X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/f17bcd1f66dab24135ca3352491ac8fa4385fe37..e89a22bfab22e4d2ee73be49dcb66b51f8d0e892:/src/reader.c diff --git a/src/reader.c b/src/reader.c index e073f028..529cebe9 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,5 +1,5 @@ /* Input parser for bison - Copyright 1984, 1986, 1989, 1992, 1998, 2000 + Copyright 1984, 1986, 1989, 1992, 1998, 2000, 2001 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -26,7 +26,6 @@ #include "quote.h" #include "getargs.h" #include "files.h" -#include "xalloc.h" #include "symtab.h" #include "lex.h" #include "gram.h" @@ -35,13 +34,11 @@ #include "reader.h" #include "conflicts.h" -/* Number of slots allocated (but not necessarily used yet) in `rline' */ -static int rline_allocated; - typedef struct symbol_list { struct symbol_list *next; bucket *sym; + int line; bucket *ruleprec; } symbol_list; @@ -63,11 +60,21 @@ static int typed; /* Incremented for each %left, %right or %nonassoc seen */ static int lastprec; -/* Incremented for each generated symbol */ -static int gensym_count; - static bucket *errtoken; static bucket *undeftoken; + + +static symbol_list * +symbol_list_new (bucket *sym) +{ + symbol_list *res = XMALLOC (symbol_list, 1); + res->next = NULL; + res->sym = sym; + res->line = lineno; + res->ruleprec = NULL; + return res; +} + /*===================\ @@ -125,7 +132,7 @@ read_signed_integer (FILE *stream) `--------------------------------------------------------------*/ static char * -get_type_name (int n, symbol_list * rule) +get_type_name (int n, symbol_list *rule) { int i; symbol_list *rp; @@ -297,13 +304,15 @@ copy_comment (FILE *fin, struct obstack *oout) /*-----------------------------------------------------------------. | FIN is pointing to a location (i.e., a `@'). Output to OOUT a | | reference to this location. STACK_OFFSET is the number of values | -| in the current rule so far, which says where to find `$0' with | +| in the current rule so far, which says where to find `@0' with | | respect to the top of the stack. | `-----------------------------------------------------------------*/ static inline void -copy_at (FILE *fin, struct obstack *oout, int stack_offset) +copy_at (FILE *fin, struct obstack *oout, + struct symbol_list *rule, int stack_offset) { + symbol_list *rp; int c; c = getc (fin); @@ -314,12 +323,25 @@ copy_at (FILE *fin, struct obstack *oout, int stack_offset) } else if (isdigit (c) || c == '-') { - int n; - char buf[4096]; + int n, i; ungetc (c, fin); n = read_signed_integer (fin); + rp = rule; + i = 0; + + while (i < n) + { + rp = rp->next; + if (rp == NULL) + { + complain (_("invalid @ value")); + return; + } + i++; + } + obstack_fgrow1 (oout, "yylsp[%d]", n - stack_offset); locations_flag = 1; } @@ -347,7 +369,7 @@ copy_dollar (FILE *fin, struct obstack *oout, symbol_list *rule, int stack_offset) { int c = getc (fin); - char *type_name = NULL; + const char *type_name = NULL; /* Get the type name if explicit. */ if (c == '<') @@ -466,8 +488,8 @@ copy_definition (void) static void parse_token_decl (symbol_class what_is, symbol_class what_is_not) { - token_t token = 0; - char *typename = 0; + token_t token = tok_undef; + char *typename = NULL; /* The symbol being defined. */ struct bucket *symbol = NULL; @@ -516,7 +538,6 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) /* symbol and symval combined are only one symbol */ nsyms--; } - translations = 1; symbol = NULL; } else if (token == tok_identifier) @@ -541,12 +562,12 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) else if (symbol && token == tok_number) { symbol->user_token_number = numval; - translations = 1; } else { complain (_("`%s' is invalid in %s"), - token_buffer, (what_is == token_sym) ? "%token" : "%nterm"); + token_buffer, + (what_is == token_sym) ? "%token" : "%nterm"); skip_to_char ('%'); } } @@ -683,7 +704,6 @@ parse_assoc_decl (associativity