da21b64ab6
not define. Obtained From: NetBSD (rev 1.18; sjg)
1232 lines
29 KiB
C
1232 lines
29 KiB
C
/*
|
||
* Copyright (c) 1988, 1989, 1990, 1993
|
||
* The Regents of the University of California. All rights reserved.
|
||
* Copyright (c) 1988, 1989 by Adam de Boor
|
||
* Copyright (c) 1989 by Berkeley Softworks
|
||
* All rights reserved.
|
||
*
|
||
* This code is derived from software contributed to Berkeley by
|
||
* Adam de Boor.
|
||
*
|
||
* Redistribution and use in source and binary forms, with or without
|
||
* modification, are permitted provided that the following conditions
|
||
* are met:
|
||
* 1. Redistributions of source code must retain the above copyright
|
||
* notice, this list of conditions and the following disclaimer.
|
||
* 2. Redistributions in binary form must reproduce the above copyright
|
||
* notice, this list of conditions and the following disclaimer in the
|
||
* documentation and/or other materials provided with the distribution.
|
||
* 3. All advertising materials mentioning features or use of this software
|
||
* must display the following acknowledgement:
|
||
* This product includes software developed by the University of
|
||
* California, Berkeley and its contributors.
|
||
* 4. Neither the name of the University nor the names of its contributors
|
||
* may be used to endorse or promote products derived from this software
|
||
* without specific prior written permission.
|
||
*
|
||
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
||
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
||
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
||
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
||
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
||
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
||
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
||
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
||
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
||
* SUCH DAMAGE.
|
||
*
|
||
* @(#)cond.c 8.2 (Berkeley) 1/2/94
|
||
*/
|
||
|
||
#include <sys/cdefs.h>
|
||
__FBSDID("$FreeBSD$");
|
||
|
||
/*-
|
||
* cond.c --
|
||
* Functions to handle conditionals in a makefile.
|
||
*
|
||
* Interface:
|
||
* Cond_Eval Evaluate the conditional in the passed line.
|
||
*
|
||
*/
|
||
|
||
#include <ctype.h>
|
||
#include <math.h>
|
||
#include "make.h"
|
||
#include "hash.h"
|
||
#include "dir.h"
|
||
#include "buf.h"
|
||
|
||
/*
|
||
* The parsing of conditional expressions is based on this grammar:
|
||
* E -> F || E
|
||
* E -> F
|
||
* F -> T && F
|
||
* F -> T
|
||
* T -> defined(variable)
|
||
* T -> make(target)
|
||
* T -> exists(file)
|
||
* T -> empty(varspec)
|
||
* T -> target(name)
|
||
* T -> symbol
|
||
* T -> $(varspec) op value
|
||
* T -> $(varspec) == "string"
|
||
* T -> $(varspec) != "string"
|
||
* T -> ( E )
|
||
* T -> ! T
|
||
* op -> == | != | > | < | >= | <=
|
||
*
|
||
* 'symbol' is some other symbol to which the default function (condDefProc)
|
||
* is applied.
|
||
*
|
||
* Tokens are scanned from the 'condExpr' string. The scanner (CondToken)
|
||
* will return And for '&' and '&&', Or for '|' and '||', Not for '!',
|
||
* LParen for '(', RParen for ')' and will evaluate the other terminal
|
||
* symbols, using either the default function or the function given in the
|
||
* terminal, and return the result as either True or False.
|
||
*
|
||
* All Non-Terminal functions (CondE, CondF and CondT) return Err on error.
|
||
*/
|
||
typedef enum {
|
||
And, Or, Not, True, False, LParen, RParen, EndOfFile, None, Err
|
||
} Token;
|
||
|
||
/*-
|
||
* Structures to handle elegantly the different forms of #if's. The
|
||
* last two fields are stored in condInvert and condDefProc, respectively.
