help with recursive searching+editing

[alexzive]

Many Thanks FM!

I am almost done with the following code (thanks also to another no-awk guy here

#!/bin/bash

c=/path/
cd $c

#crop bad elements between "small area" and "P R O B L E M"
awk 'BEGIN{RS="--"}/'"P R O B L E M   S I Z E"'/{for(i=1;i<=NF;i++)if($i == "NO"){n=i-1;print $n}}' DAT.dat > BAD.dat


while read line
do
  P="$line"
  echo $P
  (echo "/Element, type"; echo ":/[0-9]"; echo "/$P/d";echo "wq!")|ex -s INPUT.inp

done < BAD.dat

There is only one problem left, when the BAD elements appear as second third or fourth item in the line--> in this case the line should not be deleted:

*Element, type=DC2D3
432,477,478,479
433,439,481,482 --> line NOT to be deleted
434,483,484,440 --> line NOT to be deleted
435,486,485,484
436,487,488,489
437,490,491,492
438,493,488,487
439,494,488,493 --> line to be deleted
440,495,496,497 --> line to be deleted
441,491,498,496
442,496,498,497
443,493,499,494
444,500,487,501
445,500,493,487
446,493,500,499
447,499,500,498
448,502,503,504
449,505,506,507
450,508,507,509
451,506,509,507
452,510,511,512
453,510,506,505
454,510,513,506
455,514,505,515
*Elset, elset=C

Thanks!

[FeatherMonkey]

As I'm enjoying the learning curve ...

You really are missing the beauty of awk (Bloody terrible syntax though)
There is no need for a while loop

awk 'BEGIN{FS=","}NR==FNR{a[$1];next} $1 in a{print $0}' check.txt dat1.txt

OK the above still doesn't do what you want but does show that there is no need to loop. The problem with the above is it will still bring back the early and later ones.

Now I just cant get my head around at combining the expressions hopefully one of the gurus will fix it. Even though I did manage to achieve what you first wanted without the 0 padding though
**(note)

for i in $(ls -l | egrep -v "^d|^l" | awk '{print $9}');
do awk 'BEGIN{str=01;RS="--"}/'"P R O B L E M   S I Z E"'/{{print "*Element"}
for(i=1;i<=NF;i++)
if($i == "NO"){n=i-1;print $n}
else if ($i=="YES"){n=i-1;print $n}{print "*Elset"}}' $i;
done

This is actually missing the count and the #, to get that.. the var is there and just needs printing it also doesn't presume just NO but also takes into account YES. If you change print $n to $n'#'str;str++ you'll have the format you wanted(But then it won't match for the delete), if you change "*Element" to FILENAME(Or even prior should work inside{}) it'll append the filename to the top of the block but... this uses *EL as the record sets.

As mentioned I've had to crudely hack it with pipes hopefully someone will come along and tidy it up and I'll learn something new.

awk 'BEGIN{FS=","}NR==FNR{a[$1];next} $1 in a{print $0}' check.txt dat1.txt | awk 'BEGIN{RS="\*El";FS="\n"}/'"type\="'/{print $0}' | awk '/^([[:digit:]])/{print $0}'

Now personally I would then pipe this to sed -i as this'll update them but carefully test and test again(Did I say test) using test data there will be no undo(OK maybe you'll have a backup looking at the man).

**
The for loop won't actually do as intended as you won't have paths if you use the recursive flag, it looks like this would be one for find. So the for loop is rather superfluous and could be written as for i in *.*

[FeatherMonkey]

edit
If you change the print to..

{print "*Element" "\n" FILENAME}

You'll have the filename in each block as to how you'll use it I leave that one upto you.

[alexzive]

Hello,
thanks for suggestion but I still prefer the echo "/string" style with loop. I solved my problem above just adding ^ before $P.
This code does exactly what I wanted but for just a single file:

#!/bin/bash

c=/path/
cd $c

#crop bad elements between "small area" and "P R O B L E M"
awk 'BEGIN{RS="--"}/'"P R O B L E M   S I Z E"'/{for(i=1;i<=NF;i++)if($i == "NO"){n=i-1;print $n}}' DAT.dat > BAD.dat


while read line
do
  P="$line"
  echo $P
  (echo "/Element, type"; echo ":/[0-9]"; echo "/^$P/d";echo "wq!")|ex -s INPUT.inp

done < BAD.dat

making this code recursive on several files is the next challenge now (part II)
Thanks!

[dmera]

Sorry that I couldn't help earlier but sometimes I have to work at work.
Anyway is not what I would have like the script to be like (I was thinking of a awk complete solution, but time was not on my side.
Just to help get over with the last step here is the script you have and I added just a loop to do multiple files(assuming they are in the same directory.
#!/bin/bash

#c=/path/
#cd $c

#crop bad elements between "small area" and "P R O B L E M"
awk 'BEGIN{RS="--"}/'"P R O B L E M S I Z E"'/{for(i=1;i<=NF;i++)if($i == "NO"){n=i-1;print $n}}' A.dat > BAD.dat
all_files="/home/dan/progr_task/*.inp"
for the_file in $all_files; do
echo "$the_file is my file"
while read line
do
P="$line"
# echo $P
(echo "/Element, type"; echo ":/[0-9]"; echo "/^$P/d";echo "wq!")|ex -s $the_file

done < BAD.dat
done