user/bktess

   1 #!/bin/bash
   2 #Desc: A custom wrapper for running tesseract recursively on a directory
   3 # Version: 0.0.1
   4 # Depends: Debian: tesseract 4.1.1
   5 # Depends: BK-2020-03: checkapp() 0.1.1, checkfile() 0.1.2, checkdir() 0.1.2, displayMissing() 1.0.0
   6
   7 declare -Ag appRollCall # Associative array for storing app status
   8 declare -Ag fileRollCall # Associative array for storing file status
   9 declare -Ag dirRollCall # Associative array for storing dir status
  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 $*"; } # runs args as command, reports args if command fails
  15 checkapp() {
  16     # Desc: If arg is a command, save result in assoc array 'appRollCall'
  17     # Usage: checkapp arg1 arg2 arg3 ...
  18     # Version: 0.1.1
  19     # Input: global assoc. array 'appRollCall'
  20     # Output: adds/updates key(value) to global assoc array 'appRollCall'
  21     # Depends: bash 5.0.3
  22     local returnState
  23
  24     #===Process Args===
  25     for arg in "$@"; do
  26         if command -v "$arg" 1>/dev/null 2>&1; then # Check if arg is a valid command
  27             appRollCall[$arg]="true";
  28             if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
  29         else
  30             appRollCall[$arg]="false"; returnState="false";
  31         fi;
  32     done;
  33
  34     #===Determine function return code===
  35     if [ "$returnState" = "true" ]; then
  36         return 0;
  37     else
  38         return 1;
  39     fi;
  40 } # Check that app exists
  41 checkfile() {
  42     # Desc: If arg is a file path, save result in assoc array 'fileRollCall'
  43     # Usage: checkfile arg1 arg2 arg3 ...
  44     # Version: 0.1.2
  45     # Input: global assoc. array 'fileRollCall'
  46     # Output: adds/updates key(value) to global assoc array 'fileRollCall';
  47     # Output: returns 0 if app found, 1 otherwise
  48     # Depends: bash 5.0.3
  49     local returnState
  50
  51     #===Process Args===
  52     for arg in "$@"; do
  53         if [ -f "$arg" ]; then
  54             fileRollCall["$arg"]="true";
  55             if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
  56         elif [ -z "$arg" ]; then
  57             fileRollCall["(no name)"]="false"; returnState="false";
  58         else
  59             fileRollCall["$arg"]="false"; returnState="false";
  60         fi;
  61     done;
  62
  63     #===Determine function return code===
  64     if [ "$returnState" = "true" ]; then
  65         return 0;
  66     else
  67         return 1;
  68     fi;
  69 } # Check that file exists
  70 checkdir() {
  71     # Desc: If arg is a dir path, save result in assoc array 'dirRollCall'
  72     # Usage: checkdir arg1 arg2 arg3 ...
  73     # Version 0.1.2
  74     # Input: global assoc. array 'dirRollCall'
  75     # Output: adds/updates key(value) to global assoc array 'dirRollCall';
  76     # Output: returns 0 if all args are dirs; 1 otherwise
  77     # Depends: Bash 5.0.3
  78     local returnState
  79
  80     #===Process Args===
  81     for arg in "$@"; do
  82         if [ -z "$arg" ]; then
  83             dirRollCall["(Unspecified Dirname(s))"]="false"; returnState="false";
  84         elif [ -d "$arg" ]; then
  85             dirRollCall["$arg"]="true";
  86             if ! [ "$returnState" = "false" ]; then returnState="true"; fi
  87         else
  88             dirRollCall["$arg"]="false"; returnState="false";
  89         fi
  90     done
  91
  92     #===Determine function return code===
  93     if [ "$returnState" = "true" ]; then
  94         return 0;
  95     else
  96         return 1;
  97     fi
  98 } # Check that dir exists
  99 displayMissing() {
 100     # Desc: Displays missing apps, files, and dirs
 101     # Usage: displayMissing
 102     # Version 1.0.0
 103     # Input: associative arrays: appRollCall, fileRollCall, dirRollCall
 104     # Output: stderr: messages indicating missing apps, file, or dirs
 105     # Output: returns exit code 0 if nothing missing; 1 otherwise
 106     # Depends: bash 5, checkAppFileDir()
 107     local missingApps value appMissing missingFiles fileMissing
 108     local missingDirs dirMissing
 109
 110     #==BEGIN Display errors==
 111     #===BEGIN Display Missing Apps===
 112     missingApps="Missing apps  :";
 113     #for key in "${!appRollCall[@]}"; do echo "DEBUG:$key => ${appRollCall[$key]}"; done
 114     for key in "${!appRollCall[@]}"; do
 115         value="${appRollCall[$key]}";
 116         if [ "$value" = "false" ]; then
 117             #echo "DEBUG:Missing apps: $key => $value";
 118             missingApps="$missingApps""$key ";
 119             appMissing="true";
 120         fi;
 121     done;
 122     if [ "$appMissing" = "true" ]; then  # Only indicate if an app is missing.
