dburkes-people_places_things 1.1.0 → 1.1.2

data/lib/people_places_things/person_name/person_name.rb

@@ -0,0 +1,157 @@
+class PersonName
+  attr_accessor :parts
+  attr_accessor :raw
+
+  def initialize(options={})
+    @parts = options.inject({}) do |parts, (k, v)| 
+      sym_key = (k.to_sym rescue key) || key
+      parts[sym_key] = v if SUPPORTED_PARTS.include?(sym_key)
+      parts
+    end
+  end
+
+  def self.parse(string, fmt = :auto_detect)
+    raise "Unsupported Format" if !PARSE_FORMATS.include?(fmt)
+    
+    if fmt == :auto_detect
+      fmt = string.include?(',') ? :last_first_middle : :first_middle_last
+    end
+    
+    args = {}
+
+    parts = string.split(/\s|,/).collect {|p| p.strip}.reject {|p| PersonName.blank?(p) || p == ',' }
+  
+    if parts.size == 1
+      args[:last] = parts.first
+    else
+      case fmt
+        when :first_middle_last
+          if parts.size > 2 and SUPPORTED_SUFFIXES.detect {|s| s.casecmp(parts.last) == 0}
+            args[:suffix] = normalize_suffix(parts.last)
+            parts.delete_at(parts.size - 1)
+          end
+        
+          args[:first] = parts.first if parts.size > 0
+          args[:last] = parts.last if parts.size > 1
+        
+          if parts.size > 2 && ODD_LAST_NAME_PREFIXES.detect {|s| s.casecmp(parts[-2]) == 0}
+            args[:last] = "#{parts[-2]}#{args[:last]}"
+            parts.delete_at(parts.size - 2)
+          end
+            
+          args[:middle] = parts[1..(parts.size - 2)].join(' ') if parts.size > 2
+    
+        when :last_first_middle
+          args[:last] = parts.first if parts.size > 0
+        
+          if parts.size > 1 && ODD_LAST_NAME_PREFIXES.detect {|s| s.casecmp(args[:last]) == 0}
+            args[:last] << parts[1]
+            parts.delete_at(1)
+          end
+
+          if parts.size > 2 and SUPPORTED_SUFFIXES.detect {|s| s.casecmp(parts[1]) == 0}
+            args[:suffix] = normalize_suffix(parts[1])
+            parts.delete_at(1)
+          end
+
+          args[:first] = parts[1] if parts.size > 1
+          args[:middle] = parts[2..(parts.size - 1)].join(' ') if parts.size > 2
+      end
+    end
+  
+    n = PersonName.new args
+    n.raw = string
+    n
+  end
+  
+  def self.format(options, fmt=:full)
+    PersonName.new(options).to_s(fmt)
+  end
+
+  def [](which)
+    @parts[which]
+  end
+
+  def to_s(fmt = :full)
+    raise "Unsupported Format" if !OUTPUT_FORMATS.include?(fmt)
+
+    case fmt
+      when :first, :middle, :last
+        @parts[fmt]
+      
+      when :full
+        [@parts[:first], @parts[:middle], @parts[:last], @parts[:suffix]].compact.join(' ')
+        
+      when :full_reverse
+        [@parts[:last], @parts[:first], @parts[:middle], @parts[:suffix]].compact.join(' ')
+      
+      when :first_space_last
+        [@parts[:first], @parts[:last]].compact.join(' ')
+      
+      when :last_space_first
+        [@parts[:last], @parts[:first]].compact.join(' ')
+      
+      when :last_comma_first
+        [@parts[:last], @parts[:first]].compact.join(',')
+      
+      when :last_comma_space_first
+        [(@parts[:first] ? "#{@parts[:last]}," : @parts[:last]), @parts[:first]].compact.join(' ')
+    end
+  end
+
+  def first
+    @parts[:first]
+  end
+
+  def first_i
+    first[0,1] rescue nil
+  end
+
+  def middle
+    @parts[:middle]
+  end
+
+  def middle_i
+    middle[0,1] rescue nil
+  end
+
+  def last
+    @parts[:last]
+  end
+
+  def last_i
+    last[0,1] rescue nil
+  end
+  
+  def suffix
+    @parts[:suffix]
+  end
+  
+  def eql?(other, initials_only=false)
+    if other.is_a?(PersonName)
+      (self.parts.keys & other.parts.keys).all? do |k|
+        msg = (k != :last && initials_only) ? "#{k}_i" : k
+        self.send(msg).casecmp(other.send(msg)) == 0
+      end
+    else
+      false
+    end
+  end
+  
+  PARSE_FORMATS = [:first_middle_last, :last_first_middle, :auto_detect]
+  OUTPUT_FORMATS = [:first, :middle, :last, :full, :full_reverse, :first_space_last, :last_space_first, :last_comma_first, :last_comma_space_first]
+
+  private
+
+  def self.blank?(string_or_nil)
+    string_or_nil.nil? || string_or_nil !~ /\S/
+  end
+  
+  def self.normalize_suffix(suffix)
+    suffix.match(/\w+/)[0] rescue suffix
+  end
+  
+  SUPPORTED_PARTS = [:first,:middle,:last, :suffix]
+  SUPPORTED_SUFFIXES = %w(II III IV V JR JR. SR SR.)
+  ODD_LAST_NAME_PREFIXES = %w(MC ST ST.)
+end

