X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fimdb.rb;h=0c2a74ce03d007d074f0e5ed8ae335d5069e701d;hb=904106997bab0ea9894516ae324af6f3538ccf24;hp=c23cd8d663fc88b89f623bfc32e3e2914d2626a3;hpb=f0bd016c8d660b543bc5ea2643c9b9ba430198b3;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/imdb.rb b/data/rbot/plugins/imdb.rb
index c23cd8d6..0c2a74ce 100644
--- a/data/rbot/plugins/imdb.rb
+++ b/data/rbot/plugins/imdb.rb
@@ -11,13 +11,12 @@
 #
 # License:: MIT license
 
-require 'uri/common'
-
 class Imdb
-  IMDB = "http://us.imdb.com"
+  IMDB = "http://www.imdb.com"
   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 +30,7 @@ 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]
@@ -89,9 +88,9 @@ 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
@@ -100,7 +99,7 @@ class Imdb
   end
 
   def grab_info(info, body)
-    /<div class="info">\s+<h5>#{info}:<\/h5>\s+(.*?)<\/div>/mi.match(body)[1] rescue nil
+    /<div (?:id="\S+-info" )?class="info">\s+<h5>#{info}:<\/h5>\s+(.*?)<\/div>/mi.match(body)[1] rescue nil
   end
 
   def fix_article(org_tit)
@@ -154,12 +153,19 @@ class Imdb
         country = data.ircify_html.gsub(' / ','/')
       end
 
-      info << [title, "(#{country}, #{date})", extra, dir ? "[#{dir}]" : nil, ": http://us.imdb.com#{sr}"].compact.join(" ")
+      info << [title, "(#{country}, #{date})", extra, dir ? "[#{dir}]" : nil, opts[:nourl] ? nil : ": http://www.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.ircify_html]
+        }.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)
+      m = /<b>([0-9.]+)\/10<\/b>\n?\r?\s+[^<]+<a href="ratings"[^>]+>([0-9,]+) votes?<\/a>/.match(resp.body)
       if m
         ratings = "#{m[1]}/10 (#{m[2]} voters)"
       end
@@ -199,14 +205,15 @@ class Imdb
       return nil if !m
       name = m[1]
 
-      info << "#{name} : http://us.imdb.com#{sr}"
+      info << "#{name}"
+      info.last << " : http://www.imdb.com#{sr}" unless opts[:nourl]
 
       return info if opts[:name_only]
 
-      if year = opts[:movies_in_year]
+      if opts[:movies_by_year]
         filmoyear = @bot.httputil.get(IMDB + sr + "filmoyear")
         if filmoyear
-          info << filmoyear.scan(/#{TITLE_MATCH} \(#{year}\)[^\[\n]*((?:\s+\[[^\]]+\](?:\s+\([^\[<]+\))*)+)\s+</)
+          info << filmoyear.scan(/#{TITLE_MATCH} \((\d\d\d\d)\)[^\[\n]*((?:\s+\[[^\]]+\](?:\s+\([^\[<]+\))*)+)\s+</)
         end
         return info
       end
@@ -272,9 +279,25 @@ class Imdb
     return nil
   end
 
-  def year_movies(urls, year)
+  def year_movies(urls, years_txt_org, role_req)
+    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_in_year => year)
+      info = info_name(url, :movies_by_year => true)
 
       debug info.inspect
 
@@ -288,23 +311,41 @@ class Imdb
         bclip = b[1][0,5]
         quot = '&#34;'
         (aclip == quot ? 1 : -1) <=> (bclip == quot ? 1 : -1)
-      }.each { |url, pre_title, pre_roles|
+      }.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|
+          role = nil
+          extra = nil
           if txt.match(/^(.*)\s+\.\.\.\.\s+(.*)$/)
-            role_array << "#{$1} (#{$2})"
+            role = $1
+            extra = "(#{$2.ircify_html})"
           else
-            role_array << txt
+            role = txt
           end
-          role_array.last << " " + comm.ircify_html if comm
+          next if role_req and not role.match(/^#{role_req}/i)
+          if comm
+            extra ||= ""
+            extra += comm.ircify_html if comm
+          end
+          role_array << [role, extra]
         }
+        next if role_req and role_array.empty?
 
