From 7fa3c93d0b469896d3681318031601c5de3341b8 Mon Sep 17 00:00:00 2001 From: Michael Orlitzky Date: Sat, 26 Sep 2009 20:20:41 -0400 Subject: [PATCH] Fixed the Howcast downloads. Modified the Howcast class to parse an XML file containing the real video URL. Added a test for the new method. --- src/websites/howcast.rb | 29 +- test/fixtures/howcast/81134.xml | 586 ++++++++++++++++++++++++++++++++ test/howcast_test.rb | 16 + 3 files changed, 624 insertions(+), 7 deletions(-) create mode 100644 test/fixtures/howcast/81134.xml diff --git a/src/websites/howcast.rb b/src/websites/howcast.rb index c8dfcd3..58861e7 100644 --- a/src/websites/howcast.rb +++ b/src/websites/howcast.rb @@ -26,8 +26,19 @@ class Howcast < Website return url =~ VALID_HOWCAST_URL_REGEX end + + def get_video_url() + video_id = parse_video_id() + xml_url = "http://www.howcast.com/videos/#{video_id}.xml" + xml_data = self.get_page_data(xml_url) + filepath = parse_file_path_from_xml(xml_data) + + return "http://www.howcast.com#{filepath}" + end - def parse_video_id() + :private + + def parse_video_id() # This regex just pulls out the video id id_regex = /\/(\d+)-/ matches = id_regex.match(@url) @@ -39,12 +50,16 @@ class Howcast < Website return matches[1] end - - def get_video_url() - video_id = parse_video_id() - - return "http://media.howcast.com/system/videos/#{video_id}/#{video_id}.flv" + + def parse_file_path_from_xml(data) + file_path_regex = /(.*?)<\/filename>/ + matches = file_path_regex.match(data) + + if matches.nil? + raise StandardError.new("Couldn't parse the tag from the XML file.") + end + + return matches[1] end - end diff --git a/test/fixtures/howcast/81134.xml b/test/fixtures/howcast/81134.xml new file mode 100644 index 0000000..b5c9282 --- /dev/null +++ b/test/fixtures/howcast/81134.xml @@ -0,0 +1,586 @@ + + + + diff --git a/test/howcast_test.rb b/test/howcast_test.rb index a76c725..aacfcf2 100644 --- a/test/howcast_test.rb +++ b/test/howcast_test.rb @@ -42,7 +42,23 @@ class HowcastTest < Test::Unit::TestCase assert(!Howcast.owns_url?('redtube.com/6807')) end + def test_doesnt_own_misc_urls assert(!Howcast.owns_url?('http://www.howcast.com/abc')) end + + + def test_parse_file_path_from_xml + hc = Howcast.new(nil) + + page_data = nil + + File.open('test/fixtures/howcast/81134.xml') do |f| + page_data = f.read + end + + test_result = hc.send('parse_file_path_from_xml', page_data) + assert_equal('/system/videos/4/34/11/08/81134.flv', test_result) + end + end -- 2.44.2