version 1.6, 2002/10/04 20:27:16 |
version 1.7, 2002/12/02 07:16:23 |
|
|
/* $OpenBSD$ */ |
/* $OpenBSD$ */ |
/* $NetBSD: unifdef.c,v 1.6 1998/10/08 01:31:59 wsanchez Exp $ */ |
|
|
|
/* |
/* |
* Copyright (c) 1985, 1993 |
* Copyright (c) 1985, 1993 |
* The Regents of the University of California. All rights reserved. |
* The Regents of the University of California. All rights reserved. |
* |
* |
* This code is derived from software contributed to Berkeley by |
* This code is derived from software contributed to Berkeley by |
* Dave Yost. |
* Dave Yost. Support for #if and #elif was added by Tony Finch. |
* |
* |
* Redistribution and use in source and binary forms, with or without |
* Redistribution and use in source and binary forms, with or without |
* modification, are permitted provided that the following conditions |
* modification, are permitted provided that the following conditions |
|
|
*/ |
*/ |
|
|
#ifndef lint |
#ifndef lint |
static char copyright[] = |
static const char copyright[] = |
"@(#) Copyright (c) 1985, 1993\n\ |
"@(#) Copyright (c) 1985, 1993\n\ |
The Regents of the University of California. All rights reserved.\n"; |
The Regents of the University of California. All rights reserved.\n"; |
#endif /* not lint */ |
|
|
|
#ifndef lint |
|
#if 0 |
#if 0 |
static char sccsid[] = "@(#)unifdef.c 8.1 (Berkeley) 6/6/93"; |
static char sccsid[] = "@(#)unifdef.c 8.1 (Berkeley) 6/6/93"; |
#endif |
#endif |
static char rcsid[] = "$OpenBSD$"; |
static const char rcsid[] = "$OpenBSD$"; |
#endif /* not lint */ |
#endif |
|
|
/* |
/* |
* unifdef - remove ifdef'ed lines |
* unifdef - remove ifdef'ed lines |
* |
* |
* Warning: will not work correctly if input contains null characters. |
* Warning: will not work correctly if input contains nul characters. |
* |
* |
* Wishlist: |
* Wishlist: |
* provide an option which will append the name of the |
* provide an option which will append the name of the |
|
|
* provide an option which will check symbols after |
* provide an option which will check symbols after |
* #else's and #endif's to see that they match their |
* #else's and #endif's to see that they match their |
* corresponding #ifdef or #ifndef |
* corresponding #ifdef or #ifndef |
|
* generate #line directives in place of deleted code |
*/ |
*/ |
|
|
#include <stdio.h> |
|
#include <ctype.h> |
#include <ctype.h> |
|
#include <err.h> |
|
#include <stdarg.h> |
|
#include <stdbool.h> |
|
#include <stdio.h> |
|
#include <stdlib.h> |
|
#include <string.h> |
|
#include <unistd.h> |
|
|
FILE *input; |
/* types of input lines: */ |
#ifndef YES |
typedef enum { |
#define YES 1 |
LT_PLAIN, /* ordinary line */ |
#define NO 0 |
LT_TRUE, /* a true #if */ |
#endif /* YES */ |
LT_FALSE, /* a false #if */ |
#define C_COMMENT 1 |
LT_ELTRUE, /* a true #elif */ |
#define CXX_COMMENT 2 |
LT_ELFALSE, /* a false #elif */ |
typedef int Bool; |
LT_IF, /* an unknown #if */ |
|
LT_ELIF, /* an unknown #elif */ |
|
LT_ELSE, /* #else */ |
|
LT_ENDIF, /* #endif */ |
|
LT_EOF /* end of file */ |
|
} Linetype; |
|
|
char *progname; |
typedef enum { /* 0 or 1: pass thru; 1 or 2: ignore comments */ |
char *filename; |
REJ_NO, |
char text; /* -t option in effect: this is a text file */ |
REJ_IGNORE, |
char lnblank; /* -l option in effect: blank deleted lines */ |
REJ_YES |
char complement; /* -c option in effect: complement the |
} Reject_level; |
* operation */ |
|
|
|
#define MAXSYMS 100 |
typedef enum { |
char *symname[MAXSYMS]; /* symbol name */ |
NO_COMMENT = false, |
char true[MAXSYMS]; /* -Dsym */ |
C_COMMENT, |
char ignore[MAXSYMS]; /* -iDsym or -iUsym */ |
CXX_COMMENT |
char insym[MAXSYMS]; /* state: false, inactive, true */ |
} Comment_state; |
#define SYM_INACTIVE 0 /* symbol is currently inactive */ |
|
#define SYM_FALSE 1 /* symbol is currently false */ |
|
#define SYM_TRUE 2 /* symbol is currently true */ |
|
|
|
char nsyms; |
typedef enum { |
char incomment; /* inside C comment */ |
QUOTE_NONE = false, |
|
QUOTE_SINGLE, |
|
QUOTE_DOUBLE |
|
} Quote_state; |
|
|
#define QUOTE_NONE 0 |
const char *const errs[] = { |
#define QUOTE_SINGLE 1 |
|
#define QUOTE_DOUBLE 2 |
|
char inquote; /* inside single or double quotes */ |
|
int exitstat; |
|
|
|
int error(int, int, int); |
|
int findsym(char *); |
|
void flushline(Bool); |
|
int getlin(char *, int, FILE *, int); |
|
void pfile(void); |
|
void prname(void); |
|
char *skipcomment(char *); |
|
char *skipquote(char *, int); |
|
|
|
int |
|
main(argc, argv) |
|
int argc; |
|
char **argv; |
|
{ |
|
char **curarg; |
|
char *cp; |
