Make blacklist a list, and prevent Wiley duplicates

Author: fxcoudert

config.yml

@@ -7,9 +7,11 @@ throttle: 10
 # Default value: 5
 wait_time: 30
 
-# URLs matching this regular expression will not be tweeted
+# URLs matching these regular expressions will not be tweeted
 # Used e.g. for Wiley journals that republish old content with new URLs
-blacklist: "pericles.pericles-prod.literatumonline.com"
+blacklist:
+ - "pericles.pericles-prod.literatumonline.com"
+ - "www.onlinelibrary.wiley.com"
 
 # Whether to include journal Twitter handles. Not recommended: can lead
 # to the bot being banned for "large numbers of unsolicited mentions".

papersbot.py

@@ -220,9 +220,9 @@ def __init__(self, doTweet=True):
  config = {}
  self.throttle = config.get("throttle", 0)
  self.wait_time = config.get("wait_time", 5)
- self.blacklist = config.get("blacklist", "").strip()
- self.blacklist = re.compile(self.blacklist) if self.blacklist else None
  self.handles = config.get("handles", True)
+ self.blacklist = config.get("blacklist", [])
+ self.blacklist = [re.compile(s) for s in self.blacklist]
 
  # Connect to Twitter, unless requested not to
  if doTweet:
@@ -267,10 +267,11 @@ def sendTweet(self, entry):
  tweet_body = title[:length] + " " + url
 
  # Some URLs may match our blacklist
- if self.blacklist and self.blacklist.search(url):
- print(f"BLACKLISTED: {tweet_body}\n")
- self.addToPosted(entry.id)
- return
+ for regexp in self.blacklist:
+ if regexp.search(url):
+ print(f"BLACKLISTED: {tweet_body}\n")
+ self.addToPosted(entry.id)
+ return
 
  media = None
  image = findImage(entry)