/* Generate the nondeterministic finite state machine for bison,
- Copyright 1984, 1986, 1989, 2000, 2001 Free Software Foundation, Inc.
+
+ Copyright (C) 1984, 1986, 1989, 2000, 2001, 2002 Free Software
+ Foundation, Inc.
This file is part of Bison, the GNU Compiler Compiler.
The entry point is generate_states. */
#include "system.h"
+#include "bitset.h"
+#include "quotearg.h"
+#include "symtab.h"
+#include "gram.h"
#include "getargs.h"
#include "reader.h"
#include "gram.h"
#include "lalr.h"
#include "reduce.h"
-int nstates;
-int final_state;
-static state_t *first_state = NULL;
-static shifts *first_shift = NULL;
+typedef struct state_list_s
+{
+ struct state_list_s *next;
+ state_t *state;
+} state_list_t;
-static state_t *this_state = NULL;
-static state_t *last_state = NULL;
-static shifts *last_shift = NULL;
+static state_list_t *first_state = NULL;
+static state_list_t *last_state = NULL;
-static int nshifts;
-static short *shift_symbol = NULL;
-static short *redset = NULL;
-static short *shiftset = NULL;
+/*------------------------------------------------------------------.
+| A state was just discovered from another state. Queue it for |
+| later examination, in order to find its transitions. Return it. |
+`------------------------------------------------------------------*/
-static short **kernel_base = NULL;
-static int *kernel_size = NULL;
-static short *kernel_items = NULL;
+static state_t *
+state_list_append (symbol_number_t symbol,
+ size_t core_size, item_number_t *core)
+{
+ state_list_t *node = XMALLOC (state_list_t, 1);
+ state_t *state = state_new (symbol, core_size, core);
-/* hash table for states, to recognize equivalent ones. */
+ if (trace_flag & trace_automaton)
+ fprintf (stderr, "state_list_append (state = %d, symbol = %d (%s))\n",
+ nstates, symbol, symbols[symbol]->tag);
-#define STATE_HASH_SIZE 1009
-static state_t **state_hash = NULL;
+ /* If this is the endtoken, and this is not the initial state, then
+ this is the final state. */
+ if (symbol == 0 && first_state)
+ final_state = state;
+
+ node->next = NULL;
+ node->state = state;
+
+ if (!first_state)
+ first_state = node;
+ if (last_state)
+ last_state->next = node;
+ last_state = node;
+
+ return state;
+}
+
+static int nshifts;
+static symbol_number_t *shift_symbol = NULL;
+
+static rule_t **redset = NULL;
+static state_t **shiftset = NULL;
+
+static item_number_t **kernel_base = NULL;
+static int *kernel_size = NULL;
+static item_number_t *kernel_items = NULL;
\f
static void
allocate_itemsets (void)
{
- int i;
+ symbol_number_t i;
+ rule_number_t r;
+ item_number_t *rhsp;
/* Count the number of occurrences of all the symbols in RITEMS.
Note that useless productions (hence useless nonterminals) are
int count = 0;
short *symbol_count = XCALLOC (short, nsyms + nuseless_nonterminals);
- for (i = 0; ritem[i]; ++i)
- if (ritem[i] > 0)
+ for (r = 0; r < nrules; ++r)
+ for (rhsp = rules[r].rhs; *rhsp >= 0; ++rhsp)
{
count++;
- symbol_count[ritem[i]]++;
+ symbol_count[*rhsp]++;
}
/* See comments before new_itemsets. All the vectors of items
live inside KERNEL_ITEMS. The number of active items after
some symbol cannot be more than the number of times that symbol
- appears as an item, which is symbol_count[symbol].
+ appears as an item, which is SYMBOL_COUNT[SYMBOL].
