+ videourl = sorted(playlist, key=lambda k: int(k['BANDWIDTH']))[-1]['url']
+ segments, metadata = parse_segment_playlist(requests.get(videourl).text)
+ if "EXT-X-KEY" in metadata:
+ key = requests.get(metadata["EXT-X-KEY"]['URI'].strip('"')).text
+ decrypt=True
+ else:
+ decrypt=False
+ with open("%s"%video['filename'],"wb") as ofile:
+ segment=0
+ size = 0
+ for url in segments:
+ ufile = requests.get(url, stream=True).raw
+ print("\r{} MB".format(size/1024/1024))
+ sys.stdout.flush()
+ if decrypt:
+ iv=struct.pack("IIII",segment,0,0,0)
+ decryptor = AES.new(key, AES.MODE_CBC, iv)
+ while(True):
+ buf = ufile.read(4096)
+ if not buf:
+ break
+ if decrypt:
+ buf = decryptor.decrypt(buf)
+ ofile.write(buf)
+ size += len(buf)
+ segment += 1
+
+def parse_playlist(playlist):
+ if not playlist.startswith("#EXTM3U"):
+ print(playlist)
+ return False
+ playlist = playlist.splitlines()[1:]
+ items=[]
+ for (metadata_string,url) in zip(playlist[0::2], playlist[1::2]):
+ md = dict()
+ assert 'EXT-X-STREAM-INF' in metadata_string.split(':')[0]
+ for item in metadata_string.split(':')[1].split(','):
+ if '=' in item:
+ md.update([item.split('='),])
+ md['url']=url
+ items.append(md)
+ return items
+
+def parse_segment_playlist(playlist):
+ assert playlist.startswith("#EXTM3U")
+ PATTERN = re.compile(r'''((?:[^,"']|"[^"]*"|'[^']*')+)''')
+ segments = []
+ next_is_url=False
+ metadata = {}
+ for row in playlist.splitlines():
+ if next_is_url:
+ segments.append(row)
+ next_is_url=False
+ continue
+ if 'EXTINF' in row:
+ next_is_url=True
+ if "EXT-X-KEY" in row:
+ row = row.split(':',1)[1] #skip first part
+ parts = PATTERN.split(row)[1:-1] #do magic re split and keep quotes
+ metadata["EXT-X-KEY"] = dict([part.split('=',1) for part in parts if '=' in part]) #throw away the commas and make dict of the pairs
+ return(segments, metadata)
+
+def parse_videolist():
+ page_num = 1
+ soup = BeautifulSoup(requests.get("http://www.svtplay.se/ajax/videospager").text)#this call does not work for getting the pages, we use it for the page totals only
+ page_tot = int(soup.find('a',{'data-currentpage':True}).attrs['data-lastpage'])
+ videos_per_page = 8
+ video_num = 0
+ while(page_num <= page_tot):
+ base_url = "http://www.svtplay.se/ajax/videos?sida={}".format(page_num)
+ soup = BeautifulSoup(requests.get(base_url).text)
+ for article in soup.findAll('article'):
+ meta = dict(article.attrs)
+ video = {}
+ video['title'] = meta['data-title']
+ video['description'] = meta['data-description']
+ video['url'] = dict(article.find('a').attrs)['href']
+ video['thumb-url'] = dict(article.find('img',{}).attrs)['src']
+ video['num'] = video_num
+ video['total'] = page_tot * videos_per_page
+ video_num += 1
+ yield video
+ page_num += 1
+