shell issue: parameter substitution

[Jdamian]

Hi

I'm using the ${#} to check lexicographycally an IP address:

IP=192.112
echo "${IP%+([0-9].+([0-9])}"
19

Why does the pattern +([0-9]) remove just ONE digit from the first number while those same pattern removes all digits from the second number ?

[TY 007 · Accepted solution]

Hello Oscar,

Tested in a HP-UX 11.11 OS:

# echo $SHELL
/sbin/sh

# IP=192.112

# echo "${IP%+([0-9].+([0-9])}"
192.112

Thanks

[J. Bravo]

In my 11.23 box it's working fine with POSIX shell:

# IP=192.112
# echo "${IP%+([0-9].+([0-9])}"
192.112

Best regards.

[TY 007]

Hello Oscar,

Tested as well in HP-UX 11.00 OS.

# uname -r
B.11.00

# echo $SHELL
/sbin/sh

# IP=192.112

# echo "${IP%+([0-9].+([0-9])}"
192.112

Thanks

[Jdamian]

I'm sorry but my patter posted here has a missing character ')', i.e., it's wrong.

Try this next one:

echo "${IP%+([0-9]).+([0-9])}"

[AwadheshPandey]

Oscar!!

Can you clear what actually you want to print?
Regards,

Awadhesh

[Pete Randall]

# IP=192.112
# echo "${IP%+([0-9]).+([0-9])}"
19
# uname -r
B.11.11


Pete

[TY 007]

Hello Oscar,

# IP=192.112

# AA=`echo $IP | awk -F. '{print $1}'`

# echo "${AA%+([0-9])}"
19

Thanks

[TY 007]

Hello Oscar,

# IP=192.112

# BB=`echo $IP | awk -F. '{print $2}'`

# echo "${BB%+([0-9])}"
11

Thanks

[James R. Ferguson]

Hi Oscar:

# echo "${IP%\.*}"
192

# echo "${IP##[0-9]*\.}"
112

You should escape the dot which otherwise means any character.

Regards!

...JRF...

[AwadheshPandey]

http://wwwhome.cs.utwente.nl/~broekjg/bsc/scripts/phase2.txt

[Sandman!]

# echo "${IP%+([0-9]).+([0-9])}"

In the above expression the shell is doing what it has been asked of; delete the shortest matching pattern as a single percent sign was specified. On the other hand if the expression was:

# echo "${IP%%+([0-9]).+([0-9])}"

...then it deletes the longest matching pattern which is the entire string as a double percent sign was specified. See the sh-posix(1) man page for details.

~cheers

[Dennis Handly]

>JRF: You should escape the dot which otherwise means any character.

Actually no. This is a file matching pattern not a regular expression. In a pattern you have "*" and "?".

[Jdamian]

Some explanations:

I'm trying to check lexicographically an IP address.

I'm very distrustful :-( -- I don't trust external commands as grep or awk, so I prefer to use shell internal tricks


I know the difference of ${%} and ${%%} but in my example pattern

echo "${IP%+([0-9]).+([0-9])}"

I cannot understand why the same sub-pattern

+([0-9])

matches the full last byte (112) BUT only one digit from the first byte (192).

Does the 'shortest length' behaviour of ${%} apply only to the sub-pattern following the character '%' ?

Thanx in advance

[Dennis Handly]

>BUT only one digit from the first byte (192).
>Does the 'shortest length' behaviour of ${%} apply only to the sub-pattern following the character '%'?

As Sandman said, it applies to everything. Since the last pattern must match ALL of 112 to match the ".", it does. But to match 192, it only has to match the shortest.

I would suggest you trust in the grep and regular expressions. grep/awk aren't the darkside ;-)

[Sandman!]

>Does the 'shortest length' behaviour of ${%} apply only to the sub-pattern
>following the character '%' ?

Yes it does...and the difference can be seen in the following constructs...

# IP=192.112
# echo "${IP%+([0-9])}"
192.11
# echo "${IP%%+([0-9])}"
192.

[Jdamian]

thanx