-/* Output the generated parsing program for bison,
- Copyright (C) 1984, 1986, 1989, 1992, 2000, 2001, 2002
- Free Software Foundation, Inc.
+/* Output the generated parsing program for Bison.
+
+ Copyright (C) 1984, 1986, 1989, 1992, 2000, 2001, 2002, 2003, 2004,
+ 2005, 2006, 2007, 2008 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.
+ This program 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 3 of the License, 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.
+ This program 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. */
-
-
-/* The parser tables consist of these tables. Marked ones needed only
- for the semantic parser. Double marked are output only if switches
- are set.
-
- YYTRANSLATE = vector mapping yylex's token numbers into bison's
- token numbers.
-
- ++ YYTNAME = vector of string-names indexed by bison token number.
-
- ++ YYTOKNUM = vector of yylex token numbers corresponding to
- entries in YYTNAME.
-
- YYRLINE = vector of line-numbers of all rules. For yydebug
- printouts.
-
- YYRHS = vector of items of all rules. This is exactly what RITEMS
- contains. For yydebug and for semantic parser.
-
- YYPRHS[R] = index in YYRHS of first item for rule R.
-
- YYR1[R] = symbol number of symbol that rule R derives.
-
- YYR2[R] = number of symbols composing right hand side of rule R.
-
- + YYSTOS[S] = the symbol number of the symbol that leads to state
- S.
-
- YYDEFACT[S] = default rule to reduce with in state s, when YYTABLE
- doesn't specify something else to do. Zero means the default is an
- error.
-
- YYDEFGOTO[I] = default state to go to after a reduction of a rule
- that generates variable NTOKENS + I, except when YYTABLE specifies
- something else to do.
-
- YYPACT[S] = index in YYTABLE of the portion describing state S.
- The lookahead token's type is used to index that portion to find
- out what to do.
-
- If the value in YYTABLE is positive, we shift the token and go to
- that state.
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
- If the value is negative, it is minus a rule number to reduce by.
-
- If the value is zero, the default action from YYDEFACT[S] is used.
-
- YYPGOTO[I] = the index in YYTABLE of the portion describing what to
- do after reducing a rule that derives variable I + NTOKENS. This
- portion is indexed by the parser state number, S, as of before the
- text for this nonterminal was read. The value from YYTABLE is the
- state to go to if the corresponding value in YYCHECK is S.
-
- YYTABLE = a vector filled with portions for different uses, found
- via YYPACT and YYPGOTO.
-
- YYCHECK = a vector indexed in parallel with YYTABLE. It indicates,
- in a roundabout way, the bounds of the portion you are trying to
- examine.
-
- Suppose that the portion of yytable starts at index P and the index
- to be examined within the portion is I. Then if YYCHECK[P+I] != I,
- I is outside the bounds of what is actually allocated, and the
- default (from YYDEFACT or YYDEFGOTO) should be used. Otherwise,
- YYTABLE[P+I] should be used.
+#include <config.h>
+#include "system.h"
- YYFINAL = the state number of the termination state. YYFLAG = most
- negative short int. Used to flag ?? */
+#include <configmake.h>
+#include <error.h>
+#include <get-errno.h>
+#include <quotearg.h>
+#include <subpipe.h>
+#include <timevar.h>
-#include "system.h"
-#include "bitsetv.h"
-#include "quotearg.h"
-#include "error.h"
-#include "getargs.h"
+#include "complain.h"
#include "files.h"
+#include "getargs.h"
#include "gram.h"
-#include "LR0.h"
-#include "complain.h"
+#include "muscle_tab.h"
#include "output.h"
-#include "lalr.h"
#include "reader.h"
+#include "scan-code.h" /* max_left_semantic_context */
+#include "scan-skel.h"
#include "symtab.h"
-#include "conflicts.h"
-#include "muscle_tab.h"
-
-/* From lib/readpipe.h. */
-FILE *readpipe PARAMS ((const char *, ...));
-
-/* From src/scan-skel.l. */
-int skel_lex PARAMS ((void));
-extern FILE *skel_in;
-
-static int nvectors;
-static int nentries;
-static short **froms = NULL;
-static short **tos = NULL;
-static short *tally = NULL;
-static short *width = NULL;
-static short *actrow = NULL;
-static short *state_count = NULL;
-static short *order = NULL;
-static short *base = NULL;
-static short *pos = NULL;
-
-/* TABLE_SIZE is the allocated size of both TABLE and CHECK.
- We start with the original hard-coded value: SHRT_MAX
- (yes, not USHRT_MAX). */
-static size_t table_size = SHRT_MAX;
-static short *table = NULL;
-static short *check = NULL;
-static int lowzero;
-static int high;
-
-struct obstack muscle_obstack;
-static struct obstack format_obstack;
+#include "tables.h"
-int error_verbose = 0;
-
-/*----------------------------------------------------------------.
-| If TABLE (and CHECK) appear to be small to be addressed at |
-| DESIRED, grow them. Note that TABLE[DESIRED] is to be used, so |
-| the desired size is at least DESIRED + 1. |
-`----------------------------------------------------------------*/
-
-static void
-table_grow (size_t desired)
-{
- size_t old_size = table_size;
-
- while (table_size <= desired)
- table_size *= 2;
-
- if (trace_flag)
- fprintf (stderr, "growing table and check from: %d to %d\n",
- old_size, table_size);
-
- table = XREALLOC (table, short, table_size);
- check = XREALLOC (check, short, table_size);
-
- for (/* Nothing. */; old_size < table_size; ++old_size)
- {
- table[old_size] = 0;
- check[old_size] = -1;
- }
-}
+static struct obstack format_obstack;
/*-------------------------------------------------------------------.
