rss2mail 0.0.1 → 0.0.2

data/README

@@ -2,7 +2,7 @@
 
 == VERSION
 
-This documentation refers to rss2mail version 0.0.1
+This documentation refers to rss2mail version 0.0.2
 
 
 == DESCRIPTION
@@ -26,7 +26,7 @@ Rubyforge project:: <http://rubyforge.org/projects/rss2mail>
 
 == LICENSE AND COPYRIGHT
 
-Copyright (C) 2007-2008 Jens Wille
+Copyright (C) 2007-2009 Jens Wille
 
 rss2mail is free software: you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free Software

data/Rakefile

@@ -15,7 +15,7 @@ begin
       :summary      => %q{Send RSS feeds as e-mail},
       :homepage     => %q{http://rss2mail.rubyforge.org/},
       :files        => FileList['lib/**/*.rb', 'bin/*'].to_a,
-      :extra_files  => FileList['[A-Z]*', 'example/*'].to_a,
+      :extra_files  => FileList['[A-Z]*', 'templates/*', 'example/*'].to_a,
       :dependencies => %w[simple-rss hpricot unidecode ruby-nuggets]
     }
   }}

data/bin/rss2mail

@@ -5,7 +5,7 @@
 #                                                                             #
 # rss2mail -- Send RSS feeds as e-mail                                        #
 #                                                                             #
-# Copyright (C) 2007-2008 Jens Wille                                          #
+# Copyright (C) 2007-2009 Jens Wille                                          #
 #                                                                             #
 # Authors:                                                                    #
 #     Jens Wille <ww@blackwinter.de>                                          #
@@ -40,7 +40,8 @@ abort USAGE if ARGV.empty?
 options = {
   :files   => nil,
   :reload  => false,
-  :verbose => false
+  :verbose => false,
+  :debug   => false
 }
 
 OptionParser.new { |opts|
@@ -61,6 +62,10 @@ OptionParser.new { |opts|
   opts.on('-v', '--verbose', 'Be verbose') {
     options[:verbose] = true
   }
+
+  opts.on('-D', '--debug', "Print debug output and don't send any mails") {
+    options[:debug] = true
+  }
 }.parse!
 
 if target = ARGV.shift
@@ -70,16 +75,15 @@ else
 end
 
 templates = Hash.new { |h, k|
-  t = File.join(base, 'templates', "#{k}.erb")
-
-  begin
-    h[k] = File.read(t)
+  h[k] = begin
+    File.read(File.join(base, 'templates', "#{k}.erb"))
   rescue Errno::ENOENT
     # silently ignore
   end
 }
 
 feeds_files = options.delete(:files) || [File.join(base, 'feeds.yaml')]
+
 feeds_files.each { |feeds_file|
   feeds = begin
     YAML.load_file(feeds_file)
@@ -98,7 +102,5 @@ feeds_files.each { |feeds_file|
   }
 
   # write updated feed information
-  File.open(feeds_file, 'w') { |file|
-    YAML.dump(feeds, file)
-  }
+  File.open(feeds_file, 'w') { |file| YAML.dump(feeds, file) } unless options[:debug]
 }

data/example/feeds.yaml

@@ -28,3 +28,8 @@
 - :url: http://rss.slashdot.org/Slashdot/slashdotBookReviews
   :to: my.secret@e.mail
   :title: "Slashdot: Book Reviews"
+:test: 
+- :url: http://github.com/blackwinter.atom
+  :to: my.secret@e.mail
+  :title: GitHub/blackwinter
+  :unescape_html: true

data/lib/rss2mail/feed.rb

@@ -3,7 +3,7 @@
 #                                                                             #
 # A component of rss2mail, the RSS to e-mail forwarder.                       #
 #                                                                             #
-# Copyright (C) 2007-2008 Jens Wille                                          #
+# Copyright (C) 2007-2009 Jens Wille                                          #
 #                                                                             #
 # Authors:                                                                    #
 #     Jens Wille <ww@blackwinter.de>                                          #
@@ -28,9 +28,6 @@ require 'open-uri'
 require 'erb'
 
 require 'rubygems'
-require 'hpricot'
-require 'unidecode'
-require 'nuggets/util/i18n'
 require 'nuggets/string/evaluate'
 
 require 'rss2mail/rss'
@@ -39,17 +36,9 @@ module RSS2Mail
 
   class Feed
 
-    SUBSTITUTIONS = {
-      '–'     => '--',
-      '«'     => '<<',
-      '&amp;' => '&'
-    }
+    HOST = ENV['HOSTNAME'] || ENV['HOST'] || %x{hostname}.chomp
 
