X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Frss.rb;h=9e85b416bb6bc6db5a50994008653e29a13c5087;hb=41ea4232df0ed0b3f922b0476444ca8e38f3301f;hp=ae2c02339bfc548bf888a4f896a24c1680a14d10;hpb=087ff14d6117ef746bfad797a8088a7e6b74d5ac;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index ae2c0233..9e85b416 100644
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -25,21 +25,6 @@ end
 
 module ::RSS
 
-  # Make an  'unique' ID for a given item, based on appropriate bot options
-  # Currently only suppored is bot.config['rss.show_updated']: when true, the
-  # description is included in the uid hashing, otherwise it's not
-  #
-  def RSS.item_uid_for_bot(item, opts={})
-    options = { :show_updated => true}.merge(opts)
-    desc = nil
-    if options[:show_updated]
-      desc = item.content.content rescue item.description rescue nil
-    end
-    [(item.title.content rescue item.title rescue nil),
-     (item.link.href rescue item.link),
-     desc].hash
-  end
-
   # Add support for Slashdot namespace in RDF. The code is just an adaptation
   # of the DublinCore code.
   unless defined?(SLASH_PREFIX)
@@ -269,6 +254,38 @@ class RSSFeedsPlugin < Plugin
     :default => true,
     :desc => "Whether to display links from the text of a feed item.")
 
+  # Make an  'unique' ID for a given item, based on appropriate bot options
+  # Currently only suppored is bot.config['rss.show_updated']: when false,
+  # only the guid/link is accounted for.
+  
+  def block_rescue(df = nil, &block)
+    v = block.call rescue nil
+    (String === v && '' != v) ? v : nil
+  end
+
+  def make_uid(item)
+    uid = [
+      (block_rescue do item.guid.content end ||
+       block_rescue do item.guid end ||
+       block_rescue do item.link.href end ||
+       block_rescue do item.link end
+      )
+    ]
+    if @bot.config['rss.show_updated']
+      uid.push(
+        block_rescue do item.content.content end ||
+        block_rescue do item.description end
+      )
+      uid.unshift(
+        block_rescue do item.title.content end ||
+        block_rescue do item.title end
+      )
+    end
+    # debug "taking hash of #{uid.inspect}"
+    uid.hash
+  end
+
+
   # We used to save the Mutex with the RssBlob, which was idiotic. And
   # since Mutexes dumped in one version might not be resotrable in another,
   # we need a few tricks to be able to restore data from other versions of Ruby
@@ -297,6 +314,10 @@ class RSSFeedsPlugin < Plugin
   # TODO: load personal ones
   def define_filters
     @outkey = :"rss.out"
+    @bot.register_filter(:headlines, @outkey) { |s|
+      line1 = (s[:handle].empty? ? "%{date}" : "%{handle}") << "%{title}"
+      make_stream(line1, nil, s)
+    }
     @bot.register_filter(:blog, @outkey) { |s|
       author = s[:author] ? (s[:author] + " ") : ""
       abt = s[:category] ? "about #{s[:category]} " : ""
@@ -312,13 +333,13 @@ class RSSFeedsPlugin < Plugin
       make_stream(line1, line2, s, :author => author, :abt => abt)
     }
     @bot.register_filter(:news, @outkey) { |s|
-      line1 = "%{handle}%{date}%{title} @ %{link}" % s
+      line1 = "%{handle}%{date}%{title}%{at}%{link}" % s
       line2 = "%{handle}%{date}%{desc}" % s
       make_stream(line1, line2, s)
     }
     @bot.register_filter(:git, @outkey) { |s|
       author = s[:author] ? (s[:author] + " ") : ""
-      line1 = "%{handle}%{date}%{author}committed %{title} @ %{link}"
+      line1 = "%{handle}%{date}%{author}committed %{title}%{at}%{link}"
       make_stream(line1, nil, s, :author => author)
     }
     @bot.register_filter(:forum, @outkey) { |s|
@@ -336,7 +357,7 @@ class RSSFeedsPlugin < Plugin
     }
     @bot.register_filter(:trac, @outkey) { |s|
       author = s[:author].sub(/@\S+?\s*>/, "@...>") + ": " if s[:author]
-      line1 = "%{handle}%{date}%{author}%{title} @ %{link}"
+      line1 = "%{handle}%{date}%{author}%{title}%{at}%{link}"
       line2 = nil
       unless s[:item].title =~ /^(?:Changeset \[(?:[\da-f]+)\]|\(git commit\))/
         line2 = "%{handle}%{date}%{desc}"
@@ -574,13 +595,12 @@ class RSSFeedsPlugin < Plugin
       fetched = fetchRss(feed, m, false)
     end
     return unless fetched or feed.xml
-    if not fetched and feed.items
-      m.reply "using old data"
-    else
+    if fetched or not feed.items
       parsed = parseRss(feed, m)
-      m.reply "using old data" unless parsed
     end
     return unless feed.items
+    m.reply "using old data" unless fetched and parsed
+
     title = feed.title
     items = feed.items
 