assoc) if (prev == tok_identifier) { symval->user_token_number = numval; - translations = 1; } else { @@ -703,7 +723,6 @@ token_buffer); } prev = t; - } } @@ -720,6 +739,13 @@ parse_union_decl (void) { int c; int count = 0; + const char *prologue = "\ +#ifndef YYSTYPE\n\ +typedef union"; + const char *epilogue = "\ + yystype;\n\ +# define YYSTYPE yystype\n\ +#endif\n"; if (typed) complain (_("multiple %s declarations"), "%union"); @@ -732,17 +758,22 @@ parse_union_decl (void) else obstack_1grow (&attrs_obstack, '\n'); - obstack_sgrow (&attrs_obstack, "typedef union"); + obstack_sgrow (&attrs_obstack, prologue); if (defines_flag) - obstack_sgrow (&defines_obstack, "typedef union"); + obstack_sgrow (&defines_obstack, prologue); c = getc (finput); while (c != EOF) { - obstack_1grow (&attrs_obstack, c); - if (defines_flag) - obstack_1grow (&defines_obstack, c); + + /* If C contains '/', it is output by copy_comment (). */ + if (c != '/') + { + obstack_1grow (&attrs_obstack, c); + if (defines_flag) + obstack_1grow (&defines_obstack, c); + } switch (c) { @@ -764,9 +795,9 @@ parse_union_decl (void) count--; if (count <= 0) { - obstack_sgrow (&attrs_obstack, " YYSTYPE;\n"); + obstack_sgrow (&attrs_obstack, epilogue); if (defines_flag) - obstack_sgrow (&defines_obstack, " YYSTYPE;\n"); + obstack_sgrow (&defines_obstack, epilogue); /* JF don't choke on trailing semi */ c = skip_white_space (); if (c != ';') @@ -824,9 +855,8 @@ parse_thong_decl (void) token_t token; struct bucket *symbol; char *typename = 0; - int usrtoknum; + int usrtoknum = SUNDEF; - translations = 1; token = lex (); /* fetch typename or first token */ if (token == tok_typename) { @@ -856,8 +886,6 @@ parse_thong_decl (void) usrtoknum = numval; token = lex (); /* okay, did number, now get literal */ } - else - usrtoknum = 0; /* process literal string token */ @@ -879,6 +907,61 @@ parse_thong_decl (void) } +/*------------------------------------------------------------------. +| Parse a double quoted parameter. It was used for | +| %{source,header}_extension. For the moment, It is not used since | +| extension features have been removed. | +`------------------------------------------------------------------*/ + +#if 0 + +static const char * +parse_dquoted_param (const char *from) +{ + char buff[32]; + int c; + int i; + + c = skip_white_space (); + + if (c != '"') + { + ungetc (c, finput); + complain (_("invalid %s declaration"), from); + return NULL; + } + + c = getc (finput); + for (i = 0; (c >= '!') && (c <= '~'); i++) + { + if (c == '"') + break; + + if (c == '\\') + { + c = getc (finput); + if ((c < '!') && (c > '~')) + break; + } + + buff[i] = c; + c = getc (finput); + } + buff[i] = '\0'; + + if (c != '"') + { + ungetc (c, finput); + complain (_("invalid %s declaration"), from); + return NULL; + } + + return xstrdup (buff); +} + +#endif + + /*----------------------------------------------------------------. | Read from finput until `%%' is seen. Discard the `%%'. Handle | | any `%' declarations, and copy the contents of any `%{ ... %}' | @@ -888,16 +971,13 @@ parse_thong_decl (void) static void read_declarations (void) { - int c; - int tok; - for (;;) { - c = skip_white_space (); + int c = skip_white_space (); if (c == '%') { - tok = parse_percent_token (); + token_t tok = parse_percent_token (); switch (tok) { @@ -951,6 +1031,13 @@ read_declarations (void) case tok_noop: break; + case tok_stropt: + case tok_intopt: + case tok_obsolete: + abort (); + break; + + case tok_illegal: default: complain (_("unrecognized: %s"), token_buffer); skip_to_char ('%'); @@ -1032,7 +1119,7 @@ copy_action (symbol_list *rule, int stack_offset) case '@': copy_at (finput, &action_obstack, - stack_offset); + rule, stack_offset); break; case EOF: @@ -1054,7 +1141,13 @@ copy_action (symbol_list *rule, int stack_offset) } } - obstack_sgrow (&action_obstack, ";\n break;}"); + /* 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. */ + if (yacc_flag) + obstack_sgrow (&action_obstack, "}\n break;"); + else + obstack_sgrow (&action_obstack, ";\n break;}"); } /*-------------------------------------------------------------------. @@ -1126,7 +1219,7 @@ copy_guard (symbol_list *rule, int stack_offset) break; case '@': - copy_at (finput, &guard_obstack, stack_offset); + copy_at (finput, &guard_obstack, rule, stack_offset); break; case EOF: @@ -1156,20 +1249,6 @@ copy_guard (symbol_list *rule, int stack_offset) } -static void -record_rule_line (void) -{ - /* Record each rule's source line number in rline table. */ - - if (nrules >= rline_allocated) - { - rline_allocated = nrules * 2; - rline = XREALLOC (rline, short, rline_allocated); - } - rline[nrules] = lineno; -} - - /*-------------------------------------------------------------------. | Generate a dummy symbol, a nonterminal, whose name cannot conflict | | with the user's names. | @@ -1178,9 +1257,14 @@ record_rule_line (void) static bucket * gensym (void) { + /* Incremented for each generated symbol */ + static int gensym_count = 0; + static char buf[256]; + bucket *sym; - sprintf (token_buffer, "@%d", ++gensym_count); + sprintf (buf, "@%d", ++gensym_count); + token_buffer = buf; sym = getsym (token_buffer); sym->class = nterm_sym; sym->value = nvars++; @@ -1307,10 +1391,7 @@ readgram (void) nrules++; nitems++; - record_rule_line (); - - p = XCALLOC (symbol_list, 1); - p->sym = lhs; + p = symbol_list_new (lhs); crule1 = p1; if (p1) @@ -1372,36 +1453,35 @@ readgram (void) non-terminal. */ if (action_flag) { - bucket *sdummy; - /* Since the action was written out with this rule's number, we must give the new rule this number by inserting the new rule before it. */ /* Make a dummy nonterminal, a gensym. */ - sdummy = gensym (); + bucket *sdummy = gensym (); - /* Make a new rule, whose body is empty, - before the current one, so that the action - just read can belong to it. */ + /* Make a new rule, whose body is empty, before the + current one, so that the action just read can + belong to it. */ nrules++; nitems++; - record_rule_line (); - p = XCALLOC (symbol_list, 1); + p = symbol_list_new (sdummy); + /* Attach its lineno to that of the host rule. */ + p->line = crule->line; if (crule1) crule1->next = p; else grammar = p; - p->sym = sdummy; - crule1 = XCALLOC (symbol_list, 1); - p->next = crule1; + /* End of the rule. */ + crule1 = symbol_list_new (NULL); crule1->next = crule; + p->next = crule1; + /* Insert the dummy generated by that rule into this rule. */ nitems++; - p = XCALLOC (symbol_list, 1); - p->sym = sdummy; + p = symbol_list_new (sdummy); p1->next = p; p1 = p; @@ -1411,8 +1491,7 @@ readgram (void) if (t == tok_identifier) { nitems++; - p = XCALLOC (symbol_list, 1); - p->sym = symval; + p = symbol_list_new (symval); p1->next = p; p1 = p; } @@ -1426,7 +1505,7 @@ readgram (void) } /* end of read rhs of rule */ /* Put an empty link in the list to mark the end of this rule */ - p = XCALLOC (symbol_list, 1); + p = symbol_list_new (NULL); p1->next = p; p1 = p; @@ -1513,7 +1592,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1537,9 +1616,9 @@ readgram (void) "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); if (defines_flag) obstack_sgrow (&defines_obstack, "\ -#ifndef YYSTYPE\n\ -# define YYSTYPE int\n\ -#endif\n"); +# ifndef YYSTYPE\n\ +# define YYSTYPE int\n\ +# endif\n"); } /* Report any undefined symbols and consider them nonterminals. */ @@ -1596,18 +1675,101 @@ output_token_defines (struct obstack *oout) if (c != '\0') continue; - obstack_fgrow2 (oout, "#define\t%s\t%d\n", - symbol, - ((translations && !raw_flag) - ? bp->user_token_number : bp->value)); + obstack_fgrow2 (oout, "# define\t%s\t%d\n", + symbol, bp->user_token_number); if (semantic_parser) - obstack_fgrow2 (oout, "#define\tT%s\t%d\n", symbol, bp->value); + /* FIXME: This is certainly dead wrong, and should be just