swissmatch-street 0.0.1

data/lib/swissmatch/street.rb

@@ -0,0 +1,159 @@
+# encoding: utf-8
+
+
+
+require 'swissmatch/street/version'
+
+
+
+# From SwissMatch::Street
+# Parse and handle street names and numbers.
+#
+# @note
+#   All strings passed to SwissMatch are expected to be utf-8. All strings
+#   returned by SwissMatch are also in utf-8.
+#
+module SwissMatch
+
+  # Street
+  # Parse and handle street names and numbers.
+  class Street
+    # House number formats:
+    # '12'
+    # '12b'
+    # '12bis', '12BIS', '12Bis', '12 bis', … - see development/Noteworthy_things.txt
+    # '12 B'
+    # '12/14'
+    # '105-107'
+    # '16/2/22'
+    # '8-10-12'
+    # '16-2/22'
+    # '16/2-22'
+    HouseNumber                       = /\d+(?:\x20?\w{1,2}|(?:\/\d+|-\d+\w{1,2})*)/
+    FrontHouseNumber                  = /\d+(?:\x20?(?!ch|bd|av)\w{1,2}|(?:\/\d+|-\d+\w{1,2})*)/i
+    MatchStreetAndStreetNumberGerman  = /\A(.*?)(?:(?: |, ?)(#{HouseNumber}))\z/
+    MatchStreetAndStreetNumberFrench  = /\A(?:(#{FrontHouseNumber}), )(.*?)\z/
+    MatchBrokenStreetAndNumberGerman  = /\A(.*?)(?:(#{HouseNumber}))\z/
+    MatchBrokenStreetAndNumberFrench  = /\A(?:(#{FrontHouseNumber})[.,]| )(.*?)\z/
+
+    # Abbreviations that can be expanded
+    Replacements = {
+      'ch'    => 'Chemin',
+      'chem'  => 'Chemin',
+      'rte'   => 'Route',
+      'bd'    => 'Boulevard',
+      'av'    => 'Avenue',
+      'ave'   => 'Avenue',
+      'str'   => 'strasse',
+      'mte'   => 'Monte',
+      's'     => 'san',
+    }
+    # Detect the abbreviations to expand
+    ReplacementsMatch = /\b(?:ch|chem|rte|bd|ave?|mte)(?:\.|\b)|\bs(?:\.|\b)(?!$)|\Bstr(?:\.|\b)/i
+
+
+    NoCapitalize = {
+      'auf'     => 'auf',
+      'uf'      => 'uf',   # hurray for swiss german street names :D
+      'em'      => 'em',
+      'der'     => 'der',
+      'die'     => 'die',
+      'das'     => 'das',
+      'von'     => 'von',
+      'nach'    => 'nach',
+      'im'      => 'im',
+      'in'      => 'in',
+      'zum'     => 'zum',
+      'zur'     => 'zur',
+      'unteren' => 'unteren',
+      'oberen'  => 'oberen',
+
+      'd'       => 'd',
+      'de'      => 'de',
+      'des'     => 'des',
+      'du'      => 'du',
+      'l'       => 'l',
+      'le'      => 'le',
+      'la'      => 'la',
+      'les'     => 'les',
+      'vers'    => 'vers',
+
+      'il'      => 'il',
+      'dei'     => 'dei',
+      'di'      => 'di',
+      'delle'   => 'delle',
+      'della'   => 'della',
+      'al'      => 'al',
+      'alla'    => 'alla',
+      'alle'    => 'alle',
+      'ai'      => 'ai',
+    }
+
+    def self.normalize_street(street)
+      return '' unless street
+
+      street.strip.
+        squeeze(' ').
+        gsub(/\s*-\s*/, '-').
+        gsub(/\A(#{FrontHouseNumber}) /, '\1, '). # '24 Rue Baulacre'     => '24, Rue Baulacre' - but not '24 bd blabla' -> '24 Boulevard, Blabla'
+        gsub(/\s*([.,])(?=\S)/, '\1 ').           # '283,Rte.de Meyrin'   => '283, Rte. de Meyrin; '283 ,Foo' => '283, Foo'
+        gsub(ReplacementsMatch) { |m|             # ch., chem., str. etc. => chemin, strasse etc.
+          Replacements[m.downcase.chomp('.')]
+        }.
+        gsub(/\s*n°\s*/, ' ')
+    end
+
+    def self.normalize_name(name)
+      name.
+        gsub(/\b[\p{Letter}\p{Mark}\p{Connector_Punctuation}]{2,}\b/) { |word| NoCapitalize.fetch(word.downcase) { word.capitalize } }.
+        sub(/\b[\p{Letter}\p{Mark}\p{Connector_Punctuation}]{2,}\b/) { |word| word.capitalize } # [\p{Letter}\p{Mark}\p{Connector_Punctuation}] is \p{Word} without digits
+    end
+
+    def self.normalize_number(number)
+      return unless number
+      normalized = number.downcase.delete('^0-9a-z/()-')
+
+      normalized.empty? ? nil : normalized
+    end
+
+    def self.parse(street, normalize=false)
+      normalized        = normalize_street(street)
+      name, number, pos = case normalized
+        when MatchStreetAndStreetNumberGerman then [$1, $2, :end]
+        when MatchStreetAndStreetNumberFrench then [$2, $1, :begin]
+        when MatchBrokenStreetAndNumberGerman then [$1, $2, :end]
+        when MatchBrokenStreetAndNumberFrench then [$2, $1, :begin]
+        else [normalized, nil, nil]
+      end
+      name    = normalize_name(name) if normalize
+      number  = normalize_number(number) if normalize
+
+      new(name, number, pos, street)
+    end
+
+    attr_reader :original, :name, :number, :number_position, :full
+
+    def initialize(name, number=nil, number_position=:end, original=nil)
+      @name             = name
+      @number           = number
+      @number_position  = number_position
+      @original         = original
+      @full             = case number_position
+        when :end then [name, number].compact.join(" ")
+        when :begin then [number, name].compact.join(", ")
+        when nil then name.dup
+        else raise ArgumentError, "Invalid value for number_position: #{number_position.inspect}"
+      end
+    end
+
+    def original_or_full
+      @original || @full
+    end
+
+    alias to_s full
+
+    def inspect
+      "#<Street #{self}>"
+    end
+  end
+end

