X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fsearch.rb;h=b48e937e6a9ed38d4d35126a2aa104752e530978;hb=63314f4987d7acc8a39a3680320e339cc94da66b;hp=27f9519c37b4d5e3bec5bbda080380526371bed9;hpb=6fe052181de09dd4a618241e600e2473f6706343;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git diff --git a/data/rbot/plugins/search.rb b/data/rbot/plugins/search.rb index 27f9519c..b48e937e 100644 --- a/data/rbot/plugins/search.rb +++ b/data/rbot/plugins/search.rb @@ -1,70 +1,194 @@ -require 'uri' +#-- vim:sw=2:et +#++ +# +# :title: Google and Wikipedia search plugin for rbot +# +# Author:: Tom Gilbert (giblet) +# Author:: Giuseppe "Oblomov" Bilotta +# +# Copyright:: (C) 2002-2005 Tom Gilbert +# Copyright:: (C) 2006 Tom Gilbert, Giuseppe Bilotta +# Copyright:: (C) 2006-2007 Giuseppe Bilotta -Net::HTTP.version_1_2 +# TODO:: use lr=lang_ or whatever is most appropriate to let google know +# it shouldn't use the bot's location to find the preferred language +# TODO:: support localized uncyclopedias -- not easy because they have different names +# for most languages -GOOGLE_WAP_LINK = /(.*?)<\/a>/im - -class ::String - def ircify_html - txt = self - - # bold and strong -> bold - txt.gsub!(/<\/?(?:b|strong)\s*>/, "#{Bold}") - - # italic, emphasis and underline -> underline - txt.gsub!(/<\/?(?:i|em|u)\s*>/, "#{Underline}") - - ## This would be a nice addition, but the results are horrible - ## Maybe make it configurable? - # txt.gsub!(/<\/?a( [^>]*)?>/, "#{Reverse}") - - # Paragraph and br tags are converted to whitespace. - txt.gsub!(/<\/?(p|br)\s*\/?\s*>/, ' ') - txt.gsub!("\n", ' ') - - # All other tags are just removed - txt.gsub!(/<[^>]+>/, '') - - # Remove double formatting options, since they only waste bytes - txt.gsub!(/#{Bold}\s*#{Bold}/,"") - txt.gsub!(/#{Underline}\s*#{Underline}/,"") - - # And finally whitespace is squeezed - txt.gsub!(/\s+/, ' ') - - # Decode entities and strip whitespace - return Utils.decode_html_entities(txt).strip! - end -end +GOOGLE_SEARCH = "https://www.google.com/search?hl=en&oe=UTF-8&ie=UTF-8&gbv=1&q=" +GOOGLE_WAP_SEARCH = "https://www.google.com/m/search?hl=en&ie=UTF-8&gbv=1&q=" +GOOGLE_WAP_LINK = /]*>\s*]*>(.*?)\s*<\/div>/im +GOOGLE_CALC_RESULT = />Calculator<\/span>(?:<\/?[^>]+>\s*)+([^<]+)/ +GOOGLE_XPATH_DEF = "//img[@id='flex_text_audio_icon_chunk']/../../../../div[3]//text()" +DDG_API_SEARCH = "http://api.duckduckgo.com/?format=xml&no_html=1&skip_disambig=1&no_redirect=0&q=" +WOLFRAM_API_SEARCH = "http://api.wolframalpha.com/v2/query?input=%{terms}&appid=%{key}&format=plaintext" + + "&scantimeout=3.0&podtimeout=4.0&formattimeout=8.0&parsetimeout=5.0" + + "&excludepodid=SeriesRepresentations:*" +WOLFRAM_API_KEY = "9RW6XR-QTL2JT7J4W" class SearchPlugin < Plugin - BotConfig.register BotConfigIntegerValue.new('google.hits', + Config.register Config::IntegerValue.new('duckduckgo.hits', + :default => 3, :validate => Proc.new{|v| v > 0}, + :desc => "Number of hits to return from searches") + Config.register Config::IntegerValue.new('duckduckgo.first_par', + :default => 0, + :desc => "When set to n > 0, the bot will return the first paragraph from the first n search hits") + Config.register Config::IntegerValue.new('google.hits', :default => 3, :desc => "Number of hits to return from Google searches") - BotConfig.register BotConfigIntegerValue.new('google.first_par', + Config.register Config::IntegerValue.new('google.first_par', :default => 0, :desc => "When set to n > 0, the bot will return the first paragraph from the first n search hits") - BotConfig.register BotConfigIntegerValue.new('wikipedia.hits', + Config.register Config::IntegerValue.new('wikipedia.hits', :default => 3, :desc => "Number of hits to return from Wikipedia searches") - BotConfig.register BotConfigIntegerValue.new('wikipedia.first_par', + Config.register Config::IntegerValue.new('wikipedia.first_par', :default => 1, :desc => "When set to n > 0, the bot will return the first paragraph from the first n wikipedia search hits") def help(plugin, topic="") case topic + when "ddg" + "Use '#{topic} ' to return a search or calculation from " + + "DuckDuckGo. Use #{topic} define to return a definition." when "search", "google" "#{topic} => search google for " + when "gcalc" + "gcalc => use the google calculator to find the answer to " + when "gdef" + "gdef => use the google define mechanism to find a definition of " + when "wa" + "wa => searches WolframAlpha for " when "wp" "wp [] => search for on Wikipedia. You can select