X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fimdb.rb;h=8c46f3fe24fae918000cbbd5fab0130e51a5296c;hb=37c38bb9571052b75ede02602e568754c5ff8158;hp=543a9582455f534d465114b8d7bca562d821e108;hpb=a0a91c1259343ab4e3a39f39c1b8d449d1b77648;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/imdb.rb b/data/rbot/plugins/imdb.rb
index 543a9582..8c46f3fe 100644
--- a/data/rbot/plugins/imdb.rb
+++ b/data/rbot/plugins/imdb.rb
@@ -18,6 +18,7 @@ class Imdb
   TITLE_OR_NAME_MATCH = /<a href="(\/(?:title|name)\/(?:tt|nm)[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
   TITLE_MATCH = /<a href="(\/title\/tt[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
   NAME_MATCH = /<a href="(\/name\/nm[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
+  CREDIT_NAME_MATCH = /#{NAME_MATCH}<\/td><td[^>]+> \.\.\. <\/td><td[^>]+>(.+?)<\/td>/
   FINAL_ARTICLE_MATCH = /, ([A-Z]\S{0,2})$/
 
   MATCHER = {
@@ -31,7 +32,7 @@ class Imdb
   end
 
   def search(rawstr, rawopts={})
-    str = URI.escape(rawstr)
+    str = CGI.escape(rawstr)
     str << ";site=aka" if @bot.config['imdb.aka']
     opts = rawopts.dup
     opts[:type] = :both unless opts[:type]
@@ -51,7 +52,6 @@ class Imdb
 
 
     matcher = MATCHER[opts[:type]]
-    debug matcher.inspect
 
     if resp.code == "200"
       m = []
@@ -90,9 +90,9 @@ class Imdb
       type = sr.match(/^\/([^\/]+)\//)[1].downcase.intern rescue nil
       case type
       when :title
-        results << info_title(sr)
+        results << info_title(sr, opts)
       when :name
-        results << info_name(sr)
+        results << info_name(sr, opts)
       else
         results << "#{sr}"
       end
@@ -106,10 +106,14 @@ class Imdb
 
   def fix_article(org_tit)
     title = org_tit.dup
+    debug title.inspect
+    if title.match(/^"(.*)"$/)
+      return "\"#{fix_article($1)}\""
+    end
     if @bot.config['imdb.fix_article'] and title.gsub!(FINAL_ARTICLE_MATCH, '')
       art = $1.dup
       debug art.inspect
-      if art[-1,1].match(/[a-z]/)
+      if art[-1,1].match(/[A-Za-z]/)
         art << " "
       end
       return art + title
@@ -117,7 +121,7 @@ class Imdb
     return title
   end
 
-  def info_title(sr)
+  def info_title(sr, opts={})
     resp = nil
     begin
       resp = @bot.httputil.get_response(IMDB + sr, :max_redir => -1)
@@ -135,13 +139,13 @@ class Imdb
       title_date = m[1]
       pre_title, date, extra = title_date.scan(/^(.*)\((\d\d\d\d(?:\/[IV]+)?)\)\s*(.+)?$/).first
       pre_title.strip!
-      title = fix_article(pre_title)
+      title = fix_article(pre_title.ircify_html)
 
       dir = nil
       data = grab_info(/Directors?/, resp.body)
       if data
         dir = data.scan(NAME_MATCH).map { |url, name|
-          name
+          name.ircify_html
         }.join(', ')
       end
 
@@ -153,6 +157,15 @@ class Imdb
 
       info << [title, "(#{country}, #{date})", extra, dir ? "[#{dir}]" : nil, ": http://us.imdb.com#{sr}"].compact.join(" ")
 
+      return info if opts[:title_only]
+
+      if opts[:characters]
+        info << resp.body.scan(CREDIT_NAME_MATCH).map { |url, name, role|
+          "%s: %s" % [name, role]
+        }.join('; ')
+        return info
+      end
+
       ratings = "no votes"
       m = /<b>([0-9.]+)\/10<\/b>\n?\r?\s+<small>\(<a href="ratings">([0-9,]+) votes?<\/a>\)<\/small>/.match(resp.body)
       if m
@@ -177,7 +190,7 @@ class Imdb
     return nil
   end
 
