class Doko

Public Class Methods

deep(str,base_uri=nil) click to toggle source
# File lib/doko.rb, line 11
def self.deep(str,base_uri=nil)
  addrs = parse(str)
  if addrs.empty?
    addrs = links(str,base_uri).map{ |u|
      parse(u)
    }.flatten
  end
  addrs
end
new(str) click to toggle source
# File lib/doko.rb, line 38
def initialize(str)
  if str.kind_of? URI 
    str = open(str.to_s).read
  elsif str.match( /^#{URI.regexp}$/ )
    str = open(str).read
  end
  if str.match(/<html/i)
    @text =  (Nokogiri::HTML(str)/"body").text
  else
    @text = str
  end
end
parse(str) click to toggle source
# File lib/doko.rb, line 7
def self.parse(str)
  self.new(str).parse
end

Public Instance Methods

parse() click to toggle source
# File lib/doko.rb, line 51
def parse
  body = @text
  body.tr!("0-9","0-9")
  body.tr!("()","()")
  body.tr!("、",",")
  body.tr!(" "," ")
  body.tr!(".",".")
  body.tr!(":",":")
  blackchars =  ",()\n"

  addrs = body.scan(/\b([^\s,()]{2,3}(都|道|府|県)[^\s,()]{1,8}(市|区|町|村)[^#{blackchars}]+)/).map{ |m|
    clean(m[0])
  }
  if addrs.empty?
    addrs = body.scan(/([^\s]{1,6}(市|区).{1,8}(区|町|村)[^\s,()]{2,10}\d)/).map{ |m|
      clean(m[0])
    }
  end
  addrs.select{ |a|
    !a.match(/を/)
  }
end

Private Instance Methods

clean(line) click to toggle source
# File lib/doko.rb, line 77
def clean(line)
  line.gsub!(/住所(\s|\n)?/,"")
  line.gsub!(/〒\d{3}-\d{4} ?/,"")
  line.gsub!(/\s+$/,"")
  line.gsub!(/\s?電話:.+$/,"")
  line.gsub!("[MAP]","")
  line.gsub!(/(TEL|FAX):\d{2,4}-\d{2,4}-\d{2,4}.+/,"")
  line.gsub!(/(\dー)*\d/) do |t|
    t.tr("ー","-")
  end
  line.sub!(/\s$/,"")
  line.gsub!(/\s{3,}+.+$/,"")
  line
end