help with sed

metalx1000 · 09-17-2006, 03:36 PM

Here is what I'm trying to do.

I have file with tags simular to the one below

Code:

<tre>hello</tre><lkj>world</lkj><sdf>tux</sdf><tre>cool</tre>
<lkj>time</lkj><sdf>linux</sdf>

I want to write a script that will collect the info between the 'tre' and the 'sdf' tags and put them in a text files like below:

Code:

hello tux
cool linux

I've tryed to use 'sed' but I'm new to this command and can't get it right or 'sed' is not the right command to use.

spirit receiver · 09-17-2006, 04:34 PM

As sed proceeds linewise, you should tell us how many of these tags appear on a single line and specify which ones should appear on each line of the output. If there are no such rules, it's probably better to use Perl and process the entire file all at once.

metalx1000 · 09-17-2006, 04:41 PM

It has no line order. Everything is bunched together. And, It's a few pages long. How would I go about doing it in Perl, if it is possible?

kryptonite0110 · 09-17-2006, 06:56 PM

replace > with >\n, now you have:

<tre>
hello</tre>
<lkj>
world</lkj>
<sdf>
tux</sdf>
<tre>
cool</tre>
<lkj>
time</lkj>
<sdf>
linux</sdf>

then delete ^<*>$ (i don't know if ^ is the beggining or end, so it may be $<*>^) lines
then grep tre, sdf
then s/</.*>//
:edit

it's possible to optimize this in a number of ways, i'm not going to do that for you...

:edit2

i just realized, that for that to work, you'd need to put everyhing on one line (or at least the </tag> on the same line as the contents of said tag...) s/\n// may work...

ghostdog74 · 09-17-2006, 07:05 PM

you could do it with regular expressions in perl, or python
an example only, Python:

Code:

import re
s = "<tre>hello</tre><lkj>world</lkj><sdf>tux</sdf><tre>cool</tre><lkj>time</lkj><sdf>linux</sdf>"
trepat = re.compile(r"<tre>(.*?)</tre>",re.M)
sdfpat = trepat = re.compile(r"<sdf>(.*?)</sdf>",re.M)
tre = trepat.findall(s)
sdf = sdfpat.findall(s)
zip(tre,sdf) ## "transpose" function here.

output:

Code:

>>> zip(tre,sdf)
[('hello', 'tux'), ('cool', 'linux')]

can be converted to perl easiliy..

spirit receiver · 09-17-2006, 07:27 PM

Here's a Perl one-liner. It will begin a new line after each sdf-tag:

Code:

ada@barnabas:~/tmp> echo -e "<tre>hello</tre><lkj>world</lkj><sdf>tux</sdf><tre>cool</tre>\n<lkj>time</lkj><sdf>linux</sdf>" |
> perl -ne 'while ( m/<(tre|sdf)>(.*?)</g) {print "$+ "; print "\n" if ( "$1" eq "sdf" );}'
hello tux
cool linux

metalx1000 · 09-18-2006, 09:18 AM

Quote:

Originally Posted by spirit receiver

Here's a Perl one-liner. It will begin a new line after each sdf-tag:

Code:

ada@barnabas:~/tmp> echo -e "<tre>hello</tre><lkj>world</lkj><sdf>tux</sdf><tre>cool</tre>\n<lkj>time</lkj><sdf>linux</sdf>" |
> perl -ne 'while ( m/<(tre|sdf)>(.*?)</g) {print "$+ "; print "\n" if ( "$1" eq "sdf" );}'
hello tux
cool linux

Thank you everyone

This Perl one-liner did the job great.

Thanks once again.