-    SUBSTITUTIONS_RE = %r{Regexp.union(*SUBSTITUTIONS.keys)}o
-
-    TAGS_TO_KEEP = %w[a p br h1 h2 h3 h4]
-
-    attr_reader :feed, :verbose, :reload, :simple, :updated, :content, :rss
+    attr_reader :feed, :reload, :verbose, :debug, :simple, :updated, :content, :rss
 
     def initialize(feed, options = {})
       raise TypeError, "Hash expected, got #{feed.class}" unless feed.is_a?(Hash)
@@ -58,33 +47,34 @@ module RSS2Mail
       @simple  = feed[:simple]
       @updated = feed[:updated]
 
-      @verbose = options[:verbose]
       @reload  = options[:reload]
+      @verbose = options[:verbose]
+      @debug   = options[:debug]
 
       required = [:url, :to, :title]
       required.delete_if { |i| feed.has_key?(i) }
 
-      raise ArgumentError, "feed incomplete: #{required.join(', ')} missing" unless required.empty?
+      raise ArgumentError, "Feed incomplete: #{required.join(', ')} missing" unless required.empty?
     end
 
     def deliver(templates)
-      unless get && parse
-        warn "[#{feed[:title]}] Nothing to send" if verbose
+      to = [*feed[:to]]
+
+      if to.empty?
+        log 'No one to send to'
         return
       end
 
-      if rss.items.empty?
-        warn "[#{feed[:title]}] No new items" if verbose
+      unless get && parse
+        log 'Nothing to send'
         return
       end
 
-      to = [*feed[:to]]
-      if to.empty?
-        warn "[#{feed[:title]}] No one to send to" if verbose
+      if rss.items.empty?
+        log 'No new items'
         return
       end
 
-      feed_title   = feed[:title]
       content_type = feed[:content_type] || 'text/html'
       encoding     = feed[:encoding]     || 'UTF-8'
 
@@ -93,7 +83,7 @@ module RSS2Mail
       content_type_header = "Content-type: #{content_type}; charset=#{encoding}"
 
       unless template = templates[content_type[/\/(.*)/, 1]]
-        warn "[#{feed[:title]}] Template not found: #{content_type}" if verbose
+        log "Template not found: #{content_type}"
         return
       end
 
@@ -101,8 +91,8 @@ module RSS2Mail
         '/usr/bin/mail',
         '-e',
         "-a '#{content_type_header}'",
-        "-a 'From: rss2mail@blackwinter.de'",
-        "-s '[#{feed_title}] \#{subject}'",
+        "-a 'From: rss2mail@#{HOST}'",
+        "-s '[#{feed[:title]}] \#{subject}'",
         *to
       ].join(' ')
 
@@ -111,45 +101,25 @@ module RSS2Mail
       rss.items.each { |item|
         title       = item.title
         link        = item.link
-        description = item.description
+        description = item.description(feed[:unescape_html])
         date        = item.date
         author      = item.author
+        body        = item.body(feed[:body])
+        subject     = item.subject
 
-        if description && feed[:unescape_html]
-          description.gsub!(/&lt;/, '<')
-          description.gsub!(/&gt;/, '>')
-        end
-
-        if tag = feed[:body]
-          body = case tag
-            when true: open(link).read
-            else       Hpricot(open(link)).at(tag).to_s
-          end.gsub(/<\/?(.*?)>/) { |m|
-            m if TAGS_TO_KEEP.include?($1.split.first.downcase)
-          }.gsub(/<a\s+href=['"](?!http:).*?>(.*?)<\/a>/mi, '\1')
+        log "#{title} / #{date} [#{author}]", debug
+        log "<#{link}>", debug
 
-          if body_encoding = feed[:body_encoding]
-            body = Iconv.conv('UTF-8', body_encoding, body)
-          end
-        end
-
-        subject = title ? clean_subject(title) : 'NO TITLE'
-
-        _cmd = cmd.evaluate(binding)
-
-        begin
-          IO.popen(_cmd, 'w') { |mail| mail.puts ERB.new(template).result(binding) }
+        send_mail(cmd.evaluate(binding), ERB.new(template).result(binding)) {
           feed[:sent] << link
           sent += 1
-        rescue Errno::EPIPE => err
-          warn "[#{feed[:title]}] Error while sending mail (#{err.class}): #{_cmd}"
-        end
+        }
       }
 
       # only keep the last 100 entries
       feed[:sent].slice!(0...-100)
 
-      warn "[#{feed[:title]}] #{sent} items sent" if verbose
+      log "#{sent} items sent"
       sent
     end
 
@@ -161,26 +131,30 @@ module RSS2Mail
         conditions = {}
       else
         conditions = case
-          when etag  = feed[:etag]:  { 'If-None-Match'     => etag  }
-          when mtime = feed[:mtime]: { 'If-Modified-Since' => mtime }
-          else                         {}
+          when etag  = feed[:etag]  then { 'If-None-Match'     => etag  }
+          when mtime = feed[:mtime] then { 'If-Modified-Since' => mtime }
+          else                           {}
         end
       end
 
