]> code.delx.au - webdl/blob - sbs.py
Handle URLs without scheme, fixes SBS issue #27
[webdl] / sbs.py
1 #!/usr/bin/env python
2
3 from common import grab_html, grab_json, grab_xml, download_hls, Node
4
5 import json
6
7 BASE = "http://www.sbs.com.au"
8 FULL_VIDEO_LIST = BASE + "/api/video_feed/f/Bgtm9B/sbs-section-programs/?form=json"
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 doc = grab_html(VIDEO_URL % self.video_id, 0)
24 player_params = self.get_player_params(doc)
25 release_url = player_params["releaseUrls"]["html"]
26
27 doc = grab_xml(release_url, 0)
28 video = doc.xpath("//smil:video", namespaces=NS)[0]
29 video_url = video.attrib["src"]
30 if not video_url:
31 raise Exception("Unsupported video %s: %s" % (self.video_id, self.title))
32 filename = self.title + ".ts"
33 return download_hls(filename, video_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
49 class SbsNavNode(Node):
50 def create_video_node(self, entry_data):
51 SbsVideoNode(entry_data["title"], self, entry_data["id"])
52
53 def find_existing_child(self, path):
54 for child in self.children:
55 if child.title == path:
56 return child
57
58 class SbsRootNode(SbsNavNode):
59 def __init__(self, parent):
60 Node.__init__(self, "SBS", parent)
61
62 def fill_children(self):
63 full_video_list = grab_json(FULL_VIDEO_LIST, 3600)
64 category_and_entry_data = self.explode_videos_to_unique_categories(full_video_list)
65 for category_path, entry_data in category_and_entry_data:
66 nav_node = self.create_nav_node(self, category_path)
67 nav_node.create_video_node(entry_data)
68
69 def explode_videos_to_unique_categories(self, full_video_list):
70 for entry_data in full_video_list["entries"]:
71 for category_data in entry_data["media$categories"]:
72 category_path = self.calculate_category_path(
73 category_data["media$scheme"],
74 category_data["media$name"],
75 )
76 if category_path:
77 yield category_path, entry_data
78
79 def calculate_category_path(self, scheme, name):
80 if not scheme:
81 return
82 if scheme == name:
83 return
84 name = name.split("/")
85 if name[0] != scheme:
86 name.insert(0, scheme)
87 return name
88
89 def create_nav_node(self, parent, category_path):
90 if not category_path:
91 return parent
92
93 current_path = category_path[0]
94 current_node = parent.find_existing_child(current_path)
95 if not current_node:
96 current_node = SbsNavNode(current_path, parent)
97 return self.create_nav_node(current_node, category_path[1:])
98
99 def fill_nodes(root_node):
100 SbsRootNode(root_node)