X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/131e2fef683836d45262423bbdf004a014335aba..e3f1699f5554776a9211b410f3ebdaa6120e62df:/src/reader.c diff --git a/src/reader.c b/src/reader.c index cdacc071..03c0d311 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,5 +1,5 @@ /* Input parser for bison - Copyright (C) 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. @@ -21,17 +21,20 @@ #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 "options.h" #include "lex.h" #include "gram.h" #include "complain.h" #include "output.h" #include "reader.h" #include "conflicts.h" -#include "quote.h" +#include "muscle_tab.h" /* Number of slots allocated (but not necessarily used yet) in `rline' */ static int rline_allocated; @@ -61,9 +64,6 @@ 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; @@ -151,17 +151,19 @@ get_type_name (int n, symbol_list * rule) return rp->sym->type_name; } -/*-------------------------------------------------------------------. -| Dump the string from FINPUT to FOUTPUT. 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, int match) +copy_string2 (FILE *fin, struct obstack *oout, int match, int store) { int c; - putc (match, fout); + if (store) + obstack_1grow (oout, match); + c = getc (fin); while (c != match) @@ -176,14 +178,15 @@ copy_string (FILE *fin, FILE *fout, int match) continue; } - putc (c, fout); + obstack_1grow (oout, c); if (c == '\\') { c = getc (fin); if (c == EOF) fatal (_("unterminated string at end of file")); - putc (c, fout); + obstack_1grow (oout, c); + if (c == '\n') lineno++; } @@ -191,27 +194,68 @@ copy_string (FILE *fin, FILE *fout, int match) c = getc (fin); } - putc (c, fout); + if (store) + obstack_1grow (oout, c); } +/* FIXME. */ -/*---------------------------------------------------------------. -| Dump the comment from IN to OUT1 and OUT2. C is either `*' or | -| `/', depending upon the type of comments used. OUT2 might be | -| NULL. | -`---------------------------------------------------------------*/ +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 | +| 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 *in, FILE *out1, FILE *out2, int c) +copy_comment2 (FILE *fin, struct obstack *oout1, struct obstack *oout2) { int cplus_comment; int ended; + int c; + + /* We read a `/', output it. */ + obstack_1grow (oout1, '/'); + if (oout2) + obstack_1grow (oout2, '/'); + + switch ((c = getc (fin))) + { + case '/': + cplus_comment = 1; + break; + case '*': + cplus_comment = 0; + break; + default: + ungetc (c, fin); + return; + } - cplus_comment = (c == '/'); - putc (c, out1); - if (out2) - putc (c, out2); - c = getc (in); + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); + c = getc (fin); ended = 0; while (!ended) @@ -220,72 +264,72 @@ copy_comment2 (FILE *in, FILE *out1, FILE *out2, int c) { while (c == '*') { - putc (c, out1); - if (out2) - putc (c, out2); - c = getc (in); + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); + c = getc (fin); } if (c == '/') { - putc (c, out1); - if (out2) - putc (c, out2); + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); ended = 1; } } else if (c == '\n') { lineno++; - putc (c, out1); - if (out2) - putc (c, out2); + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); if (cplus_comment) ended = 1; else - c = getc (in); + c = getc (fin); } else if (c == EOF) fatal (_("unterminated comment")); else { - putc (c, out1); - if (out2) - putc (c, out2); - c = getc (in); + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); + c = getc (fin); } } } -/*------------------------------------------------------------. -| Dump the comment from FIN to FOUT. C is either `*' or `/', | -| depending upon the type of comments used. | -`------------------------------------------------------------*/ +/*-------------------------------------------------------------------. +| Dump the comment (actually the current string starting with a `/') | +| from FIN to OOUT. | +`-------------------------------------------------------------------*/ static inline void -copy_comment (FILE *fin, FILE *fout, int c) +copy_comment (FILE *fin, struct obstack *oout) { - copy_comment2 (fin, fout, NULL, c); + 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 `$0' with | | respect to the top of the stack. | `-----------------------------------------------------------------*/ static inline void -copy_at (FILE *fin, FILE *fout, int stack_offset) +copy_at (FILE *fin, struct obstack *oout, int stack_offset) { int c; c = getc (fin); if (c == '$') { - fprintf (fout, "yyloc"); + obstack_sgrow (oout, "yyloc"); locations_flag = 1; } else if (isdigit (c) || c == '-') @@ -295,7 +339,7 @@ copy_at (FILE *fin, FILE *fout, int stack_offset) ungetc (c, fin); n = read_signed_integer (fin); - fprintf (fout, "yylsp[%d]", n - stack_offset); + obstack_fgrow1 (oout, "yylsp[%d]", n - stack_offset); locations_flag = 1; } else @@ -312,44 +356,35 @@ copy_at (FILE *fin, FILE *fout, 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 `$0' with respect to the top of the stack. | `-------------------------------------------------------------------*/ static inline void -copy_dollar (FILE *fin, FILE *fout, +copy_dollar (FILE *fin, struct obstack *oout, symbol_list *rule, int stack_offset) { int c = getc (fin); - char *type_name = NULL; + const char *type_name = NULL; - /* Get the typename if explicit. */ + /* Get the type name if explicit. */ if (c == '<') { - char *cp = token_buffer; - - while ((c = getc (fin)) != '>' && c > 0) - { - if (cp == token_buffer + maxtoken) - cp = grow_token_buffer (cp); - - *cp++ = c; - } - *cp = 0; + read_type_name (fin); type_name = token_buffer; value_components_used = 1; - c = getc (fin); } if (c == '$') { - fprintf (fout, "yyval"); + obstack_sgrow (oout, "yyval"); + if (!type_name) type_name = get_type_name (0, rule); if (type_name) - fprintf (fout, ".%s", type_name); + obstack_fgrow1 (oout, ".%s", type_name); if (!type_name && typed) complain (_("$$ of `%s' has no declared type"), rule->sym->tag); @@ -363,9 +398,10 @@ copy_dollar (FILE *fin, FILE *fout, if (!type_name && n > 0) type_name = get_type_name (n, rule); - fprintf (fout, "yyvsp[%d]", n - stack_offset); + obstack_fgrow1 (oout, "yyvsp[%d]", n - stack_offset); + if (type_name) - fprintf (fout, ".%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); @@ -390,8 +426,14 @@ copy_definition (void) /* -1 while reading a character if prev char was %. */ int after_percent; +#if 0 if (!no_lines_flag) - fprintf (fattrs, "#line %d \"%s\"\n", lineno, infile); + { + obstack_fgrow2 (&attrs_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find("filename"))); + } +#endif after_percent = 0; @@ -402,7 +444,7 @@ copy_definition (void) switch (c) { case '\n': - putc (c, fattrs); + obstack_1grow (&attrs_obstack, c); lineno++; break; @@ -412,22 +454,18 @@ copy_definition (void) case '\'': case '"': - copy_string (finput, fattrs, c); + copy_string (finput, &attrs_obstack, c); break; case '/': - putc (c, fattrs); - c = getc (finput); - if (c != '*' && c != '/') - continue; - copy_comment (finput, fattrs, c); + copy_comment (finput, &attrs_obstack); break; case EOF: fatal ("%s", _("unterminated `%{' definition")); default: - putc (c, fattrs); + obstack_1grow (&attrs_obstack, c); } c = getc (finput); @@ -436,12 +474,10 @@ copy_definition (void) { if (c == '}') return; - putc ('%', fattrs); + obstack_1grow (&attrs_obstack, '%'); } after_percent = 0; - } - } @@ -454,32 +490,38 @@ copy_definition (void) static void parse_token_decl (symbol_class what_is, symbol_class what_is_not) { - int token = 0; - char *typename = 0; - struct bucket *symbol = NULL; /* pts to symbol being defined */ + token_t token = tok_undef; + char *typename = NULL; + /* The symbol being defined. */ + struct bucket *symbol = NULL; + + /* After `%token' and `%nterm', any number of symbols maybe be + defined. */ for (;;) { int tmp_char = ungetc (skip_white_space (), finput); + /* `%' (for instance from `%token', or from `%%' etc.) is the + only valid means to end this declaration. */ if (tmp_char == '%') return; if (tmp_char == EOF) 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"), @@ -498,10 +540,9 @@ 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; @@ -520,10 +561,9 @@ 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 { @@ -545,7 +585,7 @@ 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 { @@ -564,7 +604,7 @@ parse_type_decl (void) { char *name; - if (lex () != TYPENAME) + if (lex () != tok_typename) { complain ("%s", _("%type declaration has no ")); skip_to_char ('%'); @@ -575,7 +615,7 @@ parse_type_decl (void) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -588,11 +628,11 @@ 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) @@ -625,7 +665,7 @@ parse_assoc_decl (associativity assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -637,14 +677,14 @@ 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; @@ -661,11 +701,10 @@ 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 { @@ -676,7 +715,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -685,45 +724,50 @@ token_buffer); } prev = t; - } } -/*-------------------------------------------------------------------. -| Copy the union declaration into fattrs (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 parse_union_decl (void) { int c; int count = 0; + struct obstack union_obstack; if (typed) complain (_("multiple %s declarations"), "%union"); typed = 1; - if (!no_lines_flag) - fprintf (fattrs, "\n#line %d \"%s\"\n", lineno, infile); - else - fprintf (fattrs, "\n"); + /* FIXME: I'm worried: are you sure attrs_obstack is properly + filled? */ + if (no_lines_flag) + obstack_1grow (&attrs_obstack, '\n'); - fprintf (fattrs, "typedef union"); - if (fdefines) - fprintf (fdefines, "typedef union"); + obstack_init (&union_obstack); + obstack_sgrow (&union_obstack, "union"); + if (defines_flag) + obstack_sgrow (&defines_obstack, "typedef union"); c = getc (finput); while (c != EOF) { - putc (c, fattrs); - if (fdefines) - putc (c, fdefines); + /* 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) { @@ -732,13 +776,9 @@ parse_union_decl (void) break; case '/': - c = getc (finput); - if (c != '*' && c != '/') - continue; - copy_comment2 (finput, fattrs, fdefines, c); + copy_comment2 (finput, &defines_obstack, &union_obstack); break; - case '{': count++; break; @@ -749,19 +789,21 @@ parse_union_decl (void) count--; if (count <= 0) { - fprintf (fattrs, " YYSTYPE;\n"); - if (fdefines) - fprintf (fdefines, " YYSTYPE;\n"); + if (defines_flag) + obstack_sgrow (&defines_obstack, " YYSTYPE;\n"); /* 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); } + } @@ -806,14 +848,13 @@ 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; @@ -822,7 +863,7 @@ parse_thong_decl (void) /* process first token */ - if (token != IDENTIFIER) + if (token != tok_identifier) { complain (_("unrecognized item %s, expected an identifier"), token_buffer); @@ -836,17 +877,15 @@ 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 ('%'); @@ -863,88 +902,139 @@ 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 what comes after %skeleton. | +`----------------------------------*/ + +void +parse_skel_decl (void) +{ + /* Complete with parse_dquoted_param () on the CVS branch 1.29. */ +} /*----------------------------------------------------------------. | Read from finput until `%%' is seen. Discard the `%%'. Handle | | any `%' declarations, and copy the contents of any `%{ ... %}' | -| groups to fattrs. | +| groups to ATTRS_OBSTACK. | `----------------------------------------------------------------*/ 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 (); - } + case tok_define: + parse_muscle_decl (); + break; + + case tok_skel: + parse_skel_decl (); break; - case PURE_PARSER: - pure_parser = 1; + case tok_noop: break; - case NOOP: + case tok_stropt: + case tok_intopt: + case tok_obsolete: + case tok_illegal: + abort (); break; default: @@ -976,15 +1066,21 @@ 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; - fprintf (faction, "\ncase %d:\n", nrules); + obstack_fgrow1 (&action_obstack, "\ncase %d:\n", nrules); + if (!no_lines_flag) - fprintf (faction, "#line %d \"%s\"\n", lineno, infile); - putc ('{', faction); + { + obstack_fgrow2 (&action_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find ("filename"))); + } + obstack_1grow (&action_obstack, '{'); count = 1; c = getc (finput); @@ -996,41 +1092,39 @@ copy_action (symbol_list *rule, int stack_offset) switch (c) { case '\n': - putc (c, faction); + obstack_1grow (&action_obstack, c); lineno++; break; case '{': - putc (c, faction); + obstack_1grow (&action_obstack, c); count++; break; case '\'': case '"': - copy_string (finput, faction, c); + copy_string (finput, &action_obstack, c); break; case '/': - putc (c, faction); - c = getc (finput); - if (c != '*' && c != '/') - continue; - copy_comment (finput, faction, c); + copy_comment (finput, &action_obstack); break; case '$': - copy_dollar (finput, faction, rule, stack_offset); + copy_dollar (finput, &action_obstack, + rule, stack_offset); break; case '@': - copy_at (finput, faction, stack_offset); + copy_at (finput, &action_obstack, + stack_offset); break; case EOF: fatal (_("unmatched %s"), "`{'"); default: - putc (c, faction); + obstack_1grow (&action_obstack, c); } c = getc (finput); @@ -1040,12 +1134,12 @@ copy_action (symbol_list *rule, int stack_offset) if (--count) { - putc (c, faction); + obstack_1grow (&action_obstack, c); c = getc (finput); } } - fprintf (faction, ";\n break;}"); + obstack_sgrow (&action_obstack, ";\n break;}"); } /*-------------------------------------------------------------------. @@ -1068,10 +1162,12 @@ 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, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find ("filename"))); + obstack_1grow (&guard_obstack, '{'); count = 0; c = getc (finput); @@ -1081,18 +1177,18 @@ 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 @@ -1104,30 +1200,26 @@ copy_guard (symbol_list *rule, int stack_offset) case '\'': case '"': - copy_string (finput, fguard, c); + copy_string (finput, &guard_obstack, c); break; case '/': - putc (c, fguard); - c = getc (finput); - if (c != '*' && c != '/') - continue; - copy_comment (finput, fguard, c); + copy_comment (finput, &guard_obstack); break; case '$': - copy_dollar (finput, fguard, rule, stack_offset); + copy_dollar (finput, &guard_obstack, rule, stack_offset); break; case '@': - copy_at (finput, fguard, stack_offset); + copy_at (finput, &guard_obstack, stack_offset); break; case EOF: fatal ("%s", _("unterminated %guard clause")); default: - putc (c, fguard); + obstack_1grow (&guard_obstack, c); } if (c != '}' || count != 0) @@ -1136,7 +1228,7 @@ 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 == '=') @@ -1172,9 +1264,14 @@ 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++; @@ -1192,12 +1289,12 @@ 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; @@ -1207,17 +1304,17 @@ 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) @@ -1226,7 +1323,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1247,7 +1344,7 @@ get_type (void) static void readgram (void) { - int t; + token_t t; bucket *lhs = NULL; symbol_list *p; symbol_list *p1; @@ -1263,9 +1360,9 @@ 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 */ @@ -1273,7 +1370,7 @@ readgram (void) int xactions = 0; /* JF for error checking */ bucket *first_rhs = 0; - if (t == IDENTIFIER) + if (t == tok_identifier) { lhs = symval; @@ -1284,14 +1381,14 @@ 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 */ @@ -1331,28 +1428,28 @@ 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 */ @@ -1368,9 +1465,9 @@ readgram (void) { 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. */ + /* 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 (); @@ -1391,7 +1488,8 @@ readgram (void) p->next = crule1; crule1->next = crule; - /* insert the dummy generated by that rule into this rule. */ + /* Insert the dummy generated by that rule into this + rule. */ nitems++; p = XCALLOC (symbol_list, 1); p->sym = sdummy; @@ -1401,7 +1499,7 @@ readgram (void) action_flag = 0; } - if (t == IDENTIFIER) + if (t == tok_identifier) { nitems++; p = XCALLOC (symbol_list, 1); @@ -1423,14 +1521,14 @@ readgram (void) 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")); @@ -1438,7 +1536,7 @@ 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) @@ -1463,7 +1561,7 @@ 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 @@ -1473,31 +1571,31 @@ 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 (); @@ -1506,7 +1604,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1519,18 +1617,6 @@ 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. */ - fprintf (fattrs, "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); - if (fdefines) - fprintf (fdefines, "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); - } - /* Report any undefined symbols and consider them nonterminals. */ for (bp = firstsymbol; bp; bp = bp->next) @@ -1545,6 +1631,25 @@ 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 | @@ -1552,7 +1657,7 @@ readgram (void) `--------------------------------------------------------------*/ static void -output_token_defines (FILE *file) +output_token_defines (struct obstack *oout) { bucket *bp; char *cp, *symbol; @@ -1585,14 +1690,53 @@ output_token_defines (FILE *file) if (c != '\0') continue; - fprintf (file, "#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) - fprintf (file, "#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); } +} + + +/*------------------------------------------------------------------. +| 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); - putc ('\n', file); + /* Initialize all entries for literal tokens to 2, the internal + token number for $undefined., which represents all invalid + inputs. */ + for (i = 0; i <= max_user_token_number; i++) + token_translations[i] = 2; + + for (bp = firstsymbol; bp; bp = bp->next) + { + /* Non-terminal? */ + if (bp->value >= ntokens) + continue; + /* A token string alias? */ + if (bp->user_token_number == SALIAS) + continue; + + assert (bp->user_token_number != SUNDEF); + + /* A token which translation has already been set? */ + if (token_translations[bp->user_token_number] != 2) + complain (_("tokens %s and %s both assigned number %d"), + tags[token_translations[bp->user_token_number]], + bp->tag, bp->user_token_number); + token_translations[bp->user_token_number] = bp->value; + } } @@ -1605,22 +1749,21 @@ output_token_defines (FILE *file) static void packsymbols (void) { - bucket *bp; + bucket *bp = NULL; int tokno = 1; - int i; int last_user_token_number; static char DOLLAR[] = "$"; - /* 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; @@ -1673,7 +1816,7 @@ 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; @@ -1683,73 +1826,74 @@ 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 (ftable); - 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 0 + if (!no_parser_flag) + output_token_defines (&table_obstack); +#endif if (defines_flag) { - output_token_defines (fdefines); + output_token_defines (&defines_obstack); if (!pure_parser) { if (spec_name_prefix) - fprintf (fdefines, "\nextern YYSTYPE %slval;\n", - spec_name_prefix); + obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n", + spec_name_prefix); else - fprintf (fdefines, "\nextern YYSTYPE yylval;\n"); + obstack_sgrow (&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] != '@') - fprintf (fdefines, "#define\tNT%s\t%d\n", tags[i], i); - } + { + 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; + /* `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 + } } } @@ -1821,10 +1965,10 @@ packgram (void) /*-------------------------------------------------------------------. | Read in the grammar specification and record it in the format | -| described in gram.h. All guards are copied into the FGUARD file | -| and all actions into FACTION, 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. | +| 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. | `-------------------------------------------------------------------*/ void @@ -1833,15 +1977,6 @@ 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; @@ -1852,83 +1987,49 @@ reader (void) 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."); undeftoken->class = token_sym; undeftoken->user_token_number = 2; - /* Read the declaration section. Copy %{ ... %} groups to FTABLE - and FDEFINES file. Also notice any %token, %left, etc. found - there. */ - putc ('\n', ftable); - fprintf (ftable, "\ -/* %s, made from %s\n\ - by GNU bison %s. */\n\ -\n", no_parser_flag ? "Bison-generated parse tables" : "A Bison parser", infile, VERSION); - - fputs ("#define YYBISON 1 /* Identify Bison output. */\n\n", ftable); + /* Read the declaration section. Copy %{ ... %} groups to + TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left, + etc. found there. */ 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 (fdefines) - reader_output_yylsp (fdefines); - /* Write closing delimiters for actions and guards. */ - output_trailers (); - if (locations_flag) - fputs ("#define YYLSP_NEEDED\n\n", ftable); + 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 (); + symbols_output (); /* 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 (); -} - - -void -reader_output_yylsp (FILE *f) -{ - if (locations_flag) - fputs ("\ -\n\ -#ifndef YYLTYPE\n\ -typedef struct yyltype\n\ -{\n\ - int timestamp;\n\ - int first_line;\n\ - int first_column;\ -\n\ - int last_line;\n\ - int last_column;\n\ - char *text;\n\ -} yyltype;\n\ -\n\ -# define YYLTYPE yyltype\n\ -#endif\n\ -\n", - f); }