X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/9a1af7e6b2ba4581b98f18a64db7431fcd8d5b52..3628146521aa93cdd499fa5058d21c7be33014bf:/src/reader.c diff --git a/src/reader.c b/src/reader.c index 3f7cb0e3..552bdd48 100644 --- a/src/reader.c +++ b/src/reader.c @@ -1,90 +1,90 @@ /* Input parser for bison - Copyright (C) 1984, 1986, 1989 Free Software Foundation, Inc. + Copyright (C) 1984, 86, 89, 92, 98, 2000 Free Software Foundation, Inc. -This file is part of Bison, the GNU Compiler Compiler. + This file is part of Bison, the GNU Compiler Compiler. -Bison is free software; you can redistribute it and/or modify -it under the terms of the GNU General Public License as published by -the Free Software Foundation; either version 2, or (at your option) -any later version. + Bison is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. -Bison is distributed in the hope that it will be useful, -but WITHOUT ANY WARRANTY; without even the implied warranty of -MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -GNU General Public License for more details. + Bison is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. -You should have received a copy of the GNU General Public License -along with Bison; see the file COPYING. If not, write to -the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ + You should have received a copy of the GNU General Public License + along with Bison; see the file COPYING. If not, write to + the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ -/* 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. +/* 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. -The entry point is reader(). */ + The entry point is reader (). */ #include -#include #include "system.h" #include "files.h" -#include "new.h" +#include "alloc.h" #include "symtab.h" #include "lex.h" #include "gram.h" #include "machine.h" - -#define LTYPESTR "\n#ifndef YYLTYPE\ntypedef\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" +#include "complain.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' */ int rline_allocated; -extern char *program_name; extern int definesflag; extern int nolinesflag; +extern int noparserflag; +extern int rawtoknumflag; extern bucket *symval; extern int numval; -extern int failure; extern int expected_conflicts; extern char *token_buffer; - -extern int atoi (); - -extern void init_lex(); -extern void tabinit(); -extern void output_headers(); -extern void output_trailers(); -extern void free_symtab(); -extern void open_extra_files(); -extern void fatal(); -extern void fatals(); -extern void unlex(); -extern void done(); - -extern int skip_white_space(); -extern int parse_percent_token(); -extern int lex(); - -void read_declarations(); -void copy_definition(); -void parse_token_decl(); -void parse_start_decl(); -void parse_type_decl(); -void parse_assoc_decl(); -void parse_union_decl(); -void parse_expect_decl(); -void copy_action(); -void readgram(); -void record_rule_line(); -void packsymbols(); -void output_token_defines(); -void packgram(); -int read_signed_integer(); -int get_type(); +extern int maxtoken; + +extern void init_lex PARAMS((void)); +extern char *grow_token_buffer PARAMS((char *)); +extern void tabinit PARAMS((void)); +extern void output_headers PARAMS((void)); +extern void output_trailers PARAMS((void)); +extern void free_symtab PARAMS((void)); +extern void open_extra_files PARAMS((void)); +extern char *printable_version PARAMS((int)); +extern void unlex PARAMS((int)); + +extern int skip_white_space PARAMS((void)); +extern int parse_percent_token PARAMS((void)); +extern int lex PARAMS((void)); typedef struct symbol_list @@ -96,12 +96,38 @@ typedef symbol_list; +void reader PARAMS((void)); +void reader_output_yylsp PARAMS((FILE *)); +void read_declarations PARAMS((void)); +void copy_definition PARAMS((void)); +void parse_token_decl PARAMS((int, int)); +void parse_start_decl PARAMS((void)); +void parse_type_decl PARAMS((void)); +void parse_assoc_decl PARAMS((int)); +void parse_union_decl PARAMS((void)); +void parse_expect_decl PARAMS((void)); +char *get_type_name PARAMS((int, symbol_list *)); +void copy_guard PARAMS((symbol_list *, int)); +void parse_thong_decl PARAMS((void)); +void copy_action PARAMS((symbol_list *, int)); +bucket *gensym PARAMS((void)); +void readgram PARAMS((void)); +void record_rule_line PARAMS((void)); +void packsymbols PARAMS((void)); +void output_token_defines PARAMS((FILE *)); +void packgram PARAMS((void)); +int read_signed_integer PARAMS((FILE *)); + +#if 0 +static int get_type PARAMS((void)); +#endif int lineno; symbol_list *grammar; int start_flag; bucket *startval; char **tags; +int *user_toknums; /* Nonzero if components of semantic values are used, implying they must be unions. */ @@ -114,14 +140,123 @@ static int lastprec; /* incremented for each %left, %right or %nonassoc seen */ static int gensym_count; /* incremented for each generated symbol */ static bucket *errtoken; +static bucket *undeftoken; /* Nonzero if any action or guard uses the @n construct. */ static int yylsp_needed; -extern char *version_string; + +static void +skip_to_char (int target) +{ + int c; + if (target == '\n') + complain (_(" Skipping to next \\n")); + else + complain (_(" Skipping to next %c"), target); + + do + c = skip_white_space(); + while (c != target && c != EOF); + if (c != EOF) + ungetc(c, finput); +} + + +/* Dump the string from FINPUT to FOUTPUT. MATCH is the delimiter of + the string (either ' or "). */ + +static inline void +copy_string (FILE *finput, FILE *foutput, int match) +{ + int c; + + putc (match, foutput); + c = getc (finput); + + while (c != match) + { + if (c == EOF) + fatal (_("unterminated string at end of file")); + if (c == '\n') + { + complain (_("unterminated string")); + ungetc (c, finput); + c = match; /* invent terminator */ + continue; + } + + putc(c, foutput); + + if (c == '\\') + { + c = getc (finput); + if (c == EOF) + fatal (_("unterminated string at end of file")); + putc (c, foutput); + if (c == '\n') + lineno++; + } + + c = getc(finput); + } + + putc(c, foutput); +} + + +/* Dump the comment from FINPUT to FOUTPUT. C is either `*' or `/', + depending upon the type of comments used. */ + +static inline void +copy_comment (FILE *finput, FILE *foutput, int c) +{ + int cplus_comment; + register int ended; + + cplus_comment = (c == '/'); + putc (c, foutput); + c = getc (finput); + + ended = 0; + while (!ended) + { + if (!cplus_comment && c == '*') + { + while (c == '*') + { + putc(c, foutput); + c = getc(finput); + } + + if (c == '/') + { + putc(c, foutput); + ended = 1; + } + } + else if (c == '\n') + { + lineno++; + putc (c, foutput); + if (cplus_comment) + ended = 1; + else + c = getc(finput); + } + else if (c == EOF) + fatal (_("unterminated comment")); + else + { + putc (c, foutput); + c = getc (finput); + } + } +} + void -reader() +reader (void) { start_flag = 0; startval = NULL; /* start symbol not specified yet. */ @@ -163,27 +298,27 @@ reader() errtoken->user_token_number = 256; /* Value specified by posix. */ /* construct a token that represents all undefined literal tokens. */ /* it is always token number 2. */ - getsym("$illegal.")->class = STOKEN; + undeftoken = getsym("$undefined."); + undeftoken->class = STOKEN; + undeftoken->user_token_number = 2; /* Read the declaration section. Copy %{ ... %} groups to ftable and fdefines file. Also notice any %token, %left, etc. found there. */ - fprintf(ftable, "\n/* A Bison parser, made from %s", infile); - fprintf(ftable, " with Bison version %s */\n\n", version_string); + if (noparserflag) + fprintf(ftable, "\n/* Bison-generated parse tables, made from %s\n", + infile); + else + fprintf(ftable, "\n/* A Bison parser, made from %s\n", infile); + fprintf(ftable, " by %s */\n\n", VERSION_STRING); fprintf(ftable, "#define YYBISON 1 /* Identify Bison output. */\n\n"); read_declarations(); - /* output the definition of YYLTYPE into the fattrs and fdefines files. */ - /* fattrs winds up in the .tab.c file, before bison.simple. */ - fprintf(fattrs, LTYPESTR); /* 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(); /* Now we know whether we need the line-number stack. If we do, write its type into the .tab.h file. */ - if (yylsp_needed) - { - if (fdefines) - fprintf(fdefines, LTYPESTR); - } + if (fdefines) + reader_output_yylsp(fdefines); /* write closing delimiters for actions and guards. */ output_trailers(); if (yylsp_needed) @@ -198,14 +333,19 @@ reader() free_symtab(); } +void +reader_output_yylsp (FILE *f) +{ + if (yylsp_needed) + fprintf(f, LTYPESTR); +} - -/* read from finput until %% is seen. Discard the %%. -Handle any % declarations, -and copy the contents of any %{ ... %} groups to fattrs. */ +/* Read from finput until `%%' is seen. Discard the `%%'. Handle any + `%' declarations, and copy the contents of any `%{ ... %}' groups + to fattrs. */ void -read_declarations () +read_declarations (void) { register int c; register int tok; @@ -230,27 +370,29 @@ read_declarations () case TOKEN: parse_token_decl (STOKEN, SNTERM); break; - + case NTERM: parse_token_decl (SNTERM, STOKEN); break; - + case TYPE: parse_type_decl(); break; - + case START: parse_start_decl(); break; - + case UNION: parse_union_decl(); break; - + case EXPECT: parse_expect_decl(); break; - + case THONG: + parse_thong_decl(); + break; case LEFT: parse_assoc_decl(LEFT_ASSOC); break; @@ -275,38 +417,41 @@ read_declarations () pure_parser = 1; break; + case NOOP: + break; + default: - fatal("junk after `%%' in definition section"); - } + complain (_("unrecognized: %s"), token_buffer); + skip_to_char('%'); + } } else if (c == EOF) - fatal("no input grammar"); - else if (c >= 040 && c <= 0177) - fatals ("unknown character `%c' in declaration section", c); + fatal (_("no input grammar")); else - fatals ("unknown character with code 0x%x in declaration section", c); + { + complain (_("unknown character: %s"), printable_version(c)); + skip_to_char('%'); + } } } -/* copy the contents of a %{ ... %} into the definitions file. -The %{ has already been read. Return after reading the %}. */ +/* Copy the contents of a `%{ ... %}' into the definitions file. The + `%{' has already been read. Return after reading the `%}'. */ void -copy_definition () +copy_definition (void) { register int c; - register int match; - register int ended; - register int after_percent; /* -1 while reading a character if prev char was % */ - int cplus_comment; + /* -1 while reading a character if prev char was %. */ + register int after_percent; if (!nolinesflag) fprintf(fattrs, "#line %d \"%s\"\n", lineno, infile); after_percent = 0; - c = getc(finput); + c = getc (finput); for (;;) { @@ -320,85 +465,23 @@ copy_definition () case '%': after_percent = -1; break; - + case '\'': case '"': - match = c; - putc(c, fattrs); - c = getc(finput); - - while (c != match) - { - if (c == EOF || c == '\n') - fatal("unterminated string"); - - putc(c, fattrs); - - if (c == '\\') - { - c = getc(finput); - if (c == EOF) - fatal("unterminated string"); - putc(c, fattrs); - if (c == '\n') - lineno++; - } - - c = getc(finput); - } - - putc(c, fattrs); + copy_string (finput, fattrs, c); break; case '/': - putc(c, fattrs); - c = getc(finput); + putc (c, fattrs); + c = getc (finput); if (c != '*' && c != '/') continue; - - cplus_comment = (c == '/'); - putc(c, fattrs); - c = getc(finput); - - ended = 0; - while (!ended) - { - if (!cplus_comment && c == '*') - { - while (c == '*') - { - putc(c, fattrs); - c = getc(finput); - } - - if (c == '/') - { - putc(c, fattrs); - ended = 1; - } - } - else if (c == '\n') - { - lineno++; - putc(c, fattrs); - if (cplus_comment) - ended = 1; - else - c = getc(finput); - } - else if (c == EOF) - fatal("unterminated comment in `%{' definition"); - else - { - putc(c, fattrs); - c = getc(finput); - } - } - + copy_comment (finput, fattrs, c); break; case EOF: - fatal("unterminated `%{' definition"); + fatal ("%s", + _("unterminated `%{' definition")); default: putc(c, fattrs); @@ -425,80 +508,177 @@ For %token, what_is is STOKEN and what_is_not is SNTERM. For %nterm, the arguments are reversed. */ void -parse_token_decl (what_is, what_is_not) - int what_is, what_is_not; +parse_token_decl (int what_is, int what_is_not) { -/* register int start_lineno; JF */ register int token = 0; - register int prev; register char *typename = 0; + register struct bucket *symbol = NULL; /* pts to symbol being defined */ int k; -/* start_lineno = lineno; JF */ - for (;;) { - if(ungetc(skip_white_space(), finput) == '%') - return; - -/* if (lineno != start_lineno) - return; JF */ + int tmp_char = ungetc (skip_white_space (), finput); - /* we have not passed a newline, so the token now starting is in this declaration */ - prev = token; + if (tmp_char == '%') + return; + if (tmp_char == EOF) + fatal (_("Premature EOF after %s"), token_buffer); token = lex(); if (token == COMMA) - continue; + { + symbol = NULL; + continue; + } if (token == TYPENAME) { k = strlen(token_buffer); typename = NEW2(k + 1, char); strcpy(typename, token_buffer); value_components_used = 1; + symbol = NULL; + } + else if (token == 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 */ } else if (token == IDENTIFIER) { int oldclass = symval->class; + symbol = symval; - if (symval->class == what_is_not) - fatals("symbol %s redefined", symval->tag); - symval->class = what_is; + if (symbol->class == what_is_not) + complain (_("symbol %s redefined"), symbol->tag); + symbol->class = what_is; if (what_is == SNTERM && oldclass != SNTERM) - symval->value = nvars++; + symbol->value = nvars++; if (typename) { - if (symval->type_name == NULL) - symval->type_name = typename; - else - fatals("type redeclaration for %s", symval->tag); + if (symbol->type_name == NULL) + symbol->type_name = typename; + else if (strcmp(typename, symbol->type_name) != 0) + complain (_("type redeclaration for %s"), symbol->tag); } } - else if (prev == IDENTIFIER && token == NUMBER) + else if (symbol && token == NUMBER) { - symval->user_token_number = numval; + symbol->user_token_number = numval; translations = 1; } else - fatal("invalid text in %token or %nterm declaration"); + { + complain (_("`%s' is invalid in %s"), + token_buffer, + (what_is == STOKEN) ? "%token" : "%nterm"); + skip_to_char('%'); + } } } +/* parse what comes after %thong + the full syntax is + %thong token number literal + the or number may be omitted. The number specifies the + user_token_number. + + Two symbols are entered in the table, one for the token symbol and + one for the literal. Both are given the , if any, from the declaration. + The ->user_token_number of the first is SALIAS and the ->user_token_number + of the second is set to the number, if any, from the declaration. + The two symbols are linked via pointers in their ->alias fields. + + during output_defines_table, the symbol is reported + thereafter, only the literal string is retained + it is the literal string that is output to yytname +*/ + +void +parse_thong_decl (void) +{ + register int token; + register struct bucket *symbol; + register char *typename = 0; + int k, usrtoknum; + + translations = 1; + token = lex(); /* fetch typename or first token */ + if (token == TYPENAME) { + k = strlen(token_buffer); + typename = NEW2(k + 1, char); + strcpy(typename, token_buffer); + value_components_used = 1; + token = lex(); /* fetch first token */ + } + + /* process first token */ + + if (token != IDENTIFIER) + { + complain (_("unrecognized item %s, expected an identifier"), + token_buffer); + skip_to_char('%'); + return; + } + symval->class = STOKEN; + symval->type_name = typename; + symval->user_token_number = SALIAS; + symbol = symval; + + token = lex(); /* get number or literal string */ + + if (token == NUMBER) { + usrtoknum = numval; + token = lex(); /* okay, did number, now get literal */ + } + else usrtoknum = 0; + + /* process literal string token */ + + if (token != IDENTIFIER || *symval->tag != '\"') + { + complain (_("expected string constant instead of %s"), + token_buffer); + skip_to_char('%'); + return; + } + symval->class = STOKEN; + symval->type_name = typename; + symval->user_token_number = usrtoknum; + + symval->alias = symbol; + symbol->alias = symval; + + nsyms--; /* symbol and symval combined are only one symbol */ +} -/* parse what comes after %start */ +/* Parse what comes after %start */ void -parse_start_decl () +parse_start_decl (void) { if (start_flag) - fatal("multiple %start declarations"); - start_flag = 1; + complain ("%s", _("multiple %start declarations")); if (lex() != IDENTIFIER) - fatal("invalid %start declaration"); - startval = symval; + complain ("%s", _("invalid %start declaration")); + else + { + start_flag = 1; + startval = symval; + } } @@ -506,32 +686,31 @@ parse_start_decl () /* read in a %type declaration and record its information for get_type_name to access */ void -parse_type_decl () +parse_type_decl (void) { register int k; register char *name; -/* register int start_lineno; JF */ if (lex() != TYPENAME) - fatal("ill-formed %type declaration"); + { + complain ("%s", _("%type declaration has no ")); + skip_to_char('%'); + return; + } k = strlen(token_buffer); name = NEW2(k + 1, char); strcpy(name, token_buffer); -/* start_lineno = lineno; */ - for (;;) { register int t; + int tmp_char = ungetc (skip_white_space (), finput); - if(ungetc(skip_white_space(), finput) == '%') + if (tmp_char == '%') return; - -/* if (lineno != start_lineno) - return; JF */ - - /* we have not passed a newline, so the token now starting is in this declaration */ + if (tmp_char == EOF) + fatal (_("Premature EOF after %s"), token_buffer); t = lex(); @@ -545,13 +724,15 @@ parse_type_decl () case IDENTIFIER: if (symval->type_name == NULL) symval->type_name = name; - else - fatals("type redeclaration for %s", symval->tag); + else if (strcmp(name, symval->type_name) != 0) + complain (_("type redeclaration for %s"), symval->tag); break; default: - fatal("invalid %type declaration"); + complain (_("invalid %%type declaration due to item: %s"), + token_buffer); + skip_to_char('%'); } } } @@ -562,29 +743,23 @@ parse_type_decl () /* assoc is either LEFT_ASSOC, RIGHT_ASSOC or NON_ASSOC. */ void -parse_assoc_decl (assoc) -int assoc; +parse_assoc_decl (int assoc) { register int k; register char *name = NULL; -/* register int start_lineno; JF */ - register int prev = 0; /* JF added = 0 to keep lint happy */ + register int prev = 0; lastprec++; /* Assign a new precedence level, never 0. */ -/* start_lineno = lineno; */ - for (;;) { register int t; + int tmp_char = ungetc (skip_white_space (), finput); - if(ungetc(skip_white_space(), finput) == '%') + if (tmp_char == '%') return; - - /* if (lineno != start_lineno) - return; JF */ - - /* we have not passed a newline, so the token now starting is in this declaration */ + if (tmp_char == EOF) + fatal (_("Premature EOF after %s"), token_buffer); t = lex(); @@ -602,18 +777,18 @@ int assoc; case IDENTIFIER: if (symval->prec != 0) - fatals("redefining precedence of %s", symval->tag); + complain (_("redefining precedence of %s"), symval->tag); symval->prec = lastprec; symval->assoc = assoc; if (symval->class == SNTERM) - fatals("symbol %s redefined", symval->tag); + complain (_("symbol %s redefined"), symval->tag); symval->class = STOKEN; if (name) { /* record the type, if one is specified */ if (symval->type_name == NULL) symval->type_name = name; - else - fatals("type redeclaration for %s", symval->tag); + else if (strcmp(name, symval->type_name) != 0) + complain (_("type redeclaration for %s"), symval->tag); } break; @@ -623,15 +798,20 @@ int assoc; symval->user_token_number = numval; translations = 1; } - else - fatal("invalid text in association declaration"); + else + { + complain (_("invalid text (%s) - number should be after identifier"), + token_buffer); + skip_to_char('%'); + } break; case SEMICOLON: return; default: - fatal("malformatted association declaration"); + complain (_("unexpected item: %s"), token_buffer); + skip_to_char('%'); } prev = t; @@ -646,7 +826,7 @@ int assoc; definition of YYSTYPE, the type of elements of the parser value stack. */ void -parse_union_decl() +parse_union_decl (void) { register int c; register int count; @@ -654,7 +834,7 @@ parse_union_decl() int cplus_comment; if (typed) - fatal("multiple %union declarations"); + complain ("%s", _("multiple %union declarations")); typed = 1; @@ -712,7 +892,7 @@ parse_union_decl() } } if (c == EOF) - fatal("unterminated comment"); + fatal (_("unterminated comment at end of file")); if (!cplus_comment && c == '*') { @@ -738,9 +918,9 @@ parse_union_decl() case '}': if (count == 0) - fatal ("unmatched close-brace (`}')"); + complain (_("unmatched close-brace (`}')")); count--; - if (count == 0) + if (count <= 0) { fprintf(fattrs, " YYSTYPE;\n"); if (fdefines) @@ -760,7 +940,7 @@ parse_union_decl() shift-reduce conflicts. */ void -parse_expect_decl() +parse_expect_decl (void) { register int c; register int count; @@ -781,25 +961,57 @@ parse_expect_decl() ungetc (c, finput); + if (count <= 0 || count > 10) + complain ("%s", _("argument of %expect is not an integer")); expected_conflicts = atoi (buffer); } /* that's all of parsing the declaration section */ -/* Get the data type (alternative in the union) of the value for symbol n in rule rule. */ +/* FINPUT is pointing to a location (i.e., a `@'). Output to FOUTPUT + a reference to this