Take Invoice Scans from a Canon RS40 into a Hazel-Watched Folder, send them to AWS Textract and rename them into Date and Company

pdf-invoice-textract-and-rename

# Load Homebrew environment
eval "$(/opt/homebrew/bin/brew shellenv)"

# Variables
dir_path=$(dirname "$1")
base_filename=$(basename "$1")
log_file="${dir_path}/${base_filename}-error.log"

# Base64 encode the PDF file
base64_string=$(qpdf "$1" --pages . 1 -- - | base64)

# Call AWS Textract and handle errors
aws textract analyze-expense --profile textract --document "Bytes=$base64_string" 2> "$log_file" | jq -r '.ExpenseDocuments[0].SummaryFields[]' > "$1-summary.json"
if [ ${PIPESTATUS[0]} -ne 0 ]; then
    say -v Zoe "AWS Textract failed. Check log file."
    exit 1
fi

# Check if the JSON file is not empty
if [ -s "$1-summary.json" ]; then
    # Extract the date string
    date_str=$(jq -rs 'map(select(.Type.Text == "INVOICE_RECEIPT_DATE")) | sort_by(.Type.Confidence) | reverse | first | .ValueDetection.Text // empty' "$1-summary.json" | sed -e "s/,$//")

    # If the date string is not empty, process further
    if [ -n "$date_str" ]; then
        escaped_date_str=$(printf '%q' "$date_str")

        # Convert the date to ISO format using Date.parse and Chronic
        formatted_date=$(ruby -r date -r chronic -e "
require 'date';
require 'chronic';
date_str = '$escaped_date_str';
date = Chronic.parse(date_str)
if date && date.to_date != Date.today + 1
  parsed_date = date
else
  begin
    parsed_date = Date.parse(date_str)
  rescue ArgumentError
    parsed_date = nil
  end
end
puts (parsed_date&.strftime('%Y-%m-%d') || '')")

        # Check if formatted_date was successfully set
        if [ -z "$formatted_date" ]; then
            say -v Zoe "Date parse failed. Check log file."
            exit 1
        fi

        # Extract the vendor name and format it
        vendor=$(jq -rs 'map(select(.Type.Text == "VENDOR_NAME")) | sort_by(.Type.Confidence) | reverse | first | .ValueDetection.Text // empty' "$1-summary.json" | gsed -E 's/\s+/-/g')

        [ -n "$vendor" ] && vendor="-${vendor}"

        # If the formatted date is not empty, proceed with renaming the file
        if [ -n "$formatted_date" ]; then
            final_filename="${dir_path}/${formatted_date}${vendor}.pdf"
            if [ -e "$final_filename" ]; then
                final_filename="${dir_path}/${formatted_date}${vendor}-$$.pdf"
            fi
            mv "$1" "$final_filename"
        fi
    else
        say -v Zoe "Date string empty. Check log file."
        exit 1
    fi

    # Remove the JSON summary file and log file if everything was successful
    trash "$1-summary.json" "$log_file"
    say -v Zoe "Parsing Receipt ${vendor} from ${formatted_date} successful"
else
    echo "Error: JSON file is empty. Textract analysis might have failed." >&2
    say -v Zoe "Empty JSON. Check log file."
    trash "$1-summary.json"
    exit 1
fi