|
||
*/
|
||
static void CondPushBack(Token);
|
||
static int CondGetArg(char **, char **, char *, Boolean);
|
||
static Boolean CondDoDefined(int, char *);
|
||
static int CondStrMatch(void *, void *);
|
||
static Boolean CondDoMake(int, char *);
|
||
static Boolean CondDoExists(int, char *);
|
||
static Boolean CondDoTarget(int, char *);
|
||
static char * CondCvtArg(char *, double *);
|
||
static Token CondToken(Boolean);
|
||
static Token CondT(Boolean);
|
||
static Token CondF(Boolean);
|
||
static Token CondE(Boolean);
|
||
|
||
static struct If {
|
||
char *form; /* Form of if */
|
||
int formlen; /* Length of form */
|
||
Boolean doNot; /* TRUE if default function should be negated */
|
||
Boolean (*defProc)(int, char *); /* Default function to apply */
|
||
} ifs[] = {
|
||
{ "ifdef", 5, FALSE, CondDoDefined },
|
||
{ "ifndef", 6, TRUE, CondDoDefined },
|
||
{ "ifmake", 6, FALSE, CondDoMake },
|
||
{ "ifnmake", 7, TRUE, CondDoMake },
|
||
{ "if", 2, FALSE, CondDoDefined },
|
||
{ NULL, 0, FALSE, NULL }
|
||
};
|
||
|
||
static Boolean condInvert; /* Invert the default function */
|
||
static Boolean (*condDefProc) /* Default function to apply */
|
||
(int, char *);
|
||
static char *condExpr; /* The expression to parse */
|
||
static Token condPushBack=None; /* Single push-back token used in
|
||
* parsing */
|
||
|
||
#define MAXIF 30 /* greatest depth of #if'ing */
|
||
|
||
static Boolean condStack[MAXIF]; /* Stack of conditionals's values */
|
||
static int condTop = MAXIF; /* Top-most conditional */
|
||
static int skipIfLevel=0; /* Depth of skipped conditionals */
|
||
static Boolean skipLine = FALSE; /* Whether the parse module is skipping
|
||
* lines */
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondPushBack --
|
||
* Push back the most recent token read. We only need one level of
|
||
* this, so the thing is just stored in 'condPushback'.
|
||
*
|
||
* Results:
|
||
* None.
|
||
*
|
||
* Side Effects:
|
||
* condPushback is overwritten.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static void
|
||
CondPushBack (Token t)
|
||
{
|
||
condPushBack = t;
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondGetArg --
|
||
* Find the argument of a built-in function. parens is set to TRUE
|
||
* if the arguments are bounded by parens.
|
||
*
|
||
* Results:
|
||
* The length of the argument and the address of the argument.
|
||
*
|
||
* Side Effects:
|
||
* The pointer is set to point to the closing parenthesis of the
|
||
* function call.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static int
|
||
CondGetArg (char **linePtr, char **argPtr, char *func, Boolean parens)
|
||
{
|
||
char *cp;
|
||
int argLen;
|
||
Buffer buf;
|
||
|
||
cp = *linePtr;
|
||
if (parens) {
|
||
while (*cp != '(' && *cp != '\0') {
|
||
cp++;
|
||
}
|
||
if (*cp == '(') {
|
||
cp++;
|
||
}
|
||
}
|
||
|
||
if (*cp == '\0') {
|
||
/*
|
||
* No arguments whatsoever. Because 'make' and 'defined' aren't really
|
||
* "reserved words", we don't print a message. I think this is better
|
||
* than hitting the user with a warning message every time s/he uses
|
||
* the word 'make' or 'defined' at the beginning of a symbol...
|
||
*/
|
||
*argPtr = cp;
|
||
return (0);
|
||
}
|
||
|
||
while (*cp == ' ' || *cp == '\t') {
|
||
cp++;
|
||
}
|
||
|
||
/*
|
||
* Create a buffer for the argument and start it out at 16 characters
|
||
* long. Why 16? Why not?
|
||
*/
|
||
buf = Buf_Init(16);
|
||
|
||
while ((strchr(" \t)&|", *cp) == (char *)NULL) && (*cp != '\0')) {
|
||
if (*cp == '$') {
|
||
/*
|
||
* Parse the variable spec and install it as part of the argument
|
||
* if it's valid. We tell Var_Parse to complain on an undefined
|
||
* variable, so we don't do it too. Nor do we return an error,
|
||
* though perhaps we should...
|
||
*/
|
||
char *cp2;
|
||
int len;
|
||
Boolean doFree;
|
||
|
||
cp2 = Var_Parse(cp, VAR_CMD, TRUE, &len, &doFree);
|
||
|
||
Buf_AddBytes(buf, strlen(cp2), (Byte *)cp2);
|
||
if (doFree) {
|
||
free(cp2);
|
||
}
|
||
cp += len;
|
||
} else {
|
||
Buf_AddByte(buf, (Byte)*cp);
|
||
cp++;
|
||
}
|
||
}
|
||
|
||
Buf_AddByte(buf, (Byte)'\0');
|
||
*argPtr = (char *)Buf_GetAll(buf, &argLen);
|
||
Buf_Destroy(buf, FALSE);
|
||
|
||
while (*cp == ' ' || *cp == '\t') {
|
||
cp++;
|
||
}
|
||
if (parens && *cp != ')') {
|
||
Parse_Error (PARSE_WARNING, "Missing closing parenthesis for %s()",
|
||
func);
|
||
return (0);
|
||
} else if (parens) {
|
||
/*
|
||
* Advance pointer past close parenthesis.
|
||
*/
|
||
cp++;
|
||
}
|
||
|
||
*linePtr = cp;
|
||
return (argLen);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondDoDefined --
|
||
* Handle the 'defined' function for conditionals.
|
||
*
|
||
* Results:
|
||
* TRUE if the given variable is defined.
