X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/39f4191608f2bbfb5a140b9ff96ecd56f5b287bb..6b7022680ea084a5d63e73614e40647f2d0dcce9:/src/reader.c diff --git a/src/reader.c b/src/reader.c index 3c1fe2f2..75a23692 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. @@ -16,46 +17,51 @@ 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 + +#include "complain.h" +#include "conflicts.h" #include "files.h" -#include "symtab.h" -#include "symlist.h" +#include "getargs.h" #include "gram.h" -#include "complain.h" +#include "muscle_tab.h" #include "output.h" #include "reader.h" -#include "conflicts.h" -#include "muscle_tab.h" +#include "symlist.h" +#include "symtab.h" + +static void check_and_convert_grammar (void); -int lineno; -static symbol_list_t *grammar = NULL; -static int start_flag = 0; +static symbol_list *grammar = NULL; +static bool start_flag = false; merger_list *merge_functions; -/* Nonzero if %union has been seen. */ -int typed = 0; +/* Was %union seen? */ +bool typed = false; + +/* Should rules have a default precedence? */ +bool default_prec = true; /*-----------------------. | 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_at (l, _("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; } } @@ -66,54 +72,28 @@ grammar_start_symbol_set (symbol_t *s, location_t l) `----------------------------------------------------------------*/ void -prologue_augment (const char *prologue, location_t location) +prologue_augment (const char *prologue, location loc) { 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"))); - } + obstack_fgrow1 (oout, "]b4_syncline(%d, [[", loc.start.line); + MUSCLE_OBSTACK_SGROW (oout, + quotearg_style (c_quoting_style, loc.start.file)); + obstack_sgrow (oout, "]])[\n"); obstack_sgrow (oout, prologue); } - - - -/*----------------------. -| Handle the epilogue. | -`----------------------*/ - -void -epilogue_set (const char *epilogue, location_t location) -{ - if (!no_lines_flag) - { - obstack_fgrow2 (&muscle_obstack, muscle_find ("linef"), - location.first_line, - quotearg_style (c_quoting_style, - muscle_find ("filename"))); - } - obstack_sgrow (&muscle_obstack, epilogue); - obstack_1grow (&muscle_obstack, 0); - muscle_insert ("epilogue", obstack_finish (&muscle_obstack)); -} - - - /*-------------------------------------------------------------------. +/*-------------------------------------------------------------------. | Return the merger index for a merging function named NAME, whose | | arguments have type TYPE. Records the function, if new, in | -| merger_list. | +| MERGER_LIST. | `-------------------------------------------------------------------*/ static int -get_merge_function (const char* name, const char* type) +get_merge_function (uniqstr name, uniqstr type, location loc) { merger_list *syms; merger_list head; @@ -123,21 +103,23 @@ get_merge_function (const char* name, const char* type) return 0; if (type == NULL) - type = ""; + type = uniqstr_new (""); head.next = merge_functions; - for (syms = &head, n = 1; syms->next != NULL; syms = syms->next, n += 1) - if (strcmp (name, syms->next->name) == 0) + for (syms = &head, n = 1; syms->next; syms = syms->next, n += 1) + if (UNIQSTR_EQ (name, syms->next->name)) break; - if (syms->next == NULL) { - syms->next = XMALLOC (merger_list, 1); - syms->next->name = strdup (name); - syms->next->type = strdup (type); - syms->next->next = NULL; - merge_functions = head.next; - } else if (strcmp (type, syms->next->type) != 0) - warn (_("result type clash on merge function %s: `%s' vs. `%s'"), - name, type, syms->next->type); + if (syms->next == NULL) + { + syms->next = xmalloc (sizeof syms->next[0]); + syms->next->name = uniqstr_new (name); + syms->next->type = uniqstr_new (type); + 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; } @@ -148,11 +130,8 @@ get_merge_function (const char* name, const char* type) void free_merger_functions (void) { - merger_list *L0; - if (! glr_parser) - return; - L0 = merge_functions; - while (L0 != NULL) + merger_list *L0 = merge_functions; + while (L0) { merger_list *L1 = L0->next; free (L0); @@ -162,7 +141,7 @@ free_merger_functions (void) /*-------------------------------------------------------------------. -| Parse the input grammar into a one symbol_list_t structure. Each | +| Parse the input grammar into a one symbol_list structure. Each | | rule is represented by a sequence of symbols: the left hand side | | followed by the contents of the right hand side, followed by a | | null pointer instead of a symbol to terminate the