X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/8963a27bde6b9af57a795c44f680346016ad9666..e83d80b8ed8587fd05437d33f2011403f55f1f9d:/src/reader.c diff --git a/src/reader.c b/src/reader.c index cb297c4c..099ce309 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. @@ -21,62 +21,30 @@ #include "system.h" +#include "obstack.h" +#include "quotearg.h" +#include "quote.h" #include "getargs.h" #include "files.h" -#include "alloc.h" #include "symtab.h" +#include "options.h" #include "lex.h" #include "gram.h" #include "complain.h" #include "output.h" - -#define LTYPESTR "\ -\n\ -#ifndef YYLTYPE\n\ -typedef\n\ - 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\ - }\n\ -\ - yyltype;\n\ -\n\ -#define YYLTYPE yyltype\n\ -#endif\n\ -\n" - -/* Number of slots allocated (but not necessarily used yet) in `rline' */ -static int rline_allocated; - -extern bucket *symval; -extern int numval; -extern int expected_conflicts; -extern char *token_buffer; -extern int maxtoken; - -extern void tabinit PARAMS ((void)); -extern void free_symtab PARAMS ((void)); -extern char *printable_version PARAMS ((int)); +#include "reader.h" +#include "conflicts.h" +#include "muscle_tab.h" typedef struct symbol_list { struct symbol_list *next; bucket *sym; + int line; bucket *ruleprec; } symbol_list; - -extern void reader PARAMS ((void)); -extern void reader_output_yylsp PARAMS ((FILE *)); - int lineno; char **tags; short *user_toknums; @@ -88,17 +56,27 @@ static bucket *startval; they must be unions. */ static int value_components_used; -static int typed; /* nonzero if %union has been seen. */ +/* Nonzero if %union has been seen. */ +static int typed; -static int lastprec; /* incremented for each %left, %right or %nonassoc seen */ - -static int gensym_count; /* incremented for each generated symbol */ +/* Incremented for each %left, %right or %nonassoc seen */ +static int lastprec; static bucket *errtoken; static bucket *undeftoken; -/* Nonzero if any action or guard uses the @n construct. */ -static int yylsp_needed; + +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; +} + /*===================\ @@ -150,17 +128,53 @@ read_signed_integer (FILE *stream) return sign * n; } -/*-------------------------------------------------------------------. -| Dump the string from FINPUT to FOUTPUT. MATCH is the delimiter of | -| the string (either ' or "). | -`-------------------------------------------------------------------*/ +/*--------------------------------------------------------------. +| Get the data type (alternative in the union) of the value for | +| symbol N in rule RULE. | +`--------------------------------------------------------------*/ + +static char * +get_type_name (int n, symbol_list *rule) +{ + int i; + symbol_list *rp; + + if (n < 0) + { + complain (_("invalid $ value")); + return NULL; + } + + rp = rule; + i = 0; + + while (i < n) + { + rp = rp->next; + if (rp == NULL || rp->sym == NULL) + { + complain (_("invalid $ value")); + return NULL; + } + i++; + } + + return rp->sym->type_name; +} + +/*------------------------------------------------------------. +| 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) @@ -175,14 +189,15 @@ 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++; } @@ -190,27 +205,68 @@ 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_comment2 (FILE *in, FILE *out1, FILE *out2, int c) +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 *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, '/'); - cplus_comment = (c == '/'); - putc (c, out1); - if (out2) - putc (c, out2); - c = getc (in); + switch ((c = getc (fin))) + { + case '/': + cplus_comment = 1; + break; + case '*': + cplus_comment = 0; + break; + default: + ungetc (c, fin); + return; + } + + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); + c = getc (fin); ended = 0; while (!ended) @@ -219,73 +275,73 @@ 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"); - yylsp_needed = 1; + obstack_sgrow (oout, "yyloc"); + locations_flag = 1; } else if (isdigit (c) || c == '-') { @@ -294,11 +350,79 @@ copy_at (FILE *fin, FILE *fout, int stack_offset) ungetc (c, fin); n = read_signed_integer (fin); - fprintf (fout, "yylsp[%d]", n - stack_offset); - yylsp_needed = 1; + obstack_fgrow1 (oout, "yylsp[%d]", n - stack_offset); + locations_flag = 1; + } + else + { + char buf[] = "@c"; + buf[1] = c; + complain (_("%s is invalid"), quote (buf)); + } +} + + +/*-------------------------------------------------------------------. +| FIN is pointing to a wannabee semantic value (i.e., a `$'). | +| | +| Possible inputs: $[]($|integer) | +| | +| Output to OOUT a reference to this semantic value. STACK_OFFSET is | +| the number of values in the current rule so far, which says where | +| to find `$0' with respect to the top of the stack. | +`-------------------------------------------------------------------*/ + +static inline void +copy_dollar (FILE *fin, struct obstack *oout, + symbol_list *rule, int stack_offset) +{ + int c = getc (fin); + const char *type_name = NULL; + + /* Get the type name if explicit. */ + if (c == '<') + { + read_type_name (fin); + type_name = token_buffer; + value_components_used = 1; + c = getc (fin); + } + + if (c == '$') + { + obstack_sgrow (oout, "yyval"); + + if (!type_name) + type_name = get_type_name (0, rule); + if (type_name) + obstack_fgrow1 (oout, ".%s", type_name); + if (!type_name && typed) + complain (_("$$ of `%s' has no declared type"), + rule->sym->tag); + } + else if (isdigit (c) || c == '-') + { + int n; + ungetc (c, fin); + n = read_signed_integer (fin); + + if (!type_name && n > 0) + type_name = get_type_name (n, rule); + + obstack_fgrow1 (oout, "yyvsp[%d]", n - stack_offset); + + if (type_name) + obstack_fgrow1 (oout, ".%s", type_name); + if (!type_name && typed) + complain (_("$%d of `%s' has no declared type"), + n, rule->sym->tag); } else - complain (_("@%s is invalid"), printable_version (c)); + { + char buf[] = "$c"; + buf[1] = c; + complain (_("%s is invalid"), quote (buf)); + } } /*-------------------------------------------------------------------. @@ -313,8 +437,12 @@ copy_definition (void) /* -1 while reading a character if prev char was %. */ int after_percent; - if (!nolinesflag) - fprintf (fattrs, "#line %d \"%s\"\n", lineno, infile); + if (!no_lines_flag) + { + obstack_fgrow2 (&attrs_obstack, muscle_find ("linef"), + lineno, quotearg_style (c_quoting_style, + muscle_find("filename"))); + } after_percent = 0; @@ -325,7 +453,7 @@ copy_definition (void) switch (c) { case '\n': - putc (c, fattrs); + obstack_1grow (&attrs_obstack, c); lineno++; break; @@ -335,22 +463,18 @@ 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); @@ -359,67 +483,75 @@ copy_definition (void) { if (c == '}') return; - putc ('%', fattrs); + obstack_1grow (&attrs_obstack, '%'); } after_percent = 0; - } - } -/*-----------------------------------------------------------------. -| Parse what comes after %token or %nterm. For %token, what_is is | -| STOKEN and what_is_not is SNTERM. For %nterm, the arguments are | -| reversed. | -`-----------------------------------------------------------------*/ +/*-------------------------------------------------------------------. +| Parse what comes after %token or %nterm. For %token, WHAT_IS is | +| token_sym and WHAT_IS_NOT is nterm_sym. For %nterm, the arguments | +| are reversed. | +`-------------------------------------------------------------------*/ static void -parse_token_decl (int what_is, int what_is_not) +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 */ - int k; + 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) { - k = strlen (token_buffer); - typename = NEW2 (k + 1, char); - strcpy (typename, token_buffer); + 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) { - translations = 1; - symval->class = STOKEN; - symval->type_name = typename; - symval->user_token_number = symbol->user_token_number; - symbol->user_token_number = SALIAS; - - symval->alias = symbol; - symbol->alias = symval; - symbol = NULL; - - nsyms--; /* symbol and symval combined are only one symbol */ + if (symval->alias) + warn (_("symbol `%s' used more than once as a literal string"), + symval->tag); + else if (symbol->alias) + warn (_("symbol `%s' given more than one literal string"), + symbol->tag); + else + { + symval->class = token_sym; + symval->type_name = typename; + symval->user_token_number = symbol->user_token_number; + symbol->user_token_number = SALIAS; + symval->alias = symbol; + symbol->alias = symval; + /* symbol and symval combined are only one symbol */ + nsyms--; + } + symbol = NULL; } - else if (token == IDENTIFIER) + else if (token == tok_identifier) { int oldclass = symval->class; symbol = symval; @@ -427,7 +559,7 @@ parse_token_decl (int what_is, int what_is_not) if (symbol->class == what_is_not) complain (_("symbol %s redefined"), symbol->tag); symbol->class = what_is; - if (what_is == SNTERM && oldclass != SNTERM) + if (what_is == nterm_sym && oldclass != nterm_sym) symbol->value = nvars++; if (typename) @@ -438,15 +570,15 @@ parse_token_decl (int what_is, int 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 == STOKEN) ? "%token" : "%nterm"); + token_buffer, + (what_is == token_sym) ? "%token" : "%nterm"); skip_to_char ('%'); } } @@ -454,14 +586,16 @@ parse_token_decl (int what_is, int what_is_not) } -/* Parse what comes after %start */ +/*------------------------------. +| Parse what comes after %start | +`------------------------------*/ static void 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 { @@ -470,43 +604,6 @@ parse_start_decl (void) } } - - -/*--------------------------------------------------------------. -| Get the data type (alternative in the union) of the value for | -| symbol n in rule rule. | -`--------------------------------------------------------------*/ - -static char * -get_type_name (int n, symbol_list * rule) -{ - int i; - symbol_list *rp; - - if (n < 0) - { - complain (_("invalid $ value")); - return NULL; - } - - rp = rule; - i = 0; - - while (i < n) - { - rp = rp->next; - if (rp == NULL || rp->sym == NULL) - { - complain (_("invalid $ value")); - return NULL; - } - i++; - } - - return rp->sym->type_name; -} - - /*-----------------------------------------------------------. | read in a %type declaration and record its information for | | get_type_name to access | @@ -515,23 +612,20 @@ get_type_name (int n, symbol_list * rule) static void parse_type_decl (void) { - int k; char *name; - if (lex () != TYPENAME) + if (lex () != tok_typename) { complain ("%s", _("%type declaration has no ")); skip_to_char ('%'); return; } - k = strlen (token_buffer); - name = NEW2 (k + 1, char); - strcpy (name, token_buffer); + name = xstrdup (token_buffer); for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -544,11 +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) @@ -566,13 +660,14 @@ parse_type_decl (void) -/* read in a %left, %right or %nonassoc declaration and record its information. */ -/* assoc is either LEFT_ASSOC, RIGHT_ASSOC or NON_ASSOC. */ +/*----------------------------------------------------------------. +| Read in a %left, %right or %nonassoc declaration and record its | +| information. | +`----------------------------------------------------------------*/ static void -parse_assoc_decl (int assoc) +parse_assoc_decl (associativity assoc) { - int k; char *name = NULL; int prev = 0; @@ -580,7 +675,7 @@ parse_assoc_decl (int assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -592,24 +687,21 @@ parse_assoc_decl (int assoc) switch (t) { - - case TYPENAME: - k = strlen (token_buffer); - name = NEW2 (k + 1, char); - strcpy (name, token_buffer); + 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; symval->assoc = assoc; - if (symval->class == SNTERM) + if (symval->class == nterm_sym) complain (_("symbol %s redefined"), symval->tag); - symval->class = STOKEN; + symval->class = token_sym; if (name) { /* record the type, if one is specified */ if (symval->type_name == NULL) @@ -619,11 +711,10 @@ parse_assoc_decl (int assoc) } break; - case NUMBER: - if (prev == IDENTIFIER) + case tok_number: + if (prev == tok_identifier) { symval->user_token_number = numval; - translations = 1; } else { @@ -634,7 +725,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -643,43 +734,59 @@ 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; + 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 (!nolinesflag) - 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? */ + /* 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'); - 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, prologue); 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) { @@ -688,13 +795,9 @@ 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; @@ -705,49 +808,39 @@ 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, 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); } + } -/* parse the declaration %expect N which says to expect N - shift-reduce conflicts. */ + +/*-------------------------------------------------------. +| Parse the