X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/72a23c9797290f1f2c948cf3027d80fa5c9370bc..ec3bc39:/src/reader.c diff --git a/src/reader.c b/src/reader.c index a37f2d4a..93df69a6 100644 --- a/src/reader.c +++ b/src/reader.c @@ -38,23 +38,20 @@ typedef struct symbol_list { struct symbol_list *next; - bucket *sym; + symbol_t *sym; int line; /* The action is attached to the LHS of a rule. */ const char *action; int action_line; - /* The guard is attached to the LHS of a rule. */ - const char *guard; - int guard_line; - bucket *ruleprec; + symbol_t *ruleprec; } symbol_list; int lineno; static symbol_list *grammar = NULL; static int start_flag = 0; -static bucket *startval = NULL; +static symbol_t *startval = NULL; /* Nonzero if components of semantic values are used, implying they must be unions. */ @@ -66,13 +63,13 @@ static int typed = 0; /* Incremented for each %left, %right or %nonassoc seen */ static int lastprec = 0; -bucket *errtoken = NULL; -bucket *undeftoken = NULL; -bucket *eoftoken = NULL; -bucket *axiom = NULL; +symbol_t *errtoken = NULL; +symbol_t *undeftoken = NULL; +symbol_t *eoftoken = NULL; +symbol_t *axiom = NULL; static symbol_list * -symbol_list_new (bucket *sym) +symbol_list_new (symbol_t *sym) { symbol_list *res = XMALLOC (symbol_list, 1); res->next = NULL; @@ -80,14 +77,12 @@ symbol_list_new (bucket *sym) res->line = lineno; res->action = NULL; res->action_line = 0; - res->guard = NULL; - res->guard_line = 0; res->ruleprec = NULL; return res; } /*------------------------. -| Operations on buckets. | +| Operations on symbols. | `------------------------*/ @@ -97,7 +92,7 @@ symbol_list_new (bucket *sym) `-----------------------------------------------------------*/ static bool -bucket_check_defined (bucket *this) +symbol_check_defined (symbol_t *this) { if (this->class == unknown_sym) { @@ -118,7 +113,7 @@ bucket_check_defined (bucket *this) `-------------------------------------------------------------------*/ static bool -bucket_make_alias (bucket *symbol, char *typename) +symbol_make_alias (symbol_t *symbol, char *typename) { if (symval->alias) warn (_("symbol `%s' used more than once as a literal string"), @@ -131,7 +126,7 @@ bucket_make_alias (bucket *symbol, char *typename) symval->class = token_sym; symval->type_name = typename; symval->user_token_number = symbol->user_token_number; - symbol->user_token_number = SALIAS; + symbol->user_token_number = USER_NUMBER_ALIAS; symval->alias = symbol; symbol->alias = symval; /* symbol and symval combined are only one symbol */ @@ -151,10 +146,10 @@ bucket_make_alias (bucket *symbol, char *typename) `---------------------------------------------------------*/ static bool -bucket_check_alias_consistence (bucket *this) +symbol_check_alias_consistence (symbol_t *this) { /* Check only those who _are_ the aliases. */ - if (this->alias && this->user_token_number == SALIAS) + if (this->alias && this->user_token_number == USER_NUMBER_ALIAS) { if (this->prec != this->alias->prec) { @@ -188,13 +183,8 @@ bucket_check_alias_consistence (bucket *this) `-------------------------------------------------------------------*/ static bool -bucket_pack (bucket *this) +symbol_pack (symbol_t *this) { - if (getenv ("DEBUG")) - fprintf (stderr, "Packing %s, %s, number = %d\n", - this->tag, - this->class == nterm_sym ? "nterm" : "TERM", - this->number); if (this->class == nterm_sym) { this->number += ntokens; @@ -204,27 +194,25 @@ bucket_pack (bucket *this) /* This symbol and its alias are a single token defn. Allocate a tokno, and assign to both check agreement of prec and assoc fields and make both the same */ - if (this->number == -1) + if (this->number == NUMBER_UNDEFINED) { if (this == eoftoken || this->alias == eoftoken) this->number = this->alias->number = 0; else { - assert (this->alias->number != -1); + assert (this->alias->number != NUMBER_UNDEFINED); this->number = this->alias->number; } } - /* Do not do processing below for SALIASs. */ - if (this->user_token_number == SALIAS) + /* Do not do processing below for USER_NUMBER_ALIASs. */ + if (this->user_token_number == USER_NUMBER_ALIAS) return TRUE; } else /* this->class == token_sym */ { - assert (this->number != -1); + assert (this->number != NUMBER_UNDEFINED); } - if (getenv ("DEBUG")) - fprintf (stderr, "Setting %d = %s\n", this->number, this->tag); symbols[this->number] = this; return TRUE; } @@ -237,25 +225,18 @@ bucket_pack (bucket *this) `--------------------------------------------------*/ static bool -bucket_translation (bucket *this) +symbol_translation (symbol_t *this) { - if (getenv ("DEBUG")) - fprintf (stderr, "Considering Setting UserVal %s = %d (val = %d)\n", - this->tag, this->user_token_number, this->number); - /* Non-terminal? */ if (this->class == token_sym - && this->user_token_number != SALIAS) + && this->user_token_number != USER_NUMBER_ALIAS) { /* A token which translation has already been set? */ - if (token_translations[this->user_token_number] != 2) + if (token_translations[this->user_token_number] != undeftoken->number) complain (_("tokens %s and %s both assigned number %d"), symbols[token_translations[this->user_token_number]]->tag, this->tag, this->user_token_number); - if (getenv ("DEBUG")) - fprintf (stderr, "Setting UserVal %s = %d (val = %d)\n", - this->tag, this->user_token_number, this->number); token_translations[this->user_token_number] = this->number; } @@ -340,12 +321,34 @@ get_type_name (int n, symbol_list *rule) complain (_("invalid $ value")); return NULL; } - i++; + ++i; } return rp->sym->type_name; } +/*------------------------------------------------------------------. +| Copy the character C to OOUT, and insert quadigraphs when needed. | +`------------------------------------------------------------------*/ + +static inline void +copy_character (struct obstack *oout, int c) +{ + switch (c) + { + case '[': + obstack_sgrow (oout, "@<:@"); + break; + + case ']': + obstack_sgrow (oout, "@:>@"); + break; + + default: + obstack_1grow (oout, c); + } +} + /*------------------------------------------------------------. | Dump the string from FIN to OOUT if non null. MATCH is the | | delimiter of the string (either ' or "). | @@ -373,17 +376,17 @@ copy_string2 (FILE *fin, struct obstack *oout, int match, int store) continue; } - obstack_1grow (oout, c); + copy_character (oout, c); if (c == '\\') { c = getc (fin); if (c == EOF) fatal (_("unterminated string at end of file")); - obstack_1grow (oout, c); + copy_character (oout, c); if (c == '\n') - lineno++; + ++lineno; } c = getc (fin); @@ -466,7 +469,7 @@ copy_comment (FILE *fin, struct obstack *oout) } else if (c == '\n') { - lineno++; + ++lineno; obstack_1grow (oout, c); if (cplus_comment) ended = 1; @@ -477,30 +480,29 @@ copy_comment (FILE *fin, struct obstack *oout) fatal (_("unterminated comment")); else { - obstack_1grow (oout, c); + copy_character (oout, c); c = getc (fin); } } } -/*-----------------------------------------------------------------. -| FIN is pointing to a location (i.e., a `@'). Output to OOUT a | -| reference to this location. STACK_OFFSET is the number of values | -| in the current rule so far, which says where to find `$0' with | -| respect to the top of the stack. | -`-----------------------------------------------------------------*/ +/*-------------------------------------------------------------------. +| FIN is pointing to a location (i.e., a `@'). Output to OOUT a | +| reference to this location. RULE_LENGTH is the number of values in | +| the current rule so far, which says where to find `$0' with | +| respect to the top of the stack. | +`-------------------------------------------------------------------*/ static inline void -copy_at (FILE *fin, struct obstack *oout, int stack_offset) +copy_at (FILE *fin, struct obstack *oout, int rule_length) { - int c; + int c = getc (fin); + locations_flag = 1; - c = getc (fin); if (c == '$') { - obstack_sgrow (oout, "yyloc"); - locations_flag = 1; + obstack_sgrow (oout, "]b4_lhs_location["); } else if (isdigit (c) || c == '-') { @@ -508,16 +510,11 @@ copy_at (FILE *fin, struct obstack *oout, int stack_offset) ungetc (c, fin); n = read_signed_integer (fin); - if (n > stack_offset) + if (n > rule_length) complain (_("invalid value: %s%d"), "@", n); else - { - /* Offset is always 0 if parser has already popped the stack - pointer. */ - obstack_fgrow1 (oout, "yylsp[%d]", - n - (semantic_parser ? 0 : stack_offset)); - locations_flag = 1; - } + obstack_fgrow2 (oout, "]b4_rhs_location([%d], [%d])[", + rule_length, n); } else { @@ -528,19 +525,19 @@ copy_at (FILE *fin, struct obstack *oout, int stack_offset) } -/*-------------------------------------------------------------------. -| FIN is pointing to a wannabee semantic value (i.e., a `$'). | -| | -| Possible inputs: $[]($|integer) | -| | -| Output to OOUT a reference to this semantic value. STACK_OFFSET is | -| the number of values in the current rule so far, which says where | -| to find `$0' with respect to the top of the stack. | -`-------------------------------------------------------------------*/ +/*------------------------------------------------------------------. +| FIN is pointing to a wannabee semantic value (i.e., a `$'). | +| | +| Possible inputs: $[]($|integer) | +| | +| Output to OOUT a reference to this semantic value. RULE_LENGTH is | +| the number of values in the current rule so far, which says where | +| to find `$0' with respect to the top of the stack. | +`------------------------------------------------------------------*/ static inline void copy_dollar (FILE *fin, struct obstack *oout, - symbol_list *rule, int stack_offset) + symbol_list *rule, int rule_length) { int c = getc (fin); const char *type_name = NULL; @@ -556,15 +553,15 @@ copy_dollar (FILE *fin, struct obstack *oout, if (c == '$') { - obstack_sgrow (oout, "yyval"); - if (!type_name) type_name = get_type_name (0, rule); - if (type_name) - obstack_fgrow1 (oout, ".%s", type_name); if (!type_name && typed) complain (_("$$ of `%s' has no declared type"), rule->sym->tag); + if (!type_name) + type_name = ""; + obstack_fgrow1 (oout, + "]b4_lhs_value([%s])[", type_name); } else if (isdigit (c) || c == '-') { @@ -572,23 +569,19 @@ copy_dollar (FILE *fin, struct obstack *oout, ungetc (c, fin); n = read_signed_integer (fin); - if (n > stack_offset) + if (n > rule_length) complain (_("invalid value: %s%d"), "$", n); else { if (!type_name && n > 0) type_name = get_type_name (n, rule); - - /* Offset is always 0 if parser has already popped the stack - pointer. */ - obstack_fgrow1 (oout, "yyvsp[%d]", - n - (semantic_parser ? 0 : stack_offset)); - - if (type_name) - obstack_fgrow1 (oout, ".