Changeset 32340 for lang/ruby/video_scraper/lib/www/video_scraper/veoh.rb

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

Files:

• lang/ruby/video_scraper/lib/www/video_scraper/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