!logical_line)
  43
  44 let prev_plus = ref false
  45 let line_start = ref 0 (* offset of the beginning of the line *)
  46 let get_current_line_type lexbuf =
  47   let (c,l,ll) = !current_line_type in
  48   let lex_start = Lexing.lexeme_start lexbuf in
  49   let preceeding_spaces =
  50     if !line_start < 0 then 0 else lex_start - !line_start in
  51   line_start := -1;
  52   prev_plus := (c = D.PLUS);
  53   (c,l,ll,lex_start,preceeding_spaces,[],[],Ast0.NoMetaPos)
  54 let current_line_started = ref false
  55 let col_zero = ref true
  56
  57 let reset_line lexbuf =
  58   line := !line + 1;
  59   current_line_type := (D.CONTEXT,!line,!logical_line);
  60   current_line_started := false;
  61   col_zero := true;
  62   line_start := Lexing.lexeme_start lexbuf + 1
  63
  64 let started_line = ref (-1)
  65
  66 let start_line seen_char =
  67   current_line_started := true;
  68   col_zero := false;
  69   (if seen_char && not(!line = !started_line)
  70   then
  71     begin
  72       started_line := !line;
  73       logical_line := !logical_line + 1
  74     end)
  75
  76 let pass_zero _ = col_zero := false
  77
  78 let lexerr s1 s2 = raise (Lexical (Printf.sprintf "%s%s" s1 s2))
  79
  80 let add_current_line_type x =
  81   match (x,!current_line_type) with
  82     (D.MINUS,(D.CONTEXT,ln,lln))  ->
  83       current_line_type := (D.MINUS,ln,lln)
  84   | (D.MINUS,(D.UNIQUE,ln,lln))   ->
  85       current_line_type := (D.UNIQUEMINUS,ln,lln)
  86   | (D.MINUS,(D.OPT,ln,lln))      ->
  87       current_line_type := (D.OPTMINUS,ln,lln)
  88   | (D.PLUS,(D.CONTEXT,ln,lln))   ->
  89       current_line_type := (D.PLUS,ln,lln)
  90   | (D.UNIQUE,(D.CONTEXT,ln,lln)) ->
  91       current_line_type := (D.UNIQUE,ln,lln)
  92   | (D.OPT,(D.CONTEXT,ln,lln))    ->
  93       current_line_type := (D.OPT,ln,lln)
  94   | _ -> lexerr "invalid control character combination" ""
  95
  96 let check_minus_context_linetype s =
  97   match !current_line_type with
  98     (D.PLUS,_,_) -> lexerr "invalid in a + context: " s
  99   | _ -> ()
 100
 101 let check_context_linetype s =
 102   match !current_line_type with
 103     (D.CONTEXT,_,_) -> ()
 104   | _ -> lexerr "invalid in a nonempty context: " s
 105
 106 let check_plus_linetype s =
 107   match !current_line_type with
 108     (D.PLUS,_,_) -> ()
 109   | _ -> lexerr "invalid in a non + context: " s
 110
 111 let check_arity_context_linetype s =
 112   match !current_line_type with
 113     (D.CONTEXT,_,_) | (D.PLUS,_,_) | (D.UNIQUE,_,_) | (D.OPT,_,_) -> ()
 114   | _ -> lexerr "invalid in a nonempty context: " s
 115
 116 let process_include start finish str =
 117   (match !current_line_type with
 118     (D.PLUS,_,_) ->
 119       (try
 120         let _ = Str.search_forward (Str.regexp "\\.\\.\\.") str start in
 121         lexerr "... not allowed in + include" ""
 122       with Not_found -> ())
 123   | _ -> ());
 124   String.sub str (start + 1) (finish - start - 1)
 125
 126 (* ---------------------------------------------------------------------- *)
 127 type pm = PATCH | MATCH | UNKNOWN
 128
 129 let pm = ref UNKNOWN
 130
 131 let patch_or_match = function
 132     PATCH ->
 133       (match !pm with
 134         MATCH -> lexerr "- or + not allowed in the first column for a match" ""
 135       | PATCH -> ()
 136       | UNKNOWN -> Flag.sgrep_mode2 := false; pm := PATCH)
 137   | MATCH ->
 138       (match !pm with
 139         PATCH -> lexerr "* not allowed in the first column for a patch" ""
 140       | MATCH -> ()
 141       | UNKNOWN -> Flag.sgrep_mode2 := true; pm := MATCH)
 142   | _ -> failwith "unexpected argument"
