unitproc/bkshuf

   1 #!/usr/bin/env bash
   2 # Desc: Mixes input lines while also preserving some neighbors
   3 # Usage: cat file | bkshuf arg1
   4 # Version 0.0.1
   5 # Depends: bc 1.07.1, GNU Coreutils 8.32 (shuf)
   6 # Input: var: arg1  initial lines to output
   7
   8 BKSHUF_PARAM_LINEC=1000000;
   9 BKSHUF_PARAM_GSIZE=25 # lines per group for BKSHUF_PARAM_LINEC lines of input
  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 $*"; } # runs args as command, reports args if command fails
  15 read_stdin() {
  16     # Desc: Consumes stdin; outputs as stdout lines
  17     # Input: stdin (consumes)
  18     # Output: stdout (newline delimited)
  19     # Example: printf "foo\nbar\n" | read_stdin
  20     # Depends: GNU bash (version 5.1.16)
  21     # Version: 0.0.1
  22     local input_stdin output;
  23
  24     # Store stdin
  25     if [[ -p /dev/stdin ]]; then
  26         input_stdin="$(cat -)";
  27     fi;
  28
  29     # Store as output array elements
  30     ## Read in stdin
  31     if [[ -n $input_stdin ]]; then
  32         while read -r line; do
  33             output+=("$line");
  34         done < <(printf "%s\n" "$input_stdin");
  35     fi;
  36
  37     # Print to stdout
  38     printf "%s\n" "${output[@]}";
  39 }; # read stdin to stdout lines
  40 checkInt() {
  41     # Desc: Checks if arg is integer
  42     # Usage: checkInt arg
  43     # Input: arg: integer
  44     # Output: - return code 0 (if arg is integer)
  45     #         - return code 1 (if arg is not integer)
  46     # Example: if ! checkInt $arg; then echo "not int"; fi;
  47     # Version: 0.0.1
  48     local returnState
  49
  50     #===Process Arg===
  51     if [[ $# -ne 1 ]]; then
  52         die "ERROR:Invalid number of arguments:$#";
  53     fi;
  54
  55     RETEST1='^[0-9]+$'; # Regular Expression to test
  56     if [[ ! $1 =~ $RETEST1 ]] ; then
  57         returnState="false";
  58     else
  59         returnState="true";
  60     fi;
  61
  62     #===Determine function return code===
  63     if [ "$returnState" = "true" ]; then
  64         return 0;
  65     else
  66         return 1;
  67     fi;
  68 } # Checks if arg is integer
  69 consume_line() {
  70     # Desc: Outputs and destroys line from list_input starting at index ip
  71     # Usage: consume_line;
  72     # Input: var: list_input  array input lines
  73     #        var: lco         int   line count original
  74     #        var: lcr         int   line count remaining
  75     #        var: ip          int   list_input index pointer
  76     # Output: stdout:    a single non-blank element from list_input
  77     #         list_input     one element destroyed
  78     #         var: lc_out    incremented once
  79     local n line;
  80
  81     n=0; # for tracking progress iterating through remaining list_input
  82     ### Loop once through list_input indices until a non-blank line found
  83     while [[ $n -le $lco ]]; do
  84         #### check if line at ip is blank
  85         line="${list_input[$ip]}";
  86         if [[ -n $line ]]; then
  87             ##### consume line at index ip
  88             printf "%s\n" "$line" & # print to output
  89             unset "list_input[$ip]"; # destroy line in list_input array
  90             ((lc_out++));
  91             ((lcr--)); # decrement line count remaining lcr
  92             #yell "DEBUG:Consumed line ip:$ip:$line";
  93             break;
  94         fi;
  95         #### increment input index pointer, looping around if necessary
