Rename your shitty named invoice pdfs to `YYYY-MM-DD_name-of-the-biller_invoice-number.pdf` (the code below outputs the name in german, remove `Please answer in german.` if you want english output)

rename_invoice_pdfs.ruby

# usage: bundle exec ruby rename_invoice_pdfs.rb folder/*.pdf

require "openai"
require "json"
require "fileutils"
require "shellwords"

dry_run = false

client = OpenAI::Client.new(
  access_token: "your-api-key",
  log_errors: true # Highly recommended in development, so you can see what errors OpenAI is returning. Not recommended in production because it could leak private data to your logs.
)

question = "Please give me a filename for the text delimited by triple quotes. Please answer in german. The filename shoould start with the invoice date in the format year-month-day. Followd by a hyphen and contains the name of biller and the the invoice number. Please return the filename in the json field filename."


files = ARGV.filter { |f| f.end_with?('.pdf') }


files.each do |file|
  text = `docker run --rm -i kadock/pdftotext < #{Shellwords.escape(file)}`
  next if text.nil? || text.strip == ''

  response = client.chat(
    parameters: {
        model: "gpt-4-turbo", # Required.
        response_format: { type: "json_object" },
        messages: [{ role: "user", content: "#{question}\n\n\"\"\"#{text}\"\"\""}], # Required.
    }
  )
  pdf_filename = JSON.parse(response.dig("choices", 0, "message", "content"))["filename"]
  next if pdf_filename.strip == ''

  file_name = File.basename(pdf_filename, ".*").gsub(/\s+/, '_').downcase

  pdf_filename = File.dirname(file) + "/#{file_name}.pdf"
  if File.exist?(pdf_filename)
    pdf_filename = pdf_filename.gsub(/\.pdf$/, "_#{Time.now.to_i}.pdf")
  end
  puts "#{file} -> #{pdf_filename}"
  FileUtils.mv(file, pdf_filename) unless dry_run
end