X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/c3e2364794005071e7010d4dfd09d5a9cf046d47..9ca7f077a0a5ecdd00c5945c7a7bd3cf46e23c13:/src/output.c?ds=sidebyside diff --git a/src/output.c b/src/output.c index c6cb9ba8..1a1d7b3a 100644 --- a/src/output.c +++ b/src/output.c @@ -1,1426 +1,639 @@ -/* Output the generated parsing program for bison, - Copyright (C) 1984, 1986, 1989 Free Software Foundation, Inc. +/* Output the generated parsing program for Bison. -This file is part of Bison, the GNU Compiler Compiler. + Copyright (C) 1984, 1986, 1989, 1992, 2000, 2001, 2002, 2003, 2004, + 2005, 2006, 2007 Free Software Foundation, Inc. -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 file is part of Bison, the GNU Compiler Compiler. -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 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. -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, 675 Mass Ave, Cambridge, MA 02139, USA. */ + 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 this program. If not, see . */ -/* functions to output parsing data to various files. Entries are: - - output_headers () - -Output constant strings to the beginning of certain files. - - output_trailers() - -Output constant strings to the ends of certain files. - - output () - -Output the parsing tables and the parser code to ftable. - -The parser tables consist of these tables. -Starred ones needed only for the semantic parser. - -yytranslate = vector mapping yylex's token numbers into bison's token numbers. - -yytname = vector of string-names indexed by bison token number - -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 - as of before the text for this nonterminal was read. - The value from yytable is the state to go to. - -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. - -YYFINAL = the state number of the termination state. -YYFLAG = most negative short int. Used to flag ?? -YYNTBASE = ntokens. +#include +#include "system.h" -*/ +#include +#include +#include +#include +#include +#include -#include -#include "system.h" -#include "machine.h" -#include "new.h" +#include "complain.h" #include "files.h" +#include "getargs.h" #include "gram.h" -#include "state.h" - - -extern int debugflag; -extern int nolinesflag; - -extern char **tags; -extern int tokensetsize; -extern int final_state; -extern core **state_table; -extern shifts **shift_table; -extern errs **err_table; -extern reductions **reduction_table; -extern short *accessing_symbol; -extern unsigned *LA; -extern short *LAruleno; -extern short *lookaheads; -extern char *consistent; -extern short *goto_map; -extern short *from_state; -extern short *to_state; - -void output_token_translations(); -void output_gram(); -void output_stos(); -void output_rule_data(); -void output_defines(); -void output_actions(); -void token_actions(); -void save_row(); -void goto_actions(); -void save_column(); -void sort_actions(); -void pack_table(); -void output_base(); -void output_table(); -void output_check(); -void output_parser(); -void output_program(); -void free_itemset(); -void free_shifts(); -void free_reductions(); -void free_itemsets(); -int action_row(); -int default_goto(); -int matching_state(); -int pack_vector(); - -extern void berror(); -extern void fatals(); - -static int nvectors; -static int nentries; -static short **froms; -static short **tos; -static short *tally; -static short *width; -static short *actrow; -static short *state_count; -static short *order; -static short *base; -static short *pos; -static short *table; -static short *check; -static int lowzero; -static int high; - - - -#define GUARDSTR "\n#include \"%s\"\nextern int yyerror;\n\ -extern int yycost;\nextern char * yymsg;\nextern YYSTYPE yyval;\n\n\ -yyguard(n, yyvsp, yylsp)\nregister int n;\nregister YYSTYPE *yyvsp;\n\ -register YYLTYPE *yylsp;\n\ -{\n yyerror = 0;\nyycost = 0;\n yymsg = 0;\nswitch (n)\n {" - -#define ACTSTR "\n#include \"%s\"\nextern YYSTYPE yyval;\ -\nextern int yychar;\ -yyaction(n, yyvsp, yylsp)\nregister int n;\nregister YYSTYPE *yyvsp;\n\ -register YYLTYPE *yylsp;\n{\n switch (n)\n{" - -#define ACTSTR_SIMPLE "\n switch (yyn) {\n" - - -void -output_headers() -{ - if (semantic_parser) - fprintf(fguard, GUARDSTR, attrsfile); - fprintf(faction, (semantic_parser ? ACTSTR : ACTSTR_SIMPLE), attrsfile); -/* if (semantic_parser) JF moved this below - fprintf(ftable, "#include \"%s\"\n", attrsfile); - fprintf(ftable, "#include \n\n"); -*/ - - /* Rename certain symbols if -p was specified. */ - if (spec_name_prefix) - { - fprintf(ftable, "#define yyparse %sparse\n", spec_name_prefix); - fprintf(ftable, "#define yylex %slex\n", spec_name_prefix); - fprintf(ftable, "#define yyerror %serror\n", spec_name_prefix); - fprintf(ftable, "#define yylval %slval\n", spec_name_prefix); - fprintf(ftable, "#define yychar %schar\n", spec_name_prefix); - fprintf(ftable, "#define yydebug %sdebug\n", spec_name_prefix); - fprintf(ftable, "#define yynerrs %snerrs\n", spec_name_prefix); - } +#include "muscle_tab.h" +#include "output.h" +#include "reader.h" +#include "scan-code.h" /* max_left_semantic_context */ +#include "scan-skel.h" +#include "symtab.h" +#include "tables.h" + + +static struct obstack format_obstack; + + +/*-------------------------------------------------------------------. +| 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. | +`-------------------------------------------------------------------*/ + + +#define GENERATE_MUSCLE_INSERT_TABLE(Name, Type) \ + \ +static void \ +Name (char const *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); \ } +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_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) -void -output_trailers() -{ - if (semantic_parser) - { - fprintf(fguard, "\n }\n}\n"); - fprintf(faction, "\n }\n}\n"); - } - else - fprintf(faction, "\n}\n"); -} +/*--------------------------------------------------------------------. +| Print to OUT a representation of STRING escaped both for C and M4. | +`--------------------------------------------------------------------*/ -void -output() +static void +escaped_output (FILE *out, char const *string) { - int c; + char const *p; + fprintf (out, "[["); - /* output_token_defines(ftable); / * JF put out token defines FIRST */ - if (!semantic_parser) /* JF Put out other stuff */ - { - rewind(fattrs); - while ((c=getc(fattrs))!=EOF) - putc(c,ftable); - } + 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; + } - if (debugflag) - fprintf(ftable, "#ifndef YYDEBUG\n#define YYDEBUG %d\n#endif\n\n", - !!debugflag); - - if (semantic_parser) - fprintf(ftable, "#include \"%s\"\n", attrsfile); - fprintf(ftable, "#include \n\n"); - - /* Make "const" do nothing if not in ANSI C. */ - fprintf (ftable, "#ifndef __cplusplus\n#ifndef __STDC__\n#define const\n#endif\n#endif\n\n"); - - free_itemsets(); - output_defines(); - output_token_translations(); -/* if (semantic_parser) */ - /* This is now unconditional because debugging printouts can use it. */ - output_gram(); - FREE(ritem); - if (semantic_parser) - output_stos(); - output_rule_data(); - output_actions(); - output_parser(); - output_program(); + fprintf (out, "]]"); } -void -output_token_translations() -{ - register int i, j; -/* register short *sp; JF unused */ - - if (translations) - { - fprintf(ftable, - "\n#define YYTRANSLATE(x) ((unsigned)(x) <= %d ? yytranslate[x] : %d)\n", - max_user_token_number, nsyms); - - if (ntokens < 127) /* play it very safe; check maximum element value. */ - fprintf(ftable, "\nstatic const char yytranslate[] = { 0"); - else - fprintf(ftable, "\nstatic const short yytranslate[] = { 0"); - - j = 10; - for (i = 1; i <= max_user_token_number; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", token_translations[i]); - } - - fprintf(ftable, "\n};\n"); - } - else - { - fprintf(ftable, "\n#define YYTRANSLATE(x) (x)\n"); - } -} - - -void -output_gram() -{ - register int i; - register int j; - register short *sp; - - /* With