X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/6deb44470e47bb02f0ca3ea630408d013e128710..63c2d5de0fa3237edad53913dca35abda6a36f2d:/src/reader.c diff --git a/src/reader.c b/src/reader.c index 9ae0238e..22eedc80 100644 --- a/src/reader.c +++ b/src/reader.c @@ -34,6 +34,7 @@ #include "output.h" #include "reader.h" #include "conflicts.h" +#include "macrotab.h" /* Number of slots allocated (but not necessarily used yet) in `rline' */ static int rline_allocated; @@ -63,9 +64,6 @@ static int typed; /* Incremented for each %left, %right or %nonassoc seen */ static int lastprec; -/* Incremented for each generated symbol */ -static int gensym_count; - static bucket *errtoken; static bucket *undeftoken; @@ -159,11 +157,12 @@ get_type_name (int n, symbol_list * rule) `------------------------------------------------------------*/ static inline void -copy_string (FILE *fin, struct obstack *oout, int match) +copy_string2 (FILE *fin, struct obstack *oout, int match, int store) { int c; - obstack_1grow (oout, match); + if (store) + obstack_1grow (oout, match); c = getc (fin); @@ -195,9 +194,30 @@ copy_string (FILE *fin, struct obstack *oout, int match) c = getc (fin); } - obstack_1grow (oout, c); + if (store) + obstack_1grow (oout, c); +} + +/* FIXME. */ + +static inline void +copy_string (FILE *fin, struct obstack *oout, int match) +{ + copy_string2 (fin, oout, match, 1); } +/* FIXME. */ + +static inline void +copy_identifier (FILE *fin, struct obstack *oout) +{ + int c; + + while (isalnum (c = getc (fin)) || c == '_') + obstack_1grow (oout, c); + + ungetc (c, fin); +} /*-----------------------------------------------------------------. | Dump the wannabee comment from IN to OUT1 and OUT2 (which can be | @@ -315,7 +335,6 @@ copy_at (FILE *fin, struct obstack *oout, int stack_offset) else if (isdigit (c) || c == '-') { int n; - char buf[4096]; ungetc (c, fin); n = read_signed_integer (fin); @@ -347,7 +366,7 @@ copy_dollar (FILE *fin, struct obstack *oout, symbol_list *rule, int stack_offset) { int c = getc (fin); - char *type_name = NULL; + const char *type_name = NULL; /* Get the type name if explicit. */ if (c == '<') @@ -407,9 +426,11 @@ copy_definition (void) /* -1 while reading a character if prev char was %. */ int after_percent; +#if 0 if (!no_lines_flag) obstack_fgrow2 (&attrs_obstack, "#line %d %s\n", lineno, quotearg_style (c_quoting_style, infile)); +#endif after_percent = 0; @@ -466,7 +487,7 @@ 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; /* The symbol being defined. */ @@ -486,18 +507,18 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) fatal (_("Premature EOF after %s"), token_buffer); token = lex (); - if (token == COMMA) + if (token == tok_comma) { symbol = NULL; continue; } - if (token == TYPENAME) + if (token == tok_typename) { typename = xstrdup (token_buffer); value_components_used = 1; symbol = NULL; } - else if (token == IDENTIFIER && *symval->tag == '\"' && symbol) + else if (token == tok_identifier && *symval->tag == '\"' && symbol) { if (symval->alias) warn (_("symbol `%s' used more than once as a literal string"), @@ -519,7 +540,7 @@ parse_token_decl (symbol_class what_is, symbol_class what_is_not) translations = 1; symbol = NULL; } - else if (token == IDENTIFIER) + else if (token == tok_identifier) { int oldclass = symval->class; symbol = symval; @@ -538,7 +559,7 @@ 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; @@ -563,7 +584,7 @@ parse_start_decl (void) { if (start_flag) complain (_("multiple %s declarations"), "%start"); - if (lex () != IDENTIFIER) + if (lex () != tok_identifier) complain (_("invalid %s declaration"), "%start"); else { @@ -582,7 +603,7 @@ parse_type_decl (void) { char *name; - if (lex () != TYPENAME) + if (lex () != tok_typename) { complain ("%s", _("%type declaration has no ")); skip_to_char ('%'); @@ -593,7 +614,7 @@ parse_type_decl (void) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -606,11 +627,11 @@ parse_type_decl (void) switch (t) { - case COMMA: - case SEMICOLON: + case tok_comma: + case tok_semicolon: break; - case IDENTIFIER: + case tok_identifier: if (symval->type_name == NULL) symval->type_name = name; else if (strcmp (name, symval->type_name) != 0) @@ -643,7 +664,7 @@ parse_assoc_decl (associativity assoc) for (;;) { - int t; + token_t t; int tmp_char = ungetc (skip_white_space (), finput); if (tmp_char == '%') @@ -655,14 +676,14 @@ parse_assoc_decl (associativity assoc) switch (t) { - case TYPENAME: + case tok_typename: name = xstrdup (token_buffer); break; - case COMMA: + case tok_comma: break; - case IDENTIFIER: + case tok_identifier: if (symval->prec != 0) complain (_("redefining precedence of %s"), symval->tag); symval->prec = lastprec; @@ -679,8 +700,8 @@ 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; @@ -694,7 +715,7 @@ token_buffer); } break; - case SEMICOLON: + case tok_semicolon: return; default: @@ -821,14 +842,14 @@ parse_expect_decl (void) static void parse_thong_decl (void) { - int token; + token_t token; struct bucket *symbol; char *typename = 0; int usrtoknum; translations = 1; token = lex (); /* fetch typename or first token */ - if (token == TYPENAME) + if (token == tok_typename) { typename = xstrdup (token_buffer); value_components_used = 1; @@ -837,7 +858,7 @@ parse_thong_decl (void) /* process first token */ - if (token != IDENTIFIER) + if (token != tok_identifier) { complain (_("unrecognized item %s, expected an identifier"), token_buffer); @@ -851,7 +872,7 @@ 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 */ @@ -861,7 +882,7 @@ 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 ('%'); @@ -878,6 +899,88 @@ parse_thong_decl (void) nsyms--; } +/* FIXME. */ + +static void +parse_macro_decl (void) +{ + int ch = ungetc (skip_white_space (), finput); + char* macro_key; + char* macro_value; + + /* Read key. */ + if (!isalpha (ch) && ch != '_') + { + complain (_("invalid %s declaration"), "%define"); + skip_to_char ('%'); + return; + } + copy_identifier (finput, ¯o_obstack); + obstack_1grow (¯o_obstack, 0); + macro_key = obstack_finish (¯o_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, ¯o_obstack, '"', 0); + obstack_1grow (¯o_obstack, 0); + macro_value = obstack_finish (¯o_obstack); + + /* Store the (key, value) pair in the environment. */ + macro_insert (macro_key, macro_value); +} + + +/*----------------------------------. +| Parse what comes after %skeleton. | +`----------------------------------*/ + +void +parse_skel_decl (void) +{ + /* Complete with parse_dquoted_param () on the CVS branch 1.29. */ +} + +/*------------------------------------------. +| 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 | @@ -901,54 +1004,70 @@ 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 NOOP: + case tok_hdrext: + parse_header_extension_decl (); + break; + + case tok_srcext: + parse_source_extension_decl (); + break; + + case tok_define: + parse_macro_decl (); + break; + + case tok_skel: + parse_skel_decl (); + break; + + case tok_noop: break; default: @@ -1178,9 +1297,14 @@ 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++; @@ -1198,12 +1322,12 @@ static int