CODEWITHSUNDEEP
regexbashsed
suma
suma

Reputation: 31

How to match more than one pattern and delete a line using sed?

I have a line in a file as shown below

abcd    x  10.10.10.10

sometimes the same line might have some extra fields like below

abcd    123  AB  x  10.10.10.10

So now I have to delete this line from the file. My match pattern would be to search for abcd, x and 10.10.10.10 in a line and delete it as these are the only fixed values. Is there way to match multiple patterns with logical and in sed and delete the line? I have tried using regex, but since the pattern is not always the same here, that does not work. I need a solution with sed. I tried to search other websites and stackoverflow. Could not find any solution which works for me.

Upvotes: 1

Views: 196

Answers (2)

potong
potong

Reputation: 58578

This might work for you (GNU sed);

sed -r 'h;s/$/\nabcd x 10.10.10.10/;ta;:a;s/(.+)(.*\n.*)\<\1\>/\2/;ta;/^\s*\n\s*$/d;x' file

This makes a copy of the current line. Then appends the required fields to the end of the line separated by a newline. Using substitution and backreferences, matching fields are removed from the pattern space until no further matches can be made. If the remaining string consists of nothing but zero or more spaces separated by a newline this would identify a line to be deleted, otherwise the line is to be kept so the copy is reinstated.

Upvotes: 0

Ed Morton
Ed Morton

Reputation: 204721

This is a job for awk, not sed:

$ cat tst.awk
BEGIN {
    split("abcd x 10.10.10.10",flds)
    for (idx in flds) {
        targets[flds[idx]]
    }
}
{
    delete cnt
    for (i=1;i<=NF;i++) {
        cnt[$i]++
    }

    for (fld in targets) {
        if (cnt[fld] == 1) {
            delete cnt[fld]
        }
    }

    for (fld in cnt) {
        # If we reach here then one or more of these conditions is true:
        # a) there was a field that is not one we targetted
        # b) there was a target field that occurred multiple times
        # c) a target field was missing
        print
        next
    }
}

$ awk -f tst.awk file
abcd    123  AB  x  10.10.10.10

Since awk is available on all UNIX installations just like sed is, there's no reason to force yourself to try to use sed for this.

Upvotes: 2

Related Questions