|
||
*
|
||
* Side Effects:
|
||
* None.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Boolean
|
||
CondDoDefined (int argLen, char *arg)
|
||
{
|
||
char savec = arg[argLen];
|
||
char *p1;
|
||
Boolean result;
|
||
|
||
arg[argLen] = '\0';
|
||
if (Var_Value (arg, VAR_CMD, &p1) != (char *)NULL) {
|
||
result = TRUE;
|
||
} else {
|
||
result = FALSE;
|
||
}
|
||
free(p1);
|
||
arg[argLen] = savec;
|
||
return (result);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondStrMatch --
|
||
* Front-end for Str_Match so it returns 0 on match and non-zero
|
||
* on mismatch. Callback function for CondDoMake via Lst_Find
|
||
*
|
||
* Results:
|
||
* 0 if string matches pattern
|
||
*
|
||
* Side Effects:
|
||
* None
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static int
|
||
CondStrMatch(void *string, void *pattern)
|
||
{
|
||
return(!Str_Match((char *) string,(char *) pattern));
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondDoMake --
|
||
* Handle the 'make' function for conditionals.
|
||
*
|
||
* Results:
|
||
* TRUE if the given target is being made.
|
||
*
|
||
* Side Effects:
|
||
* None.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Boolean
|
||
CondDoMake (int argLen, char *arg)
|
||
{
|
||
char savec = arg[argLen];
|
||
Boolean result;
|
||
|
||
arg[argLen] = '\0';
|
||
if (Lst_Find (create, (void *)arg, CondStrMatch) == NULL) {
|
||
result = FALSE;
|
||
} else {
|
||
result = TRUE;
|
||
}
|
||
arg[argLen] = savec;
|
||
return (result);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondDoExists --
|
||
* See if the given file exists.
|
||
*
|
||
* Results:
|
||
* TRUE if the file exists and FALSE if it does not.
|
||
*
|
||
* Side Effects:
|
||
* None.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Boolean
|
||
CondDoExists (int argLen, char *arg)
|
||
{
|
||
char savec = arg[argLen];
|
||
Boolean result;
|
||
char *path;
|
||
|
||
arg[argLen] = '\0';
|
||
path = Dir_FindFile(arg, dirSearchPath);
|
||
if (path != (char *)NULL) {
|
||
result = TRUE;
|
||
free(path);
|
||
} else {
|
||
result = FALSE;
|
||
}
|
||
arg[argLen] = savec;
|
||
return (result);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondDoTarget --
|
||
* See if the given node exists and is an actual target.
|
||
*
|
||
* Results:
|
||
* TRUE if the node exists as a target and FALSE if it does not.
|
||
*
|
||
* Side Effects:
|
||
* None.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Boolean
|
||
CondDoTarget (int argLen, char *arg)
|
||
{
|
||
char savec = arg[argLen];
|
||
Boolean result;
|
||
GNode *gn;
|
||
|
||
arg[argLen] = '\0';
|
||
gn = Targ_FindNode(arg, TARG_NOCREATE);
|
||
if ((gn != NULL) && !OP_NOP(gn->type)) {
|
||
result = TRUE;
|
||
} else {
|
||
result = FALSE;
|
||
}
|
||
arg[argLen] = savec;
|
||
return (result);
|
||
}
|
||
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondCvtArg --
|
||
* Convert the given number into a double. If the number begins
|
||
* with 0x, it is interpreted as a hexadecimal integer
|
||
* and converted to a double from there. All other strings just have
|
||
* strtod called on them.
|
||
*
|
||
* Results:
|
||
* Sets 'value' to double value of string.
|
||
* Returns address of the first character after the last valid
|
||
* character of the converted number.
|
||
*
|
||
* Side Effects:
|
||
* Can change 'value' even if string is not a valid number.
|
||
*
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static char *
|
||
CondCvtArg(char *str, double *value)
|
||
{
|
||
if ((*str == '0') && (str[1] == 'x')) {
|
||
long i;
|
||
|
||
for (str += 2, i = 0; ; str++) {
|
||
int x;
|
||
if (isdigit((unsigned char) *str))
|
||
x = *str - '0';
|
||
else if (isxdigit((unsigned char) *str))
|
||
x = 10 + *str - isupper((unsigned char) *str) ? 'A' : 'a';
|
||
else {
|
||
*value = (double) i;
|
||
return str;
|
||
}
|
||
i = (i << 4) + x;
|
||
}
|
||
}
|
||
else {
|
||
char *eptr;
|
||
*value = strtod(str, &eptr);
|
||
return eptr;
|
||
}
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondToken --
|
||
* Return the next token from the input.
|
||
*
|
||
* Results:
|
||
* A Token for the next lexical token in the stream.
