X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/5af1f54933220f178509b15ec6b7f65a780630f1..e89a22bfab22e4d2ee73be49dcb66b51f8d0e892:/src/reader.c diff --git a/src/reader.c b/src/reader.c index ee72dfec..529cebe9 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. @@ -22,9 +22,10 @@ #include "system.h" #include "obstack.h" +#include "quotearg.h" +#include "quote.h" #include "getargs.h" #include "files.h" -#include "xalloc.h" #include "symtab.h" #include "lex.h" #include "gram.h" @@ -32,15 +33,12 @@ #include "output.h" #include "reader.h" #include "conflicts.h" -#include "quote.h" - -/* Number of slots allocated (but not necessarily used yet) in `rline' */ -static int rline_allocated; typedef struct symbol_list { struct symbol_list *next; bucket *sym; + int line; bucket *ruleprec; } symbol_list; @@ -62,11 +60,21 @@ 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; +} + /*===================\ @@ -124,7 +132,7 @@ 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; @@ -152,20 +160,17 @@ get_type_name (int n, symbol_list * rule) return rp->sym->type_name; } -/*-----------------------------------------------------------------. -| Dump the string from FIN to FOUT and OOUT if non null. MATCH is | -| the delimiter of the string (either ' or "). | -`-----------------------------------------------------------------*/ +/*------------------------------------------------------------. +| Dump the string from FIN to OOUT if non null. MATCH is the | +| delimiter of the string (either ' or "). | +`------------------------------------------------------------*/ static inline void -copy_string (FILE *fin, FILE *fout, struct obstack *oout, int match) +copy_string (FILE *fin, struct obstack *oout, int match) { int c; - if (fout) - putc (match, fout); - if (oout) - obstack_1grow (oout, match); + obstack_1grow (oout, match); c = getc (fin); @@ -181,20 +186,14 @@ copy_string (FILE *fin, FILE *fout, struct obstack *oout, int match) continue; } - if (fout) - putc (c, fout); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout, c); if (c == '\\') { c = getc (fin); if (c == EOF) fatal (_("unterminated string at end of file")); - if (fout) - putc (c, fout); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout, c); if (c == '\n') lineno++; @@ -203,34 +202,27 @@ copy_string (FILE *fin, FILE *fout, struct obstack *oout, int match) c = getc (fin); } - if (fout) - putc (c, fout); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout, c); } -/*----------------------------------------------------------------. -| Dump the wannabee comment from IN to OUT1 and OUT2. In fact we | -| just saw a `/', which might or might not be a comment. In any | -| case, copy what we saw. | -| | -| OUT2 might be NULL. | -`----------------------------------------------------------------*/ +/*-----------------------------------------------------------------. +| Dump the wannabee comment from IN to OUT1 and OUT2 (which can be | +| NULL). In fact we just saw a `/', which might or might not be a | +| comment. In any case, copy what we saw. | +| | +| OUT2 might be NULL. | +`-----------------------------------------------------------------*/ static inline void -copy_comment2 (FILE *fin, FILE *out1, - struct obstack *oout2, struct obstack *oout) +copy_comment2 (FILE *fin, struct obstack *oout1, struct obstack *oout2) { int cplus_comment; int ended; int c; /* We read a `/', output it. */ - if (out1) - putc ('/', out1); - if (oout) - obstack_1grow (oout, '/'); + obstack_1grow (oout1, '/'); if (oout2) obstack_1grow (oout2, '/'); @@ -247,10 +239,7 @@ copy_comment2 (FILE *fin, FILE *out1, return; } - if (out1) - putc (c, out1); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout1, c); if (oout2) obstack_1grow (oout2, c); c = getc (fin); @@ -262,10 +251,7 @@ copy_comment2 (FILE *fin, FILE *out1, { while (c == '*') { - if (out1) - putc (c, out1); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout1, c); if (oout2) obstack_1grow (oout2, c); c = getc (fin); @@ -273,10 +259,7 @@ copy_comment2 (FILE *fin, FILE *out1, if (c == '/') { - if (out1) - putc (c, out1); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout1, c); if (oout2) obstack_1grow (oout2, c); ended = 1; @@ -285,10 +268,7 @@ copy_comment2 (FILE *fin, FILE *out1, else if (c == '\n') { lineno++; - if (out1) - putc (c, out1); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout1, c); if (oout2) obstack_1grow (oout2, c); if (cplus_comment) @@ -300,10 +280,7 @@ copy_comment2 (FILE *fin, FILE *out1, fatal (_("unterminated comment")); else { - if (out1) - putc (c, out1); - if (oout) - obstack_1grow (oout, c); + obstack_1grow (oout1, c); if (oout2) obstack_1grow (oout2, c); c = getc (fin); @@ -314,50 +291,58 @@ copy_comment2 (FILE *fin, FILE *out1, /*-------------------------------------------------------------------. | Dump the comment (actually the current string starting with a `/') | -| from FIN to FOUT. | +| from FIN to OOUT. | `-------------------------------------------------------------------*/ static inline void -copy_comment (FILE *fin, FILE *fout, struct obstack *oout) +copy_comment (FILE *fin, struct obstack *oout) { - copy_comment2 (fin, fout, NULL, oout); + copy_comment2 (fin, oout, NULL); } /*-----------------------------------------------------------------. -| FIN is pointing to a location (i.e., a `@'). Output to FOUT a | +| 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 `$0' with | +| in the current rule so far, which says where to find `@0' with | | respect to the top of the stack. | `-----------------------------------------------------------------*/ static inline void -copy_at (FILE *fin, FILE *fout, struct obstack *oout, int stack_offset) +copy_at (FILE *fin, struct obstack *oout, + struct symbol_list *rule, int stack_offset) { + symbol_list *rp; int c; c = getc (fin); if (c == '$') { - if (fout) - fprintf (fout, "yyloc"); - if (oout) - obstack_grow_literal_string (oout, "yyloc"); + obstack_sgrow (oout, "yyloc"); locations_flag = 1; } else if (isdigit (c) || c == '-') { - int n; - char buf[4096]; + int n, i; ungetc (c, fin); n = read_signed_integer (fin); - sprintf (buf, "yylsp[%d]", n - stack_offset); - if (fout) - fputs (buf, fout); - if (oout) - obstack_grow (oout, buf, strlen (buf)); + rp = rule; + i = 0; + + while (i < n) + { + rp = rp->next; + if (rp == NULL) + { + complain (_("invalid @ value")); + return; + } + i++; + } + + obstack_fgrow1 (oout, "yylsp[%d]", n - stack_offset); locations_flag = 1; } else @@ -374,17 +359,17 @@ copy_at (FILE *fin, FILE *fout, struct obstack *oout, int stack_offset) | | | Possible inputs: $[]($|integer) | | | -| Output to FOUT a reference to this semantic value. STACK_OFFSET is | +| 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 `$0' with respect to the top of the stack. | `-------------------------------------------------------------------*/ static inline void -copy_dollar (FILE *fin, FILE *fout, struct obstack *oout, +copy_dollar (FILE *fin, struct obstack *oout, symbol_list *rule, int stack_offset) { int c = getc (fin); - char *type_name = NULL; + const char *type_name = NULL; /* Get the type name if explicit. */ if (c == '<') @@ -397,20 +382,12 @@ copy_dollar (FILE *fin, FILE *fout, struct obstack *oout, if (c == '$') { - if (fout) - fputs ("yyval", fout); - if (oout) - obstack_grow_literal_string (oout, "yyval"); + obstack_sgrow (oout, "yyval"); if (!type_name) type_name = get_type_name (0, rule); if (type_name) - { - if (fout) - fprintf (fout, ".%s", type_name); - if (oout) - obstack_fgrow1 (oout, ".%s", type_name); - } + obstack_fgrow1 (oout, ".