data/lib/swissmatch/street/version.rb

@@ -0,0 +1,15 @@
+# encoding: utf-8
+
+begin
+  require 'rubygems/version' # newer rubygems use this
+rescue LoadError
+  require 'gem/version' # older rubygems use this
+end
+
+module SwissMatch
+  class Street
+
+    # The version of the swissmatch-street gem.
+    Version = Gem::Version.new("0.0.1")
+  end
+end

data/swissmatch-street.gemspec

@@ -0,0 +1,38 @@
+# encoding: utf-8
+
+Gem::Specification.new do |s|
+  s.name                      = "swissmatch-street"
+  s.version                   = "0.0.1"
+  s.authors                   = "Stefan Rusterholz"
+  s.email                     = "stefan.rusterholz@gmail.com"
+  s.homepage                  = "http://github.com/apeiros/swissmatch-street"
+
+  s.description               = <<-DESCRIPTION.gsub(/^    /, '').chomp
+    Parse, tokenize, analyze, repair and handle swiss street names and numbers.
+  DESCRIPTION
+  s.summary                   = <<-SUMMARY.gsub(/^    /, '').chomp
+    Parse, tokenize, analyze, repair and handle swiss street names and numbers.
+  SUMMARY
+
+  s.files                     =
+    Dir['bin/**/*'] +
+    Dir['data/**/*'] +
+    Dir['lib/**/*'] +
+    Dir['rake/**/*'] +
+    Dir['test/**/*'] +
+    Dir['*.gemspec'] +
+    %w[
+      LICENSE.txt
+      Rakefile
+      README.markdown
+    ]
+
+  if File.directory?('bin') then
+    executables = Dir.chdir('bin') { Dir.glob('**/*').select { |f| File.executable?(f) } }
+    s.executables = executables unless executables.empty?
+  end
+
+  s.required_rubygems_version = Gem::Requirement.new("> 1.3.1")
+  s.rubygems_version          = "1.3.1"
+  s.specification_version     = 3
+end