 123         echo "$missingApps" 1>&2;
 124     fi;
 125     unset value;
 126     #===END Display Missing Apps===
 127
 128     #===BEGIN Display Missing Files===
 129     missingFiles="Missing files:";
 130     #for key in "${!fileRollCall[@]}"; do echo "DEBUG:$key => ${fileRollCall[$key]}"; done
 131     for key in "${!fileRollCall[@]}"; do
 132         value="${fileRollCall[$key]}";
 133         if [ "$value" = "false" ]; then
 134             #echo "DEBUG:Missing files: $key => $value";
 135             missingFiles="$missingFiles""$key ";
 136             fileMissing="true";
 137         fi;
 138     done;
 139     if [ "$fileMissing" = "true" ]; then  # Only indicate if an app is missing.
 140         echo "$missingFiles" 1>&2;
 141     fi;
 142     unset value;
 143     #===END Display Missing Files===
 144
 145     #===BEGIN Display Missing Directories===
 146     missingDirs="Missing dirs:";
 147     #for key in "${!dirRollCall[@]}"; do echo "DEBUG:$key => ${dirRollCall[$key]}"; done
 148     for key in "${!dirRollCall[@]}"; do
 149         value="${dirRollCall[$key]}";
 150         if [ "$value" = "false" ]; then
 151             #echo "DEBUG:Missing dirs: $key => $value";
 152             missingDirs="$missingDirs""$key ";
 153             dirMissing="true";
 154         fi;
 155     done;
 156     if [ "$dirMissing" = "true" ]; then  # Only indicate if an dir is missing.
 157         echo "$missingDirs" 1>&2;
 158     fi;
 159     unset value;
 160     #===END Display Missing Directories===
 161
 162     #==END Display errors==
 163     #==BEGIN Determine function return code===
 164     if [ "$appMissing" == "true" ] || [ "$fileMissing" == "true" ] || [ "$dirMissing" == "true" ]; then
 165         return 1;
 166     else
 167         return 0;
 168     fi
 169     #==END Determine function return code===
 170 } # Display missing apps, files, dirs
 171 checkPlumbing() {
 172     if ! checkapp tesseract parallel grep shuf find; then
 173         displayMissing; die "FATAL: Missing apps."; fi;
 174     if [[ ! -d "$1" ]]; then die "FATAL:Not a dir:$1"; fi;
 175 }; # Check dependencies and expected inputs
 176 runTesseract() {
 177     # Depends: Debian: tesseract 4.1.1
 178     # Input: arg1:  file path
 179
 180     # Check input file path
 181     if [[ ! -f "$1" ]]; then yell "File not found:$1"; return 1; fi;
 182
 183     # Print file path
 184     printf "File path:%s\n" "$1";
 185
 186     # Run tesseract
 187     if ! tesseract "$1" stdout 2>/dev/random; then yell "STATUS:Tesseract exited with error."; fi;
 188 }; # Run tesseract
 189 main() {
 190     declare -a array_files; # array for storing files to scan
 191
 192     # Inputs: arg1: dir
 193     checkPlumbing "$@";
 194
 195     # Assemble file list
 196     pat_exclusions=".ots$";
 197     while read -r line; do
 198         array_files+=("$line");
 199     done < <(find "$1" -type f | grep -Eiv "$pat_exclusions" | shuf);
 200
 201     # Run tesseract
 202     parallel --jobs="100%" runTesseract '{}' ::: "${array_files[@]}";
 203 }; # main program
 204
 205 export -f yell die must runTesseract;
 206
 207 main "$@";