diff-text

#!/bin/bash
# diff-text — OCR two images and diff the extracted text
#
# Perfect for comparing screenshots before/after a change,
# spotting differences in document versions, etc.
#
# Usage:
#   diff-text <image1> <image2>       # show text diff
#   diff-text -s <image1> <image2>    # side-by-side diff
#   diff-text -c <image1> <image2>    # copy diff to clipboard
#
# Examples:
#   diff-text before.png after.png
#   diff-text old_receipt.jpg new_receipt.jpg -s
#   diff-text v1.png v2.png | head -20
#
# Requires: auge

set -euo pipefail

copy=false
side=false
files=()

while [[ $# -gt 0 ]]; do
  case "$1" in
    -c|--copy)      copy=true; shift ;;
    -s|--side)      side=true; shift ;;
    -h|--help)
      sed -n '2,/^$/{ s/^# //; s/^#//; p; }' "$0"
      exit 0 ;;
    *)
      files+=("$1"); shift ;;
  esac
done

[[ ${#files[@]} -eq 2 ]] || { echo "usage: diff-text <image1> <image2>" >&2; exit 2; }
[[ -f "${files[0]}" ]] || { echo "error: file not found: ${files[0]}" >&2; exit 1; }
[[ -f "${files[1]}" ]] || { echo "error: file not found: ${files[1]}" >&2; exit 1; }

command -v auge >/dev/null || { echo "error: auge not found" >&2; exit 1; }

tmp1=$(mktemp /tmp/diff-text-1-XXXXXX.txt)
tmp2=$(mktemp /tmp/diff-text-2-XXXXXX.txt)
trap 'rm -f "$tmp1" "$tmp2"' EXIT

echo "OCR: ${files[0]}..." >&2
auge --ocr "${files[0]}" -q 2>/dev/null > "$tmp1"

echo "OCR: ${files[1]}..." >&2
auge --ocr "${files[1]}" -q 2>/dev/null > "$tmp2"

if $side; then
  output=$(diff --side-by-side --width=120 "$tmp1" "$tmp2" 2>/dev/null || true)
else
  output=$(diff --unified=3 \
    --label "$(basename "${files[0]}")" \
    --label "$(basename "${files[1]}")" \
    "$tmp1" "$tmp2" 2>/dev/null || true)
fi

if [[ -z "$output" ]]; then
  echo "No text differences found."
else
  echo "$output"
fi

if $copy && [[ -n "$output" ]]; then
  printf '%s' "$output" | pbcopy
  printf '\033[2m(copied to clipboard)\033[0m\n' >&2
fi