%s", type_name); if (!type_name && typed) complain (_("$%d of `%s' has no declared type"), n, rule->sym->tag); + if (!type_name) + type_name = ""; + obstack_fgrow3 (oout, "]b4_rhs_value([%d], [%d], [%s])[", + rule_length, n, type_name); } } else @@ -605,7 +598,7 @@ copy_dollar (FILE *fin, struct obstack *oout, `-------------------------------------------------------------------*/ static void -copy_definition (void) +copy_definition (struct obstack *oout) { int c; /* -1 while reading a character if prev char was %. */ @@ -613,7 +606,7 @@ copy_definition (void) if (!no_lines_flag) { - obstack_fgrow2 (&attrs_obstack, muscle_find ("linef"), + obstack_fgrow2 (oout, muscle_find ("linef"), lineno, quotearg_style (c_quoting_style, muscle_find ("filename"))); } @@ -627,8 +620,8 @@ copy_definition (void) switch (c) { case '\n': - obstack_1grow (&attrs_obstack, c); - lineno++; + obstack_1grow (oout, c); + ++lineno; break; case '%': @@ -637,18 +630,18 @@ copy_definition (void) case '\'': case '"': - copy_string (finput, &attrs_obstack, c); + copy_string (finput, oout, c); break; case '/': - copy_comment (finput, &attrs_obstack); + copy_comment (finput, oout); break; case EOF: fatal ("%s", _("unterminated `%{' definition")); default: - obstack_1grow (&attrs_obstack, c); + copy_character (oout, c); } c = getc (finput); @@ -657,7 +650,7 @@ copy_definition (void) { if (c == '}') return; - obstack_1grow (&attrs_obstack, '%'); + obstack_1grow (oout, '%'); } after_percent = 0; } @@ -677,7 +670,7 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) char *typename = NULL; /* The symbol being defined. */ - struct bucket *symbol = NULL; + symbol_t *symbol = NULL; /* After `%token' and `%nterm', any number of symbols maybe be defined. */ @@ -706,7 +699,7 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) } else if (token == tok_identifier && *symval->tag == '\"' && symbol) { - bucket_make_alias (symbol, typename); + symbol_make_alias (symbol, typename); symbol = NULL; } else if (token == tok_identifier) @@ -719,13 +712,8 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) symbol->class = what_is; if (what_is == nterm_sym && oldclass != nterm_sym) symbol->number = nvars++; - if (what_is == token_sym && symbol->number == -1) - { - symbol->number = ntokens++; - if (getenv ("DEBUG")) - fprintf (stderr, "Set %s to %d\n", - symbol->tag, symbol->number); - } + if (what_is == token_sym && symbol->number == NUMBER_UNDEFINED) + symbol->number = ntokens++; if (typename) { @@ -845,7 +833,8 @@ parse_assoc_decl (associativity assoc) char *name = NULL; int prev = 0; - lastprec++; /* Assign a new precedence level, never 0. */ + /* Assign a new precedence level, never 0. */ + ++lastprec; for (;;) { @@ -875,7 +864,7 @@ parse_assoc_decl (associativity assoc) symval->assoc = assoc; if (symval->class == nterm_sym) complain (_("symbol %s redefined"), symval->tag); - if (symval->number == -1) + if (symval->number == NUMBER_UNDEFINED) { symval->number = ntokens++; symval->class = token_sym; @@ -950,7 +939,7 @@ parse_union_decl (void) switch (c) { case '\n': - lineno++; + ++lineno; break; case '/': @@ -958,7 +947,7 @@ parse_union_decl (void) break; case '{': - count++; + ++count; break; case '}': @@ -1009,10 +998,10 @@ parse_expect_decl (void) | | | Two symbols are entered in the table, one for the token symbol and | | one for the literal. Both are given the , if any, from the | -| declaration. The ->user_token_number of the first is SALIAS and | -| the ->user_token_number of the second is set to the number, if | -| any, from the declaration. The two symbols are linked via | -| pointers in their ->alias fields. | +| declaration. The ->user_token_number of the first is | +| USER_NUMBER_ALIAS and the ->user_token_number of the second is set | +| to the number, if any, from the declaration. The two symbols are | +| linked via pointers in their ->alias fields. | | | | During OUTPUT_DEFINES_TABLE, the symbol is reported thereafter, | | only the literal string is retained it is the literal string that | @@ -1023,9 +1012,9 @@ static void parse_thong_decl (void) { token_t token; - struct bucket *symbol; + symbol_t *symbol; char *typename = 0; - int usrtoknum = SUNDEF; + int usrtoknum = USER_NUMBER_UNDEFINED; token = lex (); /* fetch typename or first token */ if (token == tok_typename) @@ -1046,7 +1035,7 @@ parse_thong_decl (void) } symval->class = token_sym; symval->type_name = typename; - symval->user_token_number = SALIAS; + symval->user_token_number = USER_NUMBER_ALIAS; symbol = symval; token = lex (); /* get number or literal string */ @@ -1172,7 +1161,7 @@ parse_skel_decl (void) /*----------------------------------------------------------------. | Read from finput until `%%' is seen. Discard the `%%'. Handle | | any `%' declarations, and copy the contents of any `%{ ... %}' | -| groups to ATTRS_OBSTACK. | +| groups to PRE_PROLOGUE_OBSTACK or POST_PROLOGUE_OBSTACK. | `----------------------------------------------------------------*/ static void @@ -1192,7 +1181,10 @@ read_declarations (void) return; case tok_percent_left_curly: - copy_definition (); + if (!typed) + copy_definition (&pre_prologue_obstack); + else + copy_definition (&post_prologue_obstack); break; case tok_token: @@ -1270,37 +1262,37 @@ read_declarations (void) } } -/*-------------------------------------------------------------------. -| Assuming that a `{' has just been seen, copy everything up to the | -| matching `}' into the actions file. STACK_OFFSET is the number of | -| values in the current rule so far, which says where to find `$0' | -| with respect to the top of the stack. | -| | -| This routine is used both for actions and guards. Only | -| ACTION_OBSTACK is used, but this is fine, since we use only | -| pointers to relevant portions inside this obstack. | -`-------------------------------------------------------------------*/ +/*------------------------------------------------------------------. +| Assuming that a `{' has just been seen, copy everything up to the | +| matching `}' into ACTION_OBSTACK. | +| | +| RULE_LENGTH is the number of values in the current rule so far, | +| which says where to find `$0' with respect to the top of the | +| stack. It is not the same as the rule->length in the case of mid | +| rule actions. | +| | +| This routine is used for actions. | +`------------------------------------------------------------------*/ static void -parse_braces (symbol_list *rule, int stack_offset) +parse_action (symbol_list *rule, int rule_length) { - int c; - int count; - - count = 1; + int count = 1; + rule->action_line = lineno; while (count > 0) { + int c; while ((c = getc (finput)) != '}') switch (c) { case '\n': - obstack_1grow (&action_obstack, c); - lineno++; + copy_character (&action_obstack, c); + ++lineno; break; case '{': - obstack_1grow (&action_obstack, c); - count++; + copy_character (&action_obstack, c); + ++count; break; case '\'': @@ -1313,51 +1305,29 @@ parse_braces (symbol_list *rule, int stack_offset) break; case '$': - copy_dollar (finput, &action_obstack, - rule, stack_offset); + copy_dollar (finput, &action_obstack, rule, rule_length); break; case '@': - copy_at (finput, &action_obstack, - stack_offset); + copy_at (finput, &action_obstack, rule_length); break; case EOF: fatal (_("unmatched %s"), "`{'"); default: - obstack_1grow (&action_obstack, c); + copy_character (&action_obstack, c); } /* Above loop exits when C is '}'. */ if (--count) - obstack_1grow (&action_obstack, c); + copy_character (&action_obstack, c); } obstack_1grow (&action_obstack, '\0'); -} - - -static void -parse_action (symbol_list *rule, int stack_offset) -{ - rule->action_line = lineno; - parse_braces (rule, stack_offset); rule->action = obstack_finish (&action_obstack); } - -static void -parse_guard (symbol_list *rule, int stack_offset) -{ - token_t t = lex (); - if (t != tok_left_curly) - complain (_("invalid %s declaration"), "%guard"); - rule->guard_line = lineno; - parse_braces (rule, stack_offset); - rule->guard = obstack_finish (&action_obstack); -} - /*-------------------------------------------------------------------. @@ -1365,14 +1335,14 @@ parse_guard (symbol_list *rule, int stack_offset) | with the user's names. | `-------------------------------------------------------------------*/ -static bucket * +static symbol_t * gensym (void) { /* Incremented for each generated symbol */ static int gensym_count = 0; static char buf[256]; - bucket *sym; + symbol_t *sym; sprintf (buf, "@%d", ++gensym_count); token_buffer = buf; @@ -1389,8 +1359,8 @@ gensym (void) | null pointer instead of a symbol to terminate the rule. The next | | symbol is the lhs of the following rule. | | | -| All guards and actions are copied out to the appropriate files, | -| labelled by the rule number they apply to. | +| 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 | @@ -1404,7 +1374,7 @@ static void readgram (void) { token_t t; - bucket *lhs = NULL; + symbol_t *lhs = NULL; symbol_list *p = NULL; symbol_list *p1 = NULL; @@ -1423,7 +1393,7 @@ readgram (void) /* Number of symbols in rhs of this rule so far */ int rulelength = 0; int xactions = 0; /* JF for error checking */ - bucket *first_rhs = 0; + symbol_t *first_rhs = 0; if (t == tok_identifier) { @@ -1450,8 +1420,8 @@ readgram (void) } /* start a new rule and record its lhs. */ - nrules++; - nitems++; + ++nrules; + ++nritems; p = symbol_list_new (lhs); @@ -1470,7 +1440,7 @@ readgram (void) { lhs->class = nterm_sym; lhs->number = nvars; - nvars++; + ++nvars; } else if (lhs->class == token_sym) complain (_("rule given for %s, which is a token"), lhs->tag); @@ -1494,7 +1464,7 @@ readgram (void) If one does, exit this rule now. */ if (t == tok_identifier) { - bucket *ssave; + symbol_t *ssave; token_t t1; ssave = symval; @@ -1523,13 +1493,13 @@ readgram (void) inserting the new rule before it. */ /* Make a dummy nonterminal, a gensym. */ - bucket *sdummy = gensym (); + symbol_t *sdummy = gensym (); /* Make a new rule, whose body is empty, before the current one, so that the action just read can belong to it. */ - nrules++; - nitems++; + ++nrules; + ++nritems; p = symbol_list_new (sdummy); /* Attach its lineno to that of the host rule. */ p->line = crule->line; @@ -1550,7 +1520,7 @@ readgram (void) /* Insert the dummy generated by that rule into this rule. */ - nitems++; + ++nritems; p = symbol_list_new (sdummy); p1->next = p; p1 = p; @@ -1560,7 +1530,7 @@ readgram (void) if (t == tok_identifier) { - nitems++; + ++nritems; p = symbol_list_new (symval); p1->next = p; p1 = p; @@ -1569,9 +1539,9 @@ readgram (void) { parse_action (crule, rulelength); action_flag = 1; - xactions++; /* JF */ + ++xactions; /* JF */ } - rulelength++; + ++rulelength; } /* end of read rhs of rule */ /* Put an empty link in the list to mark the end of this rule */ @@ -1587,15 +1557,6 @@ readgram (void) t = lex (); } - if (t == tok_guard) - { - if (!semantic_parser) - complain (_("%%guard present but %%semantic_parser not specified")); - - parse_guard (crule, rulelength); - t = lex (); - } - if (t == tok_left_curly) { /* This case never occurs -wjh */ @@ -1603,7 +1564,7 @@ readgram (void) complain (_("two actions at