code.delx.au - webdl/blob - plus7.py

   1 #!/usr/bin/env python
   2 # vim:ts=4:sts=4:sw=4:noet
   3
   4 import json
   5 from lxml.cssselect import CSSSelector
   6
   7 from common import grab_html, download_rtmp, Node
   8
   9 METADATA = "http://cosmos.bcst.yahoo.com/rest/v2/pops;id=%d;lmsoverride=1;element=stream;bw=1200"
  10 BASE = "http://au.tv.yahoo.com"
  11 BROWSE = BASE + "/plus7/browse/"
  12 HASH_URL = "http://d.yimg.com/nl/australia/au-tv/player.swf"
  13 HASH_URL = "http://d.yimg.com/m/up/ypp/au/player.swf"
  14
  15 def extract_and_remove(tokens, key):
  16         lowertokens = [x.lower() for x in tokens]
  17         pos = lowertokens.index(key)
  18
  19         value = int(tokens[pos+1])
  20         tokens = tokens[:pos] + tokens[pos+2:]
  21
  22         return value, tokens
  23
  24
  25 def demangle_title(title):
  26         # Postman Pat - Postman Pat and the Runaway Kite Series 1 Episode 1
  27         tokens = title.split()
  28
  29         try:
  30                 season, tokens = extract_and_remove(tokens, "series")
  31                 episode, tokens = extract_and_remove(tokens, "episode")
  32         except ValueError:
  33                 return title
  34
  35         try:
  36                 i = tokens.index("-") + 1
  37         except ValueError:
  38                 i = 0
  39         tokens.insert(i, "%sx%s - " % (season, str(episode).zfill(2)))
  40
  41         return " ".join(tokens)
  42
  43 class Plus7Node(Node):
  44         def __init__(self, title, parent, url):
  45                 Node.__init__(self, demangle_title(title), parent)
  46                 self.url = url
  47                 self.can_download = True
  48
  49         def get_vid(self):
  50                 doc = grab_html(self.url, 3600)
  51                 for script in doc.xpath("//script"):
  52                         if not script.text:
  53                                 continue
  54                         for line in script.text.split("\n"):
  55                                 if line.find("vid : ") <= 0:
  56                                         continue
  57                                 vid = line[line.find("'")+1 : line.rfind("'")]
  58                                 vid = int(vid)
  59                                 return vid
  60                 raise Exception("Could not find vid on page " + self.url)
  61
  62         def download(self):
  63                 vid = self.get_vid()
  64                 doc = grab_html(METADATA % vid, 0)
  65                 content = doc.xpath("//content")[0]
  66                 vbase = content.attrib["url"]
  67                 vpath = content.attrib["path"]
  68                 filename = self.title + ".flv"
  69                 return download_rtmp(filename, vbase, vpath, HASH_URL)
  70
  71
  72 class Plus7Series(Node):
  73         def __init__(self, title, parent, url):
  74                 Node.__init__(self, title, parent)
  75                 self.url = url
  76
  77         def fill_children(self):
  78                 doc = grab_html(self.url, 3600)
  79                 for item in CSSSelector("#related-episodes div.itemdetails")(doc):
  80                         title = CSSSelector("span.title")(item)[0].text
  81                         subtitle = CSSSelector("span.subtitle")(item)[0].xpath("string()")
  82                         if subtitle and subtitle.strip():
  83                                 title += " - " + subtitle.strip().replace("  ", " ")
  84                         url = CSSSelector("a")(item)[0].attrib["href"]
  85                         Plus7Node(title, self, BASE + url)
  86
  87 class Plus7Root(Node):
  88         def __init__(self, parent=None):
  89                 Node.__init__(self, "Yahoo Plus7", parent)
  90
  91         def fill_children(self):
  92                 doc = grab_html(BROWSE, 3600)
  93                 shows = []
  94                 for script in doc.xpath("//script"):
  95                         if not script.text or not script.text.startswith("var shows = "):
  96                                 continue
  97                         shows = script.text[12:]
  98                         shows = shows.rstrip("; \n")
  99                         shows = json.loads(shows)
 100                 for show in shows:
 101                         Plus7Series(show["title"], self, show["url"])
 102
 103 def fill_nodes(root_node):
 104         Plus7Root(root_node)
 105