I'm not quite sure how to do it purely in the terminal without it getting too difficult to read, but I think you could get the result you are looking for, or at least get started, with something along these lines:

Edit: updated with info from the comment. Also swapped secondary and primary since those were backwards.

Edit2: realized that while the secondary name no longer relied on placement, the principle name relied on secondary's placement.

#!/bin/bash



input_directory="/path/to/your/data"

output_directory="/path/to/your/output"



cd "$input_directory"

for file in *; do



    if [ ! -f "$file" ]; then

        continue;

    fi



    # place your secondary names inside "(mpc|flf|dtd)" seperated by '|'

    secondary=$(echo "$file" | grep -o -E "(mpc|flf|dtd)");

    princple=$(echo "$file" | grep -o -E "([0-9]+_S_[0-9]+)");

    

    # skip over and alert that a secondary match was not found for a file

    if [ "$secondary" == "" ]; then

        echo "No secondary match found!! Skipping $file";

        continue;

    fi



    destination="${output_directory}/$secondary/$princple"



    # create the directories if they don't exist

    if [ ! -d "$destination" ]; then

        mkdir -p "$destination";

    fi



    # uncomment to move the files to the new directories if the test output

    # from echo is correct

    #mv "$file" "$destination"



    # test to print result of moving the files

    from=$(readlink -f "$file")

    echo "$from -> $destination/$file"

done

grep -o -E "(mpc|flf|dtd)" searches the filename for one of the secondary name keywords, e.g. (mpc, dtd or flf), and saves that word to the secondary variable.

grep -o -E "([0-9]+_S_[0-9]+) Same idea, by looks for the xxx_S_xxx pattern.

It can be run as: bash script.sh

The input_directory and output_directory variable will need to be filled in with the correct paths. Also, the fields "(mpc|flf|dtd)" in the grep statement can be filled in with other secondaries.