[SOLVED] how to sed remove text on Nth occurrence of same word

BW-userx · 02-21-2018, 03:26 PM

from what I've read online, it states just write this.

Code:

sed 's:exit(0):\/* removed *\/:2' /home/userx/bin/main.c

the '2' indicating the 2nd occurrence of that word, nope, it works if I use nothing or g but replaces both occurrences of that word. and yes that this the correct number, it , exit(0), is only in that file 2 times.

Code:

            if (event.xbutton.time - lastTime < 250) {
                if (system(cmd) == -1) {
                fprintf(stdout, "Failed to run command:%s\n", cmd);
                exit(0);
                }
            } else {
                lastTime = event.xbutton.time;
            }
            } else if (event.xbutton.button == Button3) {
            exit(0);
            }
            /*printf("ButtonRelease\n");*/
            break;
        }
    }

running without g

Code:

    case ButtonRelease:
            if (event.xbutton.button == Button1) {
            if (event.xbutton.time - lastTime < 250) {
                if (system(cmd) == -1) {
                fprintf(stdout, "Failed to run command:%s\n", cmd);
                /* removed */;
                }
            } else {
                lastTime = event.xbutton.time;
            }
            } else if (event.xbutton.button == Button3) {
            /* removed */;
            }
            /*printf("ButtonRelease\n");*/
            break;
        }
    }
    usleep(10000);
    }
    /* we should never get here */
    return (0);
}

#ran this
userx@slackwhere101:~
$ sed 's:exit(0):\/* removed *\/:' /home/userx/bin/main.c

I only need that last one to be removed leaving the semi-colon.

danielbmartin · 02-21-2018, 03:38 PM

Maybe this thread ...
https://www.linuxquestions.org/quest...or-awk-800171/
... will be helpful.

Daniel B. Martin

.

BW-userx · 02-21-2018, 04:18 PM

Quote:

Originally Posted by danielbmartin

Maybe this thread ...
https://www.linuxquestions.org/quest...or-awk-800171/
... will be helpful.

Daniel B. Martin

.

Oh yeah there it is. I've been looking for that.
Thanks

MadeInGermany · 02-22-2018, 01:15 AM

The 2 modifier means the 2nd match within the line.
Example

Code:

exit(0); exit(0); exit(0);

Likewise the g modifier matches all occurences within the line.
Note that the repeated matches go from left to right and must not overlap (what matched once is left from the next match).

For counting the matches in the whole file you need a variable or an awful trick. And sed does not have variables...

BW-userx · 02-22-2018, 08:33 AM

Quote:

Originally Posted by MadeInGermany

The 2 modifier means the 2nd match within the line.
Example

Code:

exit(0); exit(0); exit(0);

Likewise the g modifier matches all occurrences within the line.
Note that the repeated matches go from left to right and must not overlap (what matched once is left from the next match).

For counting the matches in the whole file you need a variable or an awful trick. And sed does not have variables...

I know what the g does, but that 2 nd was not working like I was reading it is done, then I came across something that said the very first one is not counted, or some such strange this as that.

because their are only 2 exit (0) within that entire file, and I just needed to remove that 2nd one.
this is what worked for me.

Code:

sed -i -e  '0,/exit(0)/! {0,/exit(0)/ s/exit(0)/\/*removed exit to keep dockapp running *\//}' $PWD/src/main.c
sed -i -e '0,/strcat/ s/strcat/strcpy/'  $PWD/src/main.c

whereas the second line I needed to get just the very first occurrence of the same thing which was in the file 3 times, to fix that.

MadeInGermany · 02-22-2018, 09:42 AM

You are using two nested scopes (only GNU sed allows this).
The following is a general+portable sed trick that (mis-)uses the hold buffer for counting.

Code:

sed -i '
/exit(0)/{
 x
 # add a dot to the hold space
 s/$/./
 # #dots==#desired?
 /^.\{2\}$/{
  # do the replacement action
  x
  s:exit(0):/* removed */:
  x
 }
 x
}
' $PWD/src/main.c

BW-userx · 02-22-2018, 09:57 AM

Quote:

Originally Posted by MadeInGermany

You are using two nested scopes (only GNU sed allows this).
The following is a general+portable sed trick that (mis-)uses the hold buffer for counting.

Code:

sed -i '
/exit(0)/{
 x
 # add a dot to the hold space
 s/$/./
 # #dots==#desired?
 /^.\{2\}$/{
  # do the replacement action
  x
  s:exit(0):/* removed */:
  x
 }
 x
}
' $PWD/src/main.c

it is for a slackbuild, so that is all it was going to be used on, but, that worked too, (of course). as long as it works.thanks.