Replace line in text-file using C

Question

I want to change lines which contain the # symbol in a text file with heet using C.

I have tried it this way, but it did not work thoroughly, it just replaces the characters & overwrites not the whole string, like I want.

Is there any other trick to remove or delete a whole line from the file? So, we can easily replace it.

myfile.txt: (before execution)

Joy
#Smith
Lee
Sara#
Priyanka
#Addy

Code:

#include <stdio.h>
#include <string.h>

int main() {
    FILE *pFile;
    fpos_t pos1, pos2;
    int line = 0;
    char buf[68]
    char *p;
    char temp[10] = "heet";

    pFile = fopen("myfile.txt", "r+");

    printf("changes are made in this lines:\t");    
    while (!feof(pFile)) {
        ++line;
        fgetpos(pFile, &pos1);          

        if (fgets(buf, 68, pFile) == NULL)  
            break;

        fgetpos(pFile, &pos2);

        p = strchr(buf, '#');

        if (p != NULL) {
            printf("%d, " , line);
            fsetpos(pFile, &pos1);  
            fputs(temp, pFile);
        }
        fsetpos(pFile, &pos2);
    }

    fclose(pFile);
    return 0;
}

myfile.txt: (after execution)

Joy  
heetth  
Lee  
heet#  
Priyanka  
heety  

Output:

changes are made in this lines: 2, 4, 6,  

myfile.txt: (I want to get)

Joy  
heet  
Lee  
heet  
Priyanka  
heet  

Answer 1

The best way of doing what you want is to use a utility like sed. It is faster and uses less memory than anything you (or I) would write.

That aside, let's assume you want to go ahead and write it yourself anyway.

A file is just like a long array of bytes. If you want to increase or decrease the length of one line, it affects the position of every byte in the rest of the file. The result can be shorter (or longer) than the original. As the result can be shorter, modifying the file in place is a bad idea.

The following pseudo-code illustrates a simple approach:

open original file
open output file
allocate a line buffer that is large enough
read a line from the original file
do
  return an error if the buffer is too small
  manipulate the line
  write the manipulated line to the output file
  read a line from the original file
loop until read returns nothing

sed does it much smarter. I once saw an explanation on how sed works, but my google karma can't seem to find it.

Edit: How to do it using sed:

 sed -e 's/.*\#.*/heet/g' myfile.txt

The s, or substitute, command of sed can replace one string, or regular expression, with another string.

The above command is interpreted as:

replace any line that has a # somewhere in it with heet. The final g tells sed to do this globally, i.e. in the entire file.

Edit2: By default, sed writes to standard output. To rewrite the file you should redirect the output to a file and then rename it. In linux, do the following (you can run command line stuff from C with system):

sed -e 's/.*\#.*/heet/g' myfile.txt > temp_file123.txt
rm myfile.txt
mv temp_file123.txt myfile.txt

From C:

system("sed -e 's/.*\#.*/heet/g' myfile.txt > temp_file123.txt");
system("rm myfile.txt");
system("mv temp_file123.txt myfile.txt");

If you want to do it with just one call to system, put all the command line stuff in a shell script.