CODEWITHSUNDEEP
regexshellawksed
Garreth
Garreth

Reputation: 1095

Replace the line above a matched pattern

I have a html file which has a table in it as follows:

<td>unknown</td> <td>blah11</td> ... <td>unknown</td> <td>blah24</td> ...

I need to match on 'blah24' and replace the 'unknown' above with 'test' to give:

<td>test</td> <td>blah24</td>

I have tried to research sed & awk solutions but have only been able to find either 'inserts before the matched pattern' or 'replace after the matched pattern'.

I only have basic shell tools available on a stand alone machine using cygwin.

Upvotes: 0

Views: 1218

Answers (2)

123
123

Reputation: 11246

In sed you could do

sed -n '/blah24/{x;s/unknown/test/;x};x;1!p;${x;p}' file

<td>unknown</td>
<td>blah11</td>
...
<td>test</td>
<td>blah24</td>
...

Every line it exchanges the current line for the hold buffer. If it finds blah24, then it exchanges the held line, changes it to test, then swaps it back. It prints for every line except 1(as it would print a blank space).

Upvotes: 1

Ed Morton
Ed Morton

Reputation: 204628

$ awk 'NR>1{if(/blah24/) sub(/unknown/,"test",prev); print prev} {prev=$0} END{print prev}' file
<td>unknown</td>
<td>blah11</td>
...
<td>test</td>
<td>blah24</td>
...

You didn't tell us what to do if blah24 appears as part of a longer string or if you really wanted to specifically replace unknown or just whatever was within the tags or various other crucial details but hopefully you can figure all that out given the above.

Upvotes: 2

Related Questions