feat(user/randtxt.sh):Add script to sample text files in a dir

[BK-2020-03.git] / user / randtxt.sh

diff --git a/user/randtxt.sh b/user/randtxt.sh
new file mode 100755 (executable)
index 0000000..c809246
--- /dev/null
+++ b/user/randtxt.sh
@@ -0,0 +1,113 @@
+#!/bin/bash
+# Desc: Outputs text at randomish position within dir of text files
+# Usage: randtxt.sh DIR
+# Version 0.0.4
+# Depends: Bash 5.1.16, GNU findutils 4.8.0, GNU Coreutils 8.32
+
+SAMPLE=10000;
+CONTEXT="5000"; # total bytes +1 before and after point within file to print
+if [[ $(( CONTEXT/2 + CONTEXT/2 )) -lt $CONTEXT ]]; then ((CONTEXT++)); fi;
+
+
+yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
+die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
+must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
+showUsage() {
+    cat <<'EOF'
+    USAGE:
+        randtxt.sh DIR
+
+    EXAMPLE:
+      randtxt.sh ~/
+EOF
+} # Display information on how to use this script.
+checkInput() {
+    # input:  arg1
+    if [[ $# -gt 1 ]]; then die "FATAL:Too many arguments"; fi;
+    if [[ $# -lt 1 ]]; then die "FATAL:Not enough arguments"; fi;
+    if [[ ! -d "$1" ]]; then die "FATAL:Not a dir:$1"; else return 0; fi;
+    #yell "DEBUG:checkInput() finished."; # debug
+};
+getFileSizeList() {
+    # Desc: Create list of file sizes and paths
+    # input:  arg1   directory path
+    # output: array  fileSizeList
+    #         var    totalSize
+    #         var    randPoint
+    
+    declare -g -a fileSizeList;
+    mapfile -d '' -t fileSizeList < <(
+        find -- "$1" -type f -name "*.txt" -printf '%s\t%p\0' | shuf -z -n"$SAMPLE";
+    ); # Build array by feeding null-delimited lines from `find` to `mapfile`
+    #declare -p fileSizeList; # debug
+
+    declare -g totalSize=0;
+    local i size;
+    for i in "${!fileSizeList[@]}"; do
+        #declare -p i; # debug
+        #yell "DEBUG:fileSizeList[i]:${fileSizeList[i]}"; # debug
+        size="${fileSizeList[i]%%$'\t'*}";
+        #declare -p size; # debug
+        totalSize=$((totalSize + size));
+        #declare -p i size totalSize; # debug
+        #yell "=============="; # debug
+    done;
+    if [[ $totalSize -le 0 ]]; then die "FATAL:Total size is zero."; fi;
+    
+    declare -g randPoint
+    randPoint="$(shuf -n1 -i0-$((totalSize-1)); )";
+    #declare -p fileSizeList totalSize randPoint 1>&2;
+    #yell "DEBUG:fileSizeList element count:${#fileSizeList[@]}"; # debug
+    #yell "DEBUG:getFileSizeList() finished."; # debug
+};
+getRandText() {
+    # Desc: Print text within fileSizeList around randPoint
+    # input: array  fileSizeList
+    #        var    totalSize
+    #        var    randPoint
+    
+    local sum psum i size fileStart filePoint fileEnd selStart selEnd selCount file;
+    sum=0; # init byte sum
+    psum=0;
+    for i in "${!fileSizeList[@]}"; do
+        # Move search to next file
+        size="${fileSizeList[i]%%$'\t'*}";
+        sum=$(( sum + size ));
+
+        # Check if point is within file
+        if [[ $sum -gt $randPoint ]]; then
+            # Calculate important positions within file (zero-indexed)
+            fileStart=0;                              # first byte within file
+            filePoint="$(( randPoint - psum ))";      # point as byte within file
+            fileEnd="$(( size - 1 ))";                # last byte within file
+            selStart="$(( filePoint - CONTEXT/2 ))";  # start of output selection
+            selEnd="$(( filePoint + CONTEXT/2 ))";    # end of output selection
+            # Clamp selection start and end indexes
+            if [[ $selStart -lt $fileStart ]]; then selStart=$fileStart; fi;
+            if [[ $selEnd -gt $fileEnd ]]; then selEnd=$fileEnd; fi;
+            selCount="$(( selEnd - selStart + 1 ))";  # number of bytes within selection
+            # Output context
+            file="$(cut -f2- <<< "${fileSizeList[i]}"; )";
+            printf "INFO:Sample of:%s\n" "$file";
+            #head --bytes=$((selEnd + 1 )) -- "$file" | tail --bytes=+$((selStart + 1));
+            tail --bytes=+$((selStart+1)) -- "$file" | head --bytes=$((selCount));
+            printf "\n";
+            return 0;
+        fi;
+        psum=$sum; # store previous sum
+    done;
+    #yell "DEBUG:getRandText() finished."; # debug
+};
+export -f checkInput getFileSizeList getRandText;
+
+main() {
+    # Input:  arg1    directory path
+    # Output: stdout  text
+
+    checkInput "$@";      # check input arguments
+    getFileSizeList "$1"; # make fileSizeList array
+    getRandText;          # output text
+    return 0;
+};
+
+main "$@" && exit 0;