From b99dd4533ac832f73b17fa4ebd560789ef54ce5a Mon Sep 17 00:00:00 2001
From: lnu <lnu@f70e237a-67f3-0310-a06c-d2b8a7116972>
Date: Wed, 5 Apr 2006 16:22:19 +0000
Subject: - Now compares the whole items when searching for duplicates (might
 cause some false-negatives) - New always-new flag

git-svn-id: svn+ssh://svn.gna.org/svn/feed2imap/trunk/feed2imap@93 f70e237a-67f3-0310-a06c-d2b8a7116972
---
 data/doc/feed2imap/examples/feed2imaprc |  9 +++++++++
 lib/feed2imap/cache.rb                  | 35 ++++++++++++++++++---------------
 lib/feed2imap/config.rb                 |  3 ++-
 lib/feed2imap/feed2imap.rb              |  2 +-
 4 files changed, 31 insertions(+), 18 deletions(-)

diff --git a/data/doc/feed2imap/examples/feed2imaprc b/data/doc/feed2imap/examples/feed2imaprc
index 13ac3c2..35a9f6c 100644
--- a/data/doc/feed2imap/examples/feed2imaprc
+++ b/data/doc/feed2imap/examples/feed2imaprc
@@ -4,6 +4,10 @@
 # min-frequency (in HOURS) is the minimum frequency with which this particular
 #      feed will be fetched
 # disable: if set to something, the feed will be ignored
+# always-new: feed2imap tries to use a clever algorithm to determine whether an item
+#      is new or has been updated. It doesn't work well with some web apps like
+#      mediawiki. When this flag is enabled, all items which don't match exactly
+#      a previously downloaded item are considered as new items.
 #
 # If your login contains an @ character, replace it with %40. Other reserved
 # characters can be escaped in the same way (see man ascii to get their code)
@@ -17,3 +21,8 @@ feeds:
   - name: linuxfr
     url: http://linuxfr.org/backend/news/rss20.rss
     target: imap://lucaswebmail:password@imap.apinc.org/INBOX.Feeds.LinuxFR
+  -
+    name: JabberFrWiki
+    url: http://wiki.jabberfr.org/index.php?title=Special:Recentchanges&feed=rss
+    target: imaps://lucaswebmail:password@imap.apinc.org/INBOX.Feeds.JabberFR
+    always-new: true
diff --git a/lib/feed2imap/cache.rb b/lib/feed2imap/cache.rb
index 006e99e..cd75db2 100644
--- a/lib/feed2imap/cache.rb
+++ b/lib/feed2imap/cache.rb
@@ -30,9 +30,9 @@ class ItemCache
   end
 
   # Returns the really new items amongst items
-  def get_new_items(id, items)
+  def get_new_items(id, items, always_new = false)
     @channels[id] ||= CachedChannel::new
-    return @channels[id].get_new_items(items)
+    return @channels[id].get_new_items(items, always_new)
   end
 
   # Commit changes to the cache
@@ -116,7 +116,7 @@ class CachedChannel
   UPDATEDDEBUG = false
 
   # Returns the really new items amongst items
-  def get_new_items(items)
+  def get_new_items(items, always_new = false)
     # save number of new items
     @nbnewitems = items.length
     # set items' cached version if not set yet
@@ -130,7 +130,7 @@ class CachedChannel
       dups = false
       for i in 0...items.length do
         for j in i+1...items.length do
-          if items[i].cacheditem.link == items[j].cacheditem.link
+          if items[i].cacheditem == items[j].cacheditem
             if UPDATEDDEBUG
               puts "## Removed #{items[j].cacheditem.to_s}"
             end
@@ -148,6 +148,7 @@ class CachedChannel
       items.each { |i| puts "#{i.cacheditem.to_s}" }
       puts "-------Items already there :----------"
       @items.each { |i| puts "#{i.to_s}" }
+      puts "Items always considered as new: #{always_new.to_s}"
     end
     items.each do |i|
       found = false
@@ -163,18 +164,20 @@ class CachedChannel
         end
       end
       next if found
-      # Try to find an updated item
-      @items.each do |j|
-        # Do we need a better heuristic ?
-        if i.link and i.link == j.link
-          i.cacheditem.index = j.index
-          i.cacheditem.updated = true
-          updateditems.push(i)
-          found = true
-          # let's put j in front of itemstemp
-          @itemstemp.delete(j)
-          @itemstemp.unshift(i.cacheditem)
-          break
+      if not always_new
+        # Try to find an updated item
+        @items.each do |j|
+          # Do we need a better heuristic ?
+          if i.link and i.link == j.link
+            i.cacheditem.index = j.index
+            i.cacheditem.updated = true
+            updateditems.push(i)
+            found = true
+            # let's put j in front of itemstemp
+            @itemstemp.delete(j)
+            @itemstemp.unshift(i.cacheditem)
+            break
+          end
         end
       end
       next if found
diff --git a/lib/feed2imap/config.rb b/lib/feed2imap/config.rb
index 0483423..1e13a04 100644
--- a/lib/feed2imap/config.rb
+++ b/lib/feed2imap/config.rb
@@ -70,7 +70,7 @@ end
 
 # A configured feed. simple data container.
 class ConfigFeed
-  attr_reader :name, :url, :imapaccount, :folder
+  attr_reader :name, :url, :imapaccount, :folder, :always_new
   attr_accessor :body
 
   def initialize(f, imapaccount, folder)
@@ -79,6 +79,7 @@ class ConfigFeed
     @url.sub!(/^feed:/, '') if @url =~ /^feed:/
     @imapaccount, @folder = imapaccount, folder
     @freq = f['min-frequency']
+    @always_new =  (f['always-new'] and f['always-new'] != 'false')
     @freq = @freq.to_i if @freq
   end
 
diff --git a/lib/feed2imap/feed2imap.rb b/lib/feed2imap/feed2imap.rb
index 91660c5..19d6ccd 100644
--- a/lib/feed2imap/feed2imap.rb
+++ b/lib/feed2imap/feed2imap.rb
@@ -145,7 +145,7 @@ class Feed2Imap
         next
       end
       begin
-        newitems, updateditems = @cache.get_new_items(f.name, feed.items)
+        newitems, updateditems = @cache.get_new_items(f.name, feed.items, f.always_new)
       rescue
         @logger.fatal("Exception caught when selecting new items for #{f.name}: #{$!}")
         puts $!.backtrace
-- 
cgit v1.2.3-54-g00ecf