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 import sys
   6
   7 BASE = "https://www.sbs.com.au"
   8 FULL_VIDEO_LIST = BASE + "/api/video_feed/f/Bgtm9B/sbs-section-programs/"
   9 VIDEO_URL = BASE + "/ondemand/video/single/%s"
  10
  11 NS = {
  12     "smil": "http://www.w3.org/2005/SMIL21/Language",
  13 }
  14
  15
  16 class SbsVideoNode(Node):
  17     def __init__(self, title, parent, url):
  18         Node.__init__(self, title, parent)
  19         self.video_id = url.split("/")[-1]
  20         self.can_download = True
  21
  22     def download(self):
  23         with requests_cache.disabled():
  24             doc = grab_html(VIDEO_URL % self.video_id)
  25         player_params = self.get_player_params(doc)
  26
  27         error = player_params.get("error", None)
  28         if error:
  29             print("Cannot download:", error)
  30             return False
  31
  32         release_url = player_params["releaseUrls"]["html"]
  33         filename = self.title + ".ts"
  34
  35         hls_url = self.get_hls_url(release_url)
  36         if hls_url:
  37             return download_hls(filename, hls_url)
  38         else:
  39             return download_mpd(filename, release_url)
  40
  41     def get_player_params(self, doc):
  42         for script in doc.xpath("//script"):
  43             if not script.text:
  44                 continue
  45             for line in script.text.split("\n"):
  46                 s = "var playerParams = {"
  47                 if s in line:
  48                     p1 = line.find(s) + len(s) - 1
  49                     p2 = line.find("};", p1) + 1
  50                     if p1 >= 0 and p2 > 0:
  51                         return json.loads(line[p1:p2])
  52         raise Exception("Unable to find player params for %s: %s" % (self.video_id, self.title))
  53
  54     def get_hls_url(self, release_url):
  55         with requests_cache.disabled():
  56             doc = grab_xml("https:" + release_url.replace("http:", "").replace("https:", ""))
  57             video = doc.xpath("//smil:video", namespaces=NS)
  58             if not video:
  59                 return
  60             video_url = video[0].attrib["src"]
  61             return video_url
  62
  63 class SbsNavNode(Node):
  64     def create_video_node(self, entry_data):
  65         SbsVideoNode(entry_data["title"], self, entry_data["id"])
  66
  67     def find_existing_child(self, path):
  68         for child in self.children:
  69             if child.title == path:
  70                 return child
  71
  72 class SbsRootNode(SbsNavNode):
  73     def __init__(self, parent):
  74         Node.__init__(self, "SBS", parent)
  75
  76     def fill_children(self):
  77         all_video_entries = self.load_all_video_entries()
  78         category_and_entry_data = self.explode_videos_to_unique_categories(all_video_entries)
  79         for category_path, entry_data in category_and_entry_data:
  80             nav_node = self.create_nav_node(self, category_path)
  81             nav_node.create_video_node(entry_data)
  82
  83     def load_all_video_entries(self):
  84         offset = 1
  85         amount = 1000
  86         uniq = set()
  87         while True:
  88             url = append_to_qs(FULL_VIDEO_LIST, {"range": "%s-%s" % (offset, offset+amount-1)})
  89             data = grab_json(url)
  90             if "entries" not in data:
  91                 raise Exception("Missing data in SBS response", data)
  92             entries = data["entries"]
  93             if len(entries) == 0:
  94                 break
  95             for i, entry in enumerate(entries):
  96                 if entry["guid"] not in uniq:
  97                     uniq.add(entry["guid"])
  98                     yield entry
  99             offset += amount
 100             sys.stdout.write(".")
 101             sys.stdout.flush()
 102         print()
 103
 104     def explode_videos_to_unique_categories(self, all_video_entries):
 105         for entry_data in all_video_entries:
 106             for category_data in entry_data["media$categories"]:
 107                 category_path = self.calculate_category_path(
 108                     category_data["media$scheme"],
 109                     category_data["media$name"],
 110                 )
 111                 if category_path:
 112                     yield category_path, entry_data
 113
 114     def calculate_category_path(self, scheme, name):
 115         if not scheme:
 116             return
 117         if scheme == name:
 118             return
 119         name = name.split("/")
 120         if name[0] != scheme:
 121             name.insert(0, scheme)
 122         return name
 123
 124     def create_nav_node(self, parent, category_path):
 125         if not category_path:
 126             return parent
 127
 128         current_path = category_path[0]
 129         current_node = parent.find_existing_child(current_path)
 130         if not current_node:
 131             current_node = SbsNavNode(current_path, parent)
 132         return self.create_nav_node(current_node, category_path[1:])
 133
 134 def fill_nodes(root_node):
 135     SbsRootNode(root_node)