We allocate that much space for each symbol. */
- kernel_base = XCALLOC (short *, nsyms);
+ kernel_base = XCALLOC (item_number_t *, nsyms);
if (count)
- kernel_items = XCALLOC (short, count);
+ kernel_items = XCALLOC (item_number_t, count);
count = 0;
for (i = 0; i < nsyms; i++)
{
allocate_itemsets ();
- shiftset = XCALLOC (short, nsyms);
- redset = XCALLOC (short, nrules + 1);
- state_hash = XCALLOC (state_t *, STATE_HASH_SIZE);
+ shiftset = XCALLOC (state_t *, nsyms);
+ redset = XCALLOC (rule_t *, nrules);
+ state_hash_new ();
+ shift_symbol = XCALLOC (symbol_number_t, nsyms);
}
free (kernel_base);
free (kernel_size);
XFREE (kernel_items);
- free (state_hash);
+ state_hash_free ();
}
-/*----------------------------------------------------------------.
-| Find which symbols can be shifted in the current state, and for |
-| each one record which items would be active after that shift. |
-| Uses the contents of itemset. |
-| |
-| shift_symbol is set to a vector of the symbols that can be |
-| shifted. For each symbol in the grammar, kernel_base[symbol] |
-| points to a vector of item numbers activated if that symbol is |
-| shifted, and kernel_size[symbol] is their numbers. |
-`----------------------------------------------------------------*/
+/*---------------------------------------------------------------.
+| Find which symbols can be shifted in STATE, and for each one |
+| record which items would be active after that shift. Uses the |
+| contents of itemset. |
+| |
+| shift_symbol is set to a vector of the symbols that can be |
+| shifted. For each symbol in the grammar, kernel_base[symbol] |
+| points to a vector of item numbers activated if that symbol is |
+| shifted, and kernel_size[symbol] is their numbers. |
+`---------------------------------------------------------------*/
static void
-new_itemsets (void)
+new_itemsets (state_t *state)
{
int i;
- if (trace_flag)
+ if (trace_flag & trace_automaton)
fprintf (stderr, "Entering new_itemsets, state = %d\n",
- this_state->number);
+ state->number);
for (i = 0; i < nsyms; i++)
kernel_size[i] = 0;
- shift_symbol = XCALLOC (short, nsyms);
nshifts = 0;
- for (i = 0; i < nitemset; ++i)
- {
- int symbol = ritem[itemset[i]];
- if (symbol > 0)
- {
- if (!kernel_size[symbol])
- {
- shift_symbol[nshifts] = symbol;
- nshifts++;
- }
-
- kernel_base[symbol][kernel_size[symbol]] = itemset[i] + 1;
- kernel_size[symbol]++;
- }
- }
+ for (i = 0; i < nritemset; ++i)
+ if (ritem[itemset[i]] >= 0)
+ {
+ symbol_number_t symbol
+ = item_number_as_symbol_number (ritem[itemset[i]]);
+ if (!kernel_size[symbol])
+ {
+ shift_symbol[nshifts] = symbol;
+ nshifts++;
+ }
+
+ kernel_base[symbol][kernel_size[symbol]] = itemset[i] + 1;
+ kernel_size[symbol]++;
+ }
}
/*-----------------------------------------------------------------.
-| Subroutine of get_state. Create a new state for those items, if |
-| necessary. |
+| Find the state we would get to (from the current state) by |
+| shifting SYMBOL. Create a new state if no equivalent one exists |
+| already. Used by append_states. |
`-----------------------------------------------------------------*/
static state_t *
-new_state (int symbol)
+get_state (symbol_number_t symbol, size_t core_size, item_number_t *core)
{
- state_t *p;
-
- if (trace_flag)
- fprintf (stderr, "Entering new_state, state = %d, symbol = %d (%s)\n",
- this_state->number, symbol, tags[symbol]);
-
- if (nstates >= MAXSHORT)
- fatal (_("too many states (max %d)"), MAXSHORT);
-
- p = STATE_ALLOC (kernel_size[symbol]);
- p->accessing_symbol = symbol;
- p->number = nstates;
- p->nitems = kernel_size[symbol];
-
- shortcpy (p->items, kernel_base[symbol], kernel_size[symbol]);
-
- last_state->next = p;
- last_state = p;
- nstates++;
-
- return p;
-}
-
-
-/*--------------------------------------------------------------.