@@ -859,9 +879,8 @@ class RSSFeedsPlugin < Plugin
               otxt = []
 
               # These are used for checking new items vs old ones
-              uid_opts = { :show_updated => @bot.config['rss.show_updated'] }
               oids = Set.new feed.items.map { |item|
-                uid = RSS.item_uid_for_bot(item, uid_opts)
+                uid = make_uid item
                 otxt << item.to_s
                 debug [uid, item].inspect
                 debug [uid, otxt.last].inspect
@@ -880,7 +899,7 @@ class RSSFeedsPlugin < Plugin
                 # debug feed.xml
 
                 dispItems = feed.items.reject { |item|
-                  uid = RSS.item_uid_for_bot(item, uid_opts)
+                  uid = make_uid item
                   txt = item.to_s
                   if oids.include?(uid)
                     debug "rejecting old #{uid} #{item.inspect}"
@@ -937,13 +956,13 @@ class RSSFeedsPlugin < Plugin
   end
 
   def select_nonempty(*ar)
-    debug ar
+    # debug ar
     ar.each { |i| return i unless i.nil_or_empty? }
     return nil
   end
 
   def printFormattedRss(feed, item, opts=nil)
-    debug item
+    # debug item
     places = feed.watchers
     handle = feed.handle.empty? ? "" : "::#{feed.handle}:: "
     date = String.new
@@ -1019,10 +1038,13 @@ class RSSFeedsPlugin < Plugin
       desc = "(?)"
     end
 
-    link = item.link.href rescue item.link.chomp rescue nil
+    link = item.link.href rescue item.link rescue nil
+    link.strip! if link
 
     category = select_nonempty((item.category.content rescue nil), (item.dc_subject rescue nil))
+    category.strip! if category
     author = select_nonempty((item.author.name.content rescue nil), (item.dc_creator rescue nil), (item.author rescue nil))
+    author.strip! if author
 
     line1 = nil
     line2 = nil
@@ -1102,7 +1124,12 @@ class RSSFeedsPlugin < Plugin
       end
       items = []
       if rss.nil?
-        report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
+        if xml.match(/xmlns\s*=\s*(['"])http:\/\/www.w3.org\/2005\/Atom\1/) and not defined?(RSS::Atom)
+          report_problem("#{feed.handle} @ #{feed.url} looks like an Atom feed, but your Ruby/RSS library doesn't seem to support it. Consider getting the latest version from http://raa.ruby-lang.org/project/rss/", nil, m)
+        else
+          report_problem("#{feed.handle} @ #{feed.url} doesn't seem to contain an RSS or Atom feed I can read", nil, m)
+        end
+        return nil
       else
         begin
           rss.output_encoding = 'UTF-8'
@@ -1126,7 +1153,7 @@ class RSSFeedsPlugin < Plugin
         report_problem("no items found in the feed, maybe try weed?", e, m)
         return nil
       end
-      feed.title = title
+      feed.title = title.strip
       feed.items = items
       return true
     end