-  def info_name(sr)
+  def info_name(sr, opts={})
     resp = nil
     begin
       resp = @bot.httputil.get_response(IMDB + sr, :max_redir => -1)
@@ -196,6 +209,16 @@ class Imdb
 
       info << "#{name} : http://us.imdb.com#{sr}"
 
+      return info if opts[:name_only]
+
+      if opts[:movies_by_year]
+        filmoyear = @bot.httputil.get(IMDB + sr + "filmoyear")
+        if filmoyear
+          info << filmoyear.scan(/#{TITLE_MATCH} \((\d\d\d\d)\)[^\[\n]*((?:\s+\[[^\]]+\](?:\s+\([^\[<]+\))*)+)\s+</)
+        end
+        return info
+      end
+
       birth = nil
       data = grab_info("Date of Birth", resp.body)
       if data
@@ -223,7 +246,7 @@ class Imdb
           what = str.match(/<a name="[^"]+">([^<]+)<\/a>/)[1] rescue nil
           next unless what
           movies[what] = str.scan(TITLE_MATCH)[0..2].map { |url, tit|
-            fix_article(tit)
+            fix_article(tit.ircify_html)
           }
         }
       end
@@ -256,6 +279,86 @@ class Imdb
     end
     return nil
   end
+
+  def year_movies(urls, years_txt_org)
+    years_txt = years_txt_org.dup
+    years_txt.sub!(/^'/,'')
+    years_txt = "9#{years_txt}" if years_txt.match(/^\d\ds?$/)
+    years_txt = "1#{years_txt}" if years_txt.match(/^\d\d\ds?$/)
+
+    years = []
+    case years_txt
+    when /^\d\d\d\d$/
+      years << years_txt
+    when /^(\d\d\d\d)s$/
+      base = $1.to_i
+      base.upto(base+9) { |year|
+        years << year.to_s
+      }
+    end
+
+    urls.map { |url|
+      info = info_name(url, :movies_by_year => true)
+
+      debug info.inspect
+
+      name_url = info.first
+      data = info[1]
+
+      movies = []
+      # Sort by pre-title putting movies before TV series
+      data.sort { |a, b|
+        aclip = a[1][0,5]
+        bclip = b[1][0,5]
+        quot = '&#34;'
+        (aclip == quot ? 1 : -1) <=> (bclip == quot ? 1 : -1)
+      }.each { |url, pre_title, year, pre_roles|
+        next unless years.include?(year)
+        title = fix_article(pre_title.ircify_html)
+        if title[0] == ?" and not @bot.config['imdb.tv_series_in_movies']
+          next
+        end
+        title << " (#{year})" unless years.length == 1
+        role_array = []
+        pre_roles.strip.scan(/\[([^\]]+)\]((?:\s+\([^\[]+\))+)?/) { |txt, comm|
+          if txt.match(/^(.*)\s+\.\.\.\.\s+(.*)$/)
+            role_array << "#{$1} (#{$2})"
+          else
+            role_array << txt
+          end
+          role_array.last << " " + comm.ircify_html if comm
+        }
+
+        roles = role_array.join(', ')
+        movies << [roles, title].join(": ")
+      }
+
+      if movies.empty?
+        [name_url, nil]
+      else
+        [name_url, movies.join(" | ")]
+      end
+    }
+  end
+
+  def name_in_movie(name_urls, movie_urls)
+    info = []
+    movie_urls.each { |movie|
+      title_info = info_title(movie, :title_only => true)
+      valid = []
+
+      data = @bot.httputil.get(IMDB + movie + "fullcredits")
+      data.scan(CREDIT_NAME_MATCH).each { |url, name, role|
+        valid << [url, name.ircify_html, role.ircify_html] if name_urls.include?(url)
+      }
+      valid.each { |url, name, role|
+        info << "%s : %s was %s in %s" % [name, IMDB + url, role, title_info]
+      }
+    }
+    return info
+  end
+
+
 end
 
