code.delx.au - webdl/blob - sbs.py

   1 #!/usr/bin/env python
   2 # vim:ts=4:sts=4:sw=4:noet
   3
   4 from common import grab_html, grab_json, grab_xml, download_rtmp, download_urllib, Node
   5
   6 import collections
   7
   8 BASE = "http://www.sbs.com.au"
   9 MENU_URL = "/api/video_feed/f/dYtmxB/%s?startIndex=%d"
  10 VIDEO_URL = BASE + "/ondemand/video/single/%s"
  11
  12 NS = {
  13         "smil": "http://www.w3.org/2005/SMIL21/Language",
  14 }
  15
  16 SECTIONS = [
  17         "section-sbstv",
  18         "section-programs",
  19 ]
  20
  21 CATEGORY_MAP = {
  22         "Factual": "Documentary",
  23 }
  24
  25
  26 class SbsNode(Node):
  27         def __init__(self, title, parent, video_id):
  28                 Node.__init__(self, title, parent)
  29                 self.title = title
  30                 self.video_id = video_id.split("/")[-1]
  31                 self.can_download = True
  32
  33         def download(self):
  34                 doc = grab_html(VIDEO_URL % self.video_id, 0)
  35                 desc_url = None
  36                 for script in doc.xpath("//script", namespaces=NS):
  37             if not script.text:
  38                 continue
  39             for line in script.text.split("\n"):
  40                 if line.find("player.releaseUrl") < 0:
  41                     continue
  42                 desc_url = line[line.find("\"")+1 : line.rfind("\"")]
  43                                 break
  44                         if desc_url is not None:
  45                                 break
  46                 if desc_url is None:
  47                         raise Exception("Failed to get JSON URL for " + self.title)
  48
  49                 doc = grab_xml(desc_url, 0)
  50                 best_url = None
  51                 best_bitrate = 0
  52                 for video in doc.xpath("//smil:video", namespaces=NS):
  53                         bitrate = int(video.attrib["system-bitrate"])
  54                         if best_bitrate == 0 or best_bitrate < bitrate:
  55                                 best_bitrate = bitrate
  56                                 best_url = video.attrib["src"]
  57
  58                 ext = best_url.rsplit(".", 1)[1]
  59                 filename = self.title + "." + ext
  60                 best_url += "?v=2.5.14&fp=MAC%2011,1,102,55&r=FLQDD&g=YNANAXRIYFYO"
  61                 return download_urllib(filename, best_url)
  62
  63 def fill_entry(get_catnode, entry):
  64         title = entry["title"]
  65         video_id = entry["id"]
  66         info = collections.defaultdict(list)
  67         for d in entry["media$categories"]:
  68                 if not d.has_key("media$scheme"):
  69                         continue
  70                 info[d["media$scheme"]].append(d["media$name"])
  71
  72         if "Section/Promos" in info.get("Section", []):
  73                 # ignore promos
  74                 return
  75
  76         for category in info.get("Genre", ["$UnknownCategory$"]):
  77                 category = CATEGORY_MAP.get(category, category)
  78                 parent_node = get_catnode(category)
  79                 SbsNode(title, parent_node, video_id)
  80
  81
  82 def fill_section(get_catnode, section):
  83         index = 1
  84         while True:
  85                 doc = grab_json(BASE + MENU_URL % (section, index), 3600)
  86                 if len(doc.get("entries", [])) == 0:
  87                         break
  88                 for entry in doc["entries"]:
  89                         fill_entry(get_catnode, entry)
  90                 index += doc["itemsPerPage"]
  91
  92 class SbsRoot(Node):
  93         def __init__(self, parent=None):
  94                 Node.__init__(self, "SBS", parent)
  95                 self.catnodes = {}
  96
  97         def get_catnode(self, name):
  98                 try:
  99                         return self.catnodes[name]
 100                 except KeyError:
 101                         n = Node(name, self)
 102                         self.catnodes[name] = n
 103                         return n
 104
 105         def fill_children(self):
 106                 for section in SECTIONS:
 107                         fill_section(self.get_catnode, section)
 108
 109 def fill_nodes(root_node):
 110         SbsRoot(root_node)
 111