banquo_initial_data.rb

require 'csv'

def split_and_tokenise(string_to_tokenise)
  string_to_tokenise.to_s.downcase.gsub(/[^a-z0-9]/, ' ').
      split(" ").
      map { |token| Digest::SHA256.hexdigest token }.
      join(" ")
end

def tokenise(string_to_tokenise)
  string_to_tokenise.present? ? split_and_tokenise(string_to_tokenise) : nil
end

def transaction_attrs(transaction)
  {
    bank_transaction_id: transaction.id,
    bank_transaction_dated_on: transaction.dated_on,
    bank_transaction_name: tokenise(transaction.name),
    bank_transaction_memo: tokenise(transaction.memo),
    bank_transaction_amount: transaction.amount,
    bank_transaction_deleted: transaction.deleted_at.present?,
    bank_account_id: transaction.bank_account_id,
    bank_account_is_personal: transaction.bank_account.is_personal,
    bank_account_type: transaction.bank_account.type,
    company_id: transaction.bank_account.company_id,
  }
end

def explanation_attrs(explanation)
  if explanation.nil? 
    {
      bank_account_entry_id: nil,
      general_ledger_account_id: nil,
      general_ledger_account_nominal_code_base: nil,
      general_ledger_account_nominal_code_sub: nil,
      general_ledger_account_type: nil,
      # area is not exposed with accessor
      general_ledger_account_area: nil,
    }
  else
    account = explanation.general_ledger_account
    {
      bank_account_entry_id: explanation.id,
      general_ledger_account_id: account.id,
      general_ledger_account_nominal_code_base: account.base_code,
      general_ledger_account_nominal_code_sub: account.sub_code,
      general_ledger_account_type: account.account_type.name,
      # area is not exposed with accessor
      general_ledger_account_area: account.account_type.instance_variable_get(:@area),
    }
  end
end

models_to_include = [
  :bank_account,
  bank_account_entries: {general_ledger_account: :chart_of_accounts},
]

# in the initial instance we just want companys with the tracking feature
# and those that would have had it in september 2017
company_ids = []
Company.where("created_at < '2017-10-01'").find_each do |c|    
  company_ids << (c.has_feature?(:dw_banking_events) ? c.id : nil)    
end    

# ony want a sample of all data
#number_of_samples = 10
#Random.srand(666)
sampling = {
  bank_account_id: BankAccount.
    joins(:company).
    merge(Company.find(company_ids)).
    pluck("bank_accounts.id"),
}

FILE_PREFIX = 'banquo_training_data_on'
writer = Metrics::CsvFileWriter.new(FILE_PREFIX)

writer.open_csv_for_write do |csv|
  BankTransaction.
    where("bank_transactions.created_at < '2017-10-01' and bank_transactions.created_at > '2017-09-01'").
    where(sampling).
    left_outer_joins(models_to_include).
    includes(models_to_include).
    find_each do |transaction|
      explanations = transaction.bank_account_entries
      
      if explanations.length == 0
        attrs = transaction_attrs(transaction).merge(explanation_attrs(nil))

        csv << attrs.keys if csv.lineno.zero?
        csv << attrs.values
      end
      
      explanations.each do |explanation|
        attrs = transaction_attrs(transaction).merge(explanation_attrs(explanation))

        csv << attrs.keys if csv.lineno.zero?
        csv << attrs.values
      end
    end
end