|
char *cp1; |
|
char ignorethis; |
|
|
|
progname = argv[0][0] ? argv[0] : "unifdef"; |
|
|
|
for (curarg = &argv[1]; --argc > 0; curarg++) { |
|
if (*(cp1 = cp = *curarg) != '-') |
|
break; |
|
if (*++cp1 == 'i') { |
|
ignorethis = YES; |
|
cp1++; |
|
} else |
|
ignorethis = NO; |
|
if ((*cp1 == 'D' |
|
|| *cp1 == 'U' |
|
) |
|
&& cp1[1] != '\0' |
|
) { |
|
int symind; |
|
|
|
if ((symind = findsym(&cp1[1])) < 0) { |
|
if (nsyms >= MAXSYMS) { |
|
prname(); |
|
fprintf(stderr, "too many symbols.\n"); |
|
exit(2); |
|
} |
|
symind = nsyms++; |
|
symname[symind] = &cp1[1]; |
|
insym[symind] = SYM_INACTIVE; |
|
} |
|
ignore[symind] = ignorethis; |
|
true[symind] = *cp1 == 'D' ? YES : NO; |
|
} else |
|
if (ignorethis) |
|
goto unrec; |
|
else |
|
if (strcmp(&cp[1], "t") == 0) |
|
text = YES; |
|
else |
|
if (strcmp(&cp[1], "l") == 0) |
|
lnblank = YES; |
|
else |
|
if (strcmp(&cp[1], "c") == 0) |
|
complement = YES; |
|
else { |
|
unrec: |
|
prname(); |
|
fprintf(stderr, "unrecognized option: %s\n", cp); |
|
goto usage; |
|
} |
|
} |
|
if (nsyms == 0) { |
|
usage: |
|
fprintf(stderr, "\ |
|
Usage: %s [-l] [-t] [-c] [[-Dsym] [-Usym] [-iDsym] [-iUsym]]... [file]\n\ |
|
At least one arg from [-D -U -iD -iU] is required\n", progname); |
|
exit(2); |
|
} |
|
if (argc > 1) { |
|
prname(); |
|
fprintf(stderr, "can only do one file.\n"); |
|
} else |
|
if (argc == 1) { |
|
filename = *curarg; |
|
if ((input = fopen(filename, "r")) != NULL) { |
|
pfile(); |
|
(void) fclose(input); |
|
} else { |
|
prname(); |
|
fprintf(stderr, "can't open "); |
|
perror(*curarg); |
|
} |
|
} else { |
|
filename = "[stdin]"; |
|
input = stdin; |
|
pfile(); |
|
} |
|
|
|
(void) fflush(stdout); |
|
exit(exitstat); |
|
} |
|
/* types of input lines: */ |
|
typedef int Linetype; |
|
#define LT_PLAIN 0 /* ordinary line */ |
|
#define LT_TRUE 1 /* a true #ifdef of a symbol known to us */ |
|
#define LT_FALSE 2 /* a false #ifdef of a symbol known to us */ |
|
#define LT_OTHER 3 /* an #ifdef of a symbol not known to us */ |
|
#define LT_IF 4 /* an #ifdef of a symbol not known to us */ |
|
#define LT_ELSE 5 /* #else */ |
|
#define LT_ENDIF 6 /* #endif */ |
|
#define LT_LEOF 7 /* end of file */ |
|
Linetype checkline(int *); |
|
|
|
typedef int Reject_level; |
|
Reject_level reject; /* 0 or 1: pass thru; 1 or 2: ignore comments */ |
|
#define REJ_NO 0 |
|
#define REJ_IGNORE 1 |
|
#define REJ_YES 2 |
|
int doif(int, int, Reject_level, int); |
|
|
|
int linenum; /* current line number */ |
|
int stqcline; /* start of current coment or quote */ |
|
char *errs[] = { |
|
#define NO_ERR 0 |
#define NO_ERR 0 |
"", |
"", |
#define END_ERR 1 |
#define END_ERR 1 |
"", |
"", |
#define ELSE_ERR 2 |
#define ELIF_ERR 2 |
|
"Inappropriate elif", |
|
#define ELSE_ERR 3 |
"Inappropriate else", |
"Inappropriate else", |
#define ENDIF_ERR 3 |
#define ENDIF_ERR 4 |
"Inappropriate endif", |
"Inappropriate endif", |
#define IEOF_ERR 4 |
#define IEOF_ERR 5 |
"Premature EOF in ifdef", |
"Premature EOF in ifdef", |
#define CEOF_ERR 5 |
#define CEOF_ERR 6 |
"Premature EOF in comment", |
"Premature EOF in comment", |
#define Q1EOF_ERR 6 |
#define Q1EOF_ERR 7 |
"Premature EOF in quoted character", |
"Premature EOF in quoted character", |
#define Q2EOF_ERR 7 |
#define Q2EOF_ERR 8 |
"Premature EOF in quoted string" |
"Premature EOF in quoted string" |
}; |
}; |
/* States for inif arg to doif */ |
|
#define IN_NONE 0 |
|
#define IN_IF 1 |
|
#define IN_ELSE 2 |
|
|
|
|
/* |
|
* These are the operators that are supported by the expression evaluator. |
|
*/ |
|
static int op_lt(int a, int b) { return a < b; } |
|
static int op_gt(int a, int b) { return a > b; } |
|
static int op_le(int a, int b) { return a <= b; } |
|
static int op_ge(int a, int b) { return a >= b; } |
|
static int op_eq(int a, int b) { return a == b; } |
|
static int op_ne(int a, int b) { return a != b; } |
|
static int op_or(int a, int b) { return a || b; } |
|
static int op_and(int a, int b) { return a && b; } |
|
|
|
struct ops; |
|
|
|
/* |
|
* An evaluation function takes three arguments, as follows: (1) a pointer to |
|
* an element of the precedence table which lists the operators at the current |
|
* level of precedence; (2) a pointer to an integer which will receive the |
|
* value of the expression; and (3) a pointer to a char* that points to the |
|
* expression to be evaluated and that is updated to the end of the expression |
|
* when evaluation is complete. The function returns LT_FALSE if the value of |
|
* the expression is zero, LT_TRUE if it is non-zero, or LT_IF if the |