end of one rule")); parse_action (crule, rulelength); action_flag = 1; - xactions++; /* -wjh */ + ++xactions; /* -wjh */ t = lex (); } /* If $$ is being set in default way, report if any type @@ -1638,7 +1599,7 @@ readgram (void) fatal (_("no rules in the input grammar")); /* Report any undefined symbols and consider them nonterminals. */ - buckets_do (bucket_check_defined, NULL); + symbols_do (symbol_check_defined, NULL); /* Insert the initial rule, which line is that of the first rule (not that of the start symbol): @@ -1651,17 +1612,14 @@ readgram (void) p->next->next->next = symbol_list_new (NULL); p->next->next->next->next = grammar; nrules += 1; - nitems += 3; + nritems += 3; grammar = p; startval = axiom; - if (nsyms > MAXSHORT) + if (nsyms > SHRT_MAX) fatal (_("too many symbols (tokens plus nonterminals); maximum %d"), - MAXSHORT); + SHRT_MAX); - if (getenv ("DEBUG")) - fprintf (stderr, "nsyms == ntokens + nvars: %d == %d + %d\n", - nsyms, ntokens, nvars); assert (nsyms == ntokens + nvars); } @@ -1684,7 +1642,7 @@ read_additionnal_code (void) } while ((c = getc (finput)) != EOF) - obstack_1grow (&el_obstack, c); + copy_character (&el_obstack, c); obstack_1grow (&el_obstack, 0); muscle_insert ("epilogue", obstack_finish (&el_obstack)); @@ -1699,34 +1657,50 @@ read_additionnal_code (void) static void token_translations_init (void) { - int last_user_token_number = 256; + int num_256_available_p = TRUE; int i; - /* Set the user numbers. */ + /* Find the highest user token number, and whether 256, the POSIX + preferred user token number for the error token, is used. */ + max_user_token_number = 0; + for (i = 0; i < ntokens; ++i) + { + symbol_t *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 256 is not used, assign it to error, to follow POSIX. */ + if (num_256_available_p + && errtoken->user_token_number == USER_NUMBER_UNDEFINED) + errtoken->user_token_number = 256; + + /* Set the missing user numbers. */ + if (max_user_token_number < 256) + max_user_token_number = 256; + for (i = 0; i < ntokens; ++i) { - bucket *this = symbols[i]; - if (getenv ("DEBUG")) - fprintf (stderr, "UserVal %s = %d (val = %d)\n", - this->tag, this->user_token_number, this->number); - if (this->user_token_number == SUNDEF) - this->user_token_number = ++last_user_token_number; + symbol_t *this = symbols[i]; + if (this->user_token_number == USER_NUMBER_UNDEFINED) + 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; - if (getenv ("DEBUG")) - fprintf (stderr, "Now: UserVal %s = %d (val = %d)\n", - this->tag, this->user_token_number, this->number); } - token_translations = XCALLOC (short, max_user_token_number + 1); + token_translations = XCALLOC (token_number_t, max_user_token_number + 1); /* Initialize all entries for literal tokens to 2, the internal token number for $undefined., which represents all invalid inputs. */ for (i = 0; i < max_user_token_number + 1; i++) - token_translations[i] = 2; - - buckets_do (bucket_translation, NULL); + token_translations[i] = undeftoken->number; + symbols_do (symbol_translation, NULL); } @@ -1738,15 +1712,13 @@ token_translations_init (void) static void packsymbols (void) { - symbols = XCALLOC (bucket *, nsyms); + symbols = XCALLOC (symbol_t *, nsyms); - buckets_do (bucket_check_alias_consistence, NULL); - buckets_do (bucket_pack, NULL); + symbols_do (symbol_check_alias_consistence, NULL); + symbols_do (symbol_pack, NULL); token_translations_init (); - error_token_number = errtoken->number; - if (startval->class == unknown_sym) fatal (_("the start symbol %s is undefined"), startval->tag); else if (startval->class == token_sym) @@ -1764,15 +1736,11 @@ packsymbols (void) static void packgram (void) { - int itemno; + unsigned int itemno; int ruleno; symbol_list *p; - /* We use short to index items. */ - if (nitems >= MAXSHORT) - fatal (_("too many items (max %d)"), MAXSHORT); - - ritem = XCALLOC (short, nitems + 1); + ritem = XCALLOC (item_number_t, nritems); rules = XCALLOC (rule_t, nrules) - 1; itemno = 0; @@ -1781,7 +1749,7 @@ packgram (void) p = grammar; while (p) { - bucket *ruleprec = p->ruleprec; + symbol_t *ruleprec = p->ruleprec; rules[ruleno].user_number = ruleno; rules[ruleno].number = ruleno; rules[ruleno].lhs = p->sym; @@ -1790,13 +1758,13 @@ packgram (void) rules[ruleno].useful = TRUE; rules[ruleno].action = p->action; rules[ruleno].action_line = p->action_line; - rules[ruleno].guard = p->guard; - rules[ruleno].guard_line = p->guard_line; p = p->next; while (p && p->sym) { - ritem[itemno++] = p->sym->number; + /* item_number_t = token_number_t. + But the former needs to contain more: negative rule numbers. */ + ritem[itemno++] = token_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) @@ -1813,27 +1781,24 @@ packgram (void) rules[ruleno].prec = ruleprec; } ritem[itemno++] = -ruleno; - ruleno++; + ++ruleno; if (p) p = p->next; } - ritem[itemno] = 0; - nritems = itemno; - assert (nritems == nitems); + assert (itemno == nritems); if (trace_flag) ritem_print (stderr); } -/*-------------------------------------------------------------------. -| Read in the grammar specification and record it in the format | -| described in gram.h. All guards are copied into the GUARD_OBSTACK | -| and all actions into ACTION_OBSTACK, in each case forming the body | -| of a C function (YYGUARD or YYACTION) which contains a switch | -| statement to decide which guard or action to execute. | -`-------------------------------------------------------------------*/ +/*------------------------------------------------------------------. +| Read in the grammar specification and record it in the format | +| described in gram.h. All actions are copied into ACTION_OBSTACK, | +| in each case forming the body of a C function (YYACTION) which | +| contains a switch statement to decide which action to execute. | +`------------------------------------------------------------------*/ void reader (void) @@ -1845,7 +1810,7 @@ reader (void) obstack_init (&muscle_obstack); /* Initialize the symbol table. */ - buckets_new (); + symbols_new (); /* Construct the axiom symbol. */ axiom = getsym ("$axiom"); @@ -1856,19 +1821,18 @@ reader (void) errtoken = getsym ("error"); errtoken->class = token_sym; errtoken->number = ntokens++; - errtoken->user_token_number = 256; /* Value specified by POSIX. */ /* Construct a token that represents all undefined literal tokens. It is always token number 2. */ undeftoken = getsym ("$undefined."); undeftoken->class = token_sym; undeftoken->number = ntokens++; - undeftoken->user_token_number = 2; /* Initialize the obstacks. */ obstack_init (&action_obstack); - obstack_init (&attrs_obstack); obstack_init (&output_obstack); + obstack_init (&pre_prologue_obstack); + obstack_init (&post_prologue_obstack); finput = xfopen (infile, "r"); @@ -1888,7 +1852,7 @@ reader (void) } /* Read in the grammar, build grammar in list form. Write out - guards and actions. */ + actions. */ readgram (); /* Some C code is given at the end of the grammar file. */ read_additionnal_code (); @@ -1906,12 +1870,3 @@ reader (void) /* The grammar as a symbol_list is no longer needed. */ LIST_FREE (symbol_list, grammar); } - -void -grammar_free (void) -{ - XFREE (ritem); - free (rules + 1); - /* Free the symbol table data structure. */ - buckets_free (); -}