California BCC Bureau of Cannabis Control's License CSV Parser

licence_reader.rb

require 'csv'

# This LicenseReader will receive a CSV file containing business license information. Different US states will implement completely different formats,
# the following example is for California. The LicenseReader, integrated into the LIMS will allow the laboratory to type a license `C11-0001123-LIC` and
# return attribute data businessuctures compatible with the creation of the Client and License model.The output must remain constant but the input must
# be parsed using several regular expressions depending on the data made available by the regulatory agency. The class must take into consideration
# fault tolerancy when data is incomplete. Particularly badly businessucture in the example is the column Business Contact Information!
#
# "License Number","License Type","Business Owner","Business Contact Information","Business businessucture","Premise Address","Status","Issue Date","Expiration Date","Activities","Adult-Use/Medicinal"
# "C11-0001129-LIC","Cannabis - Dibusinessibutor License","RICHARD SMITH","CLH Dibusinessibution, LLC : Email- [email protected] : Phone- 8057203518 ","undefined","LOMPOC, CA 934366120 County: SANTA BARBARA ","Active","12/17/2019","12/16/2020","N/A for this license type","BOTH"
# "C11-0001130-LIC","Cannabis - Dibusinessibutor License","Steven Dang","XTRACTA DIbusinessIBUTION, LLC : Email- [email protected] : Phone- 6196345267 ","undefined","San Diego, CA 92126 County: SAN DIEGO ","Active","12/17/2019","12/16/2020","N/A for this license type","BOTH"
# "C11-0001123-LIC","Cannabis - Dibusinessibutor License","Douglas Fierro","Humboldt Hills Dibusinessibution : Email- [email protected] : Phone- 2138001789 ","undefined","SHASTA LAKE, CA 96019 County: SHASTA ","Active","12/13/2019","12/12/2020","N/A for this license type","BOTH"
# "C11-0001122-LIC","Cannabis - Dibusinessibutor License","Leo Shlovsky","ONE LED CA LLC  : Green Revolution : Email- [email protected] : Phone- 9177545717 ","undefined","DESERT HOT SPRINGS, CA 92240 County: ","Active","12/12/2019","12/11/2020","N/A for this license type","BOTH"
# "C11-0001121-LIC","Cannabis - Dibusinessibutor License","Austin Lubin","PENINSULA DIbusinessIBUTION : Email- [email protected] : Phone- 8313244917 ","undefined","SEASIDE, CA 939554321 County: MONTEREY ","Active","12/10/2019","12/09/2020","N/A for this license type","BOTH"
# "C11-0001120-LIC","Cannabis - Dibusinessibutor License","Daniel Nathanson","SWEETWATER CANYON DEVELOPMENT, LLC  : Malibu Gold : Email- [email protected] : Phone- 4243351086 ","undefined","SANTA ANA, CA 927074236 County: ORANGE ","Active","12/04/2019","12/03/2020","N/A for this license type","BOTH"
#
# Example of use
# lr = LicenseReader.new(path: csv, regs: 'ca')
# lr.query('C11-0001121-LIC') => true
# lr.client_attributes => {:business_name=>"PENINSULA DISTRIBUTION", ...
# lr.license_attributes => {:name=>"PENINSULA DISTRIBUTION", ...
#
# Parse the whole file
#
# lr.read_lines
# lr.lines.each do |line|
#   next if line[0] == 'License Number'
#
#   if lr.query(line[0])
#     puts "\nclient_attributes"
#     puts lr.client_attributes
#     puts "\nlicense_attributes"
#     puts lr.license_attributes
#     puts
#   end
# end

class LicenseReader
  attr_accessor :path, :lines, :data, :regs, :match, :query
  def initialize(path:, regs:)
    self.path = path
    self.regs = regs
    self.lines = []
    self.data  = {}
  end

  def read_lines
    self.lines = CSV.read(path)
  end

  def query(q)
    self.query = nil
    self.match = `grep '#{q}' #{path}`
    return false if match == ''

    self.query = q
    true
  end

  def client_attributes
    return {} if match == ''

    parse_line
    {
      business_name: data[:business_name],
      business_type: data[:business_type],
      url: data[:website],
      first_name: data[:first_name],
      last_name: data[:last_name],
      address: data[:location],
      phone1: data[:phone],
      email1: data[:email]
    }
  end

  def license_attributes
    return {} if match == ''

    parse_line
    {
      name: data[:dba] || data[:business_name],
      address: data[:location],
      number: data[:license_number],
      license_type: data[:license_type],
      verified: true,
      expires_on: Date.strptime(data[:expires], '%m/%d/%Y').to_s,
      deleted: false
    }
  end

  # data breaker
  def parse_line
    case regs
    when 'ca', 'california'
      license_number,
      license_type,
      name,
      business,
      business_type,
      location,
      _,
      assigned,
      expires,
      _,
      category = match.split(/\s*","\s*/)
      business_array = business.split(/\s+\:\s+/)
      md = business.match(/Email- (?<email>[^\s]+)\s*:*/)
      email = md && md[:email]
      md = business.match(/Phone- (?<phone>[^\s]+)\s*:*/)
      phone = md && md[:phone]
      md = business.match(/Website- (?<website>[^\s]+)\s*:*/)
      website = md && md[:website]
      business_name = business_array[0]
      dba = business_array[1]
      dba = nil if dba =~ /Email/ || dba =~ /Phone/ || dba =~ /Website/
      first_name, last_name = name.split(/\s+/)
      license_number.delete!('"')

      row_data = %i[
        license_number
        license_type
        name
        business
        business_type
        location
        assigned
        expires
        category
        dba
        business_name
        email
        phone
        website
        first_name
        last_name].zip([
          license_number,
          license_type,
          name,
          business,
          business_type,
          location,
          assigned,
          expires,
          category,
          dba,
          business_name,
          email,
          phone,
          website,
          first_name,
          last_name
        ])

      self.data = Hash[row_data]
    end
  end
end