Fuzzy directory match using simhash

simdir

#!/bin/bash
shingle_size=4
feature_count=1024

if [[ ! -d $1 || ! -d $2 ]]; then
  echo "Usage: <dir1> <dir2>"
  exit 1
fi;

# make sure we don't simhash simhash files
rm $1/*.sim
rm $2/*.sim

simhash -f $feature_count -s $shingle_size -w $1/*
simhash -f $feature_count -s $shingle_size -w $2/*

# TODO: If a certain match threshold is met, then exlcude the target
# file from all other match attempts. This would only work properly
# assuming a one-to-one mapping, so would want to make  it an option.

for A in "$1"/*.sim; do
  match=""
  maxresult=0
  for B in "$2"/*.sim; do
    if [[ $A != $B ]]; then
      result=$(simhash -c $A $B)
      if (( result > maxresult )); then
        maxresult=$result
        match=$B
      fi
    fi
  done
  echo "$maxresult ${A#.sim} --> ${match#.sim}"
done

# clean up after ourself
rm $1/*.sim
rm $2/*.sim