I've been wrestling with sed for a while and have been unable to arrive the correct solution to the problem.

I have a file containing a large number of XML 'documents'. Most of them are contained on one line, but occassionally, some have embedded newlines. Using sed with the N to grab the next line isn't sufficient, since there may be more than two newlines. So, looking at the sample data below, the goal is to remove the newline for every line that does *not* end with the string </AUDIT>.

So... this data:
<AUDIT>all on one line</AUDIT>
<AUDIT>on two lines
this one is</AUDIT>
<AUDIT>of three lines, this it the first
this the second
and finally, the third</AUDIT>
<AUDIT>all on one again</AUDIT>
<AUDIT>this one
stretches across
a quite a few
lines
you never know how
many</AUDIT>
.
.
.
should end up looking like this:
<AUDIT>all on one line</AUDIT>
<AUDIT>on two lines this one is</AUDIT>
<AUDIT>of three lines, this it the first this the second and finally, the third</AUDIT>
<AUDIT>all on one again</AUDIT>
<AUDIT>this one stretches across a quite a few lines you never know how many</AUDIT>

TIA,

David

Not a good way, but a simple way.

Run your program until wc -l returns the same value compared to the last run

Luke Chi

Take a look at this link:

http://main.rtfiber.com.tw/~changyj/sed/html/p.20020306a.html

It might do what you want.

Many thanks nails, that worked - I owe you a beer.

David

you are welcome.