Grouping files according to certain fields in their name

Shell Programming and Scripting
1

I have a list of fils stored insortedLst, and want to select certain fields to group specific files together:

Example of the files would be as below:

n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run3.log

I use the following command to group similar files according to similar fields

        echo $sortedLst | tr ' ' '\n' \
          | awk -F- '{ c=($4$5$6!=p && FNR!=1)?ORS:""; p=$4$5$6 } { printf("%c%s\n",c,$0) }'

I now want the user to be able to define the grouping fields himself rather than hardwiring '$4$5$6' in the awk script.

Using the code above, the output would then be shown like this:

n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run3.log
   
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run3.log
   
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run3.log
   
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run3.log
   
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run3.log
2

If the user selects a grouping field, would a grep be sufficient?

Can you give us some examples of what the user could select?

3

The user would want to list for example the log files present in the directory.

Let us assume that the files are these one. These files would have already been sorted using the numeric values rather than just alphabetical, because if one does not sort numerically certain field, everything will be mixed up.

n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.004-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.006-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.008-16x12drw-run3.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.010-16x12drw-run3.log

As an example, the three files here are grouped together since they represent the running of the same programs with same parameters, for various running instances. These will then be separated from the others by a blank line.

n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run1.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run2.log
n02-z30-dsr65-ndelt0.25-varp0.002-16x12drw-run3.log

For example
$4$5$6 gives ndelt0.25-varp0.002-16x12drw. When this changed I insert a blank line. The user would be able to supply the fields he wants for grouping things together.

As an example, he can pass to the command line argument --group=4/5/6

As you say, the user can use grep, but then he will also need to sort by numerical values and separate files related to the various runs himself. I would like for him to just run a script and does the work for him, like numeric sorting by certain fields first, then by others, then perform some group. The only things he would need to specify would be the sorting order (e.g. --sort=3/1/5) and the grouping field (e.g. --group=4/5/6 as in this example).