apex_drivewealth_security_extractor.rb

require 'tempfile'
require 'bundler/inline'
require 'date'

gemfile do
  source 'https://rubygems.org'
  ruby '3.3.0'
  gem 'pdftotext', require: true
  gem 'pry', require: true
end

# Apex
APEX_PATTERN = /APEX/i

# DriveWealth
DRIVEWEALTH_PATTERN = /DriveWealth/i

notas = Dir.glob('*.pdf').map do |file|
  [file, Pdftotext.text(file)]
end

def date_parser(date) = Date.strptime(date, '%m/%Od/%Y')
def apex_date_parser(date) = Date.strptime(date, '%m/%Od/%y')

def extract_from_apex(nota)
  nota.split("\n").grep(/\A\d/).select { !_1.include?('Paul Street 1300') }.map(&:split).map do |line|
    trade_date, settle_date, qty, ticker, price, principal, _, transaction_fee, fees, _, net_amount = line[2..12]
    { trade_date: apex_date_parser(trade_date), settle_date: apex_date_parser(settle_date), qty:, ticker:, price:, principal:,
      transaction_fee:, fees:, net_amount:, line: }
  end
end

def extract_from_drivewealth(nota)
  nota_lines = nota.split("\n")
  last_line = nil
  nota_lines.to_enum.with_index.to_a.select { |(line, _)| line.include?('Agent') }.map do |line, index|
    lines = nota_lines[index + 2..index + 8].map { _1.split(/\s+/).last }

    principal = lines[0]
    fees = lines[2]
    transaction_fee = lines[3]
    net_amount = lines[6]

    line = line.split(/\s{6,}/).map(&:strip)
    last_line = [index, line]
    ticker = line[0]
    qty = line[4]

    dates_and_price = line[-2].split(/\s+/)
    if dates_and_price.size == 2
      trade_date, settle_date = dates_and_price
      price = line[5]
    else
      price, trade_date, settle_date = dates_and_price
    end

    { trade_date: date_parser(trade_date), settle_date: date_parser(settle_date), qty:, ticker:, price:, principal:,
      fees:, transaction_fee:, net_amount:, line: }
  end
end

result = notas.flat_map do |(file, nota)|
  if nota.match(APEX_PATTERN)
    puts ">>>>> Apex: #{file}"
    extract_from_apex(nota).map { _1.merge(file:) }
  elsif nota.match(DRIVEWEALTH_PATTERN)
    puts ">>> DriveWealth: #{file}"
    extract_from_drivewealth(nota).map { _1.merge(file:) }
  else
    raise "Unknown broker for file #{file} with content: #{nota}"
  end
end