-| Find the state number for the state we would get to (from the |
-| current state) by shifting symbol. Create a new state if no |
-| equivalent one exists already. Used by append_states. |
-`--------------------------------------------------------------*/
-
-static int
-get_state (int symbol)
-{
- int key;
- int i;
state_t *sp;
- if (trace_flag)
- fprintf (stderr, "Entering get_state, state = %d, symbol = %d (%s)\n",
- this_state->number, symbol, tags[symbol]);
+ if (trace_flag & trace_automaton)
+ fprintf (stderr, "Entering get_state, symbol = %d (%s)\n",
+ symbol, symbols[symbol]->tag);
- /* Add up the target state's active item numbers to get a hash key.
- */
- key = 0;
- for (i = 0; i < kernel_size[symbol]; ++i)
- key += kernel_base[symbol][i];
- key = key % STATE_HASH_SIZE;
- sp = state_hash[key];
+ sp = state_hash_lookup (core_size, core);
+ if (!sp)
+ sp = state_list_append (symbol, core_size, core);
- if (sp)
- {
- int found = 0;
- while (!found)
- {
- if (sp->nitems == kernel_size[symbol])
- {
- found = 1;
- for (i = 0; i < kernel_size[symbol]; ++i)
- if (kernel_base[symbol][i] != sp->items[i])
- found = 0;
- }
-
- if (!found)
- {
- if (sp->link)
- {
- sp = sp->link;
- }
- else /* bucket exhausted and no match */
- {
- sp = sp->link = new_state (symbol);
- found = 1;
- }
- }
- }
- }
- else /* bucket is empty */
- {
- state_hash[key] = sp = new_state (symbol);
- }
-
- if (trace_flag)
+ if (trace_flag & trace_automaton)
fprintf (stderr, "Exiting get_state => %d\n", sp->number);
- return sp->number;
+ return sp;
}
-/*------------------------------------------------------------------.
-| Use the information computed by new_itemsets to find the state |
-| numbers reached by each shift transition from the current state. |
-| |
-| shiftset is set up as a vector of state numbers of those states. |
-`------------------------------------------------------------------*/
+/*---------------------------------------------------------------.
+| Use the information computed by new_itemsets to find the state |
+| numbers reached by each shift transition from STATE. |
+| |
+| SHIFTSET is set up as a vector of those states. |
+`---------------------------------------------------------------*/
static void
-append_states (void)
+append_states (state_t *state)
{
int i;
int j;
- int symbol;
+ symbol_number_t symbol;
- if (trace_flag)
+ if (trace_flag & trace_automaton)
fprintf (stderr, "Entering append_states, state = %d\n",
- this_state->number);
+ state->number);
/* first sort shift_symbol into increasing order */
}
for (i = 0; i < nshifts; i++)
- shiftset[i] = get_state (shift_symbol[i]);
-}
-
-
-static void
-new_states (void)
-{
- first_state = last_state = this_state = STATE_ALLOC (0);
- nstates = 1;
-}
-
-
-/*------------------------------------------------------------.
-| Save the NSHIFTS of SHIFTSET into the current linked list. |
-`------------------------------------------------------------*/
-
-static void
-save_shifts (void)
-{
- shifts *p = shifts_new (nshifts);
-
- p->number = this_state->number;
- shortcpy (p->shifts, shiftset, nshifts);
- this_state->shifts = p;
-
- if (last_shift)
- last_shift->next = p;
- else
- first_shift = p;
- last_shift = p;
-}
-
-
-/*------------------------------------------------------------------.