location. STACK_OFFSET is the number of values + in the current rule so far, which says where to find `$0' with + respect to the top of the stack. */ +static inline void +copy_at (FILE *finput, FILE *foutput, int stack_offset) +{ + int c; + + c = getc (finput); + if (c == '$') + { + fprintf (foutput, "yyloc"); + yylsp_needed = 1; + } + else if (isdigit(c) || c == '-') + { + int n; + + ungetc (c, finput); + n = read_signed_integer (finput); + + fprintf (foutput, "yylsp[%d]", n - stack_offset); + yylsp_needed = 1; + } + else + complain (_("@%s is invalid"), printable_version (c)); +} + + +/* Get the data type (alternative in the union) of the value for + symbol n in rule rule. */ char * -get_type_name(n, rule) -int n; -symbol_list *rule; +get_type_name (int n, symbol_list *rule) { - static char *msg = "invalid $ value"; - register int i; register symbol_list *rp; if (n < 0) - fatal(msg); + { + complain (_("invalid $ value")); + return NULL; + } rp = rule; i = 0; @@ -808,42 +1020,40 @@ symbol_list *rule; { rp = rp->next; if (rp == NULL || rp->sym == NULL) - fatal(msg); + { + complain (_("invalid $ value")); + return NULL; + } i++; } - return (rp->sym->type_name); + return rp->sym->type_name; } -/* after %guard is seen in the input file, -copy the actual guard into the guards file. -If the guard is followed by an action, copy that into the actions file. -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, -for the simple parser in which the stack is not popped until after the guard is run. */ +/* After `%guard' is seen in the input file, copy the actual guard + into the guards file. If the guard is followed by an action, copy + that into the actions file. 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, for the simple parser in which the + stack is not popped until after the guard is run. */ void -copy_guard(rule, stack_offset) -symbol_list *rule; -int stack_offset; +copy_guard (symbol_list *rule, int stack_offset) { register int c; register int n; register int count; - register int match; - register int ended; register char *type_name; int brace_flag = 0; - int cplus_comment; /* 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); + fprintf (fguard, "#line %d \"%s\"\n", lineno, infile); putc('{', fguard); count = 0; @@ -869,83 +1079,23 @@ int stack_offset; if (count > 0) count--; else - fatal("unmatched right brace ('}')"); + { + complain (_("unmatched right brace (`}')")); + c = getc(finput); /* skip it */ + } break; case '\'': case '"': - match = c; - putc(c, fguard); - c = getc(finput); - - while (c != match) - { - if (c == EOF || c == '\n') - fatal("unterminated string"); - - putc(c, fguard); - - if (c == '\\') - { - c = getc(finput); - if (c == EOF) - fatal("unterminated string"); - putc(c, fguard); - if (c == '\n') - lineno++; - } - - c = getc(finput); - } - - putc(c, fguard); + copy_string (finput, fguard, c); break; case '/': - putc(c, fguard); - c = getc(finput); + putc (c, fguard); + c = getc (finput); if (c != '*' && c != '/') continue; - - cplus_comment = (c == '/'); - putc(c, fguard); - c = getc(finput); - - ended = 0; - while (!ended) - { - if (!cplus_comment && c == '*') - { - while (c == '*') - { - putc(c, fguard); - c = getc(finput); - } - - if (c == '/') - { - putc(c, fguard); - ended = 1; - } - } - else if (c == '\n') - { - lineno++; - putc(c, fguard); - if (cplus_comment) - ended = 1; - else - c = getc(finput); - } - else if (c == EOF) - fatal("unterminated comment"); - else - { - putc(c, fguard); - c = getc(finput); - } - } - + copy_comment (finput, fguard, c); break; case '$': @@ -957,7 +1107,12 @@ int stack_offset; register char *cp = token_buffer; while ((c = getc(finput)) != '>' && c > 0) - *cp++ = c; + { + if (cp == token_buffer + maxtoken) + cp = grow_token_buffer(cp); + + *cp++ = c; + } *cp = 0; type_name = token_buffer; @@ -967,18 +1122,19 @@ int stack_offset; if (c == '$') { fprintf(fguard, "yyval"); - if (!type_name) type_name = rule->sym->type_name; + if (!type_name) + type_name = rule->sym->type_name; if (type_name) fprintf(fguard, ".%s", type_name); - if(!type_name && typed) /* JF */ - fprintf(stderr,"%s:%d: warning: $$ of '%s' has no declared type.\n",infile,lineno,rule->sym->tag); + if(!type_name && typed) + complain (_("$$ 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); + n = read_signed_integer (finput); + c = getc (finput); if (!type_name && n > 0) type_name = get_type_name(n, rule); @@ -986,36 +1142,25 @@ int stack_offset; fprintf(fguard, "yyvsp[%d]", n - stack_offset); if (type_name) fprintf(fguard, ".%s", type_name); - if(!type_name && typed) /* JF */ - fprintf(stderr,"%s:%d: warning: $%d of '%s' has no declared type.