implement ICapVideoProvider.get_video()

weboob/backends/youporn/backend.py

@@ -44,11 +44,15 @@ class YoupornBackend(Backend, ICapVideoProvider):
     def need_url(func):
         def inner(self, *args, **kwargs):
             url = args[0]
-            if u'youporn.com' not in url:
+            if isinstance(url, (str,unicode)) and not url.isdigit() and u'youporn.com' not in url:
                 return None
             return func(self, *args, **kwargs)
         return inner
 
+    @need_url
+    def get_video(self, _id):
+        return self.browser.get_video(_id)
+
     @need_url
     def iter_page_urls(self, mozaic_url):
         raise NotImplementedError()

weboob/backends/youporn/browser.py

@@ -36,10 +36,20 @@ class YoupornBrowser(Browser):
         # Disallow arguments
         Browser.__init__(self)
 
-    def get_video_title(self, page_url):
+    def id2url(self, _id):
-        self.location(page_url)
+        if isinstance(_id, int) or isinstance(_id, (str,unicode)) and _id.isdigit():
-        return self.page.title
+            return 'http://www.youporn.com/watch/%d' % _id
+        else:
+            return str(_id)
 
-    def get_video_url(self, page_url):
+    def get_video(self, _id):
-        self.location(page_url)
+        self.location(self.id2url(_id))
-        return self.page.url
+        return self.page.video
+
+    def get_video_title(self, _id):
+        self.location(self.id2url(_id))
+        return self.page.video.title
+
+    def get_video_url(self, _id):
+        self.location(self.id2url(_id))
+        return self.page.video.url

weboob/backends/youporn/pages/video.py

@@ -18,19 +18,75 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 
 """
 
+import re
+import datetime
+from logging import warning
+
 from .base import PornPage
+from weboob.capabilities.video import Video
 
 class VideoPage(PornPage):
+    URL_REGEXP = re.compile("https?://[w\.]*youporn.com/watch/(\d+)/?.*")
+
     def loaded(self):
         if not PornPage.loaded(self):
             return
 
+        self.video = Video(self.get_id(),
+                           self.get_title(),
+                           self.get_url())
+
+        self.set_details(self.video)
+
+    def get_id(self):
+        m = self.URL_REGEXP.match(self.url)
+        if m:
+            return int(m.group(1))
+        warning("Unable to parse ID")
+        return 0
+
+    def get_url(self):
         el = self.document.getroot().cssselect('div[id=download]')
         if el:
-            self.url = el[0].cssselect('a')[0].attrib['href']
+            return el[0].cssselect('a')[0].attrib['href']
-        else:
-            self.url = None
 
+    def get_title(self):
         el = self.document.getroot().cssselect('h1')
         if el:
-            self.title = unicode(el[0].getchildren()[0].tail).strip()
+            return unicode(el[0].getchildren()[0].tail).strip()
+
+    DATE_REGEXP = re.compile("\w+ (\w+) (\d+) (\d+):(\d+):(\d+) (\d+)")
+    MONTH2I = ['', 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
+
+    def set_details(self, v):
+        div = self.document.getroot().cssselect('div[id=details]')
+        if not div:
+            return
+
+        for li in div[0].getiterator('li'):
+            span = li.find('span')
+            name = span.text.strip()
+            value = span.tail.strip()
+
+            if name == 'Duration:':
+                duration = 0
+                for word in value.split():
+                    if word.endswith('min'):
+                        duration += 60 * int(word[:word.find('min')])
+                    elif word.endswith('sec'):
+                        duration += int(word[:word.find('sec')])
+                v.duration = duration
+            elif name == 'Submitted:':
+                v.author = li.find('i').text
+            elif name == 'Rating:':
+                v.rating = float(value[:value.find(' ')])
+            elif name == 'Date:':
+                m = self.DATE_REGEXP.match(value)
+                if m:
+                    month = self.MONTH2I.index(m.group(1))
+                    day = int(m.group(2))
+                    hour = int(m.group(3))
+                    minute = int(m.group(4))
+                    second = int(m.group(5))
+                    year = int(m.group(6))
+                    v.date = datetime.datetime(year, month, day, hour, minute, second)