Find and replace last line in a file

Shell Programming and Scripting
1

Hi I am having a file which has like this content shown below

Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,12-Jun,t

I need to replace last line like this 
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-August,t
2

Try : for given input, with the assumption that there is no empty new line at the end

$ tac file | awk -F, 'NR==1{$2=replace}1' OFS=\, replace=10-August | tac
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-August,t

OR

--edit----

$ awk -F, 'NR == 1 ; getline ; END{$2=replace;print}' OFS=\, replace=10-August file
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-August,t
3

Hi,
With sed ( no tested ) :

sed -e '$s/12-Jun/10-August/' file

This command replace first "12-Jun" by 10-August and only in last line of file.

Regards.

5

Hi i am having multiple files with different date at last line which s at 2nd position

is it possible to achieve to change the 2nd position date to one single date
for eg:

file1
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,12-Jun,t

file2
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-Jun,t

file3
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,16-Dec,t

i need to replace in all the files last line like below without affecting any other data in each of the files

file1
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-Aug,t

file2
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-Aug,t

file3
Aaa,bb,cc,dd
Xxx,yy,d,12
Dodd,10-Aug,t
6 
sed -i '$s/,[^,]*,/,10-August,/' file1 file2
7

One way using tac and awk

#!/bin/bash

# Replace here whatever you want in 2nd field
replace=10-Aug

for file in file*; do
    tac $file | \
        awk -F, 'NR==1{ $2 = replace}1' OFS=, replace=$replace | tac >fl.tmp && cat fl.tmp >$file
done

rm fl.tmp
1 Like
8

Hi ,

i would say tac is not working in my shell even i tried in bash mode, can you help me how i can use tac command in sunsol OS

9

I assume that last line is not blank

Try :

#!/bin/bash

# Replace here whatever you want in 2nd field
replace=replaces

for file in file*; do
        awk -F, 'NR == 1 ; getline ; END{$2=replace;print}' OFS=\, replace=$replace $file >fl.tmp 
        cat fl.tmp >$file
done

rm fl.tmp

If you want to run this on a Solaris/SunOS system, use /usr/xpg4/bin/awk , /usr/xpg6/bin/awk , or nawk instead of the default /usr/bin/awk .

1 Like
10

Here is an alternative way that doesn't require tac:

#!/bin/ksh
IAm=${0##*/}
if [ $# -lt 2 ]
then    printf "Usage: %s replacement_string file...\n" "$IAm" >&2
        exit 1
fi
rep="$1"
shift
for file in "$@"
do      printf "%s: Processing file %s\n" "$IAm" "$file"
        ed -s "$file" <<-EOF
                \$g/\([^,]*,\)[^,]*/s//\1$rep/
                w
                q
EOF
done

This was tested used ksh and bash, but any shell that recognizes basic Bourne shell syntax should work just as well. It was also tested using ed and ex. On most systems ed is a little bit smaller and faster than ex, but your mileage may vary.

To use it to make the changes requested in message #4 in this thread, save the above script in a file (e.g., tester), make it executable using:

chmod +x tester

and run it with:

./tester "10-Aug" file[123]
1 Like
11

hi it worked if i use tail -r instead of tac thanks for you help

12

Hi Don,

Is it possible to replace any field or coloumn irrespective of 2nd column i have mentioned or can i use the which col to be replaced by giving user input in your code ???

13

I'm not sure I understand what you want to do.

I am guessing that you want the script to take an additional operand that specifies which field in the last line of each file is to be replaced, as in:

ScriptName replacement_string field_number file...

If that is what you want, we could add code before the for loop to calculate a BRE to use in ed to select the given number of fields minus one to be copied before the given field is replaced.

Note that my script verified that the last line in the file did have the appropriate number of fields and silently ignores the request if the last line is not in the proper format. Would you prefer to get an error message if the last line is not in the proper format?

14

Hi don

Actually I need to replace the last line which comes in 4th position so is it possible to replace any position in last line using your code can you share it

15

If by "last line in 4th position", you mean that you want to change the 4th comma delimited field in the last line, you could try something like:

#!/bin/ksh
IAm=${0##*/}
replacement="$1"
field="$2"
if [ $# -ge 3 ]
then    if [ ${#field} -eq 0 ] || [ "${field}" != "${field%*[^0-9]*}" ] ||
                [ $field -eq 0 ]
        then    printf "%s: invalid field# <%s>\n" "$IAm" "$field" >&2
                shift $#
        fi
fi
if [ $# -lt 3 ]
then    printf "Usage: %s replacement_string field# file...\n" "$IAm" >&2
        exit 1
fi
if [ "$2" -eq 1 ]
then    edCommandMiddle='[^,]*/'
else    BRE=''
        while [ $((field--)) -gt 1 ]
        do      BRE="${BRE}[^,]*,"
        done
        edCommandMiddle='\('"$BRE"'\)[^,]*/\1'
fi
shift 2
# printf "ed command will be: %s\n" '$'"s/^$edCommandMiddle$replacement/"
for file in "$@"
do      printf "%s: Processing file %s\n" "$IAm" "$file"
        ed -s "$file" <<-EOF
                \$s/^$edCommandMiddle$replacement/
                w
                q
EOF
done

As before, the 1st operand you need to give to this script is the string you want to use to replace the current contents of the selected field. The 2nd operand is the field number for the field you want to change. And the remaining operands are pathnames to the files you want to change. So, if you named this script tester and you want to change the 4th field of the last line in files named file1 , file2 , and file3 to "new text", use:

./tester 'new text' 4 file[123]
2 Likes