|
* expression could not be evaluated. |
|
*/ |
|
typedef Linetype eval_fn(struct ops *, int *, const char **); |
|
|
|
eval_fn eval_table, eval_unary; |
|
|
|
/* |
|
* The precedence table. Expressions involving binary operators are evaluated |
|
* in a table-driven way by eval_table. When it evaluates a subexpression it |
|
* calls the inner function with its first argument pointing to the next |
|
* element of the table. Innermost expressions have special non-table-driven |
|
* handling. |
|
*/ |
|
struct ops { |
|
eval_fn *inner; |
|
struct op { |
|
const char *str; |
|
int (*fn)(int, int); |
|
} op[5]; |
|
} eval_ops[] = { |
|
{ eval_table, { { "||", op_or } } }, |
|
{ eval_table, { { "&&", op_and } } }, |
|
{ eval_table, { { "==", op_eq }, |
|
{ "!=", op_ne } } }, |
|
{ eval_unary, { { "<=", op_le }, |
|
{ ">=", op_ge }, |
|
{ "<", op_lt }, |
|
{ ">", op_gt } } } |
|
}; |
|
|
|
FILE *input; |
|
const char *filename; |
|
int linenum; /* current line number */ |
|
int stifline; /* start of current #if */ |
|
int stqcline; /* start of current coment or quote */ |
|
bool keepthis; /* ignore this #if's value 'cause it's const */ |
|
|
|
#define MAXLINE 1024 |
|
#define KWSIZE 8 |
|
/* tline has extra space so that it isn't overflowed when editing #elifs */ |
|
char tline[MAXLINE+KWSIZE]; /* input buffer */ |
|
char *keyword; /* used for editing #elif's */ |
|
|
|
bool complement; /* -c option in effect: do the complement */ |
|
bool debugging; /* -d option in effect: debugging reports */ |
|
bool killconsts; /* -k option in effect: eval constant #ifs */ |
|
bool lnblank; /* -l option in effect: blank deleted lines */ |
|
bool symlist; /* -s option in effect: output symbol list */ |
|
bool text; /* -t option in effect: this is a text file */ |
|
|
|
#define MAXSYMS 1000 |
|
const char *symname[MAXSYMS]; /* symbol name */ |
|
const char *value[MAXSYMS]; /* -Dsym=value */ |
|
bool ignore[MAXSYMS]; /* -iDsym or -iUsym */ |
|
|
|
int nsyms = 1; /* symbol 0 is used for tracking #ifs */ |
|
|
|
Reject_level reject; /* what kind of filtering we are doing */ |
|
Comment_state incomment; /* inside C comment */ |
|
Quote_state inquote; /* inside single or double quotes */ |
|
|
|
Linetype checkline(int *); |
|
void debug(const char *, ...); |
|
Linetype process(int); |
|
void doif(int, Linetype, bool); |
|
void elif2if(void); |
|
void elif2endif(void); |
|
void error(int, int); |
|
void addsym(bool, bool, char *); |
|
int findsym(const char *); |
|
void flushline(bool); |
|
#if 0 |
|
int getline(char *, int, FILE *, bool); |
|
#endif |
|
Linetype ifeval(const char **); |
|
const char *skipcomment(const char *); |
|
const char *skipquote(const char *, Quote_state); |
|
const char *skipsym(const char *); |
|
void usage(void); |
|
|
|
#define endsym(c) (!isalpha((unsigned char)c) && !isdigit((unsigned char)c) && c != '_') |
|
|
|
int |
|
main(int argc, char *argv[]) |
|
{ |
|
int opt; |
|
|
|
while ((opt = getopt(argc, argv, "i:D:U:I:cdklst")) != -1) |
|
switch (opt) { |
|
case 'i': /* treat stuff controlled by these symbols as text */ |
|
/* |
|
* For strict backwards-compatibility the U or D |
|
* should be immediately after the -i but it doesn't |
|
* matter much if we relax that requirement. |
|
*/ |
|
opt = *optarg++; |
|
if (opt == 'D') |
|
addsym(true, true, optarg); |
|
else if (opt == 'U') |
|
addsym(true, false, optarg); |
|
else |
|
usage(); |
|
break; |
|
case 'D': /* define a symbol */ |
|
addsym(false, true, optarg); |
|
break; |
|
case 'U': /* undef a symbol */ |
|
addsym(false, false, optarg); |
|
break; |
|
case 'I': |
|
/* ignore for compatibility with cpp */ |
|
break; |
|
case 'c': /* treat -D as -U and vice versa */ |
|
complement = true; |
|
break; |
|
case 'k': /* process constant #ifs */ |
|
killconsts = true; |
|
break; |
|
case 'd': |
|
debugging = true; |
|
break; |
|
case 'l': /* blank deleted lines instead of omitting them */ |
|
lnblank = true; |
|
break; |
|
case 's': /* only output list of symbols that control #ifs */ |
|
symlist = true; |
|
break; |
|
case 't': /* don't parse C comments or strings */ |
|
text = true; |
|
break; |
|
default: |
|
usage(); |
|
} |
|
argc -= optind; |
|
argv += optind; |
|
if (nsyms == 1 && !symlist) { |
|
warnx("must -D or -U at least one symbol"); |
|
usage(); |
|
} |
|
if (argc > 1) { |
|
errx(2, "can only do one file"); |
|
} else if (argc == 1 && strcmp(*argv, "-") != 0) { |
|
filename = *argv; |
|
if ((input = fopen(filename, "r")) != NULL) { |
|
(void) process(0); |
|