data/lib/people_places_things/phone_number/phone_number.rb

@@ -0,0 +1,38 @@
+class PhoneNumber
+  @@SUPPORTED_PARTS = [:country_code, :area_code, :number, :exchange, :suffix]
+  @@SUPPORTED_PARTS.each {|attr| attr_accessor attr}
+  attr_accessor :raw
+  
+  def initialize(parts={})
+    parts.keys.each do |k|
+      send("#{k}=", parts[k]) if @@SUPPORTED_PARTS.include?(k)
+    end
+  end
+  
+  def self.parse(string)
+    extract = string.match(/.*?([-()\d ]+)/)[1].gsub(/[^\d]/, '') rescue nil
+    raise "Unsupported Format" if !extract || extract.length < 10 || extract.length > 11
+
+    parts = {}
+    
+    if extract.length == 11
+      parts[:country_code] = extract.slice!(0..0)
+    else
+      parts[:country_code] = '1'
+    end
+    
+    raise "Unsupported Format" if parts[:country_code] != '1'
+    
+    parts[:area_code] = extract.slice!(0..2)
+    
+    parts[:number] = extract.dup
+
+    parts[:exchange] = extract.slice!(0..2)
+    
+    parts[:suffix] = extract
+    
+    ret = PhoneNumber.new parts
+    ret.raw = string
+    ret
+  end
+end

