user/randtxt.sh

   1 #!/bin/bash
   2 # Desc: Output random text selection from a text file within a directory
   3 # Usage: randtxt.sh DIR
   4 # Version 0.0.6
   5 # Depends: Bash 5.1.16, GNU findutils 4.8.0, GNU Coreutils 8.32
   6 # Example: randtxt.sh ~/Calibre\ Library/
   7
   8 SAMPLE=10000;   # ceiling for number of text files to consider
   9 CONTEXT="5000"; # total bytes +1 of text to display
  10 if [[ $(( CONTEXT/2 + CONTEXT/2 )) -lt $CONTEXT ]]; then ((CONTEXT++)); fi;
  11
  12
  13 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
  14 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  15 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  16 showUsage() {
  17     cat <<'EOF'
  18     USAGE:
  19         randtxt.sh DIR
  20
  21     EXAMPLE:
  22       randtxt.sh ~/
  23 EOF
  24 } # Display information on how to use this script.
  25 checkInput() {
  26     # input:  arg1
  27     if [[ $# -gt 1 ]]; then die "FATAL:Too many arguments"; fi;
  28     if [[ $# -lt 1 ]]; then die "FATAL:Not enough arguments"; fi;
  29     if [[ ! -d "$1" ]]; then die "FATAL:Not a dir:$1"; else return 0; fi;
  30 };
  31 getFileSizeList() {
  32     # Desc: Create list of file sizes and paths
  33     # input:  arg1   directory path
  34     # output: array  fileSizeList
  35     #         var    totalSize
  36     #         var    randPoint
  37
  38     declare -g -a fileSizeList;
  39     mapfile -d '' -t fileSizeList < <(
  40         find -- "$1" -type f -name "*.txt" -printf '%s\t%p\0' | shuf -z -n"$SAMPLE";
  41     ); # Build array by feeding null-delimited lines from `find` to `mapfile`
  42
  43     declare -g totalSize=0;
  44     local i size;
  45     for i in "${!fileSizeList[@]}"; do
  46         size="${fileSizeList[i]%%$'\t'*}";
  47         totalSize=$((totalSize + size));
  48     done;
  49     if [[ $totalSize -le 0 ]]; then die "FATAL:Total size is zero."; fi;
  50
  51     declare -g randPoint
  52     randPoint="$(shuf -n1 -i0-$((totalSize-1)); )";
  53 };
  54 getRandText() {
  55     # Desc: Print text within fileSizeList around randPoint
  56     # input: array  fileSizeList
  57     #        var    totalSize
  58     #        var    randPoint
  59
  60     local sum psum i size fileStart filePoint fileEnd selStart selEnd selCount file;
  61     sum=0; # init byte sum
  62     psum=0;
  63     for i in "${!fileSizeList[@]}"; do
  64         # Move search to next file
  65         size="${fileSizeList[i]%%$'\t'*}";
  66         sum=$(( sum + size ));
  67
  68         # Check if point is within file
  69         if [[ $sum -gt $randPoint ]]; then
  70             # Calculate important positions within file (zero-indexed)
  71             fileStart=0;                              # first byte within file
  72             filePoint="$(( randPoint - psum ))";      # point as byte within file
  73             fileEnd="$(( size - 1 ))";                # last byte within file
  74             selStart="$(( filePoint - CONTEXT/2 ))";  # start of output selection
  75             selEnd="$(( filePoint + CONTEXT/2 ))";    # end of output selection
  76             # Clamp selection start and end indexes
  77             if [[ $selStart -lt $fileStart ]]; then selStart=$fileStart; fi;
  78             if [[ $selEnd -gt $fileEnd ]]; then selEnd=$fileEnd; fi;
  79             selCount="$(( selEnd - selStart + 1 ))";  # number of bytes within selection
  80             # Output context
  81             file="$(cut -f2- <<< "${fileSizeList[i]}"; )";
  82             printf "INFO:Sample of:%s\n" "$file" 1>&2;
  83             tail --bytes=+$((selStart+1)) -- "$file" | head --bytes=$((selCount));
  84             printf "\n";
  85             return 0;
  86         fi;
  87         psum=$sum; # store previous sum
  88     done;
  89 };
  90 export -f checkInput getFileSizeList getRandText;
  91
  92 main() {
  93     # Input:  arg1    directory path
  94     # Output: stdout  text
  95
  96     checkInput "$@";      # check input arguments
  97     getFileSizeList "$1"; # make fileSizeList array
  98     getRandText;          # output text
  99     return 0;
 100 };
 101
 102 main "$@" && exit 0;