From 8c20f8435f4a583f856d86941e66909eed91f1bb Mon Sep 17 00:00:00 2001
From: Ward Wouts <ward@baboon.wizeazz.nl>
Date: Thu, 1 Apr 2021 14:35:36 +0200
Subject: [PATCH] be much more aggressive in scrubbing caches.

---
 news/article.rb | 27 ++++++++++++++++++++++-----
 1 file changed, 22 insertions(+), 5 deletions(-)

diff --git a/news/article.rb b/news/article.rb
index 5df84ef..baded62 100644
--- a/news/article.rb
+++ b/news/article.rb
@@ -114,6 +114,10 @@ def initialize(nntpservers, groupname, newsrc="~/.newsrc", maxage=0)
 	@serverlist = []
 	@serverpasses = {}
 	@maxage = maxage.to_i
+	@oldestallowed = 0
+	if @maxage != 0
+		@oldestallowed = (DateTime.now - @maxage).strftime('%Y%m%d').to_i
+	end
 
 	tmplist = nntpservers.split('|')
 	tmplist.each{ |server|
@@ -355,10 +359,12 @@ def get_articles(cachedir=false)
 						# dit wellicht alleen doen indien preselector hem uitkiest
 						# en anders een leuk regeltje aan de cache toevoegen,
 						# maar niet in het geheugen houden
-						if preselect(art[id]["subject"])
+						if art[id]["date"].to_i >= @oldestallowed && preselect(art[id]["subject"])
 							add(id.to_i, art[id]["messid"], art[id]["date"], art[id]["from"], art[id]["subject"], server)
 						end
-						cache_add(cachedir, id, art[id]["messid"], art[id]["date"], art[id]["from"], art[id]["subject"], server)
+						if art[id]["date"].to_i >= @oldestallowed
+							cache_add(cachedir, id, art[id]["messid"], art[id]["date"], art[id]["from"], art[id]["subject"], server)
+						end
 					end
 				}
 				if @maxage and @maxage > 0
@@ -826,6 +832,7 @@ def save_newsrc()
 end
 
 def cache_add(cachedir, id, messid, date, from, subject, server)
+	# also add to skip stuff
 	if @cache_buf.has_key?(server)
 		@cache_buf[server].push("#{id}|#{messid}|#{date}|#{from}|#{subject}\n")
 	else
@@ -850,7 +857,7 @@ puts "#{Time.now}	Reading & scrubbing caches"
 	@connections.keys.each{|server|
 		first = @connections[server]["first"]
 		last = @connections[server]["last"]
-		#cache_scrub(cachedir, server)
+		cache_scrub(cachedir, server)
 puts "	#{Time.now}	Reading cache for #{server}"
 		excludes[server] = {}
 		@connections[server]["skip_ids"].elements.collect!{|x| excludes[server][x]=true}
@@ -877,11 +884,13 @@ puts "	#{Time.now}	Reading cache for #{server}"
 					if first <= id_i and id_i <= last
 						if ! excludes[server].has_key?(id_i)
 							outfile.puts(line)
-							if preselect(subject)
+							if date.to_i >= @oldestallowed  && preselect(subject)
 								add(id_i, messid, date, from, subject, server)
 							end
 			# XXX alle traagheid van de cache_read zit in deze regel:
-							@connections[server]["skip_ids"].insert!(id_i)
+							if date.to_i < @oldestallowed
+								@connections[server]["skip_ids"].insert!(id_i)
+							end
 						end
 					end
 				end
@@ -936,11 +945,19 @@ p Time.now
 		outfile = File.new("#{filename}.#{server}.new", "w") or puts "Couldn't open cachefile for writing"
 		infile.each{ |line|
 			id, messid, date, subject = line.split("|", 3)
+			#puts "#{date.to_i} #{@oldestallowed}"
+			# XXX maybe also add to skipids ??
+			next if date.to_i < @oldestallowed
 			if id.to_i >= @connections[server]["first"] and
 			   id.to_i <= @connections[server]["last"]
 				outfile.puts(line)
 			end
 		}
+		if ( FileUtils.move("#{filename}.#{server}.new", "#{filename}.#{server}") )
+			puts "	#{Time.now}	Cache scrubbed for #{server}"
+		else
+			puts "Couldn't scrub #{server} cache"
+		end
 	end
 p Time.now
 end