data/lib/people_places_things/street_address/street_address.rb

@@ -0,0 +1,195 @@
+class StreetAddress
+  @@SUPPORTED_PARTS = [:number, :pre_direction, :name, :suffix, :post_direction, :unit_type, :unit]
+  @@SUPPORTED_PARTS.each {|attr| attr_accessor attr}
+  attr_accessor :raw
+  
+  def initialize(parts={})
+    parts.keys.each do |k|
+      send("#{k}=", parts[k]) if @@SUPPORTED_PARTS.include?(k)
+    end
+    
+    validate_parts
+  end
+  
+  def self.parse(string)
+    tokens = string.split(/[\s,]/).select {|s| !s.empty?}
+    parts = {}
+    
+    # Check the first token for leading numericality.  If so, set number to the first token, and delete it
+    #
+    if tokens.first =~ /(^\d+.*)/
+      parts[:number] = $1
+      tokens.shift
+    end
+    
+    # If at least two tokens remain, check next-to-last token as unit type.  If so, set unit_type and unit, and delete the tokens
+    #
+    if tokens.size > 1 
+      parts[:unit_type] = find_token(tokens[-2], UNIT_TYPES)
+      if parts[:unit_type]
+        parts[:unit] = tokens[-1]
+        tokens.slice!(tokens.size - 2, 2)
+      end
+    end
+    
+    # If at least one token remains, check last token for directionality.  If so, set post_direction and delete the token
+    #
+    if tokens.size > 0
+      parts[:post_direction] = find_token(tokens[-1], DIRECTIONS)
+      if parts[:post_direction]
+        post_direction_token = tokens[-1]
+        tokens.slice!(tokens.size - 1)
+      end
+    end
+    
+    # If at least one token remains, check last token for suffix.  If so, self set.suffix and delete the token
+    #
+    if tokens.size > 0
+      parts[:suffix] = find_token(tokens[-1], SUFFIXES)
+      tokens.slice!(tokens.size - 1) if parts[:suffix]
+    end
+    
+    # If at least two tokens remain, check first for directionality. If so, set pre_direction and delete token
+    #
+    if tokens.size > 1
+      parts[:pre_direction] = find_token(tokens.first, DIRECTIONS)
+      tokens.shift if parts[:pre_direction]
+    end
+    
+    # if any tokens remain, set joined remaining tokens as name, otherwise, set name to post_direction, if set, and set post_direction to nil
+    #
+    if tokens.size > 0
+      parts[:name] = tokens.join(' ')
+    else
+      parts[:name] = post_direction_token
+      parts[:post_direction] = nil
+    end
+    
+    ret = StreetAddress.new parts
+    ret.raw = string
+    ret
+  end
+  
+  def self.format(parts)
+    StreetAddress.new(parts).to_s
+  end
+  
+  def to_s
+    parts = []
+    parts << self.number if self.number
+    parts << DIRECTIONS[self.pre_direction].first if self.pre_direction
+    parts << self.name if self.name
+    parts << SUFFIXES[self.suffix].first if self.suffix
+    parts << DIRECTIONS[self.post_direction].first if self.post_direction
+    parts << UNIT_TYPES[self.unit_type].first if self.unit_type
+    parts << self.unit if self.unit
+    parts.join(' ')
+  end
+  
+  def self.string_for(symbol, form)
+    raise "Requested unknown form \"#{type}\" for :#{symbol}" if !SUPPORTED_FORMS.include?(form)
+    
+    val = DIRECTIONS[symbol] || SUFFIXES[symbol] || UNIT_TYPES[symbol]
+
+    if val
+      val = ((val[SUPPORTED_FORMS.index(form)] rescue nil) || (val.first rescue val))
+    end
+    
+    val
+  end
+  
+  private
+  
+  def validate_parts
+    [:pre_direction, :suffix, :post_direction, :unit_type].each do |p|
+      if self.send(p)
+        legal_values = p == :suffix ? SUFFIXES : p == :unit_type ? UNIT_TYPES : DIRECTIONS
+        raise "Invalid #{p.to_s} \"#{self.send(p)}\"" if !legal_values.include?(self.send(p))
+      end
+    end
+  end
+  
+  def self.find_token(token, values)
+    values.keys.each do |k|
+      return k if values[k].detect {|v| v.casecmp(token) == 0}
+    end
+    
+    nil
+  end
+  
+  DIRECTIONS = {
+    :north => %w(north n n.),
+    :northeast => %w(northeast ne ne. n.e.),
+    :east => %w(east e e.),
+    :southeast => %w(southeast se se. s.e.),
+    :south => %w(south s s.),
+    :southwest => %w(southwest sw sw. s.w.),
+    :west => %w(west w w.),
+    :northwest => %w(northwest nw nw. n.w.)
+  }
+  
+  SUFFIXES = {
+  	:alley => %w(alley al al.),
+  	:avenue => %w(avenue ave ave. av av.),
+  	:beach => %w(beach bch bch.),
+  	:bend => %w(bend),
+  	:boulevard => %w(boulevard blvd blvd. blv blv.),
+  	:center => %w(center ctr ctr.),
+  	:circle => %w(circle cir cir.),
+  	:cliff => %w(cliff clf clf.),
+  	:club => %w(club),
+  	:condo => %w(condo con con.),
+  	:court => %w(court ct ct. cor cor.),
+  	:cove => %w(cove),
+  	:creek => %w(creek crk crk.),
+  	:crossing => %w(crossing xing xing. crs crs.),
+  	:drive => %w(drive dr dr.),
+  	:extension => %w(extension ext ext.),
+  	:freeway => %w(freeway fwy fwy.),
+  	:gardens => %w(gardens gdns gdns.),
+  	:glen => %w(glen gl gl.),
+  	:green => %w(green grn grn.),
+  	:heights => %w(heights hts hts.),
+  	:highway => %w(highway hwy hwy. hgwy hgwy.),
+  	:hill => %w(hill),
+  	:knoll => %w(knoll knl knl.),
+  	:lake => %w(lake),
+  	:lane => %w(lane ln ln.),
+  	:landing => %w(landing lndg lndg.),
+  	:loop => %w(loop),
+  	:meadows => %w(meadows mdws mdws.),
+  	:manor => %w(manor mnr mnr.),
+  	:mountain => %w(mountain mtn mtn. mnt mnt.),
+  	:oaks => %w(oaks),
+  	:oval => %w(oval),
+  	:park => %w(park pk pk. prk prk.),
+  	:parkway => %w(parkway pkwy pkwy. pky pky.),
+  	:pier => %w(pier),
+  	:place => %w(place pl pl.),
+  	:plaza => %w(plaza plz plz.),
+  	:point => %w(point pt pt. pnt pnt.),
+  	:ridge => %w(ridge ri ri.),
+  	:road => %w(road rd rd.),
+  	:row => %w(row),
+  	:run => %w(run),
+  	:springs => %w(springs spgs spgs.),
+  	:square => %w(square sq sq.),
+  	:street => %w(street st st.),
+  	:station => %w(station sta sta.),
+  	:terrace => %w(terrace ter ter. te te.),
+  	:turnpike => %w(turnpike tpke tpke.),
+  	:trace => %w(trace trc trc.),
+  	:trail => %w(trail trl trl. tl tl.),
+  	:valley => %w(valley vly vly.),
+  	:walk => %w(walk),
+  	:way => %w(way)
+  }
+  
+  UNIT_TYPES = {
+    :suite => %w(suite ste ste.),
+    :number => %w(number # nbr nbr.),
+    :apartment => %w(apartment apt apt.)
+  }
+  
+  SUPPORTED_FORMS = [:long, :short]
+end