declaration %expect N which says to expect N | +| shift-reduce conflicts. | +`-------------------------------------------------------*/ static void parse_expect_decl (void) { - int c; - int count; - char buffer[20]; - - c = getc (finput); - while (c == ' ' || c == '\t') - c = getc (finput); - - count = 0; - while (c >= '0' && c <= '9') - { - if (count < 20) - buffer[count++] = c; - c = getc (finput); - } - buffer[count] = 0; - + int c = skip_white_space (); ungetc (c, finput); - if (count <= 0 || count > 10) - complain ("%s", _("argument of %expect is not an integer")); - expected_conflicts = atoi (buffer); + if (!isdigit (c)) + complain (_("argument of %%expect is not an integer")); + else + expected_conflicts = read_signed_integer (finput); } @@ -774,146 +867,241 @@ parse_expect_decl (void) static void parse_thong_decl (void) { - int token; + token_t token; struct bucket *symbol; char *typename = 0; - int k, usrtoknum; + int usrtoknum = SUNDEF; - translations = 1; token = lex (); /* fetch typename or first token */ - if (token == TYPENAME) + if (token == tok_typename) { - k = strlen (token_buffer); - typename = NEW2 (k + 1, char); - strcpy (typename, token_buffer); + typename = xstrdup (token_buffer); value_components_used = 1; token = lex (); /* fetch first token */ } /* process first token */ - if (token != IDENTIFIER) + if (token != tok_identifier) { complain (_("unrecognized item %s, expected an identifier"), token_buffer); skip_to_char ('%'); return; } - symval->class = STOKEN; + symval->class = token_sym; symval->type_name = typename; symval->user_token_number = SALIAS; symbol = symval; 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 ('%'); return; } - symval->class = STOKEN; + symval->class = token_sym; symval->type_name = typename; symval->user_token_number = usrtoknum; symval->alias = symbol; symbol->alias = symval; - nsyms--; /* symbol and symval combined are only one symbol */ + /* symbol and symval combined are only one symbol. */ + 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 | | 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: - parse_token_decl (STOKEN, SNTERM); + case tok_token: + parse_token_decl (token_sym, nterm_sym); break; - case NTERM: - parse_token_decl (SNTERM, STOKEN); + 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: - parse_assoc_decl (LEFT_ASSOC); + + case tok_left: + parse_assoc_decl (left_assoc); break; - case RIGHT: - parse_assoc_decl (RIGHT_ASSOC); + case tok_right: + parse_assoc_decl (right_assoc); break; - case NONASSOC: - parse_assoc_decl (NON_ASSOC); + 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 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 ('%'); @@ -923,7 +1111,9 @@ read_declarations (void) fatal (_("no input grammar")); else { - complain (_("unknown character: %s"), printable_version (c)); + char buf[] = "c"; + buf[0] = c; + complain (_("unknown character: %s"), quote (buf)); skip_to_char ('%'); } } @@ -937,21 +1127,24 @@ read_declarations (void) `-------------------------------------------------------------------*/ static void -copy_action (symbol_list * rule, int stack_offset) +copy_action (symbol_list *rule, int stack_offset) { int c; - int n; int count; - char *type_name; /* offset is always 0 if parser has already popped the stack pointer */ if (semantic_parser) stack_offset = 0; - fprintf (faction, "\ncase %d:\n", nrules); - if (!nolinesflag) - fprintf (faction, "#line %d \"%s\"\n", lineno, infile); - putc ('{', faction); + obstack_fgrow1 (&action_obstack, "\ncase %d:\n", nrules); + + if (!no_lines_flag) + { + 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); @@ -963,91 +1156,39 @@ 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 '$': - c = getc (finput); - type_name = NULL; - - if (c == '<') - { - char *cp = token_buffer; - - while ((c = getc (finput)) != '>' && c > 0) - { - if (cp == token_buffer + maxtoken) - cp = grow_token_buffer (cp); - - *cp++ = c; - } - *cp = 0; - type_name = token_buffer; - value_components_used = 1; - - c = getc (finput); - } - if (c == '$') - { - fprintf (faction, "yyval"); - if (!type_name) - type_name = get_type_name (0, rule); - if (type_name) - fprintf (faction, ".