X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/9874f80b2e813a8ad8e5562dbbaa30ecafd5e7d8..297e263a0050959e0fd139ad66e71383fc9ac4db:/src/symtab.c diff --git a/src/symtab.c b/src/symtab.c index 35b61dea..124da372 100644 --- a/src/symtab.c +++ b/src/symtab.c @@ -22,7 +22,6 @@ #include "system.h" #include -#include #include "complain.h" #include "gram.h" @@ -34,6 +33,7 @@ `-------------------------------------------------------------------*/ static symbol **symbols_sorted = NULL; +static symbol **semantic_types_sorted = NULL; /*------------------------. | Distinguished symbols. | @@ -50,10 +50,16 @@ location startsymbol_location; | Default %destructor's and %printer's. | `---------------------------------------*/ -static code_props default_tagged_destructor = CODE_PROPS_NONE_INIT; -static code_props default_tagless_destructor = CODE_PROPS_NONE_INIT; -static code_props default_tagged_printer = CODE_PROPS_NONE_INIT; -static code_props default_tagless_printer = CODE_PROPS_NONE_INIT; +static code_props default_tagged_code_props[CODE_PROPS_SIZE] = + { + CODE_PROPS_NONE_INIT, + CODE_PROPS_NONE_INIT, + }; +static code_props default_tagless_code_props[CODE_PROPS_SIZE] = + { + CODE_PROPS_NONE_INIT, + CODE_PROPS_NONE_INIT, + }; /*---------------------------------. | Create a new symbol, named TAG. | @@ -68,7 +74,7 @@ symbol_new (uniqstr tag, location loc) /* If the tag is not a string (starts with a double quote), check that it is valid for Yacc. */ - if (tag[0] != '\"' && tag[0] != '\'' && mbschr (tag, '-')) + if (tag[0] != '\"' && tag[0] != '\'' && strchr (tag, '-')) yacc_at (loc, _("POSIX Yacc forbids dashes in symbol names: %s"), tag); @@ -76,8 +82,8 @@ symbol_new (uniqstr tag, location loc) res->location = loc; res->type_name = NULL; - code_props_none_init (&res->destructor); - code_props_none_init (&res->printer); + for (int i = 0; i < CODE_PROPS_SIZE; ++i) + code_props_none_init (&res->props[i]); res->number = NUMBER_UNDEFINED; res->prec = 0; @@ -86,28 +92,43 @@ symbol_new (uniqstr tag, location loc) res->alias = NULL; res->class = unknown_sym; - res->declared = false; + res->status = undeclared; if (nsyms == SYMBOL_NUMBER_MAXIMUM) fatal (_("too many symbols in input grammar (limit is %d)"), - SYMBOL_NUMBER_MAXIMUM); + SYMBOL_NUMBER_MAXIMUM); nsyms++; return res; } +char const * +code_props_type_string (code_props_type kind) +{ + switch (kind) + { + case destructor: + return "%destructor"; + case printer: + return "%printer"; + } + assert (0); +} + /*----------------------------------------. | Create a new semantic type, named TAG. | `----------------------------------------*/ static semantic_type * -semantic_type_new (uniqstr tag) +semantic_type_new (uniqstr tag, const location *loc) { semantic_type *res = xmalloc (sizeof *res); uniqstr_assert (tag); res->tag = tag; - code_props_none_init (&res->destructor); - code_props_none_init (&res->printer); + if (loc) + res->location = *loc; + for (int i = 0; i < CODE_PROPS_SIZE; ++i) + code_props_none_init (&res->props[i]); return res; } @@ -117,13 +138,13 @@ semantic_type_new (uniqstr tag) | Print a symbol. | `-----------------*/ -#define SYMBOL_ATTR_PRINT(Attr) \ - if (s->Attr) \ +#define SYMBOL_ATTR_PRINT(Attr) \ + if (s->Attr) \ fprintf (f, " %s { %s }", #Attr, s->Attr) -#define SYMBOL_CODE_PRINT(Attr) \ - if (s->Attr.code) \ - fprintf (f, " %s { %s }", #Attr, s->Attr.code) +#define SYMBOL_CODE_PRINT(Attr) \ + if (s->props[Attr].code) \ + fprintf (f, " %s { %s }", #Attr, s->props[Attr].code) void symbol_print (symbol *s, FILE *f) @@ -142,6 +163,41 @@ 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. | @@ -176,122 +232,72 @@ symbol_type_set (symbol *sym, uniqstr type_name, location loc) if (type_name) { if (sym->type_name) - symbol_redeclaration (sym, "%type", sym->type_location, loc); + symbol_redeclaration (sym, "%type", sym->type_location, loc); uniqstr_assert (type_name); sym->type_name = type_name; sym->type_location = loc; } } -/*-----------------------------------------. -| Set the DESTRUCTOR associated with SYM. | -`-----------------------------------------*/ - -void -symbol_destructor_set (symbol *sym, code_props