X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/8efe435c059af57671105fd8671ba97bb8474264..41cce2f604996c7713d9025f8a719ea56d83783e:/src/reader.c diff --git a/src/reader.c b/src/reader.c index f66a7607..928a5408 100644 --- a/src/reader.c +++ b/src/reader.c @@ -26,7 +26,7 @@ #include "getargs.h" #include "files.h" #include "symtab.h" -#include "options.h" +#include "symlist.h" #include "gram.h" #include "complain.h" #include "output.h" @@ -34,61 +34,13 @@ #include "conflicts.h" #include "muscle_tab.h" -int lineno; -static symbol_list *grammar = NULL; +static symbol_list_t *grammar = NULL; static int start_flag = 0; +merger_list *merge_functions; /* Nonzero if %union has been seen. */ int typed = 0; - -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; -} - -/*--------------------------------------------------------------. -| 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. | `-----------------------*/ @@ -97,7 +49,7 @@ void grammar_start_symbol_set (symbol_t *s, location_t l) { if (start_flag) - complain (_("multiple %s declarations"), "%start"); + complain_at (l, _("multiple %s declarations"), "%start"); else { start_flag = 1; @@ -153,29 +105,66 @@ epilogue_set (const char *epilogue, location_t location) -/*-------------------------------------------------------------------. -| Generate a dummy symbol, a nonterminal, whose name cannot conflict | -| with the user's names. | + /*-------------------------------------------------------------------. +| Return the merger index for a merging function named NAME, whose | +| arguments have type TYPE. Records the function, if new, in | +| merger_list. | `-------------------------------------------------------------------*/ -static symbol_t * -gensym (void) +static int +get_merge_function (const char* name, const char* type, + location_t loc) { - /* Incremented for each generated symbol */ - static int gensym_count = 0; - static char buf[256]; + merger_list *syms; + merger_list head; + int n; + + if (! glr_parser) + return 0; - symbol_t *sym; + if (type == NULL) + type = ""; - sprintf (buf, "@%d", ++gensym_count); - sym = getsym (buf); - sym->class = nterm_sym; - sym->number = nvars++; - return sym; + head.next = merge_functions; + for (syms = &head, n = 1; syms->next != NULL; syms = syms->next, n += 1) + if (strcmp (name, syms->next->name) == 0) + break; + if (syms->next == NULL) + { + syms->next = XMALLOC (merger_list, 1); + syms->next->name = xstrdup (name); + syms->next->type = xstrdup (type); + syms->next->next = NULL; + merge_functions = head.next; + } + else if (strcmp (type, syms->next->type) != 0) + warn_at (loc, _("result type clash on merge function %s: `%s' vs. `%s'"), + name, type, syms->next->type); + return n; } + +/*--------------------------------------. +| Free all merge-function definitions. | +`--------------------------------------*/ + +void +free_merger_functions (void) +{ + merger_list *L0; + if (! glr_parser) + return; + L0 = merge_functions; + while (L0 != NULL) + { + merger_list *L1 = L0->next; + free (L0); + L0 = L1; + } +} + /*-------------------------------------------------------------------. -| Parse the input grammar into a one symbol_list structure. Each | +| Parse the input grammar into a one symbol_list_t 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 | @@ -193,13 +182,13 @@ gensym (void) `-------------------------------------------------------------------*/ /* The (currently) last symbol of GRAMMAR. */ -symbol_list *grammar_end = NULL; +symbol_list_t *grammar_end = NULL; /* Append S to the GRAMMAR. */ void grammar_symbol_append (symbol_t *symbol, location_t location) { - symbol_list *p = symbol_list_new (symbol, location); + symbol_list_t *p = symbol_list_new (symbol, location); if (grammar_end) grammar_end->next = p; @@ -212,8 +201,8 @@ grammar_symbol_append (symbol_t *symbol, location_t location) /* 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; +symbol_list_t *current_rule = NULL; +symbol_list_t *previous_rule_end = NULL; /*----------------------------------------------. @@ -247,7 +236,7 @@ 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 (location, _("rule given for %s, which is a token"), lhs->tag); } /* Check that the last rule (CURRENT_RULE) is properly defined. For @@ -257,29 +246,32 @@ static void grammar_current_rule_check (void) { symbol_t *lhs = current_rule->sym; + char const *lhs_type = lhs->type_name; symbol_t *first_rhs = current_rule->next->sym; /* If there is an action, then