-| Subroutine of augment_automaton. Create the next-to-final state, |
-| to which a shift has already been made in the initial state. |
-| |
-| The task of this state consists in shifting (actually, it's a |
-| goto, but shifts and gotos are both stored in SHIFTS) the start |
-| symbols, hence the name. |
-`------------------------------------------------------------------*/
-
-static void
-insert_start_shifting_state (void)
-{
- state_t *statep;
- shifts *sp;
-
- statep = STATE_ALLOC (0);
- statep->number = nstates;
-
- /* The distinctive feature of this state from the
- eof_shifting_state, is that it is labeled as post-start-symbol
- shifting. I fail to understand why this state, and the
- post-start-start can't be merged into one. But it does fail if
- you try. --akim */
- statep->accessing_symbol = start_symbol;
-
- last_state->next = statep;
- last_state = statep;
-
- /* Make a shift from this state to (what will be) the final state. */
- sp = shifts_new (1);
- statep->shifts = sp;
- sp->number = nstates++;
- sp->shifts[0] = nstates;
-
- last_shift->next = sp;
- last_shift = sp;
-}
-
-
-/*-----------------------------------------------------------------.
-| Subroutine of augment_automaton. Create the final state, which |
-| shifts `0', the end of file. The initial state shifts the start |
-| symbol, and goes to here. |
-`-----------------------------------------------------------------*/
-
-static void
-insert_eof_shifting_state (void)
-{
- state_t *statep;
- shifts *sp;
-
- /* Make the final state--the one that follows a shift from the
- next-to-final state.
- The symbol for that shift is 0 (end-of-file). */
- statep = STATE_ALLOC (0);
- statep->number = nstates;
-
- last_state->next = statep;
- last_state = statep;
-
- /* Make the shift from the final state to the termination state. */
- sp = shifts_new (1);
- statep->shifts = sp;
- sp->number = nstates++;
- sp->shifts[0] = nstates;
-
- last_shift->next = sp;
- last_shift = sp;
-}
-
-
-/*---------------------------------------------------------------.
-| Subroutine of augment_automaton. Create the accepting state. |
-`---------------------------------------------------------------*/
-
-static void
-insert_accepting_state (void)
-{
- state_t *statep;
-
- /* Note that the variable `final_state' refers to what we sometimes
- call the termination state. */
- final_state = nstates;
-
- /* Make the termination state. */
- statep = STATE_ALLOC (0);
- statep->number = nstates++;
- last_state->next = statep;
- last_state = statep;
-}
-
-
-
-
-
-/*------------------------------------------------------------------.
-| Make sure that the initial state has a shift that accepts the |
-| grammar's start symbol and goes to the next-to-final state, which |
-| has a shift going to the final state, which has a shift to the |
-| termination state. Create such states and shifts if they don't |
-| happen to exist already. |
-`------------------------------------------------------------------*/
-
-static void
-augment_automaton (void)
-{
- if (!first_state->shifts->nshifts)
- {
- /* The first state has no shifts. Make one shift, from the
- initial state to the next-to-final state. */
-
- shifts *sp = shifts_new (1);
- first_state->shifts = sp;
- sp->shifts[0] = nstates;
-
- /* Initialize the chain of shifts with sp. */
- first_shift = sp;
- last_shift = sp;
-
- /* Create the next-to-final state, with shift to
- what will be the final state. */
- insert_start_shifting_state ();
- }
- else
{
- state_t *statep = first_state->next;
- shifts *sp = first_state->shifts;
- shifts *sp1 = NULL;
- /* The states reached by shifts from FIRST_STATE are numbered
- 1..(SP->NSHIFTS). Look for one reached by START_SYMBOL.
