Find first occurrence from the end with awk or sed

Question

Is is possible to find first string occurrence from the end of file using awk or sed? Or I will need to use some scripting language like Python or Perl?

Answer 1

You can revert the order of lines in the file with tac (the opposite of cat) and then use grep:

tac file | grep -m 1 STRING

grep -m1 gives you only the first occurrence.

Alternatively you may just pipe grep to tail:

grep STRING | tail -n1

---

If you want to use awk:

awk 'BEGIN{res=""}/STRING/{res=$0}END{if(res!=""){print res}}' file

Explanation:

# Initialize result with an empty string
BEGIN{res=""}
# If STRING is found, store the current line in res.
# This will overwrite a previous result, giving you
# always only the last occurrence.
/STRING/{res=$0}
# Once the end of input has been reached print the result
# if STRING was found.
END{if(res!=""){print res}}

Answer 2

Using sed

sed -n '/string/h;${x;p}' infile

Using awk

tac infile | awk '/abc/{print;exit}' 

Test Results:

akshay@db-3325:/tmp$ cat infile
abc 1
xyz 2
pwd 3
cwd 4
aks 5
axy 6
abc 7
xyz 8
nwd 9

akshay@db-3325:/tmp$ sed -n '/abc/h;${x;p}' infile
abc 7

akshay@db-3325:/tmp$ tac infile | awk '/abc/{print;exit}' 
abc 7

Answer 3

Following tac and awk combination will help you to read the very first occurrence of any string very quickly.

tac Input_file | awk '/STRING/{print;exit}'

Explanation: tac will reverse the Input_file then awk will check for the string and as soon as it gets it will print the line and exit from it, since are only bothered about very first occurrence so no need to read whole Input_file here.