the ordinary parser, - yyprhs and yyrhs are needed only for yydebug. */ - if (!semantic_parser) - fprintf(ftable, "\n#if YYDEBUG != 0"); - - fprintf(ftable, "\nstatic const short yyprhs[] = { 0"); - - j = 10; - for (i = 1; i <= nrules; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", rrhs[i]); - } - - fprintf(ftable, "\n};\n"); - - fprintf(ftable, "\nstatic const short yyrhs[] = {%6d", ritem[0]); +/*------------------------------------------------------------------. +| Prepare the muscles related to the symbols: translate, tname, and | +| toknum. | +`------------------------------------------------------------------*/ - j = 10; - for (sp = ritem + 1; *sp; sp++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - if (*sp > 0) - fprintf(ftable, "%6d", *sp); - else - fprintf(ftable, " 0"); - } - - fprintf(ftable, "\n};\n"); - - if(!semantic_parser) - fprintf(ftable, "\n#endif\n"); -} - - -void -output_stos() +static void +prepare_symbols (void) { - register int i; - register int j; - - fprintf(ftable, "\nstatic const short yystos[] = { 0"); - - j = 10; - for (i = 1; i < nstates; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", accessing_symbol[i]); - } - - fprintf(ftable, "\n};\n"); -} - - -void -output_rule_data() -{ - register int i; - register int j; - - fprintf(ftable, "\n#if YYDEBUG != 0\nstatic const short yyrline[] = { 0"); - - j = 10; - for (i = 1; i <= nrules; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", rline[i]); - } - - /* Output the table of symbol names. */ - - fprintf(ftable, - "\n};\n\nstatic const char * const yytname[] = { \"%s\"", - tags[0]); - - j = strlen (tags[0]) + 44; - for (i = 1; i <= nsyms; i++) - { - register char *p; - putc(',', ftable); - j++; - - if (j > 75) - { - putc('\n', ftable); - j = 0; - } - - putc ('\"', ftable); - j++; - - for (p = tags[i]; p && *p; p++) - { - if (*p == '"' || *p == '\\') - { - fprintf(ftable, "\\%c", *p); - j += 2; - } - else if (*p == '\n') - { - fprintf(ftable, "\\n"); - j += 2; - } - else if (*p == '\t') - { - fprintf(ftable, "\\t"); - j += 2; - } - else if (*p == '\b') - { - fprintf(ftable, "\\b"); - j += 2; - } - else if (*p < 040 || *p >= 0177) - { - fprintf(ftable, "\\%03o", *p); - j += 4; - } - else - { - putc(*p, ftable); - j++; - } - } - - putc ('\"', ftable); - j++; - } - - fprintf(ftable, "\n};\n#endif\n\nstatic const short yyr1[] = { 0"); - - j = 10; - for (i = 1; i <= nrules; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", rlhs[i]); - } - - FREE(rlhs + 1); - - fprintf(ftable, "\n};\n\nstatic const short yyr2[] = { 0"); - - j = 10; - for (i = 1; i < nrules; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", rrhs[i + 1] - rrhs[i] - 1); - } + 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; + /* We assume that the table will be output starting at column 2. */ + int j = 2; + for (i = 0; i < nsyms; i++) + { + char const *cp = quotearg_style (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; - putc(',', ftable); - if (j >= 10) - putc('\n', ftable); + if (j + width > 75) + { + obstack_sgrow (&format_obstack, "\n "); + j = 1; + } - fprintf(ftable, "%6d\n};\n", nitems - rrhs[nrules] - 1); - FREE(rrhs + 1); -} + if (i) + obstack_1grow (&format_obstack, ' '); + MUSCLE_OBSTACK_SGROW (&format_obstack, cp); + obstack_1grow (&format_obstack, ','); + j += width; + } + obstack_sgrow (&format_obstack, " ]b4_null["); + /* Finish table and store. */ + obstack_1grow (&format_obstack, 0); + muscle_insert ("tname", obstack_finish (&format_obstack)); + } -void -output_defines() -{ - fprintf(ftable, "\n\n#define\tYYFINAL\t\t%d\n", final_state); - fprintf(ftable, "#define\tYYFLAG\t\t%d\n", MINSHORT); - fprintf(ftable, "#define\tYYNTBASE\t%d\n", ntokens); + /* Output YYTOKNUM. */ + { + int i; + int *values = xnmalloc (ntokens, sizeof *values); + 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); + } } +/*-------------------------------------------------------------. +| Prepare the muscles related to the rules: rhs, prhs, r1, r2, | +| rline, dprec, merger. | +`-------------------------------------------------------------*/ -/* compute and output yydefact, yydefgoto, yypact, yypgoto, yytable and yycheck. */ - -void -output_actions() +static void +prepare_rules (void) { - nvectors = nstates + nvars; - - froms = NEW2(nvectors, short *); - tos = NEW2(nvectors, short *); - tally = NEW2(nvectors, short); - width = NEW2(nvectors, short); - - token_actions(); - free_shifts(); - free_reductions(); - FREE(lookaheads); - FREE(LA); - FREE(LAruleno); - FREE(accessing_symbol); - - goto_actions(); - FREE(goto_map + ntokens); - FREE(from_state); - FREE(to_state); - - sort_actions(); - pack_table(); - output_base(); - output_table(); - output_check(); + rule_number r; + unsigned int i = 0; + 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 *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; + } + aver (i == nritems); + + 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_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); } +/*--------------------------------------------. +| Prepare the muscles related to the states. | +`--------------------------------------------*/ - -/* 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. */ - -void -token_actions() +static void +prepare_states (void) { - register int i; - register int j; - register int k; - - actrow = NEW2(ntokens, short); - - k = action_row(0); - fprintf(ftable, "\nstatic const short yydefact[] = {%6d", k); - save_row(0); - - j = 10; - for (i = 1; i < nstates; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - k = action_row(i); - fprintf(ftable, "%6d", k); - save_row(i); - } - - fprintf(ftable, "\n};\n"); - FREE(actrow); + state_number i; + symbol_number *values = xnmalloc (nstates, sizeof *values); + for (i = 0; i < nstates; ++i) + values[i] = states[i]->accessing_symbol; + 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 MINSHORT, a very negative number, means this situation - is an error. The parser recognizes this value specially. +/*---------------------------------. +| Output the user actions to OUT. | +`---------------------------------*/ - 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. */ - -int -action_row(state) -int state; +static void +user_actions_output (FILE *out) { - register int i; - register int j; - register int k; - register int m; - register int n; - register int count; - register int default_rule; - register int nreds; - register int max; - register int rule; - register int shift_state; - register int symbol; - register unsigned mask; - register unsigned *wordp; - register reductions *redp; - register shifts *shiftp; - register errs *errp; - int nodefault = 0; /* set nonzero to inhibit having any default reduction */ - - for (i = 0; i < ntokens; i++) - actrow[i] = 0; - - default_rule = 0; - nreds = 0; - redp = reduction_table[state]; - - if (redp) - { - nreds = redp->nreds; - - if (nreds >= 1) - { - /* loop over all the rules available here which require lookahead */ - m = lookaheads[state]; - n = lookaheads[state + 1]; - - for (i = n - 1; i >= m; i--) - { - rule = - LAruleno[i]; - wordp = LA + i * tokensetsize; - mask = 1; - - /* 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 (mask & *wordp) - actrow[j] = rule; - - mask <<= 1; - if (mask == 0) - { - mask = 1; - wordp++; - } - } - } - } - } - - shiftp = shift_table[state]; - - /* 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. */ - - if (shiftp) - { - k = shiftp->nshifts; - - for (i = 0; i < k; i++) - { - shift_state = shiftp->shifts[i]; - if (! shift_state) continue; - - symbol = accessing_symbol[shift_state]; - - if (ISVAR(symbol)) - break; - - actrow[symbol] = shift_state; - - /* do not use any default reduction if there is a shift for error */ + rule_number r; - if (symbol == error_token_number) nodefault = 1; - } - } - - errp = err_table[state]; - - /* See which tokens are an explicit error in this state - (due to %nonassoc). For them, record MINSHORT as the action. */ - - if (errp) - { - k = errp->nerrs; - - for (i = 0; i < k; i++) - { - symbol = errp->errs[i]; - actrow[symbol] = MINSHORT; - } - } - - /* now find the most common reduction and make it the default action for this state. */ - - if (nreds >= 1 && ! nodefault) - { - if (consistent[state]) - default_rule = redp->rules[0]; - else - { - max = 0; - for (i = m; i < n; i++) - { - count = 0; - rule = - LAruleno[i]; - - 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) - { - 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 (j = 0; j < ntokens; j++) + fputs ("m4_define([b4_actions], \n[", out); + for (r = 0; r < nrules; ++r) + if (rules[r].action) { - if (actrow[j] == MINSHORT) - actrow[j] = 0; + 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); } - - return (default_rule); + fputs ("])\n\n", out); } +/*--------------------------------------. +| Output the merge functions to OUT. | +`--------------------------------------*/ -void -save_row(state) -int state; -{ - register int i; - register int count; - register short *sp; - register short *sp1; - register short *sp2; - - count = 0; - for (i = 0; i < ntokens; i++) - { - if (actrow[i] != 0) - count++; - } - - if (count == 0) - return; - - froms[state] = sp1 = sp = NEW2(count, short); - tos[state] = sp2 = NEW2(count, short); - - for (i = 0; i < ntokens; i++) - { - if (actrow[i] != 0) - { - *sp1++ = i; - *sp2++ = actrow[i]; - } - } - - tally[state] = count; - width[state] = sp1[-1] - sp[0] + 1; -} - - - -/* 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. */ - -void -goto_actions() +static void +merger_output (FILE *out) { - register int i; - register int j; - register int k; - - state_count = NEW2(nstates, short); - - k = default_goto(ntokens); - fprintf(ftable, "\nstatic const short yydefgoto[] = {%6d", k); - save_column(ntokens, k); + int n; + merger_list* p; - j = 10; - for (i = ntokens + 1; i < nsyms; i++) + fputs ("m4_define([b4_mergers], \n[[", out); + for (n = 1, p = merge_functions; p != NULL; n += 1, p = p->next) { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } + if (p->type[0] == '\0') + fprintf (out, " case %d: *yy0 = %s (*yy0, *yy1); break;\n", + n, p->name); else - { - j++; - } - - k = default_goto(i); - fprintf(ftable, "%6d", k); - save_column(i, k); + fprintf (out, " case %d: yy0->%s = %s (*yy0, *yy1); break;\n", + n, p->type, p->name); } - - fprintf(ftable, "\n};\n"); - FREE(state_count); + fputs ("]])\n\n", out); } +/*--------------------------------------. +| Output the tokens definition to OUT. | +`--------------------------------------*/ - -int -default_goto(symbol) -int symbol; -{ - register int i; - register int m; - register int n; - register int default_state; - register int max; - - m = goto_map[symbol]; - n = goto_map[symbol + 1]; - - 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]]++; - - max = 0; - default_state = -1; - - for (i = 0; i < nstates; i++) - { - if (state_count[i] > max) - { - max = state_count[i]; - default_state = i; - } - } - - return (default_state); -} - - -void -save_column(symbol, default_state) -int symbol; -int default_state; +static void +token_definitions_output (FILE *out) { - register int i; - register int m; - register int n; - register short *sp; - register short *sp1; - register short *sp2; - register int count; - register int symno; - - m = goto_map[symbol]; - n = goto_map[symbol + 1]; - - count = 0; - for (i = m; i < n; i++) - { - if (to_state[i] != default_state) - count++; - } - - if (count == 0) - return; - - symno = symbol - ntokens + nstates; - - froms[symno] = sp1 = sp = NEW2(count, short); - tos[symno] = sp2 = NEW2(count, short); - - for (i = m; i < n; 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; -} - - - -/* the next few functions decide how to pack - the actions and gotos information into yytable. */ - -void -sort_actions() -{ - register int i; - register int j; - register int k; - register int t; - register int w; - - order = NEW2(nvectors, short); - nentries = 0; - - for (i = 0; i < nvectors; i++) - { - if (tally[i] > 0) - { - t = tally[i]; - w = width[i]; - 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++; - } - } -} - - -void -pack_table() -{ - register int i; - register int place; - register int state; - - base = NEW2(nvectors, short); - pos = NEW2(nentries, short); - table = NEW2(MAXTABLE, short); - check = NEW2(MAXTABLE, short); - - 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; - } - - for (i = 0; i < nvectors; i++) - { - if (froms[i]) - FREE(froms[i]); - if (tos[i]) - FREE(tos[i]); - } - - FREE(froms); - FREE(tos); - FREE(pos); -} + int i; + char const *sep = ""; - - -int -matching_state(vector) -int vector; -{ - register int i; - register int j; - register int k; - register int t; - register int w; - register int match; - register int prev; - - i = order[vector]; - 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) { - j = order[prev]; - if (width[j] != w || tally[j] != t) - return (-1); - - match = 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); - } - - return (-1); -} + 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. */ + aver (number != USER_NUMBER_ALIAS); + /* Skip error token. */ + if (sym == errtoken) + continue; -int -pack_vector(vector) -int vector; -{ - register int i; - register int j; - register int k; - register int t; - register int loc; - register int ok; - register short *from; - register short *to; - - i = order[vector]; - t = tally[i]; + /* If this string has an alias, then it is necessarily the alias + which is to be output. */ + if (sym->alias) + sym = sym->alias; - if (t == 0) - berror("pack_vector"); + /* 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; - from = froms[i]; - 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; - for (j = lowzero - from[0]; j < MAXTABLE; j++) - { - ok = 1; - - for (k = 0; ok && k < t; k++) - { - loc = j + from[k]; - if (loc > MAXTABLE) - fatals("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); - } + fprintf (out, "%s[[[%s]], %d]", + sep, sym->tag, number); + sep = ",\n"; } - - berror("pack_vector"); - return 0; /* JF keep lint happy */ + fputs ("])\n\n", out); } +/*---------------------------------------------------. +| Output the symbol destructors or printers to OUT. | +`---------------------------------------------------*/ -/* the following functions output yytable, yycheck - and the vectors whose elements index the portion starts */ - -void -output_base() +static void +symbol_code_props_output (FILE *out, char const *what, + code_props const *(*get)(symbol const *)) { - register int i; - register int j; - - fprintf(ftable, "\nstatic const short yypact[] = {%6d", base[0]); - - j = 10; - for (i = 1; i < nstates; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", base[i]); - } - - fprintf(ftable, "\n};\n\nstatic const short yypgoto[] = {%6d", base[nstates]); - - j = 10; - for (i = nstates + 1; i < nvectors; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", base[i]); - } - - fprintf(ftable, "\n};\n"); - FREE(base); + int i; + char const *sep = ""; + + fputs ("m4_define([b4_symbol_", out); + fputs (what, out); + fputs ("], \n[", out); + for (i = 0; i < nsyms; ++i) + { + 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); + } + } + fputs ("])\n\n", out); } -void -output_table() +static void +prepare_actions (void) { - register int i; - register int j; - - fprintf(ftable, "\n\n#define\tYYLAST\t\t%d\n\n", high); - fprintf(ftable, "\nstatic const short yytable[] = {%6d", table[0]); - - j = 10; - for (i = 1; i <= high; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", table[i]); - } - - fprintf(ftable, "\n};\n"); - FREE(table); + /* 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, + 0, 1, conflict_list_cnt); } + +/*---------------------------. +| Call the skeleton parser. | +`---------------------------*/ -void -output_check() +static void +output_skeleton (void) { - register int i; - register int j; - - fprintf(ftable, "\nstatic const short yycheck[] = {%6d", check[0]); - - j = 10; - for (i = 1; i <= high; i++) - { - putc(',', ftable); - - if (j >= 10) - { - putc('\n', ftable); - j = 1; - } - else - { - j++; - } - - fprintf(ftable, "%6d", check[i]); - } - - fprintf(ftable, "\n};\n"); - FREE(check); + FILE *in; + FILE *out; + int filter_fd[2]; + char const *argv[7]; + 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 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); + 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); + + argv[0] = m4; + argv[1] = full_m4sugar; + argv[2] = "-"; + argv[3] = full_m4bison; + argv[4] = full_skeleton; + argv[5] = trace_flag & trace_m4 ? "-dV" : NULL; + argv[6] = NULL; + + init_subpipe (); + pid = create_subpipe (argv, filter_fd); + free (full_m4bison); + free (full_m4sugar); + 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); + + user_actions_output (out); + merger_output (out); + token_definitions_output (out); + symbol_code_props_output (out, "destructors", &symbol_destructor_get); + symbol_code_props_output (out, "printers", &symbol_printer_get); + + muscles_m4_output (out); + xfclose (out); + + /* 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); } - - -/* copy the parser code into the ftable file at the end. */ - -void -output_parser() +static void +prepare (void) { - register int c; -#ifdef DONTDEF - FILE *fpars; -#else -#define fpars fparser -#endif - - if (pure_parser) - fprintf(ftable, "#define YYPURE 1\n\n"); - -#ifdef DONTDEF /* JF no longer needed 'cuz open_extra_files changes the - currently open parser from bison.simple to bison.hairy */ - if (semantic_parser) - fpars = fparser; - else fpars = fparser1; -#endif - - /* Loop over lines in the standard parser file. */ - - while (1) - { - int write_line = 1; - - c = getc(fpars); - - /* See if the line starts with `#line. - If so, set write_line to 0. */ - if (nolinesflag) - if (c == '#') - { - c = getc(fpars); - if (c == 'l') - { - c = getc(fpars); - if (c == 'i') - { - c = getc(fpars); - if (c == 'n') - { - c = getc(fpars); - if (c == 'e') - write_line = 0; - else - fprintf(ftable, "#lin"); - } - else - fprintf(ftable, "#li"); - } - else - fprintf(ftable, "#l"); - } - else - fprintf(ftable, "#"); - } - - /* now write out the line... */ - for ( ; c != '\n' && c != EOF; c = getc(fpars)) - if (write_line) - if (c == '$') - { - /* `$' in the parser file indicates where to put the actions. - Copy them in at this point. */ - rewind(faction); - for(c=getc(faction);c!=EOF;c=getc(faction)) - putc(c,ftable); - } - else - putc(c, ftable); - if (c == EOF) - break; - putc(c, ftable); - } + /* 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 ()); + } } -void -output_program() -{ - register int c; - extern int lineno; - - if (!nolinesflag) - fprintf(ftable, "#line %d \"%s\"\n", lineno, infile); - - c = getc(finput); - while (c != EOF) - { - putc(c, ftable); - c = getc(finput); - } -} - +/*----------------------------------------------------------. +| Output the parsing tables and the parser code to ftable. | +`----------------------------------------------------------*/ void -free_itemsets() +output (void) { - register core *cp,*cptmp; - - FREE(state_table); - - for (cp = first_state; cp; cp = cptmp) { - cptmp=cp->next; - FREE(cp); - } -} + obstack_init (&format_obstack); + prepare_symbols (); + prepare_rules (); + prepare_states (); + prepare_actions (); -void -free_shifts() -{ - register shifts *sp,*sptmp;/* JF derefrenced freed ptr */ + prepare (); - FREE(shift_table); + /* Process the selected skeleton file. */ + output_skeleton (); - for (sp = first_shift; sp; sp = sptmp) { - sptmp=sp->next; - FREE(sp); - } + obstack_free (&format_obstack, NULL); } - -void -free_reductions() +char const * +compute_pkgdatadir (void) { - register reductions *rp,*rptmp;/* JF fixed freed ptr */ - - FREE(reduction_table); - - for (rp = first_reduction; rp; rp = rptmp) { - rptmp=rp->next; - FREE(rp); - } + char const *pkgdatadir = getenv ("BISON_PKGDATADIR"); + return pkgdatadir ? pkgdatadir : PKGDATADIR; }