ocr.sh

#!/bin/sh
# Source https://github.com/sdushantha/dotfiles/blob/master/bin/bin/utils/ocr

TEXT_FILE="/tmp/ocr.txt"
IMAGE_FILE="/tmp/ocr.png"

# Check if the needed dependencies are installed
dependencies="tesseract-ocr
maim
notify-send
xclip"
for dependency in $dependencies
do
	command -v "$dependency" || {
		notify-send "ocr" "Could not find '${dependency}', is it installed?"
		exit 1
	} >/dev/null 2>&1
done

# shellcheck disable=SC2064
trap "rm -rf $TEXT_FILE $IMAGE_FILE" EXIT INT TERM

# Take screenshot by selecting the area
maim -s "$IMAGE_FILE" || exit 1

# Do the magic (∩^o^)⊃━☆ﾟ.*･｡ﾟ
# Notice how I have removing the extension .txt from the file path. This is
# because tesseract-ocr adds .txt to the given file path anyways. So if we were to
# specify /tmp/ocr.txt as the file path, tesseract-ocr would out the text to
# /tmp/ocr.txt.txt
tesseract-ocr "$IMAGE_FILE" "${TEXT_FILE%.txt}" 2>/dev/null

# Remove the new page character.
# Source: https://askubuntu.com/a/1276441/782646
sed -i 's/\x0c//' "$TEXT_FILE"

# Check if the text was detected by checking number
# of lines in the file
NUM_LINES=$(wc -l <$TEXT_FILE)
if [ "$NUM_LINES" -eq 0 ]
then
	notify-send "ocr" "no text was detected"
	exit 1
fi

# Copy text to clipboard
xclip -selection clip <"$TEXT_FILE"

# Send a notification with the text that was grabbed using OCR
notify-send "ocr" "$(cat $TEXT_FILE)"