 143
 144 (* ---------------------------------------------------------------------- *)
 145 (* identifiers, including metavariables *)
 146
 147 let metavariables = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 148
 149 let all_metavariables =
 150   (Hashtbl.create(100) : (string,(string * (D.clt -> token)) list) Hashtbl.t)
 151
 152 let type_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 153
 154 let declarer_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 155
 156 let iterator_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 157
 158 let rule_names = (Hashtbl.create(100) : (string, unit) Hashtbl.t)
 159
 160 let check_var s linetype =
 161   let fail _ =
 162     if (!Data.in_prolog || !Data.in_rule_name) &&
 163       Str.string_match (Str.regexp "<.*>") s 0
 164     then TPathIsoFile s
 165     else
 166       try (Hashtbl.find metavariables s) linetype
 167       with Not_found ->
 168         (try (Hashtbl.find type_names s) linetype
 169         with Not_found ->
 170           (try (Hashtbl.find declarer_names s) linetype
 171           with Not_found ->
 172             (try (Hashtbl.find iterator_names s) linetype
 173             with Not_found -> TIdent (s,linetype)))) in
 174   if !Data.in_meta or !Data.in_rule_name
 175   then (try Hashtbl.find rule_names s; TRuleName s with Not_found -> fail())
 176   else fail()
 177
 178 let id_tokens lexbuf =
 179   let s = tok lexbuf in
 180   let linetype = get_current_line_type lexbuf in
 181   let in_rule_name = !Data.in_rule_name in
 182   let in_meta = !Data.in_meta in
 183   let in_iso = !Data.in_iso in
 184   let in_prolog = !Data.in_prolog in
 185   match s with
 186     "identifier" when in_meta -> check_arity_context_linetype s; TIdentifier
 187   | "type" when in_meta ->       check_arity_context_linetype s; TType
 188   | "parameter" when in_meta ->  check_arity_context_linetype s; TParameter
 189   | "constant"  when in_meta ->  check_arity_context_linetype s; TConstant
 190   | "generated" when in_rule_name && not (!Flag.make_hrule = None) ->
 191       check_arity_context_linetype s; TGenerated
 192   | "expression" when in_meta || in_rule_name ->
 193       check_arity_context_linetype s; TExpression
 194   | "idexpression" when in_meta ->
 195       check_arity_context_linetype s; TIdExpression
 196   | "statement" when in_meta ->  check_arity_context_linetype s; TStatement
 197   | "function"  when in_meta ->  check_arity_context_linetype s; TFunction
 198   | "local" when in_meta ->      check_arity_context_linetype s; TLocal
 199   | "list" when in_meta ->       check_arity_context_linetype s; Tlist
 200   | "fresh" when in_meta ->      check_arity_context_linetype s; TFresh
 201   | "typedef" when in_meta ->    check_arity_context_linetype s; TTypedef
 202   | "declarer" when in_meta ->   check_arity_context_linetype s; TDeclarer
 203   | "iterator" when in_meta ->   check_arity_context_linetype s; TIterator
 204   | "name" when in_meta ->       check_arity_context_linetype s; TName
 205   | "position" when in_meta ->   check_arity_context_linetype s; TPosition
 206   | "any" when in_meta ->        check_arity_context_linetype s; TPosAny
 207   | "pure" when in_meta && in_iso ->
 208       check_arity_context_linetype s; TPure
 209   | "context" when in_meta && in_iso ->
 210       check_arity_context_linetype s; TContext
 211   | "error" when in_meta ->      check_arity_context_linetype s; TError
 212   | "words" when in_meta ->      check_context_linetype s; TWords
 213
 214   | "using" when in_rule_name || in_prolog ->  check_context_linetype s; TUsing