+      log conditions.inspect, debug
+
       begin
         open(feed[:url], conditions) { |uri|
           case
-            when etag  = uri.meta['etag']:  feed[:etag]    = etag
-            when mtime = uri.last_modified: feed[:mtime]   = mtime.rfc822
-            else                            feed[:updated] = Time.now
+            when etag  = uri.meta['etag']  then feed[:etag]    = etag
+            when mtime = uri.last_modified then feed[:mtime]   = mtime.rfc822
+            else                                feed[:updated] = Time.now
           end
 
           @content ||= uri.read
         }
+
+        log feed.values_at(:etag, :mtime, :updated).inspect, debug
       rescue OpenURI::HTTPError
-        warn "[#{feed[:title]}] Feed not found or unchanged" if verbose
+        log 'Feed not found or unchanged'
       rescue Timeout::Error, Errno::ETIMEDOUT, Errno::ECONNRESET => err
-        warn "[#{feed[:title]}] Error while getting feed: #{err} (#{err.class})"
+        error err, 'while getting feed'
       end
 
       @content
@@ -192,7 +166,7 @@ module RSS2Mail
       if content && @rss ||= begin
         RSS2Mail::RSS.new(content, simple)
       rescue SimpleRSSError => err
-        warn "[#{feed[:title]}] Error while parsing feed: #{err} (#{err.class})"
+        error err, 'while parsing feed'
       end
         sent = feed[:sent]
 
@@ -206,18 +180,33 @@ module RSS2Mail
           }
         end
       else
-        warn "[#{feed[:title]}] Nothing to parse" if verbose
+        log 'Nothing to parse'
       end
 
       @rss
     end
 
-    def clean_subject(string)
-      string.
-        replace_diacritics.
-        gsub(SUBSTITUTIONS_RE) { |m| SUBSTITUTIONS[m] }.
-        to_ascii.
-        gsub(/'/, "'\\\\''")
+    def send_mail(cmd, body)
+      return if debug
+
+      IO.popen(cmd, 'w') { |mail| mail.puts body }
+      yield if block_given?
+    rescue Errno::EPIPE => err
+      error err, 'while sending mail', cmd
+    end
+
+    def log(msg, verbose = verbose)
+      warn "[#{feed[:title]}] #{msg}" if verbose
+    end
+
+    def error(err = nil, occasion = nil, extra = nil)
+      msg = 'Error'
+
+      msg << " #{occasion}"            if occasion
+      msg << ": #{err} (#{err.class})" if err
+      msg << " [#{extra}]"             if extra
+
+      log msg, true
     end
 
   end

data/lib/rss2mail/rss.rb

@@ -1,9 +1,11 @@
+# encoding: utf-8
+
 #--
 ###############################################################################
 #                                                                             #
 # A component of rss2mail, the RSS to e-mail forwarder.                       #
 #                                                                             #
-# Copyright (C) 2007-2008 Jens Wille                                          #
+# Copyright (C) 2007-2009 Jens Wille                                          #
 #                                                                             #
 # Authors:                                                                    #
 #     Jens Wille <ww@blackwinter.de>                                          #
@@ -24,15 +26,34 @@
 ###############################################################################
 #++
 
+require 'open-uri'
 require 'rss'
 
 require 'rubygems'
 require 'simple-rss'
+require 'unidecode'
+require 'nuggets/util/i18n'
+
+begin
+  require 'hpricot'
+rescue LoadError => err
+  warn err
+end
 
 module RSS2Mail
 
   class RSS
 
+    SUBSTITUTIONS = {
+      '–'     => '--',
+      '«'     => '<<',
+      '&amp;' => '&'
+    }
+
+    SUBSTITUTIONS_RE = Regexp.union(*SUBSTITUTIONS.keys)
+
+    TAGS_TO_KEEP = %w[a p br h1 h2 h3 h4]
+
     attr_reader :content, :rss
 
     def initialize(content, simple = false)
@@ -60,32 +81,100 @@ module RSS2Mail
 
     class Item
 
-      ALIASES = {
-        :title       => %w[],
-        :link        => %w[],
-        :description => %w[summary content],
-        :date        => %w[pubDate updated],
-        :author      => %w[dc_creator]
-      }
-
       def initialize(item)
         @item = item
       end
 