|
||
*
|
||
* Side Effects:
|
||
* condPushback will be set back to None if it is used.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Token
|
||
CondToken(Boolean doEval)
|
||
{
|
||
Token t;
|
||
|
||
if (condPushBack == None) {
|
||
while (*condExpr == ' ' || *condExpr == '\t') {
|
||
condExpr++;
|
||
}
|
||
switch (*condExpr) {
|
||
case '(':
|
||
t = LParen;
|
||
condExpr++;
|
||
break;
|
||
case ')':
|
||
t = RParen;
|
||
condExpr++;
|
||
break;
|
||
case '|':
|
||
if (condExpr[1] == '|') {
|
||
condExpr++;
|
||
}
|
||
condExpr++;
|
||
t = Or;
|
||
break;
|
||
case '&':
|
||
if (condExpr[1] == '&') {
|
||
condExpr++;
|
||
}
|
||
condExpr++;
|
||
t = And;
|
||
break;
|
||
case '!':
|
||
t = Not;
|
||
condExpr++;
|
||
break;
|
||
case '\n':
|
||
case '\0':
|
||
t = EndOfFile;
|
||
break;
|
||
case '$': {
|
||
char *lhs;
|
||
char *rhs;
|
||
char *op;
|
||
int varSpecLen;
|
||
Boolean doFree;
|
||
|
||
/*
|
||
* Parse the variable spec and skip over it, saving its
|
||
* value in lhs.
|
||
*/
|
||
t = Err;
|
||
lhs = Var_Parse(condExpr, VAR_CMD, doEval,&varSpecLen,&doFree);
|
||
if (lhs == var_Error) {
|
||
/*
|
||
* Even if !doEval, we still report syntax errors, which
|
||
* is what getting var_Error back with !doEval means.
|
||
*/
|
||
return(Err);
|
||
}
|
||
condExpr += varSpecLen;
|
||
|
||
if (!isspace((unsigned char) *condExpr) &&
|
||
strchr("!=><", *condExpr) == NULL) {
|
||
Buffer buf;
|
||
char *cp;
|
||
|
||
buf = Buf_Init(0);
|
||
|
||
for (cp = lhs; *cp; cp++)
|
||
Buf_AddByte(buf, (Byte)*cp);
|
||
|
||
if (doFree)
|
||
free(lhs);
|
||
|
||
for (;*condExpr && !isspace((unsigned char) *condExpr);
|
||
condExpr++)
|
||
Buf_AddByte(buf, (Byte)*condExpr);
|
||
|
||
Buf_AddByte(buf, (Byte)'\0');
|
||
lhs = (char *)Buf_GetAll(buf, &varSpecLen);
|
||
Buf_Destroy(buf, FALSE);
|
||
|
||
doFree = TRUE;
|
||
}
|
||
|
||
/*
|
||
* Skip whitespace to get to the operator
|
||
*/
|
||
while (isspace((unsigned char) *condExpr))
|
||
condExpr++;
|
||
|
||
/*
|
||
* Make sure the operator is a valid one. If it isn't a
|
||
* known relational operator, pretend we got a
|
||
* != 0 comparison.
|
||
*/
|
||
op = condExpr;
|
||
switch (*condExpr) {
|
||
case '!':
|
||
case '=':
|
||
case '<':
|
||
case '>':
|
||
if (condExpr[1] == '=') {
|
||
condExpr += 2;
|
||
} else {
|
||
condExpr += 1;
|
||
}
|
||
break;
|
||
default:
|
||
op = "!=";
|
||
rhs = "0";
|
||
|
||
goto do_compare;
|
||
}
|
||
while (isspace((unsigned char) *condExpr)) {
|
||
condExpr++;
|
||
}
|
||
if (*condExpr == '\0') {
|
||
Parse_Error(PARSE_WARNING,
|
||
"Missing right-hand-side of operator");
|
||
goto error;
|
||
}
|
||
rhs = condExpr;
|
||
do_compare:
|
||
if (*rhs == '"') {
|
||
/*
|
||
* Doing a string comparison. Only allow == and != for
|
||
* operators.
|
||
*/
|
||
char *string;
|
||
char *cp, *cp2;
|
||
int qt;
|
||
Buffer buf;
|
||
|
||
do_string_compare:
|
||
if (((*op != '!') && (*op != '=')) || (op[1] != '=')) {
|
||
Parse_Error(PARSE_WARNING,
|
||
"String comparison operator should be either == or !=");
|
||
goto error;
|
||
}
|
||
|
||
buf = Buf_Init(0);
|
||
qt = *rhs == '"' ? 1 : 0;
|
||
|
||
for (cp = &rhs[qt];
|
||
((qt && (*cp != '"')) ||
|
||
(!qt && strchr(" \t)", *cp) == NULL)) &&
|
||
(*cp != '\0'); cp++) {
|
||
if ((*cp == '\\') && (cp[1] != '\0')) {
|
||
/*
|
||
* Backslash escapes things -- skip over next
|
||
* character, if it exists.
