src/gram.h

   1 /* Data definitions for internal representation of bison's input,
   2    Copyright (C) 1984, 1986, 1989, 1992 Free Software Foundation, Inc.
   3
   4 This file is part of Bison, the GNU Compiler Compiler.
   5
   6 Bison 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; either version 2, or (at your option)
   9 any later version.
  10
  11 Bison is distributed in the hope that it will be useful,
  12 but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 GNU General Public License for more details.
  15
  16 You should have received a copy of the GNU General Public License
  17 along with Bison; see the file COPYING.  If not, write to
  18 the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.  */
  19
  20
  21 /* representation of the grammar rules:
  22
  23 ntokens is the number of tokens, and nvars is the number of variables
  24 (nonterminals).  nsyms is the total number, ntokens + nvars.
  25
  26         (the true number of token values assigned is ntokens
  27         reduced by one for each alias declaration)
  28
  29 Each symbol (either token or variable) receives a symbol number.
  30 Numbers 0 to ntokens-1 are for tokens, and ntokens to nsyms-1 are for
  31 variables.  Symbol number zero is the end-of-input token.  This token
  32 is counted in ntokens.
  33
  34 The rules receive rule numbers 1 to nrules in the order they are written.
  35 Actions and guards are accessed via the rule number.
  36
  37 The rules themselves are described by three arrays: rrhs, rlhs and
  38 ritem.  rlhs[R] is the symbol number of the left hand side of rule R.
  39 The right hand side is stored as symbol numbers in a portion of
  40 ritem.  rrhs[R] contains the index in ritem of the beginning of the
  41 portion for rule R.
  42
  43 If rlhs[R] is -1, the rule has been thrown out by reduce.c
  44 and should be ignored.
  45
  46 The length of the portion is one greater
  47  than the number of symbols in the rule's right hand side.
  48 The last element in the portion contains minus R, which
  49 identifies it as the end of a portion and says which rule it is for.
  50
  51 The portions of ritem come in order of increasing rule number and are
  52 followed by an element which is zero to mark the end.  nitems is the
  53 total length of ritem, not counting the final zero.  Each element of
  54 ritem is called an "item" and its index in ritem is an item number.
  55
  56 Item numbers are used in the finite state machine to represent
  57 places that parsing can get to.
  58
  59 Precedence levels are recorded in the vectors sprec and rprec.
  60 sprec records the precedence level of each symbol,
  61 rprec the precedence level of each rule.
  62 rprecsym is the symbol-number of the symbol in %prec for this rule (if any).
  63
  64 Precedence levels are assigned in increasing order starting with 1 so
  65 that numerically higher precedence values mean tighter binding as they
  66 ought to.  Zero as a symbol or rule's precedence means none is
  67 assigned.
  68
  69 Associativities are recorded similarly in rassoc and sassoc.  */
  70
  71
  72 #define ISTOKEN(s)      ((s) < ntokens)
  73 #define ISVAR(s)        ((s) >= ntokens)
  74
  75
  76 extern int nitems;
  77 extern int nrules;
  78 extern int nsyms;
  79 extern int ntokens;
  80 extern int nvars;
  81
  82 extern short *ritem;
  83 extern short *rlhs;
  84 extern short *rrhs;
  85 extern short *rprec;
  86 extern short *rprecsym;
  87 extern short *sprec;
  88 extern short *rassoc;
  89 extern short *sassoc;
  90 extern short *rline;            /* Source line number of each rule */
  91
  92 extern int start_symbol;
  93
  94
  95 /* associativity values in elements of rassoc, sassoc.  */
  96
  97 #define RIGHT_ASSOC 1
  98 #define LEFT_ASSOC 2
  99 #define NON_ASSOC 3
 100
 101 /* token translation table:
 102 indexed by a token number as returned by the user's yylex routine,
 103 it yields the internal token number used by the parser and throughout bison.
 104 If translations is zero, the translation table is not used because
 105 the two kinds of token numbers are the same.
 106 (It is noted in reader.c that "Nowadays translations is always set to 1...")
 107 */
 108
 109 extern short *token_translations;
 110 extern int translations;
 111 extern int max_user_token_number;
 112
 113 /* semantic_parser is nonzero if the input file says to use the hairy parser
 114 that provides for semantic error recovery.  If it is zero, the yacc-compatible
 115 simplified parser is used.  */
 116
 117 extern int semantic_parser;
 118
 119 /* pure_parser is nonzero if should generate a parser that is all pure and reentrant. */
 120
 121 extern int pure_parser;
 122
 123 /* error_token_number is the token number of the error token.  */
 124
 125 extern int error_token_number;