a national to only search the national Wikipedia" + when "unpedia" + "unpedia => search for on Uncyclopedia" else - "search (or: google ) => search google for | wp => search for on Wikipedia" + "search (or: google ) => search google for | ddg to search DuckDuckGo | wp => search for on Wikipedia | wa => search for on WolframAlpha | unpedia => search for on Uncyclopedia" + end + end + + def duckduckgo(m, params) + what = params[:words].to_s + terms = CGI.escape what + url = DDG_API_SEARCH + terms + + hits = @bot.config['duckduckgo.hits'] + first_pars = params[:firstpar] || @bot.config['duckduckgo.first_par'] + single = params[:lucky] || (hits == 1 and first_pars == 1) + + begin + feed = @bot.httputil.get(url) + raise unless feed + rescue => e + m.reply "error duckduckgoing for #{what}" + return + end + + xml = REXML::Document.new feed + heading = xml.elements['//Heading/text()'].to_s + # answer is returned for calculations + answer = xml.elements['//Answer/text()'].to_s + # abstract is returned for definitions etc + abstract = xml.elements['//AbstractText/text()'].to_s + abfrom = "" + unless abstract.empty? + absrc = xml.elements['//AbstractSource/text()'].to_s + aburl = xml.elements['//AbstractURL/text()'].to_s + unless absrc.empty? and aburl.empty? + abfrom = " --" + abfrom << " " << absrc unless absrc.empty? + abfrom << " " << aburl unless aburl.empty? + end + end + + # but also definition (yes, you can have both, see e.g. printf) + definition = xml.elements['//Definition/text()'].to_s + deffrom = "" + unless definition.empty? + defsrc = xml.elements['//Definition/@source/text()'].to_s + defurl = xml.elements['//Definition/@url/text()'].to_s + unless defsrc.empty? and defurl.empty? + deffrom = " --" + deffrom << " " << defsrc unless defsrc.empty? + deffrom << " " << defurl unless defurl.empty? + end + end + + if heading.empty? and answer.empty? and abstract.empty? and definition.empty? + m.reply "no results" + return + end + + # if we got a one-shot answer (e.g. a calculation, return it) + unless answer.empty? + m.reply answer + return + end + + # otherwise, return the abstract, followed by as many hits as found + unless heading.empty? or abstract.empty? + m.reply "%{bold}%{heading}:%{bold} %{abstract}%{abfrom}" % { + :bold => Bold, :heading => heading, + :abstract => abstract, :abfrom => abfrom + } end + unless heading.empty? or definition.empty? + m.reply "%{bold}%{heading}:%{bold} %{abstract}%{abfrom}" % { + :bold => Bold, :heading => heading, + :abstract => definition, :abfrom => deffrom + } + end + # return zeroclick search results + links, texts = [], [] + xml.elements.each("//Results/Result/FirstURL") { |element| + links << element.text + break if links.size == hits + } + return if links.empty? + + xml.elements.each("//Results/Result/Text") { |element| + texts << " #{element.text}" + break if links.size == hits + } + # TODO see treatment of `single` in google search + + single ||= (links.length == 1) + pretty = [] + links.each_with_index do |u, i| + t = texts[i] + pretty.push("%{n}%{b}%{t}%{b}%{sep}%{u}" % { + :n => (single ? "" : "#{i}. "), + :sep => (single ? " -- " : ": "), + :b => Bold, :t => t, :u => u + }) + end + + result_string = pretty.join(" | ") + + # If we return a single, full result, change the output to a more compact representation + if single + fp = first_pars > 0 ? " -- #{Utils.get_first_pars(links, first_pars)}" : "" + m.reply("Result for %{what}: %{string}%{fp}" % { + :what => what, :string => result_string, :fp => fp + }, :overlong => :truncate) + return + end + + m.reply "Results for #{what}: #{result_string}", :split_at => /\s+\|\s+/ + + return unless first_pars > 0 + + Utils.get_first_pars urls, first_pars, :message => m end def google(m, params) what = params[:words].to_s - searchfor = URI.escape what + if what.match(/^define:/) + return google_define(m, what, params) + end + + searchfor = CGI.escape what # This method is also called by other methods to restrict searching to some sites if params[:site] site = "site:#{params[:site]}+" @@ -75,79 +199,182 @@ class SearchPlugin < Plugin # e.g.: "Wikipedia, the free encyclopedia" when doing Wikipedia searches filter = params[:filter] || "" - url = "http://www.google.com/wml/search?q=#{site}#{searchfor}" + url = GOOGLE_WAP_SEARCH + site + searchfor hits = params[:hits] || @bot.config['google.hits'] + hits = 1 if params[:lucky] + + first_pars = params[:firstpar] || @bot.config['google.first_par'] + + single = params[:lucky] || (hits == 1 