-        roles = role_array.join(', ')
-        movies << [roles, title].join(": ")
+        roles = role_array.map { |ar|
+          if role_req
+            ar[1] # works for us both if it's nil and if it's something
+          else
+            ar.compact.join(" ")
+          end
+        }.compact.join(', ')
+        roles = nil if roles.empty?
+        movies << [roles, title].compact.join(": ")
       }
 
       if movies.empty?
@@ -322,7 +363,7 @@ class Imdb
       valid = []
 
       data = @bot.httputil.get(IMDB + movie + "fullcredits")
-      data.scan(/#{NAME_MATCH}<\/td><td[^>]+> \.\.\. <\/td><td[^>]+>(.+?)<\/td>/).each { |url, name, role|
+      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|
@@ -336,43 +377,90 @@ class Imdb
 end
 
 class ImdbPlugin < Plugin
-  BotConfig.register BotConfigBooleanValue.new('imdb.aka',
+  Config.register Config::BooleanValue.new('imdb.aka',
     :default => true,
     :desc => "Look for IMDB matches also in translated titles and other 'also known as' information")
-  BotConfig.register BotConfigBooleanValue.new('imdb.popular',
+  Config.register Config::BooleanValue.new('imdb.popular',
     :default => true,
     :desc => "Display info on popular IMDB entries matching the request closely")
-  BotConfig.register BotConfigBooleanValue.new('imdb.exact',
+  Config.register Config::BooleanValue.new('imdb.exact',
     :default => true,
     :desc => "Display info on IMDB entries matching the request exactly")
-  BotConfig.register BotConfigBooleanValue.new('imdb.fix_article',
+  Config.register Config::BooleanValue.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',
+  Config.register Config::BooleanValue.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"
+    case plugin
+    when "movies"
+      "movies by <who> in <years> [as <role>] => display the movies in the <years> where which <who> was <role>; <role> can be one of actor, actress, director or anything: if it's omitted, the role is defined by the prefix: \"movies by ...\" implies director, \"movies with ...\" implies actor or actress; the years can be specified as \"in the 60s\" or as \"in 1953\""
+    when /characters?/
+      "character played by <who> in <movie> => show the character played by <who> in movie <movie>. characters in <movie> => show the actors and characters in movie <movie>"
+    else
+      "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. see also movies and characters"
+    end
   end
 
   attr_reader :i
 
+  TITLE_URL = %r{^http://(?:[^.]+\.)?imdb.com(/title/tt\d+/)}
+  NAME_URL = %r{^http://(?:[^.]+\.)?imdb.com(/name/nm\d+/)}
+  def imdb_filter(s)
+    loc = Utils.check_location(s, TITLE_URL)
+    if loc
+      sr = loc.first.match(TITLE_URL)[1]
+      extra = $2 # nothign for the time being, could be fullcredits or whatever
+      res = i.info_title(sr, :nourl => true, :characters => (extra == 'fullcredits'))
+      debug res
+      if res
+        return {:title => res.first, :content => res.last}
+      else
+        return nil
+      end
+    end
+    loc = Utils.check_location(s, NAME_URL)
+    if loc
+      sr = loc.first.match(NAME_URL)[1]
+      extra = $2 # nothing for the time being, could be filmoyear or whatever
+      res = i.info_name(sr, :nourl => true, :movies_by_year => (extra == 'filmoyear'))
+      debug res
+      if res
+        name = res.shift
+        return {:title => name, :content => res.join(". ")}
+      else
+        return nil
+      end
+    end
+    return nil
+  end
+
   def initialize
     super
     @i = Imdb.new(@bot)
+    @bot.register_filter(:imdb, :htmlinfo) { |s| imdb_filter(s) }
   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
-    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")
+      m.reply Utils.decode_html_entities(info.first.join("\n"))
     else
       m.reply info.map { |si|
         Utils.decode_html_entities si.join(" | ")
@@ -380,9 +468,23 @@ class ImdbPlugin < Plugin
     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
-    year = params[:year]
+    years = params[:years]
+    role = params[:role]
+    if role and role.downcase == 'anything'
+      role = nil
+    elsif not role
+      case params[:prefix].intern
+      when :with
+        role = /actor|actress/i
+      when :by
+        role = 'director'
+      end
+    end
 
     name_urls = i.search(who, :type => :name)
     unless name_urls
@@ -390,7 +492,7 @@ class ImdbPlugin < Plugin
       return
     end
 
-    movie_urls = i.year_movies(name_urls, year)
+    movie_urls = i.year_movies(name_urls, years, role)
     debug movie_urls.inspect
     debug movie_urls[0][1]
 
@@ -398,12 +500,14 @@ class ImdbPlugin < Plugin
       m.reply movie_urls.join("\n")
     else
       m.reply movie_urls.map { |si|
-        si[1] = "no movies in #{year}" unless si[1]
+        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
@@ -416,12 +520,28 @@ class ImdbPlugin < Plugin
 
     movie_urls = i.search(movie, :type => :title)
     unless movie_urls
-      m.reply "nothing found about #{hwo}, sorry"
+      m.reply "nothing found about #{who}, sorry"
       return
     end
 
     info = i.name_in_movie(name_urls, movie_urls)
-    m.reply info.join("\n")
+    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
@@ -429,7 +549,8 @@ end
 plugin = ImdbPlugin.new
 
 plugin.map "imdb [:type] *what", :requirements => { :type => /name|title/ }, :defaults => { :type => 'both' }
-plugin.map "movies :prefix *who in :year", :requirements => { :prefix => /with|by|from/, :year => /\d+/ }
+plugin.map "movies :prefix *who in [the] :years [as :role]", :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