 class ImdbPlugin < Plugin
@@ -271,30 +374,118 @@ class ImdbPlugin < Plugin
   BotConfig.register BotConfigBooleanValue.new('imdb.fix_article',
     :default => false,
     :desc => "Try to detect an article placed at the end and move it in front of the title")
+  BotConfig.register BotConfigBooleanValue.new('imdb.tv_series_in_movies',
+    :default => false,
+    :desc => "Whether searching movies by person/year should also return TV series")
 
   def help(plugin, topic="")
     "imdb <string> => search http://www.imdb.org for <string>: prefix <string> with 'name' or 'title' if you only want to search for people or films respectively, e.g.: imdb name ed wood"
   end
 
+  attr_reader :i
+
+  def initialize
+    super
+    @i = Imdb.new(@bot)
+  end
+
+  # Find a person or movie on IMDB. A :type (name/title, default both) can be
+  # specified to limit the search to either.
+  #
   def imdb(m, params)
-    what = params[:what].to_s
-    type = params[:type].intern
-    i = Imdb.new(@bot)
-    info = i.info(what, :type => type)
-    if !info
-      m.reply "Nothing found for #{what}"
-      return nil
+    if params[:movie]
+      movie = params[:movie].to_s
+      info = i.info(movie, :type => :title, :characters => true)
+    else
+      what = params[:what].to_s
+      type = params[:type].intern
+      info = i.info(what, :type => type)
+      if !info
+        m.reply "nothing found for #{what}"
+        return nil
+      end
     end
     if info.length == 1
       m.reply Utils.decode_html_entities info.first.join("\n")
     else
-      m.reply info.map { |i|
-        Utils.decode_html_entities i.join(" | ")
+      m.reply info.map { |si|
+        Utils.decode_html_entities si.join(" | ")
+      }.join("\n")
+    end
+  end
+
+  # Find the movies with a participation of :who in the year :year
+  # TODO: allow year to be either a year or a decade ('[in the] 1960s') 
+  #
+  def movies(m, params)
+    who = params[:who].to_s
+    years = params[:years]
+
+    name_urls = i.search(who, :type => :name)
+    unless name_urls
+      m.reply "nothing found about #{who}, sorry"
+      return
+    end
+
+    movie_urls = i.year_movies(name_urls, years)
+    debug movie_urls.inspect
+    debug movie_urls[0][1]
+
+    if movie_urls.length == 1 and movie_urls[0][1]
+      m.reply movie_urls.join("\n")
+    else
+      m.reply movie_urls.map { |si|
+        si[1] = "no movies in #{years}" unless si[1]
+        Utils.decode_html_entities si.join(" | ")
       }.join("\n")
     end
   end
+
+  # Find the character played by :who in :movie
+  #
+  def character(m, params)
+    who = params[:who].to_s
+    movie = params[:movie].to_s
+
+    name_urls = i.search(who, :type => :name)
+    unless name_urls
+      m.reply "nothing found about #{who}, sorry"
+      return
+    end
+
+    movie_urls = i.search(movie, :type => :title)
+    unless movie_urls
+      m.reply "nothing found about #{who}, sorry"
+      return
+    end
+
+    info = i.name_in_movie(name_urls, movie_urls)
+    if info.empty?
+      m.reply "nothing found about #{who} in #{movie}, sorry"
+    else
+      m.reply info.join("\n")
+    end
+  end
+
+  # Report the characters in movie :movie
+  #
+  def characters(m, params)
+    movie = params[:movie].to_s
+
+    urls = i.search(movie, :type => :title)
+    unless urls
+      m.reply "nothing found about #{movie}"
+    end
+
+  end
+
 end
 
 plugin = ImdbPlugin.new
+
 plugin.map "imdb [:type] *what", :requirements => { :type => /name|title/ }, :defaults => { :type => 'both' }
+plugin.map "movies :prefix *who in [the] :years", :requirements => { :prefix => /with|by|from/, :years => /'?\d+s?/ }
+plugin.map "character [played] by *who in *movie"
+plugin.map "character of *who in *movie"
+plugin.map "characters in *movie", :action => :imdb