Changeset 32340 for lang/ruby

Timestamp:

04/13/09 13:11:05 (4 years ago)

Author:

valda

Message:

veoh の URL のバリエーションを追加

Location:

lang/ruby/video_scraper

Files:

• lib/www/video_scraper/veoh.rb (modified) (2 diffs)
• test/www/video_scraper/test_veoh.rb (modified) (2 diffs)

lang/ruby/video_scraper/lib/www/video_scraper/veoh.rb

@@ -6 +6 @@
   module VideoScraper
     class Veoh < Base
-      url_regex [%r!\Ahttp://www\.veoh\.com/videos/([[:alnum:]]+)!,
-                 %r!\Ahttp://www\.veoh\.com/collection/\w+/watch/.*#watch%3[Dd]([[:alnum:]]+)!,
-                 %r!\Ahttp://www\.veoh\.com/collection/\w+/watch/([[:alnum:]]+)!]
+      url_regex [%r!\Ahttp://www\.veoh\.com/videos/(v\d+[[:alnum:]]+)!,
+                 %r!\Ahttp://www\.veoh\.com/collection/\w+/watch/.*#watch%3[Dd](v\d+[[:alnum:]]+)!,
+                 %r!\Ahttp://www\.veoh\.com/(?:browse|collection)/(?:[\w]+/)+watch/(v\d+[[:alnum:]]+)!]
 
       def scrape
@@ -19 +19 @@
         @thumb_url = xml.match(/fullMedResImagePath="([^"]+)"/).to_a[1]
         html = http_get(@page_url)
-        if embed_tag = html.match(/\sid="embed"\s[^>]*value="([^"]+)"/).to_a[1]
+        #logger.debug html
+        if embed_tag = html.match(/class="embedinput"\s[^>]*value="([^"]+)"/).to_a[1]
           @embed_tag = CGI.unescapeHTML(embed_tag)
         end

lang/ruby/video_scraper/test/www/video_scraper/test_veoh.rb

@@ -9 +9 @@
     assert_match %r|http://content\.veoh\.com/flash/p/\d/[[:alnum:]]{16}/[[:alnum:]]{40}\.fll\?ct=[[:alnum:]]{48}|, vs.video_url
     assert_match %r|http://p-images\.veoh\.com/image\.out\?imageId=media-[[:alnum:]]+.jpg|, vs.thumb_url
-    assert_match %r|^<embed\s.*>$|, vs.embed_tag
+    assert_match %r|^<object\s.*>$|, vs.embed_tag
   end
 
@@ -19 +19 @@
     vs = WWW::VideoScraper::Veoh.scrape('http://www.veoh.com/videos/v19937773gwSJPMk?rank=0&jsonParams=%7B%22numResults%22%3A20%2C%22rlmin%22%3A0%2C%22query%22%3A%22Shaman+King+01%22%2C%22rlmax%22%3Anull%2C%22veohOnly%22%3Atrue%2C%22order%22%3A%22default%22%2C%22range%22%3A%22a%22%2C%22sId%22%3A%22192998624295114150%22%7D&searchId=192998624295114150&rank=1', default_opt)
     assert_equal 'http://www.veoh.com/videos/v19937773gwSJPMk', vs.page_url
+    vs = WWW::VideoScraper::Veoh.scrape('http://www.veoh.com/browse/videos/category/comedy/watch/v17078605sszQzbBF')
+    assert_equal 'http://www.veoh.com/videos/v17078605sszQzbBF', vs.page_url
   end
 end