Mastodon cache contents by file extension

cache_by_ext.sh

#!/bin/sh

# Probably needs GNU AWK. Come to think of it, some OSes
# won't have a `find` with -printf like GNU does.
# Usage:
# ./cache_by_ext.sh /path/to/mastodon/system/cache

cache_dir=${1-/opt/mastodon/web/system/cache}

find "$cache_dir" -type f -printf "%p\n" \
    | awk '{
f_path=$0
ext_pos=match($f_path, /\.[^\.]+$/)
f_ext=substr($f_path, ext_pos+1) # +1 to skip over . itself

# Consider .jpeg to be .jpg.
if (f_ext == "jpeg") { f_ext="jpg" }

ext_count[f_ext]++
file_count++
}
END {
    for (key in ext_count) {
        printf "%-7s %8lu\n",
            key, ext_count[key] \
        | "sort -k 2 -rn"
    }
    close("sort -k 2 -rn")
    print "-----"
    printf "Total: %9lu files\n", file_count
}'