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

   1 import requests_cache
   2 from common import grab_html, grab_json, grab_xml, download_hls, download_mpd, 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         filename = self.title + ".ts"
  28
  29         hls_url = self.get_hls_url(release_url)
  30         if hls_url:
  31             return download_hls(filename, hls_url)
  32         else:
  33             return download_mpd(filename, release_url)
  34
  35     def get_player_params(self, doc):
  36         for script in doc.xpath("//script"):
  37             if not script.text:
  38                 continue
  39             for line in script.text.split("\n"):
  40                 s = "var playerParams = {"
  41                 if s in line:
  42                     p1 = line.find(s) + len(s) - 1
  43                     p2 = line.find("};", p1) + 1
  44                     if p1 >= 0 and p2 > 0:
  45                         return json.loads(line[p1:p2])
  46         raise Exception("Unable to find player params for %s: %s" % (self.video_id, self.title))
  47
  48     def get_hls_url(self, release_url):
  49         with requests_cache.disabled():
  50             doc = grab_xml("http:" + release_url.replace("http:", "").replace("https:", ""))
  51             video = doc.xpath("//smil:video", namespaces=NS)
  52             if not video:
  53                 return
  54             video_url = video[0].attrib["src"]
  55             return video_url
  56
  57 class SbsNavNode(Node):
  58     def create_video_node(self, entry_data):
  59         SbsVideoNode(entry_data["title"], self, entry_data["id"])
  60
  61     def find_existing_child(self, path):
  62         for child in self.children:
  63             if child.title == path:
  64                 return child
  65
  66 class SbsRootNode(SbsNavNode):
  67     def __init__(self, parent):
  68         Node.__init__(self, "SBS", parent)
  69
  70     def fill_children(self):
  71         all_video_entries = self.load_all_video_entries()
  72         category_and_entry_data = self.explode_videos_to_unique_categories(all_video_entries)
  73         for category_path, entry_data in category_and_entry_data:
  74             nav_node = self.create_nav_node(self, category_path)
  75             nav_node.create_video_node(entry_data)
  76
  77     def load_all_video_entries(self):
  78         offset = 1
  79         amount = 500
  80         while True:
  81             url = append_to_qs(FULL_VIDEO_LIST, {"range": "%s-%s" % (offset, offset+amount)})
  82             data = grab_json(url)
  83             if "entries" not in data:
  84                 raise Exception("Missing data in SBS response", data)
  85             entries = data["entries"]
  86             if len(entries) == 0:
  87                 break
  88             for entry in entries:
  89                 yield entry
  90             offset += amount
  91
  92     def explode_videos_to_unique_categories(self, all_video_entries):
  93         for entry_data in all_video_entries:
  94             for category_data in entry_data["media$categories"]:
  95                 category_path = self.calculate_category_path(
  96                     category_data["media$scheme"],
  97                     category_data["media$name"],
  98                 )
  99                 if category_path:
 100                     yield category_path, entry_data
 101
 102     def calculate_category_path(self, scheme, name):
 103         if not scheme:
 104             return
 105         if scheme == name:
 106             return
 107         name = name.split("/")
 108         if name[0] != scheme:
 109             name.insert(0, scheme)
 110         return name
 111
 112     def create_nav_node(self, parent, category_path):
 113         if not category_path:
 114             return parent
 115
 116         current_path = category_path[0]
 117         current_node = parent.find_existing_child(current_path)
 118         if not current_node:
 119             current_node = SbsNavNode(current_path, parent)
 120         return self.create_nav_node(current_node, category_path[1:])
 121
 122 def fill_nodes(root_node):
 123     SbsRootNode(root_node)