X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/6666f98f4f1997c1325197e6ecdf98d4227c849e..6c89f1c1388ef0368855fc0400d35dfaf57db3ef:/src/reader.c?ds=sidebyside diff --git a/src/reader.c b/src/reader.c index d670310a..5094bc41 100644 --- a/src/reader.c +++ b/src/reader.c @@ -27,14 +27,15 @@ The entry point is reader (). */ -#include #include "system.h" +#include "getargs.h" #include "files.h" #include "alloc.h" #include "symtab.h" #include "lex.h" #include "gram.h" -#include "machine.h" +#include "complain.h" +#include "output.h" #define LTYPESTR "\ \n\ @@ -59,13 +60,8 @@ typedef\n\ \n" /* Number of slots allocated (but not necessarily used yet) in `rline' */ -int rline_allocated; +static 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 expected_conflicts; @@ -75,21 +71,10 @@ 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 *int_to_string PARAMS((int)); extern char *printable_version PARAMS((int)); -extern void fatal PARAMS((char *)); -extern void fatals PARAMS((char *, char *)); -extern void warn PARAMS((char *)); -extern void warni PARAMS((char *, int)); -extern void warns PARAMS((char *, char *)); -extern void warnss PARAMS((char *, char *, char *)); -extern void warnsss PARAMS((char *, char *, char *, char *)); extern void unlex PARAMS((int)); -extern void done PARAMS((int)); extern int skip_white_space PARAMS((void)); extern int parse_percent_token PARAMS((void)); @@ -105,38 +90,38 @@ 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 *)); +extern void reader PARAMS((void)); +extern void reader_output_yylsp PARAMS((FILE *)); + +static void read_declarations PARAMS((void)); +static void copy_definition PARAMS((void)); +static void parse_token_decl PARAMS((int, int)); +static void parse_start_decl PARAMS((void)); +static void parse_type_decl PARAMS((void)); +static void parse_assoc_decl PARAMS((int)); +static void parse_union_decl PARAMS((void)); +static void parse_expect_decl PARAMS((void)); +static char *get_type_name PARAMS((int, symbol_list *)); +static void copy_guard PARAMS((symbol_list *, int)); +static void parse_thong_decl PARAMS((void)); +static void copy_action PARAMS((symbol_list *, int)); +static bucket *gensym PARAMS((void)); +static void readgram PARAMS((void)); +static void record_rule_line PARAMS((void)); +static void packsymbols PARAMS((void)); +static void output_token_defines PARAMS((FILE *)); +static void packgram PARAMS((void)); #if 0 static int get_type PARAMS((void)); #endif int lineno; -symbol_list *grammar; -int start_flag; -bucket *startval; char **tags; int *user_toknums; +static symbol_list *grammar; +static int start_flag; +static bucket *startval; /* Nonzero if components of semantic values are used, implying they must be unions. */ @@ -154,21 +139,170 @@ static bucket *undeftoken; /* Nonzero if any action or guard uses the @n construct. */ static int yylsp_needed; + +/*===================\ +| Low level lexing. | +\===================*/ static void skip_to_char (int target) { int c; if (target == '\n') - warn(_(" Skipping to next \\n")); + complain (_(" Skipping to next \\n")); else - warni(_(" Skipping to next %c"), target); + complain (_(" Skipping to next %c"), target); do - c = skip_white_space(); + c = skip_white_space (); while (c != target && c != EOF); if (c != EOF) - ungetc(c, finput); + ungetc (c, finput); +} + + +/*---------------------------------------------------------. +| Read a signed integer from STREAM and return its value. | +`---------------------------------------------------------*/ + +static inline int +read_signed_integer (FILE *stream) +{ + register int c = getc (stream); + register int sign = 1; + register int n = 0; + + if (c == '-') + { + c = getc (stream); + sign = -1; + } + + while (isdigit (c)) + { + n = 10 * n + (c - '0'); + c = getc (stream); + } + + ungetc (c, stream); + + return sign * n; +} + +/*-------------------------------------------------------------------. +| Dump the string from FINPUT to FOUTPUT. MATCH is the delimiter of | +| the string (either ' or "). | +`-------------------------------------------------------------------*/ + +static inline void +copy_string (FILE *fin, FILE *fout, int match) +{ + int c; + + putc (match, fout); + c = getc (fin); + + while (c != match) + { + if (c == EOF) + fatal (_("unterminated string at end of file")); + if (c == '\n') + { + complain (_("unterminated string")); + ungetc (c, fin); + c = match; /* invent terminator */ + continue; + } + + putc(c, fout); + + if (c == '\\') + { + c = getc (fin); + if (c == EOF) + fatal (_("unterminated string at end of file")); + putc (c, fout); + if (c == '\n') + lineno++; + } + + c = getc(fin); + } + + putc(c, fout); +} + + +/*---------------------------------------------------------------. +| Dump the comment from IN to OUT1 and OUT2. C is either `*' or | +| `/', depending upon the type of comments used. OUT2 might be | +| NULL. | +`---------------------------------------------------------------*/ + +static inline void +copy_comment2 (FILE *in, FILE *out1, FILE* out2, int c) +{ + int cplus_comment; + register int ended; + + cplus_comment = (c == '/'); + putc (c, out1); + if (out2) + putc (c, out2); + c = getc (in); + + ended = 0; + while (!ended) + { + if (!cplus_comment && c == '*') + { + while (c == '*') + { + putc (c, out1); + if (out2) + putc (c, out2); + c = getc (in); + } + + if (c == '/') + { + putc(c, out1); + if (out2) + putc(c, out2); + ended = 1; + } + } + else if (c == '\n') + { + lineno++; + putc (c, out1); + if (out2) + putc (c, out2); + if (cplus_comment) + ended = 1; + else + c = getc (in); + } + else if (c == EOF) + fatal (_("unterminated comment")); + else + { + putc (c, out1); + if (out2) + putc (c, out2); + c = getc (in); + } + } +} + + +/* Dump the comment from FIN to FOUT. C is either `*' or `/', + depending upon the type of comments used. */ + +static inline void +copy_comment (FILE *fin, FILE *fout, int c) +{ + copy_comment2 (fin, fout, NULL, c); } @@ -179,11 +313,12 @@ reader (void) startval = NULL; /* start symbol not specified yet. */ #if 0 - translations = 0; /* initially assume token number translation not needed. */ + /* initially assume token number translation not needed. */ + translations = 0; #endif - /* Nowadays translations is always set to 1, - since we give `error' a user-token-number - to satisfy the Posix demand for YYERRCODE==256. */ + /* Nowadays translations is always set to 1, since we give `error' a + user-token-number to satisfy the Posix demand for YYERRCODE==256. + */ translations = 1; nsyms = 1; @@ -191,7 +326,7 @@ reader (void) nrules = 0; nitems = 0; rline_allocated = 10; - rline = NEW2(rline_allocated, short); + rline = NEW2 (rline_allocated, short); typed = 0; lastprec = 0; @@ -204,50 +339,56 @@ reader (void) grammar = NULL; - init_lex(); + init_lex (); lineno = 1; - /* initialize the symbol table. */ - tabinit(); - /* construct the error token */ - errtoken = getsym("error"); + /* Initialize the symbol table. */ + tabinit (); + /* Construct the error token */ + errtoken = getsym ("error"); errtoken->class = STOKEN; - 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("$undefined."); + errtoken->user_token_number = 256; /* Value specified by POSIX. */ + /* Construct a token that represents all undefined literal tokens. + It is always token number 2. */ + undeftoken = getsym ("$undefined."); undeftoken->class = STOKEN; undeftoken->user_token_number = 2; - /* Read the declaration section. Copy %{ ... %} groups to ftable and fdefines file. - Also notice any %token, %left, etc. found there. */ - 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(); - /* 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. */ + + /* Read the declaration section. Copy %{ ... %} groups to FTABLE + and FDEFINES file. Also notice any %token, %left, etc. found + there. */ + putc ('\n', ftable); + fprintf (ftable, "\ +/* %s, made from %s\n\ + by GNU bison %s. */\n\ +\n", + noparserflag ? "Bison-generated parse tables" : "A Bison parser", + infile, + VERSION); + + fputs ("#define YYBISON 1 /* Identify Bison