const *destructor) -{ - if (sym->destructor.code) - symbol_redeclaration (sym, "%destructor", sym->destructor.location, - destructor->location); - sym->destructor = *destructor; -} - -/*------------------------------------------. -| Set the DESTRUCTOR associated with TYPE. | -`------------------------------------------*/ - -void -semantic_type_destructor_set (semantic_type *type, - code_props const *destructor) -{ - if (type->destructor.code) - semantic_type_redeclaration (type, "%destructor", - type->destructor.location, - destructor->location); - type->destructor = *destructor; -} - -/*---------------------------------------. -| Get the computed %destructor for SYM. | -`---------------------------------------*/ - -code_props const * -symbol_destructor_get (symbol const *sym) -{ - /* Per-symbol %destructor. */ - if (sym->destructor.code) - return &sym->destructor; - - /* Per-type %destructor. */ - if (sym->type_name) - { - code_props const *destructor = - &semantic_type_get (sym->type_name)->destructor; - if (destructor->code) - return destructor; - } - - /* Apply default %destructor's only to user-defined symbols. */ - if (sym->tag[0] == '$' || sym == errtoken) - return &code_props_none; - - if (sym->type_name) - return &default_tagged_destructor; - return &default_tagless_destructor; -} - -/*--------------------------------------. -| Set the PRINTER associated with SYM. | -`--------------------------------------*/ +/*--------------------------------------------------------. +| Set the DESTRUCTOR or PRINTER associated with the SYM. | +`--------------------------------------------------------*/ void -symbol_printer_set (symbol *sym, code_props const *printer) +symbol_code_props_set (symbol *sym, code_props_type kind, + code_props const *code) { - if (sym->printer.code) - symbol_redeclaration (sym, "%printer", - sym->printer.location, printer->location); - sym->printer = *printer; + if (sym->props[kind].code) + symbol_redeclaration (sym, code_props_type_string (kind), + sym->props[kind].location, + code->location); + sym->props[kind] = *code; } -/*---------------------------------------. -| Set the PRINTER associated with TYPE. | -`---------------------------------------*/ +/*-----------------------------------------------------. +| Set the DESTRUCTOR or PRINTER associated with TYPE. | +`-----------------------------------------------------*/ void -semantic_type_printer_set (semantic_type *type, code_props const *printer) +semantic_type_code_props_set (semantic_type *type, + code_props_type kind, + code_props const *code) { - if (type->printer.code) - semantic_type_redeclaration (type, "%printer", - type->printer.location, printer->location); - type->printer = *printer; + if (type->props[kind].code) + semantic_type_redeclaration (type, code_props_type_string (kind), + type->props[kind].location, + code->location); + type->props[kind] = *code; } -/*------------------------------------. -| Get the computed %printer for SYM. | -`------------------------------------*/ +/*---------------------------------------------------. +| Get the computed %destructor or %printer for SYM. | +`---------------------------------------------------*/ code_props const * -symbol_printer_get (symbol const *sym) +symbol_code_props_get (symbol const *sym, + code_props_type kind) { - /* Per-symbol %printer. */ - if (sym->printer.code) - return &sym->printer; + /* Per-symbol code props. */ + if (sym->props[kind].code) + return &sym->props[kind]; - /* Per-type %printer. */ + /* Per-type code props. */ if (sym->type_name) { - code_props const *printer = &semantic_type_get (sym->type_name)->printer; - if (printer->code) - return printer; + code_props const *code = + &semantic_type_get (sym->type_name, NULL)->props[kind]; + if (code->code) + return code; } - /* Apply the default %printer only to user-defined symbols. */ + /* Apply default code props's only to user-defined symbols. */ if (sym->tag[0] == '$' || sym == errtoken) return &code_props_none; if (sym->type_name) - return &default_tagged_printer; - return &default_tagless_printer; + return &default_tagged_code_props[kind]; + return &default_tagless_code_props[kind]; } /*-----------------------------------------------------------------. @@ -305,7 +311,7 @@ symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) if (a != undef_assoc) { if (sym->prec != 0) - symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, + symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, loc); sym->prec = prec; sym->assoc = a; @@ -324,10 +330,12 @@ symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) void symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) { + bool warned = false; if (sym->class != unknown_sym && sym->class != class) { complain_at (loc, _("symbol %s redefined"), sym->tag); - sym->declared = false; + // Don't report both "redefined" and "redeclared". + warned = true; } if (class == nterm_sym && sym->class != nterm_sym) @@ -339,9 +347,9 @@ symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) if (declaring) { - if (sym->declared) - warn_at (loc, _("symbol %s redeclared"), sym->tag); - sym->declared = true; + if (sym->status == declared && !warned) + warn_at (loc, _("symbol %s redeclared"), sym->tag); + sym->status = declared; } } @@ -369,7 +377,7 @@ symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) { endtoken = sym; /* It is always mapped to 0, so it was already counted in - NTOKENS. */ + NTOKENS. */ if (endtoken->number != NUMBER_UNDEFINED) --ntokens; endtoken->number = 0; @@ -387,14 +395,68 @@ symbol_check_defined (symbol *sym) { if (sym->class == unknown_sym) { - complain_at - (sym->location, - _("symbol %s is used, but is not defined as a token and has no rules"), - sym->tag); + switch (sym->status) + { + case used: + warn_at (sym->location, + _("symbol %s is used, but is not defined as a token" + " and has no rules"), + sym->tag); + break; + case undeclared: + case needed: + complain_at (sym->location, + _("symbol %s is used, but is not defined as a token" + " and has no rules"), + sym->tag); + break; + case declared: + /* If declared, then sym->class != unknown_sym. */ + assert (0); + } + sym->class = nterm_sym; sym->number = nvars++; } + for (int i = 0; i < 2; ++i) + if (sym->props[i].kind == CODE_PROPS_NONE && sym->type_name) + { + semantic_type *sem_type = semantic_type_get (sym->type_name, NULL); + if (sem_type + && sem_type->props[i].kind != CODE_PROPS_NONE) + sem_type->props[i].is_used = true; + } + + /* Set the semantic type status associated to the current symbol to + 'declared' so that we could check semantic types unnecessary uses. */ + if (sym->type_name) + { + semantic_type *sem_type = semantic_type_get (sym->type_name, NULL); + if (sem_type) + sem_type->status = declared; + } + + return true; +} + +static inline bool +semantic_type_check_defined (semantic_type *sem_type) +{ + if (sem_type->status == declared) + { + for (int i = 0; i < 2; ++i) + if (sem_type->props[i].kind != CODE_PROPS_NONE + && ! sem_type->props[i].is_used) + warn_at (sem_type->location, + _("useless %s for type <%s>"), + code_props_type_string (i), sem_type->tag); + } + else + warn_at (sem_type->location, + _("type <%s> is used, but is not associated to any symbol"), + sem_type->tag); + return true; } @@ -404,15 +466,22 @@ symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) return symbol_check_defined (sym); } +static bool +semantic_type_check_defined_processor (void *sem_type, + void *null ATTRIBUTE_UNUSED) +{ + return semantic_type_check_defined (sem_type); +} + void symbol_make_alias (symbol *sym, symbol *str, location loc) { if (str->alias) - warn_at (loc, _("symbol '%s' used more than once as a literal string"), + warn_at (loc, _("symbol %s used more than once as a literal string"), str->tag); else if (sym->alias) - warn_at (loc, _("symbol '%s' given more than one literal string"), + warn_at (loc, _("symbol %s given more than one literal string"), sym->tag); else { @@ -446,42 +515,32 @@ symbol_check_alias_consistency (symbol *this) if (str->type_name != sym->type_name) { if (str->type_name) - symbol_type_set (sym, str->type_name, str->type_location); + symbol_type_set (sym, str->type_name, str->type_location); else - symbol_type_set (str, sym->type_name, sym->type_location); + symbol_type_set (str, sym->type_name, sym->type_location); } - if (str->destructor.code || sym->destructor.code) - { - if (str->destructor.code) - symbol_destructor_set (sym, &str->destructor); - else - symbol_destructor_set (str, &sym->destructor); - } - - if (str->printer.code || sym->printer.code) - { - if (str->printer.code) - symbol_printer_set (sym, &str->printer); - else - symbol_printer_set (str, &sym->printer); - } + for (int i = 0; i < CODE_PROPS_SIZE; ++i) + if (str->props[i].code) + symbol_code_props_set (sym, i, &str->props[i]); + else if (sym->props[i].code) + symbol_code_props_set (str, i, &sym->props[i]); if (sym->prec || str->prec) { if (str->prec) - symbol_precedence_set (sym, str->prec, str->assoc, - str->prec_location); + symbol_precedence_set (sym, str->prec, str->assoc, + str->prec_location); else - symbol_precedence_set (str, sym->prec, sym->assoc, - sym->prec_location); + symbol_precedence_set (str, sym->prec, sym->assoc, + sym->prec_location); } } static bool symbol_check_alias_consistency_processor (void *this, - void *null ATTRIBUTE_UNUSED) + void *null ATTRIBUTE_UNUSED) { symbol_check_alias_consistency (this); return true; @@ -547,7 +606,7 @@ symbol_translation (symbol *this) { /* A token which translation has already been set? */ if (token_translations[this->user_token_number] != undeftoken->number) - user_token_number_redeclaration + user_token_number_redeclaration (this->user_token_number, symbols[token_translations[this->user_token_number]], this); @@ -635,15 +694,15 @@ void symbols_new (void) { symbol_table = hash_initialize (HT_INITIAL_CAPACITY, - NULL, - hash_symbol_hasher, - hash_symbol_comparator, - free); + NULL, + hash_symbol_hasher, + hash_symbol_comparator, + free); semantic_type_table = hash_initialize (HT_INITIAL_CAPACITY, - NULL, - hash_semantic_type_hasher, - hash_semantic_type_comparator, - free); + NULL, + hash_semantic_type_hasher, + hash_semantic_type_comparator, + free); } @@ -679,7 +738,7 @@ symbol_from_uniqstr (const uniqstr key, location loc) `-----------------------------------------------------------------------*/ semantic_type * -semantic_type_from_uniqstr (const uniqstr key) +semantic_type_from_uniqstr (const uniqstr key, const location *loc) { semantic_type probe; semantic_type *entry; @@ -690,7 +749,7 @@ semantic_type_from_uniqstr (const uniqstr key) if (!entry) { /* First insertion in the hash. */ - entry = semantic_type_new (key); + entry = semantic_type_new (key, loc); if (!hash_insert (semantic_type_table, entry)) xalloc_die (); } @@ -716,9 +775,9 @@ symbol_get (const char *key, location loc) `-----------------------------------------------------------------------*/ semantic_type * -semantic_type_get (const char *key) +semantic_type_get (const char *key, const location *loc) { - return semantic_type_from_uniqstr (uniqstr_new (key)); + return semantic_type_from_uniqstr (uniqstr_new (key), loc); } @@ -781,20 +840,20 @@ symbols_cmp_qsort (void const *a, void const *b) } static void -symbols_do (Hash_processor processor, void *processor_data) +symbols_do (Hash_processor processor, void *processor_data, + struct hash_table *table, symbol **sorted) { - size_t count = hash_get_n_entries (symbol_table); - if (!symbols_sorted) + size_t count = hash_get_n_entries (table); + if (!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); + sorted = xnmalloc (count, sizeof *sorted); + hash_get_entries (table, (void**)sorted, count); + qsort (sorted, count, sizeof *sorted, symbols_cmp_qsort); } { size_t i; for (i = 0; i < count; ++i) - processor (symbols_sorted[i], processor_data); + processor (sorted[i], processor_data); } } @@ -806,7 +865,10 @@ symbols_do (Hash_processor processor, void *processor_data) void symbols_check_defined (void) { - symbols_do (symbol_check_defined_processor, NULL); + symbols_do (symbol_check_defined_processor, NULL, + symbol_table, symbols_sorted); + symbols_do (semantic_type_check_defined_processor, NULL, + semantic_type_table, semantic_types_sorted); } /*------------------------------------------------------------------. @@ -827,12 +889,12 @@ symbols_token_translations_init (void) { symbol *this = symbols[i]; if (this->user_token_number != USER_NUMBER_UNDEFINED) - { - if (this->user_token_number > max_user_token_number) - max_user_token_number = this->user_token_number; - if (this->user_token_number == 256) - num_256_available_p = false; - } + { + if (this->user_token_number > max_user_token_number) + max_user_token_number = this->user_token_number; + if (this->user_token_number == 256) + num_256_available_p = false; + } } /* If 256 is not used, assign it to error, to follow POSIX. */ @@ -848,20 +910,20 @@ symbols_token_translations_init (void) { symbol *this = symbols[i]; if (this->user_token_number == USER_NUMBER_UNDEFINED) - this->user_token_number = ++max_user_token_number; + this->user_token_number = ++max_user_token_number; if (this->user_token_number > max_user_token_number) - max_user_token_number = this->user_token_number; + max_user_token_number = this->user_token_number; } token_translations = xnmalloc (max_user_token_number + 1, - sizeof *token_translations); + 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); + symbols_do (symbol_translation_processor, NULL, + symbol_table, symbols_sorted); } @@ -873,10 +935,11 @@ symbols_token_translations_init (void) void symbols_pack (void) { - symbols_do (symbol_check_alias_consistency_processor, NULL); + symbols_do (symbol_check_alias_consistency_processor, NULL, + symbol_table, symbols_sorted); symbols = xcalloc (nsyms, sizeof *symbols); - symbols_do (symbol_pack_processor, NULL); + symbols_do (symbol_pack_processor, NULL, symbol_table, symbols_sorted); /* Aliases leave empty slots in symbols, so remove them. */ { @@ -906,12 +969,12 @@ symbols_pack (void) if (startsymbol->class == unknown_sym) fatal_at (startsymbol_location, - _("the start symbol %s is undefined"), - startsymbol->tag); + _("the start symbol %s is undefined"), + startsymbol->tag); else if (startsymbol->class == token_sym) fatal_at (startsymbol_location, - _("the start symbol %s is a token"), - startsymbol->tag); + _("the start symbol %s is a token"), + startsymbol->tag); } @@ -920,53 +983,29 @@ symbols_pack (void) `--------------------------------------------------*/ void -default_tagged_destructor_set (code_props const *destructor) -{ - if (default_tagged_destructor.code) - { - complain_at (destructor->location, - _("redeclaration for default tagged %%destructor")); - complain_at (default_tagged_destructor.location, - _("previous declaration")); - } - default_tagged_destructor = *destructor; -} - -void -default_tagless_destructor_set (code_props const *destructor) -{ - if (default_tagless_destructor.code) - { - complain_at (destructor->location, - _("redeclaration for default tagless %%destructor")); - complain_at (default_tagless_destructor.location, - _("previous declaration")); - } - default_tagless_destructor = *destructor; -} - -void -default_tagged_printer_set (code_props const *printer) +default_tagged_code_props_set (code_props_type kind, code_props const *code) { - if (default_tagged_printer.code) + if (default_tagged_code_props[kind].code) { - complain_at (printer->location, - _("redeclaration for default tagged %%printer")); - complain_at (default_tagged_printer.location, - _("previous declaration")); + complain_at (code->location, + _("redeclaration for default tagged %s"), + code_props_type_string (kind)); + complain_at (default_tagged_code_props[kind].location, + _("previous declaration")); } - default_tagged_printer = *printer; + default_tagged_code_props[kind] = *code; } void -default_tagless_printer_set (code_props const *printer) +default_tagless_code_props_set (code_props_type kind, code_props const *code) { - if (default_tagless_printer.code) + if (default_tagless_code_props[kind].code) { - complain_at (printer->location, - _("redeclaration for default tagless %%printer")); - complain_at (default_tagless_printer.location, - _("previous declaration")); + complain_at (code->location, + _("redeclaration for default tagless %s"), + code_props_type_string (kind)); + complain_at (default_tagless_code_props[kind].location, + _("previous declaration")); } - default_tagless_printer = *printer; + default_tagless_code_props[kind] = *code; }