\n",infile,lineno,n,rule->sym->tag); + if (!type_name && typed) + complain (_("$%d of `%s' has no declared type"), + n, rule->sym->tag); continue; } else - fatals("$%c is invalid",c); /* JF changed style */ - + complain (_("$%s is invalid"), printable_version (c)); break; case '@': - c = getc(finput); - if (isdigit(c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer(finput); - c = getc(finput); - } - else - fatals("@%c is invalid",c); /* JF changed style */ - - fprintf(fguard, "yylsp[%d]", n - stack_offset); - yylsp_needed = 1; - - continue; + copy_at (finput, fguard, stack_offset); + break; case EOF: - fatal("unterminated %guard clause"); + fatal ("%s", + _("unterminated %guard clause")); default: - putc(c, fguard); + putc (c, fguard); } if (c != '}' || count != 0) @@ -1026,12 +1171,12 @@ int stack_offset; fprintf(fguard, ";\n break;}"); if (c == '{') - copy_action(rule, stack_offset); + copy_action (rule, stack_offset); else if (c == '=') { - c = getc(finput); + c = getc(finput); /* why not skip_white_space -wjh */ if (c == '{') - copy_action(rule, stack_offset); + copy_action (rule, stack_offset); } else ungetc(c, finput); @@ -1039,31 +1184,27 @@ int stack_offset; -/* Assuming that a { has just been seen, copy everything up to the matching } -into the actions file. -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. */ +/* Assuming that a `{' has just been seen, copy everything up to the + matching `}' into the actions file. STACK_OFFSET is the number of + values in the current rule so far, which says where to find `$0' + with respect to the top of the stack. */ void -copy_action(rule, stack_offset) -symbol_list *rule; -int stack_offset; +copy_action (symbol_list *rule, int stack_offset) { register int c; register int n; register int count; - register int match; - register int ended; register char *type_name; - int cplus_comment; /* offset is always 0 if parser has already popped the stack pointer */ - if (semantic_parser) stack_offset = 0; + if (semantic_parser) + stack_offset = 0; - fprintf(faction, "\ncase %d:\n", nrules); + fprintf (faction, "\ncase %d:\n", nrules); if (!nolinesflag) - fprintf(faction, "#line %d \"%s\"\n", lineno, infile); - putc('{', faction); + fprintf (faction, "#line %d \"%s\"\n", lineno, infile); + putc ('{', faction); count = 1; c = getc(finput); @@ -1086,31 +1227,7 @@ int stack_offset; case '\'': case '"': - match = c; - putc(c, faction); - c = getc(finput); - - while (c != match) - { - if (c == EOF || c == '\n') - fatal("unterminated string"); - - putc(c, faction); - - if (c == '\\') - { - c = getc(finput); - if (c == EOF) - fatal("unterminated string"); - putc(c, faction); - if (c == '\n') - lineno++; - } - - c = getc(finput); - } - - putc(c, faction); + copy_string (finput, faction, c); break; case '/': @@ -1118,46 +1235,7 @@ int stack_offset; c = getc(finput); if (c != '*' && c != '/') continue; - - cplus_comment = (c == '/'); - putc(c, faction); - c = getc(finput); - - ended = 0; - while (!ended) - { - if (!cplus_comment && c == '*') - { - while (c == '*') - { - putc(c, faction); - c = getc(finput); - } - - if (c == '/') - { - putc(c, faction); - ended = 1; - } - } - else if (c == '\n') - { - lineno++; - putc(c, faction); - if (cplus_comment) - ended = 1; - else - c = getc(finput); - } - else if (c == EOF) - fatal("unterminated comment"); - else - { - putc(c, faction); - c = getc(finput); - } - } - + copy_comment (finput, faction, c); break; case '$': @@ -1169,7 +1247,12 @@ int stack_offset; register char *cp = token_buffer; while ((c = getc(finput)) != '>' && c > 0) - *cp++ = c; + { + if (cp == token_buffer + maxtoken) + cp = grow_token_buffer(cp); + + *cp++ = c; + } *cp = 0; type_name = token_buffer; value_components_used = 1; @@ -1179,11 +1262,13 @@ int stack_offset; if (c == '$') { fprintf(faction, "yyval"); - if (!type_name) type_name = get_type_name(0, rule); + if (!type_name) + type_name = get_type_name(0, rule); if (type_name) fprintf(faction, ".%s", type_name); - if(!type_name && typed) /* JF */ - fprintf(stderr,"%s:%d: warning: $$ of '%s' has no declared type.\n",infile,lineno,rule->sym->tag); + if(!type_name && typed) + complain (_("$$ of `%s' has no declared type"), + rule->sym->tag); } else if (isdigit(c) || c == '-') { @@ -1197,33 +1282,22 @@ int stack_offset; fprintf(faction, "yyvsp[%d]", n - stack_offset); if (type_name) fprintf(faction, ".%s", type_name); - if(!type_name && typed) /* JF */ - fprintf(stderr,"%s:%d: warning: $%d of '%s' has no declared type.\n",infile,lineno,n,rule->sym->tag); + if(!type_name && typed) + complain (_("$%d of `%s' has no declared type"), + n, rule->sym->tag); continue; } else - fatals("$%c is invalid",c); /* JF changed format */ + complain (_("$%s is invalid"), printable_version (c)); break; case '@': - c = getc(finput); - if (isdigit(c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer(finput); - c = getc(finput); - } - else - fatal("invalid @-construct"); - - fprintf(faction, "yylsp[%d]", n - stack_offset); - yylsp_needed = 1; - - continue; + copy_at (finput, faction, stack_offset); + break; case EOF: - fatal("unmatched '{'"); + fatal (_("unmatched `{'")); default: putc(c, faction); @@ -1250,7 +1324,7 @@ int stack_offset; whose name cannot conflict with the user's names. */ bucket * -gensym() +gensym (void) { register bucket *sym; @@ -1258,7 +1332,7 @@ gensym() sym = getsym(token_buffer); sym->class = SNTERM; sym->value = nvars++; - return (sym); + return sym; } /* Parse the input grammar into a one symbol_list structure. @@ -1271,10 +1345,10 @@ All guards and actions are copied out to the appropriate files, labelled by the rule number they apply to. */ void -readgram() +readgram (void) { register int t; - register bucket *lhs; + register bucket *lhs = NULL; register symbol_list *p; register symbol_list *p1; register bucket *bp; @@ -1299,21 +1373,26 @@ readgram() if (t == IDENTIFIER) { lhs = symval; - + + if (!start_flag) + { + startval = lhs; + start_flag = 1; + } + t = lex(); if (t != COLON) - fatal("ill-formed rule"); + { + complain (_("ill-formed rule: initial symbol not followed by colon")); + unlex(t); + } } - if (nrules == 0) + if (nrules == 0 && t == BAR) { - if (t == BAR) - fatal("grammar starts with vertical bar"); - - if (!start_flag) - startval = lhs; + complain (_("grammar starts with vertical bar")); + lhs = symval; /* BOGUS: use a random symval */ } - /* start a new rule and record its lhs. */ nrules++; @@ -1342,13 +1421,19 @@ readgram() nvars++; } else if (lhs->class == STOKEN) - fatals("rule given for %s, which is a token", lhs->tag); + complain (_("rule given for %s, which is a token"), lhs->tag); /* read the rhs of the rule. */ for (;;) { t = lex(); + if (t == PREC) + { + t = lex(); + crule->ruleprec = symval; + t = lex(); + } if (! (t == IDENTIFIER || t == LEFT_CURLY)) break; @@ -1379,7 +1464,7 @@ readgram() register bucket *sdummy; /* Since the action was written out with this rule's */ - /* number, we must write give the new rule this number */ + /* number, we must give the new rule this number */ /* by inserting the new rule before it. */ /* Make a dummy nonterminal, a gensym. */ @@ -1425,7 +1510,7 @@ readgram() 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); @@ -1434,6 +1519,7 @@ readgram() if (t == PREC) { + complain (_("two @prec's in a row")); t = lex(); crule->ruleprec = symval; t = lex(); @@ -1441,39 +1527,48 @@ readgram() if (t == GUARD) { if (! semantic_parser) - fatal("%guard present but %semantic_parser not specified"); + complain ("%s", + _("%guard present but %semantic_parser not specified")); copy_guard(crule, rulelength); t = lex(); } else if (t == LEFT_CURLY) { - if (actionflag) fatal("two actions at end of one rule"); + /* This case never occurs -wjh */ + if (actionflag) + complain (_("two actions at end of one rule")); copy_action(crule, rulelength); + actionflag = 1; + xactions++; /* -wjh */ t = lex(); } - /* If $$ is being set in default way, - warn if any type mismatch. */ - else if (!xactions && first_rhs && lhs->type_name != first_rhs->type_name) + /* If $$ is being set in default way, report if any type + mismatch. */ + else if (!xactions + && first_rhs + && lhs->type_name != first_rhs->type_name) { - if (lhs->type_name == 0 || first_rhs->type_name == 0 + if (lhs->type_name == 0 + || first_rhs->type_name == 0 || strcmp(lhs->type_name,first_rhs->type_name)) - fprintf(stderr, "%s:%d: warning: type clash ('%s' '%s') on default action\n", - infile, - lineno, - lhs->type_name ? lhs->type_name : "", - first_rhs->type_name ? first_rhs->type_name : ""); + complain (_("type clash (`%s' `%s') on default action"), + lhs->type_name ? lhs->type_name : "", + first_rhs->type_name ? first_rhs->type_name : ""); } /* Warn if there is no default for $$ but we need one. */ else if (!xactions && !first_rhs && lhs->type_name != 0) - fprintf(stderr, - "%s:%d: warning: empty rule for typed nonterminal, and no action\n", - infile, - lineno); + complain (_("empty rule for typed nonterminal, and no action")); if (t == SEMICOLON) t = lex(); } - /* these things can appear as alternatives to rules. */ +#if 0 + /* these things can appear as alternatives to rules. */ +/* NO, they cannot. + a) none of the documentation allows them + b) most of them scan forward until finding a next % + thus they may swallow lots of intervening rules +*/ else if (t == TOKEN) { parse_token_decl(STOKEN, SNTERM); @@ -1503,15 +1598,22 @@ readgram() parse_start_decl(); t = lex(); } +#endif + else - fatal("invalid input"); + { + complain (_("invalid input: %s"), token_buffer); + t = lex(); + } } + /* grammar has been read. Do some checking */ + if (nsyms > MAXSHORT) - fatals("too many symbols (tokens plus nonterminals); maximum %d", + fatal (_("too many symbols (tokens plus nonterminals); maximum %d"), MAXSHORT); if (nrules == 0) - fatal("no input grammar"); + fatal (_("no rules in the input grammar")); if (typed == 0 /* JF put out same default YYSTYPE as YACC does */ && !value_components_used) @@ -1529,9 +1631,8 @@ readgram() for (bp = firstsymbol; bp; bp = bp->next) if (bp->class == SUNKNOWN) { - fprintf(stderr, "symbol %s used, not defined as token, and no rules for it\n", - bp->tag); - failure = 1; + complain (_("symbol %s is used, but is not defined as a token and has no rules"), + bp->tag); bp->class = SNTERM; bp->value = nvars++; } @@ -1541,24 +1642,25 @@ readgram() void -record_rule_line () +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 (rline, + rline = (short *) xrealloc ((char *) rline, rline_allocated * sizeof (short)); } rline[nrules] = lineno; } +#if 0 /* read in a %type declaration and record its information for get_type_name to access */ - -int -get_type() +/* this is unused. it is only called from the #if 0 part of readgram */ +static int +get_type (void) { register int k; register int t; @@ -1567,7 +1669,10 @@ get_type() t = lex(); if (t != TYPENAME) - fatal("ill-formed %type declaration"); + { + complain (_("ill-formed %type declaration")); + return t; + } k = strlen(token_buffer); name = NEW2(k + 1, char); @@ -1580,7 +1685,7 @@ get_type() switch (t) { case SEMICOLON: - return (lex()); + return lex(); case COMMA: break; @@ -1588,24 +1693,25 @@ get_type() case IDENTIFIER: if (symval->type_name == NULL) symval->type_name = name; - else - fatals("type redeclaration for %s", symval->tag); + else if (strcmp(name, symval->type_name) != 0) + complain (_("type