#!/usr/bin/env sh
# OCR a screen region (default) or an image file → clipboard.
#
# Usage:
#   ocr           # interactive: select a region with slurp, OCR it
#   ocr <file>    # OCR an image file
#
# Requires: tesseract (+tesseract-data-eng, tesseract-data-por),
#           grim, slurp, wl-clipboard, libnotify.
#
# Override languages via TESSERACT_LANG (e.g. TESSERACT_LANG=eng).

set -eu

lang="${TESSERACT_LANG:-eng+por}"

if [ "${1:-}" ]; then
  [ -r "$1" ] || {
    notify-send -u critical "📋 OCR" "Cannot read: $1"
    exit 1
  }
  text="$(tesseract "$1" - -l "$lang" 2>/dev/null || true)"
else
  region="$(slurp 2>/dev/null)" || exit 0
  text="$(grim -g "$region" - | tesseract - - -l "$lang" 2>/dev/null || true)"
fi

# Trim trailing whitespace per line; collapse runs of blank lines; drop
# leading blanks.
text="$(printf '%s\n' "$text" | awk '
  { sub(/[[:space:]]+$/, "") }
  NF { print; blank = 0; next }
  !blank { print; blank = 1 }
' | sed -e '/./,$!d')"

if [ -z "$text" ]; then
  notify-send -u low "📋 OCR" "No text detected."
  exit 1
fi

printf '%s' "$text" | wl-copy
notify-send -t 3000 "📋 OCR copied" "$(printf '%s' "$text" | head -c 200)"