A tricky sed problem 2

[butterfm]

Hi,

I have similar to the following text :

I would like this[Unloading randomclass1]
[Unloading randomclass2]
[Unloading randomclass3]
[Unloading randomclass4]
to be a single sentence

I need to end up with :

I would like this to be a single sentence

There number of lines between the required parts is unknown as is the text (it differs from line to line)

What I have so far is this :

sed '/Unloading/{
N
s/\[Unloading.*\n//g
}' somefile

But all this gives me is :

I would like this[Unloading randomclass2]
[Unloading randomclass4]
to be a single sentence

because it doesn't re-process the line again once it has brought the next line up.

Any sed Guru's who may be able to help ?

Thanks,
Matt.

 

[TrojanWarBlade]

This is not that simple really because the filter needs to know when to insert a space and when not to.
I have a perl pipeline that does it.

echo 'I would like this[Unloading randomclass1]
[Unloading randomclass2]
[Unloading randomclass3]
[Unloading randomclass4]
to be a single sentence
' | perl -ne 's/\[Unloading randomclass.\]\n/ /g;print;' | perl -lne 's/\s\s+/ /g;print'

produces: "I would like this to be a single sentence"

Is that any use?


Trojan.

 

[feherke]

Hi

At first glance :

sed -n 's/\[Unloading.*\]//; H; ${ g; s/\n//g; p }' somefile

This will output the whole text in one line. You did not specified if there is other text to keep in separate line beside the mentioned sentece.

Feherke.

http://rootshell.be/~feherke/

 

[TrojanWarBlade]

Pinching feherke's idea I got:

echo 'I would like this[Unloading randomclass1]
[Unloading randomclass2]
[Unloading randomclass3]
[Unloading randomclass4]
to be a single sentence
' | sed -n 's/\[Unloading.*\]/ /; H; ${ g; s/\n//g; p }' | sed -n 's/   */ /g; p'

Trojan.

 

[butterfm]

Thanks guys. Apologies but I didn't give quite enough detail.

The file may be something like this any number of times.

I would like this[Unloading randomclass1]
[Unloading randomclass2]
[Unloading randomclass3]
[Unloading randomclass4]
to be a single sentence
Ignore this line here
I would like this[Unloading randomclass1]
[Unloading randomclass2]
[Unloading randomclass3]
[Unloading randomclass4]
to be a single sentence

The results I would like to get out are :


I would like this to be a single sentence
Ignore this line here
I would like this to be a single sentence

Apologies for the lack of detail.

Thanks,
Matt.

 

[butterfm]

TrojanWarBlade,

Your solution is very nearly there. The only issue remaining is that I get everything on one line.

e.g.

I would like this to be a single sentenceIgnore this line hereI would like this to be a single sentence

Thanks,
Matt

 

[feherke]

Hi

With minimal changes, but supposed that you allways have more then one consecutive Unloading things :

sed -n 's/\[Unloading.*\]//; H; ${ g; s/\n\{2,\}/ /g; p }' sentence.txt

Feherke.

http://rootshell.be/~feherke/

 

[butterfm]

Thanks very much to both of you. Feherke's solution works a treat. As far as I can see there'll always be more than one consecutive "Unloading" line.

I think I need to do a lot more reading up on sed.

Many thanks,
Matt.

 

[TrojanWarBlade]

feherke is obviously much more of a "sed" guy than I.
Well done feherke.


Trojan.

 

[feherke]

Hi

Thanks, but my [tt]sed[/tt] knowledge is just above the minimal. I was close to cry when I saw what Aurelio Marinho Jargas can do :

http://aurelio.net/sed/sokoban/

Feherke.

http://rootshell.be/~feherke/

 

[TrojanWarBlade]

I used to write awk and sed many years ago but I haven't bothered for nearly 10 years since I got into perl.
I find that sed and awk can be a little more succinct on occasion but generally, perl is far more consistent, powerful, versatile and efficient. Hence the lack of use of any other scripting languages.


Trojan.

 

[feherke]

Hi

Right, [tt]Perl[/tt] is much powerfull. I love it too.

perl -pe 's/\[.*\]\n//; s/\b$/ /' somefile

Feherke.

http://rootshell.be/~feherke/

 

[duncdude]

Hi feherke - i just looked at that aurelio link you posted... and now i am crying too

That is just not normal - he hasn't put a picture of himself on the web page... because he must be green with big ears!


Kind Regards
Duncan

 

[TrojanWarBlade]

Come on Duncdude,
Don't pretend that you couldn't have writtent that Sokoban in your sleep! (or in your dreams maybe? ;-) )


Trojan.

 

[duncdude]

more akin to nightmares than dreams! :-(


Kind Regards
Duncan

 

[futurelet]

Here's a Gawk or Mawk version:

BEGIN {RS="(\\[[^\n]*\\]\n)+"; ORS=" "}
8

This part is executed before any lines are read:

BEGIN {RS="(\\[[^\n]*\\]\n)+"; ORS=" "}

allowing the main loop to be quite small:

8

[tt]perl -pe 's/\[.*\]\n//; s/\b$/ /' somefile[/tt]

Wouldn't this add a space to
[tt]Ignore this line here[/tt]

 

[feherke]

Hi

Yes, you are right, futurelet. As usual. Thanks.

perl -pe 'if (s/\[.*\]\n//) { s/\b$/ / }' somefile

Feherke.

http://rootshell.be/~feherke/