X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/f16b08196c780556cbf50691e2944960aebc46f6..8b9e021fd1e3b03568d4b2167c47f33ba822a1c8:/src/symtab.c diff --git a/src/symtab.c b/src/symtab.c index 83851351..54ab0569 100644 --- a/src/symtab.c +++ b/src/symtab.c @@ -1,7 +1,7 @@ /* Symbol table manager for Bison. - Copyright (C) 1984, 1989, 2000, 2001, 2002, 2004, 2005, 2006, 2007 Free - Software Foundation, Inc. + Copyright (C) 1984, 1989, 2000-2002, 2004-2010 Free Software + Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -28,6 +28,13 @@ #include "gram.h" #include "symtab.h" +/*-------------------------------------------------------------------. +| Symbols sorted by tag. Allocated by the first invocation of | +| symbols_do, after which no more symbols should be created. | +`-------------------------------------------------------------------*/ + +static symbol **symbols_sorted = NULL; + /*------------------------. | Distinguished symbols. | `------------------------*/ @@ -58,6 +65,13 @@ symbol_new (uniqstr tag, location loc) symbol *res = xmalloc (sizeof *res); uniqstr_assert (tag); + + /* If the tag is not a string (starts with a double quote), check + that it is valid for Yacc. */ + if (tag[0] != '\"' && tag[0] != '\'' && strchr (tag, '-')) + yacc_at (loc, _("POSIX Yacc forbids dashes in symbol names: %s"), + tag); + res->tag = tag; res->location = loc; @@ -128,6 +142,41 @@ symbol_print (symbol *s, FILE *f) #undef SYMBOL_ATTR_PRINT #undef SYMBOL_CODE_PRINT + +/*----------------------------------. +| Whether S is a valid identifier. | +`----------------------------------*/ + +static bool +is_identifier (uniqstr s) +{ + static char const alphanum[26 + 26 + 1 + 10] = + "abcdefghijklmnopqrstuvwxyz" + "ABCDEFGHIJKLMNOPQRSTUVWXYZ" + "_" + "0123456789"; + if (!s || ! memchr (alphanum, *s, sizeof alphanum - 10)) + return false; + for (++s; *s; ++s) + if (! memchr (alphanum, *s, sizeof alphanum)) + return false; + return true; +} + + +/*-----------------------------------------------. +| Get the identifier associated to this symbol. | +`-----------------------------------------------*/ +uniqstr +symbol_id_get (symbol const *sym) +{ + aver (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS); + if (sym->alias) + sym = sym->alias; + return is_identifier (sym->tag) ? sym->tag : 0; +} + + /*------------------------------------------------------------------. | Complain that S's WHAT is redeclared at SECOND, and was first set | | at FIRST. | @@ -150,6 +199,7 @@ semantic_type_redeclaration (semantic_type *s, const char *what, location first, } + /*-----------------------------------------------------------------. | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 | | as TYPE_NAME. | @@ -168,7 +218,6 @@ symbol_type_set (symbol *sym, uniqstr type_name, location loc) } } - /*-----------------------------------------. | Set the DESTRUCTOR associated with SYM. | `-----------------------------------------*/ @@ -341,9 +390,7 @@ symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) { int *user_token_numberp; - aver (sym->class == token_sym); - - if (sym->user_token_number != USER_NUMBER_ALIAS) + if (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) user_token_numberp = &sym->user_token_number; else user_token_numberp = &sym->alias->user_token_number; @@ -392,29 +439,24 @@ symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) } -/*------------------------------------------------------------------. -| Declare the new symbol SYM. Make it an alias of SYMVAL, and type | -| SYMVAL with SYM's type. | -`------------------------------------------------------------------*/ - void -symbol_make_alias (symbol *sym, symbol *symval, location loc) +symbol_make_alias (symbol *sym, symbol *str, location loc) { - if (symval->alias) + if (str->alias) warn_at (loc, _("symbol `%s' used more than once as a literal string"), - symval->tag); + str->tag); else if (sym->alias) warn_at (loc, _("symbol `%s' given more than one literal string"), sym->tag); else { - symval->class = token_sym; - symval->user_token_number = sym->user_token_number; - sym->user_token_number = USER_NUMBER_ALIAS; - symval->alias = sym; - sym->alias = symval; - symval->number = sym->number; - symbol_type_set (symval, sym->type_name, loc); + str->class = token_sym; + str->user_token_number = sym->user_token_number; + sym->user_token_number = USER_NUMBER_HAS_STRING_ALIAS; + str->alias = sym; + sym->alias = str; + str->number = sym->number; + symbol_type_set (str, sym->type_name, loc); } } @@ -427,46 +469,47 @@ symbol_make_alias (symbol *sym, symbol *symval, location loc) static inline void symbol_check_alias_consistency (symbol *this) { - symbol *alias = this; - symbol *orig = this->alias; + symbol *sym = this; + symbol *str = this->alias; - /* Check only those that _are_ the aliases. */ - if (!(this->alias && this->user_token_number == USER_NUMBER_ALIAS)) + /* Check only the symbol in the symbol-string pair. */ + if (!