there is nothing we can do: the user - is allowed to shoot in her foot. */ + is allowed to shoot herself in the foot. */ if (current_rule->action) return; - /* If $$ is being set in default way, report if any type mismatch. - */ + /* Don't worry about the default action if $$ is untyped, since $$'s + value can't be used. */ + if (! lhs_type) + return; + + /* If $$ 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); + complain_at (current_rule->location, + _("type clash (`%s' `%s') on default action"), + lhs_type, rhs_type); } /* Warn if there is no default for $$ but we need one. */ else - { - if (lhs->type_name) - complain (_("empty rule for typed nonterminal, and no action")); - } + complain_at (current_rule->location, + _("empty rule for typed nonterminal, and no action")); } @@ -313,9 +305,9 @@ 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; - symbol_list *midrule = symbol_list_new (dummy, dummy_location); + symbol_t *dummy = dummy_symbol_get (dummy_location); + symbol_list_t *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. */ @@ -346,13 +338,42 @@ grammar_midrule_action (void) /* 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_t *precsym, location_t location) { if (current_rule->ruleprec) - complain (_("two @prec's in a row")); + complain_at (location, _("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) +{ + if (! glr_parser) + warn_at (location, _("%s affects only GLR parsers"), "%dprec"); + if (dprec <= 0) + complain_at (location, + _("%s must be followed by positive number"), "%dprec"); + else if (current_rule->dprec != 0) + complain_at (location, _("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 (const char* name, location_t location) +{ + if (! glr_parser) + warn_at (location, _("%s affects only GLR parsers"), "%merge"); + if (current_rule->merger != 0) + complain_at (location, _("only one %s allowed per rule"), "%merge"); + current_rule->merger = + get_merge_function (name, current_rule->sym->type_name, location); +} + /* Attach a SYMBOL to the current rule. If needed, move the previous action as a mid-rule action. */ @@ -365,7 +386,6 @@ grammar_current_rule_symbol_append (symbol_t *symbol, location_t location) grammar_symbol_append (symbol, location); } - /* Attach an ACTION to the current rule. If needed, move the previous action as a mid-rule action. */ @@ -387,17 +407,13 @@ 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_t ruleno = 0; + symbol_list_t *p = grammar; ritem = XCALLOC (item_number_t, nritems); - rules = XCALLOC (rule_t, nrules) - 1; - - itemno = 0; - ruleno = 1; + rules = XCALLOC (rule_t, nrules); - p = grammar; while (p) { symbol_t *ruleprec = p->ruleprec; @@ -406,9 +422,11 @@ packgram (void) rules[ruleno].lhs = p->sym; rules[ruleno].rhs = ritem + itemno; 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) @@ -431,7 +449,7 @@ packgram (void) rules[ruleno].precsym = ruleprec; rules[ruleno].prec = ruleprec; } - ritem[itemno++] = -ruleno; + ritem[itemno++] = rule_number_as_item_number (ruleno); ++ruleno; if (p) @@ -440,7 +458,7 @@ packgram (void) assert (itemno == nritems); - if (trace_flag) + if (trace_flag & trace_sets) ritem_print (stderr); } @@ -455,24 +473,23 @@ 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++; @@ -483,11 +500,16 @@ reader (void) finput = xfopen (infile, "r"); gram_in = finput; - 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); + /* If something went wrong during the parsing, don't try to + continue. */ + if (complaint_issued) + exit (EXIT_FAILURE); + /* Grammar has been read. Do some checking */ if (nrules == 0) fatal (_("no rules in the input grammar")); @@ -495,25 +517,25 @@ reader (void) /* 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 (not that of the start symbol): - axiom: %start EOF. */ + accept: %start EOF. */ { - symbol_list *p = symbol_list_new (axiom, empty_location); + symbol_list_t *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; @@ -521,9 +543,9 @@ reader (void) grammar = p; } - if (nsyms > SHRT_MAX) + if (SYMBOL_NUMBER_MAX < nsyms) fatal (_("too many symbols (tokens plus nonterminals); maximum %d"), - SHRT_MAX); + SYMBOL_NUMBER_MAX); assert (nsyms == ntokens + nvars); @@ -536,6 +558,6 @@ reader (void) /* Convert the grammar into the format described in gram.h. */ packgram (); - /* The grammar as a symbol_list is no longer needed. */ - LIST_FREE (symbol_list, grammar); + /* The grammar as a symbol_list_t is no longer needed. */ + LIST_FREE (symbol_list_t, grammar); }