 215   | "disable" when in_rule_name ->  check_context_linetype s; TDisable
 216   | "extends" when in_rule_name -> check_context_linetype s; TExtends
 217   | "depends" when in_rule_name -> check_context_linetype s; TDepends
 218   | "on" when in_rule_name      -> check_context_linetype s; TOn
 219   | "ever" when in_rule_name    -> check_context_linetype s; TEver
 220   | "never" when in_rule_name   -> check_context_linetype s; TNever
 221   | "exists" when in_rule_name  -> check_context_linetype s; TExists
 222   | "forall" when in_rule_name  -> check_context_linetype s; TForall
 223   | "reverse" when in_rule_name -> check_context_linetype s; TReverse
 224   | "script" when in_rule_name -> check_context_linetype s; TScript
 225
 226   | "char" ->       Tchar     linetype
 227   | "short" ->      Tshort    linetype
 228   | "int" ->        Tint      linetype
 229   | "double" ->     Tdouble   linetype
 230   | "float" ->      Tfloat    linetype
 231   | "long" ->       Tlong     linetype
 232   | "void" ->       Tvoid     linetype
 233   | "struct" ->     Tstruct   linetype
 234   | "union" ->      Tunion    linetype
 235   | "enum" ->       Tenum     linetype
 236   | "unsigned" ->   Tunsigned linetype
 237   | "signed" ->     Tsigned   linetype
 238
 239   | "auto"  ->      Tauto     linetype
 240   | "register" ->   Tregister linetype
 241   | "extern" ->     Textern   linetype
 242   | "static" ->     Tstatic   linetype
 243   | "inline" ->     Tinline   linetype
 244   | "typedef" ->    Ttypedef  linetype
 245
 246   | "const" ->      Tconst    linetype
 247   | "volatile" ->   Tvolatile linetype
 248
 249   | "if" ->         TIf       linetype
 250   | "else" ->       TElse     linetype
 251   | "while" ->      TWhile    linetype
 252   | "do" ->         TDo       linetype
 253   | "for" ->        TFor      linetype
 254   | "switch" ->     TSwitch   linetype
 255   | "case" ->       TCase     linetype
 256   | "default" ->    TDefault  linetype
 257   | "return" ->     TReturn   linetype
 258   | "break" ->      TBreak    linetype
 259   | "continue" ->   TContinue linetype
 260   | "goto" ->       TGoto     linetype
 261
 262   | "sizeof" ->     TSizeof   linetype
 263
 264   | "Expression"     -> TIsoExpression
 265   | "ArgExpression"  -> TIsoArgExpression
 266   | "TestExpression" -> TIsoTestExpression
 267   | "Statement"      -> TIsoStatement
 268   | "Declaration"    -> TIsoDeclaration
 269   | "Type"           -> TIsoType
 270   | "TopLevel"       -> TIsoTopLevel
 271
 272   | s -> check_var s linetype
 273
 274 let mkassign op lexbuf =
 275   TAssign (Ast.OpAssign op, (get_current_line_type lexbuf))
 276
 277 let init _ =
 278   line := 1;
 279   logical_line := 0;
 280   prev_plus := false;
 281   line_start := 0;
 282   current_line_started := false;
 283   col_zero := true;
 284   pm := UNKNOWN;
 285   Data.in_rule_name := false;
 286   Data.in_meta := false;
 287   Data.in_prolog := false;
 288   Data.inheritable_positions := [];
 289   Hashtbl.clear all_metavariables;
 290   Hashtbl.clear Data.all_metadecls;
 291   Hashtbl.clear metavariables;
 292   Hashtbl.clear type_names;
 293   Hashtbl.clear rule_names;
 294   let get_name (_,x) = x in
 295   Data.add_id_meta :=
 296     (fun name constraints pure ->
 297       let fn clt = TMetaId(name,constraints,pure,clt) in
 298       Hashtbl.replace metavariables (get_name name) fn);
 299   Data.add_type_meta :=
 300     (fun name pure ->
 301       let fn clt = TMetaType(name,pure,clt) in
 302       Hashtbl.replace metavariables (get_name name) fn);
 303   Data.add_param_meta :=
 304     (function name -> function pure ->
