Question

I have a “file” as a resource. I can only use read(), write() and fstat() it. This file is a text file which I would like to parse.

Normally I use fgets() to read the text file line by line and parse it. How can I do this in this case?

FILE *fp;
char buffer[128];

fp = fopen( "/home/txtfile", "r" );
if (fp == NULL){
    perror("file missing");
}
while (fgets (buffer, sizeof(buffer), fp) != NULL) {
       //some code
}

How can I do the same with read() ?

Is this right?

int fd = open("/dev/file",O_RDONLY);
if (fd==-1) {
    printf("Failed to open file!!!\n");
}
while (fgets (buffer, sizeof (buffer), fd) != NULL) { 
    //some code 
}

Answer 1

Unless your file is huge, if you’re using read(), it would be easier to read in the entire file, then operate on the memory buffer, rather than in discrete chunks. That is, unless each line is of a fixed length.

I’d do something like this:

int rc;

int fd = open("data", O_RDONLY);  // open the file for reading
if (fd == -1) {
    // error
}

// to be thorough, do a stat() here to find how big to make the buffer
struct stat sb;
rc = fstat(fd, &sb);
if (rc == -1) {
    // error
}

char *buffer = calloc(1, sb.st_size);

int bytes_read = 0;
// read in entire file; each read() can be incomplete, hence why it's in a loop,
// and reading from/writing to increasing sections of the memory and file
while ((rc = read(fd, (buffer + bytes_read), (sb.st_size - bytes_read))) > 0) {
    if (rc == -1) {
        // error
    }

    bytes_read += rc;
}

close(fd);

// Now, to read it line-by-line...
char line[128]; // 128 is arbitrary
int pos = 0;
while ((rc = sscanf(buffer + pos, "%127[^\n]\n", line)) > 0) {
    pos += strlen(line) + 1;
    // do stuff with line
}

return 0;

Then you can operate on your memory buffer line-by-line by scanning for newlines, or using sscanf(). Also make sure to free() your buffer!

Edit: I’ve added some example code for using sscanf() to handle your buffer. If you know the format of the lines (you say you’re parsing them) you might be able to make better use of sscanf() by using the format specifiers. All of this is untested, by the way.