#define GENERATE_MUSCLE_INSERT_TABLE(Name, Type) \
\
static void \
-Name (const char *name, \
+Name (char const *name, \
Type *table_data, \
Type first, \
int begin, \
int end) \
{ \
- long int max = first; \
+ Type min = first; \
+ Type max = first; \
+ long int lmin; \
+ long int lmax; \
int i; \
int j = 1; \
\
else \
++j; \
obstack_fgrow1 (&format_obstack, "%6d", table_data[i]); \
- if (table_data[i] > max) \
+ if (table_data[i] < min) \
+ min = table_data[i]; \
+ if (max < table_data[i]) \
max = table_data[i]; \
} \
obstack_1grow (&format_obstack, 0); \
muscle_insert (name, obstack_finish (&format_obstack)); \
\
- /* Build `NAME_max' in the obstack. */ \
+ lmin = min; \
+ lmax = max; \
+ /* Build `NAME_min' and `NAME_max' in the obstack. */ \
+ obstack_fgrow1 (&format_obstack, "%s_min", name); \
+ obstack_1grow (&format_obstack, 0); \
+ MUSCLE_INSERT_LONG_INT (obstack_finish (&format_obstack), lmin); \
obstack_fgrow1 (&format_obstack, "%s_max", name); \
obstack_1grow (&format_obstack, 0); \
- MUSCLE_INSERT_LONG_INT (obstack_finish (&format_obstack), max); \
+ MUSCLE_INSERT_LONG_INT (obstack_finish (&format_obstack), lmax); \
}
-GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_int_table, int)
GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_unsigned_int_table, unsigned int)
-GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_short_table, short)
-GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_token_number_table, token_number_t)
-GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_item_number_table, item_number_t)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_int_table, int)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_base_table, base_number)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_rule_number_table, rule_number)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_symbol_number_table, symbol_number)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_item_number_table, item_number)
+GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_state_number_table, state_number)
-/*-----------------------------------------------------------------.
-| Prepare the muscles related to the tokens: translate, tname, and |
-| toknum. |
-`-----------------------------------------------------------------*/
+/*--------------------------------------------------------------------.
+| Print to OUT a representation of STRING escaped both for C and M4. |
+`--------------------------------------------------------------------*/
static void
-prepare_tokens (void)
+escaped_output (FILE *out, char const *string)
{
- muscle_insert_token_number_table ("translate",
- token_translations,
- 0, 1, max_user_token_number + 1);
+ char const *p;
+ fprintf (out, "[[");
+
+ for (p = quotearg_style (c_quoting_style, string); *p; p++)
+ switch (*p)
+ {
+ case '$': fputs ("$][", out); break;
+ case '@': fputs ("@@", out); break;
+ case '[': fputs ("@{", out); break;
+ case ']': fputs ("@}", out); break;
+ default: fputc (*p, out); break;
+ }
+
+ fprintf (out, "]]");
+}
+
+
+/*------------------------------------------------------------------.
+| Prepare the muscles related to the symbols: translate, tname, and |
+| toknum. |
+`------------------------------------------------------------------*/
+
+static void
+prepare_symbols (void)
+{
+ MUSCLE_INSERT_BOOL ("token_table", token_table_flag);
+ MUSCLE_INSERT_INT ("tokens_number", ntokens);
+ MUSCLE_INSERT_INT ("nterms_number", nvars);
+ MUSCLE_INSERT_INT ("undef_token_number", undeftoken->number);
+ MUSCLE_INSERT_INT ("user_token_number_max", max_user_token_number);
+
+ muscle_insert_symbol_number_table ("translate",
+ token_translations,
+ token_translations[0],
+ 1, max_user_token_number + 1);
+ /* tname -- token names. */
{
int i;
- int j = 0;
+ /* We assume that the table will be output starting at column 2. */
+ int j = 2;
+ struct quoting_options *qo = clone_quoting_options (0);
+ set_quoting_style (qo, c_quoting_style);
+ set_quoting_flags (qo, QA_SPLIT_TRIGRAPHS);
for (i = 0; i < nsyms; i++)
{
- /* Be sure not to use twice the same quotearg slot. */
- const char *cp =
- quotearg_n_style (1, c_quoting_style,
- quotearg_style (escape_quoting_style,
- symbols[i]->tag));
- /* Width of the next token, including the two quotes, the coma
- and the space. */
- int strsize = strlen (cp) + 2;
-
- if (j + strsize > 75)
+ char *cp = quotearg_alloc (symbols[i]->tag, -1, qo);
+ /* Width of the next token, including the two quotes, the
+ comma and the space. */
+ int width = strlen (cp) + 2;
+
+ if (j + width > 75)
{
- obstack_sgrow (&format_obstack, "\n ");
- j = 2;
+ obstack_sgrow (&format_obstack, "\n ");
+ j = 1;
}
- obstack_sgrow (&format_obstack, cp);
- obstack_sgrow (&format_obstack, ", ");
- j += strsize;
+ if (i)
+ obstack_1grow (&format_obstack, ' ');
+ MUSCLE_OBSTACK_SGROW (&format_obstack, cp);
+ free (cp);
+ obstack_1grow (&format_obstack, ',');
+ j += width;
}
- /* Add a NULL entry to list of tokens (well, 0, as NULL might not be
- defined). */
- obstack_sgrow (&format_obstack, "0");
+ free (qo);
+ obstack_sgrow (&format_obstack, " ]b4_null[");
/* Finish table and store. */
obstack_1grow (&format_obstack, 0);
muscle_insert ("tname", obstack_finish (&format_obstack));
}
- /* Output YYTOKNUM. */
+ /* Output YYTOKNUM. */
{
int i;
- short *values = XCALLOC (short, ntokens + 1);
- for (i = 0; i < ntokens + 1; ++i)
+ int *values = xnmalloc (ntokens, sizeof *values);
+ for (i = 0; i < ntokens; ++i)
values[i] = symbols[i]->user_token_number;
- muscle_insert_short_table ("toknum", values,
- 0, 1, ntokens + 1);
+ muscle_insert_int_table ("toknum", values,
+ values[0], 1, ntokens);
free (values);
}
}
/*-------------------------------------------------------------.
