trouble casting char to double

Hi I have the following code:

Code:

i = 1;
while (fscanf(infile, "%[^:]:%[^:]:%s\n", line1,line2,line3) != EOF) {
        printf("Line was composed of %s, %s, %s\n",line1,line2,line3);
        for(j = 1 ; j <= NumInput ; j++){
                printf("char  :i=%d,j=%d and value=%c\n",i,j,line2[j-1]);
                printf("float :i=%d,j=%d and value=%f\n",i,j,line2[j-1]);
                printf("int   :i=%d,j=%d and value=%d\n",i,j,line2[j-1]-48);
                printf("float :i=%d,j=%d and value=%f\n",i,j,(double)line2[j-1]-48);
        /*      printf("atof  :i=%d,j=%d and value=%f\n",i,j,atof(line2[j-1]));   */
                printf("------\n");
                exit(1);
        }
        i++;
    }

(The infile has lines like: C0001:1001:1(string:0s and 1s:0 or 1))
The output is:

Code:

Line was composed of C003, 1100, 1
char   :i=1,j=1 and value=1
float  :i=1,j=1 and value=-0.523882
int    :i=1,j=1 and value=1
float  :i=1,j=1 and value=1.000000
------

I want to be able to cast the line2[j] to a float, since it appears to be a char. But if I uncomment the atof line i get the following error:

Code:

cc pnn.c -O -lm -o pnn
pnn.c: In function ‘main’:
pnn.c:74: warning: passing argument 1 of ‘atof’ makes pointer from integer without a cast

Is the (double)line2[2]-48 the best way to do what I want to do?

-deprekate

Code:

line2[j]

is not a pointer; the subscript dereferences the pointer. To get a pointer, you can use

Code:

&line2[j]

or

line2+j

"makes pointer from integer" is probably because line2[x] is a single char, and atof expects a char *pointer. You could try &line2[x], but you will likely have a problem with that because the "string" pointed to is not null terminated, methinks.

Originally Posted by MK27

but you will likely have a problem with that because the "string" pointed to is not null terminated, methinks.

since the last char of line2[] definitely has a NULL terminator, it doesn't matter where in the array you start reading the string.

QuantumPete

Originally Posted by drunken_scot

Code:

line2[j]

is not a pointer; the subscript dereferences the pointer. To get a pointer, you can use

Code:

&line2[j]

or

line2+j

I tried both ways:

Code:

                printf("char  :i=%d,j=%d and value=%c\n",i,j,line2[j-1]);
                printf("float :i=%d,j=%d and value=%f\n",i,j,line2[j-1]);
                printf("int   :i=%d,j=%d and value=%d\n",i,j,line2[j-1]-48);
                printf("float :i=%d,j=%d and value=%f\n",i,j,(double)line2[j-1]-48);
                printf("atof  :i=%d,j=%d and value=%f\n",i,j,atof(&line2[j-1]));
                printf("atof  :i=%d,j=%d and value=%f\n",i,j,atof(line2+(j-1)));
                printf("------\n");
...

Neither of those ways work.

Code:

Line was composed of 003, 0111, 1
char  :i=1,j=1 and value=0
float :i=1,j=1 and value=-0.062139
int   :i=1,j=1 and value=0
float :i=1,j=1 and value=0.000000
atof  :i=1,j=1 and value=111.000000
atof  :i=1,j=1 and value=111.000000
------

I guess my (double)line2[j-1]-48 workaround is the best/and only way to do it?
(the correct value should be the first character of the 2nd term of "Line composed of ____")

Excuse me, but I do not understand what you are actually trying to achieve.

If you have this input:

Code:

C0001:1001:1

and you want to translate that to a floating point value, yes?

What sort of data is the original data, if I may ask?

--
Mats

Originally Posted by matsp

Excuse me, but I do not understand what you are actually trying to achieve.

If you have this input:

Code:

C0001:1001:1

and you want to translate that to a floating point value, yes?

No I want to take the second term(1001) of that line(: are delimiters), and then pull out each character of that term("1","0","0","1") each into its own variable.

So I want to cast a single char(which happens to also be an underlying integer) to a float(using atof() presumably)", however I guess that atof() only accepts a pointer to a char and not a character/integer.

The printf's are just there to illustrate, I really end up setting an array cell to the value of that character/integer.

Originally Posted by matsp

What sort of data is the original data, if I may ask?
Mats

The original data is a simple text file with lines like: C0001:1001:1

Originally Posted by deprekate

No I want to take the second term(1001) of that line(: are delimiters), and then pull out each character of that term("1","0","0","1") each into its own variable.

So I want to cast a single char(which happens to also be an underlying integer) to a float(using atof() presumably)", however I guess that atof() only accepts a pointer to a char and not a character/integer.

The printf's are just there to illustrate, I really end up setting an array cell to the value of that character/integer.


The original data is a simple text file with lines like: C0001:1001:1

Why a float?

Anyway, if you have a character that you know is numeric, then subtracting '0' must needs give you the numeric value of that character. If you want it to be a float, you can assign it to a float, just like you can say float foo = 4.

Right, so lets put this a different way: If we have the string C0001:1001:1, and you want to extract the part that says 1001, what value do you want to get from it? 9, 1001 or 1001.0f?

I agree with tabstop - there is something not completely understood here. I'm not sure where the lack of understanding is, but it certainly doesn't seem sensible to use float casts where they are at the moment.

--
Mats

also what about

ffe0 - is this string possible?

Originally Posted by matsp

Right, so lets put this a different way: If we have the string C0001:1001:1, and you want to extract the part that says 1001, what value do you want to get from it? 9, 1001 or 1001.0f?

I agree with tabstop - there is something not completely understood here. I'm not sure where the lack of understanding is, but it certainly doesn't seem sensible to use float casts where they are at the moment.

--
Mats

So I take the string C001:1001:1 and I want to get teh 1001 part. and then I want to get each of the characters of 1001. So "1","0","0","1". and I will be putting these into a double array. arr[1]=1,arr[2]=0;arr[3]=0;arr[4]=1.

This is simplified since I don't know the length that the second term(1001) will be: it could be (0101011,...)

I need to use float casts because downsteam in the code it will be required that they are float.

Originally Posted by vart

also what about

ffe0 - is this string possible?

No the second term will always be in the form of binary:so only 0's and 1's

So, you need a loop that iterates over the substring, and extracts each digit using the line2[j]-'0' form. Casting to float is probably not necessary, as C allows conversion from int to float and float to int without casts.

--
Mats

Originally Posted by matsp

So, you need a loop that iterates over the substring, and extracts each digit using the line2[j]-'0' form. Casting to float is probably not necessary, as C allows conversion from int to float and float to int without casts.

--
Mats

Yes although I believe I tried the -'0' method and it did not work. I had to subtract 48: hence the

Code:

(double)line2[j]-48

Originally Posted by deprekate

Yes although I believe I tried the -'0' method and it did not work. I had to subtract 48

Since '0' and 48 have the same value (and type!) where ASCII is concerned, that should not be the case.

Originally Posted by laserlight

Since '0' and 48 have the same value (and type!) where ASCII is concerned, that should not be the case.

Assuming of course the compiler and the input text are both using ASCII or a character set derived from ASCII - but that is highly likely, so I don't really see how it can go wrong.

--
Mats