 305       let fn clt = TMetaParam(name,pure,clt) in
 306       Hashtbl.replace metavariables (get_name name) fn);
 307   Data.add_paramlist_meta :=
 308     (function name -> function lenname -> function pure ->
 309       let fn clt = TMetaParamList(name,lenname,pure,clt) in
 310       Hashtbl.replace metavariables (get_name name) fn);
 311   Data.add_const_meta :=
 312     (fun tyopt name constraints pure ->
 313       let fn clt = TMetaConst(name,constraints,pure,tyopt,clt) in
 314       Hashtbl.replace metavariables (get_name name) fn);
 315   Data.add_err_meta :=
 316     (fun name constraints pure ->
 317       let fn clt = TMetaErr(name,constraints,pure,clt) in
 318       Hashtbl.replace metavariables (get_name name) fn);
 319   Data.add_exp_meta :=
 320     (fun tyopt name constraints pure ->
 321       let fn clt = TMetaExp(name,constraints,pure,tyopt,clt) in
 322       Hashtbl.replace metavariables (get_name name) fn);
 323   Data.add_idexp_meta :=
 324     (fun tyopt name constraints pure ->
 325       let fn clt = TMetaIdExp(name,constraints,pure,tyopt,clt) in
 326       Hashtbl.replace metavariables (get_name name) fn);
 327   Data.add_local_idexp_meta :=
 328     (fun tyopt name constraints pure ->
 329       let fn clt = TMetaLocalIdExp(name,constraints,pure,tyopt,clt) in
 330       Hashtbl.replace metavariables (get_name name) fn);
 331   Data.add_explist_meta :=
 332     (function name -> function lenname -> function pure ->
 333       let fn clt = TMetaExpList(name,lenname,pure,clt) in
 334       Hashtbl.replace metavariables (get_name name) fn);
 335   Data.add_stm_meta :=
 336     (function name -> function pure ->
 337       let fn clt = TMetaStm(name,pure,clt) in
 338       Hashtbl.replace metavariables (get_name name) fn);
 339   Data.add_stmlist_meta :=
 340     (function name -> function pure ->
 341       let fn clt = TMetaStmList(name,pure,clt) in
 342       Hashtbl.replace metavariables (get_name name) fn);
 343   Data.add_func_meta :=
 344     (fun name constraints pure ->
 345       let fn clt = TMetaFunc(name,constraints,pure,clt) in
 346       Hashtbl.replace metavariables (get_name name) fn);
 347   Data.add_local_func_meta :=
 348     (fun name constraints pure ->
 349       let fn clt = TMetaLocalFunc(name,constraints,pure,clt) in
 350       Hashtbl.replace metavariables (get_name name) fn);
 351   Data.add_iterator_meta :=
 352     (fun name constraints pure ->
 353       let fn clt = TMetaIterator(name,constraints,pure,clt) in
 354       Hashtbl.replace metavariables (get_name name) fn);
 355   Data.add_declarer_meta :=
 356     (fun name constraints pure ->
 357       let fn clt = TMetaDeclarer(name,constraints,pure,clt) in
 358       Hashtbl.replace metavariables (get_name name) fn);
 359   Data.add_pos_meta :=
 360     (fun name constraints any ->
 361       let fn ((d,ln,_,_,_,_,_,_) as clt) =
 362         (if d = Data.PLUS
 363         then
 364           failwith
 365             (Printf.sprintf "%d: positions only allowed in minus code" ln));
 366         TMetaPos(name,constraints,any,clt) in
 367       Hashtbl.replace metavariables (get_name name) fn);
 368   Data.add_type_name :=
 369     (function name ->
 370       let fn clt = TTypeId(name,clt) in
 371       Hashtbl.replace type_names name fn);
 372   Data.add_declarer_name :=
 373     (function name ->
 374       let fn clt = TDeclarerId(name,clt) in
 375       Hashtbl.replace declarer_names name fn);
 376   Data.add_iterator_name :=
 377     (function name ->
 378       let fn clt = TIteratorId(name,clt) in
 379       Hashtbl.replace iterator_names name fn);
