X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/d7020c20785f7684de4f71ede4fe871fc16f61e7..dd3127cf7a8534b1689d19452e310138d8db784f:/src/reader.c?ds=sidebyside diff --git a/src/reader.c b/src/reader.c index bd13e044..7e23c0ef 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 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -21,6 +21,9 @@ #include "system.h" +#include "obstack.h" +#include "quotearg.h" +#include "quote.h" #include "getargs.h" #include "files.h" #include "xalloc.h" @@ -32,8 +35,6 @@ #include "reader.h" #include "conflicts.h" -extern char *printable_version PARAMS ((int)); - /* Number of slots allocated (but not necessarily used yet) in `rline' */ static int rline_allocated; @@ -62,14 +63,8 @@ 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; - -/* Nonzero if any action or guard uses the @n construct. */ -static int yylsp_needed; /*===================\ @@ -121,17 +116,52 @@ 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_string (FILE *fin, struct obstack *oout, int match) { int c; - putc (match, fout); + obstack_1grow (oout, match); + c = getc (fin); while (c != match) @@ -146,14 +176,15 @@ 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++; } @@ -161,27 +192,47 @@ copy_string (FILE *fin, FILE *fout, int match) c = getc (fin); } - putc (c, fout); + obstack_1grow (oout, c); } -/*---------------------------------------------------------------. -| Dump the comment from IN to OUT1 and OUT2. C is either `*' or | -| `/', depending upon the type of comments used. OUT2 might be | -| NULL. | -`---------------------------------------------------------------*/ +/*-----------------------------------------------------------------. +| Dump the wannabee comment from IN to OUT1 and OUT2 (which can be | +| NULL). In fact we just saw a `/', which might or might not be a | +| comment. In any case, copy what we saw. | +| | +| OUT2 might be NULL. | +`-----------------------------------------------------------------*/ static inline void -copy_comment2 (FILE *in, FILE *out1, FILE *out2, int c) +copy_comment2 (FILE *fin, struct obstack *oout1, struct obstack *oout2) { int cplus_comment; int ended; + int c; - cplus_comment = (c == '/'); - putc (c, out1); - if (out2) - putc (c, out2); - c = getc (in); + /* 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; + } + + obstack_1grow (oout1, c); + if (oout2) + obstack_1grow (oout2, c); + c = getc (fin); ended = 0; while (!ended) @@ -190,73 +241,73 @@ 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 == '-') { @@ -265,11 +316,79 @@ 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 - complain (_("@%s is invalid"), printable_version (c)); + { + 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 + { + char buf[] = "$c"; + buf[1] = c; + complain (_("%s is invalid"), quote (buf)); + } } /*-------------------------------------------------------------------. @@ -284,8 +403,9 @@ 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, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); after_percent = 0; @@ -296,7 +416,7 @@ copy_definition (void) switch (c) { case '\n': - putc (c, fattrs); + obstack_1grow (&attrs_obstack, c); lineno++; break; @@ -306,22 +426,18 @@ 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); @@ -330,12 +446,10 @@ copy_definition (void) { if (c == '}') return; - putc ('%', fattrs); + obstack_1grow (&attrs_obstack, '%'); } after_percent = 0; - } - } @@ -348,49 +462,60 @@ copy_definition (void) static void parse_token_decl (symbol_class what_is, symbol_class what_is_not) { - int token = 0; + token_t token = 0; char *typename = 0; - struct bucket *symbol = NULL; /* pts to symbol being defined */ - int k; + /* 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 = XCALLOC (char, k + 1); - 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) { + 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--; + } translations = 1; - 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 = NULL; - - nsyms--; /* symbol and symval combined are only one symbol */ + symbol = NULL; } - else if (token == IDENTIFIER) + else if (token == tok_identifier) { int oldclass = symval->class; symbol = symval; @@ -409,7 +534,7 @@ 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; @@ -434,7 +559,7 @@ 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 { @@ -443,43 +568,6 @@ 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 | @@ -488,23 +576,20 @@ 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 = XCALLOC (char, k + 1); - 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 == '%') @@ -517,11 +602,11 @@ 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) @@ -547,7 +632,6 @@ parse_type_decl (void) static void parse_assoc_decl (associativity assoc) { - int k; char *name = NULL; int prev = 0; @@ -555,7 +639,7 @@ parse_assoc_decl (associativity assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -567,17 +651,14 @@ parse_assoc_decl (associativity assoc) switch (t) { - - case TYPENAME: - k = strlen (token_buffer); - name = XCALLOC (char, k + 1); - 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; @@ -594,8 +675,8 @@ 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; @@ -609,7 +690,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -624,11 +705,11 @@ token_buffer); -/*-------------------------------------------------------------------. -| 