Grep Problem 1

[kvh71]

I am trying to check a line of data using grep. The data I have in the file should be in the format

0123456/1,xxx,xxx etc

or

012345678/1,xxx,xxx etc

The first field must either be 7digits/1digit or 9digits/1digit, but cannot be 8digits/1digit.

So far i have '[0-9]\{7\}/[0-9]' but can someone give me a clue how to check for both formats in one line. I tried '[0-9]\{7|9\}/[0-9]' but it didn't seem to work.

Any ideas ??

 

[PHV]

Try something like this:

grep -E '([0-9]\{7\}|[0-9]\{9\})/[0-9]'

Hope This Help
PH.

 

[kvh71]

I tried this but I keep getting

grep: illegal option -- E

I'm on a server using Oracle 9i, would this make any difference to the options available to me?

 

[dchoulette]

Try without the -E option: it could be implicit with your version of grep.
If this does not work, you could also try 'egrep' instead of 'grep -E'.

--------------------
"When all else has failed, read the manuals."

Denis

 

[kvh71]

It's still not working correctly.
The code and files I am using are

egrep -v '^([0-9]\{7\}|[0-9]\{9\})/[0-9],[0-9A-Z]\{1,12\},[^ ]\{1,6\}$' test_saw.txt >> errors_saw.txt

test_saw.txt :-

0123456/1,AB,21
0321654/2,BC,22
02345678/3,CD,23
0987654/4,DE, 24
0654321/5,Ef,25

errors_saw.txt

0123456/1,AB,21
0321654/2,BC,22
02345678/3,CD,23
0987654/4,DE, 24
0654321/5,Ef,25

 

[dchoulette]

It works for me if I suppress the \ before { and }:

$

egrep -v '^([0-9]{7}|[0-9]{9})/[0-9],[0-9A-Z]{1,12},[^ ]{1,6}$' test_saw.txt

02345678/3,CD,23
0987654/4,DE, 24
0654321/5,Ef,25

--------------------
"When all else has failed, read the manuals."

Denis

 

[kvh71]

Thanks for you patience, but if I leave out the \ it produces no results at all. Could this be an Oracle issue?

 

[dchoulette]

I don't think so. It must be a grep issue: you apply a grep command to a simple text file.
It seems your grep does not behave like mine
(

$

grep --version

grep (GNU grep) 2.4.2

).

We could try another tool. Do you have perl ?
If so try:

$

perl -n -e 'print unless m!^([0-9]{7}|[0-9]{9})/[0-9],[0-9A-Z]{1,12},[^ ]{1,6}$!' tmp/test_saw.txt

02345678/3,CD,23
0987654/4,DE, 24
0654321/5,Ef,25

And just to be sure, the three lines result I get is what you are expecting, is'nt it ?

--------------------
"When all else has failed, read the manuals."

Denis

 

[kvh71]

Tried the Perl and it worked.

Thank you for your time

 

[Annihilannic]

How about egrep '^.......(..)?/'. It's not as strict about numerics being in there, but if you wanted to be you could replace each . with [0-9].

Annihilannic.

 

[kvh71]

This also works, cheers.

It's always good to have options.

 

[ernieah]

"perl" is MUCH better at "pattern matching" than "grep", or even "egrep". And, often perl is beter because it can deal with "other" requirements of the problem-to-be solved, also.

End-of-memo: Best to you..
from ernieah.

 

[vgersh99]

your ratio of the problem complexity to the 'foot print' of the tools to solve it may vary as well.

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+