[SOLVED] Replace string in several files at position

ceantuco · 04-28-2017, 12:37 PM

Hi Everyone,

I have over 35k files and I would like to replace the first 12 characters of certain string on all files with 'x'.

I used grep -ir 'string_name' /home/user/files/* to list the files that contain the specific string I want to change. How can I use SED or AWK to accomplish this task?

Please advise

Thank you!

Didier Spaier · 04-28-2017, 02:59 PM

Lets set A as the initial string and B as the modified string.

If you use GNU sed, write:

Code:

sed -i s/$A/$B/g /home/user/files/*

Turbocapitalist · 04-28-2017, 03:05 PM

Which first twelve? Can you describe your pattern a little more?

I'd anchor the pattern to the beginning of the line so that matches only the first letters not anywhere else on the line. Leaving off the /g modifier also restricts the pattern to a single attempt per line:

Code:

sed -e "s/^$A/$B/;" /path/to/files/*.txt

And be sure to leave off the -i until you are sure the pattern works.

See:

Code:

man sed
man 7 regex

Also, it is best if you have a backup using tar before experimenting.

ceantuco · 04-28-2017, 03:09 PM

Thank you for your responses!

Okay so basically each file contain x number of entries as shown below:

Acct: 1111111111111111

and I would like to replace all entries on all files as follows:

Acct: XXXXXXXXXXXX1111

Yes, I will back up the data before I start experimenting. Wouldn't want to destroy all information within those files.

Thank you!

Turbocapitalist · 04-28-2017, 03:13 PM

I'm not sure about which pattern you want to zap but if you want to replace the first twelve letters of all lines, then the following would do it:

Code:

sed -r -e 's/^.{12}/xxxxxxxxxxxx/' *.txt

If you want a more precise search pattern, then replace .{12} with a more detailed pattern.

ceantuco · 05-01-2017, 07:55 AM

Thank you Turbocapitalist. I will give your suggestion above a try. If it does not work, I will have to be more specific with what I would like to accomplish.

ceantuco · 05-01-2017, 10:23 AM

I tried what suggested but it did not work. Basically, my files contain many lines of text which some of these lines start with:

Code:

  ACCT: 3478913473413536

which I would like to change to:

Code:

  ACCT: XXXXXXXXXXXX3536

I tried the code below to test and I was able to locate the line where "Acct: " is present and replace it with "Cess: "

Code:

grep -irl "ACCT: " file.txt | xargs sed -i 's/ACCT: /Cess: /gI'

What I need is a away to tell SED to replace the first 12 digits after "ACCT: " with "X"

Hopefully the explanation above is easier to understand.

Thank you!

pan64 · 05-01-2017, 10:25 AM

Code:

sed '/^ACCT:/s/[0-9][0-9] .... /XXXXXXX/' filename

this should work, but you need to count [0-9] and XXX

ceantuco · 05-01-2017, 10:42 AM

Thank you Pan64; unfortunately, it did not work.

pan64 · 05-01-2017, 10:52 AM

so please post what did you really try, what's happened, what did you expect.....

ceantuco · 05-01-2017, 10:54 AM

Of course; I tried the command below:

Code:

sed '/^ACCT:/s/[0-9][0-9] .... /XXXXXXX/' filename

Then I opened the file to see if the string(s) has been changed but they were not:

Code:

  ACCT: 3478913473413536

I expected:

Code:

  ACCT: XXXXXXXXXXXX3536

Thank you

pan64 · 05-01-2017, 11:06 AM

of course, you need to repeat [0-9] 12 times, if you want to replace 12 digits. or course you must not add ..... and spaces.
of course, you need to specify the replacement string exactly too.

ceantuco · 05-01-2017, 11:13 AM

I was not trying to be sarcastic with 'of course'.... no need to repeat it x times.

would this work?

Code:

sed '/^ACCT:/s/[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]/ACCT: XXXXXXXXXXXX/' filename

sundialsvcs · 05-01-2017, 11:14 AM

s/^.d{12}/xxxxxxxxxxxx/ ...?

I mean this to be an expression which replaces "12 occurrences of 'anything', anchored at the start of the string," with 12 x's.

pan64 · 05-01-2017, 11:24 AM

try that, and you will see. but I would really like to ask you to understand/learn how sed works:

Code:

/^ACCT:/   look for lines beginning with ACCT:
s          means the command substitution
/          delimiter
something  to search for
/          delimiter
string     replacement string
/          closing delimiter
           optional modifiers, at this time there is no any

from this point of view probably the replacement string is not ok (what you specified), but you need to know that exactly

{12} works only with sed -r