|
||
*/
|
||
cp++;
|
||
Buf_AddByte(buf, (Byte)*cp);
|
||
} else if (*cp == '$') {
|
||
int len;
|
||
Boolean freeIt;
|
||
|
||
cp2 = Var_Parse(cp, VAR_CMD, doEval,&len, &freeIt);
|
||
if (cp2 != var_Error) {
|
||
Buf_AddBytes(buf, strlen(cp2), (Byte *)cp2);
|
||
if (freeIt) {
|
||
free(cp2);
|
||
}
|
||
cp += len - 1;
|
||
} else {
|
||
Buf_AddByte(buf, (Byte)*cp);
|
||
}
|
||
} else {
|
||
Buf_AddByte(buf, (Byte)*cp);
|
||
}
|
||
}
|
||
|
||
Buf_AddByte(buf, (Byte)0);
|
||
|
||
string = (char *)Buf_GetAll(buf, (int *)0);
|
||
Buf_Destroy(buf, FALSE);
|
||
|
||
DEBUGF(COND, ("lhs = \"%s\", rhs = \"%s\", op = %.2s\n",
|
||
lhs, string, op));
|
||
/*
|
||
* Null-terminate rhs and perform the comparison.
|
||
* t is set to the result.
|
||
*/
|
||
if (*op == '=') {
|
||
t = strcmp(lhs, string) ? False : True;
|
||
} else {
|
||
t = strcmp(lhs, string) ? True : False;
|
||
}
|
||
free(string);
|
||
if (rhs == condExpr) {
|
||
if (!qt && *cp == ')')
|
||
condExpr = cp;
|
||
else
|
||
condExpr = cp + 1;
|
||
}
|
||
} else {
|
||
/*
|
||
* rhs is either a float or an integer. Convert both the
|
||
* lhs and the rhs to a double and compare the two.
|
||
*/
|
||
double left, right;
|
||
char *string;
|
||
|
||
if (*CondCvtArg(lhs, &left) != '\0')
|
||
goto do_string_compare;
|
||
if (*rhs == '$') {
|
||
int len;
|
||
Boolean freeIt;
|
||
|
||
string = Var_Parse(rhs, VAR_CMD, doEval,&len,&freeIt);
|
||
if (string == var_Error) {
|
||
right = 0.0;
|
||
} else {
|
||
if (*CondCvtArg(string, &right) != '\0') {
|
||
if (freeIt)
|
||
free(string);
|
||
goto do_string_compare;
|
||
}
|
||
if (freeIt)
|
||
free(string);
|
||
if (rhs == condExpr)
|
||
condExpr += len;
|
||
}
|
||
} else {
|
||
char *c = CondCvtArg(rhs, &right);
|
||
if (c == rhs)
|
||
goto do_string_compare;
|
||
if (rhs == condExpr) {
|
||
/*
|
||
* Skip over the right-hand side
|
||
*/
|
||
condExpr = c;
|
||
}
|
||
}
|
||
|
||
DEBUGF(COND, ("left = %f, right = %f, op = %.2s\n", left,
|
||
right, op));
|
||
switch(op[0]) {
|
||
case '!':
|
||
if (op[1] != '=') {
|
||
Parse_Error(PARSE_WARNING,
|
||
"Unknown operator");
|
||
goto error;
|
||
}
|
||
t = (left != right ? True : False);
|
||
break;
|
||
case '=':
|
||
if (op[1] != '=') {
|
||
Parse_Error(PARSE_WARNING,
|
||
"Unknown operator");
|
||
goto error;
|
||
}
|
||
t = (left == right ? True : False);
|
||
break;
|
||
case '<':
|
||
if (op[1] == '=') {
|
||
t = (left <= right ? True : False);
|
||
} else {
|
||
t = (left < right ? True : False);
|
||
}
|
||
break;
|
||
case '>':
|
||
if (op[1] == '=') {
|
||
t = (left >= right ? True : False);
|
||
} else {
|
||
t = (left > right ? True : False);
|
||
}
|
||
break;
|
||
default:
|
||
break;
|
||
}
|
||
}
|
||
error:
|
||
if (doFree)
|
||
free(lhs);
|
||
break;
|
||
}
|
||
default: {
|
||
Boolean (*evalProc)(int, char *);
|
||
Boolean invert = FALSE;
|
||
char *arg;
|
||
int arglen;
|
||
|
||
if (strncmp (condExpr, "defined", 7) == 0) {
|
||
/*
|
||
* Use CondDoDefined to evaluate the argument and
|
||
* CondGetArg to extract the argument from the 'function
|
||
* call'.
|
||
*/
|
||
evalProc = CondDoDefined;
|
||
condExpr += 7;
|
||
arglen = CondGetArg (&condExpr, &arg, "defined", TRUE);
|
||
if (arglen == 0) {
|
||
condExpr -= 7;
|
||
goto use_default;
|
||
}
|
||
} else if (strncmp (condExpr, "make", 4) == 0) {
|
||
/*
|
||
* Use CondDoMake to evaluate the argument and
|
||
* CondGetArg to extract the argument from the 'function
|
||
* call'.