(void) fclose(input); |
|
} else |
|
err(2, "can't open %s", *argv); |
|
} else { |
|
filename = "[stdin]"; |
|
input = stdin; |
|
(void) process(0); |
|
} |
|
|
|
exit(0); |
|
} |
|
|
void |
void |
pfile() |
usage(void) |
{ |
{ |
reject = REJ_NO; |
fprintf (stderr, "usage: unifdef [-cdklst] [[-Dsym[=val]]" |
(void) doif(-1, IN_NONE, reject, 0); |
"[-Usym] [-iDsym[=val]] [-iUsym]] ... [file]\n"); |
return; |
exit (2); |
} |
} |
|
|
int |
/* |
doif(thissym, inif, prevreject, depth) |
* This function processes #if lines and alters the pass-through |
int thissym; /* index of the symbol who was last ifdef'ed */ |
* state accordingly. All the complicated state transition suff is |
int inif; /* YES or NO we are inside an ifdef */ |
* dealt with in this function, as well as checking that the |
Reject_level prevreject;/* previous value of reject */ |
* #if/#elif/#else/#endif lines happen in the correct order. Lines |
int depth; /* depth of ifdef's */ |
* between #if lines are handled by a recursive call to process(). |
|
*/ |
|
void |
|
doif(int depth, Linetype lineval, bool ignoring) |
{ |
{ |
Linetype lineval; |
Reject_level savereject; |
Reject_level thisreject; |
bool active; |
int doret; /* tmp return value of doif */ |
bool donetrue; |
int cursym; /* index of the symbol returned by checkline */ |
bool inelse; |
int stline; /* line number when called this time */ |
int saveline; |
|
|
stline = linenum; |
debug("#if line %d code %d depth %d", |
|
linenum, lineval, depth); |
|
saveline = stifline; |
|
stifline = linenum; |
|
savereject = reject; |
|
inelse = false; |
|
donetrue = false; |
|
if (lineval == LT_IF || reject != REJ_NO) { |
|
active = false; |
|
ignoring = false; |
|
flushline(true); |
|
} else if (ignoring) { |
|
active = false; |
|
flushline(true); |
|
if (lineval == LT_FALSE) |
|
reject = REJ_IGNORE; |
|
else |
|
donetrue = true; |
|
} else { |
|
active = true; |
|
flushline(false); |
|
if (lineval == LT_FALSE) |
|
reject = REJ_YES; |
|
else |
|
donetrue = true; |
|
} |
|
debug("active %d ignore %d", active, ignoring); |
for (;;) { |
for (;;) { |
switch (lineval = checkline(&cursym)) { |
switch (lineval = process(depth)) { |
case LT_PLAIN: |
case LT_ELIF: |
flushline(YES); |
debug("#elif start %d line %d code %d depth %d", |
|
stifline, linenum, lineval, depth); |
|
if (inelse) |
|
error(ELIF_ERR, depth); |
|
donetrue = false; |
|
reject = savereject; |
|
if (active) { |
|
active = false; |
|
elif2if(); |
|
flushline(true); |
|
} else { |
|
ignoring = false; |
|
flushline(true); |
|
} |
|
debug("active %d ignore %d", active, ignoring); |
break; |
break; |
|
case LT_ELTRUE: |
case LT_TRUE: |
case LT_ELFALSE: |
case LT_FALSE: |
debug("#elif start %d line %d code %d depth %d", |
thisreject = reject; |
stifline, linenum, lineval, depth); |
if (lineval == LT_TRUE) |
if (inelse) |
insym[cursym] = SYM_TRUE; |
error(ELIF_ERR, depth); |
|
if (active) |
|
flushline(false); |
else { |
else { |
if (reject != REJ_YES) |
ignoring = false; |
reject = ignore[cursym] ? REJ_IGNORE : REJ_YES; |
active = true; |
insym[cursym] = SYM_FALSE; |
elif2endif(); |
|
flushline(true); |
} |
} |
if (ignore[cursym]) |
if (lineval == LT_ELFALSE) |
flushline(YES); |
reject = REJ_YES; |
else { |
else { |
exitstat = 1; |
reject = REJ_NO; |
flushline(NO); |
donetrue = true; |
} |
} |
if ((doret = doif(cursym, IN_IF, thisreject, depth + 1)) != NO_ERR) |
debug("active %d ignore %d", active, ignoring); |
return error(doret, stline, depth); |
|
break; |
break; |
|
|
case LT_IF: |
|
case LT_OTHER: |
|
flushline(YES); |
|
if ((doret = doif(-1, IN_IF, reject, depth + 1)) != NO_ERR) |
|
return error(doret, stline, depth); |
|
break; |
|
|
|
case LT_ELSE: |
case LT_ELSE: |
if (inif != IN_IF) |
debug("#else start %d line %d code %d depth %d", |
return error(ELSE_ERR, linenum, depth); |
stifline, linenum, lineval, depth); |
inif = IN_ELSE; |
if (inelse) |
if (thissym >= 0) { |
error(ELSE_ERR, depth); |
if (insym[thissym] == SYM_TRUE) { |
if (active) { |
reject = ignore[thissym] ? REJ_IGNORE : REJ_YES; |
flushline(false); |
insym[thissym] = SYM_FALSE; |
reject = REJ_YES; |
} else { /* (insym[thissym] == |
if (reject == REJ_YES && !donetrue) |
* SYM_FALSE) */ |
reject = REJ_NO; |
reject = prevreject; |
} else { |
insym[thissym] = SYM_TRUE; |
flushline(true); |
|
if (ignoring) { |
|
if (reject == REJ_IGNORE) |
|
reject = REJ_NO; |
} |
} |
if (!ignore[thissym]) { |
|
flushline(NO); |
|
break; |
|
} |
|
} |
} |
flushline(YES); |
inelse = true; |
|
debug("active %d ignore %d", active, ignoring); |