| Prepare the muscles related to the rules: rhs, prhs, r1, r2, |
-| rline. |
+| rline, dprec, merger. |
`-------------------------------------------------------------*/
static void
prepare_rules (void)
{
- long int max;
- int r;
+ rule_number r;
unsigned int i = 0;
- item_number_t *rhs = XMALLOC (item_number_t, nritems);
- unsigned int *prhs = XMALLOC (unsigned int, nrules + 1);
- unsigned int *rline = XMALLOC (unsigned int, nrules + 1);
- token_number_t *r1 = XMALLOC (token_number_t, nrules + 1);
- unsigned int *r2 = XMALLOC (unsigned int, nrules + 1);
-
- for (r = 1; r < nrules + 1; ++r)
+ item_number *rhs = xnmalloc (nritems, sizeof *rhs);
+ unsigned int *prhs = xnmalloc (nrules, sizeof *prhs);
+ unsigned int *rline = xnmalloc (nrules, sizeof *rline);
+ symbol_number *r1 = xnmalloc (nrules, sizeof *r1);
+ unsigned int *r2 = xnmalloc (nrules, sizeof *r2);
+ int *dprec = xnmalloc (nrules, sizeof *dprec);
+ int *merger = xnmalloc (nrules, sizeof *merger);
+
+ for (r = 0; r < nrules; ++r)
{
- item_number_t *rhsp;
+ item_number *rhsp = NULL;
/* Index of rule R in RHS. */
prhs[r] = i;
/* RHS of the rule R. */
/* Separator in RHS. */
rhs[i++] = -1;
/* Line where rule was defined. */
- rline[r] = rules[r].line;
+ rline[r] = rules[r].location.start.line;
+ /* Dynamic precedence (GLR). */
+ dprec[r] = rules[r].dprec;
+ /* Merger-function index (GLR). */
+ merger[r] = rules[r].merger;
}
- assert (i == nritems);
+ aver (i == nritems);
muscle_insert_item_number_table ("rhs", rhs, ritem[0], 1, nritems);
- muscle_insert_unsigned_int_table ("prhs", prhs, 0, 1, nrules + 1);
- muscle_insert_unsigned_int_table ("rline", rline, 0, 1, nrules + 1);
- muscle_insert_token_number_table ("r1", r1, 0, 1, nrules + 1);
- muscle_insert_unsigned_int_table ("r2", r2, 0, 1, nrules + 1);
+ muscle_insert_unsigned_int_table ("prhs", prhs, 0, 0, nrules);
+ muscle_insert_unsigned_int_table ("rline", rline, 0, 0, nrules);
+ muscle_insert_symbol_number_table ("r1", r1, 0, 0, nrules);
+ muscle_insert_unsigned_int_table ("r2", r2, 0, 0, nrules);
+ muscle_insert_int_table ("dprec", dprec, 0, 0, nrules);
+ muscle_insert_int_table ("merger", merger, 0, 0, nrules);
+
+ MUSCLE_INSERT_INT ("rules_number", nrules);
+ MUSCLE_INSERT_INT ("max_left_semantic_context", max_left_semantic_context);
free (rhs);
free (prhs);
free (rline);
free (r1);
free (r2);
+ free (dprec);
+ free (merger);
}
/*--------------------------------------------.
static void
prepare_states (void)
{
- size_t i;
- token_number_t *values =
- (token_number_t *) alloca (sizeof (token_number_t) * nstates);
+ state_number i;
+ symbol_number *values = xnmalloc (nstates, sizeof *values);
for (i = 0; i < nstates; ++i)
values[i] = states[i]->accessing_symbol;
- muscle_insert_token_number_table ("stos", values,
- 0, 1, nstates);
+ muscle_insert_symbol_number_table ("stos", values,
+ 0, 1, nstates);
+ free (values);
+
+ MUSCLE_INSERT_INT ("last", high);
+ MUSCLE_INSERT_INT ("final_state_number", final_state->number);
+ MUSCLE_INSERT_INT ("states_number", nstates);
}
-/*------------------------------------------------------------------.
-| Decide what to do for each type of token if seen as the lookahead |
-| token in specified state. The value returned is used as the |
-| default action (yydefact) for the state. In addition, actrow is |
-| filled with what to do for each kind of token, index by symbol |
-| number, with zero meaning do the default action. The value |
-| SHRT_MIN, a very negative number, means this situation is an |
-| error. The parser recognizes this value specially. |
-| |
-| This is where conflicts are resolved. The loop over lookahead |
-| rules considered lower-numbered rules last, and the last rule |
-| considered that likes a token gets to handle it. |
-`------------------------------------------------------------------*/
-static int
-action_row (state_t *state)
+/*-----------------------------------------------.
+| For each symbol type, its tags and type name. |
+`-----------------------------------------------*/
+
+static void
+type_names_output (FILE *out)
{
int i;
- int default_rule = 0;
- reductions *redp = state->reductions;
- shifts *shiftp = state->shifts;
- errs *errp = state->errs;
- /* set nonzero to inhibit having any default reduction */
- int nodefault = 0;
-
- for (i = 0; i < ntokens; i++)
- actrow[i] = 0;
-
- if (redp->nreds >= 1)
- {
- int j;
- /* loop over all the rules available here which require
- lookahead */
- for (i = state->nlookaheads - 1; i >= 0; --i)
- /* and find each token which the rule finds acceptable
- to come next */
- for (j = 0; j < ntokens; j++)
- /* and record this rule as the rule to use if that
- token follows. */
- if (bitset_test (LA[state->lookaheadsp + i], j))
- actrow[j] = -LArule[state->lookaheadsp + i]->number;
- }
-
- /* Now see which tokens are allowed for shifts in this state. For
- them, record the shift as the thing to do. So shift is preferred
- to reduce. */
- for (i = 0; i < shiftp->nshifts; i++)
- {
- token_number_t symbol;
- int shift_state = shiftp->shifts[i];
- if (!shift_state)
- continue;
-
- symbol = states[shift_state]->accessing_symbol;
-
- if (ISVAR (symbol))
- break;
-
- actrow[symbol] = shift_state;
- /* Do not use any default reduction if there is a shift for
- error */
- if (symbol == errtoken->number)
- nodefault = 1;
- }
-
- /* See which tokens are an explicit error in this state (due to
- %nonassoc). For them, record SHRT_MIN as the action. */
- for (i = 0; i < errp->nerrs; i++)
+ fputs ("m4_define([b4_type_names],\n[", out);
+ for (i = 0; i < nsyms; ++i)
{
- int symbol = errp->errs[i];
- actrow[symbol] = SHRT_MIN;
+ symbol *sym = symbols[i];
+ /* Symbol-name, Symbol-number, optional typename. */
+ fprintf (out, "%s[", i ? ",\n" : "");
+ escaped_output (out, sym->tag);
+ fprintf (out, ", %d, [[%s]]]",
+ sym->number,
+ sym->type_name ? sym->type_name : "");
}
-
- /* Now find the most common reduction and make it the default action
- for this state. */
-
- if (redp->nreds >= 1 && !nodefault)
- {
- if (state->consistent)
- default_rule = redp->rules[0];
- else
- {
- int max = 0;
- for (i = 0; i < state->nlookaheads; i++)
- {
- int count = 0;
- int rule = -LArule[state->lookaheadsp + i]->number;
- int j;
-
- for (j = 0; j < ntokens; j++)
- if (actrow[j] == rule)
- count++;
-
- if (count > max)
- {
- max = count;
- default_rule = rule;
- }
- }
-
- /* actions which match the default are replaced with zero,
- which means "use the default" */
-
- if (max > 0)
- {
- int j;
- for (j = 0; j < ntokens; j++)
- if (actrow[j] == default_rule)
- actrow[j] = 0;
-
- default_rule = -default_rule;
- }
- }
- }
-
- /* If have no default rule, the default is an error.
