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

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