I would break the program in different tasks.
The first step is to be able to read a pair of lines, the first line which tells you the number of numbers to read, and then the second line to read the actual numbers. For this, a function called something like read_set
might be useful. It should be able to return the numbers read, and signal end of file as well as errors. For this, we can define a data structure such as:
struct numbers {
long *data; /* or choose a type depending upon your needs */
size_t len;
};
and then we can declare our function with the prototype:
int read_set(FILE *fp, struct numbers *num);
The function will allocate memory for num->data
, and set num->len
to the correct value. It returns 0 for success, and a set of error conditions otherwise. We might get fancy and use an enum
for return statuses later. For now, let's say that 0 = success, 1 = end of file, and everything else is an error.
The caller then calls read_set()
in a loop:
struct numbers numbers;
int status;
while ((status = read_set(fp, &numbers)) == 0) {
/* process numbers->data, and then free it */
}
if (status == 1) {
/* hit end of file, everything is OK */
} else {
/* handle error */
}
For implementing read_set()
: it has to read two lines. There are many implementations of reading a full line in C, so you can use any of them, and read a line first, then sscanf()
/strtoul()
it for one number (check its return value!). Once you have the number of numbers, n
, you can read the next line in memory, and do:
num->data = malloc(n * sizeof *num->data);
num->len = n;
You can then repeatedly call sscanf()
or strtol()
to store numbers in num->data
. You should put in checks to make sure exactly n
numbers are on that line.
Note that you can write read_set()
in other ways too: read a line character by character, and parse the numbers as you read them. This has the advantage of going over the data only once, and not needing a big buffer to store the whole input line in memory, but the disadvantage is doing low-level stuff yourself and reading data character-by-character may be slow.