IMDB = "http://www.imdb.com"
TITLE_OR_NAME_MATCH = /<a\s+href="(\/(?:title|name)\/(?:tt|nm)[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
TITLE_MATCH = /<a\s+href="(\/title\/tt[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
- NAME_MATCH = /<a\s+href="(\/name\/nm[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
+ NAME_MATCH = /<a\s+onclick="[^"]+"\s+href="(\/name\/nm[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
+ CHAR_MATCH = /<a\s+href="(\/character\/ch[0-9]+\/?)[^"]*"(?:[^>]*)>([^<]*)<\/a>/
CREDIT_NAME_MATCH = /#{NAME_MATCH}\s*<\/td>\s*<td[^>]+>\s*\.\.\.\s*<\/td>\s*<td[^>]+>\s*(.+?)\s*<\/td>/m
FINAL_ARTICLE_MATCH = /, ([A-Z]\S{0,2})$/
DESC_MATCH = /<meta name="description" content="(.*?)\. (.*?)\. (.*?)\."\s*\/>/
MATCHER = {
+ :character => CHAR_MATCH,
:title => TITLE_MATCH,
:name => NAME_MATCH,
:both => TITLE_OR_NAME_MATCH
end
def search(rawstr, rawopts={})
+ # allow the user to search directly for (movie) IDs
+ if rawstr.match /$tt\d+^/ then
+ return ["/movie/#{rawstr}/"]
+ end
str = CGI.escape(rawstr)
- str << ";site=aka" if @bot.config['imdb.aka']
+ str << "&site=aka" if @bot.config['imdb.aka']
opts = rawopts.dup
opts[:type] = :both unless opts[:type]
return do_search(str, opts)
matcher = MATCHER[opts[:type]]
if resp.code == "200"
+ if opts[:all]
+ return resp.body.scan(matcher).map { |m| m.first }.compact.uniq
+ end
m = []
m << matcher.match(resp.body) if @bot.config['imdb.popular']
if resp.body.match(/\(Exact Matches\)<\/b>/) and @bot.config['imdb.exact']
def info_title(sr, opts={})
resp = nil
begin
+ # movie urls without tailing / trigger a redirect
+ sr += '/' if sr[-1,1] != '/'
resp = @bot.httputil.get_response(IMDB + sr, :max_redir => -1)
rescue Exception => e
error e.message
title_date = m[1].ircify_html
debug title_date
# note that the date dash for series is a - (ndash), not a - (minus sign)
- pre_title, extra, date, junk = title_date.scan(/^(.*)\((.+?\s+)?(\d\d\d\d(?:–(?:\d\d\d\d)?)?(?:\/[IV]+)?)\)\s*(.+)?$/).first
+ # also, the second date, if missing, is an no-break space
+ pre_title, extra, date, junk = title_date.scan(/^(.*)\((.+?\s+)?(\d\d\d\d(?:–(?:\d\d\d\d| )?)?(?:\/[IV]+)?)\)\s*(.+)?$/).first
extra.strip! if extra
pre_title.strip!
title = fix_article(pre_title)
end
ratings = "no votes"
- m = resp.body.match(/<b>([0-9.]+)<\/b><span [^>]+>\/10<\/span><\/span>\s*[^<]+<a\s+[^>]*href="ratings"[^>]+>([0-9,]+) votes?<\/a>/m)
- if m
- ratings = "#{m[1]}/10 (#{m[2]} voters)"
+ # parse imdb rating value:
+ if resp.body.match(/itemprop="ratingValue">([^<]+)</)
+ ratings = "#{$1}/10"
+ end
+ # parse imdb rating count:
+ if resp.body.match(/itemprop="ratingCount">([^<]+)</)
+ ratings += " (#{$1} voters)"
end
genre = Array.new
- resp.body.scan(/<a href="\/genre\/[^"]+">([^<]+)<\/a>/) do |gnr|
- genre << gnr
+ resp.body.scan(/<a\s+href="\/genre\/[^\?]+\?[^"]+"\s+>([^<]+)<\/a>/) do |gnr|
+ genre << gnr.first.strip
end
plot = resp.body.match(DESC_MATCH)[3] rescue nil
}
end
- urls.map { |url|
- info = info_name(url, :movies_by_year => true)
+ urls.map { |u|
+ info = info_name(u, :movies_by_year => true)
debug info.inspect
end
def name_in_movie(name_urls, movie_urls)
+ debug name_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)
+ data.scan(CREDIT_NAME_MATCH).each { |url, name, role_data|
+ ch_url, role = role_data.scan(CHAR_MATCH).first
+ debug [ch_url, role]
+ wanted = name_urls.include?(url) || name_urls.include?(ch_url)
+ valid << [url, name.ircify_html, role.ircify_html] if wanted
}
valid.each { |url, name, role|
info << "%s : %s was %s in %s" % [name, IMDB + url, role, title_info]
# 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"
+ movie_urls = i.search(movie, :type => :title)
+ unless movie_urls
+ m.reply "movie #{who} not found, sorry"
return
end
- movie_urls = i.search(movie, :type => :title)
- unless movie_urls
+ if params[:actor]
+ who = params[:actor].to_s
+ type = :name
+ else
+ who = params[:character].to_s
+ type = :character
+ end
+
+ name_urls = i.search(who, :type => type, :all => true)
+ unless name_urls
m.reply "nothing found about #{who}, sorry"
return
end
plugin.map "imdb [:type] *what", :requirements => { :type => /name|title/ }, :defaults => { :type => 'both' }
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 "character [played] by *actor in *movie"
+plugin.map "character of *character in *movie"
plugin.map "characters in *movie", :action => :imdb