- This is typical of `start: start ... ;': there is a state
- with the item `start: start . ...'. We want to add a `shift
- on EOF to eof-shifting state here. */
- while (statep->accessing_symbol != start_symbol
- && statep->number < sp->nshifts)
- statep = statep->next;
-
- if (statep->accessing_symbol == start_symbol)
- {
- /* We already have a next-to-final state, i.e., for `start:
- start . ...'. Make sure it has a shift to what will be
- the final state. */
- while (sp && sp->number < statep->number)
- {
- sp1 = sp;
- sp = sp->next;
- }
-
- if (sp && sp->number == statep->number)
- {
- int i;
- shifts *sp2 = shifts_new (sp->nshifts + 1);
- sp2->number = statep->number;
- statep->shifts = sp2;
- sp2->shifts[0] = nstates;
- for (i = sp->nshifts; i > 0; i--)
- sp2->shifts[i] = sp->shifts[i - 1];
-
- /* Patch sp2 into the chain of shifts in place of sp,
- following sp1. */
- sp2->next = sp->next;
- sp1->next = sp2;
- if (sp == last_shift)
- last_shift = sp2;
- XFREE (sp);
- }
- else
- {
- shifts *sp2 = shifts_new (1);
- sp2->number = statep->number;
- statep->shifts = sp2;
- sp2->shifts[0] = nstates;
-
- /* Patch sp2 into the chain of shifts between sp1 and sp. */
- sp2->next = sp;
- sp1->next = sp2;
- if (sp == 0)
- last_shift = sp2;
- }
- insert_eof_shifting_state ();
- }
- else
- {
- /* There is no state for `start: start . ...'. */
- int i, k;
- shifts *sp2;
- sp = first_shift;
-
- /* There is no next-to-final state as yet. */
- /* Add one more shift in first_shift,
- going to the next-to-final state (yet to be made). */
- sp2 = shifts_new (sp->nshifts + 1);
- first_state->shifts = sp2;
- /* Stick this shift into the vector at the proper place. */
- statep = first_state->next;
- for (k = 0, i = 0; i < sp->nshifts; k++, i++)
- {
- if (statep->accessing_symbol > start_symbol && i == k)
- sp2->shifts[k++] = nstates;
- sp2->shifts[k] = sp->shifts[i];
- statep = statep->next;
- }
- if (i == k)
- sp2->shifts[k++] = nstates;
-
- /* Patch sp2 into the chain of shifts
- in place of sp, at the beginning. */
- sp2->next = sp->next;
- first_shift = sp2;
- if (last_shift == sp)
- last_shift = sp2;
-
- XFREE (sp);
-
- /* Create the next-to-final state, with shift to what will
- be the final state. Corresponds to `start: start . ...'. */
- insert_start_shifting_state ();
- }
+ symbol = shift_symbol[i];
+ shiftset[i] = get_state (symbol,
+ kernel_size[symbol], kernel_base[symbol]);
}
-
- insert_accepting_state ();
}
`----------------------------------------------------------------*/
static void
-save_reductions (void)
+save_reductions (state_t *state)
{
- int count;
+ int count = 0;
int i;
/* Find and count the active items that represent ends of rules. */
-
- count = 0;
- for (i = 0; i < nitemset; ++i)
+ for (i = 0; i < nritemset; ++i)
{
int item = ritem[itemset[i]];
if (item < 0)
- redset[count++] = -item;
+ redset[count++] = &rules[item_number_as_rule_number (item)];
}
/* Make a reductions structure and copy the data into it. */
-
- if (count)
- {
- reductions *p = REDUCTIONS_ALLOC (count);
- p->nreds = count;
- shortcpy (p->rules, redset, count);
-
- this_state->reductions = p;
- }
+ state_reductions_set (state, count, redset);
}
\f
-/*--------------------.
-| Build STATE_TABLE. |
-`--------------------*/
+/*---------------.