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 ATTRS_OBSTACK (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) @@ -641,22 +722,23 @@ parse_union_decl (void) typed = 1; - if (!nolinesflag) - fprintf (fattrs, "\n#line %d \"%s\"\n", lineno, infile); + if (!no_lines_flag) + obstack_fgrow2 (&attrs_obstack, "\n#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); else - fprintf (fattrs, "\n"); + obstack_1grow (&attrs_obstack, '\n'); - fprintf (fattrs, "typedef union"); - if (fdefines) - fprintf (fdefines, "typedef union"); + obstack_sgrow (&attrs_obstack, "typedef union"); + if (defines_flag) + obstack_sgrow (&defines_obstack, "typedef union"); c = getc (finput); while (c != EOF) { - putc (c, fattrs); - if (fdefines) - putc (c, fdefines); + obstack_1grow (&attrs_obstack, c); + if (defines_flag) + obstack_1grow (&defines_obstack, c); switch (c) { @@ -665,13 +747,9 @@ parse_union_decl (void) break; case '/': - c = getc (finput); - if (c != '*' && c != '/') - continue; - copy_comment2 (finput, fattrs, fdefines, c); + copy_comment2 (finput, &defines_obstack, &attrs_obstack); break; - case '{': count++; break; @@ -682,9 +760,9 @@ parse_union_decl (void) count--; if (count <= 0) { - fprintf (fattrs, " YYSTYPE;\n"); - if (fdefines) - fprintf (fdefines, " YYSTYPE;\n"); + obstack_sgrow (&attrs_obstack, " YYSTYPE;\n"); + if (defines_flag) + obstack_sgrow (&defines_obstack, " YYSTYPE;\n"); /* JF don't choke on trailing semi */ c = skip_white_space (); if (c != ';') @@ -706,28 +784,13 @@ parse_union_decl (void) 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); } @@ -754,25 +817,23 @@ 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; translations = 1; token = lex (); /* fetch typename or first token */ - if (token == TYPENAME) + if (token == tok_typename) { - k = strlen (token_buffer); - typename = XCALLOC (char, k + 1); - 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); @@ -786,7 +847,7 @@ 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 */ @@ -796,7 +857,7 @@ parse_thong_decl (void) /* 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 ('%'); @@ -809,14 +870,44 @@ parse_thong_decl (void) symval->alias = symbol; symbol->alias = symval; - nsyms--; /* symbol and symval combined are only one symbol */ + /* symbol and symval combined are only one symbol. */ + nsyms--; } +/*------------------------------------------. +| Parse what comes after %header_extension. | +`------------------------------------------*/ + +static void +parse_header_extension_decl (void) +{ + char buff[32]; + + if (header_extension) + complain (_("multiple %%header_extension declarations")); + fscanf (finput, "%s", buff); + header_extension = xstrdup (buff); +} + +/*------------------------------------------. +| Parse what comes after %source_extension. | +`------------------------------------------*/ + +static void +parse_source_extension_decl (void) +{ + char buff[32]; + + if (src_extension) + complain (_("multiple %%source_extension declarations")); + fscanf (finput, "%s", buff); + src_extension = xstrdup (buff); +} /*----------------------------------------------------------------. | 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 @@ -835,65 +926,62 @@ read_declarations (void) 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_hdrext: + parse_header_extension_decl (); break; - case PURE_PARSER: - pure_parser = 1; + case tok_srcext: + parse_source_extension_decl (); break; - case NOOP: + case tok_noop: break; default: @@ -905,7 +993,9 @@ 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 ('%'); } } @@ -919,21 +1009,26 @@ 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; + 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); - if (!nolinesflag) - fprintf (faction, "#line %d \"%s\"\n", lineno, infile); - putc ('{', faction); + sprintf (buf, "\ncase %d:\n", nrules); + obstack_grow (&action_obstack, buf, strlen (buf)); + + if (!no_lines_flag) + { + sprintf (buf, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); + obstack_grow (&action_obstack, buf, strlen (buf)); + } + obstack_1grow (&action_obstack, '{'); count = 1; c = getc (finput); @@ -945,91 +1040,39 @@ 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); @@ -1039,12 +1082,12 @@ 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;}"); } /*-------------------------------------------------------------------. @@ -1057,22 +1100,21 @@ 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, "#line %d %s\n", + lineno, quotearg_style (c_quoting_style, infile)); + obstack_1grow (&guard_obstack, '{'); count = 0; c = getc (finput); @@ -1082,18 +1124,18 @@ 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 @@ -1105,79 +1147,26 @@ 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) @@ -1186,7 +1175,7 @@ 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 == '=') @@ -1222,9 +1211,14 @@ 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++; @@ -1242,34 +1236,32 @@ 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 = XCALLOC (char, k + 1); - 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) @@ -1278,7 +1270,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1299,30 +1291,33 @@ 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; @@ -1333,14 +1328,14 @@ 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 */ @@ -1380,28 +1375,28 @@ 