data/test/lib/helper.rb

@@ -0,0 +1,31 @@
+require 'stringio'
+
+module TestSuite
+  attr_accessor :name
+end
+
+module Kernel
+  def suite(name, &block)
+    klass = Class.new(Test::Unit::TestCase, &block)
+    klass.extend TestSuite
+    klass.name = "Suite #{name}"
+
+    klass
+  end
+  module_function :suite
+end
+
+class Test::Unit::TestCase
+  def self.test(desc, &impl)
+    define_method("test #{desc}", &impl)
+  end
+
+  def capture_stdout
+    captured  = StringIO.new
+    $stdout   = captured
+    yield
+    captured.string
+  ensure
+    $stdout = STDOUT
+  end
+end

data/test/runner.rb

@@ -0,0 +1,20 @@
+# run with `ruby test/runner.rb`
+# if you only want to run a single test-file: `ruby test/runner.rb testfile.rb`
+
+$LOAD_PATH << File.expand_path('../../lib', __FILE__)
+$LOAD_PATH << File.expand_path('../../test/lib', __FILE__)
+TEST_DIR    = File.expand_path('../../test', __FILE__)
+
+require 'test/unit'
+require 'helper'
+
+if ENV['COVERAGE']
+  require 'simplecov'
+  SimpleCov.start
+end
+
+units = ARGV.empty? ? Dir["#{TEST_DIR}/unit/**/*.rb"] : ARGV
+
+units.each do |unit|
+  load unit
+end