-      def method_missing(method, *args, &block)
-        if aliases = ALIASES[method]
-          [method, *aliases].each { |name|
-            begin
-              res = @item.send(name)
-              return res if res
-            rescue NoMethodError
-            end
+      def title
+        @title ||= value_for(:title, :content)
+      end
+
+      def link
+        @link ||= value_for(:link, :href)
+      end
+
+      def description(unescape_html = false)
+        @description ||= get_description(unescape_html)
+      end
+
+      def date
+        @date ||= value_for({ :date => %w[pubDate updated dc_date] }, :content) { |field, value|
+          field == 'updated' && value.respond_to?(:content) ? Time.at(value.content.to_i) : value
+        }
+      end
+
+      def author
+        @author ||= value_for({ :author => %w[contributor dc_creator] }, %w[name content])
+      end
+
+      def body(tag = nil, encoding = nil)
+        @body ||= get_body(tag, encoding)
+      end
+
+      def subject
+        @subject ||= title ? clean_subject(title) : 'NO TITLE'
+      end
+
+      private
+
+      def value_for(field, methods = nil, &block)
+        value = get_value_for(field, &block)
+
+        if methods
+          [*methods].each { |method|
+            break unless value.respond_to?(method)
+            value = value.send(method)
           }
+        end
+
+        value.respond_to?(:strip) ? value.strip : value
+      end
+
+      def get_value_for(fields, &block)
+        fields = fields.is_a?(Hash) ? fields.to_a.flatten : [*fields]
+
+        fields.each { |field|
+          begin
+            value = @item.send(field)
+            value = block[field, value] if block
+            return value if value
+          rescue NoMethodError
+          end
+        }
+
+        nil
+      end
 
-          nil
-        else
-          super
+      def get_description(unescape_html)
+        description = value_for({ :description => %w[summary content] }, :content)
+
+        if description && unescape_html
+          description.gsub!(/&lt;/, '<')
+          description.gsub!(/&gt;/, '>')
+        end
+
+        description
+      end
+
+      def get_body(tag, encoding)
+        body = case tag
+          when nil  then return
+          when true then open(link).read
+          else           defined?(Hpricot) ? Hpricot(open(link)).at(tag).to_s : open(link).read
         end
+
+        body.gsub!(/<\/?(.*?)>/) { |m| m if TAGS_TO_KEEP.include?($1.split.first.downcase) }
+        body.gsub!(/<a\s+href=['"](?!http:).*?>(.*?)<\/a>/mi, '\1')
+
+        encoding ? Iconv.conv('UTF-8', encoding, body) : body
+      end
+
+      def clean_subject(string)
+        string.
+          replace_diacritics.
+          gsub(SUBSTITUTIONS_RE) { |m| SUBSTITUTIONS[m] }.
+          to_ascii.
+          gsub(/'/, "'\\\\''")
       end
 
     end

data/lib/rss2mail/util.rb

@@ -3,7 +3,7 @@
 #                                                                             #
 # A component of rss2mail, the RSS to e-mail forwarder.                       #
 #                                                                             #
-# Copyright (C) 2007-2008 Jens Wille                                          #
+# Copyright (C) 2007-2009 Jens Wille                                          #
 #                                                                             #
 # Authors:                                                                    #
 #     Jens Wille <ww@blackwinter.de>                                          #

data/lib/rss2mail/version.rb

@@ -4,7 +4,7 @@ module RSS2Mail
 
     MAJOR = 0
     MINOR = 0
-    TINY  = 1
+    TINY  = 2
 
     class << self
 

data/lib/rss2mail.rb

@@ -3,7 +3,7 @@
 #                                                                             #
 # rss2mail -- Send RSS feeds as e-mail                                        #
 #                                                                             #
-# Copyright (C) 2007-2008 Jens Wille                                          #
+# Copyright (C) 2007-2009 Jens Wille                                          #
 #                                                                             #
 # Authors:                                                                    #
 #     Jens Wille <ww@blackwinter.de>                                          #

data/templates/html.erb

@@ -0,0 +1,12 @@
+<html>
+<body>
+<%= title %><% if date %> / <%= date %><% end %><% if author %> [<%= author %>]<% end %><br />
+<% if description %><br />
+<%= description %><br />
+<% end %><br />
+<% if body %><br />
+<%= body %><br />
+<% end %><br />
+&lt;<%= link %>&gt;
+</body>
+</html>

data/templates/plain.erb

@@ -0,0 +1,8 @@
+<%= title %><% if date %> / <%= date %><% end %><% if author %> [<%= author %>]<% end %>
+<% if description %>
+<%= description %>
+<% end %>
+<% if body %>
+<%= body %>
+<% end %>
+<<%= link %>>

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: rss2mail
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors: 
 - Jens Wille
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-12-19 00:00:00 +01:00
+date: 2009-03-02 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -73,17 +73,19 @@ files:
 - COPYING
 - ChangeLog
 - README
+- templates/plain.erb
+- templates/html.erb
 - example/feeds.yaml
 has_rdoc: true
 homepage: http://rss2mail.rubyforge.org/
 post_install_message: 
 rdoc_options: 
 - --line-numbers
+- --main
+- README
 - --inline-source
 - --title
 - rss2mail Application documentation
-- --main
-- README
 - --charset
 - UTF-8
 - --all