bellevue_reporter_rss_renderer.py

   1 import generic_news_rss_renderer as gnrss
   2 import re
   3
   4
   5 class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
   6     def __init__(self, name_to_timeout_dict, feed_site, feed_uris, page_title):
   7         super(bellevue_reporter_rss_renderer, self).__init__(
   8             name_to_timeout_dict, feed_site, feed_uris, page_title
   9         )
  10         self.debug = 1
  11
  12     def debug_prefix(self):
  13         return "bellevue_reporter(%s)" % (self.page_title)
  14
  15     def get_headlines_page_prefix(self):
  16         return "bellevue-reporter"
  17
  18     def get_details_page_prefix(self):
  19         return "bellevue-reporter-details"
  20
  21     def should_use_https(self):
  22         return True
  23
  24     def munge_description(self, description):
  25         description = re.sub("<[^>]+>", "", description)
  26         description = re.sub(
  27             "Bellevue\s+Reporter\s+Bellevue\s+Reporter", "", description
  28         )
  29         description = re.sub("\s*\-\s*Your local homepage\.\s*", "", description)
  30         return description
  31
  32     def item_is_interesting_for_headlines(self, title, description, item):
  33         if self.is_item_older_than_n_days(item, 10):
  34             self.debug_print("%s: is too old!" % title)
  35             return False
  36         if (
  37             title.find("NFL") != -1
  38             or re.search("[Ll]ive [Ss]tream", title) != None
  39             or re.search("[Ll]ive[Ss]tream", title) != None
  40             or re.search("[Ll]ive [Ss]tream", description) != None
  41         ):
  42             self.debug_print("%s: looks like it's about football." % title)
  43             return False
  44         return True
  45
  46     def item_is_interesting_for_article(self, title, description, item):
  47         if self.is_item_older_than_n_days(item, 10):
  48             self.debug_print("%s: is too old!" % title)
  49             return False
  50         if (
  51             title.find(" NFL") != -1
  52             or re.search("[Ll]ive [Ss]tream", title) != None
  53             or re.search("[Ll]ive[Ss]tream", title) != None
  54             or re.search("[Ll]ive [Ss]tream", description) != None
  55         ):
  56             self.debug_print("%s: looks like it's about football." % title)
  57             return False
  58         return True
  59
  60
  61 # Test
  62 # x = bellevue_reporter_rss_renderer(
  63 #    {"Fetch News" : 1,
  64 #     "Shuffle News" : 1},
  65 #    "www.bellevuereporter.com",
  66 #    [ "/feed/" ],
  67 #    "Test" )
  68 # d = """
  69 # <DIV style="padding:8px;
  70 #     font-size:44pt;
  71 #     -webkit-column-break-inside:avoid;"><P>
  72 # <B>Task force will tackle issues of racial justice, police reform</B>
  73 # <BR>Bellevue Reporter
  74 # Bellevue Reporter - Your local homepage.
  75 # Inslee names civil rights activists, pastors, and cops to panel that may forge ideas f#or new laws Task force will tackle issues of racial justice, police reform
  76 # Wire Service
  77 # </DIV>"""
  78 # d = x.munge_description(d)
  79 # print d
  80 # if x.fetch_news() == 0:
  81 #    print "Error fetching news, no items fetched."
  82 # x.shuffle_news()