- So replace any action which says "error" with "use default". */
-
- if (default_rule == 0)
- for (i = 0; i < ntokens; i++)
- if (actrow[i] == SHRT_MIN)
- actrow[i] = 0;
-
- return default_rule;
+ fputs ("])\n\n", out);
}
+/*---------------------------------.
+| Output the user actions to OUT. |
+`---------------------------------*/
+
static void
-save_row (int state)
+user_actions_output (FILE *out)
{
- int i;
- int count;
- short *sp;
- short *sp1;
- short *sp2;
-
- count = 0;
- for (i = 0; i < ntokens; i++)
- if (actrow[i] != 0)
- count++;
-
- if (count == 0)
- return;
+ rule_number r;
- froms[state] = sp1 = sp = XCALLOC (short, count);
- tos[state] = sp2 = XCALLOC (short, count);
-
- for (i = 0; i < ntokens; i++)
- if (actrow[i] != 0)
+ fputs ("m4_define([b4_actions], \n[", out);
+ for (r = 0; r < nrules; ++r)
+ if (rules[r].action)
{
- *sp1++ = i;
- *sp2++ = actrow[i];
+ fprintf (out, "b4_case(%d, [b4_syncline(%d, ", r + 1,
+ rules[r].action_location.start.line);
+ escaped_output (out, rules[r].action_location.start.file);
+ fprintf (out, ")\n[ %s]])\n\n", rules[r].action);
}
-
- tally[state] = count;
- width[state] = sp1[-1] - sp[0] + 1;
+ fputs ("])\n\n", out);
}
-
-/*------------------------------------------------------------------.
-| Figure out the actions for the specified state, indexed by |
-| lookahead token type. |
-| |
-| The YYDEFACT table is output now. The detailed info is saved for |
-| putting into YYTABLE later. |
-`------------------------------------------------------------------*/
+/*------------------------------------.
+| Output the merge functions to OUT. |
+`------------------------------------*/
static void
-token_actions (void)
+merger_output (FILE *out)
{
- size_t i;
- short *yydefact = XCALLOC (short, nstates);
+ int n;
+ merger_list* p;
- actrow = XCALLOC (short, ntokens);
- for (i = 0; i < nstates; ++i)
+ fputs ("m4_define([b4_mergers], \n[[", out);
+ for (n = 1, p = merge_functions; p != NULL; n += 1, p = p->next)
{
- yydefact[i] = action_row (states[i]);
- save_row (i);
+ if (p->type[0] == '\0')
+ fprintf (out, " case %d: *yy0 = %s (*yy0, *yy1); break;\n",
+ n, p->name);
+ else
+ fprintf (out, " case %d: yy0->%s = %s (*yy0, *yy1); break;\n",
+ n, p->type, p->name);
}
-
- muscle_insert_short_table ("defact", yydefact,
- yydefact[0], 1, nstates);
- XFREE (actrow);
- XFREE (yydefact);
-}
-
-
-/*-----------------------------.
-| Output the actions to OOUT. |
-`-----------------------------*/
-
-void
-actions_output (FILE *out)
-{
- int rule;
- for (rule = 1; rule < nrules + 1; ++rule)
- if (rules[rule].action)
- {
- fprintf (out, " case %d:\n", rule);
-
- if (!no_lines_flag)
- fprintf (out, muscle_find ("linef"),
- rules[rule].action_line,
- quotearg_style (c_quoting_style,
- muscle_find ("filename")));
- /* As a Bison extension, add the ending semicolon. Since some
- Yacc don't do that, help people using bison as a Yacc
- finding their missing semicolons. */
- fprintf (out, "{ %s%s }\n break;\n\n",
- rules[rule].action,
- yacc_flag ? ";" : "");
- }
+ fputs ("]])\n\n", out);
}
+/*--------------------------------------.
+| Output the tokens definition to OUT. |
+`--------------------------------------*/
-/*---------------------------------------.