%s", type_name); - if (!type_name && typed) - complain (_("$$ of `%s' has no declared type"), - rule->sym->tag); - } - else if (isdigit (c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer (finput); - c = getc (finput); - - if (!type_name && n > 0) - type_name = get_type_name (n, rule); - - fprintf (faction, "yyvsp[%d]", n - stack_offset); - if (type_name) - fprintf (faction, ".%s", type_name); - if (!type_name && typed) - complain (_("$%d of `%s' has no declared type"), - n, rule->sym->tag); - continue; - } - else - complain (_("$%s is invalid"), printable_version (c)); - + 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); @@ -1057,12 +1198,12 @@ 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;}"); } /*-------------------------------------------------------------------. @@ -1075,22 +1216,22 @@ copy_action (symbol_list * rule, int stack_offset) `-------------------------------------------------------------------*/ static void -copy_guard (symbol_list * rule, int stack_offset) +copy_guard (symbol_list *rule, int stack_offset) { int c; - int n; int count; - char *type_name; int brace_flag = 0; /* offset is always 0 if parser has already popped the stack pointer */ if (semantic_parser) stack_offset = 0; - fprintf (fguard, "\ncase %d:\n", nrules); - if (!nolinesflag) - fprintf (fguard, "#line %d \"%s\"\n", lineno, infile); - putc ('{', fguard); + obstack_fgrow1 (&guard_obstack, "\ncase %d:\n", nrules); + if (!no_lines_flag) + 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); @@ -1100,18 +1241,18 @@ 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 @@ -1123,79 +1264,26 @@ 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 '$': - c = getc (finput); - type_name = NULL; - - if (c == '<') - { - char *cp = token_buffer; - - while ((c = getc (finput)) != '>' && c > 0) - { - if (cp == token_buffer + maxtoken) - cp = grow_token_buffer (cp); - - *cp++ = c; - } - *cp = 0; - type_name = token_buffer; - - c = getc (finput); - } - - if (c == '$') - { - fprintf (fguard, "yyval"); - if (!type_name) - type_name = rule->sym->type_name; - if (type_name) - fprintf (fguard, ".%s", type_name); - if (!type_name && typed) - complain (_("$$ of `%s' has no declared type"), - rule->sym->tag); - } - else if (isdigit (c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer (finput); - c = getc (finput); - - if (!type_name && n > 0) - type_name = get_type_name (n, rule); - - fprintf (fguard, "yyvsp[%d]", n - stack_offset); - if (type_name) - fprintf (fguard, ".%s", type_name); - if (!type_name && typed) - complain (_("$%d of `%s' has no declared type"), - n, rule->sym->tag); - continue; - } - else - complain (_("$%s is invalid"), printable_version (c)); + 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) @@ -1204,7 +1292,7 @@ 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 == '=') @@ -1218,21 +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 = (short *) xrealloc ((char *) rline, - rline_allocated * sizeof (short)); - } - rline[nrules] = lineno; -} - - /*-------------------------------------------------------------------. | Generate a dummy symbol, a nonterminal, whose name cannot conflict | | with the user's names. | @@ -1241,11 +1314,16 @@ 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 = SNTERM; + sym->class = nterm_sym; sym->value = nvars++; return sym; } @@ -1261,34 +1339,32 @@ 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; } - k = strlen (token_buffer); - name = NEW2 (k + 1, char); - strcpy (name, token_buffer); + name = xstrdup (token_buffer); 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) @@ -1297,7 +1373,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1318,30 +1394,33 @@ get_type (void) static void readgram (void) { - int t; + token_t t; bucket *lhs = NULL; symbol_list *p; symbol_list *p1; bucket *bp; - symbol_list *crule; /* points to first symbol_list of current rule. */ - /* its symbol is the lhs of the rule. */ - symbol_list *crule1; /* points to the symbol_list preceding crule. */ + /* Points to first symbol_list of current rule. its symbol is the + lhs of the rule. */ + symbol_list *crule; + /* Points to the symbol_list preceding crule. */ + symbol_list *crule1; p1 = NULL; 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 actionflag = 0; - int rulelength = 0; /* number of symbols in rhs of this rule so far */ + int action_flag = 0; + /* Number of symbols in rhs of this rule so far */ + int rulelength = 0; int xactions = 0; /* JF for error checking */ bucket *first_rhs = 0; - if (t == IDENTIFIER) + if (t == tok_identifier) { lhs = symval; @@ -1352,14 +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 */ @@ -1369,10 +1448,7 @@ readgram (void) nrules++; nitems++; - record_rule_line (); - - p = NEW (symbol_list); - p->sym = lhs; + p = symbol_list_new (lhs); crule1 = p1; if (p1) @@ -1385,13 +1461,13 @@ readgram (void) /* mark the rule's lhs as a nonterminal if not already so. */ - if (lhs->class == SUNKNOWN) + if (lhs->class == unknown_sym) { - lhs->class = SNTERM; + lhs->class = nterm_sym; lhs->value = nvars; nvars++; } - else if (lhs->class == STOKEN) + else if (lhs->class == token_sym) complain (_("rule given for %s, which is a token"), lhs->tag); /* read the rhs of the rule. */ @@ -1399,28 +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 */ @@ -1432,89 +1508,86 @@ readgram (void) /* If we just passed an action, that action was in the middle of a rule, so make a dummy rule to reduce it to a non-terminal. */ - if (actionflag) + 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. */ + /* 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 = NEW (symbol_list); + 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 = NEW (symbol_list); - p->next = crule1; + /* End of the rule. */ + crule1 = symbol_list_new (NULL); crule1->next = crule; - /* insert the dummy generated by that rule into this rule. */ + p->next = crule1; + + /* Insert the dummy generated by that rule into this + rule. */ nitems++; - p = NEW (symbol_list); - p->sym = sdummy; + p = symbol_list_new (sdummy); p1->next = p; p1 = p; - actionflag = 0; + action_flag = 0; } - if (t == IDENTIFIER) + if (t == tok_identifier) { nitems++; - p = NEW (symbol_list); - p->sym = symval; + p = symbol_list_new (symval); p1->next = p; p1 = p; } else /* handle an action. */ { copy_action (crule, rulelength); - actionflag = 1; + action_flag = 1; xactions++; /* JF */ } rulelength++; } /* end of read rhs of rule */ /* Put an empty link in the list to mark the end of this rule */ - p = NEW (symbol_list); + 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 ("%s", - _ - ("%guard present but %semantic_parser not specified")); + complain (_("%%guard present but %%semantic_parser not specified")); copy_guard (crule, rulelength); t = lex (); } - else if (t == LEFT_CURLY) + else if (t == tok_left_curly) { /* This case never occurs -wjh */ - if (actionflag) + if (action_flag) complain (_("two actions at end of one rule")); copy_action (crule, rulelength); - actionflag = 1; + action_flag = 1; xactions++; /* -wjh */ t = lex (); } @@ -1533,7 +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 @@ -1543,31 +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 (STOKEN, SNTERM); + parse_token_decl (token_sym, nterm_sym); t = lex (); } - else if (t == NTERM) + else if (t == tok_nterm) { - parse_token_decl (SNTERM, STOKEN); + 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 (); @@ -1576,7 +1649,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1589,31 +1662,39 @@ readgram (void) if (nrules == 0) fatal (_("no rules in the input grammar")); - if (typed == 0 /* JF put out same default YYSTYPE as YACC does */ - && !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) - if (bp->class == SUNKNOWN) + if (bp->class == unknown_sym) { complain (_ ("symbol %s is used, but is not defined as a token and has no rules"), -bp->tag); - bp->class = SNTERM; + bp->tag); + bp->class = nterm_sym; bp->value = nvars++; } 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 | @@ -1621,7 +1702,7 @@ bp->tag); `--------------------------------------------------------------*/ static void -output_token_defines (FILE *file) +output_token_defines (struct obstack *oout) { bucket *bp; char *cp, *symbol; @@ -1654,48 +1735,86 @@ output_token_defines (FILE *file) if (c != '\0') continue; - fprintf (file, "#define\t%s\t%d\n", symbol, - ((translations && !rawtoknumflag) - ? 