Processing directory structures

[Steve Givens]

I need some help processing directory structures. I have an environment variable which contains the following (example):

admin/sql
admin/bin
admin/SID
admin/SID/adump
admin/SID/bdump
admin/log
archive
archive/SID
archive/SID/dmp

I want to process (actually build a "cp -R" command) for those entries that do not contain SID or those that contain SID but only as the last word. I don't want to process the entries that contain SID but have something after it (e.g. - SID/dmp).

My current code is attached. I'm not sure how to check for the last condition of something after SID.

I would imagine there is also probably a better way of handling the from and to directories so that I don't have to use 2 variables.

Thanks in advance for your help.

Steve

[Steven E. Protter]

Shalom Steve,

Nice name.

Your code crashes my browser so I can't look.

basename

That will isolate the last variable and can be used to extract it into another variabel.

SEP

[Steve Givens]

Not sure why the attachment would crash your browser, but here is the code in question:

#!/bin/ksh
ORACLE_SID=${ORACLE_SID:-\wiltst1} #Default ORACLE_SID
SCRIPT_DIR=${SCRIPT_DIR:-\/oracle} #Default directory for scripts

FromDirNames=`find ${SCRIPT_DIR}/OB -type d | tail +2 | sed 's,'${SCRIPT_DIR}'/OB/,,'`
ToDirNames=`echo ${FromDirNames} | sed 's,SID,'${ORACLE_SID}',g'`

echo "\n\nFromDirNames=${FromDirNames}\n"
echo "\n\nToDirNames=${ToDirNames}\n"

for subdir in ${ToDirNames}
do
if [ -d "${ORACLE_BASE}/$subdir" ] ; then
echo "${ORACLE_BASE}/$subdir already exists"
else
echo "Creating ${ORACLE_BASE}/$subdir"
Cmd="cp -R ${SCRIPT_DIR}/OB/$subdir ${ORACLE_BASE}/$subdir"
FinalCmd="`echo ${Cmd} | sed 's/'${ORACLE_SID}'/SID/1'`"
echo "\t Cmd=${Cmd}"
echo "\t FinalCmd=${FinalCmd}"
fi
done # for subdir loop

[Mel Burslan]

did you try using grep to weed out the entries you do not want like


find ${SCRIPT_DIR}/OB -type d | grep -v SID$

this should eliminate tje lines ending with "SID" from the find output.

[James R. Ferguson]

Hi Steve:

Given that you have a variable loaded with (space-delimited) directories, like:

# echo ${VAR}
admin/sql admin/bin admin/SID admin/SID/adump admin/SID/bdump admin/log archive
archive/SID archive/SID/dmp

# echo ${VAR}|perl -nle '@a=split / /;for (@a) {print if m%/SID\z%;print unless m%SID%}'

...will yield:

admin/sql
admin/bin
admin/SID
admin/log
archive
archive/SID

...You could capture this output too:

# NEWVAR=`echo ${VAR}|perl -nle '@a=split / /;for (@a) {print if m%/SID\z%;print unless m%SID%}'`

Regards!

...JRF...

[Sandman!]

imho...modify the find command instead of parsing the list of directories i.e.

change...

FromDirNames=`find ${SCRIPT_DIR}/OB -type d | tail +2 | sed 's,'${SCRIPT_DIR}'/OB/,,'`

to...

FromDirNames=`find ${SCRIPT_DIR}/OB ! -path "${SCRIPT_DIR}/OB/*/SID/*" -type d | tail +2 | sed 's,'${SCRIPT_DIR}'/OB/,,'`

[Steve Givens]

I knew I'd have a Perl fan respond to this, so thanks.

What I'm working on is a deployment script that needs to run on any environment. Right now I just need to worry about Sun, AIX and HP-UX. Any concerns with Perl compatibility?

[Peter Nikitka]

Hi,

ANY environment - you are really aiming high :-)

But the UNIXs you mentioned
- as long as only such one liners are in play, they can be adopted to ANY perl-version
- as long as these OSs are in play, recent perl distributions of the same perl version are found at

http://activestate.com

mfG Peter

[Steve Givens]

Mel, actually I'm trying to do the opposite - keep lines with no SID in it and lines that end in SID.

Sandman, your syntax is not working for me:
"-path is not a valid option". Is that not supported on AIX?

[James R. Ferguson]

Hi Steve:

No, the '-path' switch is not valid with AIX's 'find'. If you want easy portability, you want Perl.

Using Perl, you could leverage the 'File::Find' core module if you want.

Regards!

...JRF...