as + above. --akim. */ + obstack_fgrow2 (oout, "# define\tT%s\t%d\n", symbol, bp->value); } obstack_1grow (oout, '\n'); } +/*--------------------. +| Output the header. | +`--------------------*/ + +static void +symbols_output (void) +{ + if (defines_flag) + { + output_token_defines (&defines_obstack); + + if (!pure_parser) + { + if (spec_name_prefix) + obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", + spec_name_prefix); + else + obstack_sgrow (&defines_obstack, + "\nextern YYSTYPE yylval;\n"); + } + + if (semantic_parser) + { + int i; + + for (i = ntokens; i < nsyms; i++) + { + /* don't make these for dummy nonterminals made by gensym. */ + if (*tags[i] != '@') + obstack_fgrow2 (&defines_obstack, + "# define\tNT%s\t%d\n", tags[i], i); + } +#if 0 + /* `fdefines' is now a temporary file, so we need to copy its + contents in `done', so we can't close it here. */ + fclose (fdefines); + fdefines = NULL; +#endif + } + } +} + + +/*------------------------------------------------------------------. +| Set TOKEN_TRANSLATIONS. Check that no two symbols share the same | +| number. | +`------------------------------------------------------------------*/ + +static void +token_translations_init (void) +{ + bucket *bp = NULL; + int i; + + token_translations = XCALLOC (short, max_user_token_number + 1); + + /* Initialize all entries for literal tokens to 2, the internal + token number for $undefined., which represents all invalid + inputs. */ + for (i = 0; i <= max_user_token_number; i++) + token_translations[i] = 2; + + for (bp = firstsymbol; bp; bp = bp->next) + { + /* Non-terminal? */ + if (bp->value >= ntokens) + continue; + /* A token string alias? */ + if (bp->user_token_number == SALIAS) + continue; + + assert (bp->user_token_number != SUNDEF); + + /* A token which translation has already been set? */ + if (token_translations[bp->user_token_number] != 2) + complain (_("tokens %s and %s both assigned number %d"), + tags[token_translations[bp->user_token_number]], + bp->tag, bp->user_token_number); + token_translations[bp->user_token_number] = bp->value; + } +} + + /*------------------------------------------------------------------. | Assign symbol numbers, and write definition of token names into | | FDEFINES. Set up vectors TAGS and SPREC of names and precedences | @@ -1617,22 +1779,21 @@ output_token_defines (struct obstack *oout) static void packsymbols (void) { - bucket *bp; + bucket *bp = NULL; int tokno = 1; - int i; int last_user_token_number; static char DOLLAR[] = "$"; - /* int lossage = 0; JF set but not used */ - tags = XCALLOC (char *, nsyms + 1); - tags[0] = DOLLAR; user_toknums = XCALLOC (short, nsyms + 1); - user_toknums[0] = 0; sprec = XCALLOC (short, nsyms); sassoc = XCALLOC (short, nsyms); + /* The EOF token. */ + tags[0] = DOLLAR; + user_toknums[0] = 0; + max_user_token_number = 256; last_user_token_number = 256; @@ -1685,7 +1846,7 @@ packsymbols (void) if (bp->class == token_sym) { - if (translations && !(bp->user_token_number)) + if (bp->user_token_number == SUNDEF) bp->user_token_number = ++last_user_token_number; if (bp->user_token_number > max_user_token_number) max_user_token_number = bp->user_token_number; @@ -1695,34 +1856,9 @@ packsymbols (void) user_toknums[bp->value] = bp->user_token_number; sprec[bp->value] = bp->prec; sassoc[bp->value] = bp->assoc; - } - if (translations) - { - int j; - - token_translations = XCALLOC (short, max_user_token_number + 1); - - /* initialize all entries for literal tokens to 2, the internal - token number for $undefined., which represents all invalid - inputs. */ - for (j = 0; j <= max_user_token_number; j++) - token_translations[j] = 2; - - for (bp = firstsymbol; bp; bp = bp->next) - { - if (bp->value >= ntokens) - continue; /* non-terminal */ - if (bp->user_token_number == SALIAS) - continue; - if (token_translations[bp->user_token_number] != 2) - complain (_("tokens %s and %s both