output. */\n\n", ftable); + read_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 (); + /* Now we know whether we need the line-number stack. If we do, + write its type into the .tab.h file. */ if (fdefines) - reader_output_yylsp(fdefines); - /* write closing delimiters for actions and guards. */ - output_trailers(); + reader_output_yylsp (fdefines); + /* Write closing delimiters for actions and guards. */ + output_trailers (); if (yylsp_needed) - fprintf(ftable, "#define YYLSP_NEEDED\n\n"); - /* assign the symbols their symbol numbers. - Write #defines for the token symbols into fdefines if requested. */ - packsymbols(); - /* convert the grammar into the format described in gram.h. */ - packgram(); - /* free the symbol table data structure - since symbols are now all referred to by symbol number. */ - free_symtab(); + fputs ("#define YYLSP_NEEDED\n\n", ftable); + /* Assign the symbols their symbol numbers. Write #defines for the + token symbols into FDEFINES if requested. */ + packsymbols (); + /* Convert the grammar into the format described in gram.h. */ + packgram (); + /* Free the symbol table data structure since symbols are now all + referred to by symbol number. */ + free_symtab (); } void @@ -261,7 +402,7 @@ reader_output_yylsp (FILE *f) `%' declarations, and copy the contents of any `%{ ... %}' groups to fattrs. */ -void +static void read_declarations (void) { register int c; @@ -301,7 +442,7 @@ read_declarations (void) break; case UNION: - parse_union_decl(); + parse_union_decl (); break; case EXPECT: @@ -338,39 +479,37 @@ read_declarations (void) break; default: - warns(_("unrecognized: %s"), token_buffer); + complain (_("unrecognized: %s"), token_buffer); skip_to_char('%'); } } else if (c == EOF) - fatal(_("no input grammar")); + fatal (_("no input grammar")); else { - warns (_("unknown character: %s"), printable_version(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 +static void 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 (;;) { @@ -387,89 +526,20 @@ copy_definition (void) case '\'': case '"': - match = c; - putc(c, fattrs); - c = getc(finput); - - while (c != match) - { - if (c == EOF) - fatal(_("unterminated string at end of file")); - if (c == '\n') - { - warn(_("unterminated string")); - ungetc(c, finput); - c = match; - continue; - } - - putc(c, fattrs); - - if (c == '\\') - { - c = getc(finput); - if (c == EOF) - fatal(_("unterminated string at end of file")); - 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); @@ -495,7 +565,7 @@ copy_definition (void) For %token, what_is is STOKEN and what_is_not is SNTERM. For %nterm, the arguments are reversed. */ -void +static void parse_token_decl (int what_is, int what_is_not) { register int token = 0; @@ -510,7 +580,7 @@ parse_token_decl (int what_is, int what_is_not) if (tmp_char == '%') return; if (tmp_char == EOF) - fatals ("Premature EOF after %s", token_buffer); + fatal (_("Premature EOF after %s"), token_buffer); token = lex(); if (token == COMMA) @@ -547,7 +617,7 @@ parse_token_decl (int what_is, int what_is_not) symbol = symval; if (symbol->class == what_is_not) - warns(_("symbol %s redefined"), symbol->tag); + complain (_("symbol %s redefined"), symbol->tag); symbol->class = what_is; if (what_is == SNTERM && oldclass != SNTERM) symbol->value = nvars++; @@ -557,7 +627,7 @@ parse_token_decl (int what_is, int what_is_not) if (symbol->type_name == NULL) symbol->type_name = typename; else if (strcmp(typename, symbol->type_name) != 0) - warns(_("type redeclaration for %s"), symbol->tag); + complain (_("type redeclaration for %s"), symbol->tag); } } else if (symbol && token == NUMBER) @@ -567,8 +637,8 @@ parse_token_decl (int what_is, int what_is_not) } else { - warnss(_("`%s' is invalid in %s"), - token_buffer, + complain (_("`%s' is invalid in %s"), + token_buffer, (what_is == STOKEN) ? "%token" : "%nterm"); skip_to_char('%'); } @@ -593,7 +663,7 @@ parse_token_decl (int what_is, int what_is_not) it is the literal string that is output to yytname */ -void +static void parse_thong_decl (void) { register int