break; |
break; |
|
|
case LT_ENDIF: |
case LT_ENDIF: |
if (inif == IN_NONE) |
debug("#endif start %d line %d code %d depth %d", |
return error(ENDIF_ERR, linenum, depth); |
stifline, linenum, lineval, depth); |
if (thissym >= 0) { |
if (active) |
insym[thissym] = SYM_INACTIVE; |
flushline(false); |
reject = prevreject; |
else |
if (!ignore[thissym]) { |
flushline(true); |
flushline(NO); |
reject = savereject; |
return NO_ERR; |
stifline = saveline; |
} |
return; |
} |
default: |
flushline(YES); |
/* bug */ |
return NO_ERR; |
abort(); |
|
|
case LT_LEOF:{ |
|
int err; |
|
err = incomment |
|
? CEOF_ERR |
|
: inquote == QUOTE_SINGLE |
|
? Q1EOF_ERR |
|
: inquote == QUOTE_DOUBLE |
|
? Q2EOF_ERR |
|
: NO_ERR; |
|
if (inif != IN_NONE) { |
|
if (err != NO_ERR) |
|
(void) error(err, stqcline, depth); |
|
return error(IEOF_ERR, stline, depth); |
|
} else |
|
if (err != NO_ERR) |
|
return error(err, stqcline, depth); |
|
else |
|
return NO_ERR; |
|
} |
|
} |
} |
} |
} |
} |
} |
#define endsym(c) (!isalpha (c) && !isdigit (c) && c != '_') |
|
|
|
#define MAXLINE 256 |
/* |
char tline[MAXLINE]; |
* The main file processing routine. This function deals with passing |
|
* through normal non-#if lines, correct nesting of #if sections, and |
|
* checking that things terminate correctly at the end of file. The |
|
* complicated stuff is delegated to doif(). |
|
*/ |
|
Linetype |
|
process(int depth) |
|
{ |
|
Linetype lineval; |
|
int cursym; |
|
|
|
for (;;) { |
|
linenum++; |
|
if (fgets(tline, MAXLINE, input) == NULL) { |
|
if (incomment) |
|
error(CEOF_ERR, depth); |
|
if (inquote == QUOTE_SINGLE) |
|
error(Q1EOF_ERR, depth); |
|
if (inquote == QUOTE_DOUBLE) |
|
error(Q2EOF_ERR, depth); |
|
if (depth != 0) |
|
error(IEOF_ERR, depth); |
|
return (LT_EOF); |
|
} |
|
switch (lineval = checkline(&cursym)) { |
|
case LT_PLAIN: |
|
flushline(true); |
|
break; |
|
case LT_IF: |
|
case LT_TRUE: |
|
case LT_FALSE: |
|
doif(depth + 1, lineval, ignore[cursym]); |
|
break; |
|
case LT_ELIF: |
|
case LT_ELTRUE: |
|
case LT_ELFALSE: |
|
case LT_ELSE: |
|
case LT_ENDIF: |
|
if (depth != 0) |
|
return (lineval); |
|
if (lineval == LT_ENDIF) |
|
error(ENDIF_ERR, depth); |
|
if (lineval == LT_ELSE) |
|
error(ELSE_ERR, depth); |
|
error(ELIF_ERR, depth); |
|
default: |
|
/* bug */ |
|
abort(); |
|
} |
|
} |
|
} |
|
|
|
/* |
|
* Parse a line and determine its type. |
|
*/ |
Linetype |
Linetype |
checkline(cursym) |
checkline(int *cursym) |
int *cursym; /* if LT_TRUE or LT_FALSE returned, set this |
|
* to sym index */ |
|
{ |
{ |
char *cp; |
const char *cp; |
char *symp; |
char *symp; |
char *scp; |
|
Linetype retval; |
Linetype retval; |
#define KWSIZE 8 |
char kw[KWSIZE]; |
char keyword[KWSIZE]; |
|
|
|
linenum++; |
|
if (getlin(tline, sizeof tline, input, NO) == EOF) |
|
return LT_LEOF; |
|
|
|
retval = LT_PLAIN; |
retval = LT_PLAIN; |
if (*(cp = tline) != '#' |
cp = skipcomment(tline); |
|| incomment |
if (*cp != '#' || incomment || inquote == QUOTE_SINGLE || |
|| inquote == QUOTE_SINGLE |
inquote == QUOTE_DOUBLE) |
|| inquote == QUOTE_DOUBLE |
|
) |
|
goto eol; |
goto eol; |
|
|
cp = skipcomment(++cp); |
cp = skipcomment(++cp); |
symp = keyword; |
keyword = (char *)cp; |
|
symp = kw; |
while (!endsym(*cp)) { |
while (!endsym(*cp)) { |
*symp = *cp++; |
*symp = *cp++; |
if (++symp >= &keyword[KWSIZE]) |
if (++symp >= &kw[KWSIZE]) |
goto eol; |
goto eol; |
} |
} |
*symp = '\0'; |
*symp = '\0'; |
|
|
if (strcmp(keyword, "ifdef") == 0) { |
if (strcmp(kw, "ifdef") == 0) { |
retval = YES; |
retval = LT_TRUE; |
goto ifdef; |
goto ifdef; |
} else |
} else if (strcmp(kw, "ifndef") == 0) { |
if (strcmp(keyword, "ifndef") == 0) { |
retval = LT_FALSE; |
retval = NO; |
ifdef: |
ifdef: |
cp = skipcomment(++cp); |
scp = cp = skipcomment(++cp); |
if (incomment) { |
if (incomment) { |
retval = LT_PLAIN; |
retval = LT_PLAIN; |
goto eol; |
goto eol; |
} |
} { |
if ((*cursym = findsym(cp)) == 0) |
int symind; |
retval = LT_IF; |
|
else if (value[*cursym] == NULL) |
|
retval = (retval == LT_TRUE) |
|
? LT_FALSE : LT_TRUE; |
|
} else if (strcmp(kw, "if") == 0) { |
|
retval = ifeval(&cp); |
|
cp = skipcomment(cp); |
|
if (*cp != '\n' || keepthis) |
|
retval = LT_IF; |
|
*cursym = 0; |
|
} else if (strcmp(kw, "elif") == 0) { |
|
retval = ifeval(&cp); |
|
cp = skipcomment(cp); |
|
if (*cp != '\n' || keepthis) |
|
retval = LT_ELIF; |
|
if (retval == LT_IF) |
|
retval = LT_ELIF; |
|
if (retval == LT_TRUE) |
|
retval = LT_ELTRUE; |
|
if (retval == LT_FALSE) |
|
retval = LT_ELFALSE; |
|
*cursym = 0; |
|
