X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/b85810aec0e5172c813b1cdd511bfce8a0cd4b4f..f61aad93992064785ef0f74d9bd5d19f016ba390:/src/output.c diff --git a/src/output.c b/src/output.c index d5eba988..42646cbf 100644 --- a/src/output.c +++ b/src/output.c @@ -1,5 +1,6 @@ -/* Output the generated parsing program for bison, - Copyright 1984, 1986, 1989, 1992, 2000, 2001 +/* Output the generated parsing program for Bison. + + Copyright (C) 1984, 1986, 1989, 1992, 2000, 2001, 2002, 2003 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -20,944 +21,485 @@ 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. - - 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 "system.h" - YYFINAL = the state number of the termination state. YYFLAG = most - negative short int. Used to flag ?? */ +#include +#include +#include +#include +#include -#include "system.h" -#include "quotearg.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 "symtab.h" -#include "conflicts.h" -#include "muscle_tab.h" -#include "skeleton.h" - -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; -static short *table = NULL; -static short *check = NULL; -static int lowzero; -static int high; - -struct obstack muscle_obstack; -static struct obstack format_obstack; - -int error_verbose = 0; - -/* Returns the number of lines of S. */ -size_t -get_lines_number (const char *s) -{ - size_t lines = 0; - - size_t i; - for (i = 0; s[i]; ++i) - if (s[i] == '\n') - ++lines; - - return lines; -} - - -/* FIXME. */ - -static inline void -output_table_data (struct obstack *oout, - short *table_data, - short first, - int begin, - int end) -{ - int i; - int j = 1; - - obstack_fgrow1 (oout, "%6d", first); - for (i = begin; i < end; ++i) - { - obstack_1grow (oout, ','); - if (j >= 10) - { - obstack_sgrow (oout, "\n "); - j = 1; - } - else - ++j; - obstack_fgrow1 (oout, "%6d", table_data[i]); - } - obstack_1grow (oout, 0); -} - - -static void -output_token_translations (void) -{ - output_table_data (&format_obstack, token_translations, - 0, 1, max_user_token_number + 1); - muscle_insert ("translate", obstack_finish (&format_obstack)); - XFREE (token_translations); -} - - -static void -output_gram (void) -{ - { - int i; - short *values = XCALLOC (short, nrules + 1); - for (i = 0; i < nrules + 1; ++i) - values[i] = rules[i].rhs; - output_table_data (&format_obstack, values, - 0, 1, nrules + 1); - XFREE (values); - } - - muscle_insert ("prhs", obstack_finish (&format_obstack)); - - { - short *yyrhs; - int i; - - yyrhs = XMALLOC (short, nritems); - - for (i = 1; i < nritems; ++i) - yyrhs[i] = ritem[i] >= 0 ? ritem[i] : -1; - - output_table_data (&format_obstack, yyrhs, - ritem[0], 1, nritems); - muscle_insert ("rhs", obstack_finish (&format_obstack)); - - XFREE (yyrhs); - } - -#if 0 - if (!semantic_parser) - obstack_sgrow (&table_obstack, "\n#endif\n"); -#endif -} - - -static void -output_stos (void) -{ - int i; - short *values = (short *) alloca (sizeof (short) * nstates); - for (i = 0; i < nstates; ++i) - values[i] = states[i]->accessing_symbol; - output_table_data (&format_obstack, values, - 0, 1, nstates); - muscle_insert ("stos", obstack_finish (&format_obstack)); -} +#include "tables.h" +/* From src/scan-skel.l. */ +void scan_skel (FILE *); -static void -output_rule_data (void) -{ - int i; - int j; - short *short_tab = NULL; - - { - short *values = XCALLOC (short, nrules + 1); - for (i = 0; i < nrules + 1; ++i) - values[i] = rules[i].line; - output_table_data (&format_obstack, values, - 0, 1, nrules + 1); - muscle_insert ("rline", obstack_finish (&format_obstack)); - XFREE (values); - } +static struct obstack format_obstack; - j = 0; - 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) - { - obstack_sgrow (&format_obstack, "\n "); - j = 2; - } - - obstack_sgrow (&format_obstack, cp); - obstack_sgrow (&format_obstack, ", "); - j += strsize; - } - /* add a NULL entry to list of tokens */ - obstack_sgrow (&format_obstack, "NULL"); +int error_verbose = 0; - /* Finish table and store. */ - obstack_1grow (&format_obstack, 0); - muscle_insert ("tname", obstack_finish (&format_obstack)); - /* Output YYTOKNUM. */ - { - short *values = XCALLOC (short, ntokens + 1); - for (i = 0; i < ntokens + 1; ++i) - values[i] = symbols[i]->user_token_number; - output_table_data (&format_obstack, values, - 0, 1, ntokens + 1); - muscle_insert ("toknum", obstack_finish (&format_obstack)); - XFREE (values); - } +/*-------------------------------------------------------------------. +| Create a function NAME which associates to the muscle NAME the | +| result of formatting the FIRST and then TABLE_DATA[BEGIN..END[ (of | +| TYPE), and to the muscle NAME_max, the max value of the | +| TABLE_DATA. | +`-------------------------------------------------------------------*/ - /* Output YYR1. */ - { - short *values = XCALLOC (short, nrules + 1); - for (i = 0; i < nrules + 1; ++i) - values[i] = rules[i].lhs; - output_table_data (&format_obstack, values, - 0, 1, nrules + 1); - muscle_insert ("r1", obstack_finish (&format_obstack)); - XFREE (values); - } - /* Output YYR2. */ - short_tab = XMALLOC (short, nrules + 1); - for (i = 1; i < nrules; i++) - short_tab[i] = rules[i + 1].rhs - rules[i].rhs - 1; - short_tab[nrules] = nritems - rules[nrules].rhs - 1; - output_table_data (&format_obstack, short_tab, - 0, 1, nrules + 1); - muscle_insert ("r2", obstack_finish (&format_obstack)); - XFREE (short_tab); +#define GENERATE_MUSCLE_INSERT_TABLE(Name, Type) \ + \ +static void \ +Name (const char *name, \ + Type *table_data, \ + Type first, \ + int begin, \ + int end) \ +{ \ + Type min = first; \ + Type max = first; \ + long int lmin; \ + long int lmax; \ + int i; \ + int j = 1; \ + \ + obstack_fgrow1 (&format_obstack, "%6d", first); \ + for (i = begin; i < end; ++i) \ + { \ + obstack_1grow (&format_obstack, ','); \ + if (j >= 10) \ + { \ + obstack_sgrow (&format_obstack, "\n "); \ + j = 1; \ + } \ + else \ + ++j; \ + obstack_fgrow1 (&format_obstack, "%6d", table_data[i]); \ + 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)); \ + \ + 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), lmax); \ } -/*------------------------------------------------------------------. -| 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 | -| MINSHORT, 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) -{ - 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 (BITISSET (LA (state->lookaheadsp + i), j)) - actrow[j] = -LAruleno[state->lookaheadsp + i]; - } - - /* 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++) - { - int symbol; - int shift_state = shiftp->shifts[i]; - if (!shift_state) - continue; - - symbol = states[shift_state]->accessing_symbol; +GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_unsigned_int_table, unsigned int) +GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_int_table, int) +GENERATE_MUSCLE_INSERT_TABLE(muscle_insert_short_table, short) +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) - if (ISVAR (symbol)) - break; - - actrow[symbol] = shift_state; - - /* Do not use any default reduction if there is a shift for - error */ - if (symbol == error_token_number) - nodefault = 1; - } - - /* See which tokens are an explicit error in this state (due to - %nonassoc). For them, record MINSHORT as the action. */ - for (i = 0; i < errp->nerrs; i++) - { - int symbol = errp->errs[i]; - actrow[symbol] = MINSHORT; - } - - /* 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 = -LAruleno[state->lookaheadsp + i]; - 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] == MINSHORT) - actrow[i] = 0; - - return default_rule; -} +/*----------------------------------------------------------------------. +| Print to OUT a representation of FILENAME escaped both for C and M4. | +`----------------------------------------------------------------------*/ static void -save_row (int state) +escaped_file_name_output (FILE *out, char const *filename) { - 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; + char const *p; + fprintf (out, "[["); - froms[state] = sp1 = sp = XCALLOC (short, count); - tos[state] = sp2 = XCALLOC (short, count); - - for (i = 0; i < ntokens; i++) - if (actrow[i] != 0) + for (p = quotearg_style (c_quoting_style, filename); *p; p++) + switch (*p) { - *sp1++ = i; - *sp2++ = actrow[i]; + case '$': fputs ("$][", out); break; + case '@': fputs ("@@", out); break; + case '[': fputs ("@{", out); break; + case ']': fputs ("@}", out); break; + default: fputc (*p, out); break; } - tally[state] = count; - width[state] = sp1[-1] - sp[0] + 1; + fprintf (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. | +| Prepare the muscles related to the symbols: translate, tname, and | +| toknum. | `------------------------------------------------------------------*/ static void -token_actions (void) +prepare_symbols (void) { - int i; - short *yydefact = XCALLOC (short, nstates); - - actrow = XCALLOC (short, ntokens); - for (i = 0; i < nstates; ++i) - { - yydefact[i] = action_row (states[i]); - save_row (i); - } - - output_table_data (&format_obstack, yydefact, - yydefact[0], 1, nstates); - muscle_insert ("defact", obstack_finish (&format_obstack)); - - XFREE (actrow); - XFREE (yydefact); -} + 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); -/*-----------------------------. -| Output the actions to OOUT. | -`-----------------------------*/ - -void -actions_output (FILE *out, size_t *line) -{ - int rule; - for (rule = 1; rule < nrules + 1; ++rule) - if (rules[rule].action) + /* tname -- token names. */ + { + int i; + int j = 0; + for (i = 0; i < nsyms; i++) { - 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 ? ";" : ""); - - /* We always output 4 '\n' per action. */ - *line += 4; - /* Plus one if !no_lines_flag. */ - if (!no_lines_flag) - ++*line; - /* Get the number of lines written by the user. */ - *line += get_lines_number (rules[rule].action); + /* Be sure not to use twice the same QUOTEARG slot: + SYMBOL_TAG_GET uses slot 0. */ + const char *cp = + quotearg_n_style (1, c_quoting_style, + symbols[i]->tag); + /* 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; + } + + MUSCLE_OBSTACK_SGROW (&format_obstack, cp); + obstack_sgrow (&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"); -/*----------------------------. -| Output the guards to OOUT. | -`----------------------------*/ + /* Finish table and store. */ + obstack_1grow (&format_obstack, 0); + muscle_insert ("tname", obstack_finish (&format_obstack)); + } -void -guards_output (FILE *out, size_t *line) -{ - 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].guard_line, - quotearg_style (c_quoting_style, - muscle_find ("filename"))); - fprintf (out, "{ %s; }\n break;\n\n", - rules[rule].guard); - - /* We always output 4 '\n' per action. */ - *line += 4; - /* Plus one if !no_lines_flag. */ - if (!no_lines_flag) - ++*line; - /* Get the number of lines written by the user. */ - *line += get_lines_number (rules[rule].guard); - } + /* Output YYTOKNUM. */ + { + int i; + int *values = MALLOC (values, ntokens); + for (i = 0; i < ntokens; ++i) + values[i] = symbols[i]->user_token_number; + muscle_insert_int_table ("toknum", values, + values[0], 1, ntokens); + free (values); + } } -/*---------------------------------------. -| Output the tokens definition to OOUT. | -`---------------------------------------*/ +/*-------------------------------------------------------------. +| Prepare the muscles related to the rules: rhs, prhs, r1, r2, | +| rline, dprec, merger. | +`-------------------------------------------------------------*/ -void -token_definitions_output (FILE *out, size_t *line) +static void +prepare_rules (void) { - int i; - for (i = 0; i < ntokens; ++i) + rule_number r; + unsigned int i = 0; + item_number *rhs = MALLOC (rhs, nritems); + unsigned int *prhs = MALLOC (prhs, nrules); + unsigned int *rline = MALLOC (rline, nrules); + symbol_number *r1 = MALLOC (r1, nrules); + unsigned int *r2 = MALLOC (r2, nrules); + short *dprec = MALLOC (dprec, nrules); + short *merger = MALLOC (merger, nrules); + + for (r = 0; r < nrules; ++r) { - bucket *symbol = symbols[i]; - int number = symbol->user_token_number; - - if (number == SALIAS) - continue; - /* Skip error token. */ - if (symbol->value == error_token_number) - continue; - if (symbol->tag[0] == '\'') - continue; /* skip literal character */ - if (symbol->tag[0] == '\"') - { - /* use literal string only if given a symbol with an alias */ - if (symbol->alias) - symbol = symbol->alias; - else - continue; - } - - /* Don't #define nonliteral tokens whose names contain periods - or '$' (as does the default value of the EOF token). */ - if (strchr (symbol->tag, '.') || strchr (symbol->tag, '$')) - continue; - - fprintf (out, "# define %s\t%d\n", - symbol->tag, number); - ++*line; - if (semantic_parser) - { - /* FIXME: This is probably wrong, and should be just as - above. --akim. */ - fprintf (out, "# define T%s\t%d\n", symbol->tag, symbol->value); - ++*line; - } + item_number *rhsp = NULL; + /* Index of rule R in RHS. */ + prhs[r] = i; + /* RHS of the rule R. */ + for (rhsp = rules[r].rhs; *rhsp >= 0; ++rhsp) + rhs[i++] = *rhsp; + /* LHS of the rule R. */ + r1[r] = rules[r].lhs->number; + /* Length of rule R's RHS. */ + r2[r] = i - prhs[r]; + /* Separator in RHS. */ + rhs[i++] = -1; + /* Line where rule was defined. */ + rline[r] = rules[r].location.start.line; + /* Dynamic precedence (GLR). */ + dprec[r] = rules[r].dprec; + /* Merger-function index (GLR). */ + merger[r] = rules[r].merger; } + if (i != nritems) + abort (); + + muscle_insert_item_number_table ("rhs", rhs, ritem[0], 1, nritems); + 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_short_table ("dprec", dprec, 0, 0, nrules); + muscle_insert_short_table ("merger", merger, 0, 0, nrules); + + MUSCLE_INSERT_INT ("rules_number", nrules); + + free (rhs); + free (prhs); + free (rline); + free (r1); + free (r2); + free (dprec); + free (merger); } +/*--------------------------------------------. +| Prepare the muscles related to the states. | +`--------------------------------------------*/ static void -save_column (int symbol, int default_state) +prepare_states (void) { - 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]; - } + state_number i; + symbol_number *values = MALLOC (values, nstates); + for (i = 0; i < nstates; ++i) + values[i] = states[i]->accessing_symbol; + muscle_insert_symbol_number_table ("stos", values, + 0, 1, nstates); + free (values); - tally[symno] = count; - width[symno] = sp1[-1] - sp[0] + 1; + MUSCLE_INSERT_INT ("last", high); + MUSCLE_INSERT_INT ("final_state_number", final_state->number); + MUSCLE_INSERT_INT ("states_number", nstates); } -static int -default_goto (int symbol) -{ - int i; - int m = goto_map[symbol]; - int 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; +/*---------------------------------. +| Output the user actions to OUT. | +`---------------------------------*/ - for (i = m; i < n; i++) - state_count[to_state[i]]++; +static void +user_actions_output (FILE *out) +{ + rule_number r; - for (i = 0; i < nstates; i++) - if (state_count[i] > max) + fputs ("m4_define([b4_actions], \n[[", out); + for (r = 0; r < nrules; ++r) + if (rules[r].action) { - max = state_count[i]; - default_state = i; + fprintf (out, " case %d:\n", r + 1); + + fprintf (out, "]b4_syncline([[%d]], ", + rules[r].action_location.start.line); + escaped_file_name_output (out, rules[r].action_location.start.file); + fprintf (out, ")[\n"); + fprintf (out, " %s\n break;\n\n", + rules[r].action); } - - return default_state; + fputs ("]])\n\n", out); } - -/*-------------------------------------------------------------------. -| 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. | -`-------------------------------------------------------------------*/ +/*--------------------------------------. +| Output the merge functions to OUT. | +`--------------------------------------*/ static void -goto_actions (void) +merger_output (FILE *out) { - int i; - short *yydefgoto = XMALLOC (short, nsyms - ntokens); + int n; + merger_list* p; - state_count = XCALLOC (short, nstates); - for (i = ntokens; i < nsyms; ++i) + fputs ("m4_define([b4_mergers], \n[[", out); + for (n = 1, p = merge_functions; p != NULL; n += 1, p = p->next) { - int default_state = default_goto (i); - save_column (i, default_state); - yydefgoto[i - ntokens] = default_state; + if (p->type[0] == '\0') + fprintf (out, " case %d: yyval = %s (*yy0, *yy1); break;\n", + n, p->name); + else + fprintf (out, " case %d: yyval.