 380   Data.init_rule := (function _ -> Hashtbl.clear metavariables);
 381   Data.install_bindings :=
 382     (function parent ->
 383       List.iter (function (name,fn) -> Hashtbl.add metavariables name fn)
 384         (Hashtbl.find all_metavariables parent))
 385
 386 let drop_spaces s =
 387   let len = String.length s in
 388   let rec loop n =
 389     if n = len
 390     then n
 391     else
 392       if List.mem (String.get s n) [' ';'\t']
 393       then loop (n+1)
 394       else n in
 395   let start = loop 0 in
 396   String.sub s start (len - start)
 397 }
 398
 399 (* ---------------------------------------------------------------------- *)
 400 (* tokens *)
 401
 402 let letter = ['A'-'Z' 'a'-'z' '_']
 403 let digit  = ['0'-'9']
 404
 405 let dec = ['0'-'9']
 406 let oct = ['0'-'7']
 407 let hex = ['0'-'9' 'a'-'f' 'A'-'F']
 408
 409 let decimal = ('0' | (['1'-'9'] dec*))
 410 let octal   = ['0']        oct+
 411 let hexa    = ("0x" |"0X") hex+
 412
 413 let pent   = dec+
 414 let pfract = dec+
 415 let sign = ['-' '+']
 416 let exp  = ['e''E'] sign? dec+
 417 let real = pent exp | ((pent? '.' pfract | pent '.' pfract? ) exp?)
 418
 419
 420 rule token = parse
 421   | [' ' '\t'  ]+             { start_line false; token lexbuf }
 422   | ['\n' '\r' '\011' '\012'] { reset_line lexbuf; token lexbuf }
 423
 424   | "//" [^ '\n']* { start_line false; token lexbuf }
 425
 426   | "@@" { start_line true; TArobArob }
 427   | "@"  { pass_zero();
 428            if !Data.in_rule_name or not !current_line_started
 429            then (start_line true; TArob)
 430            else (check_minus_context_linetype "@"; TPArob) }
 431
 432   | "WHEN" | "when"
 433       { start_line true; check_minus_context_linetype (tok lexbuf);
 434         TWhen (get_current_line_type lexbuf) }
 435
 436   | "..."
 437       { start_line true; check_minus_context_linetype (tok lexbuf);
 438         TEllipsis (get_current_line_type lexbuf) }
 439 (*
 440   | "ooo"
 441       { start_line true; check_minus_context_linetype (tok lexbuf);
 442         TCircles (get_current_line_type lexbuf) }
 443
 444   | "***"
 445       { start_line true; check_minus_context_linetype (tok lexbuf);
 446         TStars (get_current_line_type lexbuf) }
 447 *)
 448   | "<..." { start_line true; check_context_linetype (tok lexbuf);
 449              TOEllipsis (get_current_line_type lexbuf) }
 450   | "...>" { start_line true; check_context_linetype (tok lexbuf);
 451              TCEllipsis (get_current_line_type lexbuf) }
 452   | "<+..." { start_line true; check_context_linetype (tok lexbuf);
 453              TPOEllipsis (get_current_line_type lexbuf) }
 454   | "...+>" { start_line true; check_context_linetype (tok lexbuf);
 455              TPCEllipsis (get_current_line_type lexbuf) }
 456 (*
 457   | "<ooo" { start_line true; check_context_linetype (tok lexbuf);
 458              TOCircles (get_current_line_type lexbuf) }
 459   | "ooo>" { start_line true; check_context_linetype (tok lexbuf);
 460              TCCircles (get_current_line_type lexbuf) }
 461
 462   | "<***" { start_line true; check_context_linetype (tok lexbuf);
 463              TOStars (get_current_line_type lexbuf) }
 464   | "***>" { start_line true; check_context_linetype (tok lexbuf);
 465              TCStars (get_current_line_type lexbuf) }
 466 *)
 467   | "-" { pass_zero();
 468           if !current_line_started
 469           then (start_line true; TMinus (get_current_line_type lexbuf))
 470           else (patch_or_match PATCH;
 471                 add_current_line_type D.MINUS; token lexbuf) }
 472   | "+" { pass_zero();
