Class: AtlasEngine::ValidationTranscriber::AddressParserBase

Inherits:

Object

• Object
• AtlasEngine::ValidationTranscriber::AddressParserBase

Extended by:

T::Helpers, T::Sig

Includes:

Formatter

Defined in:

app/lib/atlas_engine/validation_transcriber/address_parser_base.rb

Direct Known Subclasses

At::ValidationTranscriber::AddressParser, Au::ValidationTranscriber::AddressParser, Be::ValidationTranscriber::AddressParser, Ch::Locales::Fr::ValidationTranscriber::AddressParser, Cz::ValidationTranscriber::AddressParser, De::ValidationTranscriber::AddressParser, Dk::ValidationTranscriber::AddressParser, Es::ValidationTranscriber::AddressParser, Fr::ValidationTranscriber::AddressParser, Gg::ValidationTranscriber::AddressParser, Kr::ValidationTranscriber::AddressParser, Lt::ValidationTranscriber::AddressParser, Lu::ValidationTranscriber::AddressParser, Nl::ValidationTranscriber::AddressParser, Pl::ValidationTranscriber::AddressParser, Pt::ValidationTranscriber::AddressParser, Si::ValidationTranscriber::AddressParser, AddressParserNorthAmerica, AddressParserOceanic

Constant Summary

AddressComponents =

T.type_alias { T::Hash[Symbol, String] }

BUILDING_NAME =

Parts that we slot into address format regular expressions

"(?<building_name>[\\w ]+)"

BUILDING_NUM =

"(?<building_num>(" \
'([[:digit:]]+\s)?([[:digit:]]+/[[:digit:]]+)|' \
'[[:digit:]][[:alpha:][:digit:]/\-]*|' \
'[[:alpha:]][[:digit:]][[:alpha:][:digit:]/\-]*' \
"))"

NUMERIC_ONLY_BUILDING_NUM =

"(?<building_num>(" \
'([[:digit:]]+\s+)?[[:digit:]][[:digit:]/]*[[:digit:]]|' \
"[[:digit:]]+" \
"))"

NON_NUMERIC_STREET =

"(?<street>[^[:digit:]/ -].*)"

STREET =

"(?<street>.+)"

STREET_NO_COMMAS =

"(?<street>[^,]+)"

UNIT_TYPE =

'(?<unit_type>[[:alpha:]]+\.?)'

UNIT_NUM =

'(?<unit_num>[[:alpha:][:digit:]/\-]+)'

UNIT_NUM_NO_HYPHEN =

"(?<unit_num>[[:alpha:][:digit:]/]+)"

PO_BOX =

%r{(?:^|\s|/)(?:p(?:ost)?\.?\s*o(?:ffice)?\.?\s*box|box|postal\s*box)\s+(\d+)(?:$|\s)}i

Instance Method Summary

• #initialize(address:, preprocessor: nil) ⇒ AddressParserBase constructor

  A new instance of AddressParserBase.

• #parse ⇒ Object

Methods included from Formatter

#build_address, #strip_trailing_punctuation, #strip_word

Constructor Details

#initialize(address:, preprocessor: nil) ⇒ AddressParserBase

Returns a new instance of AddressParserBase.

Raises:

• (ArgumentError)

# File 'app/lib/atlas_engine/validation_transcriber/address_parser_base.rb', line 47

def initialize(address:, preprocessor: nil)
  raise ArgumentError, "country_code cannot be blank in address" if address.country_code.blank?

  @constants = T.let(
    AtlasEngine::ValidationTranscriber::Constants.instance,
    AtlasEngine::ValidationTranscriber::Constants,
  )
  @country_regex_formats = T.let(nil, T.nilable(T::Array[Regexp]))
  @address = T.let(address, AddressValidation::TAddress)

  @preprocessor = T.let(
    preprocessor || AddressParserPreprocessor.new(address: @address),
    T.nilable(ValidationTranscriber::AddressParserPreprocessor),
  )
end

Instance Method Details

#parse ⇒ Object

# File 'app/lib/atlas_engine/validation_transcriber/address_parser_base.rb', line 64

def parse
  candidates = []

  address_lines = @preprocessor&.generate_combinations

  return candidates if address_lines&.empty?

  address_lines&.each do |address_line|
    address_line, po_box = extract_po_box(address_line)

    country_regex_formats.each do |format|
      m = format.match(address_line)
      next if m.nil?

      captures = m.named_captures.symbolize_keys

      next if ridiculous?(captures, @address)

      captures = captures.compact_blank.transform_values! do |value|
        strip_trailing_punctuation(value)
      end
      captures[:po_box] = po_box if po_box

      candidates << captures
    end

    if po_box && candidates.empty?
      candidates << { po_box: po_box }
    end
  end

  candidates.uniq
end