Create ASCII histogram of the file sizes in the folder hierarchy

histogram.sh

#!/bin/zsh
# Create ASCII histogram of the file sizes in the folder hierarchy

target="$1"

max_width_columns=120

if [ -z "$target" ]; then
  echo "Usage: $0 /path/to/directory"
  exit 1
fi

# Check if the directory exists
if [ ! -d "$target" ]; then
  echo "Error: Directory '$target' not found"
  exit 1
fi

# Print the histogram header
echo "File size distribution histogram for '$target'"

sizes=$(find  "$target" -type f -exec stat -f '%z' {} +)
# Alternative to get actual disk size -- find "$target" -type f -exec du {} + |

dump_the_histogram()
{
  counts=$(echo "$sizes" | awk -v bin_size_bytes="$bin_size_bytes" '
      {
        count[int($1/bin_size_bytes)]++
      } END {
        for (size in count)
          print (size + ((bin_size_bytes >1)?1:0)) * bin_size_bytes,count[size]
      }' |
    sort -n)

  max_count=$(echo "$counts" | awk 'BEGIN {max = 0} {if ($2 > max) max = $2} END {print max}')
  scale=$(bc -le "$max_count / $max_width_columns")

  if [ "$max_count" -gt "$max_width_columns" ]; then
    printf "One star represents about %.0f files. Omitting results with counts smaller than that\n" "$scale"
  else
    scale=1
  fi

  echo "$counts" | awk -v scale=$scale '
  {
    size = $1
    count = $2

    if (scale <= count)
    {
      units = "B  ";
      if (size > 16*1024)
      {
        units = "kiB";
        size = size / 1024
        if (size > 16*1024)
        {
          units = "MiB";
          size = size / 1024
        }
      }
      printf("%10d %s | %10d | ", size, units, count );

      for (i = 0; i < int(count/scale); i++)
        printf("*");
      printf("\n");
    }
  }'
}

# https://en.wikipedia.org/wiki/Histogram
data_count=$(echo "$sizes" | wc -l )

bin_size_bytes=$(awk -v count="$data_count" 'BEGIN{print int(2 * count**(1/3)) + 1}')
echo "--- Bin size by Rice: $bin_size_bytes bytes"
dump_the_histogram

bin_size_bytes=$(awk -v count="$data_count" 'BEGIN{ print int(1 + log(count)/log(2))}')
echo "--- Bin size by Sturges: $bin_size_bytes bytes"
dump_the_histogram

bin_size_bytes=1024
echo "--- Bin size fixed: $bin_size_bytes bytes"
dump_the_histogram

bin_size_bytes=4096
echo "--- Bin size fixed: $bin_size_bytes bytes"
dump_the_histogram

bin_size_bytes=16384
echo "--- Bin size fixed: $bin_size_bytes bytes"
dump_the_histogram