-| Output the tokens definition to OOUT. |
-`---------------------------------------*/
-
-void
+static void
token_definitions_output (FILE *out)
{
int i;
- int first = 1;
+ char const *sep = "";
+
+ fputs ("m4_define([b4_tokens], \n[", out);
for (i = 0; i < ntokens; ++i)
{
- symbol_t *symbol = symbols[i];
- int number = symbol->user_token_number;
+ symbol *sym = symbols[i];
+ int number = sym->user_token_number;
/* At this stage, if there are literal aliases, they are part of
SYMBOLS, so we should not find symbols which are the aliases
here. */
- assert (number != USER_NUMBER_ALIAS);
+ aver (number != USER_NUMBER_ALIAS);
/* Skip error token. */
- if (symbol == errtoken)
+ if (sym == errtoken)
continue;
/* If this string has an alias, then it is necessarily the alias
which is to be output. */
- if (symbol->alias)
- symbol = symbol->alias;
+ if (sym->alias)
+ sym = sym->alias;
/* Don't output literal chars or strings (when defined only as a
string). Note that must be done after the alias resolution:
think about `%token 'f' "f"'. */
- if (symbol->tag[0] == '\'' || symbol->tag[0] == '\"')
+ if (sym->tag[0] == '\'' || sym->tag[0] == '\"')
continue;
/* Don't #define nonliteral tokens whose names contain periods
or '$' (as does the default value of the EOF token). */
- if (strchr (symbol->tag, '.') || strchr (symbol->tag, '$'))
+ if (strchr (sym->tag, '.') || strchr (sym->tag, '$'))
continue;
- fprintf (out, "%s[[[%s]], [%d]]",
- first ? "" : ",\n", symbol->tag, number);
-
- first = 0;
- }
-}
-
-
-static void
-save_column (int symbol, int default_state)
-{
- int i;
- short *sp;
- short *sp1;
- short *sp2;
- int count;
- int symno = symbol - ntokens + nstates;
-
- short begin = goto_map[symbol];
- short end = goto_map[symbol + 1];
-
- count = 0;
- for (i = begin; i < end; i++)
- if (to_state[i] != default_state)
- count++;
-
- if (count == 0)
- return;
-
- froms[symno] = sp1 = sp = XCALLOC (short, count);
- tos[symno] = sp2 = XCALLOC (short, count);
-
- for (i = begin; i < end; i++)
- if (to_state[i] != default_state)
- {
- *sp1++ = from_state[i];
- *sp2++ = to_state[i];
- }
-
- tally[symno] = count;
- width[symno] = sp1[-1] - sp[0] + 1;
-}
-
-static int
-default_goto (int symbol)
-{
- size_t i;
- size_t m = goto_map[symbol];
- size_t n = goto_map[symbol + 1];
- int default_state = -1;
- int max = 0;
-
- if (m == n)
- return -1;
-
- for (i = 0; i < nstates; i++)
- state_count[i] = 0;
-
- for (i = m; i < n; i++)
- state_count[to_state[i]]++;
-
- for (i = 0; i < nstates; i++)
- if (state_count[i] > max)
- {
- max = state_count[i];
- default_state = i;
- }
-
- return default_state;
-}
-
-
-/*-------------------------------------------------------------------.
-| Figure out what to do after reducing with each rule, depending on |
-| the saved state from before the beginning of parsing the data that |
-| matched this rule. |
-| |
-| The YYDEFGOTO table is output now. The detailed info is saved for |
-| putting into YYTABLE later. |
-`-------------------------------------------------------------------*/
-
-static void
-goto_actions (void)
-{
- int i;
- short *yydefgoto = XMALLOC (short, nsyms - ntokens);
-
- state_count = XCALLOC (short, nstates);
- for (i = ntokens; i < nsyms; ++i)
- {
- int default_state = default_goto (i);
- save_column (i, default_state);
- yydefgoto[i - ntokens] = default_state;
+ fprintf (out, "%s[[[%s]], %d]",
+ sep, sym->tag, number);
+ sep = ",\n";
}
-
- muscle_insert_short_table ("defgoto", yydefgoto,
- yydefgoto[0], 1, nsyms - ntokens);
- XFREE (state_count);
- XFREE (yydefgoto);
+ fputs ("])\n\n", out);
}
-/* The next few functions decide how to pack the actions and gotos
- information into yytable. */
+/*---------------------------------------------------.
+| Output the symbol destructors or printers to OUT. |
+`---------------------------------------------------*/
static void
-sort_actions (void)
+symbol_code_props_output (FILE *out, char const *what,
+ code_props const *(*get)(symbol const *))
{
int i;
+ char const *sep = "";
- order = XCALLOC (short, nvectors);
- nentries = 0;
-
- for (i = 0; i < nvectors; i++)
- if (tally[i] > 0)
- {
- int k;
- int t = tally[i];
- int w = width[i];
- int j = nentries - 1;
-
- while (j >= 0 && (width[order[j]] < w))
- j--;
-
- while (j >= 0 && (width[order[j]] == w) && (tally[order[j]] < t))
- j--;
-
- for (k = nentries - 1; k > j; k--)
- order[k + 1] = order[k];
-
- order[j + 1] = i;
- nentries++;
- }
-}
-
-
-static int
-matching_state (int vector)
-{
- int i = order[vector];
- int t;
- int w;
- int prev;
-
- if (i >= (int) nstates)
- return -1;
-
- t = tally[i];
- w = width[i];
-
- for (prev = vector - 1; prev >= 0; prev--)
+ fputs ("m4_define([b4_symbol_", out);
+ fputs (what, out);
+ fputs ("], \n[", out);
+ for (i = 0; i < nsyms; ++i)
{
- int j = order[prev];
- int k;
- int match = 1;
-
- if (width[j] != w || tally[j] != t)
- return -1;
-
- for (k = 0; match && k < t; k++)
- if (tos[j][k] != tos[i][k] || froms[j][k] != froms[i][k])
- match = 0;
-
- if (match)
- return j;
+ symbol *sym = symbols[i];