token; @@ -615,8 +685,8 @@ parse_thong_decl (void) if (token != IDENTIFIER) { - warns(_("unrecognized item %s, expected an identifier"), - token_buffer); + complain (_("unrecognized item %s, expected an identifier"), + token_buffer); skip_to_char('%'); return; } @@ -637,8 +707,8 @@ parse_thong_decl (void) if (token != IDENTIFIER || *symval->tag != '\"') { - warns(_("expected string constant instead of %s"), - token_buffer); + complain (_("expected string constant instead of %s"), + token_buffer); skip_to_char('%'); return; } @@ -653,15 +723,15 @@ parse_thong_decl (void) } -/* parse what comes after %start */ +/* Parse what comes after %start */ -void +static void parse_start_decl (void) { if (start_flag) - warn(_("multiple %start declarations")); - if (lex() != IDENTIFIER) - warn(_("invalid %start declaration")); + complain (_("multiple %s declarations"), "%start"); + if (lex () != IDENTIFIER) + complain (_("invalid %s declaration"), "%start"); else { start_flag = 1; @@ -673,7 +743,7 @@ parse_start_decl (void) /* read in a %type declaration and record its information for get_type_name to access */ -void +static void parse_type_decl (void) { register int k; @@ -681,7 +751,7 @@ parse_type_decl (void) if (lex() != TYPENAME) { - warn(_("%type declaration has no ")); + complain ("%s", _("%type declaration has no ")); skip_to_char('%'); return; } @@ -698,7 +768,7 @@ parse_type_decl (void) if (tmp_char == '%') return; if (tmp_char == EOF) - fatals ("Premature EOF after %s", token_buffer); + fatal (_("Premature EOF after %s"), token_buffer); t = lex(); @@ -713,12 +783,13 @@ parse_type_decl (void) if (symval->type_name == NULL) symval->type_name = name; else if (strcmp(name, symval->type_name) != 0) - warns(_("type redeclaration for %s"), symval->tag); + complain (_("type redeclaration for %s"), symval->tag); break; default: - warns(_("invalid %%type declaration due to item: `%s'"), token_buffer); + complain (_("invalid %%type declaration due to item: %s"), + token_buffer); skip_to_char('%'); } } @@ -729,7 +800,7 @@ parse_type_decl (void) /* read in a %left, %right or %nonassoc declaration and record its information. */ /* assoc is either LEFT_ASSOC, RIGHT_ASSOC or NON_ASSOC. */ -void +static void parse_assoc_decl (int assoc) { register int k; @@ -746,7 +817,7 @@ parse_assoc_decl (int assoc) if (tmp_char == '%') return; if (tmp_char == EOF) - fatals ("Premature EOF after %s", token_buffer); + fatal (_("Premature EOF after %s"), token_buffer); t = lex(); @@ -764,18 +835,18 @@ parse_assoc_decl (int assoc) case IDENTIFIER: if (symval->prec != 0) - warns(_("redefining precedence of %s"), symval->tag); + complain (_("redefining precedence of %s"), symval->tag); symval->prec = lastprec; symval->assoc = assoc; if (symval->class == SNTERM) - warns(_("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 if (strcmp(name, symval->type_name) != 0) - warns(_("type redeclaration for %s"), symval->tag); + complain (_("type redeclaration for %s"), symval->tag); } break; @@ -787,7 +858,7 @@ parse_assoc_decl (int assoc) } else { - warns(_("invalid text (%s) - number should be after identifier"), + complain (_("invalid text (%s) - number should be after identifier"), token_buffer); skip_to_char('%'); } @@ -797,7 +868,7 @@ parse_assoc_decl (int assoc) return; default: - warns(_("unexpected item: %s"), token_buffer); + complain (_("unexpected item: %s"), token_buffer); skip_to_char('%'); } @@ -812,38 +883,33 @@ parse_assoc_decl (int assoc) where it is made into the definition of YYSTYPE, the type of elements of the parser value stack. */ -void +static void parse_union_decl (void) { register int c; - register int count; - register int in_comment; - int cplus_comment; + register int count = 0; if (typed) - warn(_("multiple %union declarations")); + complain (_("multiple %s declarations"), "%union"); typed = 1; if (!nolinesflag) - fprintf(fattrs, "\n#line %d \"%s\"\n", lineno, infile); + fprintf (fattrs, "\n#line %d \"%s\"\n", lineno, infile); else - fprintf(fattrs, "\n"); + fprintf (fattrs, "\n"); - fprintf(fattrs, "typedef union"); + fprintf (fattrs, "typedef union"); if (fdefines) - fprintf(fdefines, "typedef union"); + fprintf (fdefines, "typedef union"); - count = 0; - in_comment = 0; - - c = getc(finput); + c = getc (finput); while (c != EOF) { - putc(c, fattrs); + putc (c, fattrs); if (fdefines) - putc(c, fdefines); + putc (c, fdefines); switch (c) { @@ -852,50 +918,10 @@ parse_union_decl (void) break; case '/': - c = getc(finput); + c = getc (finput); if (c != '*' && c != '/') - ungetc(c, finput); - else - { - putc(c, fattrs); - if (fdefines) - putc(c, fdefines); - cplus_comment = (c == '/'); - in_comment = 1; - c = getc(finput); - while (in_comment) - { - putc(c, fattrs); - if (fdefines) - putc(c, fdefines); - - if (c == '\n') - { - lineno++; - if (cplus_comment) - { - in_comment = 0; - break; - } - } - if (c == EOF) - fatal(_("unterminated comment at end of file")); - - if (!cplus_comment && c == '*') - { - c = getc(finput); - if (c == '/') - { - putc('/', fattrs); - if (fdefines) - putc('/', fdefines); - in_comment = 0; - } - } - else - c = getc(finput); - } - } + continue; + copy_comment2 (finput, fattrs, fdefines, c); break; @@ -905,28 +931,29 @@ parse_union_decl (void) case '}': if (count == 0) - warn (_("unmatched close-brace (`}')")); + complain (_("unmatched %s"), "`}'"); count--; if (count <= 0) { - fprintf(fattrs, " YYSTYPE;\n"); + fprintf (fattrs, " YYSTYPE;\n"); if (fdefines) - fprintf(fdefines, " YYSTYPE;\n"); + fprintf (fdefines, " YYSTYPE;\n"); /* JF don't choke on trailing semi */ - c=skip_white_space(); - if(c!=';') ungetc(c,finput); + c = skip_white_space (); + if (c != ';') + ungetc (c,finput); return; } } - c = getc(finput); + c = getc (finput); } } /* parse the declaration %expect N which says to expect N shift-reduce conflicts. */ -void +static void parse_expect_decl (void) { register int c; @@ -949,25 +976,54 @@ parse_expect_decl (void) ungetc (c, finput); if (count <= 0 || count > 10) - warn(_("argument of %expect is not an integer")); + 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. */ +/* FIN is pointing to a location (i.e., a `@'). Output to FOUT + a reference to this location. STACK_OFFSET is the number of values + in the current rule so far, which says where to find `$0' with + respect to the top of the stack. */ +static inline void +copy_at (FILE *fin, FILE *fout, int stack_offset) +{ + int c; -char * + c = getc (fin); + if (c == '$') + { + fprintf (fout, "yyloc"); + yylsp_needed = 1; + } + else if (isdigit(c) || c == '-') + { + int n; + + ungetc (c, fin); + n = read_signed_integer (fin); + + fprintf (fout, "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. */ + +static char * get_type_name (int n, symbol_list *rule) { - static char *msg = N_("invalid $ value"); - register int i; register symbol_list *rp; if (n < 0) { - warn(_(msg)); + complain (_("invalid $ value")); return NULL; } @@ -979,16 +1035,17 @@ get_type_name (int n, symbol_list *rule) rp = rp->next; if (rp == NULL || rp->sym == NULL) { - warn(_(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 @@ -996,17 +1053,14 @@ get_type_name (int n, symbol_list *rule) 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 +static void 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; @@ -1040,92 +1094,22 @@ copy_guard (symbol_list *rule, int stack_offset) count--; else { - warn(_("unmatched right brace (`}')")); + complain (_("unmatched %s"), "`}'"); c = getc(finput); /* skip it */ } break; case '\'': case '"': - match = c; - putc(c, fguard); - c = getc(finput); - - while (c != match) - { - if (c == EOF) - fatal(_("unterminated string at end of file")); - if (c == '\n') - { - warn(_("unterminated string")); - ungetc(c, finput); - c = match; /* invent terminator */ - continue; - } - - 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 '$': @@ -1157,7 +1141,8 @@ copy_guard (symbol_list *rule, int stack_offset) if (type_name) fprintf(fguard, ".