rule. The next | @@ -170,23 +149,16 @@ free_merger_functions (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_t *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_t *p = symbol_list_new (symbol, location); + symbol_list *p = symbol_list_new (sym, loc); if (grammar_end) grammar_end->next = p; @@ -194,13 +166,18 @@ 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_t *current_rule = NULL; -symbol_list_t *previous_rule_end = NULL; +symbol_list *current_rule = NULL; +static symbol_list *previous_rule_end = NULL; /*----------------------------------------------. @@ -208,25 +185,22 @@ symbol_list_t *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; @@ -234,41 +208,74 @@ grammar_rule_begin (symbol_t *lhs, location_t location) ++nvars; } else if (lhs->class == token_sym) - complain_at (location, _("rule given for %s, which is a token"), lhs->tag); + complain_at (loc, _("rule given for %s, which is a token"), lhs->tag); +} + + +/*----------------------------------------------------------------------. +| 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 (s->sym->destructor + || (s->midrule && s->midrule->used)); } -/* Check that the last rule (CURRENT_RULE) is properly defined. For - instance, there should be no type clash on the default action. */ +/*----------------------------------------------------------------. +| 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 $$ is being set in default way, report if any type mismatch. - */ - if (first_rhs) + Don't worry about the default action if $$ is untyped, since $$'s + value can't be used. */ + if (!r->action && r->sym->type_name) { - const char *lhs_type = lhs->type_name ? lhs->type_name : ""; - const char *rhs_type = first_rhs->type_name ? first_rhs->type_name : ""; - if (strcmp (lhs_type, rhs_type)) - complain_at (current_rule->location, - _("type clash (`%s' `%s') on default action"), + symbol *first_rhs = r->next->sym; + /* If $$ 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 $$ but we need one. */ + else + warn_at (r->location, + _("empty rule for typed nonterminal, and no action")); } - /* Warn if there is no default for $$ but we need one. */ - else - { - if (lhs->type_name) - complain_at (current_rule->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, $$ = $1, `uses' both. */ + || (!r->action && (n == 0 || n == 1)))) + { + if (n) + warn_at (r->location, _("unused value: $%d"), n); + else + warn_at (r->location, _("unset value: $$")); + } + } } @@ -277,12 +284,12 @@ 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; + grammar_rule_check (current_rule); } @@ -302,9 +309,9 @@ grammar_midrule_action (void) /* Make a DUMMY nonterminal, whose location is that of the midrule action. Create the MIDRULE. */ - location_t dummy_location = current_rule->action_location; - symbol_t *dummy = dummy_symbol_get (dummy_location); - symbol_list_t *midrule = symbol_list_new (dummy, dummy_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 that the action just read can belong to it. */ @@ -315,6 +322,10 @@ grammar_midrule_action (void) midrule->action = current_rule->action; midrule->action_location = dummy_location; current_rule->action = NULL; + /* If $$ was used in the action, the LHS of the enclosing rule was + incorrectly flagged as used. */ + midrule->used = current_rule->used; + current_rule->used = false; if (previous_rule_end) previous_rule_end->next = midrule; @@ -322,37 +333,39 @@ 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); + grammar_rule_check (midrule); + 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; } /* Set the precedence symbol of the current rule to PRECSYM. */ void -grammar_current_rule_prec_set (symbol_t *precsym, location_t location) +grammar_current_rule_prec_set (symbol *precsym, location loc) { if (current_rule->ruleprec) - complain_at (location, _("two @prec's in a row")); + complain_at (loc, _("only one %s allowed per rule"), "%prec"); current_rule->ruleprec = precsym; } /* Attach dynamic precedence DPREC to the current rule. */ void -grammar_current_rule_dprec_set (int dprec, location_t location) +grammar_current_rule_dprec_set (int dprec, location loc) { if (! glr_parser) - warn_at (location, _("%%dprec affects only GLR parsers")); + warn_at (loc, _("%s affects only GLR parsers"), "%dprec"); if (dprec <= 0) - complain_at (location, _("%%dprec must be followed by positive number")); + complain_at (loc, _("%s must be followed by positive number"), "%dprec"); else if (current_rule->dprec != 0) - complain_at (location, _("only one %%dprec allowed per