data/lib/people_places_things.rb

@@ -0,0 +1,8 @@
+require File.join(File.dirname(__FILE__), 'people_places_things', 'street_address', 'street_address')
+require File.join(File.dirname(__FILE__), 'people_places_things', 'person_name', 'person_name')
+require File.join(File.dirname(__FILE__), 'people_places_things', 'ansi_counties', 'ansi_counties')
+require File.join(File.dirname(__FILE__), 'people_places_things', 'phone_number', 'phone_number')
+
+module PeoplePlacesThings
+  VERSION = '1.1.2'
+end

data/people_places_things.gemspec

@@ -0,0 +1,61 @@
+# Generated by jeweler
+# DO NOT EDIT THIS FILE
+# Instead, edit Jeweler::Tasks in Rakefile, and run `rake gemspec`
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{people_places_things}
+  s.version = "1.1.2"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Danny Burkes"]
+  s.date = %q{2009-08-28}
+  s.description = %q{Parsers and formatters for person names, street addresses, city/state/zip, phone numbers, etc.}
+  s.email = %q{dburkes@netable.com}
+  s.extra_rdoc_files = [
+    "README.rdoc"
+  ]
+  s.files = [
+    ".gitignore",
+     "README.rdoc",
+     "Rakefile",
+     "VERSION",
+     "lib/people_places_things.rb",
+     "lib/people_places_things/ansi_counties/ansi_counties.rb",
+     "lib/people_places_things/ansi_counties/data/data.yml",
+     "lib/people_places_things/ansi_counties/data/process_data.rb",
+     "lib/people_places_things/ansi_counties/data/raw.txt",
+     "lib/people_places_things/person_name/person_name.rb",
+     "lib/people_places_things/phone_number/phone_number.rb",
+     "lib/people_places_things/street_address/street_address.rb",
+     "people_places_things.gemspec",
+     "spec/ansi_counties_spec.rb",
+     "spec/helper.rb",
+     "spec/person_name_spec.rb",
+     "spec/phone_number_spec.rb",
+     "spec/street_address_spec.rb"
+  ]
+  s.has_rdoc = true
+  s.homepage = %q{http://github.com/dburkes/people_places_things}
+  s.rdoc_options = ["--charset=UTF-8"]
+  s.require_paths = ["lib"]
+  s.rubygems_version = %q{1.3.1}
+  s.summary = %q{Parsers and formatters for person names, street addresses, city/state/zip, phone numbers, etc.}
+  s.test_files = [
+    "spec/ansi_counties_spec.rb",
+     "spec/helper.rb",
+     "spec/person_name_spec.rb",
+     "spec/phone_number_spec.rb",
+     "spec/street_address_spec.rb"
+  ]
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 2
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+    else
+    end
+  else
+  end
+end