 473           if !current_line_started
 474           then (start_line true; TPlus (get_current_line_type lexbuf))
 475           else if !Data.in_meta
 476           then TPlus0
 477           else (patch_or_match PATCH;
 478                 add_current_line_type D.PLUS; token lexbuf) }
 479   | "?" { pass_zero();
 480           if !current_line_started
 481           then (start_line true; TWhy (get_current_line_type lexbuf))
 482           else if !Data.in_meta
 483           then TWhy0
 484           else (add_current_line_type D.OPT; token lexbuf) }
 485   | "!" { pass_zero();
 486           if !current_line_started
 487           then (start_line true; TBang (get_current_line_type lexbuf))
 488           else if !Data.in_meta
 489           then TBang0
 490           else (add_current_line_type D.UNIQUE; token lexbuf) }
 491   | "(" { if not !col_zero
 492           then (start_line true; TOPar (get_current_line_type lexbuf))
 493           else
 494             (start_line true; check_context_linetype (tok lexbuf);
 495              TOPar0 (get_current_line_type lexbuf))}
 496   | "\\(" { start_line true; TOPar0 (get_current_line_type lexbuf) }
 497   | "|" { if not (!col_zero)
 498           then (start_line true; TOr(get_current_line_type lexbuf))
 499           else (start_line true;
 500                 check_context_linetype (tok lexbuf);
 501                 TMid0 (get_current_line_type lexbuf))}
 502   | "\\|" { start_line true; TMid0 (get_current_line_type lexbuf) }
 503   | ")" { if not !col_zero
 504           then (start_line true; TCPar (get_current_line_type lexbuf))
 505           else
 506             (start_line true; check_context_linetype (tok lexbuf);
 507              TCPar0 (get_current_line_type lexbuf))}
 508   | "\\)" { start_line true; TCPar0 (get_current_line_type lexbuf) }
 509
 510   | '[' { start_line true; TOCro (get_current_line_type lexbuf)   }
 511   | ']' { start_line true; TCCro (get_current_line_type lexbuf)   }
 512   | '{' { start_line true; TOBrace (get_current_line_type lexbuf) }
 513   | '}' { start_line true; TCBrace (get_current_line_type lexbuf) }
 514
 515   | "->"           { start_line true; TPtrOp (get_current_line_type lexbuf)  }
 516   | '.'            { start_line true; TDot (get_current_line_type lexbuf)    }
 517   | ','            { start_line true; TComma (get_current_line_type lexbuf)  }
 518   | ";"            { start_line true;
 519                      if !Data.in_meta
 520                      then TMPtVirg (* works better with tokens_all *)
 521                      else TPtVirg (get_current_line_type lexbuf) }
 522
 523
 524   | '*'            { pass_zero();
 525                      if !current_line_started
 526                      then
 527                        (start_line true; TMul (get_current_line_type lexbuf))
 528                      else
 529                        (patch_or_match MATCH;
 530                         add_current_line_type D.MINUS; token lexbuf) }
 531   | '/'            { start_line true;
 532                      TDmOp (Ast.Div,get_current_line_type lexbuf) }
 533   | '%'            { start_line true;
 534                      TDmOp (Ast.Mod,get_current_line_type lexbuf) }
 535   | '~'            { start_line true;  TTilde (get_current_line_type lexbuf) }
 536
 537   | "++"           { start_line true;  TInc (get_current_line_type lexbuf) }
 538   | "--"           { start_line true;  TDec (get_current_line_type lexbuf) }
 539
 540   | "="            { start_line true; TEq (get_current_line_type lexbuf) }
 541
 542   | "-="           { start_line true; mkassign Ast.Minus lexbuf }