|
||
*/
|
||
evalProc = CondDoMake;
|
||
condExpr += 4;
|
||
arglen = CondGetArg (&condExpr, &arg, "make", TRUE);
|
||
if (arglen == 0) {
|
||
condExpr -= 4;
|
||
goto use_default;
|
||
}
|
||
} else if (strncmp (condExpr, "exists", 6) == 0) {
|
||
/*
|
||
* Use CondDoExists to evaluate the argument and
|
||
* CondGetArg to extract the argument from the
|
||
* 'function call'.
|
||
*/
|
||
evalProc = CondDoExists;
|
||
condExpr += 6;
|
||
arglen = CondGetArg(&condExpr, &arg, "exists", TRUE);
|
||
if (arglen == 0) {
|
||
condExpr -= 6;
|
||
goto use_default;
|
||
}
|
||
} else if (strncmp(condExpr, "empty", 5) == 0) {
|
||
/*
|
||
* Use Var_Parse to parse the spec in parens and return
|
||
* True if the resulting string is empty.
|
||
*/
|
||
int length;
|
||
Boolean doFree;
|
||
char *val;
|
||
|
||
condExpr += 5;
|
||
|
||
for (arglen = 0;
|
||
condExpr[arglen] != '(' && condExpr[arglen] != '\0';
|
||
arglen += 1)
|
||
continue;
|
||
|
||
if (condExpr[arglen] != '\0') {
|
||
val = Var_Parse(&condExpr[arglen - 1], VAR_CMD,
|
||
FALSE, &length, &doFree);
|
||
if (val == var_Error) {
|
||
t = Err;
|
||
} else {
|
||
/*
|
||
* A variable is empty when it just contains
|
||
* spaces... 4/15/92, christos
|
||
*/
|
||
char *p;
|
||
for (p = val; *p && isspace((unsigned char)*p); p++)
|
||
continue;
|
||
t = (*p == '\0') ? True : False;
|
||
}
|
||
if (doFree) {
|
||
free(val);
|
||
}
|
||
/*
|
||
* Advance condExpr to beyond the closing ). Note that
|
||
* we subtract one from arglen + length b/c length
|
||
* is calculated from condExpr[arglen - 1].
|
||
*/
|
||
condExpr += arglen + length - 1;
|
||
} else {
|
||
condExpr -= 5;
|
||
goto use_default;
|
||
}
|
||
break;
|
||
} else if (strncmp (condExpr, "target", 6) == 0) {
|
||
/*
|
||
* Use CondDoTarget to evaluate the argument and
|
||
* CondGetArg to extract the argument from the
|
||
* 'function call'.
|
||
*/
|
||
evalProc = CondDoTarget;
|
||
condExpr += 6;
|
||
arglen = CondGetArg(&condExpr, &arg, "target", TRUE);
|
||
if (arglen == 0) {
|
||
condExpr -= 6;
|
||
goto use_default;
|
||
}
|
||
} else {
|
||
/*
|
||
* The symbol is itself the argument to the default
|
||
* function. We advance condExpr to the end of the symbol
|
||
* by hand (the next whitespace, closing paren or
|
||
* binary operator) and set to invert the evaluation
|
||
* function if condInvert is TRUE.
|
||
*/
|
||
use_default:
|
||
invert = condInvert;
|
||
evalProc = condDefProc;
|
||
arglen = CondGetArg(&condExpr, &arg, "", FALSE);
|
||
}
|
||
|
||
/*
|
||
* Evaluate the argument using the set function. If invert
|
||
* is TRUE, we invert the sense of the function.
|
||
*/
|
||
t = (!doEval || (* evalProc) (arglen, arg) ?
|
||
(invert ? False : True) :
|
||
(invert ? True : False));
|
||
free(arg);
|
||
break;
|
||
}
|
||
}
|
||
} else {
|
||
t = condPushBack;
|
||
condPushBack = None;
|
||
}
|
||
return (t);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondT --
|
||
* Parse a single term in the expression. This consists of a terminal
|
||
* symbol or Not and a terminal symbol (not including the binary
|
||
* operators):
|
||
* T -> defined(variable) | make(target) | exists(file) | symbol
|
||
* T -> ! T | ( E )
|
||
*
|
||
* Results:
|
||
* True, False or Err.
|
||
*
|
||
* Side Effects:
|
||
* Tokens are consumed.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Token
|
||
CondT(Boolean doEval)
|
||
{
|
||
Token t;
|
||
|
||
t = CondToken(doEval);
|
||
|
||
if (t == EndOfFile) {
|
||
/*
|
||
* If we reached the end of the expression, the expression
|
||
* is malformed...
|
||
*/
|
||
t = Err;
|
||
} else if (t == LParen) {
|
||
/*
|
||
* T -> ( E )
|
||
*/
|
||
t = CondE(doEval);
|
||
if (t != Err) {
|
||
if (CondToken(doEval) != RParen) {
|
||
t = Err;
|
||
}
|
||
}
|
||
} else if (t == Not) {
|
||
t = CondT(doEval);
|
||
if (t == True) {
|
||
t = False;
|
||
} else if (t == False) {
|
||
t = True;
|
||
}
|
||
}
|
||
return (t);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondF --
|
||
* Parse a conjunctive factor (nice name, wot?)
