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_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 + "/api/video_feed/f/dYtmxB/CxeOeDELXKEv/%s?form=json"
  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_json(VIDEO_URL % self.video_id, 0)
  35                 best_url = None
  36                 best_bitrate = 0
  37                 for d in doc["media$content"]:
  38                         bitrate = d["plfile$bitrate"]
  39                         if bitrate > best_bitrate or best_url is None:
  40                                 best_bitrate = bitrate
  41                                 best_url = d["plfile$url"]
  42
  43                 doc = grab_xml(best_url, 3600)
  44                 if doc.xpath("//smil:meta/@base", namespaces=NS):
  45                         vbase = doc.xpath("//smil:meta/@base", namespaces=NS)[0]
  46                         vpath = doc.xpath("//smil:video/@src", namespaces=NS)[0]
  47                         ext = vpath.rsplit(".", 1)[1]
  48                         filename = self.title + "." + ext
  49                         return download_rtmp(filename, vbase, vpath)
  50                 else:
  51                         from lxml import etree
  52                         url = doc.xpath("//smil:video/@src", namespaces=NS)[0]
  53                         ext = url.rsplit(".", 1)[1]
  54                         filename = self.title + "." + ext
  55                         url += "?v=2.5.14&fp=MAC%2011,1,102,55&r=FLQDD&g=YNANAXRIYFYO"
  56                         return download_urllib(filename, url)
  57
  58 def fill_entry(get_catnode, entry):
  59         title = entry["title"]
  60         video_id = entry["id"]
  61         info = collections.defaultdict(list)
  62         for d in entry["media$categories"]:
  63                 if not d.has_key("media$scheme"):
  64                         continue
  65                 info[d["media$scheme"]].append(d["media$name"])
  66
  67         if "Section/Promos" in info.get("Section", []):
  68                 # ignore promos
  69                 return
  70
  71         for category in info.get("Genre", ["$UnknownCategory$"]):
  72                 category = CATEGORY_MAP.get(category, category)
  73                 parent_node = get_catnode(category)
  74                 SbsNode(title, parent_node, video_id)
  75
  76
  77 def fill_section(get_catnode, section):
  78         index = 1
  79         while True:
  80                 doc = grab_json(BASE + MENU_URL % (section, index), 3600)
  81                 if len(doc.get("entries", [])) == 0:
  82                         break
  83                 for entry in doc["entries"]:
  84                         fill_entry(get_catnode, entry)
  85                 index += doc["itemsPerPage"]
  86
  87 class SbsRoot(Node):
  88         def __init__(self, title, parent=None):
  89                 Node.__init__(self, title, parent)
  90                 self.catnodes = {}
  91
  92         def get_catnode(self, name):
  93                 try:
  94                         return self.catnodes[name]
  95                 except KeyError:
  96                         n = Node(name, self)
  97                         self.catnodes[name] = n
  98                         return n
  99
 100         def get_children(self):
 101                 if self.children:
 102                         return self.children
 103                 for section in SECTIONS:
 104                         fill_section(self.get_catnode, section)
 105                 return self.children
 106
 107 def fill_nodes(root_node):
 108         SbsRoot("SBS", root_node)
 109