%s", type_name); if(!type_name && typed) - warns(_("$$ of `%s' has no declared type"), rule->sym->tag); + complain (_("$$ of `%s' has no declared type"), + rule->sym->tag); } else if (isdigit(c) || c == '-') { @@ -1172,39 +1157,21 @@ copy_guard (symbol_list *rule, int stack_offset) if (type_name) fprintf(fguard, ".%s", type_name); if (!type_name && typed) - warnss (_("$%s of `%s' has no declared type"), - int_to_string(n), rule->sym->tag); + complain (_("$%d of `%s' has no declared type"), + n, rule->sym->tag); continue; } else - warns(_("$%s is invalid"), printable_version(c)); + complain (_("$%s is invalid"), printable_version (c)); break; case '@': - c = getc (finput); - if (c == '$') - { - fprintf (fguard, "yyloc"); - yylsp_needed = 1; - } - else if (isdigit(c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer (finput); - c = getc (finput); - fprintf (fguard, "yylsp[%d]", n - stack_offset); - yylsp_needed = 1; - continue; - } - else - { - warns (_("@%s is invalid"), printable_version (c)); - n = 1; - } + copy_at (finput, fguard, stack_offset); break; case EOF: - fatal (_("unterminated %%guard clause")); + fatal ("%s", + _("unterminated %guard clause")); default: putc (c, fguard); @@ -1218,12 +1185,12 @@ copy_guard (symbol_list *rule, 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); /* why not skip_white_space -wjh */ if (c == '{') - copy_action(rule, stack_offset); + copy_action (rule, stack_offset); } else ungetc(c, finput); @@ -1236,16 +1203,13 @@ copy_guard (symbol_list *rule, int stack_offset) values in the current rule so far, which says where to find `$0' with respect to the top of the stack. */ -void +static void copy_action (symbol_list *rule, int stack_offset) { register int c; register int n; register int count; - register int match; - register int ended; register char *type_name; - int cplus_comment; /* offset is always 0 if parser has already popped the stack pointer */ if (semantic_parser) @@ -1277,85 +1241,15 @@ copy_action (symbol_list *rule, int stack_offset) case '\'': case '"': - match = c; - putc(c, faction); - c = getc(finput); - - while (c != match) - { - if (c == '\n') - { - warn(_("unterminated string")); - ungetc(c, finput); - c = match; - continue; - } - else if (c == EOF) - fatal(_("unterminated string at end of file")); - - putc(c, faction); - - if (c == '\\') - { - c = getc(finput); - if (c == EOF) - fatal(_("unterminated string")); - putc(c, faction); - if (c == '\n') - lineno++; - } - - c = getc(finput); - } - - putc(c, faction); + copy_string (finput, faction, c); break; case '/': - putc(c, faction); - c = getc(finput); + putc (c, faction); + c = getc (finput); if (c != '*' && c != '/') continue; - - cplus_comment = (c == '/'); - putc(c, faction); - c = getc(finput); - - ended = 0; - while (!ended) - { - if (!cplus_comment && c == '*') - { - while (c == '*') - { - putc(c, faction); - c = getc(finput); - } - - if (c == '/') - { - putc(c, faction); - ended = 1; - } - } - else if (c == '\n') - { - lineno++; - putc(c, faction); - if (cplus_comment) - ended = 1; - else - c = getc(finput); - } - else if (c == EOF) - fatal(_("unterminated comment")); - else - { - putc(c, faction); - c = getc(finput); - } - } - + copy_comment (finput, faction, c); break; case '$': @@ -1387,8 +1281,8 @@ copy_action (symbol_list *rule, int stack_offset) if (type_name) fprintf(faction, ".%s", type_name); if(!type_name && typed) - warns(_("$$ of `%s' has no declared type"), - rule->sym->tag); + complain (_("$$ of `%s' has no declared type"), + rule->sym->tag); } else if (isdigit(c) || c == '-') { @@ -1403,40 +1297,21 @@ copy_action (symbol_list *rule, int stack_offset) if (type_name) fprintf(faction, ".