lib/namor/namor.rb in namor-0.1.0 vs lib/namor/namor.rb in namor-0.2.0

@@ -5,16 +5,28 @@
 
   def config(opts)
     @config = opts
   end
 
-  def extract(name)
-    return [] if name.nil?
-
+  # clean up a single name component
+  # * output all converted to uppercase
+  # * strip leading ZZ+ or XX+ (frequently used as invalid-account prefixes)
+  # * remove any words that are in the user-provided suppression list
+  # * remove words from list of common suffixes (Jr, Sr etc)
+  # * remove anything inside parenthesis
+  # * remove punctuation
+  # * squeeze whitespace & trim spaces from ends
+  def scrub(name, opts = {})
     suppression_list = @config[:suppress] || []
-    suppression_re = suppression_list.join('|')
+    suppression_re = (suppression_list + (opts[:suppress]||[])).compact.map(&:upcase).join('|')
 
-    detitled_name = name.upcase.gsub(/\b(#{suppression_re})\b/i, '').gsub(/\b(MD|JR|SR|I+|IV)\b/i, '').gsub(/\([^\(]*\)/, '').gsub(/[_.'-]/, '').gsub(/,\s*$/, '').gsub(/ +/, ' ').strip
+    name && name.upcase.gsub(/^[ZX]{2,}/, '').gsub(/\b(#{suppression_re})\b/i, '').gsub(/\b(JR|SR|II|III|IV)\b/i, '').gsub(/\([^\(]*\)/, '').gsub(/\./, ' ').gsub(/[_'-]/, '').gsub(/,\s*$/, '').gsub(/ +/, ' ').strip
+  end
+
+  def extract(name, opts = {})
+    return [] if name.nil?
+
+    detitled_name = scrub(name, opts)
 
     if detitled_name =~ /,/
       # "last, first[ middle]"
       lastname, firstname = detitled_name.split(/\s*,\s*/)
       lastname.gsub!(/ /, '')