rule")); + complain_at (loc, _("only one %s allowed per rule"), "%dprec"); current_rule->dprec = dprec; } @@ -360,38 +373,36 @@ grammar_current_rule_dprec_set (int dprec, location_t location) rule. */ void -grammar_current_rule_merge_set (const char* name, location_t location) +grammar_current_rule_merge_set (uniqstr name, location loc) { if (! glr_parser) - warn_at (location, _("%%merge affects only GLR parsers")); + warn_at (loc, _("%s affects only GLR parsers"), "%merge"); if (current_rule->merger != 0) - complain_at (location, _("only one %%merge allowed per rule")); + complain_at (loc, _("only one %s allowed per rule"), "%merge"); current_rule->merger = - get_merge_function (name, current_rule->sym->type_name); + get_merge_function (name, current_rule->sym->type_name, loc); } -/* Attach a SYMBOL to the current rule. If needed, move the previous +/* 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 (); + /* There's no need to invoke grammar_midrule_action here, since the + scanner already did it if necessary. */ current_rule->action = action; - current_rule->action_location = location; + current_rule->action_location = loc; } @@ -403,40 +414,42 @@ grammar_current_rule_action_append (const char *action, location_t location) static void packgram (void) { - unsigned int itemno; - int ruleno; - symbol_list_t *p; + unsigned int itemno = 0; + rule_number ruleno = 0; + symbol_list *p = grammar; + + ritem = xnmalloc (nritems + 1, sizeof *ritem); - ritem = XCALLOC (item_number_t, nritems); - rules = XCALLOC (rule_t, nrules) - 1; + /* This sentinel is used by build_relations in gram.c. */ + *ritem++ = 0; - itemno = 0; - ruleno = 1; + rules = xnmalloc (nrules, sizeof *rules); - p = grammar; while (p) { - symbol_t *ruleprec = p->ruleprec; + symbol *ruleprec = p->ruleprec; 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].useful = true; rules[ruleno].action = p->action; rules[ruleno].action_location = p->action_location; - rules[ruleno].dprec = p->dprec; - rules[ruleno].merger = p->merger; p = p->next; while (p && p->sym) { - /* item_number_t = symbol_number_t. + /* 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) + if (p->sym->class == token_sym && default_prec) rules[ruleno].prec = p->sym; if (p) p = p->next; @@ -449,7 +462,7 @@ packgram (void) rules[ruleno].precsym = ruleprec; rules[ruleno].prec = ruleprec; } - ritem[itemno++] = -ruleno; + ritem[itemno++] = rule_number_as_item_number (ruleno); ++ruleno; if (p) @@ -458,7 +471,7 @@ packgram (void) assert (itemno == nritems); - if (trace_flag) + if (trace_flag & trace_sets) ritem_print (stderr); } @@ -472,16 +485,13 @@ packgram (void) void reader (void) { - gram_control_t gram_control; - lineno = 1; - /* Initialize the symbol table. */ symbols_new (); - /* Construct the axiom symbol. */ - axiom = symbol_get ("$axiom", empty_location); - axiom->class = nterm_sym; - axiom->number = nvars++; + /* Construct the accept symbol. */ + accept = symbol_get ("$accept", empty_location); + accept->class = nterm_sym; + accept->number = nvars++; /* Construct the error token */ errtoken = symbol_get ("error", empty_location); @@ -490,7 +500,7 @@ reader (void) /* Construct a token that represents all undefined literal tokens. It is always token number 2. */ - undeftoken = symbol_get ("$undefined.", empty_location); + undeftoken = symbol_get ("$undefined", empty_location); undeftoken->class = token_sym; undeftoken->number = ntokens++; @@ -498,40 +508,54 @@ 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"); + gram__flex_debug = trace_flag & trace_scan; + gram_debug = trace_flag & trace_parse; scanner_initialize (); - gram_parse (&gram_control); + gram_parse (); - /* Grammar has been read. Do some checking */ + if (! complaint_issued) + check_and_convert_grammar (); + + xfclose (gram_in); +} + + +/*-------------------------------------------------------------. +| 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 = symbol_get ("$", empty_location); - eoftoken->class = token_sym; - eoftoken->number = 0; + endtoken = symbol_get ("$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_t *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; @@ -539,13 +563,7 @@ 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. */ @@ -554,6 +572,6 @@ reader (void) /* Convert the grammar into the format described in gram.h. */ packgram (); - /* The grammar as a symbol_list_t is no longer needed. */ - LIST_FREE (symbol_list_t, grammar); + /* The grammar as a symbol_list is no longer needed. */ + LIST_FREE (symbol_list, grammar); }