scr

#!/bin/bash
# Screenshot OCR tool
# requires: imagemagick, tesseract-ocr
# recommended to be installed with incron
# e.g. /home/iain/working/screenshots IN_CLOSE_WRITE /home/iain/bin/scr ocr $@/$#

op=$1
defaultfolder="${HOME}/working/screenshots"
convertpattern='s/\/\./\//g; s/\.txt\(:\|$\)/\1/'
  
if [ -n "$3" ]; then
  folder=$3
else
  folder=$defaultfolder
fi

case $op in
ocr)
  # OCR an image
  # text saved in hidden file next to image
  
  fn=$2
  txt=$(dirname "$fn")/.$(basename "$fn")

  echo "Doing OCR on $fn..."

  if [ -f "$txt.txt" ]; then
    echo "Already got a txt file"
    exit 1
  fi

  fnnocase=$(echo "$fn" | tr A-Z a-z)

  if [[ ! $fnnocase =~ \.(png|jpg)$ ]] || [[ $fn =~ \.lg\.png$ ]]; then
    echo "Not a screenshot"
    exit 1
  fi
  
  
  # Increase size to improve OCR accuracy
  convert -limit memory 8000MB -modulate 100,0 -resize 400% "$fn" "$fn.lg.png"

  # OCR to text file
  tesseract "$fn.lg.png" "$txt"

  # Write text to file metadata
  exiftool -overwrite_original -comment="$(cat "$txt.txt")" "$fn"
  exiftool -overwrite_original -Caption-Abstract="$(cat "$txt.txt")" "$fn"

  # Delete large version
  rm "$fn.lg.png"
  ;;

ocrall|oa)
  # OCR all images in the given dir
  ls "$folder" | while read f; do $0 ocr "$folder/$f"; done
  ;;

search|s)
  # Find txt in a screenshot dir
  search=$2
  grep -d skip -i "$search" "$folder"/.* | sed "$convertpattern" 2>/dev/null
  ;;

open|o)
  # Open each matching image
  $0 search "$2" "$folder" | cut -d: -f 1 | sort -u | while read f; do xdg-open "$f"; done
  ;;

cat|c)
  # Dump out all text for matching images
  grep -l -d skip -i "$2" "$folder"/.* | sort -u | while read f; do echo $f | sed "$convertpattern"; cat $f; done 2>/dev/null
  ;;

figure|fig|f)
  # Output base64 encoded HTML figures of each image matching
  $0 search "$2" "$folder" | cut -d: -f 1 | sort -u | while read f; 
  do 
    mimetype=$(file -bN --mime-type "$f")
    content=$(base64 -w0 < "$f")
    echo "\n<figure><img src=\"data:$mimetype;base64,$content\" width=\"100%\" alt=\"\" />\n<figcaption>: </figcaption>\n</figure>\n"
  done
esac