@@ -39,12 +39,7 @@ parser = HansardParser.new(people) # First check that there isn't already a patch file patch_file_path = "#{File.dirname(__FILE__)}/data/patches/#{house}.#{date}.xml.patch" -if File.exists?(patch_file_path) - puts "Patch file #{patch_file_path} already exists..." - exit -end - -File.open("original.xml", "w") {|f| f << parser.hansard_xml_source_data_on_date(date, house)} -FileUtils.cp("original.xml", "patched.xml") +File.open("original.xml", "w") {|f| f << parser.unpatched_hansard_xml_source_data_on_date(date, house)} +File.open("patched.xml", "w") {|f| f << parser.hansard_xml_source_data_on_date(date, house)} system("mate --wait patched.xml") system("diff -u original.xml patched.xml > #{patch_file_path}") create_patch.rb @@ -50,8 +50,9 @@ class HansardParser end # Returns the XML file loaded from aph.gov.au as plain text which contains all the Hansard data - # Returns nil if it doesn't exist - def hansard_xml_source_data_on_date(date, house) + # Returns nil it it doesn't exist + # This is the original data without any patches applied at this end + def unpatched_hansard_xml_source_data_on_date(date, house) agent = MechanizeProxy.new agent.cache_subdirectory = cache_subdirectory(date, house) @@ -67,15 +68,22 @@ class HansardParser @logger.error "Link to XML download is missing" nil else - text = agent.click(link).body - # Now check whether there is a patch for that day and if so apply it - patch_file_path = "#{File.dirname(__FILE__)}/../data/patches/#{house}.#{date}.xml.patch" - if File.exists?(patch_file_path) - puts "Using patch file: #{patch_file_path}" - Patch::patch(text, File.read(patch_file_path)) - else - text - end + agent.click(link).body + end + end + end + + # Returns the XML file loaded from aph.gov.au as plain text which contains all the Hansard data + # Returns nil if it doesn't exist + def hansard_xml_source_data_on_date(date, house) + text = unpatched_hansard_xml_source_data_on_date(date, house) + if text + # Now check whether there is a patch for that day and if so apply it + patch_file_path = "#{File.dirname(__FILE__)}/../data/patches/#{house}.#{date}.xml.patch" + if File.exists?(patch_file_path) + Patch::patch(text, File.read(patch_file_path)) + else + text end end end lib/hansard_parser.rb 510cf8fd6c091e25d7a24710be6d5adb71dd675a Matthew Landauer matthew@openaustralia.org http://github.com/mlandauer/openaustralia-parser/commit/45c3356a06cacb4e78393879c2f45453fdc62ad1 45c3356a06cacb4e78393879c2f45453fdc62ad1 2009-04-10T19:31:32-07:00 2009-04-10T19:28:02-07:00 create_patch script will now take into account pre existing patch 549d2cb6020f7ac58bceabd807507d582edb31d7 Matthew Landauer matthew@openaustralia.org