%s", type_name); if(!type_name && typed) - warnss(_("$%s of `%s' has no declared type"), - int_to_string(n), rule->sym->tag); + complain (_("$%d of `%s' has no declared type"), + n, rule->sym->tag); continue; } else - warns(_("$%s is invalid"), printable_version(c)); + complain (_("$%s is invalid"), printable_version (c)); break; case '@': - c = getc (finput); - if (c == '$') - { - fprintf (faction, "yyloc"); - yylsp_needed = 1; - } - else if (isdigit(c) || c == '-') - { - ungetc (c, finput); - n = read_signed_integer (finput); - c = getc (finput); - fprintf (faction, "yylsp[%d]", n - stack_offset); - yylsp_needed = 1; - continue; - } - else - { - warns (_("@%s is invalid"), printable_version (c)); - n = 1; - } + copy_at (finput, faction, stack_offset); break; case EOF: - fatal(_("unmatched `{'")); + fatal (_("unmatched %s"), "`{'"); default: putc(c, faction); @@ -1462,7 +1337,7 @@ copy_action (symbol_list *rule, int stack_offset) /* generate a dummy symbol, a nonterminal, whose name cannot conflict with the user's names. */ -bucket * +static bucket * gensym (void) { register bucket *sym; @@ -1471,7 +1346,7 @@ gensym (void) sym = getsym(token_buffer); sym->class = SNTERM; sym->value = nvars++; - return (sym); + return sym; } /* Parse the input grammar into a one symbol_list structure. @@ -1483,7 +1358,7 @@ The next symbol is the lhs of the following rule. All guards and actions are copied out to the appropriate files, labelled by the rule number they apply to. */ -void +static void readgram (void) { register int t; @@ -1522,14 +1397,14 @@ readgram (void) t = lex(); if (t != COLON) { - warn(_("ill-formed rule: initial symbol not followed by colon")); + complain (_("ill-formed rule: initial symbol not followed by colon")); unlex(t); } } if (nrules == 0 && t == BAR) { - warn(_("grammar starts with vertical bar")); + complain (_("grammar starts with vertical bar")); lhs = symval; /* BOGUS: use a random symval */ } /* start a new rule and record its lhs. */ @@ -1560,7 +1435,7 @@ readgram (void) nvars++; } else if (lhs->class == STOKEN) - warns(_("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. */ @@ -1658,7 +1533,7 @@ readgram (void) if (t == PREC) { - warn(_("two @prec's in a row")); + complain (_("two @prec's in a row")); t = lex(); crule->ruleprec = symval; t = lex(); @@ -1666,7 +1541,8 @@ readgram (void) if (t == GUARD) { if (! semantic_parser) - warn(_("%%guard present but %%semantic_parser not specified")); + complain ("%s", + _("%guard present but %semantic_parser not specified")); copy_guard(crule, rulelength); t = lex(); @@ -1675,13 +1551,13 @@ readgram (void) { /* This case never occurs -wjh */ if (actionflag) - warn(_("two actions at end of one rule")); + complain (_("two actions at end of one rule")); copy_action(crule, rulelength); actionflag = 1; xactions++; /* -wjh */ t = lex(); } - /* If $$ is being set in default way, warn if any type + /* If $$ is being set in default way, report if any type mismatch. */ else if (!xactions && first_rhs @@ -1690,13 +1566,13 @@ readgram (void) if (lhs->type_name == 0 || first_rhs->type_name == 0 || strcmp(lhs->type_name,first_rhs->type_name)) - warnss(_("type clash (`%s' `%s') on default action"), - 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) - warn(_("empty rule for typed nonterminal, and no action")); + complain (_("empty rule for typed nonterminal, and no action")); if (t == SEMICOLON) t = lex(); } @@ -1740,7 +1616,7 @@ readgram (void) else { - warns(_("invalid input: %s"), token_buffer); + complain (_("invalid input: %s"), token_buffer); t = lex(); } } @@ -1748,10 +1624,10 @@ readgram (void) /* grammar has been read. Do some checking */ if (nsyms > MAXSHORT) - fatals(_("too many symbols (tokens plus nonterminals); maximum %s"), - int_to_string(MAXSHORT)); + fatal (_("too many symbols (tokens plus nonterminals); maximum %d"), + MAXSHORT); if (nrules == 0) - fatal(_("no rules in the input grammar")); + fatal (_("no rules in the input grammar")); if (typed == 0 /* JF put out same default YYSTYPE as YACC does */ && !value_components_used) @@ -1769,8 +1645,8 @@ readgram (void) for (bp = firstsymbol; bp; bp = bp->next) if (bp->class == SUNKNOWN) { - warns(_("symbol %s is used, but is not defined as a token and has no rules"), - bp->tag); + complain (_("symbol %s is used, but is not defined as a token and has no rules"), + bp->tag); bp->class = SNTERM; bp->value = nvars++; } @@ -1779,7 +1655,7 @@ readgram (void) } -void +static void record_rule_line (void) { /* Record each