From: Matthew Vernon <matthew@rapun.sel.cam.ac.uk>
Date: Fri, 4 Jun 2010 15:35:46 +0000 (+0100)
Subject: revised url-tracking now ready to go (still commented out, though)
X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~matthewv/git?a=commitdiff_plain;h=c0b84406e088cfa837f29e7ea44e3ee87c7a158e;p=irc.git

revised url-tracking now ready to go (still commented out, though)
---

diff --git a/Servus-chiark.py b/Servus-chiark.py
index 3fcdc01..37cccf9 100644
--- a/Servus-chiark.py
+++ b/Servus-chiark.py
@@ -141,7 +141,12 @@ class game:
 	losetime=time.time()+300000
 
 # initialise the urldb on startup
-c.urldb={}
+urldb={}
+lastexp=time.time()
+#expire urls if not asked about or seen for >71 hours
+expirelen=71*60*60
+#do an expiry run every hour
+expirevery=60*60
 
 # Command processing: whenever something is said that the bot can hear,
 # "command" is invoked and must decide what to do.  This configuration
@@ -163,7 +168,7 @@ commands = {"karma": (c.karmaq,karmadb),
 	    "die": quit,
 	    "define": c.defineq,
             "google": c.googleq,
-#	    "url": urlq,
+#	    "url": (c.urlq,urldb),
 	    "units": c.unitq,
 	    "help": c.helpq,
 #	    "game": (c.gameq,game),
@@ -189,9 +194,14 @@ def command(bot, cmd, nick, conn, public):
     except IndexError:
 	    command=""
 
+    t=time.time()
+    if t - lastexp > expirevery:
+	    c.urlexpire(urldb,expirelen)
+	    lastexp=t
+
 #    if public:
 #      if c.urlre.search(cmd):
-#        c.dourl(bot,conn,nick,cmd)
+#        c.dourl(bot,conn,nick,cmd,urldb)
          
     # karma: up
     if command.endswith("++"):
diff --git a/commands.py b/commands.py
index eac7bb7..8dd8e59 100644
--- a/commands.py
+++ b/commands.py
@@ -307,10 +307,29 @@ def nicetime(tempus):
     tm="%d hours ago"%int(tempus/3600)
   return tm
 
+### class to store URL data
+class UrlLog:
+    "contains meta-data about a URL seen on-channel"
+    def __init__(self,url,nick):
+        self.nick=nick
+        self.url=url
+        self.first=time.time()
+        self.count=1
+        self.lastseen=time.time()
+        self.lastasked=time.time()
+    def recenttime(self):
+        return max(self.lastseen,self.lastasked)
+    def firstmen(self):
+        return nicetime(time.time()-self.first)
+    def urltype(self):
+        z=max(len(urlcomplaints), self.count-1)
+        return urlcomplaints(z)
+        
 urlre = re.compile("(https?://[^ ]+)( |$)")
-urlcomplaints = [" contemporary","n interesting"," fascinating","n overused"," vastly overused"]
+urlcomplaints = ["a contemporary","an interesting","a fascinating","an overused","a vastly overused"]
 
-def urlq(bot, cmd, nick, conn, public):
+### Deal with /msg bot url or ~url in channel
+def urlq(bot, cmd, nick, conn, public,urldb):
   if (not urlre.search(cmd)):
     bot.automsg(False,nick,"Please use 'url' only with http URLs")
     return
@@ -319,45 +338,42 @@ def urlq(bot, cmd, nick, conn, public):
 
   url=canonical_url(url)
   if (url in urldb):
-    users = urldb[url]
-    complaint="The url %s was mentioned %s by %s"%(url,nicetime(time.time()-users[-1][1]),users[-1][0])
+    T = urldb[url]
+    complaint="That's %s URL that was first mentioned %s by %s" % \
+               (T.urltype(),T.firstmen(),T.nick)
     if (public):
       complaint=complaint+". Furthermore it defeats the point of this command to use it other than via /msg."
     bot.automsg(False,nick,complaint)
+    T.lastasked=time.time()
   else:
     if (public):
       bot.automsg(False,nick,"That URL was unique. There is little point in using !url out loud; please use it via /msg")
     else:
       conn.privmsg(bot.channel,"%s would like to draw your attention to %s"%(nick,url))
-    urldb[url]=[[nick,time.time()]]
+    urldb[url]=UrlLog(url,nick)
 
-def dourl(bot,conn,nick,command):
+### Deal with URLs spotted in channel
+def dourl(bot,conn,nick,command,urldb):
   urlstring=urlre.search(command).group(1)
   urlstring=canonical_url(urlstring)
 
   if urlstring in urldb:
     T=urldb[urlstring]
-    uci = len(T)
-    if uci >= len(urlcomplaints):
-      uci = len(urlcomplaints)
-    message="observes a"+urlcomplaints[uci-1]+" URL: mentioned by "
-    if (len(T)>5): 
-      cutoff=len(T)-5
-    else:
-      cutoff=-1
-    for t in range(len(T)-1,cutoff,-1):
-      tempus = time.time()-T[t][1]
-      message += T[t][0]+" ("+nicetime(tempus)+")"
-      if (t!=cutoff+1):
-        message += ", "
-    if (cutoff != -1):
-      message += ", amongst others"
+    message="observes %s URL, first mentioned %s by %s" % \
+             (T.urltype(),T.firstmen(),T.nick)
     conn.action(bot.channel, message)
-    urldb[urlstring]=[[nick,time.time()]]+urldb[urlstring]
+    T.lastseen=time.time()
   else:
-    urldb[urlstring]=[[nick,time.time()]]
+    urldb[urlstring]=UrlLog(urlstring,nick)
+
+### Expire old urls
+def urlexpire(urldb,expire):
+    urls=urldb.keys()
+    for u in urls:
+        if time.time() - urldb[u].recenttime() > expire:
+            del urldb[u]
 
-# canonicalise BBC URLs
+# canonicalise BBC URLs (internal use only)
 def canonical_url(urlstring):
   if (urlstring.find("news.bbc.co.uk") != -1):
     for middle in ("/low/","/mobile/"):