assigned number %d"), - tags[token_translations[bp->user_token_number]], - bp->tag, bp->user_token_number); - token_translations[bp->user_token_number] = bp->value; - } - } + token_translations_init (); error_token_number = errtoken->value; @@ -1735,36 +1871,6 @@ packsymbols (void) fatal (_("the start symbol %s is a token"), startval->tag); start_symbol = startval->value; - - if (defines_flag) - { - output_token_defines (&defines_obstack); - - if (!pure_parser) - { - if (spec_name_prefix) - obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", - spec_name_prefix); - else - obstack_sgrow (&defines_obstack, - "\nextern YYSTYPE yylval;\n"); - } - - if (semantic_parser) - for (i = ntokens; i < nsyms; i++) - { - /* don't make these for dummy nonterminals made by gensym. */ - if (*tags[i] != '@') - obstack_fgrow2 (&defines_obstack, - "#define\tNT%s\t%d\n", tags[i], i); - } -#if 0 - /* `fdefines' is now a temporary file, so we need to copy its - contents in `done', so we can't close it here. */ - fclose (fdefines); - fdefines = NULL; -#endif - } } @@ -1780,14 +1886,8 @@ packgram (void) int ruleno; symbol_list *p; - bucket *ruleprec; - ritem = XCALLOC (short, nitems + 1); - rlhs = XCALLOC (short, nrules) - 1; - rrhs = XCALLOC (short, nrules) - 1; - rprec = XCALLOC (short, nrules) - 1; - rprecsym = XCALLOC (short, nrules) - 1; - rassoc = XCALLOC (short, nrules) - 1; + rule_table = XCALLOC (rule_t, nrules) - 1; itemno = 0; ruleno = 1; @@ -1795,9 +1895,11 @@ packgram (void) p = grammar; while (p) { - rlhs[ruleno] = p->sym->value; - rrhs[ruleno] = itemno; - ruleprec = p->ruleprec; + bucket *ruleprec = p->ruleprec; + rule_table[ruleno].lhs = p->sym->value; + rule_table[ruleno].rhs = itemno; + rule_table[ruleno].line = p->line; + rule_table[ruleno].useful = TRUE; p = p->next; while (p && p->sym) @@ -1807,8 +1909,8 @@ packgram (void) of the last token in it. */ if (p->sym->class == token_sym) { - rprec[ruleno] = p->sym->prec; - rassoc[ruleno] = p->sym->assoc; + rule_table[ruleno].prec = p->sym->prec; + rule_table[ruleno].assoc = p->sym->assoc; } if (p) p = p->next; @@ -1818,9 +1920,9 @@ packgram (void) the specified symbol's precedence replaces the default. */ if (ruleprec) { - rprec[ruleno] = ruleprec->prec; - rassoc[ruleno] = ruleprec->assoc; - rprecsym[ruleno] = ruleprec->value; + rule_table[ruleno].prec = ruleprec->prec; + rule_table[ruleno].assoc = ruleprec->assoc; + rule_table[ruleno].precsym = ruleprec->value; } ritem[itemno++] = -ruleno; @@ -1831,6 +1933,9 @@ packgram (void) } ritem[itemno] = 0; + + if (trace_flag) + ritem_print (stderr); } /*-------------------------------------------------------------------. @@ -1847,33 +1952,20 @@ reader (void) start_flag = 0; startval = NULL; /* start symbol not specified yet. */ -#if 0 - /* initially assume token number translation not needed. */ - translations = 0; -#endif - /* Nowadays translations is always set to 1, since we give `error' a - user-token-number to satisfy the Posix demand for YYERRCODE==256. - */ - translations = 1; - nsyms = 1; nvars = 0; nrules = 0; nitems = 0; - rline_allocated = 10; - rline = XCALLOC (short, rline_allocated); typed = 0; lastprec = 0; - gensym_count = 0; - semantic_parser = 0; pure_parser = 0; grammar = NULL; - init_lex (); + lex_init (); lineno = 1; /* Initialize the symbol table. */ @@ -1891,7 +1983,6 @@ reader (void) /* Read the declaration section. Copy %{ ... %} groups to TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left, etc. found there. */ - obstack_1grow (&table_obstack, '\n'); obstack_fgrow3 (&table_obstack, "\ /* %s, made from %s\n\ by GNU bison %s. */\n\ @@ -1920,9 +2011,8 @@ reader (void) packsymbols (); /* Convert the grammar into the format described in gram.h. */ packgram (); - /* Free the symbol table data structure since symbols are now all - referred to by symbol number. */ - free_symtab (); + /* Output the headers. */ + symbols_output (); }