X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fdict.rb;h=c9bfabafadef5fb75491d6590f492d7ec8caa48a;hb=052217de30c59206d7025b582d4604557a747470;hp=db779889592bd6bf98478c1210933c23252f85ad;hpb=6f9bfa43ac907700fcba394e0f6b9d987b1192fb;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/dict.rb b/data/rbot/plugins/dict.rb
index db779889..c9bfabaf 100644
--- a/data/rbot/plugins/dict.rb
+++ b/data/rbot/plugins/dict.rb
@@ -12,9 +12,11 @@
 #   * the Oxford dictionary for (British) English
 #   * the De Mauro/Paravia dictionary for Italian
 #   * the Chambers dictionary for English (accepts both US and UK)
+#   * the Littré dictionary for French
 #
 # Other plugins can use this one to check if a given word is valid in italian
-# or english by using the is_italian?/is_british?/is_english? methods
+# or english or french by using the is_italian?, is_british?, is_english?,
+# is_french? methods
 #
 # TODO: cache results and reuse them if get_cached returns a cache copy
 
@@ -29,13 +31,37 @@ class DictPlugin < Plugin
     :default => 0,
     :desc => "When set to n > 0, the bot will return the first paragraph from the first n dictionary hits")
 
+  def demauro_filter(s)
+    # check if it's a page we can handle
+    loc = Utils.check_location(s, @dmurlrx)
+    # the location might be not good, but we might still be able to handle the
+    # page
+    if !loc and s[:text] !~ /<!-- Il dizionario della lingua italiana Paravia: /
+      debug "not our business"
+      return
+    end
+    # we want to grab the content from the WAP page, since it's in a much
+    # cleaner HTML, so first try to get the word ID
+    if s[:text] !~ %r{<li><a href="(\d+)" title="vai al lemma precedente" accesskey="p">lemma precedente</a></li>}
+      return
+    end
+    id = $1.to_i + 1
+    title = s[:text].ircify_html_title
+    content = @bot.filter(:htmlinfo, URI.parse(@dmwaplemma % id))[:content]
+    return {:title => title, :content => content.sub(/^\S+\s+-\s+/,'')}
+  end
+
   def initialize
     super
-    @dmurl = "http://www.demauroparavia.it/"
+    @dmurl = "http://old.demauroparavia.it/"
+    @dmurlrx = %r{http://(?:www|old\.)?demauroparavia\.it/(\d+)}
     @dmwapurl = "http://wap.demauroparavia.it/index.php?lemma=%s"
     @dmwaplemma = "http://wap.demauroparavia.it/lemma.php?ID=%s"
     @oxurl = "http://www.askoxford.com/concise_oed/%s"
     @chambersurl = "http://www.chambersharrap.co.uk/chambers/features/chref/chref.py/main?query=%s&title=21st"
+    @littreurl = "http://francois.gannaz.free.fr/Littre/xmlittre.php?requete=%s"
+
+    @bot.register_filter(:demauro, :htmlinfo) { |s| demauro_filter(s) }
   end
 
 
@@ -47,8 +73,10 @@ class DictPlugin < Plugin
       return "oxford <word> => provides a link to the definition of <word> (it can also be an expression) from the Concise Oxford dictionary"
     when "chambers"
       return "chambers <word> => provides a link to the definition of <word> (it can also be an expression) from the Chambers 21st Century Dictionary"
+    when "littre"
+      return "littre <word> => provides a link to the definition of <word> (it can also be an expression) from the Littré online dictionary"
     end
-    return "<dictionary> <word>: check for <word> on <dictionary> where <dictionary> can be one of: demauro, oxford, chambers"
+    return "<dictionary> <word>: check for <word> on <dictionary> where <dictionary> can be one of: demauro, oxford, chambers, littre"
   end
 
   def demauro(m, params)
@@ -84,7 +112,8 @@ class DictPlugin < Plugin
     text += entries[0...hits].map { |ar|
       n += 1
       urls << @dmwaplemma % ar[2]
-      "#{n}. #{Bold}#{ar[0]}#{Bold} - #{ar[1].gsub(/<\/?em>/,'')}: #{@dmurl}#{ar[2]}"
+      key = ar[1].ircify_html
+      "#{n}. #{Bold}#{ar[0]}#{Bold} - #{key}: #{@dmurl}#{ar[2]}"
     }.join(" | ")
     m.reply text
 
@@ -112,9 +141,9 @@ class DictPlugin < Plugin
         url << "?view=uk"
       end
       h = @bot.httputil.get(url, :max_redir => 5)
-      if h and h.match(%r!<h2>#{word}(?:<sup>1</sup>)?</h2>!)
+      if h and h.match(/<h2>#{word}<\/h2>(.*)Perform/m)
         m.reply("#{word} : #{url}") unless justcheck
-        defn = $'
+        defn = $1
         m.reply("#{Bold}%s#{Bold}: %s" % [word, defn.ircify_html(:nbsp => :space)], :overlong => :truncate)
         return true
       end
@@ -165,10 +194,51 @@ class DictPlugin < Plugin
     return chambers(nil, :word => word, :justcheck => true)
   end
 
+  def littre(m, params)
+    justcheck = params[:justcheck]
+
+    word = params[:word].to_s.downcase
+    url = @littreurl % CGI.escape(word)
+    xml = nil
+    info = @bot.httputil.get_response(url) rescue nil
+    xml = info.body if info
+    head ||= xml.match(/<div class="entree">(.*?)<\/div>/)[1] rescue nil
+    case xml
+    when nil
+      info = info ? " (#{info.code} - #{info.message})" : ""
+      return false if justcheck
+      m.reply "An error occurred while looking for #{word}#{info}"
+      return
+    when /Erreur : le mot <STRONG>.*?<\/STRONG> n'a pas./
+      return false if justcheck
+      if head
+        m.reply "Nothing found for #{word}, I'll assume you meant #{head}"
+      else
+        m.reply "Nothing found for #{word}"
+        return
+      end
+    end
+    return true if justcheck
+    entete = xml.match(/<div class="entete">(.*?)<\/div>/m)[1] rescue nil
+    m.reply "#{head}: #{url} : #{entete.ircify_html rescue nil}"
+    entries = xml.scan(/<span class="variante">(.*?)<\!--variante-->/m)
+    hits = @bot.config['dict.hits']
+    n = 0
+    entries[0...hits].map { |ar|
+      n += 1
+      m.reply(("#{Bold}#{n}#{Bold} %s" % ar).ircify_html, :overlong => :truncate)
+    }
+  end
+
+  def is_french?(word)
+    return littre(nil, :word => word, :justcheck => true)
+  end
+
 end
 
 plugin = DictPlugin.new
-plugin.map 'demauro :word', :action => 'demauro'
-plugin.map 'oxford *word', :action => 'oxford'
-plugin.map 'chambers *word', :action => 'chambers'
+plugin.map 'demauro :word', :action => 'demauro', :thread => true
+plugin.map 'oxford *word', :action => 'oxford', :thread => true
+plugin.map 'chambers *word', :action => 'chambers', :thread => true
+plugin.map 'littre *word', :action => 'littre', :thread => true