get_type (void) { int k; - int t; + token_t token; char *name; - t = lex (); + token = lex (); - if (t != TYPENAME) + if (token != tok_typename) { complain (_("invalid %s declaration"), "%type"); return t; @@ -1213,17 +1337,17 @@ get_type (void) for (;;) { - t = lex (); + token = lex (); - switch (t) + switch (token) { - case SEMICOLON: + case tok_semicolon: return lex (); - case COMMA: + case tok_comma: break; - case IDENTIFIER: + case tok_identifier: if (symval->type_name == NULL) symval->type_name = name; else if (strcmp (name, symval->type_name) != 0) @@ -1232,7 +1356,7 @@ get_type (void) break; default: - return t; + return token; } } } @@ -1253,7 +1377,7 @@ get_type (void) static void readgram (void) { - int t; + token_t t; bucket *lhs = NULL; symbol_list *p; symbol_list *p1; @@ -1269,9 +1393,9 @@ readgram (void) t = lex (); - while (t != TWO_PERCENTS && t != ENDFILE) + while (t != tok_two_percents && t != tok_eof) { - if (t == IDENTIFIER || t == BAR) + if (t == tok_identifier || t == tok_bar) { int action_flag = 0; /* Number of symbols in rhs of this rule so far */ @@ -1279,7 +1403,7 @@ readgram (void) int xactions = 0; /* JF for error checking */ bucket *first_rhs = 0; - if (t == IDENTIFIER) + if (t == tok_identifier) { lhs = symval; @@ -1290,14 +1414,14 @@ 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 */ @@ -1337,28 +1461,28 @@ 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 */ @@ -1408,7 +1532,7 @@ readgram (void) action_flag = 0; } - if (t == IDENTIFIER) + if (t == tok_identifier) { nitems++; p = XCALLOC (symbol_list, 1); @@ -1430,14 +1554,14 @@ readgram (void) p1->next = p; p1 = p; - if (t == PREC) + if (t == tok_prec) { complain (_("two @prec's in a row")); t = lex (); crule->ruleprec = symval; t = lex (); } - if (t == GUARD) + if (t == tok_guard) { if (!semantic_parser) complain (_("%%guard present but %%semantic_parser not specified")); @@ -1445,7 +1569,7 @@ readgram (void) copy_guard (crule, rulelength); t = lex (); } - else if (t == LEFT_CURLY) + else if (t == tok_left_curly) { /* This case never occurs -wjh */ if (action_flag) @@ -1470,7 +1594,7 @@ 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 @@ -1480,31 +1604,31 @@ 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 (); @@ -1513,7 +1637,7 @@ readgram (void) else { - complain (_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), quote (token_buffer)); t = lex (); } } @@ -1526,22 +1650,6 @@ readgram (void) if (nrules == 0) fatal (_("no rules in the input grammar")); - /* JF put out same default YYSTYPE as YACC does */ - if (typed == 0 - && !value_components_used) - { - /* We used to use `unsigned long' as YYSTYPE on MSDOS, - but it seems better to be consistent. - Most programs should declare their own type anyway. */ - obstack_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. */ for (bp = firstsymbol; bp; bp = bp->next) @@ -1556,6 +1664,25 @@ readgram (void) ntokens = nsyms - nvars; } + +/* At the end of the grammar file, some C source code must + be stored. It is going to be associated to the epilogue + directive. */ +static void +read_additionnal_code (void) +{ + char c; + struct obstack el_obstack; + + obstack_init (&el_obstack); + + while ((c = getc (finput)) != EOF) + obstack_1grow (&el_obstack, c); + + obstack_1grow (&el_obstack, 0); + macro_insert ("epilogue", obstack_finish (&el_obstack)); +} + /*--------------------------------------------------------------. | For named tokens, but not literal ones, define the name. The | @@ -1596,15 +1723,12 @@ output_token_defines (struct obstack *oout) if (c != '\0') continue; - obstack_fgrow2 (oout, "#define\t%s\t%d\n", + obstack_fgrow2 (oout, "# define\t%s\t%d\n", symbol, - ((translations && !raw_flag) - ? bp->user_token_number : bp->value)); + (translations ? bp->user_token_number : bp->value)); if (semantic_parser) - obstack_fgrow2 (oout, "#define\tT%s\t%d\n", symbol, bp->value); + obstack_fgrow2 (oout, "# define\tT%s\t%d\n", symbol, bp->value); } - - obstack_1grow (oout, '\n'); } @@ -1623,8 +1747,6 @@ packsymbols (void) int last_user_token_number; static char DOLLAR[] = "$"; - /* int lossage = 0; JF set but not used */ - tags = XCALLOC (char *, nsyms + 1); tags[0] = DOLLAR; user_toknums = XCALLOC (short, nsyms + 1); @@ -1726,8 +1848,14 @@ packsymbols (void) error_token_number = errtoken->value; + output_token_defines (&output_obstack); + obstack_1grow (&output_obstack, 0); + macro_insert ("tokendef", obstack_finish (&output_obstack)); + +#if 0 if (!no_parser_flag) output_token_defines (&table_obstack); +#endif if (startval->class == unknown_sym) fatal (_("the start symbol %s is undefined"), startval->tag); @@ -1756,7 +1884,7 @@ packsymbols (void) /* don't make these for dummy nonterminals made by gensym. */ if (*tags[i] != '@') obstack_fgrow2 (&defines_obstack, - "#define\tNT%s\t%d\n", tags[i], i); + "# define\tNT%s\t%d\n", tags[i], i); } #if 0 /* `fdefines' is now a temporary file, so we need to copy its @@ -1866,8 +1994,6 @@ reader (void) typed = 0; lastprec = 0; - gensym_count = 0; - semantic_parser = 0; pure_parser = 0; @@ -1876,12 +2002,17 @@ reader (void) init_lex (); lineno = 1; + /* Initialize the macro obstack. */ + obstack_init (¯o_obstack); + /* Initialize the symbol table. */ tabinit (); + /* Construct the error token */ errtoken = getsym ("error"); errtoken->class = token_sym; errtoken->user_token_number = 256; /* Value specified by POSIX. */ + /* Construct a token that represents all undefined literal tokens. It is always token number 2. */ undeftoken = getsym ("$undefined."); @@ -1891,30 +2022,18 @@ reader (void) /* Read the declaration section. Copy %{ ... %} groups to TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left, etc. found there. */ - obstack_1grow (&table_obstack, '\n'); - obstack_fgrow3 (&table_obstack, "\ -/* %s, made from %s\n\ - by GNU bison %s. */\n\ -\n", - no_parser_flag ? "Bison-generated parse tables" : "A Bison parser", - infile, VERSION); - - obstack_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 (); /* 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 0 if (defines_flag) reader_output_yylsp (&defines_obstack); - /* Write closing delimiters for actions and guards. */ - output_trailers (); - if (locations_flag) - obstack_sgrow (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); +#endif /* Assign the symbols their symbol numbers. Write #defines for the token symbols into FDEFINES if requested. */ packsymbols (); @@ -1924,30 +2043,3 @@ reader (void) referred to by symbol number. */ free_symtab (); } - - -/*------------------------------------------------------------------. -| Define YYLTYPE. Cannot be in the skeleton since we might have to | -| output it in the headers if --defines is used. | -`------------------------------------------------------------------*/ - -void -reader_output_yylsp (struct obstack *oout) -{ - if (locations_flag) - obstack_sgrow (oout, "\ -\n\ -#ifndef YYLTYPE\n\ -typedef struct yyltype\n\ -{\n\ - int first_line;\n\ - int first_column;\n\ -\n\ - int last_line;\n\ - int last_column;\n\ -} yyltype;\n\ -\n\ -# define YYLTYPE yyltype\n\ -#endif\n\ -\n"); -}