} else if (strcmp(kw, "else") == 0) |
|
retval = LT_ELSE; |
|
else if (strcmp(kw, "endif") == 0) |
|
retval = LT_ENDIF; |
|
|
if ((symind = findsym(scp)) >= 0) |
|
retval = (retval ^ true[*cursym = symind]) |
|
? LT_FALSE : LT_TRUE; |
|
else |
|
retval = LT_OTHER; |
|
} |
|
} else |
|
if (strcmp(keyword, "if") == 0) |
|
retval = LT_IF; |
|
else |
|
if (strcmp(keyword, "else") == 0) |
|
retval = LT_ELSE; |
|
else |
|
if (strcmp(keyword, "endif") == 0) |
|
retval = LT_ENDIF; |
|
|
|
eol: |
eol: |
if (!text && reject != REJ_IGNORE) |
if (!text && reject != REJ_IGNORE) { |
for (; *cp;) { |
for (; *cp;) { |
if (incomment) |
if (incomment) |
cp = skipcomment(cp); |
cp = skipcomment(cp); |
|
else if (inquote == QUOTE_SINGLE) |
|
cp = skipquote(cp, QUOTE_SINGLE); |
|
else if (inquote == QUOTE_DOUBLE) |
|
cp = skipquote(cp, QUOTE_DOUBLE); |
|
else if (*cp == '/' && (cp[1] == '*' || cp[1] == '/')) |
|
cp = skipcomment(cp); |
|
else if (*cp == '\'') |
|
cp = skipquote(cp, QUOTE_SINGLE); |
|
else if (*cp == '"') |
|
cp = skipquote(cp, QUOTE_DOUBLE); |
else |
else |
if (inquote == QUOTE_SINGLE) |
cp++; |
cp = skipquote(cp, QUOTE_SINGLE); |
|
else |
|
if (inquote == QUOTE_DOUBLE) |
|
cp = skipquote(cp, QUOTE_DOUBLE); |
|
else |
|
if (*cp == '/' && (cp[1] == '*' || cp[1] == '/')) |
|
cp = skipcomment(cp); |
|
else |
|
if (*cp == '\'') |
|
cp = skipquote(cp, QUOTE_SINGLE); |
|
else |
|
if (*cp == '"') |
|
cp = skipquote(cp, QUOTE_DOUBLE); |
|
else |
|
cp++; |
|
} |
} |
return retval; |
} |
|
return (retval); |
} |
} |
|
|
/* |
/* |
* Skip over comments and stop at the next charaacter |
* Turn a #elif line into a #if. This function is used when we are |
* position that is not whitespace. |
* processing a #if/#elif/#else/#endif sequence that starts off with a |
|
* #if that we understand (and therefore it has been deleted) which is |
|
* followed by a #elif that we don't understand and therefore must be |
|
* kept. We turn it into a #if to keep the nesting correct. |
*/ |
*/ |
char * |
void |
skipcomment(cp) |
elif2if(void) |
char *cp; |
|
{ |
{ |
|
strncpy(keyword, "if ", 4); |
|
} |
|
|
|
/* |
|
* Turn a #elif line into a #endif. This is used in the opposite |
|
* situation to elif2if, i.e. a #if that we don't understand is |
|
* followed by a #elif that we do; rather than deleting the #elif (as |
|
* we would for a #if) we turn it into a #endif to keep the nesting |
|
* correct. |
|
*/ |
|
void |
|
elif2endif(void) |
|
{ |
|
strcpy(keyword, "endif\n"); |
|
} |
|
|
|
/* |
|
* Function for evaluating the innermost parts of expressions, |
|
* viz. !expr (expr) defined(symbol) symbol number |
|
* We reset the keepthis flag when we find a non-constant subexpression. |
|
*/ |
|
Linetype |
|
eval_unary(struct ops *ops, int *valp, const char **cpp) |
|
{ |
|
const char *cp; |
|
char *ep; |
|
int sym; |
|
|
|
cp = skipcomment(*cpp); |
|
if(*cp == '!') { |
|
debug("eval%d !", ops - eval_ops); |
|
cp++; |
|
if (eval_unary(ops, valp, &cp) == LT_IF) |
|
return (LT_IF); |
|
*valp = !*valp; |
|
} else if (*cp == '(') { |
|
cp++; |
|
debug("eval%d (", ops - eval_ops); |
|
if (eval_table(eval_ops, valp, &cp) == LT_IF) |
|
return (LT_IF); |
|
cp = skipcomment(cp); |
|
if (*cp++ != ')') |
|
return (LT_IF); |
|
} else if (isdigit((unsigned char)*cp)) { |
|
debug("eval%d number", ops - eval_ops); |
|
*valp = strtol(cp, &ep, 0); |
|
cp = skipsym(cp); |
|
} else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) { |
|
cp = skipcomment(cp+7); |
|
debug("eval%d defined", ops - eval_ops); |
|
if (*cp++ != '(') |
|
return (LT_IF); |
|
cp = skipcomment(cp); |
|
sym = findsym(cp); |
|
if (sym == 0 && !symlist) |
|
return (LT_IF); |
|
*valp = (value[sym] != NULL); |
|
cp = skipsym(cp); |
|
cp = skipcomment(cp); |
|
if (*cp++ != ')') |
|
return (LT_IF); |
|
keepthis = false; |
|
} else if (!endsym(*cp)) { |
|
debug("eval%d symbol", ops - eval_ops); |
|
sym = findsym(cp); |
|
if (sym == 0 && !symlist) |
|
return (LT_IF); |
|
if (value[sym] == NULL) |
|
*valp = 0; |
|
else { |
|
*valp = strtol(value[sym], &ep, 0); |
|
if (*ep != '\0' || ep == value[sym]) |
|
return (LT_IF); |
|
} |
|
cp = skipsym(cp); |
|
keepthis = false; |
|
} else |
|
return (LT_IF); |
|
|
|
*cpp = cp; |
|
debug("eval%d = %d", ops - eval_ops, *valp); |
|
return (*valp ? LT_TRUE : LT_FALSE); |
|
} |
|
|
|
/* |
|
* Table-driven evaluation of binary operators. |
|
*/ |
|
Linetype |
|
eval_table(struct ops *ops, int *valp, const char **cpp) |
|
{ |
|
const char *cp; |
|
struct op *op; |
|
int val; |
|
|
|
debug("eval%d", ops - eval_ops); |
|