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 */ @@ -1413,13 +1408,13 @@ 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 (); @@ -1440,17 +1435,18 @@ 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; p1->next = p; p1 = p; - actionflag = 0; + action_flag = 0; } - if (t == IDENTIFIER) + if (t == tok_identifier) { nitems++; p = XCALLOC (symbol_list, 1); @@ -1461,7 +1457,7 @@ readgram (void) else /* handle an action. */ { copy_action (crule, rulelength); - actionflag = 1; + action_flag = 1; xactions++; /* JF */ } rulelength++; @@ -1472,30 +1468,28 @@ 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 ("%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 (); } @@ -1514,7 +1508,7 @@ 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 @@ -1524,31 +1518,31 @@ 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 (); @@ -1557,7 +1551,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1570,15 +1564,20 @@ readgram (void) if (nrules == 0) fatal (_("no rules in the input grammar")); - if (typed == 0 /* JF put out same default YYSTYPE as YACC does */ + /* 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"); + obstack_sgrow (&attrs_obstack, + "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n"); + if (defines_flag) + obstack_sgrow (&defines_obstack, "\ +# ifndef YYSTYPE\n\ +# define YYSTYPE int\n\ +# endif\n"); } /* Report any undefined symbols and consider them nonterminals. */ @@ -1588,7 +1587,7 @@ readgram (void) { complain (_ ("symbol %s is used, but is not defined as a token and has no rules"), -bp->tag); + bp->tag); bp->class = nterm_sym; bp->value = nvars++; } @@ -1602,7 +1601,7 @@ bp->tag); `--------------------------------------------------------------*/ static void -output_token_defines (FILE *file) +output_token_defines (struct obstack *oout) { bucket *bp; char *cp, *symbol; @@ -1635,14 +1634,14 @@ 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, + (translations ? bp->user_token_number : bp->value)); if (semantic_parser) - fprintf (file, "#define\tT%s\t%d\n", symbol, bp->value); + obstack_fgrow2 (oout, "# define\tT%s\t%d\n", symbol, bp->value); } - putc ('\n', file); + obstack_1grow (oout, '\n'); } @@ -1764,8 +1763,8 @@ packsymbols (void) error_token_number = errtoken->value; - if (!noparserflag) - output_token_defines (ftable); + if (!no_parser_flag) + output_token_defines (&table_obstack); if (startval->class == unknown_sym) fatal (_("the start symbol %s is undefined"), startval->tag); @@ -1774,17 +1773,18 @@ packsymbols (void) start_symbol = startval->value; - if (definesflag) + 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) @@ -1792,7 +1792,8 @@ packsymbols (void) { /* don't make these for dummy nonterminals made by gensym. */ if (*tags[i] != '@') - fprintf (fdefines, "#define\tNT%s\t%d\n", tags[i], 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 @@ -1871,10 +1872,10 @@ 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 @@ -1902,11 +1903,8 @@ reader (void) typed = 0; lastprec = 0; - gensym_count = 0; - semantic_parser = 0; pure_parser = 0; - yylsp_needed = 0; grammar = NULL; @@ -1925,16 +1923,19 @@ reader (void) 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, "\ + /* Read the declaration section. Copy %{ ... %} groups to + TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left, + etc. found there. */ + obstack_1grow (&table_obstack, '\n'); + obstack_fgrow3 (&table_obstack, "\ /* %s, made from %s\n\ by GNU bison %s. */\n\ -\n", noparserflag ? "Bison-generated parse tables" : "A Bison parser", infile, VERSION); +\n", + no_parser_flag ? "Bison-generated parse tables" : "A Bison parser", + infile, VERSION); - fputs ("#define YYBISON 1 /* Identify Bison output. */\n\n", ftable); + obstack_sgrow (&table_obstack, + "#define YYBISON 1 /* Identify Bison output. */\n\n"); read_declarations (); /* Start writing the guard and action files, if they are needed. */ output_headers (); @@ -1943,12 +1944,12 @@ reader (void) readgram (); /* 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); + if (defines_flag) + reader_output_yylsp (&defines_obstack); /* Write closing delimiters for actions and guards. */ output_trailers (); - if (yylsp_needed) - fputs ("#define YYLSP_NEEDED\n\n", ftable); + if (locations_flag) + obstack_sgrow (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); /* Assign the symbols their symbol numbers. Write #defines for the token symbols into FDEFINES if requested. */ packsymbols (); @@ -1960,30 +1961,28 @@ reader (void) } +/*------------------------------------------------------------------. +| Define YYLTYPE. Cannot be in the skeleton since we might have to | +| output it in the headers if --defines is used. | +`------------------------------------------------------------------*/ + void -reader_output_yylsp (FILE *f) +reader_output_yylsp (struct obstack *oout) { - if (yylsp_needed) - fputs ("\ + if (locations_flag) + obstack_sgrow (oout, "\ \n\ #ifndef YYLTYPE\n\ -typedef\n\ - struct yyltype\n\ -\ - {\n\ - int timestamp;\n\ - int first_line;\n\ - int first_column;\ +typedef struct yyltype\n\ +{\n\ + int first_line;\n\ + int first_column;\n\ \n\ - int last_line;\n\ - int last_column;\n\ - char *text;\n\ - }\n\ -\ - yyltype;\n\ + int last_line;\n\ + int last_column;\n\ +} yyltype;\n\ \n\ -#define YYLTYPE yyltype\n\ +# define YYLTYPE yyltype\n\ #endif\n\ -\n", - f); +\n"); }