sed with delimiter, bourne

uribo · 04-18-2003, 12:02 AM

this line is part of /etc/shadow file -->
shelly:$$/gtrygDi$rtklncAD.lv3w4Nr6nG17.:12147:0:99999:7::12783:

i need to sed the "encrypted password" (i.e. the second field) and replace it with a "*".

becos of the delimiter, i have problem.

can anyone tell me if sed can do the work? if not, what function can i use? thanks!

jailbait · 04-18-2003, 11:00 PM

I have changed that field with my text editor, kwrite. Once I screw it up I usually use my rescue system command line text editor, mp, to fix it.

cuckoopint · 04-19-2003, 01:12 AM

ok, you need to explain more.

Code:

awk -F : '{print $2}' /etc/shadow    #this will list all second 'row' entries
#I guess you can grep for 'shelly' (or some other matching pattern)
sed -i s/$var/* /etc/shadow

is that what you want to do? I recommend trying this out on some temporary file first...
; )

uribo · 04-19-2003, 09:26 AM

hi,

yes, that is what i want... but can i have a one line command?

cuckoopint · 04-19-2003, 09:33 AM

make a bash script, or use the ';' or '&&'

ie.

ls blah; grep blah && cp blah

(I would recommend the script)

; )

uribo · 04-19-2003, 09:44 AM

hi,

yes, it is part of my script. i tried this:

`cat /etc/shadow |awk -F: '{print $2}' |grep "^shelly" | sed "s/"$2"/*/"` > temp.doc && mv temp.doc /etc/shadow

it erases my whole file... why???

cuckoopint · 04-19-2003, 11:15 AM

ok, the problem I have is that I don't know how to quote variable in sed. It doesn't seem to like acting like everyone else.

Code:

#!/bin/sh
for i in `grep "^shelly" shadow | awk -F: '{print $2}'`
do
        echo $i
        sed -i s/"$i"/*/ /etc/shadow
done

Now, you had a couple problems. First, you must grep, then awk.
and after 'sed', you cannot move the file back. (because you would be only moving the * to /etc/shadow). Instead you use the -i option to replace within the file. Again, the only problem is that sed doesn't like "$i"

BTW, I only put the echo in there for testing.
; )

uribo · 04-19-2003, 10:37 PM

hi,

thanks very much for yur prompt reply. i tried many ways, in the end, i have to do this:

cat /etc/shadow | sed "/^"shelly"/ s/^$[a-zA-Z]*:$$[a-zA-Z0-9$/.]*:$$[0-9]*:$$[0-9]*:$$[0-9]*:$$[0-9]*:$$[0-9]*:$$[0-9]*:$$/\1*\:\3\4\5\6\7\8/" > temp.doc && mv temp.doc /etc/shadow

if you happened to figure out how to call a variable, please post here again. thanks!

uribo · 04-19-2003, 11:30 PM

I did it... almost giving up hope!

for i in `grep "^shelly" /etc/shadow | awk -F: '{print $2}'` ; do
sed 's_'"$i"'_*_' /etc/shadow
done

cuckoopint · 04-19-2003, 11:51 PM

Cool.

But didn't you want that to be sed -i 's_ ... ?

Well, thanks for the heads up.

uribo · 04-20-2003, 12:16 AM

if i use sed -i..., it will erase my whole file. that is why i need to direct out to temp file first.

moses · 04-20-2003, 11:53 PM

Code:

 awk -F: '{print $1FS"*"FS$3FS$4FS$5FS$6FS$7FS$8FS}' shadow

moses · 04-20-2003, 11:55 PM

Quote:

Originally posted by jailbait
I have changed that field with my text editor, kwrite. Once I screw it up I usually use my rescue system command line text editor, mp, to fix it.

If you use vipw -s, it'll check syntax before writing and you won't have to worry about screwing up the syntax.
(man vipw)

uribo · 04-22-2003, 12:22 AM

hi moses,

one question --> how can i update shadow file after insert "*" in "mark" field?

i tried:

cat /etc/shadow | awk -F: '{ if ($1 == "mark") print $1FS"*"FS$3FS$4FS$5FS$6FS$7FS$8FS}' > temp.doc && mv temp.doc /etc/shadow

& it replace the whole file content with only "mark" entry.

thanks

Mik · 04-22-2003, 09:35 AM

Well as you can see above there are many ways to solve your problem. Here would be another way to do it:

sed -e "s/$^shelly$:[^:]*/\1:*/" /etc/passwd > temp.doc && mv temp.doc /etc/passwd