X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/affac6132ae6ee354e1497e456b5c81a0c84334f..868d2d96387042d8aba99fe427e76d186a7a4759:/src/reader.c diff --git a/src/reader.c b/src/reader.c index b55b0fa9..928be1b8 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,7 +1,7 @@ /* Input parser for Bison Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000, 2001, 2002, 2003, - 2005 Free Software Foundation, Inc. + 2005, 2006 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -22,6 +22,7 @@ #include #include "system.h" +#include #include @@ -31,17 +32,23 @@ #include "getargs.h" #include "gram.h" #include "muscle_tab.h" -#include "output.h" #include "reader.h" #include "symlist.h" #include "symtab.h" +#include "scan-gram.h" +#include "scan-code.h" + +static void check_and_convert_grammar (void); static symbol_list *grammar = NULL; static bool start_flag = false; merger_list *merge_functions; /* Was %union seen? */ -bool typed = false; +bool union_seen = false; + +/* Was a tag seen? */ +bool tag_seen = false; /* Should rules have a default precedence? */ bool default_prec = true; @@ -64,18 +71,20 @@ grammar_start_symbol_set (symbol *sym, location loc) } -/*----------------------------------------------------------------. -| There are two prologues: one before %union, one after. Augment | -| the current one. | -`----------------------------------------------------------------*/ +/*---------------------------------------------------------------------. +| There are two prologues: one before the first %union and one after. | +| Augment the one specified by POST. | +`---------------------------------------------------------------------*/ void -prologue_augment (const char *prologue, location loc) +prologue_augment (const char *prologue, location loc, bool post) { struct obstack *oout = - !typed ? &pre_prologue_obstack : &post_prologue_obstack; + !post ? &pre_prologue_obstack : &post_prologue_obstack; obstack_fgrow1 (oout, "]b4_syncline(%d, [[", loc.start.line); + /* FIXME: Protection of M4 characters missing here. See + output.c:escaped_output. */ MUSCLE_OBSTACK_SGROW (oout, quotearg_style (c_quoting_style, loc.start.file)); obstack_sgrow (oout, "]])[\n"); @@ -84,14 +93,13 @@ prologue_augment (const char *prologue, location loc) -/*-------------------------------------------------------------------. -| Return the merger index for a merging function named NAME, whose | -| arguments have type TYPE. Records the function, if new, in | -| MERGER_LIST. | -`-------------------------------------------------------------------*/ +/*------------------------------------------------------------------------. +| Return the merger index for a merging function named NAME. Records the | +| function, if new, in MERGER_LIST. | +`------------------------------------------------------------------------*/ static int -get_merge_function (uniqstr name, uniqstr type, location loc) +get_merge_function (uniqstr name) { merger_list *syms; merger_list head; @@ -100,9 +108,6 @@ get_merge_function (uniqstr name, uniqstr type, location loc) if (! glr_parser) return 0; - if (type == NULL) - type = uniqstr_new (""); - head.next = merge_functions; for (syms = &head, n = 1; syms->next; syms = syms->next, n += 1) if (UNIQSTR_EQ (name, syms->next->name)) @@ -111,16 +116,51 @@ get_merge_function (uniqstr name, uniqstr type, location loc) { syms->next = xmalloc (sizeof syms->next[0]); syms->next->name = uniqstr_new (name); - syms->next->type = uniqstr_new (type); + /* After all symbol type declarations have been parsed, packgram invokes + record_merge_function_type to set the type. */ + syms->next->type = NULL; syms->next->next = NULL; merge_functions = head.next; } - else if (!UNIQSTR_EQ (type, syms->next->type)) - warn_at (loc, _("result type clash on merge function %s: <%s> != <%s>"), - name, type, syms->next->type); return n; } +/*-------------------------------------------------------------------------. +| For the existing merging function with index MERGER, record the result | +| type as TYPE as required by the lhs of the rule whose %merge declaration | +| is at DECLARATION_LOC. | +`-------------------------------------------------------------------------*/ + +static void +record_merge_function_type (int merger, uniqstr type, location declaration_loc) +{ + int merger_find; + merger_list *merge_function; + + if (merger <= 0) + return; + + if (type == NULL) + type = uniqstr_new (""); + + merger_find = 1; + for (merge_function = merge_functions; + merge_function != NULL && merger_find != merger; + merge_function = merge_function->next) + merger_find += 1; + assert (merge_function != NULL && merger_find == merger); + if (merge_function->type != NULL && !UNIQSTR_EQ (merge_function->type, type)) + { + complain_at (declaration_loc, + _("result type clash on merge function `%s': <%s> != <%s>"), + merge_function->name, type, merge_function->type); + complain_at (merge_function->type_declaration_location, + _("previous declaration")); + } + merge_function->type = uniqstr_new (type); + merge_function->type_declaration_location = declaration_loc; +} + /*--------------------------------------. | Free all merge-function definitions. | `--------------------------------------*/ @@ -153,7 +193,7 @@ free_merger_functions (void) static symbol_list *grammar_end = NULL; /* Append SYM to the grammar. */ -void +static void grammar_symbol_append (symbol *sym, location loc) { symbol_list *p = symbol_list_new (sym, loc); @@ -164,12 +204,17 @@ grammar_symbol_append (symbol *sym, location loc) grammar = p; grammar_end = p; + + /* A null SYM stands for an end of rule; it is not an actual + part of it. */ + if (sym) + ++nritems; } /* The rule currently being defined, and the previous rule. CURRENT_RULE points to the first LHS of the current rule, while PREVIOUS_RULE_END points to the *end* of the previous rule (NULL). */ -symbol_list *current_rule = NULL; +static symbol_list *current_rule = NULL; static symbol_list *previous_rule_end = NULL; @@ -178,7 +223,7 @@ static symbol_list *previous_rule_end = NULL; `----------------------------------------------*/ void -grammar_rule_begin (symbol *lhs, location loc) +grammar_current_rule_begin (symbol *lhs, location loc) { if (!start_flag) { @@ -189,8 +234,6 @@ grammar_rule_begin (symbol *lhs, location loc) /* Start a new rule and record its lhs. */ ++nrules; - ++nritems; - previous_rule_end = grammar_end; grammar_symbol_append (lhs, loc); current_rule = grammar_end; @@ -207,17 +250,27 @@ grammar_rule_begin (symbol *lhs, location loc) } -/*------------------------------------------------------------------. -| Check that the last rule (CURRENT_RULE) is properly defined. For | -| instance, there should be no type clash on the default action. | -`------------------------------------------------------------------*/ +/*----------------------------------------------------------------------. +| A symbol should be used if it has a destructor, or if it is a | +| mid-rule symbol (i.e., the generated LHS replacing a mid-rule | +| action) that was assigned to, as in "exp: { $$ = 1; } { $$ = $1; }". | +`----------------------------------------------------------------------*/ -static void -grammar_current_rule_check (void) +static bool +symbol_should_be_used (symbol_list const *s) { - symbol *lhs = current_rule->sym; - char const *lhs_type = lhs->type_name; + return (s->sym->destructor + || (s->midrule && s->midrule->used)); +} + +/*----------------------------------------------------------------. +| Check that the rule R is properly defined. For instance, there | +| should be no type clash on the default action. | +`----------------------------------------------------------------*/ +static void +grammar_rule_check (const symbol_list *r) +{ /* Type check. If there is an action, then there is nothing we can do: the user @@ -225,36 +278,42 @@ grammar_current_rule_check (void) Don't worry about the default action if $$ is untyped, since $$'s value can't be used. */ - if (!current_rule->action && lhs_type) + if (!r->action && r->sym->type_name) { - symbol *first_rhs = current_rule->next->sym; + symbol *first_rhs = r->next->sym; /* If $$ is being set in default way, report if any type mismatch. */ if (first_rhs) { + char const *lhs_type = r->sym->type_name; const char *rhs_type = first_rhs->type_name ? first_rhs->type_name : ""; if (!UNIQSTR_EQ (lhs_type, rhs_type)) - warn_at (current_rule->location, + warn_at (r->location, _("type clash on default action: <%s> != <%s>"), lhs_type, rhs_type); } /* Warn if there is no default for $$ but we need one. */ else - warn_at (current_rule->location, + warn_at (r->location, _("empty rule for typed nonterminal, and no action")); } - /* Check that all the symbol values are used. */ - if (typed) - { - symbol_list *l = current_rule; - int n = 1; - for (l = current_rule->next; l && l->sym; l = l->next, ++n) - /* The default action `uses' $1. */ - if (! (!current_rule->action && n == 1) - && l->sym->type_name && !l->used) - warn_at (current_rule->location, _("unused value: $%d"), n); - } + /* Check that symbol values that should be used are in fact used. */ + { + symbol_list const *l = r; + int n = 0; + for (; l && l->sym; l = l->next, ++n) + if (! (l->used + || !symbol_should_be_used (l) + /* The default action, $$ = $1, `uses' both. */ + || (!r->action && (n == 0 || n == 1)))) + { + if (n) + warn_at (r->location, _("unused value: $%d"), n); + else + warn_at (r->location, _("unset value: $$")); + } + } } @@ -263,12 +322,11 @@ grammar_current_rule_check (void) `-------------------------------------*/ void -grammar_rule_end (location loc) +grammar_current_rule_end (location loc) { /* Put an empty link in the list to mark the end of this rule */ grammar_symbol_append (NULL, grammar_end->location); current_rule->location = loc; - grammar_current_rule_check (); } @@ -301,6 +359,9 @@ grammar_midrule_action (void) midrule->action = current_rule->action; midrule->action_location = dummy_location; current_rule->action = NULL; + /* The action has not been translated yet, so $$ use hasn't been + detected yet. */ + midrule->used = false; if (previous_rule_end) previous_rule_end->next = midrule; @@ -308,14 +369,17 @@ grammar_midrule_action (void) grammar = midrule; /* End the dummy's rule. */ - previous_rule_end = symbol_list_new (NULL, dummy_location); - previous_rule_end->next = current_rule; + midrule->next = symbol_list_new (NULL, dummy_location); + midrule->next->next = current_rule; - midrule->next = previous_rule_end; + previous_rule_end = midrule->next; /* Insert the dummy nonterminal replacing the midrule action into - the current rule. */ + the current rule. Bind it to its dedicated rule. */ grammar_current_rule_symbol_append (dummy, dummy_location); + grammar_end->midrule = midrule; + midrule->midrule_parent_rule = current_rule; + midrule->midrule_parent_rhs_index = symbol_list_length (current_rule->next); } /* Set the precedence symbol of the current rule to PRECSYM. */ @@ -352,8 +416,8 @@ grammar_current_rule_merge_set (uniqstr name, location loc) warn_at (loc, _("%s affects only GLR parsers"), "%merge"); if (current_rule->merger != 0) complain_at (loc, _("only one %s allowed per rule"), "%merge"); - current_rule->merger = - get_merge_function (name, current_rule->sym->type_name, loc); + current_rule->merger = get_merge_function (name); + current_rule->merger_declaration_location = loc; } /* Attach SYM to the current rule. If needed, move the previous @@ -364,18 +428,18 @@ grammar_current_rule_symbol_append (symbol *sym, location loc) { if (current_rule->action) grammar_midrule_action (); - ++nritems; grammar_symbol_append (sym, loc); } -/* Attach an ACTION to the current rule. If needed, move the previous - action as a mid-rule action. */ +/* Attach an ACTION to the current rule. */ void grammar_current_rule_action_append (const char *action, location loc) { if (current_rule->action) grammar_midrule_action (); + /* After all symbol declarations have been parsed, packgram invokes + translate_rule_action. */ current_rule->action = action; current_rule->action_location = loc; } @@ -393,12 +457,19 @@ packgram (void) rule_number ruleno = 0; symbol_list *p = grammar; - ritem = xnmalloc (nritems, sizeof *ritem); + ritem = xnmalloc (nritems + 1, sizeof *ritem); + + /* This sentinel is used by build_relations in gram.c. */ + *ritem++ = 0; + rules = xnmalloc (nrules, sizeof *rules); while (p) { + int rule_length = 0; symbol *ruleprec = p->ruleprec; + record_merge_function_type (p->merger, p->sym->type_name, + p->merger_declaration_location); rules[ruleno].user_number = ruleno; rules[ruleno].number = ruleno; rules[ruleno].lhs = p->sym; @@ -409,21 +480,36 @@ packgram (void) rules[ruleno].precsym = NULL; rules[ruleno].location = p->location; rules[ruleno].useful = true; - rules[ruleno].action = p->action; + rules[ruleno].action = p->action ? translate_rule_action (p) : NULL; rules[ruleno].action_location = p->action_location; - p = p->next; - while (p && p->sym) + /* If this rule contains midrules, rest assured that + grammar_midrule_action inserted the midrules into grammar before this + rule. Thus, the midrule actions have already been scanned in order to + set `used' flags for this rule's rhs, so grammar_rule_check will work + properly. */ + /* Don't check the generated rule 0. It has no action, so some rhs + symbols may appear unused, but the parsing algorithm ensures that + %destructor's are invoked appropriately. */ + if (p != grammar) + grammar_rule_check (p); + + for (p = p->next; p && p->sym; p = p->next) { + ++rule_length; + + /* Don't allow rule_length == INT_MAX, since that might + cause confusion with strtol if INT_MAX == LONG_MAX. */ + if (rule_length == INT_MAX) + fatal_at (rules[ruleno].location, _("rule is too long")); + /* item_number = symbol_number. But the former needs to contain more: negative rule numbers. */ ritem[itemno++] = symbol_number_as_item_number (p->sym->number); /* A rule gets by default the precedence and associativity - of the last token in it. */ + of its last token. */ if (p->sym->class == token_sym && default_prec) rules[ruleno].prec = p->sym; - if (p) - p = p->next; } /* If this rule has a %prec, @@ -433,15 +519,17 @@ packgram (void) rules[ruleno].precsym = ruleprec; rules[ruleno].prec = ruleprec; } + /* An item ends by the rule number (negated). */ ritem[itemno++] = rule_number_as_item_number (ruleno); + assert (itemno < ITEM_NUMBER_MAX); ++ruleno; + assert (ruleno < RULE_NUMBER_MAX); if (p) p = p->next; } - if (itemno != nritems) - abort (); + assert (itemno == nritems); if (trace_flag & trace_sets) ritem_print (stderr); @@ -484,15 +572,25 @@ reader (void) gram__flex_debug = trace_flag & trace_scan; gram_debug = trace_flag & trace_parse; - scanner_initialize (); + gram_scanner_initialize (); gram_parse (); - /* If something went wrong during the parsing, don't try to - continue. */ - if (complaint_issued) - return; + if (! complaint_issued) + check_and_convert_grammar (); + + xfclose (gram_in); +} + - /* Grammar has been read. Do some checking */ +/*-------------------------------------------------------------. +| Check the grammar that has just been read, and convert it to | +| internal form. | +`-------------------------------------------------------------*/ + +static void +check_and_convert_grammar (void) +{ + /* Grammar has been read. Do some checking. */ if (nrules == 0) fatal (_("no rules in the input grammar")); @@ -509,7 +607,7 @@ reader (void) endtoken->user_token_number = 0; } - /* Insert the initial rule, which line is that of the first rule + /* Insert the initial rule, whose line is that of the first rule (not that of the start symbol): accept: %start EOF. */ @@ -525,10 +623,7 @@ reader (void) grammar = p; } - if (! (nsyms <= SYMBOL_NUMBER_MAXIMUM && nsyms == ntokens + nvars)) - abort (); - - xfclose (gram_in); + assert (nsyms <= SYMBOL_NUMBER_MAXIMUM && nsyms == ntokens + nvars); /* Assign the symbols their symbol numbers. Write #defines for the token symbols into FDEFINES if requested. */