redeclaration for %s"), symval->tag); break; default: - return (t); + return t; } } } +#endif - -/* assign symbol numbers, and write definition of token names into fdefines. -Set up vectors tags and sprec of names and precedences of symbols. */ +/* Assign symbol numbers, and write definition of token names into + fdefines. Set up vectors tags and sprec of names and precedences + of symbols. */ void -packsymbols() +packsymbols (void) { register bucket *bp; register int tokno = 1; @@ -1616,6 +1722,8 @@ packsymbols() tags = NEW2(nsyms + 1, char *); tags[0] = "$"; + user_toknums = NEW2(nsyms + 1, int); + user_toknums[0] = 0; sprec = NEW2(nsyms, short); sassoc = NEW2(nsyms, short); @@ -1629,16 +1737,57 @@ packsymbols() { bp->value += ntokens; } - else + else if (bp->alias) + { + /* this symbol and its alias are a single token defn. + allocate a tokno, and assign to both check agreement of + ->prec and ->assoc fields and make both the same */ + if (bp->value == 0) + bp->value = bp->alias->value = tokno++; + + if (bp->prec != bp->alias->prec) + { + if (bp->prec != 0 && bp->alias->prec != 0 + && bp->user_token_number == SALIAS) + complain (_("conflicting precedences for %s and %s"), + bp->tag, bp->alias->tag); + if (bp->prec != 0) + bp->alias->prec = bp->prec; + else + bp->prec = bp->alias->prec; + } + + if (bp->assoc != bp->alias->assoc) + { + if (bp->assoc != 0 && bp->alias->assoc != 0 + && bp->user_token_number == SALIAS) + complain (_("conflicting assoc values for %s and %s"), + bp->tag, bp->alias->tag); + if (bp->assoc != 0) + bp->alias->assoc = bp->assoc; + else + bp->assoc = bp->alias->assoc; + } + + if (bp->user_token_number == SALIAS) + continue; /* do not do processing below for SALIASs */ + + } + else /* bp->class == STOKEN */ + { + bp->value = tokno++; + } + + if (bp->class == STOKEN) { if (translations && !(bp->user_token_number)) 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; - bp->value = tokno++; } tags[bp->value] = bp->tag; + user_toknums[bp->value] = bp->user_token_number; sprec[bp->value] = bp->prec; sassoc[bp->value] = bp->assoc; @@ -1650,37 +1799,36 @@ packsymbols() token_translations = NEW2(max_user_token_number+1, short); - /* initialize all entries for literal tokens to 2, - the internal token number for $illegal., which represents all invalid inputs. */ + /* 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; - } + token_translations[i] = 2; - for (bp = firstsymbol; bp; bp = bp->next) - { - if (bp->value >= ntokens) continue; - if (translations) - { - if (token_translations[bp->user_token_number] != 2) - { - /* JF made this a call to fatals() */ - fatals( "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; - } + 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; + } } error_token_number = errtoken->value; - output_token_defines(ftable); + if (! noparserflag) + output_token_defines(ftable); if (startval->class == SUNKNOWN) - fatals("the start symbol %s is undefined", startval->tag); + fatal (_("the start symbol %s is undefined"), startval->tag); else if (startval->class == STOKEN) - fatals("the start symbol %s is a token", startval->tag); + fatal (_("the start symbol %s is a token"), startval->tag); start_symbol = startval->value; @@ -1711,38 +1859,45 @@ packsymbols() #endif } } - + +/* For named tokens, but not literal ones, define the name. The value + is the user token number. */ void -output_token_defines(file) -FILE *file; +output_token_defines (FILE *file) { bucket *bp; + register char *cp, *symbol; + register char c; for (bp = firstsymbol; bp; bp = bp->next) { - if (bp->value >= ntokens) continue; - - /* For named tokens, but not literal ones, define the name. */ - /* The value is the user token number. */ + symbol = bp->tag; /* get symbol */ - if ('\'' != *tags[bp->value] && bp != errtoken) + if (bp->value >= ntokens) continue; + if (bp->user_token_number == SALIAS) continue; + if ('\'' == *symbol) continue; /* skip literal character */ + if (bp == errtoken) continue; /* skip error token */ + if ('\"' == *symbol) { - register char *cp = tags[bp->value]; - register char c; + /* use literal string only if given a symbol with an alias */ + if (bp->alias) + symbol = bp->alias->tag; + else + continue; + } - /* Don't #define nonliteral tokens whose names contain periods. */ + /* Don't #define nonliteral tokens whose names contain periods. */ + cp = symbol; + while ((c = *cp++) && c != '.'); + if (c != '\0') continue; - while ((c = *cp++) && c != '.'); - if (!c) - { - fprintf(file, "#define\t%s\t%d\n", tags[bp->value], - (translations ? bp->user_token_number : bp->value)); - if (semantic_parser) - fprintf(file, "#define\tT%s\t%d\n", tags[bp->value], - bp->value); - } - } + fprintf (file, "#define\t%s\t%d\n", symbol, + ((translations && ! rawtoknumflag) + ? bp->user_token_number + : bp->value)); + if (semantic_parser) + fprintf (file, "#define\tT%s\t%d\n", symbol, bp->value); } putc('\n', file); @@ -1753,7 +1908,7 @@ FILE *file; /* convert the rules into the representation using rrhs, rlhs and ritems. */ void -packgram() +packgram (void) { register int itemno; register int ruleno; @@ -1814,8 +1969,7 @@ packgram() /* Read a signed integer from STREAM and return its value. */ int -read_signed_integer (stream) - FILE *stream; +read_signed_integer (FILE *stream) { register int c = getc(stream); register int sign = 1;