seattletimes_rss_renderer.py

   1 import datetime
   2 import generic_news_rss_renderer as gnrss
   3
   4 class seattletimes_rss_renderer(gnrss.generic_news_rss_renderer):
   5     interesting_categories = frozenset([
   6         'Nation',
   7         'World',
   8         'Life',
   9         'Technology'
  10         'Local News',
  11         'Food',
  12         'Drink',
  13         'Today File',
  14         'Seahawks',
  15         'Oddities',
  16         'Packfic NW',
  17         'Home',
  18         'Garden',
  19         'Travel',
  20         'Outdoors',
  21     ])
  22
  23     def __init__(self, name_to_timeout_dict, feed_site, feed_uris, page_title):
  24         super(seattletimes_rss_renderer, self).__init__(
  25             name_to_timeout_dict,
  26             feed_site,
  27             feed_uris,
  28             page_title)
  29
  30     def debug_prefix(self):
  31         return "seattletimes"
  32
  33     def get_headlines_page_prefix(self):
  34         return "seattletimes-nonnews"
  35
  36     def get_details_page_prefix(self):
  37         return "seattletimes-details-nonnews"
  38
  39     def should_use_https(self):
  40         return True
  41
  42     def item_is_interesting_for_headlines(self, title, description, item):
  43         if item.tag != "item":
  44             self.debug_print("Item.tag isn't item?!")
  45             return False
  46         if self.is_item_older_than_n_days(item, 14):
  47             self.debug_print("%s: is too old!" % title)
  48             return False
  49
  50         details = {}
  51         for detail in item.getchildren():
  52             self.debug_print("detail %s => %s (%s)" % (detail.tag,
  53                                                        detail.attrib,
  54                                                        detail.text))
  55             if detail.text != None:
  56                 details[detail.tag] = detail.text
  57         if "category" not in details:
  58             self.debug_print("No category in details?!")
  59             self.debug_print(details)
  60             return False
  61
  62         interesting = False
  63         for x in seattletimes_rss_renderer.interesting_categories:
  64             if x in details["category"]:
  65                 self.debug_print("%s looks like a good category." % x)
  66                 interesting = True
  67         if not interesting:
  68             return False
  69         return True
  70
  71     def item_is_interesting_for_article(self, title, description, item):
  72         if self.is_item_older_than_n_days(item, 14):
  73             self.debug_print("%s: is too old!" % title)
  74             return False
  75         return len(description) >= 65
  76
  77 #x = seattletimes_rss_renderer({"Test", 123},
  78 #                              "www.seattletimes.com",
  79 #                              [ "/life/feed/" ],
  80 #                              "nonnews")
  81 #x.periodic_render("Fetch News")
  82 #x.periodic_render("Shuffle News")