%s = %s (*yy0, *yy1); break;\n", + n, p->type, p->name); } - - output_table_data (&format_obstack, yydefgoto, - yydefgoto[0], 1, nsyms - ntokens); - muscle_insert ("defgoto", obstack_finish (&format_obstack)); - - 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 tokens definition to OUT. | +`--------------------------------------*/ static void -sort_actions (void) +token_definitions_output (FILE *out) { int i; + int first = 1; - 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 >= nstates) - return -1; - - t = tally[i]; - w = width[i]; - - for (prev = vector - 1; prev >= 0; prev--) + fputs ("m4_define([b4_tokens], \n[", out); + for (i = 0; i < ntokens; ++i) { - int j = order[prev]; - int k; - int match = 1; - - if (width[j] != w || tally[j] != t) - return -1; + symbol *sym = symbols[i]; + int number = sym->user_token_number; - for (k = 0; match && k < t; k++) - if (tos[j][k] != tos[i][k] || froms[j][k] != froms[i][k]) - match = 0; + /* At this stage, if there are literal aliases, they are part of + SYMBOLS, so we should not find symbols which are the aliases + here. */ + if (number == USER_NUMBER_ALIAS) + abort (); - if (match) - return j; - } + /* Skip error token. */ + if (sym == errtoken) + continue; - return -1; -} + /* If this string has an alias, then it is necessarily the alias + which is to be output. */ + 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 (sym->tag[0] == '\'' || sym->tag[0] == '\"') + continue; -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]; + /* Don't #define nonliteral tokens whose names contain periods + or '$' (as does the default value of the EOF token). */ + if (strchr (sym->tag, '.') || strchr (sym->tag, '$')) + continue; - assert (t); + fprintf (out, "%s[[[%s]], [%d]]", + first ? "" : ",\n", sym->tag, number); - for (j = lowzero - from[0]; j < MAXTABLE; j++) - { - int k; - int ok = 1; - - for (k = 0; ok && k < t; k++) - { - loc = j + from[k]; - if (loc > MAXTABLE) - fatal (_("maximum table size (%d) exceeded"), MAXTABLE); - - 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; - } + first = 0; } -#define pack_vector_succeeded 0 - assert (pack_vector_succeeded); - return 0; + fputs ("])\n\n", out); } +/*---------------------------------------. +| Output the symbol destructors to OUT. | +`---------------------------------------*/ + static void -pack_table (void) +symbol_destructors_output (FILE *out) { int i; - int place; - int state; - - base = XCALLOC (short, nvectors); - pos = XCALLOC (short, nentries); - table = XCALLOC (short, MAXTABLE); - check = XCALLOC (short, MAXTABLE); - - lowzero = 0; - high = 0; - - for (i = 0; i < nvectors; i++) - base[i] = MINSHORT; - - for (i = 0; i < MAXTABLE; 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; - } + int first = 1; - for (i = 0; i < nvectors; i++) - { - XFREE (froms[i]); - XFREE (tos[i]); - } - - XFREE (froms); - XFREE (tos); - XFREE (pos); + fputs ("m4_define([b4_symbol_destructors], \n[", out); + for (i = 0; i < nsyms; ++i) + if (symbols[i]->destructor) + { + symbol *sym = symbols[i]; + + /* Filename, lineno, + Symbol-name, Symbol-number, + destructor, typename. */ + fprintf (out, "%s[", + first ? "" : ",\n"); + escaped_file_name_output (out, sym->destructor_location.start.file); + fprintf (out, ", [[%d]], [[%s]], [[%d]], [[%s]], [[%s]]]", + sym->destructor_location.start.line, + sym->tag, + sym->number, + sym->destructor, + sym->type_name); + + first = 0; + } + fputs ("])\n\n", out); } -/* the following functions output yytable, yycheck - and the vectors whose elements index the portion starts */ - -static