data/spec/phone_number_spec.rb

@@ -0,0 +1,39 @@
+require 'spec/helper'
+
+describe PhoneNumber do
+  it "should parse ten digits" do
+    phone = PhoneNumber.parse '4045551212'
+    phone.area_code.should == '404'
+    phone.number.should == '5551212'
+    phone.exchange.should == '555'
+    phone.suffix.should == '1212'
+  end
+
+  it "should parse eleven digits" do
+    phone = PhoneNumber.parse '14045551212'
+    phone.area_code.should == '404'
+    phone.number.should == '5551212'
+    phone.exchange.should == '555'
+    phone.suffix.should == '1212'
+  end
+  
+  it "should ignore certain characters" do
+    phone = PhoneNumber.parse '1 (404) 555-1212'
+    phone.area_code.should == '404'
+    phone.number.should == '5551212'
+    phone.exchange.should == '555'
+    phone.suffix.should == '1212'
+  end
+  
+  it "should support international format, at least for US numbers, for now" do
+    phone = PhoneNumber.parse '+1 404 555-1212'
+    phone.area_code.should == '404'
+    phone.number.should == '5551212'
+    phone.exchange.should == '555'
+    phone.suffix.should == '1212'
+  end
+  
+  it "should throw exception on unsupported format" do
+    lambda { PhoneNumber.parse('40455512') }.should raise_error
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: dburkes-people_places_things
 version: !ruby/object:Gem::Version 
-  version: 1.1.0
+  version: 1.1.2
 platform: ruby
 authors: 
 - Danny Burkes
@@ -22,9 +22,27 @@ extensions: []
 extra_rdoc_files: 
 - README.rdoc
 files: 
+- .gitignore
 - README.rdoc
+- Rakefile
+- VERSION
+- lib/people_places_things.rb
+- lib/people_places_things/ansi_counties/ansi_counties.rb
+- lib/people_places_things/ansi_counties/data/data.yml
+- lib/people_places_things/ansi_counties/data/process_data.rb
+- lib/people_places_things/ansi_counties/data/raw.txt
+- lib/people_places_things/person_name/person_name.rb
+- lib/people_places_things/phone_number/phone_number.rb
+- lib/people_places_things/street_address/street_address.rb
+- people_places_things.gemspec
+- spec/ansi_counties_spec.rb
+- spec/helper.rb
+- spec/person_name_spec.rb
+- spec/phone_number_spec.rb
+- spec/street_address_spec.rb
 has_rdoc: true
 homepage: http://github.com/dburkes/people_places_things
+licenses: 
 post_install_message: 
 rdoc_options: 
 - --charset=UTF-8
@@ -45,12 +63,13 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.2.0
+rubygems_version: 1.3.5
 signing_key: 
 specification_version: 2
 summary: Parsers and formatters for person names, street addresses, city/state/zip, phone numbers, etc.
 test_files: 
-- spec/person_name_spec.rb
-- spec/street_address_spec.rb
 - spec/ansi_counties_spec.rb
 - spec/helper.rb
+- spec/person_name_spec.rb
+- spec/phone_number_spec.rb
+- spec/street_address_spec.rb