X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/337bab46f26130e30430789b21aa7cf44fbf7c87..e83d80b8ed8587fd05437d33f2011403f55f1f9d:/src/reader.c diff --git a/src/reader.c b/src/reader.c index dc49be3e..099ce309 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,5 +1,5 @@ /* Input parser for bison - Copyright 1984, 1986, 1989, 1992, 1998, 2000 + Copyright 1984, 1986, 1989, 1992, 1998, 2000, 2001 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -26,22 +26,21 @@ #include "quote.h" #include "getargs.h" #include "files.h" -#include "xalloc.h" #include "symtab.h" +#include "options.h" #include "lex.h" #include "gram.h" #include "complain.h" #include "output.h" #include "reader.h" #include "conflicts.h" - -/* Number of slots allocated (but not necessarily used yet) in `rline' */ -static int rline_allocated; +#include "muscle_tab.h" typedef struct symbol_list { struct symbol_list *next; bucket *sym; + int line; bucket *ruleprec; } symbol_list; @@ -63,11 +62,21 @@ static int typed; /* Incremented for each %left, %right or %nonassoc seen */ static int lastprec; -/* Incremented for each generated symbol */ -static int gensym_count; - static bucket *errtoken; static bucket *undeftoken; + + +static symbol_list * +symbol_list_new (bucket *sym) +{ + symbol_list *res = XMALLOC (symbol_list, 1); + res->next = NULL; + res->sym = sym; + res->line = lineno; + res->ruleprec = NULL; + return res; +} + /*===================\ @@ -125,7 +134,7 @@ read_signed_integer (FILE *stream) `--------------------------------------------------------------*/ static char * -get_type_name (int n, symbol_list * rule) +get_type_name (int n, symbol_list *rule) { int i; symbol_list *rp; @@ -159,11 +168,12 @@ get_type_name (int n, symbol_list * rule) `------------------------------------------------------------*/ static inline void -copy_string (FILE *fin, struct obstack *oout, int match) +copy_string2 (FILE *fin, struct obstack *oout, int match, int store) { int c; - obstack_1grow (oout, match); + if (store) + obstack_1grow (oout, match); c = getc (fin); @@ -195,9 +205,30 @@ copy_string (FILE *fin, struct obstack *oout, int match) c = getc (fin); } - obstack_1grow (oout, c); + if (store) + obstack_1grow (oout, c); +} + +/* FIXME. */ + +static inline void +copy_string (FILE *fin, struct obstack *oout, int match) +{ + copy_string2 (fin, oout, match, 1); } +/* FIXME. */ + +static inline void +copy_identifier (FILE *fin, struct obstack *oout) +{ + int c; + + while (isalnum (c = getc (fin)) || c == '_') + obstack_1grow (oout, c); + + ungetc (c, fin); +} /*-----------------------------------------------------------------. | Dump the wannabee comment from IN to OUT1 and OUT2 (which can be | @@ -309,13 +340,12 @@ copy_at (FILE *fin, struct obstack *oout, int stack_offset) c = getc (fin); if (c == '$') { - obstack_grow_string (oout, "yyloc"); + obstack_sgrow (oout, "yyloc"); locations_flag = 1; } else if (isdigit (c) || c == '-') { int n; - char buf[4096]; ungetc (c, fin); n = read_signed_integer (fin); @@ -347,7 +377,7 @@ copy_dollar (FILE *fin, struct obstack *oout, symbol_list *rule, int stack_offset) { int c = getc (fin); - char *type_name = NULL; + const char *type_name = NULL; /* Get the type name if explicit. */ if (c == '<') @@ -360,7 +390,7 @@ copy_dollar (FILE *fin, struct obstack *oout, if (c == '$') { - obstack_grow_string (oout, "yyval"); + obstack_sgrow (oout, "yyval"); if (!type_name) type_name = get_type_name (0, rule); @@ -408,8 +438,11 @@ copy_definition (void) int after_percent; if (!no_lines_flag) - obstack_fgrow2 (&attrs_obstack, "#line %d %s\n", - lineno, quotearg_style (c_quoting_style, infile)); + { + obstack_fgrow2 (&attrs_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find("filename"))); + } after_percent = 0; @@ -466,8 +499,8 @@ copy_definition (void) static void parse_token_decl (symbol_class what_is, symbol_class what_is_not) { - int token = 0; - char *typename = 0; + token_t token = tok_undef; + char *typename = NULL; /* The symbol being defined. */ struct bucket *symbol = NULL; @@ -486,18 +519,18 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) fatal (_("Premature EOF after %s"), token_buffer); token = lex (); - if (token == COMMA) + if (token == tok_comma) { symbol = NULL; continue; } - if (token == TYPENAME) + if (token == tok_typename) { typename = xstrdup (token_buffer); value_components_used = 1; symbol = NULL; } - else if (token == IDENTIFIER && *symval->tag == '\"' && symbol) + else if (token == tok_identifier && *symval->tag == '\"' && symbol) { if (symval->alias) warn (_("symbol `%s' used more than once as a literal string"), @@ -516,10 +549,9 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) /* symbol and symval combined are only one symbol */ nsyms--; } - translations = 1; symbol = NULL; } - else if (token == IDENTIFIER) + else if (token == tok_identifier) { int oldclass = symval->class; symbol = symval; @@ -538,15 +570,15 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) complain (_("type redeclaration for %s"), symbol->tag); } } - else if (symbol && token == NUMBER) + else if (symbol && token == tok_number) { symbol->user_token_number = numval; - translations = 1; } else { complain (_("`%s' is invalid in %s"), - token_buffer, (what_is == token_sym) ? "%token" : "%nterm"); + token_buffer, + (what_is == token_sym) ? "%token" : "%nterm"); skip_to_char ('%'); } } @@ -563,7 +595,7 @@ parse_start_decl (void) { if (start_flag) complain (_("multiple %s declarations"), "%start"); - if (lex () != IDENTIFIER) + if (lex () != tok_identifier) complain (_("invalid %s declaration"), "%start"); else { @@ -582,7 +614,7 @@ parse_type_decl (void) { char *name; - if (lex () != TYPENAME) + if (lex () != tok_typename) { complain ("%s", _("%type declaration has no ")); skip_to_char ('%'); @@ -593,7 +625,7 @@ parse_type_decl (void) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -606,11 +638,11 @@ parse_type_decl (void) switch (t) { - case COMMA: - case SEMICOLON: + case tok_comma: + case tok_semicolon: break; - case IDENTIFIER: + case tok_identifier: if (symval->type_name == NULL) symval->type_name = name; else if (strcmp (name, symval->type_name) != 0) @@ -643,7 +675,7 @@ parse_assoc_decl (associativity assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -655,14 +687,14 @@ parse_assoc_decl (associativity assoc) switch (t) { - case TYPENAME: + case tok_typename: name = xstrdup (token_buffer); break; - case COMMA: + case tok_comma: break; - case IDENTIFIER: + case tok_identifier: if (symval->prec != 0) complain (_("redefining precedence of %s"), symval->tag); symval->prec = lastprec; @@ -679,11 +711,10 @@ parse_assoc_decl (associativity assoc) } break; - case NUMBER: - if (prev == IDENTIFIER) + case tok_number: + if (prev == tok_identifier) { symval->user_token_number = numval; - translations = 1; } else { @@ -694,7 +725,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -703,16 +734,15 @@ token_buffer); } prev = t; - } } /*--------------------------------------------------------------. -| Copy the union declaration into ATTRS_OBSTACK (and fdefines), | -| where it is made into the definition of YYSTYPE, the type of | -| elements of the parser value stack. | +| Copy the union declaration into the stype muscle | +| (and fdefines), where it is made into the definition of | +| YYSTYPE, the type of elements of the parser value stack. | `--------------------------------------------------------------*/ static void @@ -720,29 +750,43 @@ parse_union_decl (void) { int c; int count = 0; + struct obstack union_obstack; + const char *prologue = "\ +#ifndef YYSTYPE\n\ +typedef union"; + const char *epilogue = "\ + yystype;\n\ +# define YYSTYPE yystype\n\ +#endif\n"; if (typed) complain (_("multiple %s declarations"), "%union"); typed = 1; - if (!no_lines_flag) - obstack_fgrow2 (&attrs_obstack, "\n#line %d %s\n", - lineno, quotearg_style (c_quoting_style, infile)); - else + /* FIXME: I'm worried: are you sure attrs_obstack is properly + filled? */ + /* I don't see any reasons to keep this line, because we should + create a special skeleton for this option. */ + if (no_lines_flag) obstack_1grow (&attrs_obstack, '\n'); - obstack_grow_string (&attrs_obstack, "typedef union"); + obstack_init (&union_obstack); + obstack_sgrow (&union_obstack, "union"); if (defines_flag) - obstack_grow_string (&defines_obstack, "typedef union"); + obstack_sgrow (&defines_obstack, prologue); c = getc (finput); while (c != EOF) { - obstack_1grow (&attrs_obstack, c); - if (defines_flag) - obstack_1grow (&defines_obstack, c); + /* If C contains '/', it is output by copy_comment (). */ + if (c != '/') + { + obstack_1grow (&union_obstack, c); + if (defines_flag) + obstack_1grow (&defines_obstack, c); + } switch (c) { @@ -751,7 +795,7 @@ parse_union_decl (void) break; case '/': - copy_comment2 (finput, &defines_obstack, &attrs_obstack); + copy_comment2 (finput, &defines_obstack, &union_obstack); break; case '{': @@ -764,19 +808,21 @@ parse_union_decl (void) count--; if (count <= 0) { - obstack_grow_string (&attrs_obstack, " YYSTYPE;\n"); if (defines_flag) - obstack_grow_string (&defines_obstack, " YYSTYPE;\n"); + obstack_sgrow (&defines_obstack, epilogue); /* JF don't choke on trailing semi */ c = skip_white_space (); if (c != ';') ungetc (c, finput); + obstack_1grow (&union_obstack, 0); + muscle_insert ("stype", obstack_finish (&union_obstack)); return; } } c = getc (finput); } + } @@ -821,14 +867,13 @@ parse_expect_decl (void) static void parse_thong_decl (void) { - int token; + token_t token; struct bucket *symbol; char *typename = 0; - int usrtoknum; + int usrtoknum = SUNDEF; - translations = 1; token = lex (); /* fetch typename or first token */ - if (token == TYPENAME) + if (token == tok_typename) { typename = xstrdup (token_buffer); value_components_used = 1; @@ -837,7 +882,7 @@ parse_thong_decl (void) /* process first token */ - if (token != IDENTIFIER) + if (token != tok_identifier) { complain (_("unrecognized item %s, expected an identifier"), token_buffer); @@ -851,17 +896,15 @@ parse_thong_decl (void) token = lex (); /* get number or literal string */ - if (token == NUMBER) + if (token == tok_number) { usrtoknum = numval; token = lex (); /* okay, did number, now get literal */ } - else - usrtoknum = 0; /* process literal string token */ - if (token != IDENTIFIER || *symval->tag != '\"') + if (token != tok_identifier || *symval->tag != '\"') { complain (_("expected string constant instead of %s"), token_buffer); skip_to_char ('%'); @@ -878,6 +921,102 @@ parse_thong_decl (void) nsyms--; } +static void +parse_muscle_decl (void) +{ + int ch = ungetc (skip_white_space (), finput); + char* muscle_key; + char* muscle_value; + + /* Read key. */ + if (!isalpha (ch) && ch != '_') + { + complain (_("invalid %s declaration"), "%define"); + skip_to_char ('%'); + return; + } + copy_identifier (finput, &muscle_obstack); + obstack_1grow (&muscle_obstack, 0); + muscle_key = obstack_finish (&muscle_obstack); + + /* Read value. */ + ch = skip_white_space (); + if (ch != '"') + { + ungetc (ch, finput); + if (ch != EOF) + { + complain (_("invalid %s declaration"), "%define"); + skip_to_char ('%'); + return; + } + else + fatal (_("Premature EOF after %s"), "\""); + } + copy_string2 (finput, &muscle_obstack, '"', 0); + obstack_1grow (&muscle_obstack, 0); + muscle_value = obstack_finish (&muscle_obstack); + + /* Store the (key, value) pair in the environment. */ + muscle_insert (muscle_key, muscle_value); +} + + + +/*---------------------------------. +| Parse a double quoted parameter. | +`---------------------------------*/ + +static const char * +parse_dquoted_param (const char *from) +{ + struct obstack param_obstack; + const char *param = NULL; + int c; + + obstack_init (¶m_obstack); + c = skip_white_space (); + + if (c != '"') + { + complain (_("invalid %s declaration"), from); + ungetc (c, finput); + skip_to_char ('%'); + return NULL; + } + + for (;;) + { + if (literalchar (NULL, &c, '\"')) + obstack_1grow (¶m_obstack, c); + else + break; + } + + obstack_1grow (¶m_obstack, '\0'); + param = obstack_finish (¶m_obstack); + + if (c != '"' || strlen (param) == 0) + { + complain (_("invalid %s declaration"), from); + if (c != '"') + ungetc (c, finput); + skip_to_char ('%'); + return NULL; + } + + return param; +} + +/*----------------------------------. +| Parse what comes after %skeleton. | +`----------------------------------*/ + +static void +parse_skel_decl (void) +{ + skeleton = parse_dquoted_param ("%skeleton"); +} /*----------------------------------------------------------------. | Read from finput until `%%' is seen. Discard the `%%'. Handle | @@ -888,76 +1027,81 @@ parse_thong_decl (void) static void read_declarations (void) { - int c; - int tok; - for (;;) { - c = skip_white_space (); + int c = skip_white_space (); if (c == '%') { - tok = parse_percent_token (); + token_t tok = parse_percent_token (); switch (tok) { - case TWO_PERCENTS: + case tok_two_percents: return; - case PERCENT_LEFT_CURLY: + case tok_percent_left_curly: copy_definition (); break; - case TOKEN: + case tok_token: parse_token_decl (token_sym, nterm_sym); break; - case NTERM: + case tok_nterm: parse_token_decl (nterm_sym, token_sym); break; - case TYPE: + case tok_type: parse_type_decl (); break; - case START: + case tok_start: parse_start_decl (); break; - case UNION: + case tok_union: parse_union_decl (); break; - case EXPECT: + case tok_expect: parse_expect_decl (); break; - case THONG: + + case tok_thong: parse_thong_decl (); break; - case LEFT: + case tok_left: parse_assoc_decl (left_assoc); break; - case RIGHT: + case tok_right: parse_assoc_decl (right_assoc); break; - case NONASSOC: + case tok_nonassoc: parse_assoc_decl (non_assoc); break; - case SEMANTIC_PARSER: - semantic_parser = 1; + case tok_define: + parse_muscle_decl (); break; - case PURE_PARSER: - pure_parser = 1; + case tok_skel: + parse_skel_decl (); break; - case NOOP: + case tok_noop: break; + case tok_stropt: + case tok_intopt: + case tok_obsolete: + abort (); + break; + + case tok_illegal: default: complain (_("unrecognized: %s"), token_buffer); skip_to_char ('%'); @@ -987,20 +1131,18 @@ copy_action (symbol_list *rule, int stack_offset) { int c; int count; - char buf[4096]; /* offset is always 0 if parser has already popped the stack pointer */ if (semantic_parser) stack_offset = 0; - sprintf (buf, "\ncase %d:\n", nrules); - obstack_grow (&action_obstack, buf, strlen (buf)); + obstack_fgrow1 (&action_obstack, "\ncase %d:\n", nrules); if (!no_lines_flag) { - sprintf (buf, "#line %d %s\n", - lineno, quotearg_style (c_quoting_style, infile)); - obstack_grow (&action_obstack, buf, strlen (buf)); + obstack_fgrow2 (&action_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find ("filename"))); } obstack_1grow (&action_obstack, '{'); @@ -1061,7 +1203,7 @@ copy_action (symbol_list *rule, int stack_offset) } } - obstack_grow_string (&action_obstack, ";\n break;}"); + obstack_sgrow (&action_obstack, ";\n break;}"); } /*-------------------------------------------------------------------. @@ -1086,8 +1228,9 @@ copy_guard (symbol_list *rule, int stack_offset) obstack_fgrow1 (&guard_obstack, "\ncase %d:\n", nrules); if (!no_lines_flag) - obstack_fgrow2 (&guard_obstack, "#line %d %s\n", - lineno, quotearg_style (c_quoting_style, infile)); + obstack_fgrow2 (&guard_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find ("filename"))); obstack_1grow (&guard_obstack, '{'); count = 0; @@ -1149,7 +1292,7 @@ copy_guard (symbol_list *rule, int stack_offset) c = skip_white_space (); - obstack_grow_string (&guard_obstack, ";\n break;}"); + obstack_sgrow (&guard_obstack, ";\n break;}"); if (c == '{') copy_action (rule, stack_offset); else if (c == '=') @@ -1163,20 +1306,6 @@ copy_guard (symbol_list *rule, int stack_offset) } -static void -record_rule_line (void) -{ - /* Record each rule's source line number in rline table. */ - - if (nrules >= rline_allocated) - { - rline_allocated = nrules * 2; - rline = XREALLOC (rline, short, rline_allocated); - } - rline[nrules] = lineno; -} - - /*-------------------------------------------------------------------. | Generate a dummy symbol, a nonterminal, whose name cannot conflict | | with the user's names. | @@ -1185,9 +1314,14 @@ record_rule_line (void) static bucket * gensym (void) { + /* Incremented for each generated symbol */ + static int gensym_count = 0; + static char buf[256]; + bucket *sym; - sprintf (token_buffer, "@%d", ++gensym_count); + sprintf (buf, "@%d", ++gensym_count); + token_buffer = buf; sym = getsym (token_buffer); sym->class = nterm_sym; sym->value = nvars++; @@ -1205,12 +1339,12 @@ static int get_type (void) { int k; - int t; + token_t