]> code.delx.au - webdl/blob - sbs.py
Disable Channel 9 because it's been broken for a while
[webdl] / 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 with requests_cache.disabled():
28 doc = grab_xml(release_url if not release_url.startswith("//") else "http:" + release_url)
29 video = doc.xpath("//smil:video", namespaces=NS)[0]
30 video_url = video.attrib["src"]
31 if not video_url:
32 raise Exception("Unsupported video %s: %s" % (self.video_id, self.title))
33 filename = self.title + ".ts"
34 return download_hls(filename, video_url)
35
36 def get_player_params(self, doc):
37 for script in doc.xpath("//script"):
38 if not script.text:
39 continue
40 for line in script.text.split("\n"):
41 s = "var playerParams = {"
42 if s in line:
43 p1 = line.find(s) + len(s) - 1
44 p2 = line.find("};", p1) + 1
45 if p1 >= 0 and p2 > 0:
46 return json.loads(line[p1:p2])
47 raise Exception("Unable to find player params for %s: %s" % (self.video_id, self.title))
48
49
50 class SbsNavNode(Node):
51 def create_video_node(self, entry_data):
52 SbsVideoNode(entry_data["title"], self, entry_data["id"])
53
54 def find_existing_child(self, path):
55 for child in self.children:
56 if child.title == path:
57 return child
58
59 class SbsRootNode(SbsNavNode):
60 def __init__(self, parent):
61 Node.__init__(self, "SBS", parent)
62
63 def fill_children(self):
64 all_video_entries = self.load_all_video_entries()
65 category_and_entry_data = self.explode_videos_to_unique_categories(all_video_entries)
66 for category_path, entry_data in category_and_entry_data:
67 nav_node = self.create_nav_node(self, category_path)
68 nav_node.create_video_node(entry_data)
69
70 def load_all_video_entries(self):
71 offset = 1
72 amount = 500
73 while True:
74 url = append_to_qs(FULL_VIDEO_LIST, {"range": "%s-%s" % (offset, offset+amount)})
75 data = grab_json(url)
76 if "entries" not in data:
77 raise Exception("Missing data in SBS response", data)
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)