grep help

seth1788 · 03-28-2017, 07:54 AM

Hello,

Looking for assistance on a small script. I have a text file with Names numbers and special notes. Not everyone has a special note. I am looking to search the file and output the matching results. Search and output all Johns (special notes if any). Search for all managers(and their name and number). The special note is always led with " ---->".

Format of file:

Code:

John Doe 111-111-1111
Mary Jane 222-222-2222
 ----> Manager
John Smith 333-333-3333
 ----> Director

What I have so far:

Code:

if grep -i -q -A1 $1 /home/username/file.txt; then
   grep -i -A1 $1 /home/username/file.txt;
else
    grep -i $1 /home/username/file.txt;
fi

It seems like what I have is displaying some results, but is far from complete. Any pointers would be greatly appreaciated.

Thanks!

Turbocapitalist · 03-28-2017, 08:20 AM

Welcome.

Unless there is a constraint requiring grep, I'd try awk. Though I'd hope there is a way with sed too.

Code:

awk -v n="John" '$1 ~ n { print; a++; } a && $1 ~ "---->" { print; a=0; next; } a=0;' /home/username/file.txt

The basic awk statements are implied if-then statements unless stated explicitly otherwise. The -v option passes an environment variable into the script as an awk variable.

Code:

man awk

!!! · 03-28-2017, 12:51 PM

MIGHT -z with some tricky \n regex work? See #15 in http://gnu.org/software/grep/manual/grep.html
But basically as above post says, grep wasn't designed for the complexity of --after having conditionality.
I thought of possibly some obscure tricks with tr, joining all the ----> with prior line, so basic grep could operate on ONE line as designed, then maybe |tr to re-split that, re-inserting the newline, but obscure tricks probably are NOT the goal here.

A useful learning exercise for OriginalPoster might be to explain what/why/how their attempt produced not the desired result. Awaiting reply from OP. Welcome to LQ!

syg00 · 03-28-2017, 04:55 PM

A small correction if I might be so bold

Quote:

Originally Posted by Turbocapitalist

Code:

awk -v n="John" '$1 ~ n { print; a++; next} a && $1 ~ "---->" { print; a=0; next; } a=0;' /home/username/file.txt

Doesn't handle the second part of the request, but does indicate what might be done. Bit difficult if no prior awk knowledge.

I think this is beyond the scope of grep - holding the data to check the next line needs some logic ability. Could be done in sed, but would be ugly.

DDukes · 03-29-2017, 12:02 AM

@ OP

Your code of grep -i -A1 $1 /home/username/file.txt can work if the members with no special notes was replaced with a line space.

I did a test with your file entries. I enter your entries like so in a file:

Quote:

John Doe 111-111-1111

Mary Jane 222-222-2222
----> Manager
John Smith 333-333-3333
----> Director

Then ran:

Code:


grep -i -A1 "john" members 
John Doe 111-111-1111


John Smith 333-333-3333
 ----> Director

As you can see, it prints all johns with special notes if applicable .

Another test with Mary

Code:

grep -i -A1 "mary" members 
Mary Jane 222-222-2222
 ----> Manager

The key is to keep things in a uniform pattern. Use a line space for members with no special notes and the grep -i -A1 should do ok.

BTW, awk and grep/egrep can have more flexibility of output results if you create a database of members using columns a special character as a field separator.

Quote:

The pipe symbol is use here as the field separator.
- cheers

Shadow_7 · 03-30-2017, 02:24 PM

If the file is standardized and predictable you could bump through each line and retain information in a bash script. Although awk is probably a more efficient search mechanism, especially on large data sets. If you're just wanting neighboring lines on the quick, your could grep -v and remove john, then diff that with the original file.

$ cat data_original.txt | grep -i -v john > data_new.txt
$ diff -c3 data_new.txt data_original.txt | less