token; char *name; - t = lex (); + token = lex (); - if (t != TYPENAME) + if (token != tok_typename) { complain (_("invalid %s declaration"), "%type"); return t; @@ -1220,17 +1354,17 @@ get_type (void) for (;;) { - t = lex (); + token = lex (); - switch (t) + switch (token) { - case SEMICOLON: + case tok_semicolon: return lex (); - case COMMA: + case tok_comma: break; - case IDENTIFIER: + case tok_identifier: if (symval->type_name == NULL) symval->type_name = name; else if (strcmp (name, symval->type_name) != 0) @@ -1239,7 +1373,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1260,7 +1394,7 @@ get_type (void) static void readgram (void) { - int t; + token_t t; bucket *lhs = NULL; symbol_list *p; symbol_list *p1; @@ -1276,9 +1410,9 @@ readgram (void) t = lex (); - while (t != TWO_PERCENTS && t != ENDFILE) + while (t != tok_two_percents && t != tok_eof) { - if (t == IDENTIFIER || t == BAR) + if (t == tok_identifier || t == tok_bar) { int action_flag = 0; /* Number of symbols in rhs of this rule so far */ @@ -1286,7 +1420,7 @@ readgram (void) int xactions = 0; /* JF for error checking */ bucket *first_rhs = 0; - if (t == IDENTIFIER) + if (t == tok_identifier) { lhs = symval; @@ -1297,14 +1431,14 @@ readgram (void) } t = lex (); - if (t != COLON) + if (t != tok_colon) { complain (_("ill-formed rule: initial symbol not followed by colon")); unlex (t); } } - if (nrules == 0 && t == BAR) + if (nrules == 0 && t == tok_bar) { complain (_("grammar starts with vertical bar")); lhs = symval; /* BOGUS: use a random symval */ @@ -1314,10 +1448,7 @@ readgram (void) nrules++; nitems++; - record_rule_line (); - - p = XCALLOC (symbol_list, 1); - p->sym = lhs; + p = symbol_list_new (lhs); crule1 = p1; if (p1) @@ -1344,28 +1475,28 @@ readgram (void) for (;;) { t = lex (); - if (t == PREC) + if (t == tok_prec) { t = lex (); crule->ruleprec = symval; t = lex (); } - if (!(t == IDENTIFIER || t == LEFT_CURLY)) + if (!(t == tok_identifier || t == tok_left_curly)) break; /* If next token is an identifier, see if a colon follows it. If one does, exit this rule now. */ - if (t == IDENTIFIER) + if (t == tok_identifier) { bucket *ssave; - int t1; + token_t t1; ssave = symval; t1 = lex (); unlex (t1); symval = ssave; - if (t1 == COLON) + if (t1 == tok_colon) break; if (!first_rhs) /* JF */ @@ -1379,47 +1510,45 @@ readgram (void) non-terminal. */ if (action_flag) { - bucket *sdummy; - /* Since the action was written out with this rule's number, we must give the new rule this number by inserting the new rule before it. */ /* Make a dummy nonterminal, a gensym. */ - sdummy = gensym (); + bucket *sdummy = gensym (); - /* Make a new rule, whose body is empty, - before the current one, so that the action - just read can belong to it. */ + /* Make a new rule, whose body is empty, before the + current one, so that the action just read can + belong to it. */ nrules++; nitems++; - record_rule_line (); - p = XCALLOC (symbol_list, 1); + p = symbol_list_new (sdummy); + /* Attach its lineno to that of the host rule. */ + p->line = crule->line; if (crule1) crule1->next = p; else grammar = p; - p->sym = sdummy; - crule1 = XCALLOC (symbol_list, 1); - p->next = crule1; + /* End of the rule. */ + crule1 = symbol_list_new (NULL); crule1->next = crule; + p->next = crule1; + /* Insert the dummy generated by that rule into this rule. */ nitems++; - p = XCALLOC (symbol_list, 1); - p->sym = sdummy; + p = symbol_list_new (sdummy); p1->next = p; p1 = p; action_flag = 0; } - if (t == IDENTIFIER) + if (t == tok_identifier) { nitems++; - p = XCALLOC (symbol_list, 1); - p->sym = symval; + p = symbol_list_new (symval); p1->next = p; p1 = p; } @@ -1433,18 +1562,18 @@ readgram (void) } /* end of read rhs of rule */ /* Put an empty link in the list to mark the end of this rule */ - p = XCALLOC (symbol_list, 1); + p = symbol_list_new (NULL); p1->next = p; p1 = p; - if (t == PREC) + if (t == tok_prec) { complain (_("two @prec's in a row")); t = lex (); crule->ruleprec = symval; t = lex (); } - if (t == GUARD) + if (t == tok_guard) { if (!semantic_parser) complain (_("%%guard present but %%semantic_parser not specified")); @@ -1452,7 +1581,7 @@ readgram (void) copy_guard (crule, rulelength); t = lex (); } - else if (t == LEFT_CURLY) + else if (t == tok_left_curly) { /* This case never occurs -wjh */ if (action_flag) @@ -1477,7 +1606,7 @@ readgram (void) /* Warn if there is no default for $$ but we need one. */ else if (!xactions && !first_rhs && lhs->type_name != 0) complain (_("empty rule for typed nonterminal, and no action")); - if (t == SEMICOLON) + if (t == tok_semicolon) t = lex (); } #if 0 @@ -1487,31 +1616,31 @@ readgram (void) b) most of them scan forward until finding a next % thus they may swallow lots of intervening rules */ - else if (t == TOKEN) + else if (t == tok_token) { parse_token_decl (token_sym, nterm_sym); t = lex (); } - else if (t == NTERM) + else if (t == tok_nterm) { parse_token_decl (nterm_sym, token_sym); t = lex (); } - else if (t == TYPE) + else if (t == tok_type) { t = get_type (); } - else if (t == UNION) + else if (t == tok_union) { parse_union_decl (); t = lex (); } - else if (t == EXPECT) + else if (t == tok_expect) { parse_expect_decl (); t = lex (); } - else if (t == START) + else if (t == tok_start) { parse_start_decl (); t = lex (); @@ -1520,7 +1649,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1533,22 +1662,6 @@ readgram (void) if (nrules == 0) fatal (_("no rules in the input grammar")); - /* JF put out same default YYSTYPE as YACC does */ - if (typed == 0 - && !value_components_used) - { - /* We used to use `unsigned long' as YYSTYPE on MSDOS, - but it seems better to be consistent. - Most programs should declare their own type anyway. */ - obstack_grow_string (&attrs_obstack, - "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); - if (defines_flag) - obstack_grow_string (&defines_obstack, "\ -#ifndef YYSTYPE\n\ -# define YYSTYPE int\n\ -#endif\n"); - } - /* Report any undefined symbols and consider them nonterminals. */ for (bp = firstsymbol; bp; bp = bp->next) @@ -1563,6 +1676,25 @@ readgram (void) ntokens = nsyms - nvars; } + +/* At the end of the grammar file, some C source code must + be stored. It is going to be associated to the epilogue + directive. */ +static void +read_additionnal_code (void) +{ + char c; + struct obstack el_obstack; + + obstack_init (&el_obstack); + + while ((c = getc (finput)) != EOF) + obstack_1grow (&el_obstack, c); + + obstack_1grow (&el_obstack, 0); + muscle_insert ("epilogue", obstack_finish (&el_obstack)); +} + /*--------------------------------------------------------------. | For named tokens, but not literal ones, define the name. The | @@ -1603,15 +1735,53 @@ output_token_defines (struct obstack *oout) if (c != '\0') continue; - obstack_fgrow2 (oout, "#define\t%s\t%d\n", - symbol, - ((translations && !raw_flag) - ? bp->user_token_number : bp->value)); + obstack_fgrow2 (oout, "# define\t%s\t%d\n", + symbol, bp->user_token_number); if (semantic_parser) - obstack_fgrow2 (oout, "#define\tT%s\t%d\n", symbol, bp->value); + /* FIXME: This is certainly dead wrong, and should be just as + above. --akim. */ + obstack_fgrow2 (oout, "# define\tT%s\t%d\n", symbol, bp->value); } +} + - obstack_1grow (oout, '\n'); +/*------------------------------------------------------------------. +| Set TOKEN_TRANSLATIONS. Check that no two symbols share the same | +| number. | +`------------------------------------------------------------------*/ + +static void +token_translations_init (void) +{ + bucket *bp = NULL; + int i; + + token_translations = XCALLOC (short, max_user_token_number + 1); + + /* Initialize all entries for literal tokens to 2, the internal + token number for $undefined., which represents all invalid + inputs. */ + for (i = 0; i <= max_user_token_number; i++) + token_translations[i] = 2; + + for (bp = firstsymbol; bp; bp = bp->next) + { + /* Non-terminal? */ + if (bp->value >= ntokens) + continue; + /* A token string alias? */ + if (bp->user_token_number == SALIAS) + continue; + + assert (bp->user_token_number != SUNDEF); + + /* A token which translation has already been set? */ + if (token_translations[bp->user_token_number] != 2) + complain (_("tokens %s and %s both assigned number %d"), + tags[token_translations[bp->user_token_number]], + bp->tag, bp->user_token_number); + token_translations[bp->user_token_number] = bp->value; + } } @@ -1624,22 +1794,21 @@ output_token_defines (struct obstack *oout) static void packsymbols (void) { - bucket *bp; + bucket *bp = NULL; int tokno = 1; - int i; int last_user_token_number; static char DOLLAR[] = "$"; - /* int lossage = 0; JF set but not used */ - tags = XCALLOC (char *, nsyms + 1); - tags[0] = DOLLAR; user_toknums = XCALLOC (short, nsyms + 1); - user_toknums[0] = 0; sprec = XCALLOC (short, nsyms); sassoc = XCALLOC (short, nsyms); + /* The EOF token. */ + tags[0] = DOLLAR; + user_toknums[0] = 0; + max_user_token_number = 256; last_user_token_number = 256; @@ -1692,7 +1861,7 @@ packsymbols (void) if (bp->class == token_sym) { - if (translations && !(bp->user_token_number)) + if (bp->user_token_number == SUNDEF) bp->user_token_number = ++last_user_token_number; if (bp->user_token_number > max_user_token_number) max_user_token_number = bp->user_token_number; @@ -1702,75 +1871,62 @@ packsymbols (void) user_toknums[bp->value] = bp->user_token_number; sprec[bp->value] = bp->prec; sassoc[bp->value] = bp->assoc; - } - if (translations) - { - int j; - - token_translations = XCALLOC (short, 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 (j = 0; j <= max_user_token_number; j++) - token_translations[j] = 2; - - for (bp = firstsymbol; bp; bp = bp->next) - { - if (bp->value >= ntokens) - continue; /* non-terminal */ - if (bp->user_token_number == SALIAS) - continue; - if (token_translations[bp->user_token_number] != 2) - complain (_("tokens %s and %s both assigned number %d"), - tags[token_translations[bp->user_token_number]], - bp->tag, bp->user_token_number); - token_translations[bp->user_token_number] = bp->value; - } - } + token_translations_init (); error_token_number = errtoken->value; - if (!no_parser_flag) - output_token_defines (&table_obstack); - if (startval->class == unknown_sym) fatal (_("the start symbol %s is undefined"), startval->tag); else if (startval->class == token_sym) fatal (_("the start symbol %s is a token"), startval->tag); start_symbol = startval->value; +} + + +/*-----------------------------------. +| Output definition of token names. | +`-----------------------------------*/ + +static void +symbols_output (void) +{ + { + struct obstack tokendefs; + obstack_init (&tokendefs); + output_token_defines (&tokendefs); + obstack_1grow (&tokendefs, 0); + muscle_insert ("tokendef", xstrdup (obstack_finish (&tokendefs))); + obstack_free (&tokendefs, NULL); + } if (defines_flag) { output_token_defines (&defines_obstack); if (!pure_parser) + obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", + spec_name_prefix); + if (semantic_parser) { - if (spec_name_prefix) - obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", - spec_name_prefix); - else - obstack_grow_string (&defines_obstack, - "\nextern YYSTYPE yylval;\n"); - } + int i; - if (semantic_parser) - for (i = ntokens; i < nsyms; i++) - { - /* don't make these for dummy nonterminals made by gensym. */ - if (*tags[i] != '@') - obstack_fgrow2 (&defines_obstack, - "#define\tNT%s\t%d\n", tags[i], i); - } + for (i = ntokens; i < nsyms; i++) + { + /* don't make these for dummy nonterminals made by gensym. */ + if (*tags[i] != '@') + obstack_fgrow2 (&defines_obstack, + "# define\tNT%s\t%d\n", tags[i], i); + } #if 0 - /* `fdefines' is now a temporary file, so we need to copy its - contents in `done', so we can't close it here. */ - fclose (fdefines); - fdefines = NULL; + /* `fdefines' is now a temporary file, so we need to copy its + contents in `done', so we can't close it here. */ + fclose (fdefines); + fdefines = NULL; #endif + } } } @@ -1787,14 +1943,8 @@ packgram (void) int ruleno; symbol_list *p; - bucket *ruleprec; - ritem = XCALLOC (short, nitems + 1); - rlhs = XCALLOC (short, nrules) - 1; - rrhs = XCALLOC (short, nrules) - 1; - rprec = XCALLOC (short, nrules) - 1; - rprecsym = XCALLOC (short, nrules) - 1; - rassoc = XCALLOC (short, nrules) - 1; + rule_table = XCALLOC (rule_t, nrules) - 1; itemno = 0; ruleno = 1; @@ -1802,9 +1952,11 @@ packgram (void) p = grammar; while (p) { - rlhs[ruleno] = p->sym->value; - rrhs[ruleno] = itemno; - ruleprec = p->ruleprec; + bucket *ruleprec = p->ruleprec; + rule_table[ruleno].lhs = p->sym->value; + rule_table[ruleno].rhs = itemno; + rule_table[ruleno].line = p->line; + rule_table[ruleno].useful = TRUE; p = p->next; while (p && p->sym) @@ -1814,8 +1966,8 @@ packgram (void) of the last token in it. */ if (p->sym->class == token_sym) { - rprec[ruleno] = p->sym->prec; - rassoc[ruleno] = p->sym->assoc; + rule_table[ruleno].prec = p->sym->prec; + rule_table[ruleno].assoc = p->sym->assoc; } if (p) p = p->next; @@ -1825,9 +1977,9 @@ packgram (void) the specified symbol's precedence replaces the default. */ if (ruleprec) { - rprec[ruleno] = ruleprec->prec; - rassoc[ruleno] = ruleprec->assoc; - rprecsym[ruleno] = ruleprec->value; + rule_table[ruleno].prec = ruleprec->prec; + rule_table[ruleno].assoc = ruleprec->assoc; + rule_table[ruleno].precsym = ruleprec->value; } ritem[itemno++] = -ruleno; @@ -1838,6 +1990,9 @@ packgram (void) } ritem[itemno] = 0; + + if (trace_flag) + ritem_print (stderr); } /*-------------------------------------------------------------------. @@ -1854,41 +2009,33 @@ reader (void) start_flag = 0; startval = NULL; /* start symbol not specified yet. */ -#if 0 - /* initially assume token number translation not needed. */ - translations = 0; -#endif - /* Nowadays translations is always set to 1, since we give `error' a - user-token-number to satisfy the Posix demand for YYERRCODE==256. - */ - translations = 1; - nsyms = 1; nvars = 0; nrules = 0; nitems = 0; - rline_allocated = 10; - rline = XCALLOC (short, rline_allocated); typed = 0; lastprec = 0; - gensym_count = 0; - semantic_parser = 0; pure_parser = 0; grammar = NULL; - init_lex (); + lex_init (); lineno = 1; + /* Initialize the muscle obstack. */ + obstack_init (&muscle_obstack); + /* Initialize the symbol table. */ tabinit (); + /* Construct the error token */ errtoken = getsym ("error"); errtoken->class = token_sym; 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."); @@ -1898,63 +2045,22 @@ reader (void) /* Read the declaration section. Copy %{ ... %} groups to TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left, etc. found there. */ - obstack_1grow (&table_obstack, '\n'); - obstack_fgrow3 (&table_obstack, "\ -/* %s, made from %s\n\ - by GNU bison %s. */\n\ -\n", - no_parser_flag ? "Bison-generated parse tables" : "A Bison parser", - infile, VERSION); - - obstack_grow_string (&table_obstack, - "#define YYBISON 1 /* Identify Bison output. */\n\n"); read_declarations (); - /* Start writing the guard and action files, if they are needed. */ - output_headers (); /* Read in the grammar, build grammar in list form. Write out guards and actions. */ readgram (); + /* Some C code is given at the end of the grammar file. */ + read_additionnal_code (); + /* Now we know whether we need the line-number stack. If we do, - write its type into the .tab.h file. */ - if (defines_flag) - reader_output_yylsp (&defines_obstack); - /* Write closing delimiters for actions and guards. */ - output_trailers (); - if (locations_flag) - obstack_grow_string (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); + write its type into the .tab.h file. + This is no longer need with header skeleton. */ + /* Assign the symbols their symbol numbers. Write #defines for the token symbols into FDEFINES if requested. */ packsymbols (); /* Convert the grammar into the format described in gram.h. */ packgram (); - /* Free the symbol table data structure since symbols are now all - referred to by symbol number. */ - free_symtab (); -} - - -/*------------------------------------------------------------------. -| Define YYLTYPE. Cannot be in the skeleton since we might have to | -| output it in the headers if --defines is used. | -`------------------------------------------------------------------*/ - -void -reader_output_yylsp (struct obstack *oout) -{ - if (locations_flag) - obstack_grow_string (oout, "\ -\n\ -#ifndef YYLTYPE\n\ -typedef struct yyltype\n\ -{\n\ - int first_line;\n\ - int first_column;\n\ -\n\ - int last_line;\n\ - int last_column;\n\ -} yyltype;\n\ -\n\ -# define YYLTYPE yyltype\n\ -#endif\n\ -\n"); + /* Output the headers. */ + symbols_output (); }