cp = *cpp; |
|
if (ops->inner(ops+1, valp, &cp) == LT_IF) |
|
return (LT_IF); |
|
for (;;) { |
|
cp = skipcomment(cp); |
|
for (op = ops->op; op->str != NULL; op++) |
|
if (strncmp(cp, op->str, strlen(op->str)) == 0) |
|
break; |
|
if (op->str == NULL) |
|
break; |
|
cp += strlen(op->str); |
|
debug("eval%d %s", ops - eval_ops, op->str); |
|
if (ops->inner(ops+1, &val, &cp) == LT_IF) |
|
return LT_IF; |
|
*valp = op->fn(*valp, val); |
|
} |
|
|
|
*cpp = cp; |
|
debug("eval%d = %d", ops - eval_ops, *valp); |
|
return (*valp ? LT_TRUE : LT_FALSE); |
|
} |
|
|
|
/* |
|
* Evaluate the expression on a #if or #elif line. If we can work out |
|
* the result we return LT_TRUE or LT_FALSE accordingly, otherwise we |
|
* return just a generic LT_IF. If the expression is constant and |
|
* we are not processing constant #ifs then the keepthis flag is true. |
|
*/ |
|
Linetype |
|
ifeval(const char **cpp) |
|
{ |
|
int val; |
|
|
|
debug("eval %s", *cpp); |
|
keepthis = killconsts ? false : true; |
|
return (eval_table(eval_ops, &val, cpp)); |
|
} |
|
|
|
/* |
|
* Skip over comments and stop at the next character position that is |
|
* not whitespace. |
|
*/ |
|
const char * |
|
skipcomment(const char *cp) |
|
{ |
if (incomment) |
if (incomment) |
goto inside; |
goto inside; |
for (;; cp++) { |
for (;; cp++) { |
while (*cp == ' ' || *cp == '\t') |
while (*cp == ' ' || *cp == '\t') |
cp++; |
cp++; |
if (text) |
if (text) |
return cp; |
return (cp); |
if (cp[0] != '/') |
if (cp[0] != '/') |
return cp; |
return (cp); |
|
|
if (cp[1] == '*') { |
if (cp[1] == '*') { |
if (!incomment) { |
if (!incomment) { |
|
|
stqcline = linenum; |
stqcline = linenum; |
} |
} |
} else |
} else |
return cp; |
return (cp); |
|
|
cp += 2; |
cp += 2; |
inside: |
inside: |
|
|
for (;;) { |
for (;;) { |
for (; *cp != '*'; cp++) |
for (; *cp != '*'; cp++) |
if (*cp == '\0') |
if (*cp == '\0') |
return cp; |
return (cp); |
if (*++cp == '/') { |
if (*++cp == '/') { |
incomment = NO; |
incomment = NO_COMMENT; |
break; |
break; |
} |
} |
} |
} |
} |
} else if (incomment == CXX_COMMENT) { |
else if (incomment == CXX_COMMENT) { |
|
for (; *cp != '\n'; cp++) |
for (; *cp != '\n'; cp++) |
if (*cp == '\0') |
if (*cp == '\0') |
return cp; |
return (cp); |
incomment = NO; |
incomment = NO_COMMENT; |
} |
} |
} |
} |
} |
} |
|
|
/* |
/* |
* Skip over a quoted string or character and stop at the next charaacter |
* Skip over a quoted string or character and stop at the next charaacter |
* position that is not whitespace. |
* position that is not whitespace. |
*/ |
*/ |
char * |
const char * |
skipquote(cp, type) |
skipquote(const char *cp, Quote_state type) |
char *cp; |
|
int type; |
|
{ |
{ |
char qchar; |
char qchar; |
|
|
|
|
goto inside; |
goto inside; |
for (;; cp++) { |
for (;; cp++) { |
if (*cp != qchar) |
if (*cp != qchar) |
return cp; |
return (cp); |
cp++; |
cp++; |
inquote = type; |
inquote = type; |
stqcline = linenum; |
stqcline = linenum; |
|
|
if (*cp == qchar) |
if (*cp == qchar) |
break; |
break; |
if (*cp == '\0' || (*cp == '\\' && *++cp == '\0')) |
if (*cp == '\0' || (*cp == '\\' && *++cp == '\0')) |
return cp; |
return (cp); |
} |
} |
inquote = QUOTE_NONE; |
inquote = QUOTE_NONE; |
} |
} |
} |
} |
|
|
/* |
/* |
* findsym - look for the symbol in the symbol table. |
* Skip over an identifier. |
* if found, return symbol table index, |
|
* else return -1. |
|
*/ |
*/ |
|
const char * |
|
skipsym(const char *cp) |
|
{ |
|
while (!endsym(*cp)) |
|
++cp; |
|
return (cp); |
|
} |
|
|
|
/* |
|
* Look for the symbol in the symbol table. If is is found, we return |
|
* the symbol table index, else we return 0. |
|
*/ |
int |
int |
findsym(str) |
findsym(const char *str) |
char *str; |
|
{ |
{ |
char *cp; |
const char *cp; |
char *symp; |
const char *symp; |
int symind; |
int symind; |
char chr; |
|
|
|
for (symind = 0; symind < nsyms; ++symind) { |
if (symlist) { |
if (insym[symind] == SYM_INACTIVE) { |
for (cp = str; !endsym(*cp); cp++) |
for (symp = symname[symind], cp = str |
continue; |
; *symp && *cp == *symp |
printf("%.*s\n", (int)(cp-str), str); |
; cp++, symp++ |
} |
) |
for (symind = 1; symind < nsyms; ++symind) { |
continue; |
for (cp = str, symp = symname[symind]; |
chr = *cp; |
*cp && *symp && *cp == *symp; cp++, symp++) |
if (*symp == '\0' && endsym(chr)) |
continue; |
return symind; |
if (*symp == '\0' && endsym(*cp)) { |
|
debug("findsym %s %s", symname[symind], |
|
value[symind] ? value[symind] : ""); |
|
return (symind); |
} |
} |
} |
} |
return -1; |
return (0); |
} |
} |
|
|
/* |
/* |