 543   | "+="           { start_line true; mkassign Ast.Plus lexbuf }
 544
 545   | "*="           { start_line true; mkassign Ast.Mul lexbuf }
 546   | "/="           { start_line true; mkassign Ast.Div lexbuf }
 547   | "%="           { start_line true; mkassign Ast.Mod lexbuf }
 548
 549   | "&="           { start_line true; mkassign Ast.And lexbuf }
 550   | "|="           { start_line true; mkassign Ast.Or lexbuf }
 551   | "^="           { start_line true; mkassign Ast.Xor lexbuf }
 552
 553   | "<<="          { start_line true; mkassign Ast.DecLeft lexbuf }
 554   | ">>="          { start_line true; mkassign Ast.DecRight lexbuf }
 555
 556   | ":"            { start_line true; TDotDot (get_current_line_type lexbuf) }
 557
 558   | "=="           { start_line true; TEqEq   (get_current_line_type lexbuf) }
 559   | "!="           { start_line true; TNotEq  (get_current_line_type lexbuf) }
 560   | ">="           { start_line true;
 561                      TLogOp(Ast.SupEq,get_current_line_type lexbuf) }
 562   | "<="           { start_line true;
 563                      TLogOp(Ast.InfEq,get_current_line_type lexbuf) }
 564   | "<"            { start_line true;
 565                      TLogOp(Ast.Inf,get_current_line_type lexbuf) }
 566   | ">"            { start_line true;
 567                      TLogOp(Ast.Sup,get_current_line_type lexbuf) }
 568
 569   | "&&"           { start_line true; TAndLog (get_current_line_type lexbuf) }
 570   | "||"           { start_line true; TOrLog  (get_current_line_type lexbuf) }
 571
 572   | ">>"           { start_line true;
 573                      TShOp(Ast.DecRight,get_current_line_type lexbuf) }
 574   | "<<"           { start_line true;
 575                      TShOp(Ast.DecLeft,get_current_line_type lexbuf) }
 576
 577   | "&"            { start_line true; TAnd    (get_current_line_type lexbuf) }
 578   | "^"            { start_line true; TXor(get_current_line_type lexbuf) }
 579
 580   | ( ("#" [' ' '\t']*  "define" [' ' '\t']+))
 581     ( (letter (letter |digit)*) as ident)
 582       { start_line true;
 583         let (arity,line,lline,offset,col,strbef,straft,pos) as lt =
 584           get_current_line_type lexbuf in
 585         let off = String.length "#define " in
 586         (* -1 in the code below because the ident is not at the line start *)
 587         TDefine
 588           (lt,
 589            check_var ident
 590              (arity,line,lline,offset+off,(-1),[],[],Ast0.NoMetaPos)) }
 591   | ( ("#" [' ' '\t']*  "define" [' ' '\t']+))
 592     ( (letter (letter | digit)*) as ident)
 593     '('
 594       { start_line true;
 595         let (arity,line,lline,offset,col,strbef,straft,pos) as lt =
 596           get_current_line_type lexbuf in
 597         let off = String.length "#define " in
 598         TDefineParam
 599         (lt,
 600          check_var ident
 601            (* why pos here but not above? *)
 602            (arity,line,lline,offset+off,(-1),strbef,straft,pos),
 603          offset + off + (String.length ident)) }
 604   | "#" [' ' '\t']* "include" [' ' '\t']* '"' [^ '"']+ '"'
 605       { TIncludeL
 606           (let str = tok lexbuf in
 607           let start = String.index str '"' in
 608           let finish = String.rindex str '"' in
 609           start_line true;
 610           (process_include start finish str,get_current_line_type lexbuf)) }
 611   | "#" [' ' '\t']* "include" [' ' '\t']* '<' [^ '>']+ '>'
 612       { TIncludeNL
 613           (let str = tok lexbuf in
 614           let start = String.index str '<' in
 615           let finish = String.rindex str '>' in