rule's source line number in rline table. */ @@ -1808,7 +1684,7 @@ get_type (void) if (t != TYPENAME) { - warn(_("ill-formed %type declaration")); + complain (_("invalid %s declaration"), "%type"); return t; } @@ -1823,7 +1699,7 @@ get_type (void) switch (t) { case SEMICOLON: - return (lex()); + return lex(); case COMMA: break; @@ -1832,33 +1708,35 @@ get_type (void) if (symval->type_name == NULL) symval->type_name = name; else if (strcmp(name, symval->type_name) != 0) - warns(_("type redeclaration for %s"), symval->tag); + 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 +static void packsymbols (void) { register bucket *bp; register int tokno = 1; register int i; register int last_user_token_number; + static char DOLLAR[] = "$"; /* int lossage = 0; JF set but not used */ tags = NEW2(nsyms + 1, char *); - tags[0] = "$"; + tags[0] = DOLLAR; user_toknums = NEW2(nsyms + 1, int); user_toknums[0] = 0; @@ -1876,34 +1754,38 @@ packsymbols (void) } 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) - warnss(_("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; - } + /* 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->assoc != bp->alias->assoc) { - if (bp->assoc != 0 && bp->alias->assoc != 0 - && bp->user_token_number == SALIAS) - warnss(_("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->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 */ + if (bp->user_token_number == SALIAS) + continue; /* do not do processing below for SALIASs */ } else /* bp->class == STOKEN */ @@ -1928,25 +1810,27 @@ packsymbols (void) if (translations) { - register int i; + register int j; token_translations = NEW2(max_user_token_number+1, short); - /* initialize all entries for literal tokens to 2, - the internal token number for $undefined., - which represents all invalid inputs. */ - for (i = 0; i <= max_user_token_number; i++) - token_translations[i] = 2; + /* initialize all entries for literal tokens to 2, the internal + token number for $undefined., which represents all invalid + inputs. */ + for (j = 0; j <= max_user_token_number; j++) + token_translations[j] = 2; for (bp = firstsymbol; bp; bp = bp->next) { - if (bp->value >= ntokens) continue; /* non-terminal */ - if (bp->user_token_number == SALIAS) continue; + if (bp->value >= ntokens) + continue; /* non-terminal */ + if (bp->user_token_number == SALIAS) + continue; if (token_translations[bp->user_token_number] != 2) - warnsss(_("tokens %s and %s both assigned number %s"), - tags[token_translations[bp->user_token_number]], - bp->tag, - int_to_string(bp->user_token_number)); + 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; } } @@ -1957,9 +1841,9 @@ packsymbols (void) 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; @@ -1991,10 +1875,10 @@ packsymbols (void) } } -/* For named tokens, but not literal ones, define the name. - The value is the user token number. -*/ -void +/* For named tokens, but not literal ones, define the name. The value + is the user token number. */ + +static void output_token_defines (FILE *file) { bucket *bp; @@ -2023,12 +1907,12 @@ output_token_defines (FILE *file) while ((c = *cp++) && c != '.'); if (c != '\0') continue; - fprintf(file, "#define\t%s\t%d\n", symbol, - ((translations && ! rawtoknumflag) - ? bp->user_token_number - : 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); + fprintf (file, "#define\tT%s\t%d\n", symbol, bp->value); } putc('\n', file); @@ -2038,7 +1922,7 @@ output_token_defines (FILE *file) /* convert the rules into the representation using rrhs, rlhs and ritems. */ -void +static void packgram (void) { register int itemno; @@ -2096,29 +1980,3 @@ packgram (void) ritem[itemno] = 0; } - -/* Read a signed integer from STREAM and return its value. */ - -int -read_signed_integer (FILE *stream) -{ - register int c = getc(stream); - register int sign = 1; - register int n; - - if (c == '-') - { - c = getc(stream); - sign = -1; - } - n = 0; - while (isdigit(c)) - { - n = 10*n + (c - '0'); - c = getc(stream); - } - - ungetc(c, stream); - - return n * sign; -}