get substring of filename

[mshakeelazeem]

i have filename like the following

1_40_37573.dbf
1_41_37573.dbf
1_42_37573.dbf

the output of "echo $filename | cut -c3-4" is
40
41
42

i also have filenames like
1_101_37573.dbf
1_102_37573.dbf
1_103_37573.dbf

so if i use the above command i'll get the output
10
10
10
actually i want to extract the substring between 2 underscores
e.g. in case of 1_40_37573.dbf i need 40 and in case of 1_101_37373.dbf i need 101 and so on

can anybody suggest to accomplish the task?

regards,
Muhammad Shakeel Azeem

[tpl]

welcome to the forum
here's one way

awk 'BEGIN {FS = "_"}{print $2}' <filename

[scm]

You can do it in two steps with the shell's built-in variable editing facility, which is quicker than running an external filter
VAR=${filename#*_}
VAR=${VAR%_*}
echo $VAR

[Sathors]

If you want to use nearly the same expression, you can accomplish it simply with :

echo $filename | cut -d_ -f2

Have a look at the man of cut, it is a very powerful command.
Here the -d option tells which caracter should be considered as a delimiter, here it is the underscore caracter, and then with the -f option you tell it which field (which substring) to cut, here the second one.
So if you have 1_40_.... or 16_468_.... or sausage_ham_... you'll always get the second field (40, 468, and ham there).

[mshakeelazeem]

Here is another requirement
Actually i am copying from remote location "/u02/flash_rec_area/" which contains files like

1_41_732936151.dbf
1_42_732936151.dbf
1_43_732936151.dbf
1_44_732936151.dbf
1_45_732936151.dbf
1_46_732936151.dbf
1_47_732936151.dbf
1_48_732936151.dbf
1_49_732936151.dbf
1_50_732936151.dbf
1_51_732936151.dbf
1_52_732936151.dbf

2_13_732936151.dbf
2_14_732936151.dbf
2_15_732936151.dbf
2_16_732936151.dbf
2_17_732936151.dbf
2_18_732936151.dbf

now i want to just ignore the files start with 2_ ,to accomplish i am using the code mentioned below but its also ignoring the files like 1_42_732936151.dbf and 1_52_732936151.dbf which is against my requirement ,please suggest

for filename in `ssh rac1 'ls '/u02/flash_rec_area/'' | sort | grep -v ' '2_' '`
do
echo $filename
done

thanks and regards,

[Sathors]

You can do

ls --format=single-column | egrep -v "^2_"

According to the man the option -C is the equivalent to --format=single-column but it doesn't work for me.
And the egrep will remove each line which begins with "2_".

EDIT : i forgot the SSH command, but I think it should work fine with it.

[mshakeelazeem]

its working on local computer but not getting the desired results with ssh

[oracle@standby dbscripts]$ pwd
/u01/app/dbscripts

[oracle@standby dbscripts]$ ls -al
total 48
drwxr-xr-x 2 oracle oinstall 4096 Oct 29 11:55 .
drwxrwxr-x 8 oracle oinstall 4096 Oct 26 15:26 ..
-rwxr--r-- 1 oracle oinstall 1499 Oct 29 10:56 archivemove1.sh
-rwxr--r-- 1 oracle oinstall 1528 Oct 29 10:57 archivemove2.sh
-rwxr--r-- 1 oracle oinstall 1410 Oct 29 11:11 archivemove.sh
-rw-r--r-- 1 oracle oinstall 39 Oct 26 12:52 cmdfile.rcv
-rwxr--r-- 1 oracle oinstall 820 Oct 26 19:06 generic.sh
-rw-r--r-- 1 oracle oinstall 258 Oct 26 12:50 getrecid1.sql
-rw-r--r-- 1 oracle oinstall 258 Oct 26 12:50 getrecid2.sql
-rwxr--r-- 1 oracle oinstall 72 Oct 26 16:30 recover1.sh
-rw-r--r-- 1 oracle oinstall 790 Sep 21 16:18 recover.sh
-rwxr--r-- 1 oracle oinstall 268 Oct 29 14:59 test.sh

[oracle@standby dbscripts]$ cat test.sh
#!/bin/sh
ls --format=single-column | egrep -v ''^2_''
for filename in `ssh rac1 'ls --format=single-column '/uninstall/primdir'' | sort | egrep -v ''^2_''`
do
echo $filename
done

look at the output of test.sh

[oracle@standby dbscripts]$ ./test.sh
archivemove1.sh
archivemove2.sh
archivemove.sh
cmdfile.rcv
generic.sh
getrecid1.sql
getrecid2.sql
recover1.sh
recover.sh
test.sh
1_10_234.arc
1_1_234.arc
1_41_732936151.dbf
1_42_732936151.dbf
1_43_732936151.dbf
1_44_732936151.dbf
1_45_732936151.dbf
1_46_732936151.dbf
1_47_732936151.dbf
1_48_732936151.dbf
1_49_732936151.dbf
1_50_732936151.dbf
1_51_732936151.dbf
1_52_732936151.dbf

[mshakeelazeem]

my mistake

its working

[scm]

You can do

ls --format=single-column | egrep -v "^2_"

According to the man the option -C is the equivalent to --format=single-column but it doesn't work for me.

The -1 (that's the digit one) option does single column output in ls, but of course, it does single column output anyway when its stdout is not connected to a terminal (try "ls | cat" for evidence). The -C flag is used to do multi-column, even into a pipe.