 616           start_line true;
 617           (process_include start finish str,get_current_line_type lexbuf)) }
 618   | "#" [' ' '\t']* "if" [^'\n']*
 619   | "#" [' ' '\t']* "ifdef" [^'\n']*
 620   | "#" [' ' '\t']* "ifndef" [^'\n']*
 621   | "#" [' ' '\t']* "else" [^'\n']*
 622   | "#" [' ' '\t']* "elif" [^'\n']*
 623   | "#" [' ' '\t']* "endif" [^'\n']*
 624   | "#" [' ' '\t']* "error" [^'\n']*
 625       { start_line true; check_plus_linetype (tok lexbuf);
 626         TPragma (tok lexbuf) }
 627   | "---" [^'\n']*
 628       { (if !current_line_started
 629       then lexerr "--- must be at the beginning of the line" "");
 630         start_line true;
 631         TMinusFile
 632           (let str = tok lexbuf in
 633           (drop_spaces(String.sub str 3 (String.length str - 3)),
 634            (get_current_line_type lexbuf))) }
 635   | "+++" [^'\n']*
 636       { (if !current_line_started
 637       then lexerr "+++ must be at the beginning of the line" "");
 638         start_line true;
 639         TPlusFile
 640           (let str = tok lexbuf in
 641           (drop_spaces(String.sub str 3 (String.length str - 3)),
 642            (get_current_line_type lexbuf))) }
 643
 644   | letter (letter | digit)*
 645       { start_line true; id_tokens lexbuf }
 646
 647   | "'" { start_line true;
 648           TChar(char lexbuf,get_current_line_type lexbuf) }
 649   | '"' { start_line true;
 650           TString(string lexbuf,(get_current_line_type lexbuf)) }
 651   | (real as x)    { start_line true;
 652                      TFloat(x,(get_current_line_type lexbuf)) }
 653   | ((( decimal | hexa | octal)
 654       ( ['u' 'U']
 655       | ['l' 'L']
 656       | (['l' 'L'] ['u' 'U'])
 657       | (['u' 'U'] ['l' 'L'])
 658       | (['u' 'U'] ['l' 'L'] ['l' 'L'])
 659       | (['l' 'L'] ['l' 'L'])
 660       )?
 661     ) as x) { start_line true; TInt(x,(get_current_line_type lexbuf)) }
 662
 663   | "<=>"          { TIso }
 664   | "=>"           { TRightIso }
 665
 666   | eof            { EOF }
 667
 668   | _ { lexerr "unrecognised symbol, in token rule: " (tok lexbuf) }
 669
 670
 671 and char = parse
 672   | (_ as x) "'"                                     { String.make 1 x }
 673   | (("\\" (oct | oct oct | oct oct oct)) as x  "'") { x }
 674   | (("\\x" (hex | hex hex)) as x  "'")       { x }
 675   | (("\\" (_ as v)) as x "'")
 676         { (match v with
 677             | 'n' -> ()  | 't' -> ()   | 'v' -> ()  | 'b' -> ()
 678             | 'r' -> ()  | 'f' -> () | 'a' -> ()
 679             | '\\' -> () | '?'  -> () | '\'' -> ()  | '"' -> ()
 680             | 'e' -> ()
 681             | _ -> lexerr "unrecognised symbol: " (tok lexbuf)
 682             );
 683           x
 684         }
 685   | _ { lexerr "unrecognised symbol: " (tok lexbuf) }
 686
 687 and string  = parse
 688   | '"'                                       { "" }
 689   | (_ as x)                   { Common.string_of_char x ^ string lexbuf }
 690   | ("\\" (oct | oct oct | oct oct oct)) as x { x ^ string lexbuf }
 691   | ("\\x" (hex | hex hex)) as x              { x ^ string lexbuf }
 692   | ("\\" (_ as v)) as x
 693        {
 694          (match v with
 695          | 'n' -> ()  | 't' -> ()   | 'v' -> ()  | 'b' -> () | 'r' -> ()
 696          | 'f' -> () | 'a' -> ()
 697          | '\\' -> () | '?'  -> () | '\'' -> ()  | '"' -> ()
 698          | 'e' -> ()
 699          | '\n' -> ()
 700          | _ -> lexerr "unrecognised symbol:" (tok lexbuf)
 701          );
 702           x ^ string lexbuf
 703        }
 704   | _ { lexerr "unrecognised symbol: " (tok lexbuf) }