replace a specifk line in a file with sed or awk

[Simon Jespersen]

Hey!
I want to replace line nr 220 in a file with the text "my newline"
How can that be done with se or awk.

[James R. Ferguson · Accepted solution]

Hi Simon:

# sed -e '220s/.*/my newline/' file

Regards!

...JRF...

[Hein van den Heuvel]

awk '{if (NR==220) $0="my newline";print}' old > new


hth,
Hein.

[Simon Jespersen]

Thanks for thje reply, but i get strange happening

with the sed line it just behave like im doing a cat on the file
# sed -e '2s/.*/mynewline/' passwd_sijes

Here im trying to replace line 2 in the file im just getting all the file directet to stdout
a71223:
password = SyNHeFTvoiMGw
lastupdate = 1135680886
flags = ADMCHG

a74040:
password = SyNHeFTvoiMGw
lastupdate = 1157718667
flags = ADMCHG

a05243:
password = SyNHeFTvoiMGw
lastupdate = 1157718669
flags = ADMCHG

a57659:
password = SyNHeFTvoiMGw
lastupdate = 1157718670
flags = ADMCHG


can you se why this is happening.

[Simon Jespersen]

How to i use an ksh shell var with awk and sed.
i have the line number in an variable that i want to parse to the awk program. Se $lineNr
cat myfile | awk '{if (NR==$lineNr) $0="my newline";print}' old > new

rhanks

[James R. Ferguson]

Hi (again) Simon:

You must redirect the STDOUT of 'sed' to a new file:

# sed -e '220s/.*/my newline/' file file.new
# mv file.new file

If you would like to do "inplace" updates without having to move the output back over the original file, use Perl:

# perl -pi.old -e 's/.*/mynewline/ if $.==220' file

This will modify line # 220, replacing it with "mynewline". The 'file" will be modified inplace with a backup copy named 'file.old'. If you don't want the backup, do:

# perl -pi -e 's/.*/mynewline/ if $.==220' file

Regards!

...JRF...

[spex]

Hello,

> How to i use an ksh shell var with awk and sed.

$ LN=2
$ awk -v lineNr=${LN} '{if (NR==lineNr) $0="my newline";print}' < old > new1
$ sed "${LN}s/.*/myline/" < old > new2

PCS

[James R. Ferguson]

Hi (again):

> How to i use an ksh shell var with awk and sed.i have the line number in an variable that i want to parse to the awk program. Se $lineNr

cat myfile | awk '{if (NR==$lineNr) $0="my newline";print}' old > new

You pass variables to 'awk' using the '-v' switch and argument:

# awk -v lineNr=220 '{if (NR==lineNr) $0="my newline";print}' file file.new

*Notice* that the 'lineNr' variable does not have the sigil ($) when referenced within the 'awk' script, nor does it have it when it is declared. However:

# lineNR=220
# awk -v lineNr=${lineNr} '{if (NR==lineNr) $0="my newline";print}' file file.new

...is necessary since you are now using a shell variable's value.

Do not use 'cat' to open and read a file and then pipe that output to 'awk'! That's a waste of a process. 'awk' uses the commandline arguments as filenames to process.

For 'sed' use double quotes like this:

# L=220
# sed -e "${L}s/.*/mynewline/" file > file.new
# mv file.new file

Regards!

...JRF...

[Simon Jespersen]

cool stuff, thanks a lot.

[Peter Nikitka]

Hi,

one addition to JRF's sed lines:
Keep in mind, when putting such stuff in a script, to check the success of the commands.

Being not root or acting via NFS or ..., you may have write access to the directory but no read access the the file or vice versa, so something like
# L=220
# sed -e "${L}s/.*/mynewline/" file > file.new
# mv file.new file

may destroy 'file'. I suggest always
sed -e "${L}s/.*/mynewline/" file > file.new && mv file.new file

which else won't try a 'mv' when 'file.new' cannot be created.

mfG Peter