Fixes performance problem decoding large arrays

When an array has many items (a couple thousand) it may take seconds to decode.
Replace the slow regex with the standard array decoder from the 0.19 version of the
PG gem.

Author: betamatt

activerecord-postgres-array.gemspec

@@ -13,10 +13,11 @@ Gem::Specification.new do |s|
   s.rubygems_version = %q{1.3.7}
   s.summary = s.description
 
-  s.add_dependency "activerecord"
+  # The gem is no longer required after Rails 3.2
+  s.add_dependency "activerecord", "<4.0"
+  s.add_dependency "pg", '>= 0.19'
   s.add_development_dependency 'rake'
   s.add_development_dependency 'rspec', '~> 2.12.0'
-  s.add_development_dependency 'pg'
   s.add_development_dependency 'activerecord-postgres-hstore'
   s.add_development_dependency 'combustion', '~> 0.3.1'
 end

lib/activerecord-postgres-array/string.rb

@@ -20,11 +20,9 @@ def from_postgres_array(base_type = :string)
     if empty?
       []
     else
-      elements = match(/\{(.*)\}/m).captures.first.gsub(/\\"/, '$ESCAPED_DOUBLE_QUOTE$').split(/(?:,)(?=(?:[^"]|"[^"]*")*$)/m)
-      elements = elements.map do |e|
-        res = e.gsub('$ESCAPED_DOUBLE_QUOTE$', '"').gsub("\\\\", "\\").gsub(/^"/, '').gsub(/"$/, '').gsub("''", "'").strip
-        res == 'NULL' ? nil : res
-      end
+      # elements = match(/\{(.*)\}/m).captures.first.gsub(/\\"/, '$ESCAPED_DOUBLE_QUOTE$').split(/(?:,)(?=(?:[^"]|"[^"]*")*$)/m)
+      decoder = PG::TextDecoder::Array.new name: "#{base_type}[]", delimiter: ","
+      elements = decoder.decode(self)
 
       if base_type == :decimal
         elements.collect(&:to_d)

spec/string_ext_spec.rb

@@ -99,11 +99,11 @@
     end
 
     it 'correctly handles backslashes' do
-      '\'{"\\\\","\\""}\''.from_postgres_array.should == ["\\","\""]
+      '{"\\\\","\\""}'.from_postgres_array.should == ["\\","\""]
     end
 
     it 'correctly handles multi line content' do
       "{A\nB\nC,X\r\nY\r\nZ}".from_postgres_array.should == ["A\nB\nC", "X\r\nY\r\nZ"]
     end
   end
-end
+end