bellevue_reporter_rss_renderer.py

   1 #!/usr/bin/env python3
   2
   3 import re
   4 from typing import List, Dict
   5 import xml
   6
   7 import generic_news_rss_renderer as gnrss
   8
   9
  10 class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
  11     """Read the Bellevue Reporter's RSS feed."""
  12
  13     def __init__(
  14         self,
  15         name_to_timeout_dict: Dict[str, int],
  16         feed_site: str,
  17         feed_uris: List[str],
  18         page_title: str,
  19     ):
  20         super(bellevue_reporter_rss_renderer, self).__init__(
  21             name_to_timeout_dict, feed_site, feed_uris, page_title
  22         )
  23         self.debug = True
  24
  25     def debug_prefix(self) -> str:
  26         return "bellevue_reporter(%s)" % (self.page_title)
  27
  28     def get_headlines_page_prefix(self) -> str:
  29         return "bellevue-reporter"
  30
  31     def get_details_page_prefix(self) -> str:
  32         return "bellevue-reporter-details"
  33
  34     def should_use_https(self) -> bool:
  35         return True
  36
  37     def munge_description(self, description: str) -> str:
  38         description = re.sub("<[^>]+>", "", description)
  39         description = re.sub(
  40             "Bellevue\s+Reporter\s+Bellevue\s+Reporter", "", description
  41         )
  42         description = re.sub("\s*\-\s*Your local homepage\.\s*", "", description)
  43         return description
  44
  45     @staticmethod
  46     def looks_like_football(title: str, description: str) -> bool:
  47         return (
  48             title.find("NFL") != -1
  49             or re.search("[Ll]ive [Ss]tream", title) != None
  50             or re.search("[Ll]ive[Ss]tream", title) != None
  51             or re.search("[Ll]ive [Ss]tream", description) != None
  52         )
  53
  54     def item_is_interesting_for_headlines(
  55         self, title: str, description: str, item: xml.etree.ElementTree.Element
  56     ) -> bool:
  57         if self.is_item_older_than_n_days(item, 10):
  58             self.debug_print("%s: is too old!" % title)
  59             return False
  60         if bellevue_reporter_rss_renderer.looks_like_football(title, description):
  61             self.debug_print("%s: looks like it's about football." % title)
  62             return False
  63         return True
  64
  65     def item_is_interesting_for_article(
  66         self, title: str, description: str, item: xml.etree.ElementTree.Element
  67     ) -> bool:
  68         if self.is_item_older_than_n_days(item, 10):
  69             self.debug_print("%s: is too old!" % title)
  70             return False
  71         if bellevue_reporter_rss_renderer.looks_like_football(title, description):
  72             self.debug_print("%s: looks like it's about football." % title)
  73             return False
  74         return True
  75
  76
  77 # Test
  78 # x = bellevue_reporter_rss_renderer(
  79 #    {"Fetch News" : 1,
  80 #     "Shuffle News" : 1},
  81 #    "www.bellevuereporter.com",
  82 #    [ "/feed/" ],
  83 #    "Test" )
  84 # d = """
  85 # <DIV style="padding:8px;
  86 #     font-size:44pt;
  87 #     -webkit-column-break-inside:avoid;"><P>
  88 # <B>Task force will tackle issues of racial justice, police reform</B>
  89 # <BR>Bellevue Reporter
  90 # Bellevue Reporter - Your local homepage.
  91 # 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
  92 # Wire Service
  93 # </DIV>"""
  94 # d = x.munge_description(d)
  95 # print(d)
  96 # if x.fetch_news() == 0:
  97 #    print("Error fetching news, no items fetched.")
  98 # x.shuffle_news()