+ char const *code = (*get) (sym)->code;
+ if (code)
+ {
+ location loc = (*get) (sym)->location;
+ /* Filename, lineno,
+ Symbol-name, Symbol-number,
+ code, optional typename. */
+ fprintf (out, "%s[", sep);
+ sep = ",\n";
+ escaped_output (out, loc.start.file);
+ fprintf (out, ", %d, ", loc.start.line);
+ escaped_output (out, sym->tag);
+ fprintf (out, ", %d, [[%s]]", sym->number, code);
+ if (sym->type_name)
+ fprintf (out, ", [[%s]]", sym->type_name);
+ fputc (']', out);
+ }
}
-
- return -1;
-}
-
-
-static int
-pack_vector (int vector)
-{
- int i = order[vector];
- int j;
- int t = tally[i];
- int loc = 0;
- short *from = froms[i];
- short *to = tos[i];
-
- assert (t);
-
- for (j = lowzero - from[0]; j < (int) table_size; j++)
- {
- int k;
- int ok = 1;
-
- for (k = 0; ok && k < t; k++)
- {
- loc = j + from[k];
- if (loc > (int) table_size)
- table_grow (loc);
-
- if (table[loc] != 0)
- ok = 0;
- }
-
- for (k = 0; ok && k < vector; k++)
- if (pos[k] == j)
- ok = 0;
-
- if (ok)
- {
- for (k = 0; k < t; k++)
- {
- loc = j + from[k];
- table[loc] = to[k];
- check[loc] = from[k];
- }
-
- while (table[lowzero] != 0)
- lowzero++;
-
- if (loc > high)
- high = loc;
-
- return j;
- }
- }
-#define pack_vector_succeeded 0
- assert (pack_vector_succeeded);
- return 0;
+ fputs ("])\n\n", out);
}
static void
-pack_table (void)
+prepare_actions (void)
{
- int i;
- int place;
- int state;
-
- base = XCALLOC (short, nvectors);
- pos = XCALLOC (short, nentries);
- table = XCALLOC (short, table_size);
- check = XCALLOC (short, table_size);
+ /* Figure out the actions for the specified state, indexed by
+ lookahead token type. */
- lowzero = 0;
- high = 0;
-
- for (i = 0; i < nvectors; i++)
- base[i] = SHRT_MIN;
-
- for (i = 0; i < (int) table_size; i++)
- check[i] = -1;
-
- for (i = 0; i < nentries; i++)
- {
- state = matching_state (i);
-
- if (state < 0)
- place = pack_vector (i);
- else
- place = base[state];
-
- pos[i] = place;
- base[order[i]] = place;
- }
-
- for (i = 0; i < nvectors; i++)
- {
- XFREE (froms[i]);
- XFREE (tos[i]);
- }
+ muscle_insert_rule_number_table ("defact", yydefact,
+ yydefact[0], 1, nstates);
- XFREE (froms);
- XFREE (tos);
- XFREE (pos);
-}
+ /* Figure out what to do after reducing with each rule, depending on
+ the saved state from before the beginning of parsing the data
+ that matched this rule. */
+ muscle_insert_state_number_table ("defgoto", yydefgoto,
+ yydefgoto[0], 1, nsyms - ntokens);
-/* the following functions output yytable, yycheck
- and the vectors whose elements index the portion starts */
-static void
-output_base (void)
-{
- /* Output pact. */
- muscle_insert_short_table ("pact", base,
+ /* Output PACT. */
+ muscle_insert_base_table ("pact", base,
base[0], 1, nstates);
+ MUSCLE_INSERT_INT ("pact_ninf", base_ninf);
- /* Output pgoto. */
- muscle_insert_short_table ("pgoto", base,
+ /* Output PGOTO. */
+ muscle_insert_base_table ("pgoto", base,
base[nstates], nstates + 1, nvectors);
- XFREE (base);
-}
-
-
-static void
-output_table (void)
-{
- muscle_insert_short_table ("table", table,
- table[0], 1, high + 1);
- XFREE (table);
-}
-
-
-static void
-output_check (void)
-{
- muscle_insert_short_table ("check", check,
- check[0], 1, high + 1);
- XFREE (check);
-}
-
-/*-----------------------------------------------------------------.
-| Compute and output yydefact, yydefgoto, yypact, yypgoto, yytable |
-| and yycheck. |
-`-----------------------------------------------------------------*/
-static void
-output_actions (void)
-{
- size_t i;
- nvectors = nstates + nvars;
-
- froms = XCALLOC (short *, nvectors);
- tos = XCALLOC (short *, nvectors);
- tally = XCALLOC (short, nvectors);
- width = XCALLOC (short, nvectors);
-
- token_actions ();
- bitsetv_free (LA);
- free (LArule);
-
- goto_actions ();
- XFREE (goto_map + ntokens);
- XFREE (from_state);
- XFREE (to_state);
-
- sort_actions ();
- pack_table ();
-
- output_base ();
- output_table ();
-
- output_check ();
-
- for (i = 0; i < nstates; ++i)
- {
- free (states[i]->shifts);
- XFREE (states[i]->reductions);
- free (states[i]->errs);
- free (states[i]);
- }
- XFREE (states);
+ muscle_insert_base_table ("table", table,
+ table[0], 1, high + 1);
+ MUSCLE_INSERT_INT ("table_ninf", table_ninf);
+
+ muscle_insert_base_table ("check", check,
+ check[0], 1, high + 1);
+
+ /* GLR parsing slightly modifies YYTABLE and YYCHECK (and thus
+ YYPACT) so that in states with unresolved conflicts, the default
+ reduction is not used in the conflicted entries, so that there is
+ a place to put a conflict pointer.
+
+ This means that YYCONFLP and YYCONFL are nonsense for a non-GLR
+ parser, so we could avoid accidents by not writing them out in
+ that case. Nevertheless, it seems even better to be able to use
+ the GLR skeletons even without the non-deterministic tables. */
+ muscle_insert_unsigned_int_table ("conflict_list_heads", conflict_table,
+ conflict_table[0], 1, high + 1);
+ muscle_insert_unsigned_int_table ("conflicting_rules", conflict_list,
+ 0, 1, conflict_list_cnt);
}
-
\f
/*---------------------------.
