UNIX for Dummies Questions & Answers

I think I saw another post where someone said he an issue with a 'while within a while' but there were no solutions on it.

I have an input file of data columns separated by pipes "|" and I want to set each column to a variable and do something with it. This I can do.

When I get to a certain row of a certain value, I want to take a subset (i.e. grep for a certain key) of all rows from the same input file and do another while statement, but for some reason or other, the second while only runs once and it breaks back out to first while. This is just pseudo code:

while read a b c ; do
a=s/thing
b=s/thing
c=s/thing
if a=somethingspecial then
grep "somethingiwant" myinputfile > anewfile
while read a b c; do
a1=s/thing
b1=s/thing
c1=s/thing
do something as many times as grep found the key
done <anewfile
fi
done < myinputfile

The problem is that the inner while runs once even though I may have say, 20 rows to process. Any thoughts?

Thank you.
Gianni

This is crazy. I use double loops just like that all the time. They work fine. Real code example....

Code:

$ cat doubleloop
#! /usr/bin/ksh

while read count ; do
        echo count = $count >temp
        yes some stuff | sed ${count}q >> temp
        while read line ; do
                echo this came from temp file: $line
        done < temp
done < countfile
exit 0
$
$
$
$
$ cat countfile
4
6
2
7
2
$
$
$
$
$
$ ./doubleloop
this came from temp file: count = 4
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: count = 6
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: count = 2
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: count = 7
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: some stuff
this came from temp file: count = 2
this came from temp file: some stuff
this came from temp file: some stuff
$

It does sound crazy and not logical but that's what I experienced.

If I were to build on your example, what happens if you got to count '2' in your first while and you want to pull all '2's from your input file "countfile" (i.e grep '2' into another temp file) and process that temp file in the inner while?

I noticed in my situation that the inner while will pull only the second, third, fourth, etc. occurrence of '2' from countfile...and when it processes the temp file, it will do it once and go break out. This is on aix with ksh.

My work around was to duplicate the countfile into say countfile2 and grep for all '2' from countfile2 inside the inner while, then it processed all the records as expected and returned to the next record of the countfile.
In this example, '2' is simplistic but in my example, it's a table column name. I'm grepping for all records with this column so I can do some more processing with these columns. In my case, I could duplicate the file and it's small enough, but if it's a large file, it wouldn't be ideal to have two identical files just to handle a while issue...

I admit it's weird and not logical but I know I've encountered it before a few times in the past. My solution was always to switch to a 'for loop' with inner while. However, this time, I need the while loop in a while since I have way more variables to deal with then just a 'for x in... ' loop.

Gianni

see the red fonts. they are the same. try naming them differently

That was a typo on my part in the example. However, the read a1 b1 c1 didn't resolve the issue. I do have it correct in my code (granted I called it s/thing totally different).

For the grep issue also, I'm wondering if the input file, when opened by a "while read", is in memory and when you do a grep of that file that has been read to a certain point, there's still a pointer in memory to that input file so then grep only sees everything after that first occurrence?

The first occurrence is like getting to the first '2' in Perderabo's example (a 'key word' I'm interested in doing more with) and trying to pull all '2's from the input file (countfile) again but grep doesn't see the first '2' anymore in the inner while...it only pulls the 2nd, 3rd, 4th occurrence, etc. when I sent the output to tmpcountfile ( grep $x countfile > tmpcountfile .... while read a1; do s/thing done<tmpcountfile).

It's hard to explain but I tried simple tests and it confirmed my issue with inner while. I opened the inner while tmpcountfile and the first occurrence is not an output from the grep when done within the inner while.
...
Gianni

Please post a real executable shell script that illustrates your problem.