* src/scan-gram.l (YY_INIT, YY_GROW, YY_FINISH): Rename as...

[bison.git] / src / state.h

/* Type definitions for nondeterministic finite state machine for bison,
   Copyright 1984, 1989, 2000, 2001 Free Software Foundation, Inc.

   This file is part of Bison, the GNU Compiler Compiler.

   Bison is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2, or (at your option)
   any later version.

   Bison is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with Bison; see the file COPYING.  If not, write to
   the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
   Boston, MA 02111-1307, USA.  */


/* These type definitions are used to represent a nondeterministic
   finite state machine that parses the specified grammar.  This
   information is generated by the function generate_states in the
   file LR0.

   Each state of the machine is described by a set of items --
   particular positions in particular rules -- that are the possible
   places where parsing could continue when the machine is in this
   state.  These symbols at these items are the allowable inputs that
   can follow now.

   A core represents one state.  States are numbered in the number
   field.  When generate_states is finished, the starting state is
   state 0 and nstates is the number of states.  (A transition to a
   state whose state number is nstates indicates termination.)  All
   the cores are chained together and first_state points to the first
   one (state 0).

   For each state there is a particular symbol which must have been
   the last thing accepted to reach that state.  It is the
   accessing_symbol of the core.

   Each core contains a vector of NITEMS items which are the indices
   in the ritems vector of the items that are selected in this state.

   The link field is used for chaining symbols that hash states by
   their itemsets.  This is for recognizing equivalent states and
   combining them when the states are generated.

   The two types of transitions are shifts (push the lookahead token
   and read another) and reductions (combine the last n things on the
   stack via a rule, replace them with the symbol that the rule
   derives, and leave the lookahead token alone).  When the states are
   generated, these transitions are represented in two other lists.

   Each shifts structure describes the possible shift transitions out
   of one state, the state whose number is in the number field.  The
   shifts structures are linked through next and first_shift points to
   them.  Each contains a vector of numbers of the states that shift
   transitions can go to.  The accessing_symbol fields of those
   states' cores say what kind of input leads to them.

   A shift to state zero should be ignored.  Conflict resolution
   deletes shifts by changing them to zero.

   Each reductions structure describes the possible reductions at the
   state whose number is in the number field.  The data is a list of
   nreds rules, represented by their rule numbers.  first_reduction
   points to the list of these structures.

   Conflict resolution can decide that certain tokens in certain
   states should explicitly be errors (for implementing %nonassoc).
   For each state, the tokens that are errors for this reason are
   recorded in an errs structure, which has the state number in its
   number field.  The rest of the errs structure is full of token
   numbers.

   There is at least one shift transition present in state zero.  It
   leads to a next-to-final state whose accessing_symbol is the
   grammar's start symbol.  The next-to-final state has one shift to
   the final state, whose accessing_symbol is zero (end of input).
   The final state has one shift, which goes to the termination state
   (whose number is nstates-1).  The reason for the extra state at the
   end is to placate the parser's strategy of making all decisions one
   token ahead of its actions.  */

#ifndef STATE_H_
# define STATE_H_


/*---------.
| Shifts.  |
`---------*/

typedef struct shifts
{
  short nshifts;
  short shifts[1];
} shifts;

shifts *shifts_new PARAMS ((int n));


/* What is the symbol which is shifted by SHIFTS->shifts[Shift]?  Can
   be a token (amongst which the error token), or non terminals in
   case of gotos.  */

#define SHIFT_SYMBOL(Shifts, Shift) \
  (states[Shifts->shifts[Shift]]->accessing_symbol)

/* Is the SHIFTS->shifts[Shift] a real shift? (as opposed to gotos.) */

#define SHIFT_IS_SHIFT(Shifts, Shift) \
  (ISTOKEN (SHIFT_SYMBOL (Shifts, Shift)))

/* Is the SHIFTS->shifts[Shift] a goto?. */

#define SHIFT_IS_GOTO(Shifts, Shift) \
  (!SHIFT_IS_SHIFT (Shifts, Shift))

/* Is the SHIFTS->shifts[Shift] then handling of the error token?. */

#define SHIFT_IS_ERROR(Shifts, Shift) \
  (SHIFT_SYMBOL (Shifts, Shift) == errtoken->number)

/* When resolving a SR conflicts, if the reduction wins, the shift is
   disabled.  */

#define SHIFT_DISABLE(Shifts, Shift) \
  (Shifts->shifts[Shift] = 0)

#define SHIFT_IS_DISABLED(Shifts, Shift) \
  (Shifts->shifts[Shift] == 0)


/*-------.
| Errs.  |
`-------*/

typedef struct errs
{
  short nerrs;
  short errs[1];
} errs;

errs *errs_new PARAMS ((int n));
errs *errs_dup PARAMS ((errs *src));


/*-------------.
| Reductions.  |
`-------------*/

typedef struct reductions
{
  short nreds;
  short rules[1];
} reductions;

reductions *reductions_new PARAMS ((int n));


/*----------.
| State_t.  |
`----------*/

typedef struct state_s
{
  struct state_s *next;
  struct state_s *link;

  short number;
  symbol_number_t accessing_symbol;
  shifts     *shifts;
  reductions *reductions;
  errs       *errs;

  /* Nonzero if no lookahead is needed to decide what to do in state S.  */
  char consistent;

  /* Used in LALR, not LR(0). */
  /* Pseudo pointer into LA. */
  short lookaheadsp;
  int nlookaheads;

  /* If some conflicts were solved thanks to precedence/associativity,
     a human readable description of the resolution.  */
  const char *solved_conflicts;

  /* Its items.  Must be last, since ITEMS can be arbitrarily large.
     */
  unsigned short nitems;
  item_number_t items[1];
} state_t;

#define STATE_ALLOC(Nitems)						\
  (state_t *) xcalloc ((unsigned) (sizeof (state_t) 			\
                                  + (Nitems - 1) * sizeof (item_number_t)), 1)

#endif /* !STATE_H_ */