| Call the skeleton parser. |
static void
output_skeleton (void)
{
- /* Store the definition of all the muscles. */
- const char *tempdir = getenv ("TMPDIR");
- char *tempfile = NULL;
- FILE *out = NULL;
- int fd;
-
- if (tempdir == NULL)
- tempdir = DEFAULT_TMPDIR;
- tempfile = xmalloc (strlen (tempdir) + 11);
- sprintf (tempfile, "%s/bsnXXXXXX", tempdir);
- fd = mkstemp (tempfile);
- if (fd == -1)
- error (EXIT_FAILURE, errno, "%s", tempfile);
-
- out = fdopen (fd, "w");
- if (out == NULL)
- error (EXIT_FAILURE, errno, "%s", tempfile);
-
- /* There are no comments, especially not `#': we do want M4 expansion
- after `#': think of CPP macros! */
- fputs ("m4_changecom()\n", out);
+ FILE *in;
+ FILE *out;
+ int filter_fd[2];
+ char const *argv[9];
+ pid_t pid;
+
+ /* Compute the names of the package data dir and skeleton files. */
+ char const m4sugar[] = "m4sugar/m4sugar.m4";
+ char const m4bison[] = "bison.m4";
+ char *full_m4sugar;
+ char *full_m4bison;
+ char *full_skeleton;
+ char const *p;
+ char const *m4 = (p = getenv ("M4")) ? p : M4;
+ char const *pkgdatadir = compute_pkgdatadir ();
+ size_t skeleton_size = strlen (skeleton) + 1;
+ size_t pkgdatadirlen = strlen (pkgdatadir);
+ while (pkgdatadirlen && pkgdatadir[pkgdatadirlen - 1] == '/')
+ pkgdatadirlen--;
+ full_skeleton = xmalloc (pkgdatadirlen + 1
+ + (skeleton_size < sizeof m4sugar
+ ? sizeof m4sugar : skeleton_size));
+ strncpy (full_skeleton, pkgdatadir, pkgdatadirlen);
+ full_skeleton[pkgdatadirlen] = '/';
+ strcpy (full_skeleton + pkgdatadirlen + 1, m4sugar);
+ full_m4sugar = xstrdup (full_skeleton);
+ strcpy (full_skeleton + pkgdatadirlen + 1, m4bison);
+ full_m4bison = xstrdup (full_skeleton);
+ if (strchr (skeleton, '/'))
+ strcpy (full_skeleton, skeleton);
+ else
+ strcpy (full_skeleton + pkgdatadirlen + 1, skeleton);
+
+ /* Test whether m4sugar.m4 is readable, to check for proper
+ installation. A faulty installation can cause deadlock, so a
+ cheap sanity check is worthwhile. */
+ xfclose (xfopen (full_m4sugar, "r"));
+
+ /* Create an m4 subprocess connected to us via two pipes. */
+
+ if (trace_flag & trace_tools)
+ fprintf (stderr, "running: %s %s - %s %s\n",
+ m4, full_m4sugar, full_m4bison, full_skeleton);
+
+ /* Some future version of GNU M4 (most likely 1.6) may treat the -dV in a
+ position-dependent manner. Keep it as the first argument so that all
+ files are traced.
+
+ See the thread starting at
+ <http://lists.gnu.org/archive/html/bug-bison/2008-07/msg00000.html>
+ for details. */
+ {
+ int i = 0;
+ argv[i++] = m4;
+ argv[i++] = "-I";
+ argv[i++] = pkgdatadir;
+ if (trace_flag & trace_m4)
+ argv[i++] = "-dV";
+ argv[i++] = full_m4sugar;
+ argv[i++] = "-";
+ argv[i++] = full_m4bison;
+ argv[i++] = full_skeleton;
+ argv[i++] = NULL;
+ }
+ /* When POSIXLY_CORRECT is set, some future versions of GNU M4 (most likely
+ 2.0) may drop some of the GNU extensions that Bison's skeletons depend
+ upon. So that the next release of Bison is forward compatible with those
+ future versions of GNU M4, we unset POSIXLY_CORRECT here.
+
+ FIXME: A user might set POSIXLY_CORRECT to affect processes run from
+ macros like m4_syscmd in a custom skeleton. For now, Bison makes no
+ promises about the behavior of custom skeletons, so this scenario is not a
+ concern. However, we eventually want to eliminate this shortcoming. The
+ next release of GNU M4 (1.4.12 or 1.6) will accept the -g command-line
+ option as a no-op, and later releases will accept it to indicate that
+ POSIXLY_CORRECT should be ignored. Once the GNU M4 versions that accept
+ -g are pervasive, Bison should use -g instead of unsetting
+ POSIXLY_CORRECT.
+
+ See the thread starting at
+ <http://lists.gnu.org/archive/html/bug-bison/2008-07/msg00000.html>
+ for details. */
+ unsetenv ("POSIXLY_CORRECT");
+ init_subpipe ();
+ pid = create_subpipe (argv, filter_fd);
+ free (full_m4sugar);
+ free (full_m4bison);
+ free (full_skeleton);
+
+ out = fdopen (filter_fd[0], "w");
+ if (! out)
+ error (EXIT_FAILURE, get_errno (),
+ "fdopen");
+
+ /* Output the definitions of all the muscles. */
fputs ("m4_init()\n", out);
- fputs ("m4_define([b4_actions], \n[[", out);
- actions_output (out);
- fputs ("]])\n\n", out);
-
- fputs ("m4_define([b4_tokens], \n[", out);
+ type_names_output (out);
+ user_actions_output (out);
+ merger_output (out);
token_definitions_output (out);
- fputs ("])\n\n", out);
+ symbol_code_props_output (out, "destructors", &symbol_destructor_get);
+ symbol_code_props_output (out, "printers", &symbol_printer_get);
muscles_m4_output (out);
-
- fputs ("m4_wrap([m4_divert_pop(0)])\n", out);
- fputs ("m4_divert_push(0)dnl\n", out);
xfclose (out);
- /* Invoke m4 on the definition of the muscles, and the skeleton. */
- {
- const char *bison_pkgdatadir = getenv ("BISON_PKGDATADIR");
- const char *m4 = getenv ("M4");
- if (!m4)
- m4 = M4;
- if (!bison_pkgdatadir)
- bison_pkgdatadir = PKGDATADIR;