void -output_base (void) -{ - /* Output pact. */ - output_table_data (&format_obstack, base, - base[0], 1, nstates); - muscle_insert ("pact", obstack_finish (&format_obstack)); - - /* Output pgoto. */ - output_table_data (&format_obstack, base, - base[nstates], nstates + 1, nvectors); - muscle_insert ("pgoto", obstack_finish (&format_obstack)); - - XFREE (base); -} +/*------------------------------------. +| Output the symbol printers to OUT. | +`------------------------------------*/ static void -output_table (void) +symbol_printers_output (FILE *out) { - output_table_data (&format_obstack, table, - table[0], 1, high + 1); - muscle_insert ("table", obstack_finish (&format_obstack)); - XFREE (table); -} - + int i; + int first = 1; -static void -output_check (void) -{ - output_table_data (&format_obstack, check, - check[0], 1, high + 1); - muscle_insert ("check", obstack_finish (&format_obstack)); - XFREE (check); + fputs ("m4_define([b4_symbol_printers], \n[", out); + for (i = 0; i < nsyms; ++i) + if (symbols[i]->printer) + { + symbol *sym = symbols[i]; + + /* Filename, lineno, + Symbol-name, Symbol-number, + printer, typename. */ + fprintf (out, "%s[", + first ? "" : ",\n"); + escaped_file_name_output (out, sym->printer_location.start.file); + fprintf (out, ", [[%d]], [[%s]], [[%d]], [[%s]], [[%s]]]", + sym->printer_location.start.line, + sym->tag, + sym->number, + sym->printer, + sym->type_name); + + first = 0; + } + fputs ("])\n\n", out); } -/* compute and output yydefact, yydefgoto, yypact, yypgoto, yytable - and yycheck. */ static void -output_actions (void) +prepare_actions (void) { - int i; - nvectors = nstates + nvars; - - froms = XCALLOC (short *, nvectors); - tos = XCALLOC (short *, nvectors); - tally = XCALLOC (short, nvectors); - width = XCALLOC (short, nvectors); - - token_actions (); - XFREE (LA); - XFREE (LAruleno); - - 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); + /* Figure out the actions for the specified state, indexed by + lookahead token type. */ + + muscle_insert_rule_number_table ("defact", yydefact, + yydefact[0], 1, nstates); + + /* 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); + + + /* Output PACT. */ + muscle_insert_base_table ("pact", base, + base[0], 1, nstates); + MUSCLE_INSERT_INT ("pact_ninf", base_ninf); + + /* Output PGOTO. */ + muscle_insert_base_table ("pgoto", base, + base[nstates], nstates + 1, nvectors); + + 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, + conflict_list[0], 1, conflict_list_cnt); } @@ -968,105 +510,119 @@ output_actions (void) static void output_skeleton (void) { - /* Find the right skeleton file. */ - if (!skeleton) - { - if (semantic_parser) - skeleton = skeleton_find ("BISON_HAIRY", BISON_HAIRY); - else - skeleton = skeleton_find ("BISON_SIMPLE", BISON_SIMPLE); - } + FILE *in; + FILE *out; + int filter_fd[2]; + char const *argv[6]; + pid_t pid; + + /* Compute the names of the package data dir and skeleton file. + Test whether m4sugar.m4 is readable, to check for proper + installation. A faulty installation can cause deadlock, so a + cheap sanity check is worthwhile. */ + char const m4sugar[] = "m4sugar/m4sugar.m4"; + char *full_m4sugar; + char *full_cm4; + char *full_path; + char const *p; + char const *m4 = (p = getenv ("M4")) ? p : M4; + char const *pkgdatadir = (p = getenv ("BISON_PKGDATADIR")) ? p : PKGDATADIR; + size_t skeleton_size = strlen (skeleton) + 1; + size_t pkgdatadirlen = strlen (pkgdatadir); + while (pkgdatadirlen && pkgdatadir[pkgdatadirlen - 1] == '/') + pkgdatadirlen--; + full_path = xmalloc (pkgdatadirlen + 1 + + (skeleton_size < sizeof m4sugar + ? sizeof m4sugar : skeleton_size)); + strcpy (full_path, pkgdatadir); + full_path[pkgdatadirlen] = '/'; + strcpy (full_path + pkgdatadirlen + 1, m4sugar); + full_m4sugar = xstrdup (full_path); + strcpy (full_path + pkgdatadirlen + 1, "c.m4"); + full_cm4 = xstrdup (full_path); + strcpy (full_path + pkgdatadirlen + 1, skeleton); + 