|
||
* F -> T && F | T
|
||
*
|
||
* Results:
|
||
* True, False or Err
|
||
*
|
||
* Side Effects:
|
||
* Tokens are consumed.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Token
|
||
CondF(Boolean doEval)
|
||
{
|
||
Token l, o;
|
||
|
||
l = CondT(doEval);
|
||
if (l != Err) {
|
||
o = CondToken(doEval);
|
||
|
||
if (o == And) {
|
||
/*
|
||
* F -> T && F
|
||
*
|
||
* If T is False, the whole thing will be False, but we have to
|
||
* parse the r.h.s. anyway (to throw it away).
|
||
* If T is True, the result is the r.h.s., be it an Err or no.
|
||
*/
|
||
if (l == True) {
|
||
l = CondF(doEval);
|
||
} else {
|
||
(void) CondF(FALSE);
|
||
}
|
||
} else {
|
||
/*
|
||
* F -> T
|
||
*/
|
||
CondPushBack (o);
|
||
}
|
||
}
|
||
return (l);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* CondE --
|
||
* Main expression production.
|
||
* E -> F || E | F
|
||
*
|
||
* Results:
|
||
* True, False or Err.
|
||
*
|
||
* Side Effects:
|
||
* Tokens are, of course, consumed.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
static Token
|
||
CondE(Boolean doEval)
|
||
{
|
||
Token l, o;
|
||
|
||
l = CondF(doEval);
|
||
if (l != Err) {
|
||
o = CondToken(doEval);
|
||
|
||
if (o == Or) {
|
||
/*
|
||
* E -> F || E
|
||
*
|
||
* A similar thing occurs for ||, except that here we make sure
|
||
* the l.h.s. is False before we bother to evaluate the r.h.s.
|
||
* Once again, if l is False, the result is the r.h.s. and once
|
||
* again if l is True, we parse the r.h.s. to throw it away.
|
||
*/
|
||
if (l == False) {
|
||
l = CondE(doEval);
|
||
} else {
|
||
(void) CondE(FALSE);
|
||
}
|
||
} else {
|
||
/*
|
||
* E -> F
|
||
*/
|
||
CondPushBack (o);
|
||
}
|
||
}
|
||
return (l);
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* Cond_Eval --
|
||
* Evaluate the conditional in the passed line. The line
|
||
* looks like this:
|
||
* #<cond-type> <expr>
|
||
* where <cond-type> is any of if, ifmake, ifnmake, ifdef,
|
||
* ifndef, elif, elifmake, elifnmake, elifdef, elifndef
|
||
* and <expr> consists of &&, ||, !, make(target), defined(variable)
|
||
* and parenthetical groupings thereof.
|
||
*
|
||
* Results:
|
||
* COND_PARSE if should parse lines after the conditional
|
||
* COND_SKIP if should skip lines after the conditional
|
||
* COND_INVALID if not a valid conditional.
|
||
*
|
||
* Side Effects:
|
||
* None.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
int
|
||
Cond_Eval (char *line)
|
||
{
|
||
struct If *ifp;
|
||
Boolean isElse;
|
||
Boolean value = FALSE;
|
||
int level; /* Level at which to report errors. */
|
||
|
||
level = PARSE_FATAL;
|
||
|
||
for (line++; *line == ' ' || *line == '\t'; line++) {
|
||
continue;
|
||
}
|
||
|
||
/*
|
||
* Find what type of if we're dealing with. The result is left
|
||
* in ifp and isElse is set TRUE if it's an elif line.
|
||
*/
|
||
if (line[0] == 'e' && line[1] == 'l') {
|
||
line += 2;
|
||
isElse = TRUE;
|
||
} else if (strncmp (line, "endif", 5) == 0) {
|
||
/*
|
||
* End of a conditional section. If skipIfLevel is non-zero, that
|
||
* conditional was skipped, so lines following it should also be
|
||
* skipped. Hence, we return COND_SKIP. Otherwise, the conditional
|
||
* was read so succeeding lines should be parsed (think about it...)
|
||
* so we return COND_PARSE, unless this endif isn't paired with
|
||
* a decent if.
