user/randtxt.sh

   1 #!/bin/bash
   2 # Desc: Outputs text at randomish position within dir of text files
   3 # Usage: randtxt.sh DIR
   4 # Version 0.0.4
   5 # Depends: Bash 5.1.16, GNU findutils 4.8.0, GNU Coreutils 8.32
   6
   7 SAMPLE=10000;
   8 CONTEXT="5000"; # total bytes +1 before and after point within file to print
   9 if [[ $(( CONTEXT/2 + CONTEXT/2 )) -lt $CONTEXT ]]; then ((CONTEXT++)); fi;
  10
  11
  12 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
  13 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  14 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  15 showUsage() {
  16     cat <<'EOF'
  17     USAGE:
  18         randtxt.sh DIR
  19
  20     EXAMPLE:
  21       randtxt.sh ~/
  22 EOF
  23 } # Display information on how to use this script.
  24 checkInput() {
  25     # input:  arg1
  26     if [[ $# -gt 1 ]]; then die "FATAL:Too many arguments"; fi;
  27     if [[ $# -lt 1 ]]; then die "FATAL:Not enough arguments"; fi;
  28     if [[ ! -d "$1" ]]; then die "FATAL:Not a dir:$1"; else return 0; fi;
  29     #yell "DEBUG:checkInput() finished."; # debug
  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     #declare -p fileSizeList; # debug
  43
  44     declare -g totalSize=0;
  45     local i size;
  46     for i in "${!fileSizeList[@]}"; do
  47         #declare -p i; # debug
  48         #yell "DEBUG:fileSizeList[i]:${fileSizeList[i]}"; # debug
  49         size="${fileSizeList[i]%%$'\t'*}";
  50         #declare -p size; # debug
  51         totalSize=$((totalSize + size));
  52         #declare -p i size totalSize; # debug
  53         #yell "=============="; # debug
  54     done;
  55     if [[ $totalSize -le 0 ]]; then die "FATAL:Total size is zero."; fi;
  56
  57     declare -g randPoint
  58     randPoint="$(shuf -n1 -i0-$((totalSize-1)); )";
  59     #declare -p fileSizeList totalSize randPoint 1>&2;
  60     #yell "DEBUG:fileSizeList element count:${#fileSizeList[@]}"; # debug
  61     #yell "DEBUG:getFileSizeList() finished."; # debug
  62 };
  63 getRandText() {
  64     # Desc: Print text within fileSizeList around randPoint
  65     # input: array  fileSizeList
  66     #        var    totalSize
  67     #        var    randPoint
  68
  69     local sum psum i size fileStart filePoint fileEnd selStart selEnd selCount file;
  70     sum=0; # init byte sum
  71     psum=0;
  72     for i in "${!fileSizeList[@]}"; do
  73         # Move search to next file
  74         size="${fileSizeList[i]%%$'\t'*}";
  75         sum=$(( sum + size ));
  76
  77         # Check if point is within file
  78         if [[ $sum -gt $randPoint ]]; then
  79             # Calculate important positions within file (zero-indexed)
  80             fileStart=0;                              # first byte within file
  81             filePoint="$(( randPoint - psum ))";      # point as byte within file
  82             fileEnd="$(( size - 1 ))";                # last byte within file
  83             selStart="$(( filePoint - CONTEXT/2 ))";  # start of output selection
  84             selEnd="$(( filePoint + CONTEXT/2 ))";    # end of output selection
  85             # Clamp selection start and end indexes
  86             if [[ $selStart -lt $fileStart ]]; then selStart=$fileStart; fi;
  87             if [[ $selEnd -gt $fileEnd ]]; then selEnd=$fileEnd; fi;
  88             selCount="$(( selEnd - selStart + 1 ))";  # number of bytes within selection
  89             # Output context
  90             file="$(cut -f2- <<< "${fileSizeList[i]}"; )";
  91             printf "INFO:Sample of:%s\n" "$file";
  92             #head --bytes=$((selEnd + 1 )) -- "$file" | tail --bytes=+$((selStart + 1));
  93             tail --bytes=+$((selStart+1)) -- "$file" | head --bytes=$((selCount));
  94             printf "\n";
  95             return 0;
  96         fi;
  97         psum=$sum; # store previous sum
  98     done;
  99     #yell "DEBUG:getRandText() finished."; # debug
 100 };
 101 export -f checkInput getFileSizeList getRandText;
 102
 103 main() {
 104     # Input:  arg1    directory path
 105     # Output: stdout  text
 106
 107     checkInput "$@";      # check input arguments
 108     getFileSizeList "$1"; # make fileSizeList array
 109     getRandText;          # output text
 110     return 0;
 111 };
 112
 113 main "$@" && exit 0;