All you fuckers.

[kiosk.git] / bellevue_reporter_rss_renderer.py

diff --git a/bellevue_reporter_rss_renderer.py b/bellevue_reporter_rss_renderer.py
index b8fd27bafed6e9fe97b4562d6fb044615e69c7e7..fec70e00d83cb62022f526c90515e35ad8b44165 100644 (file)
--- a/bellevue_reporter_rss_renderer.py
+++ b/bellevue_reporter_rss_renderer.py
@@ -1,5 +1,6 @@
 #!/usr/bin/env python3
 
+import logging
 import re
 from typing import List, Dict
 import xml
@@ -8,6 +9,9 @@ import xml.etree.ElementTree as ET
 import generic_news_rss_renderer as gnrss
 
 
+logger = logging.getLogger(__name__)
+
+
 class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
     """Read the Bellevue Reporter's RSS feed."""
 
@@ -18,13 +22,7 @@ class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
         feed_uris: List[str],
         page_title: str,
     ):
-        super(bellevue_reporter_rss_renderer, self).__init__(
-            name_to_timeout_dict, feed_site, feed_uris, page_title
-        )
-        self.debug = True
-
-    def debug_prefix(self) -> str:
-        return f"bellevue_reporter({self.page_title})"
+        super().__init__(name_to_timeout_dict, feed_site, feed_uris, page_title)
 
     def get_headlines_page_prefix(self) -> str:
         return "bellevue-reporter"
@@ -60,31 +58,45 @@ class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
     @staticmethod
     def looks_like_spam(title: str, description: str) -> bool:
         return (
-            'marketplace' in description
-            or 'national-marketplace' in description
-            or re.search('[Ww]eed', title) is not None
-            or re.search('[Cc]annabis', title) is not None
-            or re.search('[Cc]annabis', description) is not None
-            or 'THC' in title
-            or re.search('[Ll]ose [Ww]eight', title) is not None
-            or re.search('[Ll]ose [Ww]eight', description) is not None
+            description is not None
+            and title is not None
+            and (
+                "marketplace" in description
+                or "national-marketplace" in description
+                or re.search("[Ww]eed", title) is not None
+                or re.search("[Cc]annabis", title) is not None
+                or re.search("[Cc]annabis", description) is not None
+                or "THC" in title
+                or re.search("[Tt]op.[Rr]ated", title) is not None
+                or re.search("[Ll]ose [Ww]eight", title) is not None
+                or re.search("[Ll]ose [Ww]eight", description) is not None
+            )
         )
 
+    @staticmethod
+    def looks_very_boring(title: str, description: str) -> bool:
+        return description.lower() in title.lower()
+
     def item_is_interesting_for_headlines(
         self, title: str, description: str, item: xml.etree.ElementTree.Element
     ) -> bool:
         unfiltered_description = item.findtext("description")
         if self.is_item_older_than_n_days(item, 10):
-            self.debug_print(f'{title}: is too old!')
+            logger.info(f"{title}: is too old!")
             return False
-        if bellevue_reporter_rss_renderer.looks_like_spam(title, unfiltered_description):
-            self.debug_print(f'{title}: looks like spam')
+        if bellevue_reporter_rss_renderer.looks_like_spam(
+            title, unfiltered_description
+        ):
+            logger.debug(f"{title}: looks like spam")
             return False
         if bellevue_reporter_rss_renderer.looks_like_football(title, description):
-            self.debug_print(f'{title}: looks like it\'s about football.')
+            logger.debug(f"{title}: looks like it's about football.")
             return False
         if bellevue_reporter_rss_renderer.looks_like_review(title, description):
-            self.debug_print(f'{title}: looks like a review.')
+            logger.debug(f"{title}: looks like a review.")
+            return False
+        if bellevue_reporter_rss_renderer.looks_very_boring(title, description):
+            logger.debug(f"{title}: looks very boring.")
             return False
         return True
 
@@ -93,16 +105,18 @@ class bellevue_reporter_rss_renderer(gnrss.generic_news_rss_renderer):
     ) -> bool:
         unfiltered_description = item.findtext("description")
         if self.is_item_older_than_n_days(item, 10):
-            self.debug_print(f'{title}: is too old!')
+            logger.debug(f"{title}: is too old!")
             return False
-        if bellevue_reporter_rss_renderer.looks_like_spam(title, unfiltered_description):
-            self.debug_print(f'{title}: looks like spam')
+        if bellevue_reporter_rss_renderer.looks_like_spam(
+            title, unfiltered_description
+        ):
+            logger.debug(f"{title}: looks like spam")
             return False
         if bellevue_reporter_rss_renderer.looks_like_football(title, description):
-            self.debug_print(f'{title}: looks like it\'s about football.')
+            logger.debug(f"{title}: looks like it's about football.")
             return False
         if bellevue_reporter_rss_renderer.looks_like_review(title, description):
-            self.debug_print(f'{title}: looks like a review.')
+            logger.debug(f"{title}: looks like a review.")
             return False
         return True