+| Build STATES. |
+`---------------*/
static void
-set_state_table (void)
+set_states (void)
{
- /* NSTATES + 1 because lookahead for the pseudo state number NSTATES
- might be used (see conflicts.c). It is too opaque for me to
- provide a probably less hacky implementation. --akim */
- state_table = XCALLOC (state_t *, nstates + 1);
-
- {
- state_t *sp;
- for (sp = first_state; sp; sp = sp->next)
- state_table[sp->number] = sp;
- }
-
- /* Pessimization, but simplification of the code: make sure all the
- states have a shifts, even if reduced to 0 shifts. */
- {
- int i;
- for (i = 0; i < nstates; i++)
- if (!state_table[i]->shifts)
- state_table[i]->shifts = shifts_new (0);
- }
-
- /* Initializing the lookaheads members. Please note that it must be
- performed after having set some of the other members which are
- used below. Change with extreme caution. */
- {
- int i;
- int count = 0;
- for (i = 0; i < nstates; i++)
- {
- int k;
- reductions *rp = state_table[i]->reductions;
- shifts *sp = state_table[i]->shifts;
-
- state_table[i]->lookaheads = count;
-
- if (rp
- && (rp->nreds > 1 || (sp->nshifts && SHIFT_IS_SHIFT (sp, 0))))
- count += rp->nreds;
- else
- state_table[i]->consistent = 1;
-
- for (k = 0; k < sp->nshifts; k++)
- if (SHIFT_IS_ERROR (sp, k))
- {
- state_table[i]->consistent = 0;
- break;
- }
- }
+ states = XCALLOC (state_t *, nstates);
- /* Seems to be needed by conflicts.c. */
- state_table[nstates] = STATE_ALLOC (0);
- state_table[nstates]->lookaheads = count;
- }
+ while (first_state)
+ {
+ state_list_t *this = first_state;
+
+ /* Pessimization, but simplification of the code: make sure all
+ the states have valid transitions and reductions members,
+ even if reduced to 0. It is too soon for errs, which are
+ computed later, but set_conflicts. */
+ state_t *state = this->state;
+ if (!state->transitions)
+ state_transitions_set (state, 0, 0);
+ if (!state->reductions)
+ state_reductions_set (state, 0, 0);
+
+ states[state->number] = state;
+
+ first_state = this->next;
+ free (this);
+ }
+ first_state = NULL;
+ last_state = NULL;
}
+
/*-------------------------------------------------------------------.
| Compute the nondeterministic finite state machine (see state.h for |
| details) from the grammar. |
void
generate_states (void)
{
+ state_list_t *list = NULL;
allocate_storage ();
- new_closure (nitems);
- new_states ();
+ new_closure (nritems);
+
+ /* Create the initial state. The 0 at the lhs is the index of the
+ item of this initial rule. */
+ kernel_base[0][0] = 0;
+ kernel_size[0] = 1;
+ state_list_append (0, kernel_size[0], kernel_base[0]);
- while (this_state)
+ list = first_state;
+
+ while (list)
{
- if (trace_flag)
+ state_t *state = list->state;
+ if (trace_flag & trace_automaton)
fprintf (stderr, "Processing state %d (reached by %s)\n",
- this_state->number, tags[this_state->accessing_symbol]);
+ state->number,
+ symbols[state->accessing_symbol]->tag);
/* Set up ruleset and itemset for the transitions out of this
state. ruleset gets a 1 bit for each rule that could reduce
now. itemset gets a vector of all the items that could be
accepted next. */
- closure (this_state->items, this_state->nitems);
- /* record the reductions allowed out of this state */
- save_reductions ();
- /* find the itemsets of the states that shifts can reach */
- new_itemsets ();
- /* find or create the core structures for those states */
- append_states ();
-
- /* create the shifts structures for the shifts to those states,
- now that the state numbers transitioning to are known */
- save_shifts ();
-
- /* states are queued when they are created; process them all */
- this_state = this_state->next;
+ closure (state->items, state->nitems);
+ /* Record the reductions allowed out of this state. */
+ save_reductions (state);
+ /* Find the itemsets of the states that shifts can reach. */
+ new_itemsets (state);
+ /* Find or create the core structures for those states. */
+ append_states (state);
+
+ /* Create the shifts structures for the shifts to those states,
+ now that the state numbers transitioning to are known. */
+ state_transitions_set (state, nshifts, shiftset);
+
+ /* States are queued when they are created; process them all.
+ */
+ list = list->next;
}
/* discard various storage */
free_closure ();
free_storage ();
- /* set up initial and final states as parser wants them */
- augment_automaton ();
-
- /* Set up STATE_TABLE. */
- set_state_table ();
+ /* Set up STATES. */
+ set_states ();
}
projects / bison.git / blobdiff