* getlin - expands tabs if asked for |
* Add a symbol to the symbol table. |
* and (if compiled in) treats form-feed as an end-of-line |
|
*/ |
*/ |
|
void |
|
addsym(bool ignorethis, bool definethis, char *sym) |
|
{ |
|
int symind; |
|
char *val; |
|
|
|
symind = findsym(sym); |
|
if (symind == 0) { |
|
if (nsyms >= MAXSYMS) |
|
errx(2, "too many symbols"); |
|
symind = nsyms++; |
|
} |
|
symname[symind] = sym; |
|
ignore[symind] = ignorethis; |
|
val = (char *)skipsym(sym); |
|
if (definethis) { |
|
if (*val == '=') { |
|
value[symind] = val+1; |
|
*val = '\0'; |
|
} else if (*val == '\0') |
|
value[symind] = ""; |
|
else |
|
usage(); |
|
} else { |
|
if (*val != '\0') |
|
usage(); |
|
value[symind] = NULL; |
|
} |
|
} |
|
|
|
#if 0 |
|
/* |
|
* Read a line from the input and expand tabs if requested and (if |
|
* compiled in) treats form-feed as an end-of-line. |
|
*/ |
int |
int |
getlin(line, maxline, inp, expandtabs) |
getline(char *line, int maxline, FILE *inp, bool expandtabs) |
char *line; |
|
int maxline; |
|
FILE *inp; |
|
int expandtabs; |
|
{ |
{ |
int tmp; |
int tmp; |
int num; |
int num; |
int chr; |
int chr; |
#ifdef FFSPECIAL |
#ifdef FFSPECIAL |
static char havechar = NO; /* have leftover char from last time */ |
static bool havechar = false; /* have leftover char from last time */ |
static char svchar; |
static char svchar; |
#endif /* FFSPECIAL */ |
#endif /* FFSPECIAL */ |
|
|
num = 0; |
num = 0; |
#ifdef FFSPECIAL |
#ifdef FFSPECIAL |
if (havechar) { |
if (havechar) { |
havechar = NO; |
havechar = false; |
chr = svchar; |
chr = svchar; |
goto ent; |
goto ent; |
} |
} |
#endif /* FFSPECIAL */ |
#endif /* FFSPECIAL */ |
while (num + 8 < maxline) { /* leave room for tab */ |
while (num + 8 < maxline) { /* leave room for tab */ |
chr = getc(inp); |
chr = getc(inp); |
if (chr == EOF) |
if (chr == EOF) |
return EOF; |
return (EOF); |
if (isprint(chr)) { |
if (0 && isprint(chr)) { |
#ifdef FFSPECIAL |
#ifdef FFSPECIAL |
ent: |
ent: |
#endif /* FFSPECIAL */ |
#endif /* FFSPECIAL */ |
*line++ = chr; |
*line++ = chr; |
num++; |
num++; |
} else |
} else |
switch (chr) { |
switch (chr) { |
|
case EOF: |
|
return (EOF); |
|
|
case '\t': |
case '\t': |
if (expandtabs) { |
if (expandtabs) { |
num += tmp = 8 - (num & 7); |
num += tmp = 8 - (num & 7); |
|
|
while (--tmp); |
while (--tmp); |
break; |
break; |
} |
} |
default: |
|
*line++ = chr; |
|
num++; |
|
break; |
|
|
|
case '\n': |
case '\n': |
*line = '\n'; |
*line = '\n'; |
|
|
*line = '\f'; |
*line = '\f'; |
else { |
else { |
*line = '\n'; |
*line = '\n'; |
havechar = YES; |
havechar = true; |
svchar = chr; |
svchar = chr; |
} |
} |
goto end; |
goto end; |
#endif /* FFSPECIAL */ |
#endif /* FFSPECIAL */ |
|
default: |
|
*line++ = chr; |
|
num++; |
|
break; |
} |
} |
} |
} |
end: |
end: |
*++line = '\0'; |
*++line = '\0'; |
return num; |
return (num); |
} |
} |
|
#endif |
|
|
|
/* |
|
* Write a line to the output or not, according to the current |
|
* filtering state. |
|
*/ |
void |
void |
flushline(keep) |
flushline(bool keep) |
Bool keep; |
|
{ |
{ |
if ((keep && reject != REJ_YES) ^ complement) { |
if (symlist) |
char *line = tline; |
return; |
FILE *out = stdout; |
if ((keep && reject != REJ_YES) ^ complement) |
char chr; |
fputs(tline, stdout); |
|
else if (lnblank) |
while ((chr = *line++)) |
putc('\n', stdout); |
putc(chr, out); |
|
} else |
|
if (lnblank) |
|
putc('\n', stdout); |
|
return; |
|
} |
} |
|
|
void |
void |
prname() |
debug(const char *msg, ...) |
{ |
{ |
fprintf(stderr, "%s: ", progname); |
va_list ap; |
return; |
|
|
if (debugging) { |
|
va_start(ap, msg); |
|
vwarnx(msg, ap); |
|
va_end(ap); |
|
} |
} |
} |
|
|
int |
void |
error(err, line, depth) |
error(int code, int depth) |
int err; /* type of error & index into error string |
|
* array */ |
|
int line; /* line number */ |
|
int depth; /* how many ifdefs we are inside */ |
|
{ |
{ |
if (err == END_ERR) |
if (incomment || inquote) |
return err; |
errx(2, "error in %s line %d: %s (#if depth %d)", |
|
filename, stqcline, errs[code], depth); |
prname(); |
else |
|
errx(2, "error in %s line %d: %s" |
#ifndef TESTING |
" (#if depth %d start line %d)", |
fprintf(stderr, "Error in %s line %d: %s.\n", filename, line, errs[err]); |
filename, linenum, errs[code], depth, stifline); |
#else /* TESTING */ |
|
fprintf(stderr, "Error in %s line %d: %s. ", filename, line, errs[err]); |
|
fprintf(stderr, "ifdef depth: %d\n", depth); |
|
#endif /* TESTING */ |
|
|
|
exitstat = 2; |
|
return depth > 1 ? IEOF_ERR : END_ERR; |
|
} |
} |