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); + + /* Initialize all entries for literal tokens to 2, the internal + token number for $undefined., which represents all invalid + inputs. */ + for (i = 0; i <= max_user_token_number; i++) + token_translations[i] = 2; + + for (bp = firstsymbol; bp; bp = bp->next) + { + /* Non-terminal? */ + if (bp->value >= ntokens) + continue; + /* A token string alias? */ + if (bp->user_token_number == SALIAS) + continue; - putc ('\n', file); + assert (bp->user_token_number != SUNDEF); + + /* A token which translation has already been set? */ + if (token_translations[bp->user_token_number] != 2) + complain (_("tokens %s and %s both assigned number %d"), + tags[token_translations[bp->user_token_number]], + bp->tag, bp->user_token_number); + token_translations[bp->user_token_number] = bp->value; + } } /*------------------------------------------------------------------. | Assign symbol numbers, and write definition of token names into | -| fdefines. Set up vectors tags and sprec of names and precedences | +| FDEFINES. Set up vectors TAGS and SPREC of names and precedences | | of symbols. | `------------------------------------------------------------------*/ static void packsymbols (void) { - bucket *bp; + bucket *bp = NULL; int tokno = 1; - int i; int last_user_token_number; static char DOLLAR[] = "$"; - /* int lossage = 0; JF set but not used */ + tags = XCALLOC (char *, nsyms + 1); + user_toknums = XCALLOC (short, nsyms + 1); + + sprec = XCALLOC (short, nsyms); + sassoc = XCALLOC (short, nsyms); - tags = NEW2 (nsyms + 1, char *); + /* The EOF token. */ tags[0] = DOLLAR; - user_toknums = NEW2 (nsyms + 1, short); user_toknums[0] = 0; - sprec = NEW2 (nsyms, short); - sassoc = NEW2 (nsyms, short); - max_user_token_number = 256; last_user_token_number = 256; for (bp = firstsymbol; bp; bp = bp->next) { - if (bp->class == SNTERM) + if (bp->class == nterm_sym) { bp->value += ntokens; } @@ -1735,14 +1854,14 @@ packsymbols (void) continue; /* do not do processing below for SALIASs */ } - else /* bp->class == STOKEN */ + else /* bp->class == token_sym */ { bp->value = tokno++; } - if (bp->class == STOKEN) + 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; @@ -1752,73 +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 = NEW2 (max_user_token_number + 1, short); - - /* 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 (!noparserflag) - output_token_defines (ftable); - - if (startval->class == SUNKNOWN) + if (startval->class == unknown_sym) fatal (_("the start symbol %s is undefined"), startval->tag); - else if (startval->class == STOKEN) + else if (startval->class == token_sym) fatal (_("the start symbol %s is a token"), startval->tag); start_symbol = startval->value; +} - if (definesflag) + +/*-----------------------------------. +| 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 (fdefines); + 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) - fprintf (fdefines, "\nextern YYSTYPE %slval;\n", - spec_name_prefix); - else - fprintf (fdefines, "\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] != '@') - fprintf (fdefines, "#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 + } } } @@ -1835,14 +1943,8 @@ packgram (void) int ruleno; symbol_list *p; - bucket *ruleprec; - - ritem = NEW2 (nitems + 1, short); - rlhs = NEW2 (nrules, short) - 1; - rrhs = NEW2 (nrules, short) - 1; - rprec = NEW2 (nrules, short) - 1; - rprecsym = NEW2 (nrules, short) - 1; - rassoc = NEW2 (nrules, short) - 1; + ritem = XCALLOC (short, nitems + 1); + rule_table = XCALLOC (rule_t, nrules) - 1; itemno = 0; ruleno = 1; @@ -1850,9 +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) @@ -1860,10 +1964,10 @@ packgram (void) ritem[itemno++] = p->sym->value; /* A rule gets by default the precedence and associativity of the last token in it. */ - if (p->sym->class == STOKEN) + 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; @@ -1873,9 +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; @@ -1886,14 +1990,17 @@ 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 | -| 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 @@ -1902,85 +2009,58 @@ 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 = NEW2 (rline_allocated, short); typed = 0; lastprec = 0; - gensym_count = 0; - semantic_parser = 0; pure_parser = 0; - yylsp_needed = 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 = STOKEN; + 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 = STOKEN; + 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", noparserflag ? "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 (yylsp_needed) - 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 (); /* 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 (yylsp_needed) - fprintf (f, LTYPESTR); + /* Output the headers. */ + symbols_output (); }