how to paser this string

[Gemini_2]

Hi..guru..

I have not script for a while....and need your help again..

say if I have a line like this..

list="Only in /usr/local/bin Only in /usr/lib/bin Only in /home/smith"

I parse that one string into 3 strings such as

Only in /usr/local/bin
Only in /usr/lib/bin
Only in /home/smith

I used to know how to do this...

please help out!

[James R. Ferguson · Accepted solution]

Hi Gemini:

# echo ${list} | perl -nle 'print $1 while /(Only in\s+\S+)/g'

Regards!

...JRF...

[James R. Ferguson]

Hi (again):

...and if you want a (longer) 'awk' solution:

# echo ${list} | awk '{while (match($0,/Only in [A-Za-z\/]+/)) {S=substr($0,RSTART,RLENGTH);print S;$0=substr($0,RLENGTH)}}'

Regards!

...JRF...

[Hein van den Heuvel]

We can help you, but it would be best if on top of the example (which is a great start!) you could help describe how the string is build.

Speficifcally... Will a new chunk always start with space+"Only in", or just "Only", or...

$ echo $list | perl -pe 's% (Only in)%\n$1%g'
Only in /usr/local/bin
Only in /usr/lib/bin
Only in /home/smith

If it always starts with whatever the first chunk starts with, and there is a space, then you could use:

$ echo $list | perl -pe '$x=substr($_,0,5); s/\s$x/\n$x/g'
Only in /usr/local/bin
Only in /usr/lib/bin
Only in /home/smith
$


Hein.

[Gemini_2]

After I read everyone's reply, I dont feel as bad. the Awk command is soo long, but very educational. thank you.

I would like to try the perl command, but, I am still not getting this one quite right, please help me.

the input is

Only in /usr/local/bin: a.config Only in /usr/lib/bin: b.conf Only in /home/smith: c.conf


Notice there is no "space" before "Only in", but now, I add the file name...how do I include the file name....

[James R. Ferguson]

Hi (again) Gemini:

OK, given:

# list="Only in /usr/local/bin: a.config Only in /usr/lib/bin: b.conf Only in /home/smit
h: c.conf"

...then:

# echo ${list} | perl -nle 'print $1 while /(Only in\s+\S+\s\S+)/g'

...yields:

Only in /usr/local/bin: a.config
Only in /usr/lib/bin: b.conf
Only in /home/smith: c.conf

Regards!

...JRF...

[James R. Ferguson]

Hi:

...and for similar changes to the 'awk':

# echo ${list} | awk '{while (match($0,/Only in ([A-Za-z\/\:]+ [A-Za-z\.]+ *)/)) {S=substr($0,RSTART,RLENGTH);print S;$0=substr($0,RLENGTH)}}'

Note, of course, that with the 'awk' snippet, filenames containing numbers or special characters would fail the match. I'll leave that as an exercise for you!

The Perl solutions using '\s' for a whitespace and '\S' for a nonwhitespace is easier (as Perl is!). For that matter, Hein's solutions are quite elegant too, and required no changes.

Regards!

...JRF...

[Sandman!]

Yet another (shorter) awk solution :)

# echo $list | awk -F"Only" '{for(i=1;i<=NF;++i) if($i) print FS$i}'

~cheers

[James R. Ferguson]

Hi Sandman:

That's a VERY elegant 'awk' solution! Shorter AND sensitive only to matching the keyword "Only". I always learn better ways with 'awk' and 'sed' reading your contributions.

/* NO POINTS for this post, please */

Regards!

...JRF...

[Sandman!]

That coming from the master (JRF) himself is quite the compliment.

:D

[Gemini_2]

Actually, I parse the string because I want to do the following.

I want to diff two directories, and only display the "Only In" result.

Is there an option in the diff that can do that? so, I dont even have to parse the result.

Mimosa

[James R. Ferguson]

Hi (again) Gemini:

> I want to diff two directories, and only display the "Only In" result

See the manpages for 'dircmp' :

http://docs.hp.com/en/B2355-60105/dircmp.1.html

Regards!

...JRF...

[Sandman!]

Not sure I understand...could you post the diff(1) command you are using?

[A. Clay Stephenson]

It would have really helped had you asked the real question rather than leading everyone of a Wild Goose chase. I think Hein alluded to this earlier. Anyway, you can leverage the comm command suppressing the entries found in both directories.

Something like this:
---------------------------------------
#!/usr/bin/sh

typeset TDIR=${TMPDIR:-/var/tmp}
typeset PROG=${0##*/}
typeset T1=${TDIR}/X${$}_1.tmp
typeset T2=${TDIR}/X${$}_2.tmp

trap 'eval rm -f ${T1} ${T2}' 0 1 2 15

typeset -i STAT=0

if [[ ${#} -ge 2 ]]
then
typeset D1=${1}
typeset D2=${2}
shift 2
if [[ -d "${D1}" && -d "${D2}" ]]
then
ls D1 | sort > ${T1}
ls D2 | sort > ${T2}
comm -3 ${T1} ${T2}
STAT=${?}
else
echo "${PROG}: ${D1} and/or ${D2} not found and/or not directories." >&2
STAT=254
fi
else
echo "Usage: ${PROG} dir1 dir2" >&2
STAT=255
fi
exit ${STAT}
-----------------------------------------

If I didn't make no typing booboo's; that should work. Use it like "dirdiff.sh dir1 dir2". Files that are unique to dir1 appear in column1 of stdout and files unique to dir2 appear in column2.

Man comm for details. You could easily separate this into 3 output files but that is left for you.

[A. Clay Stephenson]

Ooops, I'm stupid I did have a minor error, D1 rather than ${D1}.
----------------------------------------
#!/usr/bin/sh

typeset TDIR=${TMPDIR:-/var/tmp}
typeset PROG=${0##*/}
typeset T1=${TDIR}/X${$}_1.tmp
typeset T2=${TDIR}/X${$}_2.tmp

trap 'eval rm -f ${T1} ${T2}' 0 1 2 15

typeset -i STAT=0

if [[ ${#} -ge 2 ]]
then
typeset D1=${1}
typeset D2=${2}
shift 2
if [[ -d "${D1}" && -d "${D2}" ]]
then
ls ${D1} | sort > ${T1}
ls ${D2} | sort > ${T2}
comm -3 ${T1} ${T2}
STAT=${?}
else
echo "${PROG}: ${D1} and/or ${D2} not found and/or not directories." >&2
STAT=254
fi
else
echo "Usage: ${PROG} dir1 dir2" >&2
STAT=255
fi
exit ${STAT}


-----------------------------------------

Dircmp may be too intensive for you because it not only does file names but also compares the files themselves.

[Gemini_2]

No....I dont want to waste everyone's time.

I want to learn different ways to parse the strings anyway. I read every post and try to understand it.

you guys are being extremely helpful.


really appreciate.

[Sandman!]

Imho if the diff(1) you are using is constructed as shown...

# VAR=$(diff /dir1/dir2 /dir1/dir3)

then echo'ing the doubly-quoted VAR will do the trick w/o any scripting...

# echo "$VAR"

~hope it helps

[Arturo Galbiati]

Hi,
why not the old good Unix command diff?

/> ls -1 t1
a
b

/>ls -1 t2
b
c

/> diff t1 t2
Only in t1: a
Only in t2: c

It shows only the file that are not in both dirs.


HTH,
Art