  96         if [[ $ip -le $lco ]]; then ((ip++)); else ip=0; fi;
  97         #### track progress through list_input
  98         ((n++));
  99     done;
 100 }; # consume and output line in list_input array starting at index ip
 101 decide_read() {
 102     # Desc: Decides whether to read another line in list_input array
 103     #   by comparing $RANDOM to p_seq_int
 104     # Usage:  if decide_read; then something; fi;
 105     # Input: var: p_seq_int probability (int [0 32767])
 106
 107     if [[ $RANDOM -lt $p_seq_int ]]; then
 108         return 0;
 109     else
 110         return 1;
 111     fi;
 112 }; # returns 0 with probability p_seq; 1 otherwise
 113
 114 main() {
 115     declare par_l0 par_s0 s_exp s;
 116     declare -a list_input;
 117
 118     # Check positional arguments
 119     if [[ $# -gt 0 ]] && ! checkInt "$1"; then
 120         die "FATAL:Not an integer:$1";
 121     else
 122         lc_out_max="$1"; # output line count
 123     fi;
 124
 125
 126     # store input lines from stdin (like `shuf`)
 127     while read -r line; do
 128         if [[ -z $line ]]; then continue; fi; # skip blank lines
 129         #yell "DEBUG:INPUT:$line";
 130         list_input+=("$line");
 131     done < <( read_stdin; );
 132
 133     # calc line count (lc)
 134     lc="${#list_input[@]}";
 135     #lc="$(printf "%s\n" "${list_input[@]}" | wc -l )";
 136     #yell "DEBUG:lc:$lc";
 137     #yell "DEBUG:list_input:$(declare -p list_input)";
 138
 139     # calculate group size s
 140     par_l0="$BKSHUF_PARAM_LINEC";
 141     par_s0="$BKSHUF_PARAM_GSIZE";
 142     s_exp="(( $par_s0 - 1 )/( ( l( $par_l0 ) )^2 ))*(l( $lc ))^2+1";
 143     s="$(echo "scale=12; $s_exp" | bc -l)";
 144
 145     # calculate probabilities p_jump, p_seq
 146     p_jump="$(echo "scale=12; 1 / ( $s )" | bc -l)";
 147     p_seq="$(echo "scale=12; 1 - $p_jump" | bc -l)";
 148     p_seq_int="$(echo "scale=0; ($p_seq * 32767)/1" | bc -l)"; # p_seq as int [0 32767] for $RANDOM range
 149
 150     # generate output
 151     lco="${#list_input[@]}"; # save original input line count
 152     lcr="$lco";
 153     lc_out="0"; # init output line counter
 154     if [[ -z "$lc_out_max" ]]; then lc_out_max="$lco"; fi;
 155     ip="$(shuf -i0-$(( lco - 1 )) -n1)"; # init input index pointer
 156     n_loop1="0";
 157     #yell "DEBUG:max_blanks:$max_blanks"
 158     while [[ $lcr -ge 1 ]] && [[ $lc_out -lt $lc_out_max ]]; do
 159         ## Select random unconsumed input line and consume it to output
 160         ip="$(shuf -i0-$(( lco - 1 )) -n1)"; # init input index pointer
 161         consume_line;
 162
 163         ## Consume the next sequential line with probability p_seq.
 164         while decide_read && [[ $lcr -ge 1 ]] && [[ $lc_out -lt $lc_out_max ]]; do
 165             consume_line;
 166         done;
 167
 168         ((n_loop1++)); # increment jump counter
 169     done;
 170     #yell "DEBUG:n_loop1:$n_loop1"; # count jumps
 171
 172 }; # main program
 173
 174 main "$@";
 175
 176 # Author: Steven Baltakatei Sandoval
 177 # License: GPLv3+
 178
 179 # Dependency information
 180
 181 # bc 1.07.1
 182 # Copyright 1991-1994, 1997, 1998, 2000, 2004, 2006, 2008, 2012-2017 Free Software Foundation, Inc.
 183
 184
 185 # shuf (GNU coreutils) 8.32
 186 # Copyright (C) 2020 Free Software Foundation, Inc.
 187 # License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html>.
 188 # This is free software: you are free to change and redistribute it.
 189 # There is NO WARRANTY, to the extent permitted by law.
 190
 191 # Written by Paul Eggert.
 192
 193