%s", type_name); if (!type_name && typed) complain (_("$$ of `%s' has no declared type"), rule->sym->tag); @@ -424,18 +401,10 @@ copy_dollar (FILE *fin, FILE *fout, struct obstack *oout, if (!type_name && n > 0) type_name = get_type_name (n, rule); - if (fout) - fprintf (fout, "yyvsp[%d]", n - stack_offset); - if (oout) - obstack_fgrow1 (oout, "yyvsp[%d]", n - stack_offset); + obstack_fgrow1 (oout, "yyvsp[%d]", n - stack_offset); if (type_name) - { - if (fout) - fprintf (fout, ".%s", type_name); - if (oout) - obstack_fgrow1 (oout, ".%s", type_name); - } + obstack_fgrow1 (oout, ".%s", type_name); if (!type_name && typed) complain (_("$%d of `%s' has no declared type"), n, rule->sym->tag); @@ -461,7 +430,8 @@ copy_definition (void) int after_percent; if (!no_lines_flag) - obstack_fgrow2 (&attrs_obstack, "#line %d \"%s\"\n", lineno, infile); + obstack_fgrow2 (&attrs_obstack, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); after_percent = 0; @@ -482,11 +452,11 @@ copy_definition (void) case '\'': case '"': - copy_string (finput, 0, &attrs_obstack, c); + copy_string (finput, &attrs_obstack, c); break; case '/': - copy_comment (finput, 0, &attrs_obstack); + copy_comment (finput, &attrs_obstack); break; case EOF: @@ -518,8 +488,8 @@ 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; @@ -538,18 +508,18 @@ 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"), @@ -568,10 +538,9 @@ 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; @@ -590,15 +559,15 @@ 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 ('%'); } } @@ -615,7 +584,7 @@ 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 { @@ -634,7 +603,7 @@ parse_type_decl (void) { char *name; - if (lex () != TYPENAME) + if (lex () != tok_typename) { complain ("%s", _("%type declaration has no ")); skip_to_char ('%'); @@ -645,7 +614,7 @@ parse_type_decl (void) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -658,11 +627,11 @@ 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) @@ -695,7 +664,7 @@ parse_assoc_decl (associativity assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -707,14 +676,14 @@ 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; @@ -731,11 +700,10 @@ 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 { @@ -746,7 +714,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -755,7 +723,6 @@ token_buffer); } prev = t; - } } @@ -772,6 +739,13 @@ parse_union_decl (void) { int c; int count = 0; + 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"); @@ -779,21 +753,27 @@ parse_union_decl (void) typed = 1; if (!no_lines_flag) - obstack_fgrow2 (&attrs_obstack, "\n#line %d \"%s\"\n", lineno, infile); + obstack_fgrow2 (&attrs_obstack, "\n#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); else obstack_1grow (&attrs_obstack, '\n'); - obstack_grow_literal_string (&attrs_obstack, "typedef union"); + obstack_sgrow (&attrs_obstack, prologue); if (defines_flag) - obstack_grow_literal_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_grow_literal_string (&defines_obstack, c); + + /* If C contains '/', it is output by copy_comment (). */ + if (c != '/') + { + obstack_1grow (&attrs_obstack, c); + if (defines_flag) + obstack_1grow (&defines_obstack, c); + } switch (c) { @@ -802,7 +782,7 @@ parse_union_decl (void) break; case '/': - copy_comment2 (finput, 0, &defines_obstack, &attrs_obstack); + copy_comment2 (finput, &defines_obstack, &attrs_obstack); break; case '{': @@ -815,9 +795,9 @@ parse_union_decl (void) count--; if (count <= 0) { - obstack_grow_literal_string (&attrs_obstack, " YYSTYPE;\n"); + obstack_sgrow (&attrs_obstack, epilogue); if (defines_flag) - obstack_grow_literal_string (&defines_obstack, " YYSTYPE;\n"); + obstack_sgrow (&defines_obstack, epilogue); /* JF don't choke on trailing semi */ c = skip_white_space (); if (c != ';') @@ -872,14 +852,13 @@ 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; @@ -888,7 +867,7 @@ parse_thong_decl (void) /* process first token */ - if (token != IDENTIFIER) + if (token != tok_identifier) { complain (_("unrecognized item %s, expected an identifier"), token_buffer); @@ -902,17 +881,15 @@ 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 ('%'); @@ -930,6 +907,61 @@ parse_thong_decl (void) } +/*------------------------------------------------------------------. +| Parse a double quoted parameter. It was used for | +| %{source,header}_extension. For the moment, It is not used since | +| extension features have been removed. | +`------------------------------------------------------------------*/ + +#if 0 + +static const char * +parse_dquoted_param (const char *from) +{ + char buff[32]; + int c; + int i; + + c = skip_white_space (); + + if (c != '"') + { + ungetc (c, finput); + complain (_("invalid %s declaration"), from); + return NULL; + } + + c = getc (finput); + for (i = 0; (c >= '!') && (c <= '~'); i++) + { + if (c == '"') + break; + + if (c == '\\') + { + c = getc (finput); + if ((c < '!') && (c > '~')) + break; + } + + buff[i] = c; + c = getc (finput); + } + buff[i] = '\0'; + + if (c != '"') + { + ungetc (c, finput); + complain (_("invalid %s declaration"), from); + return NULL; + } + + return xstrdup (buff); +} + +#endif + + /*----------------------------------------------------------------. | Read from finput until `%%' is seen. Discard the `%%'. Handle | | any `%' declarations, and copy the contents of any `%{ ... %}' | @@ -939,80 +971,73 @@ 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: - if (semantic_parser == 0) - { - semantic_parser = 1; - open_extra_files (); - } - break; - - case PURE_PARSER: - pure_parser = 1; + case tok_noop: break; - case NOOP: + case tok_stropt: + case tok_intopt: + case tok_obsolete: + abort (); break; + case tok_illegal: default: complain (_("unrecognized: %s"), token_buffer); skip_to_char ('%'); @@ -1053,7 +1078,8 @@ copy_action (symbol_list *rule, int stack_offset) if (!no_lines_flag) { - sprintf (buf, "#line %d \"%s\"\n", lineno, infile); + sprintf (buf, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); obstack_grow (&action_obstack, buf, strlen (buf)); } obstack_1grow (&action_obstack, '{'); @@ -1079,21 +1105,21 @@ copy_action (symbol_list *rule, int stack_offset) case '\'': case '"': - copy_string (finput, 0, &action_obstack, c); + copy_string (finput, &action_obstack, c); break; case '/': - copy_comment (finput, 0, &action_obstack); + copy_comment (finput, &action_obstack); break; case '$': - copy_dollar (finput, 0, &action_obstack, + copy_dollar (finput, &action_obstack, rule, stack_offset); break; case '@': - copy_at (finput, 0, &action_obstack, - stack_offset); + copy_at (finput, &action_obstack, + rule, stack_offset); break; case EOF: @@ -1115,8 +1141,13 @@ copy_action (symbol_list *rule, int stack_offset) } } - obstack_grow_literal_string (&action_obstack, - ";\n break;}"); + /* As a Bison extension, add the ending semicolon. Since some Yacc + don't do that, help people using bison as a Yacc finding their + missing semicolons. */ + if (yacc_flag) + obstack_sgrow (&action_obstack, "}\n break;"); + else + obstack_sgrow (&action_obstack, ";\n break;}"); } /*-------------------------------------------------------------------. @@ -1139,10 +1170,11 @@ copy_guard (symbol_list *rule, int stack_offset) if (semantic_parser) stack_offset = 0; - fprintf (fguard, "\ncase %d:\n", nrules); + obstack_fgrow1 (&guard_obstack, "\ncase %d:\n", nrules); if (!no_lines_flag) - fprintf (fguard, "#line %d \"%s\"\n", lineno, infile); - putc ('{', fguard); + obstack_fgrow2 (&guard_obstack, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); + obstack_1grow (&guard_obstack, '{'); count = 0; c = getc (finput); @@ -1152,18 +1184,18 @@ copy_guard (symbol_list *rule, int stack_offset) switch (c) { case '\n': - putc (c, fguard); + obstack_1grow (&guard_obstack, c); lineno++; break; case '{': - putc (c, fguard); + obstack_1grow (&guard_obstack, c); brace_flag = 1; count++; break; case '}': - putc (c, fguard); + obstack_1grow (&guard_obstack, c); if (count > 0) count--; else @@ -1175,26 +1207,26 @@ copy_guard (symbol_list *rule, int stack_offset) case '\'': case '"': - copy_string (finput, fguard, 0, c); + copy_string (finput, &guard_obstack, c); break; case '/': - copy_comment (finput, fguard, 0); + copy_comment (finput, &guard_obstack); break; case '$': - copy_dollar (finput, fguard, 0, rule, stack_offset); + copy_dollar (finput, &guard_obstack, rule, stack_offset); break; case '@': - copy_at (finput, fguard, 0, stack_offset); + copy_at (finput, &guard_obstack, rule, stack_offset); break; case EOF: fatal ("%s", _("unterminated %guard clause")); default: - putc (c, fguard); + obstack_1grow (&guard_obstack, c); } if (c != '}' || count != 0) @@ -1203,7 +1235,7 @@ copy_guard (symbol_list *rule, int stack_offset) c = skip_white_space (); - fprintf (fguard, ";\n break;}"); + obstack_sgrow (&guard_obstack, ";\n break;}"); if (c == '{') copy_action (rule, stack_offset); else if (c == '=') @@ -1217,20 +1249,6 @@ 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. | @@ -1239,9 +1257,14 @@ 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++; @@ -1259,12 +1282,12 @@ 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; @@ -1274,17 +1297,17 @@ 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) @@ -1293,7 +1316,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1314,7 +1337,7 @@ get_type (void) static void readgram (void) { - int t; + token_t t; bucket *lhs = NULL; symbol_list *p; symbol_list *p1; @@ -1330,9 +1353,9 @@ 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 */ @@ -1340,7 +1363,7 @@ readgram (void) int xactions = 0; /* JF for error checking */ bucket *first_rhs = 0; - if (t == IDENTIFIER) + if (t == tok_identifier) { lhs = symval; @@ -1351,14 +1374,14 @@ 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 */ @@ -1368,10 +1391,7 @@ readgram (void) nrules++; nitems++; - record_rule_line (); - - p = XCALLOC (symbol_list, 1); - p->sym = lhs; + p = symbol_list_new (lhs); crule1 = p1; if (p1) @@ -1398,28 +1418,28 @@ 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 */ @@ -1433,47 +1453,45 @@ 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; } @@ -1487,18 +1505,18 @@ 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")); @@ -1506,7 +1524,7 @@ 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) @@ -1531,7 +1549,7 @@ 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 @@ -1541,31 +1559,31 @@ 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 (); @@ -1574,7 +1592,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1594,13 +1612,13 @@ readgram (void) /* 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_literal_string (&attrs_obstack, - "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); + obstack_sgrow (&attrs_obstack, + "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); if (defines_flag) - obstack_grow_literal_string (&defines_obstack, "\ -#ifndef YYSTYPE\n\ -# define YYSTYPE int\n\ -#endif\n"); + obstack_sgrow (&defines_obstack, "\ +# ifndef YYSTYPE\n\ +# define YYSTYPE int\n\ +# endif\n"); } /* Report any undefined symbols and consider them nonterminals. */ @@ -1657,18 +1675,101 @@ 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'); } +/*--------------------. +| Output the header. | +`--------------------*/ + +static void +symbols_output (void) +{ + if (defines_flag) + { + output_token_defines (&defines_obstack); + + if (!pure_parser) + { + if (spec_name_prefix) + obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", + spec_name_prefix); + else + obstack_sgrow (&defines_obstack, + "\nextern YYSTYPE yylval;\n"); + } + + if (semantic_parser) + { + int 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; +#endif + } + } +} + + +/*------------------------------------------------------------------. +| 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 $undefined., which represents all invalid + inputs. */ + for (i = 0; i <= max_user_token_number; i++) + token_translations[i] = 2; + + for (bp = firstsymbol; bp; bp = bp->next) + { + /* Non-terminal? */ + if (bp->value >= ntokens) + continue; + /* A token string alias? */ + if (bp->user_token_number == SALIAS) + continue; + + assert (bp->user_token_number != SUNDEF); + + /* A token which translation has already been set? */ + if (token_translations[bp->user_token_number] != 2) + complain (_("tokens %s and %s both assigned number %d"), + tags[token_translations[bp->user_token_number]], + bp->tag, bp->user_token_number); + token_translations[bp->user_token_number] = bp->value; + } +} + + /*------------------------------------------------------------------. | Assign symbol numbers, and write definition of token names into | | FDEFINES. Set up vectors TAGS and SPREC of names and precedences | @@ -1678,22 +1779,21 @@ output_token_defines (struct obstack *oout) static void packsymbols (void) { - bucket *bp; + bucket *bp = NULL; int tokno = 1; - int i; int last_user_token_number; static char DOLLAR[] = "$"; - /* 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; @@ -1746,7 +1846,7 @@ 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; @@ -1756,34 +1856,9 @@ 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; @@ -1796,36 +1871,6 @@ packsymbols (void) fatal (_("the start symbol %s is a token"), startval->tag); start_symbol = startval->value; - - if (defines_flag) - { - output_token_defines (&defines_obstack); - - if (!pure_parser) - { - if (spec_name_prefix) - obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", - spec_name_prefix); - else - obstack_grow_literal_string (&defines_obstack, - "\nextern YYSTYPE yylval;\n"); - } - - 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); - } -#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; -#endif - } } @@ -1841,14 +1886,8 @@ 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; @@ -1856,9 +1895,11 @@ 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) @@ -1868,8 +1909,8 @@ 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; @@ -1879,9 +1920,9 @@ 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; @@ -1892,11 +1933,14 @@ packgram (void) } ritem[itemno] = 0; + + 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 FGUARD file | +| 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. | @@ -1908,33 +1952,20 @@ 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 symbol table. */ @@ -1952,7 +1983,6 @@ 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\ @@ -1960,7 +1990,7 @@ reader (void) no_parser_flag ? "Bison-generated parse tables" : "A Bison parser", infile, VERSION); - obstack_grow_literal_string (&table_obstack, + obstack_sgrow (&table_obstack, "#define YYBISON 1 /* Identify Bison output. */\n\n"); read_declarations (); /* Start writing the guard and action files, if they are needed. */ @@ -1975,34 +2005,36 @@ reader (void) /* Write closing delimiters for actions and guards. */ output_trailers (); if (locations_flag) - obstack_grow_literal_string (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); + obstack_sgrow (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); /* 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 (); + /* Output the headers. */ + symbols_output (); } +/*------------------------------------------------------------------. +| 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_literal_string (oout, "\ + obstack_sgrow (oout, "\ \n\ #ifndef YYLTYPE\n\ typedef struct yyltype\n\ {\n\ - int timestamp;\n\ int first_line;\n\ - int first_column;\ + int first_column;\n\ \n\ int last_line;\n\ int last_column;\n\ - char *text;\n\ } yyltype;\n\ \n\ # define YYLTYPE yyltype\n\