data/test/unit/lib/swissmatch/street.rb

@@ -0,0 +1,102 @@
+# encoding: utf-8
+
+require 'swissmatch/street'
+include SwissMatch
+
+suite "Street" do
+  test "Street.new with original" do
+    original        = ' Beispielstrasse 15 '
+    street_name     = 'Beispielstrasse'
+    street_number   = '15'
+    number_position = :end
+
+    street          = Street.new(street_name, street_number, number_position, original)
+
+    assert_equal street_name, street.name
+    assert_equal street_number, street.number
+    assert_equal number_position, street.number_position
+    assert_equal original, street.original
+    assert_equal 'Beispielstrasse 15', street.full
+    assert_equal original, street.original_or_full
+  end
+
+  test "Street.new without original" do
+    street_name     = 'Beispielstrasse'
+    street_number   = '15'
+    number_position = :end
+
+    street          = Street.new(street_name, street_number, number_position)
+
+    assert_equal street_name, street.name
+    assert_equal street_number, street.number
+    assert_equal number_position, street.number_position
+    assert_equal nil, street.original
+    assert_equal 'Beispielstrasse 15', street.full
+    assert_equal 'Beispielstrasse 15', street.original_or_full
+  end
+
+  [
+    '12',
+    '12b',
+    '12 B',
+    '12/14',
+    '16/2/22',
+    '105-107',
+    '12/A',
+  ].each do |number|
+    test "House number #{number.inspect}" do
+      assert Street::HouseNumber =~ number
+    end
+  end
+
+  {
+    %{12,Rue Quelquechose}    => %{24, Rue Quelquechose},
+    %{12/345 Foo}             => %{12/345, Foo},
+    %{1A ch. des Choses}      => %{1A, Chemin des Choses},
+    %{ Beispielstr. 15 }      => %{Beispielstrasse 15},
+    %{12b,Rte. d'Anywhere}    => %{25b, Route d'Anywhere},
+    %{1, Av. de Blé}          => %{1, Avenue de Blé},
+  }.each do |original, expected|
+    test "Street.normalize_street #{original.inspect}, false" do
+      actual  = Street.normalize_street(original)
+
+      assert_equal expected, actual
+    end
+  end
+
+  {
+    %{Beispielstrasse 15}         => ['Beispielstrasse',        '15',       :end,   'Beispielstrasse 15'],
+    %{Beispielstrasse 45/4}       => ['Beispielstrasse',        '45/4',     :end,   'Beispielstrasse 45/4'],
+    %{24,Rue Example}             => [%{Rue Example},           '24',       :begin, %{24, Rue Example}],
+    %{70/141 Example}             => [%{Example},               '70/141',   :begin, %{70/141, Example}],
+    %{6A ch. des Cornillons}      => [%{Chemin des Cornillons}, '6a',       :begin, %{6a, Chemin des Cornillons}],
+    %{ Beispielstr. 15 }          => [%{Beispielstrasse},       '15',       :end,   %{Beispielstrasse 15}],
+    %{25b,Rte. d'Yverdon}         => [%{Route d'Yverdon},       '25b',      :begin, %{25b, Route d'Yverdon}],
+    %{6, Av. de Budé}             => [%{Avenue de Budé},        '6',        :begin, %{6, Avenue de Budé}],
+    %{Rue Ferdinand Hodler,19}    => [%{Rue Ferdinand Hodler},  '19',       :end,   %{Rue Ferdinand Hodler 19}],
+    %{Burgstrasse37}              => [%{Burgstrasse},           '37',       :end,   %{Burgstrasse 37}],
+    %{Ave. de Casino 8-10-12}     => [%{Avenue de Casino},      '8-10-12',  :end,   %{Avenue de Casino 8-10-12}],
+    %{Ave Bel- Air 49 B}          => [%{Avenue Bel-Air},        '49b',      :end,   %{Avenue Bel-Air 49b}],
+    %{39 rue Louis Faure}         => [%{Rue Louis Faure},       '39',       :begin, %{39, Rue Louis Faure}],
+    %{Rte d'Alle 13}              => [%{Route d'Alle},          '13',       :end,   %{Route d'Alle 13}],
+    %{Via Filagni, 2/a}           => [%{Via Filagni},           '2/a',      :end,   %{Via Filagni 2/a}],
+    %{Riehenring 189/A}           => [%{Riehenring},            '189a',     :end,   %{Riehenring 189a}],
+    %{Lorraine 12c/9}             => [%{Lorraine},              '12c/9',    :end,   %{Lorraine 12c/9}],
+    %{Lwaldmannstrasse 67 / J2}   => [%{Lwaldmannstrasse},      '189a',     :end,   %{Lwaldmannstrasse 67/j2}],
+    %{Kaysersbergerstrasse 56/3.} => [%{Kaysersbergerstrasse},  '56/3',     :end,   %{Kaysersbergerstrasse 56/3}],
+    %{Rue Montfalcon 2bis}        => [%{Rue Montfalcon},        '2bis',     :end,   %{Rue Montfalcon 2bis}],
+    %{Rue Montfalcon 2 bis}       => [%{Rue Montfalcon},        '2bis',     :end,   %{Rue Montfalcon 2bis}],
+    %{Elsässerstrasse 261-4}      => [%{Elsässerstrasse},       '261-4',    :end,   %{Elsässerstrasse 261-4}],
+  }.each do |original, (street_name, street_number, number_position, full)|
+    test "Street.parse #{original.inspect}" do
+      street  = Street.parse(original, true)
+
+      assert_equal street_name, street.name
+      assert_equal street_number, street.number
+      assert_equal number_position, street.number_position
+      assert_equal original, street.original
+      assert_equal full, street.full
+      assert_equal original, street.original_or_full
+    end
+  end
+end

metadata

@@ -0,0 +1,55 @@
+--- !ruby/object:Gem::Specification
+name: swissmatch-street
+version: !ruby/object:Gem::Version
+  version: 0.0.1
+  prerelease: 
+platform: ruby
+authors:
+- Stefan Rusterholz
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-08-08 00:00:00.000000000 Z
+dependencies: []
+description: Parse, tokenize, analyze, repair and handle swiss street names and numbers.
+email: stefan.rusterholz@gmail.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- data/sty21072012.txt
+- lib/swissmatch/street/version.rb
+- lib/swissmatch/street.rb
+- test/lib/helper.rb
+- test/runner.rb
+- test/unit/lib/swissmatch/street.rb
+- swissmatch-street.gemspec
+- LICENSE.txt
+- Rakefile
+- README.markdown
+homepage: http://github.com/apeiros/swissmatch-street
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>'
+    - !ruby/object:Gem::Version
+      version: 1.3.1
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.24
+signing_key: 
+specification_version: 3
+summary: Parse, tokenize, analyze, repair and handle swiss street names and numbers.
+test_files: []
+has_rdoc: