X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/8efe435c059af57671105fd8671ba97bb8474264..ec5479ce351b0365549e9c7d570b0ee86a4a6589:/src/reader.c diff --git a/src/reader.c b/src/reader.c index f66a7607..5084618f 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,6 +1,7 @@ -/* Input parser for bison - Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000, 2001, 2002 - Free Software Foundation, Inc. +/* Input parser for Bison + + Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000, 2001, 2002, 2003, + 2005, 2006 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -16,163 +17,166 @@ 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, Inc., 59 Temple Place - Suite 330, - Boston, MA 02111-1307, USA. */ - + the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, + Boston, MA 02110-1301, USA. */ +#include #include "system.h" -#include "quotearg.h" -#include "quote.h" -#include "getargs.h" -#include "files.h" -#include "symtab.h" -#include "options.h" -#include "gram.h" +#include + +#include + #include "complain.h" -#include "output.h" -#include "reader.h" #include "conflicts.h" +#include "files.h" +#include "getargs.h" +#include "gram.h" #include "muscle_tab.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); -int lineno; static symbol_list *grammar = NULL; -static int start_flag = 0; +static bool start_flag = false; +merger_list *merge_functions; -/* Nonzero if %union has been seen. */ -int typed = 0; +/* Was %union seen? */ +bool union_seen = false; -static symbol_list * -symbol_list_new (symbol_t *sym, location_t location) -{ - symbol_list *res = XMALLOC (symbol_list, 1); - res->next = NULL; - res->sym = sym; - res->location = location; - res->action = NULL; - res->ruleprec = NULL; - return res; -} +/* Was a tag seen? */ +bool tag_seen = false; +/* Should rules have a default precedence? */ +bool default_prec = true; -/*--------------------------------------------------------------. -| Get the data type (alternative in the union) of the value for | -| symbol N in rule RULE. | -`--------------------------------------------------------------*/ - -char * -get_type_name (int n, symbol_list *rule) -{ - int i; - symbol_list *rp; - - if (n < 0) - { - complain (_("invalid $ value")); - return NULL; - } - - rp = rule; - i = 0; - - while (i < n) - { - rp = rp->next; - if (rp == NULL || rp->sym == NULL) - { - complain (_("invalid $ value")); - return NULL; - } - ++i; - } - - return rp->sym->type_name; -} - - /*-----------------------. | Set the start symbol. | `-----------------------*/ void -grammar_start_symbol_set (symbol_t *s, location_t l) +grammar_start_symbol_set (symbol *sym, location loc) { if (start_flag) - complain (_("multiple %s declarations"), "%start"); + complain_at (loc, _("multiple %s declarations"), "%start"); else { - start_flag = 1; - startsymbol = s; - startsymbol_location = l; + start_flag = true; + startsymbol = sym; + startsymbol_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_t location) +prologue_augment (const char *prologue, location loc, bool post) { struct obstack *oout = - !typed ? &pre_prologue_obstack : &post_prologue_obstack; - - if (!no_lines_flag) - { - obstack_fgrow2 (oout, muscle_find ("linef"), - location.first_line, - quotearg_style (c_quoting_style, - muscle_find ("filename"))); - } + !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"); obstack_sgrow (oout, prologue); } + +/*------------------------------------------------------------------------. +| Return the merger index for a merging function named NAME. Records the | +| function, if new, in MERGER_LIST. | +`------------------------------------------------------------------------*/ - -/*----------------------. -| Handle the epilogue. | -`----------------------*/ - -void -epilogue_set (const char *epilogue, location_t location) +static int +get_merge_function (uniqstr name) { - if (!no_lines_flag) + merger_list *syms; + merger_list head; + int n; + + if (! glr_parser) + return 0; + + head.next = merge_functions; + for (syms = &head, n = 1; syms->next; syms = syms->next, n += 1) + if (UNIQSTR_EQ (name, syms->next->name)) + break; + if (syms->next == NULL) { - obstack_fgrow2 (&muscle_obstack, muscle_find ("linef"), - location.first_line, - quotearg_style (c_quoting_style, - muscle_find ("filename"))); + syms->next = xmalloc (sizeof syms->next[0]); + syms->next->name = uniqstr_new (name); + /* 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; } - obstack_sgrow (&muscle_obstack, epilogue); - obstack_1grow (&muscle_obstack, 0); - muscle_insert ("epilogue", obstack_finish (&muscle_obstack)); + 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; -/*-------------------------------------------------------------------. -| Generate a dummy symbol, a nonterminal, whose name cannot conflict | -| with the user's names. | -`-------------------------------------------------------------------*/ + if (merger <= 0) + return; -static symbol_t * -gensym (void) -{ - /* Incremented for each generated symbol */ - static int gensym_count = 0; - static char buf[256]; + if (type == NULL) + type = uniqstr_new (""); - symbol_t *sym; + 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. | +`--------------------------------------*/ - sprintf (buf, "@%d", ++gensym_count); - sym = getsym (buf); - sym->class = nterm_sym; - sym->number = nvars++; - return sym; +void +free_merger_functions (void) +{ + merger_list *L0 = merge_functions; + while (L0) + { + merger_list *L1 = L0->next; + free (L0); + L0 = L1; + } } + /*-------------------------------------------------------------------. | Parse the input grammar into a one symbol_list structure. Each | @@ -183,23 +187,16 @@ gensym (void) | | | All actions are copied out, labelled by the rule number they apply | | to. | -| | -| Bison used to allow some %directives in the rules sections, but | -| this is no longer consider appropriate: (i) the documented grammar | -| doesn't claim it, (ii), it would promote bad style, (iii), error | -| recovery for %directives consists in skipping the junk until a `%' | -| is seen and helrp synchronizing. This scheme is definitely wrong | -| in the rules section. | `-------------------------------------------------------------------*/ /* The (currently) last symbol of GRAMMAR. */ -symbol_list *grammar_end = NULL; +static symbol_list *grammar_end = NULL; -/* Append S to the GRAMMAR. */ -void -grammar_symbol_append (symbol_t *symbol, location_t location) +/* Append SYM to the grammar. */ +static void +grammar_symbol_append (symbol *sym, location loc) { - symbol_list *p = symbol_list_new (symbol, location); + symbol_list *p = symbol_list_new (sym, loc); if (grammar_end) grammar_end->next = p; @@ -207,13 +204,18 @@ grammar_symbol_append (symbol_t *symbol, location_t location) 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; -symbol_list *previous_rule_end = NULL; +static symbol_list *current_rule = NULL; +static symbol_list *previous_rule_end = NULL; /*----------------------------------------------. @@ -221,25 +223,22 @@ symbol_list *previous_rule_end = NULL; `----------------------------------------------*/ void -grammar_rule_begin (symbol_t *lhs, location_t location) +grammar_current_rule_begin (symbol *lhs, location loc) { if (!start_flag) { startsymbol = lhs; - startsymbol_location = location; - start_flag = 1; + startsymbol_location = loc; + start_flag = true; } /* Start a new rule and record its lhs. */ ++nrules; - ++nritems; - previous_rule_end = grammar_end; - grammar_symbol_append (lhs, location); + grammar_symbol_append (lhs, loc); current_rule = grammar_end; /* Mark the rule's lhs as a nonterminal if not already so. */ - if (lhs->class == unknown_sym) { lhs->class = nterm_sym; @@ -247,39 +246,74 @@ grammar_rule_begin (symbol_t *lhs, location_t location) ++nvars; } else if (lhs->class == token_sym) - complain (_("rule given for %s, which is a token"), lhs->tag); + complain_at (loc, _("rule given for %s, which is a token"), lhs->tag); } -/* 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 bool +symbol_should_be_used (symbol_list const *s) +{ + return (symbol_destructor_get (s->sym) + || (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_current_rule_check (void) +grammar_rule_check (const symbol_list *r) { - symbol_t *lhs = current_rule->sym; - symbol_t *first_rhs = current_rule->next->sym; + /* Type check. - /* If there is an action, then there is nothing we can do: the user - is allowed to shoot in her foot. */ - if (current_rule->action) - return; + If there is an action, then there is nothing we can do: the user + is allowed to shoot herself in the foot. - /* If $$ is being set in default way, report if any type mismatch. - */ - if (first_rhs) - { - const char *lhs_type = lhs->type_name ? lhs->type_name : ""; - const char *rhs_type = first_rhs->type_name ? first_rhs->type_name : ""; - if (strcmp (lhs_type, rhs_type)) - complain (_("type clash (`%s' `%s') on default action"), - lhs_type, rhs_type); - } - /* Warn if there is no default for $$ but we need one. */ - else + Don't worry about the default action if $$ is untyped, since $$'s + value can't be used. */ + if (!r->action && r->sym->type_name) { - if (lhs->type_name) - complain (_("empty rule for typed nonterminal, and no action")); + symbol *first_rhs = r->next->sym; + /* If $$ is being set in default way, report if any type mismatch. */ + if (first_rhs) + { + char const *lhs_type = r->sym->type_name; + const char *rhs_type = + first_rhs->type_name ? first_rhs->type_name : ""; + if (!UNIQSTR_EQ (lhs_type, rhs_type)) + warn_at (r->location, + _("type clash on default action: <%s> != <%s>"), + lhs_type, rhs_type); + } + /* Warn if there is no default for $$ but we need one. */ + else + warn_at (r->location, + _("empty rule for typed nonterminal, and no action")); } + + /* 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, $$ = $1, `uses' both. */ + || (!r->action && (n == 0 || n == 1)))) + { + if (n) + warn_at (r->location, _("unused value: $%d"), n); + else + warn_at (r->location, _("unset value: $$")); + } + } } @@ -288,12 +322,11 @@ grammar_current_rule_check (void) `-------------------------------------*/ void -grammar_rule_end (location_t location) +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 = location; - grammar_current_rule_check (); + current_rule->location = loc; } @@ -313,8 +346,8 @@ grammar_midrule_action (void) /* Make a DUMMY nonterminal, whose location is that of the midrule action. Create the MIDRULE. */ - symbol_t *dummy = gensym (); - location_t dummy_location = current_rule->action_location; + location dummy_location = current_rule->action_location; + symbol *dummy = dummy_symbol_get (dummy_location); symbol_list *midrule = symbol_list_new (dummy, dummy_location); /* Make a new rule, whose body is empty, before the current one, so @@ -326,6 +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; @@ -333,49 +369,79 @@ 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. */ void -grammar_current_rule_prec_set (symbol_t *precsym) +grammar_current_rule_prec_set (symbol *precsym, location loc) { if (current_rule->ruleprec) - complain (_("two @prec's in a row")); + complain_at (loc, _("only one %s allowed per rule"), "%prec"); current_rule->ruleprec = precsym; } -/* Attach a SYMBOL to the current rule. If needed, move the previous +/* Attach dynamic precedence DPREC to the current rule. */ + +void +grammar_current_rule_dprec_set (int dprec, location loc) +{ + if (! glr_parser) + warn_at (loc, _("%s affects only GLR parsers"), "%dprec"); + if (dprec <= 0) + complain_at (loc, _("%s must be followed by positive number"), "%dprec"); + else if (current_rule->dprec != 0) + complain_at (loc, _("only one %s allowed per rule"), "%dprec"); + current_rule->dprec = dprec; +} + +/* Attach a merge function NAME with argument type TYPE to current + rule. */ + +void +grammar_current_rule_merge_set (uniqstr name, location loc) +{ + if (! glr_parser) + 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->merger_declaration_location = loc; +} + +/* Attach SYM to the current rule. If needed, move the previous action as a mid-rule action. */ void -grammar_current_rule_symbol_append (symbol_t *symbol, location_t location) +grammar_current_rule_symbol_append (symbol *sym, location loc) { if (current_rule->action) grammar_midrule_action (); - ++nritems; - grammar_symbol_append (symbol, location); + 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_t location) +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 = location; + current_rule->action_location = loc; } @@ -387,41 +453,63 @@ grammar_current_rule_action_append (const char *action, location_t location) static void packgram (void) { - unsigned int itemno; - int ruleno; - symbol_list *p; + unsigned int itemno = 0; + rule_number ruleno = 0; + symbol_list *p = grammar; - ritem = XCALLOC (item_number_t, nritems); - rules = XCALLOC (rule_t, nrules) - 1; + ritem = xnmalloc (nritems + 1, sizeof *ritem); - itemno = 0; - ruleno = 1; + /* This sentinel is used by build_relations in gram.c. */ + *ritem++ = 0; + + rules = xnmalloc (nrules, sizeof *rules); - p = grammar; while (p) { - symbol_t *ruleprec = p->ruleprec; + 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; rules[ruleno].rhs = ritem + itemno; + rules[ruleno].prec = NULL; + rules[ruleno].dprec = p->dprec; + rules[ruleno].merger = p->merger; + rules[ruleno].precsym = NULL; rules[ruleno].location = p->location; - rules[ruleno].useful = TRUE; - rules[ruleno].action = p->action; + rules[ruleno].useful = true; + 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) { - /* item_number_t = symbol_number_t. + ++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. */ - if (p->sym->class == token_sym) + 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, @@ -431,8 +519,11 @@ packgram (void) rules[ruleno].precsym = ruleprec; rules[ruleno].prec = ruleprec; } - ritem[itemno++] = -ruleno; + /* 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; @@ -440,7 +531,7 @@ packgram (void) assert (itemno == nritems); - if (trace_flag) + if (trace_flag & trace_sets) ritem_print (stderr); } @@ -454,25 +545,22 @@ packgram (void) void reader (void) { - gram_control_t gram_control; - lineno = 1; - /* Initialize the symbol table. */ symbols_new (); - /* Construct the axiom symbol. */ - axiom = getsym ("$axiom"); - axiom->class = nterm_sym; - axiom->number = nvars++; + /* Construct the accept symbol. */ + accept = symbol_get ("$accept", empty_location); + accept->class = nterm_sym; + accept->number = nvars++; /* Construct the error token */ - errtoken = getsym ("error"); + errtoken = symbol_get ("error", empty_location); errtoken->class = token_sym; errtoken->number = ntokens++; /* Construct a token that represents all undefined literal tokens. It is always token number 2. */ - undeftoken = getsym ("$undefined."); + undeftoken = symbol_get ("$undefined", empty_location); undeftoken->class = token_sym; undeftoken->number = ntokens++; @@ -480,40 +568,54 @@ reader (void) obstack_init (&pre_prologue_obstack); obstack_init (&post_prologue_obstack); - finput = xfopen (infile, "r"); - gram_in = finput; + gram_in = xfopen (grammar_file, "r"); - gram_debug = !!getenv ("parse"); - gram__flex_debug = !!getenv ("scan"); - scanner_initialize (); - gram_parse (&gram_control); + gram__flex_debug = trace_flag & trace_scan; + gram_debug = trace_flag & trace_parse; + gram_scanner_initialize (); + gram_parse (); + + 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")); /* Report any undefined symbols and consider them nonterminals. */ symbols_check_defined (); - /* If the user did not define her EOFTOKEN, do it now. */ - if (!eoftoken) + /* If the user did not define her ENDTOKEN, do it now. */ + if (!endtoken) { - eoftoken = getsym ("$"); - eoftoken->class = token_sym; - eoftoken->number = 0; + endtoken = symbol_get ("$end", empty_location); + endtoken->class = token_sym; + endtoken->number = 0; /* Value specified by POSIX. */ - eoftoken->user_token_number = 0; + 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): - axiom: %start EOF. */ + accept: %start EOF. */ { - symbol_list *p = symbol_list_new (axiom, empty_location); + symbol_list *p = symbol_list_new (accept, empty_location); p->location = grammar->location; p->next = symbol_list_new (startsymbol, empty_location); - p->next->next = symbol_list_new (eoftoken, empty_location); + p->next->next = symbol_list_new (endtoken, empty_location); p->next->next->next = symbol_list_new (NULL, empty_location); p->next->next->next->next = grammar; nrules += 1; @@ -521,13 +623,7 @@ reader (void) grammar = p; } - if (nsyms > SHRT_MAX) - fatal (_("too many symbols (tokens plus nonterminals); maximum %d"), - SHRT_MAX); - - assert (nsyms == ntokens + nvars); - - xfclose (finput); + assert (nsyms <= SYMBOL_NUMBER_MAXIMUM && nsyms == ntokens + nvars); /* Assign the symbols their symbol numbers. Write #defines for the token symbols into FDEFINES if requested. */