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

   1 import requests_cache
   2 from common import grab_html, grab_json, grab_xml, download_hls, Node, append_to_qs
   3
   4 import json
   5
   6 BASE = "http://www.sbs.com.au"
   7 FULL_VIDEO_LIST = BASE + "/api/video_search/v2/?m=1&filters={section}{Programs}"
   8 VIDEO_URL = BASE + "/ondemand/video/single/%s"
   9
  10 NS = {
  11     "smil": "http://www.w3.org/2005/SMIL21/Language",
  12 }
  13
  14
  15 class SbsVideoNode(Node):
  16     def __init__(self, title, parent, url):
  17         Node.__init__(self, title, parent)
  18         self.video_id = url.split("/")[-1]
  19         self.can_download = True
  20
  21     def download(self):
  22         with requests_cache.disabled():
  23             doc = grab_html(VIDEO_URL % self.video_id)
  24         player_params = self.get_player_params(doc)
  25         release_url = player_params["releaseUrls"]["html"]
  26
  27         video_url = self.get_video_url(release_url)
  28         if not video_url:
  29             raise Exception("Unsupported video %s: %s" % (self.video_id, self.title))
  30         filename = self.title + ".ts"
  31         return download_hls(filename, video_url)
  32
  33     def get_player_params(self, doc):
  34         for script in doc.xpath("//script"):
  35             if not script.text:
  36                 continue
  37             for line in script.text.split("\n"):
  38                 s = "var playerParams = {"
  39                 if s in line:
  40                     p1 = line.find(s) + len(s) - 1
  41                     p2 = line.find("};", p1) + 1
  42                     if p1 >= 0 and p2 > 0:
  43                         return json.loads(line[p1:p2])
  44         raise Exception("Unable to find player params for %s: %s" % (self.video_id, self.title))
  45
  46     def get_video_url(self, release_url):
  47         with requests_cache.disabled():
  48             doc = grab_xml(release_url if not release_url.startswith("//") else "http:" + release_url)
  49             video = doc.xpath("//smil:video", namespaces=NS)
  50             if not video:
  51                 return
  52             video_url = video[0].attrib["src"]
  53             return video_url
  54
  55 class SbsNavNode(Node):
  56     def create_video_node(self, entry_data):
  57         SbsVideoNode(entry_data["title"], self, entry_data["id"])
  58
  59     def find_existing_child(self, path):
  60         for child in self.children:
  61             if child.title == path:
  62                 return child
  63
  64 class SbsRootNode(SbsNavNode):
  65     def __init__(self, parent):
  66         Node.__init__(self, "SBS", parent)
  67
  68     def fill_children(self):
  69         all_video_entries = self.load_all_video_entries()
  70         category_and_entry_data = self.explode_videos_to_unique_categories(all_video_entries)
  71         for category_path, entry_data in category_and_entry_data:
  72             nav_node = self.create_nav_node(self, category_path)
  73             nav_node.create_video_node(entry_data)
  74
  75     def load_all_video_entries(self):
  76         offset = 1
  77         amount = 500
  78         while True:
  79             url = append_to_qs(FULL_VIDEO_LIST, {"range": "%s-%s" % (offset, offset+amount)})
  80             data = grab_json(url)
  81             if "entries" not in data:
  82                 raise Exception("Missing data in SBS response", data)
  83             entries = data["entries"]
  84             if len(entries) == 0:
  85                 break
  86             for entry in entries:
  87                 yield entry
  88             offset += amount
  89
  90     def explode_videos_to_unique_categories(self, all_video_entries):
  91         for entry_data in all_video_entries:
  92             for category_data in entry_data["media$categories"]:
  93                 category_path = self.calculate_category_path(
  94                     category_data["media$scheme"],
  95                     category_data["media$name"],
  96                 )
  97                 if category_path:
  98                     yield category_path, entry_data
  99
 100     def calculate_category_path(self, scheme, name):
 101         if not scheme:
 102             return
 103         if scheme == name:
 104             return
 105         name = name.split("/")
 106         if name[0] != scheme:
 107             name.insert(0, scheme)
 108         return name
 109
 110     def create_nav_node(self, parent, category_path):
 111         if not category_path:
 112             return parent
 113
 114         current_path = category_path[0]
 115         current_node = parent.find_existing_child(current_path)
 116         if not current_node:
 117             current_node = SbsNavNode(current_path, parent)
 118         return self.create_nav_node(current_node, category_path[1:])
 119
 120 def fill_nodes(root_node):
 121     SbsRootNode(root_node)