|
||
*/
|
||
if (skipIfLevel != 0) {
|
||
skipIfLevel -= 1;
|
||
return (COND_SKIP);
|
||
} else {
|
||
if (condTop == MAXIF) {
|
||
Parse_Error (level, "if-less endif");
|
||
return (COND_INVALID);
|
||
} else {
|
||
skipLine = FALSE;
|
||
condTop += 1;
|
||
return (COND_PARSE);
|
||
}
|
||
}
|
||
} else {
|
||
isElse = FALSE;
|
||
}
|
||
|
||
/*
|
||
* Figure out what sort of conditional it is -- what its default
|
||
* function is, etc. -- by looking in the table of valid "ifs"
|
||
*/
|
||
for (ifp = ifs; ifp->form != (char *)0; ifp++) {
|
||
if (strncmp (ifp->form, line, ifp->formlen) == 0) {
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (ifp->form == (char *) 0) {
|
||
/*
|
||
* Nothing fit. If the first word on the line is actually
|
||
* "else", it's a valid conditional whose value is the inverse
|
||
* of the previous if we parsed.
|
||
*/
|
||
if (isElse && (line[0] == 's') && (line[1] == 'e')) {
|
||
if (condTop == MAXIF) {
|
||
Parse_Error (level, "if-less else");
|
||
return (COND_INVALID);
|
||
} else if (skipIfLevel == 0) {
|
||
value = !condStack[condTop];
|
||
} else {
|
||
return (COND_SKIP);
|
||
}
|
||
} else {
|
||
/*
|
||
* Not a valid conditional type. No error...
|
||
*/
|
||
return (COND_INVALID);
|
||
}
|
||
} else {
|
||
if (isElse) {
|
||
if (condTop == MAXIF) {
|
||
Parse_Error (level, "if-less elif");
|
||
return (COND_INVALID);
|
||
} else if (skipIfLevel != 0) {
|
||
/*
|
||
* If skipping this conditional, just ignore the whole thing.
|
||
* If we don't, the user might be employing a variable that's
|
||
* undefined, for which there's an enclosing ifdef that
|
||
* we're skipping...
|
||
*/
|
||
return(COND_SKIP);
|
||
}
|
||
} else if (skipLine) {
|
||
/*
|
||
* Don't even try to evaluate a conditional that's not an else if
|
||
* we're skipping things...
|
||
*/
|
||
skipIfLevel += 1;
|
||
return(COND_SKIP);
|
||
}
|
||
|
||
/*
|
||
* Initialize file-global variables for parsing
|
||
*/
|
||
condDefProc = ifp->defProc;
|
||
condInvert = ifp->doNot;
|
||
|
||
line += ifp->formlen;
|
||
|
||
while (*line == ' ' || *line == '\t') {
|
||
line++;
|
||
}
|
||
|
||
condExpr = line;
|
||
condPushBack = None;
|
||
|
||
switch (CondE(TRUE)) {
|
||
case True:
|
||
if (CondToken(TRUE) == EndOfFile) {
|
||
value = TRUE;
|
||
break;
|
||
}
|
||
goto err;
|
||
/*FALLTHRU*/
|
||
case False:
|
||
if (CondToken(TRUE) == EndOfFile) {
|
||
value = FALSE;
|
||
break;
|
||
}
|
||
/*FALLTHRU*/
|
||
case Err:
|
||
err:
|
||
Parse_Error (level, "Malformed conditional (%s)",
|
||
line);
|
||
return (COND_INVALID);
|
||
default:
|
||
break;
|
||
}
|
||
}
|
||
if (!isElse) {
|
||
condTop -= 1;
|
||
} else if ((skipIfLevel != 0) || condStack[condTop]) {
|
||
/*
|
||
* If this is an else-type conditional, it should only take effect
|
||
* if its corresponding if was evaluated and FALSE. If its if was
|
||
* TRUE or skipped, we return COND_SKIP (and start skipping in case
|
||
* we weren't already), leaving the stack unmolested so later elif's
|
||
* don't screw up...
|
||
*/
|
||
skipLine = TRUE;
|
||
return (COND_SKIP);
|
||
}
|
||
|
||
if (condTop < 0) {
|
||
/*
|
||
* This is the one case where we can definitely proclaim a fatal
|
||
* error. If we don't, we're hosed.
|
||
*/
|
||
Parse_Error (PARSE_FATAL, "Too many nested if's. %d max.", MAXIF);
|
||
return (COND_INVALID);
|
||
} else {
|
||
condStack[condTop] = value;
|
||
skipLine = !value;
|
||
return (value ? COND_PARSE : COND_SKIP);
|
||
}
|
||
}
|
||
|
||
/*-
|
||
*-----------------------------------------------------------------------
|
||
* Cond_End --
|
||
* Make sure everything's clean at the end of a makefile.
|
||
*
|
||
* Results:
|
||
* None.
|
||
*
|
||
* Side Effects:
|
||
* Parse_Error will be called if open conditionals are around.
|
||
*
|
||
*-----------------------------------------------------------------------
|
||
*/
|
||
void
|
||
Cond_End(void)
|
||
{
|
||
if (condTop != MAXIF) {
|
||
Parse_Error(PARSE_FATAL, "%d open conditional%s", MAXIF-condTop,
|
||
MAXIF-condTop == 1 ? "" : "s");
|
||
}
|
||
condTop = MAXIF;
|
||
}
|