Add character to specific columns using sed or awk and make it a permanent change

Shell Programming and Scripting
1

Hi,

I am writing a shell script where I want that # should be added in all those lines as the first character where the pattern matches.

file has lot of functions defined
a.sh

#!/bin/bash

 fn a {
beautiful evening
sunny day
}
fn b {
}

fn c {
 hello world .its a beautiful day
 my mom is beautiful.
 my doll is beautiful
 all things bright and beautiful
 all creatures great and small
}

so when i execute the command on a.sh, the file should only be modified for places that has beautiful word

so the new file should be
a.sh

#!/bin/bash

 fn a {
# beautiful evening
sunny day
}
fn b {

}

fn c {
# hello world .its a beautiful day
# my mom is beautiful.
# my doll is beautiful
# all things bright and beautiful
 all creatures great and small
}

command i could create was

cat a.sh | grep -i "beautiful" | awk -F" " '{$1="#" $1;}1' OFS=" "

but it doesnt modify in the file. It only gives op on comd prompt. I dont want to create a new file. I want it modified in original file

2

Please use code tags as required by forum rules!

Try

awk '/beautiful/ {printf "#"} 1' file > tmp$$; mv tmp$$ file

search these forums for "in place file replacement".

1 Like
3

Sorry new to unix.com. Didnt explore much.
Thanks!!! It resolved the issue.
Can you explain temp$$?

4

Hello ashima,

You are most welcome to forum, hope you will enjoy learning here. echo $$ will be used to write PID of the current shell.
So tmp$$ is just a Temporary file created in above code, which is later renamed to actual file. Complete explanation is as follows.

 
awk '/beautiful/  #### Searching for string beautiful here.
{printf "#"}      #### If above condition is TRUE then printing string #
1'                #### awk works on pattern of condition and action/operation so if we give condition here 1 which means we are making condition TRUE so default action will be happening here which is print for awk.
file > tmp$$;     #### mentioning Input_file named file here as putting output into file named tmp with $$ shell's pid.
mv tmp$$ file     #### renaming tmp$$ file to actual Input_file which is file.

Thanks,
R. Singh

1 Like
5

You could also try ed or ex with any shell that has a syntax based on the Bourne shell (e.g., ash, bash, ksh, zsh...):

#!/bin/ksh
ed -s a.sh <<-EOF
	g/beautiful/s/^/#/
	w
	q
EOF

These utilities might use a temp file internally, but you won't need to create one on your own this way.

1 Like
6 
sed -n 'p; /{/{ :L n; /}/p; s/\(.*\)/#\1/p; bL; };n;p;' file6
#!/bin/bash

 fn a {
# beautiful evening
# added
# sunny day
 }
# }
# fn b {
 }
# }
#
# fn c {
#  hello world .its a beautiful day
#  my mom is beautiful.
#  my doll is beautiful
#  all things bright and beautiful
#  all creatures great and small
 }
# }

I have tried branching

/}/p;

with

{/}/{p; bL;}}

I just get an error. I have branched successfully with L but can't seem to be able inside another Curly set.
I know I don't yet have the correct output for # on last line of curly set, right now I just want to understand the branching stuff, then I will try to fix the other problem after.
Thanks

7

Not sure what you want to achieve. Your script prefixes almost - not all - lines with a # , except when an opening brace is encoutered. That could be done more easily with e.g. sed '/{/ !s/^/#/' file

n;p; doesn't mean a thing; if you drop it, sed will automatically read a new line and print it.
You don't match the pattern (beautiful) which is required.

8

I know this doesn't work, my question is how to do the, if not, THEN in sed? I guess I use the branching but had less than success with it, anyway let me know.

sed -n 'p;/{/ !s/^/#/; s/^\(.*\)/# \1/' file6

If not

{

then add

#

with the other sustitute statement. Thanks

P.S. Simply put, If I am too confusing maybe show me how to do this particular example with sed, the original question.

9

That is not the original question. The original question was how to put an octothorps at the start of every line that contains the string beautiful . The obvious way to do that in sed is:

sed '/beautiful/s/^/#/' file

but the original request was also to put the results back in the input file. Just using standard sed features (i.e., no -i option), that can be done with:

sed '/beautiful/s/^/#/' file > file.$$ && cp file.$$ file;rm -f file.$$

But, if you did want to add an octothorpe to the lines that do NOT contain beautiful, you can do that in sed with:

sed '/beautiful/n;s/^/#/' file
10

Sorry Don will ask the question as a new one,
thanks

11 
while read line
do
   echo "$line" |grep beautiful >/dev/null 2>&1
   if [ $? = 0 ]
   then
     echo "#$line"
   else
     echo "$line"
   fi
done <a.sh

If you want to write it using a script.

12

Rudic can you explain !s
sed '/{/ !s/^/#/' file

13

From the standards:

Or, translated from standardese, perform the specified substitution (insert an octothorpe at the start of a line) on lines that do not contain an open brace character.

1 Like