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_cm4, full_path); + + argv[0] = m4; + argv[1] = full_m4sugar; + argv[2] = "-"; + argv[3] = full_cm4; + argv[4] = full_path; + argv[5] = NULL; + + init_subpipe (); + pid = create_subpipe (argv, filter_fd); + free (full_m4sugar); + free (full_cm4); + free (full_path); + + 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); + + user_actions_output (out); + merger_output (out); + token_definitions_output (out); + symbol_destructors_output (out); + symbol_printers_output (out); + + muscles_m4_output (out); + + fputs ("m4_wrap([m4_divert_pop(0)])\n", out); + fputs ("m4_divert_push(0)dnl\n", out); + xfclose (out); - /* Parse the skeleton file and output the needed parsers. */ - muscle_insert ("skeleton", skeleton); - process_skeleton (skeleton); + /* Read and process m4's output. */ + timevar_push (TV_M4); + 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", MINSHORT); - MUSCLE_INSERT_INT ("pure", pure_parser); - MUSCLE_INSERT_INT ("nsym", nsyms); + /* Flags. */ MUSCLE_INSERT_INT ("debug", debug_flag); - MUSCLE_INSERT_INT ("final", final_state); - MUSCLE_INSERT_INT ("maxtok", 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_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); + MUSCLE_INSERT_INT ("error_verbose", error_verbose); + MUSCLE_INSERT_INT ("locations_flag", locations_flag); + MUSCLE_INSERT_INT ("pure", pure_parser); + MUSCLE_INSERT_INT ("synclines_flag", !no_lines_flag); -/*-------------------------. -| Output the header file. | -`-------------------------*/ + /* File names. */ + MUSCLE_INSERT_STRING ("prefix", spec_name_prefix ? spec_name_prefix : "yy"); -static void -header_output (void) -{ - size_t dummy_line; - FILE *out = xfopen (spec_defines_file, "w"); - char *macro_name = compute_header_macro (); - - fprintf (out, "#ifndef %s\n", macro_name); - fprintf (out, "# define %s\n\n", macro_name); - - token_definitions_output (out, &dummy_line); - fprintf (out, "\ -#ifndef YYSTYPE\n\ -typedef %s -yystype;\n\ -# define YYSTYPE yystype\n\ -#endif\n", - muscle_find ("stype")); - - if (!pure_parser) - fprintf (out, "\nextern YYSTYPE %slval;\n", - spec_name_prefix ? spec_name_prefix : "yy"); - - if (locations_flag) - { - fputs ("\n\n", out); - fprintf (out, "\ -#ifndef YYLTYPE\n\ -typedef struct yyltype\n\ -{\n\ - int first_line;\n\ - int first_column;\n\ - int last_line;\n\ - int last_column;\n\ -} yyltype;\n\ -# define YYLTYPE yyltype\n\ -#endif\n"); - if (!pure_parser) - fprintf (out, "\nextern YYLTYPE %slloc;\n", - spec_name_prefix ? spec_name_prefix : "yy"); - } + /* User Code. */ + 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)); - if (semantic_parser) + /* Find the right skeleton file. */ + if (!skeleton) { - int i; - - for (i = ntokens; i < nsyms; i++) - /* don't make these for dummy nonterminals made by gensym. */ - if (*symbols[i]->tag != '@') - fprintf (out, "# define NT%s\t%d\n", symbols[i]->tag, i); + if (glr_parser) + skeleton = "glr.c"; + else + skeleton = "yacc.c"; } - fprintf (out, "\n#endif /* not %s */\n", macro_name); - free (macro_name); - xfclose (out); + /* Parse the skeleton file and output the needed parsers. */ + MUSCLE_INSERT_C_STRING ("skeleton", skeleton); } @@ -1079,30 +635,17 @@ output (void) { obstack_init (&format_obstack); - output_token_translations (); - output_gram (); - - XFREE (ritem); - if (semantic_parser) - output_stos (); - output_rule_data (); - output_actions (); + prepare_symbols (); + prepare_rules (); + prepare_states (); + prepare_actions (); prepare (); - /* Copy definitions in directive. */ - obstack_1grow (&attrs_obstack, 0); - muscle_insert ("prologue", obstack_finish (&attrs_obstack)); /* Process the selected skeleton file. */ output_skeleton (); - /* Output the header if needed. */ - if (defines_flag) - header_output (); - - free (rules + 1); - obstack_free (&muscle_obstack, NULL); obstack_free (&format_obstack, NULL); - obstack_free (&action_obstack, NULL); - obstack_free (&attrs_obstack, NULL); + obstack_free (&pre_prologue_obstack, NULL); + obstack_free (&post_prologue_obstack, NULL); }