Very Computer

hi group!

i have a huge text file with lines that llok like that:

123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
 (of course the lines differ :)

somethimes in that textfiles ther are errors, there are sometimes
newlines there that should not be, like this:

123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblah
blahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon
123456     blahblahbla      moreblahblahblah    andsoon

what i want now is a sed script, that does the following:

if a line starts with a letter or parenthesis, join it with the
preceeding line. i have managed to do so with this example:

cat test| sed -e :a -e '$!N;s/\n[A-Za-z\(\)]//;ta' -e "P;D"

but the problem is that this deletes the first letter of the wrong row
because of
s/\n[A-Za-z\(\)]//
.

so this:

123456     blahblahbla      moreblah
blahblah    andsoon

results in:

123456     blahblahbla      moreblahlahblah    andsoon
                                                                ^^
there should be a "b"

can i somehow preserve that letter or paste what was deleted but
without the preceeding newline?

big thanks in advance!!

Quote:

> cat test| sed -e :a -e '$!N;s/\n[A-Za-z\(\)]//;ta' -e "P;D"

> but the problem is that this deletes the first letter of the wrong row
> because of
> s/\n[A-Za-z\(\)]//
> .

It looks like the following works:
cat test| sed -e :a -e '$!N;s/\n\([A-Za-z()]\)/\1/;ta' -e "P;D"

Berend

Yeeee!!
thank you very much, you're a life saver :)
the sed manpage could really use some examples i think

best wishes, armin