#!/bin/dash

# written by andrewt@unsw.edu.au for COMP(2041|9044)
# Improved version of plagiarism_detection.comments.sh

# change all C strings to the letter 's'
# and change all identifiers to the letter 'v'.
# Hence changes in strings & identifiers will be ignored.

# mktemp provide suitable temporary filename, robustly & securely
TMP_FILE1=$(mktemp)
TMP_FILE2=$(mktemp)

# trap allows use to remove temporary files if program interrupted
trap 'rm -f $TMP_FILE1 $TMP_FILE2' EXIT

# s/"["]*"/s/g changes strings to the letter 's'
# It won't match a few C strings which is OK for our purposes

# s/[a-zA-Z_][a-zA-Z0-9_]*/v/g changes variable names to 'v'
# It will also change function names, keywords etc. which is OK for our purposes.

transform() {
    sed '
        s/\/\/.*//
        s/"[^"]"/s/g
        s/[a-zA-Z_][a-zA-Z0-9_]*/v/g
        ' $1
}

for file1 in "$@"
do
    for file2 in "$@"
    do
        test "$file1" = "$file2" &&
            break # avoid comparing pairs of assignments twice

        transform "$file1" >$TMP_FILE1
        transform "$file2" >$TMP_FILE2

        if diff -iBw $TMP_FILE1 $TMP_FILE2 >/dev/null
        then
            echo "$file1 is a copy of $file2"
        fi
    done
done