- if (trace_flag)
- fprintf (stderr,
- "running: %s -I %s m4sugar/m4sugar.m4 %s %s\n",
- m4, bison_pkgdatadir, tempfile, skeleton);
- skel_in = readpipe (m4,
- "-I", bison_pkgdatadir,
- "m4sugar/m4sugar.m4",
- tempfile,
- skeleton,
- NULL);
- if (!skel_in)
- error (EXIT_FAILURE, errno, "cannot run m4");
- skel_lex ();
-
- /* If `debugging', keep this file alive. */
- if (!trace_flag)
- unlink (tempfile);
- }
+ /* Read and process m4's output. */
+ timevar_push (TV_M4);
+ end_of_output_subpipe (pid, filter_fd);
+ in = fdopen (filter_fd[1], "r");
+ if (! in)
+ error (EXIT_FAILURE, get_errno (),
+ "fdopen");
+ scan_skel (in);
+ xfclose (in);
+ reap_subpipe (pid, m4);
+ timevar_pop (TV_M4);
}
static void
prepare (void)
{
- MUSCLE_INSERT_INT ("last", high);
- MUSCLE_INSERT_INT ("flag", SHRT_MIN);
- MUSCLE_INSERT_INT ("pure", pure_parser);
- MUSCLE_INSERT_INT ("nsym", nsyms);
- MUSCLE_INSERT_INT ("debug", debug_flag);
- MUSCLE_INSERT_INT ("final", final_state);
- MUSCLE_INSERT_INT ("undef_token_number", undeftoken->number);
- MUSCLE_INSERT_INT ("user_token_number_max", max_user_token_number);
- MUSCLE_INSERT_INT ("error_verbose", error_verbose);
- MUSCLE_INSERT_STRING ("prefix", spec_name_prefix ? spec_name_prefix : "yy");
-
- /* FIXME: This is wrong: the muscles should decide whether they hold
- a copy or not, but the situation is too obscure currently. */
- MUSCLE_INSERT_STRING ("output_infix", output_infix ? output_infix : "");
- MUSCLE_INSERT_STRING ("output_prefix", short_base_name);
- MUSCLE_INSERT_STRING ("output_parser_name", parser_file_name);
- MUSCLE_INSERT_STRING ("output_header_name", spec_defines_file);
-
- MUSCLE_INSERT_INT ("nnts", nvars);
- MUSCLE_INSERT_INT ("nrules", nrules);
- MUSCLE_INSERT_INT ("nstates", nstates);
- MUSCLE_INSERT_INT ("ntokens", ntokens);
-
- MUSCLE_INSERT_INT ("locations_flag", locations_flag);
- MUSCLE_INSERT_INT ("defines_flag", defines_flag);
-
- /* Copy definitions in directive. */
- obstack_1grow (&pre_prologue_obstack, 0);
- obstack_1grow (&post_prologue_obstack, 0);
- muscle_insert ("pre_prologue", obstack_finish (&pre_prologue_obstack));
- muscle_insert ("post_prologue", obstack_finish (&post_prologue_obstack));
-
- /* Find the right skeleton file. */
- if (!skeleton)
- skeleton = "bison.simple";
-
- /* Parse the skeleton file and output the needed parsers. */
- muscle_insert ("skeleton", skeleton);
+ /* BISON_USE_PUSH_FOR_PULL is for the test suite and should not be documented
+ for the user. */
+ char const *use_push_for_pull_env = getenv ("BISON_USE_PUSH_FOR_PULL");
+ bool use_push_for_pull_flag = false;
+ if (use_push_for_pull_env != NULL
+ && use_push_for_pull_env[0] != '\0'
+ && 0 != strcmp (use_push_for_pull_env, "0"))
+ use_push_for_pull_flag = true;
+
+ /* Flags. */
+ MUSCLE_INSERT_BOOL ("debug_flag", debug_flag);
+ MUSCLE_INSERT_BOOL ("defines_flag", defines_flag);
+ MUSCLE_INSERT_BOOL ("error_verbose_flag", error_verbose);
+ MUSCLE_INSERT_BOOL ("glr_flag", glr_parser);
+ MUSCLE_INSERT_BOOL ("locations_flag", locations_flag);
+ MUSCLE_INSERT_BOOL ("nondeterministic_flag", nondeterministic_parser);
+ MUSCLE_INSERT_BOOL ("synclines_flag", !no_lines_flag);
+ MUSCLE_INSERT_BOOL ("tag_seen_flag", tag_seen);
+ MUSCLE_INSERT_BOOL ("use_push_for_pull_flag", use_push_for_pull_flag);
+ MUSCLE_INSERT_BOOL ("yacc_flag", yacc_flag);
+
+ /* File names. */
+ if (spec_name_prefix)
+ MUSCLE_INSERT_STRING ("prefix", spec_name_prefix);
+
+ MUSCLE_INSERT_STRING ("file_name_all_but_ext", all_but_ext);
+
+#define DEFINE(Name) MUSCLE_INSERT_STRING (#Name, Name ? Name : "")
+ DEFINE (dir_prefix);
+ DEFINE (parser_file_name);
+ DEFINE (spec_defines_file);
+ DEFINE (spec_file_prefix);
+ DEFINE (spec_graph_file);
+ DEFINE (spec_name_prefix);
+ DEFINE (spec_outfile);
+ DEFINE (spec_verbose_file);
+#undef DEFINE
+
+ /* Find the right skeleton file, and add muscles about the skeletons. */
+ if (skeleton)
+ MUSCLE_INSERT_C_STRING ("skeleton", skeleton);
+ else
+ skeleton = language->skeleton;
+
+ /* About the skeletons. */
+ {
+ /* b4_pkgdatadir is used inside m4_include in the skeletons, so digraphs
+ would never be expanded. Hopefully no one has M4-special characters in
+ his Bison installation path. */
+ MUSCLE_INSERT_STRING_RAW ("pkgdatadir", compute_pkgdatadir ());
+ }
}
{
obstack_init (&format_obstack);
- prepare_tokens ();
+ prepare_symbols ();
prepare_rules ();
prepare_states ();
- output_actions ();
+ prepare_actions ();
prepare ();
/* Process the selected skeleton file. */
output_skeleton ();
- obstack_free (&muscle_obstack, NULL);
obstack_free (&format_obstack, NULL);
- obstack_free (&action_obstack, NULL);
- obstack_free (&pre_prologue_obstack, NULL);
- obstack_free (&post_prologue_obstack, NULL);
+}
+
+char const *
+compute_pkgdatadir (void)
+{
+ char const *pkgdatadir = getenv ("BISON_PKGDATADIR");
+ return pkgdatadir ? pkgdatadir : PKGDATADIR;
}
projects / bison.git / blobdiff