user/list_large_dirs.sh

   1 #!/usr/bin/env bash
   2 # Desc: Lists directories immediately containing large amounts of data
   3 # Usage: script.sh [int megabytes] [dir]
   4 # Example: A directory named `vacation_photos` located somewhere in $HOME/
   5 #   containing more than 100MB of files (without checking subdirectories of
   6 #   `vacation_photos`) could be found by running:
   7 #   $ script.sh 100 "$HOME"
   8 #   /home/johndoe/Pictures/unsorted/notjunk/vacation_photos
   9 # Version: 0.1.0
  10
  11 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
  12 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  13 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  14 checkInt() {
  15     # Desc: Checks if arg is integer
  16     # Usage: checkInt arg
  17     # Input: arg: integer
  18     # Output: - return code 0 (if arg is integer)
  19     #         - return code 1 (if arg is not integer)
  20     # Example: if ! checkInt $arg; then echo "not int"; fi;
  21     # Version: 0.0.1
  22     local returnState
  23
  24     #===Process Arg===
  25     if [[ $# -ne 1 ]]; then
  26         die "ERROR:Invalid number of arguments:$#";
  27     fi;
  28
  29     RETEST1='^[0-9]+$'; # Regular Expression to test
  30     if [[ ! $1 =~ $RETEST1 ]] ; then
  31         returnState="false";
  32     else
  33         returnState="true";
  34     fi;
  35
  36     #===Determine function return code===
  37     if [ "$returnState" = "true" ]; then
  38         return 0;
  39     else
  40         return 1;
  41     fi;
  42 } # Checks if arg is integer
  43 check_depends() {
  44     local flag_return=0;
  45
  46     if ! command -v du 1>/dev/random 2>&1; then flag_return=1; fi;
  47
  48     return "$flag_return";
  49 }; # returns 1 if missing dependencies
  50 size_dir_contents() {
  51     # usage: size_dir_contents [path]
  52     # depends: find, du
  53     local bytes re;
  54
  55     bytes=0;
  56     re="^[0-9]+$";
  57
  58     #yell "DEBUG:Checking dir:$1";
  59     while read -r addend; do
  60         #yell "DEBUG:addend:$addend";
  61         if [[ $addend =~ $re ]]; then
  62             bytes="$((bytes + addend))";
  63         fi;
  64     done < <( find "$1" -maxdepth 1 -type f -exec du -b --summarize '{}' \; | cut -f1 );
  65
  66     echo "$bytes" && return 0;
  67 }; # return size of dir immediate contents
  68 main() {
  69     #size_large="100000000"; # 100 MB
  70
  71     #yell "DEBUG:arg1:$1"; # debug
  72     #yell "DEBUG:arg2:$2"; # debug
  73
  74     # check dependencies
  75     if ! check_depends; then die "FATAL:Missing dependencies."; fi;
  76
  77     # check args
  78     if ! checkInt "$1"; then
  79         die "FATAL:Not an int:$1";
  80     else
  81         size_l_mb="$1"
  82         size_l_b="$(( size_l_mb * 1000000 ))";
  83     fi;
  84     if [[ ! -d "$2" ]]; then
  85         die "FATAL:Not a dir:$2";
  86     else
  87         dir_in="$2";
  88     fi;
  89
  90     # check each dir size contents
  91     while IFS= read -r -d $'\0' pathdir; do
  92         #yell "DEBUG:Checking:pathdir:$pathdir"; # debug
  93         if [[ ! -d "$pathdir" ]]; then continue; fi;
  94         dir_size="$(size_dir_contents "$pathdir";)"
  95         #yell "DEBUG:dir_size:$dir_size"; # debug
  96         if [[ $dir_size -gt "$size_l_b" ]]; then
  97             printf "%s\n" "$pathdir"; # output
  98         fi;
  99         #sleep 1; # debug
 100     done < <(find "$dir_in" -type d -print0);
 101
 102     return 0;
 103 };
 104
 105 main "$@";