(this->alias + && this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS)) return; - if (orig->type_name != alias->type_name) + if (str->type_name != sym->type_name) { - if (orig->type_name) - symbol_type_set (alias, orig->type_name, orig->type_location); + if (str->type_name) + symbol_type_set (sym, str->type_name, str->type_location); else - symbol_type_set (orig, alias->type_name, alias->type_location); + symbol_type_set (str, sym->type_name, sym->type_location); } - if (orig->destructor.code || alias->destructor.code) + if (str->destructor.code || sym->destructor.code) { - if (orig->destructor.code) - symbol_destructor_set (alias, &orig->destructor); + if (str->destructor.code) + symbol_destructor_set (sym, &str->destructor); else - symbol_destructor_set (orig, &alias->destructor); + symbol_destructor_set (str, &sym->destructor); } - if (orig->printer.code || alias->printer.code) + if (str->printer.code || sym->printer.code) { - if (orig->printer.code) - symbol_printer_set (alias, &orig->printer); + if (str->printer.code) + symbol_printer_set (sym, &str->printer); else - symbol_printer_set (orig, &alias->printer); + symbol_printer_set (str, &sym->printer); } - if (alias->prec || orig->prec) + if (sym->prec || str->prec) { - if (orig->prec) - symbol_precedence_set (alias, orig->prec, orig->assoc, - orig->prec_location); + if (str->prec) + symbol_precedence_set (sym, str->prec, str->assoc, + str->prec_location); else - symbol_precedence_set (orig, alias->prec, alias->assoc, - alias->prec_location); + symbol_precedence_set (str, sym->prec, sym->assoc, + sym->prec_location); } } @@ -487,31 +530,11 @@ symbol_check_alias_consistency_processor (void *this, static inline bool symbol_pack (symbol *this) { + aver (this->number != NUMBER_UNDEFINED); if (this->class == nterm_sym) - { - this->number += ntokens; - } - else if (this->alias) - { - /* This symbol and its alias are a single token defn. - Allocate a tokno, and assign to both check agreement of - prec and assoc fields and make both the same */ - if (this->number == NUMBER_UNDEFINED) - { - if (this == endtoken || this->alias == endtoken) - this->number = this->alias->number = 0; - else - { - aver (this->alias->number != NUMBER_UNDEFINED); - this->number = this->alias->number; - } - } - /* Do not do processing below for USER_NUMBER_ALIASes. */ - if (this->user_token_number == USER_NUMBER_ALIAS) - return true; - } - else /* this->class == token_sym */ - aver (this->number != NUMBER_UNDEFINED); + this->number += ntokens; + else if (this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS) + return true; symbols[this->number] = this; return true; @@ -524,7 +547,26 @@ symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED) } +static void +user_token_number_redeclaration (int num, symbol *first, symbol *second) +{ + /* User token numbers are not assigned during the parsing, but in a + second step, via a traversal of the symbol table sorted on tag. + However, error messages make more sense if we keep the first + declaration first. */ + if (location_cmp (first->location, second->location) > 0) + { + symbol* tmp = first; + first = second; + second = tmp; + } + complain_at (second->location, + _("user token number %d redeclaration for %s"), + num, second->tag); + complain_at (first->location, _("previous declaration for %s"), + first->tag); +} /*--------------------------------------------------. | Put THIS in TOKEN_TRANSLATIONS if it is a token. | @@ -535,14 +577,14 @@ symbol_translation (symbol *this) { /* Non-terminal? */ if (this->class == token_sym - && this->user_token_number != USER_NUMBER_ALIAS) + && this->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) { /* A token which translation has already been set? */ if (token_translations[this->user_token_number] != undeftoken->number) - complain_at (this->location, - _("tokens %s and %s both assigned number %d"), - symbols[token_translations[this->user_token_number]]->tag, - this->tag, this->user_token_number); + user_token_number_redeclaration + (this->user_token_number, + symbols[token_translations[this->user_token_number]], + this); token_translations[this->user_token_number] = this->number; } @@ -656,8 +698,10 @@ symbol_from_uniqstr (const uniqstr key, location loc) if (!entry) { /* First insertion in the hash. */ + aver (!symbols_sorted); entry = symbol_new (key, loc); - hash_insert (symbol_table, entry); + if (!hash_insert (symbol_table, entry)) + xalloc_die (); } return entry; } @@ -681,7 +725,8 @@ semantic_type_from_uniqstr (const uniqstr key) { /* First insertion in the hash. */ entry = semantic_type_new (key); - hash_insert (semantic_type_table, entry); + if (!hash_insert (semantic_type_table, entry)) + xalloc_die (); } return entry; } @@ -748,6 +793,7 @@ symbols_free (void) hash_free (symbol_table); hash_free (semantic_type_table); free (symbols); + free (symbols_sorted); } @@ -756,13 +802,36 @@ symbols_free (void) | terminals. | `---------------------------------------------------------------*/ +static int +symbols_cmp (symbol const *a, symbol const *b) +{ + return strcmp (a->tag, b->tag); +} + +static int +symbols_cmp_qsort (void const *a, void const *b) +{ + return symbols_cmp (*(symbol * const *)a, *(symbol * const *)b); +} + static void symbols_do (Hash_processor processor, void *processor_data) { - hash_do_for_each (symbol_table, processor, processor_data); + size_t count = hash_get_n_entries (symbol_table); + if (!symbols_sorted) + { + symbols_sorted = xnmalloc (count, sizeof *symbols_sorted); + hash_get_entries (symbol_table, (void**)symbols_sorted, count); + qsort (symbols_sorted, count, sizeof *symbols_sorted, + symbols_cmp_qsort); + } + { + size_t i; + for (i = 0; i < count; ++i) + processor (symbols_sorted[i], processor_data); + } } - /*--------------------------------------------------------------. | Check that all the symbols are defined. Report any undefined | | symbols and consider them nonterminals. | @@ -821,9 +890,8 @@ symbols_token_translations_init (void) token_translations = xnmalloc (max_user_token_number + 1, sizeof *token_translations); - /* Initialize all entries for literal tokens to 2, the internal - token number for $undefined, which represents all invalid inputs. - */ + /* Initialize all entries for literal tokens to the internal token + number for $undefined, which represents all invalid inputs. */ for (i = 0; i < max_user_token_number + 1; i++) token_translations[i] = undeftoken->number; symbols_do (symbol_translation_processor, NULL);