NNFL Submission Sanitiser

nnfl_submission_sanitiser.sh

#!/bin/bash

# Unzip everything 
pushd submissions
mkdir ../submissions_cleaned
find -name '*.zip' -exec sh -c 'unzip -jod ../submissions_cleaned/"${1%.*}" "$1"' _ {} \;
popd

# Remove spaces from all file name
find $1 -type f -print0 | \
  while read -d $'\0' f; do mv -v "$f" "${f// /_}"; done

pushd submissions_cleaned
for sol_dir in `find . -type d`
do
    # Find the best match and replace
    missing_files=`diff -rq $sol_dir ../source | grep "Only in ../source".*.ipynb | grep -v .ipynb_checkpoints | sed 's/Only in \.\.\/source://'`
    unkown_files=`diff -rq $sol_dir ../source | grep "Only in $sol_dir".*.ipynb | grep -v .ipynb_checkpoints | sed "s|Only in \$sol_dir:||"`

    for missing_file in $missing_files
    do
        best_match="-1"
        best_match_file="none"
        for unkown_file in $unkown_files
        do 
            common_lines=`comm --nocheck-order -12 ../source/$missing_file $sol_dir/$unkown_file | wc -l`
            if test $common_lines -gt $best_match
            then
                best_match=$common_lines
                best_match_file=$unkown_file
            fi
        done
        if test $best_match -gt "0"
        then
            echo "Renaming $best_match_file to $missing_file with $best_match common lines."
            mv $sol_dir/$unkown_file $sol_dir/$missing_file
        fi
    done

    # Remove garbage
    unkown_files=`diff -rq $sol_dir ../source | grep "Only in $sol_dir" | sed "s|Only in \$sol_dir:||"`
    for unkown_file in $unkown_files
    do
    rm -f $sol_dir/$unkown_file
    done

    # Move the files in nbgraders format 
    mv $sol_dir ps1
    mkdir -p $sol_dir/ps1 
    mv ps1 $sol_dir
done
popd