and first_pars == 1) begin - wml = @bot.httputil.get_cached(url) + wml = @bot.httputil.get(url) + raise unless wml rescue => e m.reply "error googling for #{what}" return end results = wml.scan(GOOGLE_WAP_LINK) + if results.length == 0 m.reply "no results found for #{what}" return end - urls = Array.new - results = results[0...hits].map { |res| - n = res[0] - t = Utils.decode_html_entities res[2].gsub(filter, '').strip - u = URI.unescape res[1] - urls.push(u) - "#{n}. #{Bold}#{t}#{Bold}: #{u}" - }.join(" | ") - m.reply "Results for #{what}: #{results}", :split_at => /\s+\|\s+/ + results = results.map {|result| + url = CGI::parse(Utils.decode_html_entities(result[0]))['q'].first + title = Utils.decode_html_entities(result[1].gsub(/<\/?[^>]+>/, '')) + [url, title] unless url.empty? or title.empty? + }.reject {|item| not item}[0..hits] - first_pars = params[:firstpar] || @bot.config['google.first_par'] + result_string = results.map {|url, title| "#{Bold}#{title}#{NormalText}: #{url}"} - idx = 0 - while first_pars > 0 and urls.length > 0 - url.replace(urls.shift) - idx += 1 - - # FIXME what happens if some big file is returned? We should share - # code with the url plugin to only retrieve partial file content! - xml = @bot.httputil.get_cached(url) - if xml.nil? - debug "Unable to retrieve #{url}" - next - end - # We get the first par after the first main heading, if possible - header_found = xml.match(/]*)?>(.*?)<\/h1>/im) - txt = String.new - if header_found - debug "Found header: #{header_found[1].inspect}" - while txt.empty? - header_found = $' - candidate = header_found[/]*)?>.*?<\/p>/im] - break unless candidate - txt.replace candidate.ircify_html - end - end - # If we haven't found a first par yet, try to get it from the whole - # document - if txt.empty? - header_found = xml - while txt.empty? - candidate = header_found[/]*)?>.*?<\/p>/im] - break unless candidate - txt.replace candidate.ircify_html - header_found = $' - end - end - # Nothing yet, try title - if txt.empty? - debug "No first par found\n#{xml}" - # FIXME only do this if the 'url' plugin is loaded - txt.replace @bot.plugins['url'].get_title_from_html(xml) - next if txt.empty? - end - m.reply "[#{idx}] #{txt}", :overlong => :truncate - first_pars -=1 + if params[:lucky] + m.reply result_string.first + Utils.get_first_pars([results.map {|url, title| url}.first], first_pars, :message => m) + return end + + m.reply "Results for #{what}: #{result_string.join(' | ')}", :split_at => /\s+\|\s+/ + + return unless first_pars > 0 + + Utils.get_first_pars(results.map {|url, title| url}, first_pars, :message => m) + end + + def google_define(m, what, params) + begin + wml = @bot.httputil.get(GOOGLE_SEARCH + CGI.escape(what)) + raise unless wml + rescue => e + m.reply "error googling for #{what}" + return + end + + begin + related_index = wml.index(/Related phrases:/, 0) + raise unless related_index + defs_index = wml.index(/Definitions of /, related_index) + raise unless defs_index + defs_end = wml.index(/ e + m.reply "no results found for #{what}" + return + end + + related = wml[related_index...defs_index] + defs = wml[defs_index...defs_end] + + m.reply defs.ircify_html(:a_href => Underline), :split_at => (Underline + ' ') + + end + + def lucky(m, params) + params.merge!(:lucky => true) + google(m, params) + end + + def gcalc(m, params) + what = params[:words].to_s + searchfor = CGI.escape(what) + + debug "Getting gcalc thing: #{searchfor.inspect}" + url = GOOGLE_WAP_SEARCH + searchfor + + begin + html = @bot.httputil.get(url) + rescue => e + m.reply "error googlecalcing #{what}" + return + end + + debug "#{html.size} bytes of html recieved" + debug html + + candidates = html.match(GOOGLE_CALC_RESULT) + debug "candidates: #{candidates.inspect}" + + if candidates.nil? + m.reply "couldn't calculate #{what}" + return + end + result = candidates[1].remove_nonascii + + debug "replying with: #{result.inspect}" + m.reply result.ircify_html + end + + def gdef(m, params) + what = params[:words].to_s + searchfor = CGI.escape("define " + what) + + debug "Getting gdef thing: #{searchfor.inspect}" + url = GOOGLE_WAP_SEARCH + searchfor + + begin + resp = @bot.httputil.get(url, resp: true) + rescue => e + m.reply "error googledefining #{what}" + return + end + + results = resp.xpath(GOOGLE_XPATH_DEF).map(&:content) + + if results.empty? + m.reply "couldn't find a definition for #{what} on Google" + else + m.reply "#{results.first} -- #{results[1..-1].join(' ')}" + end + end + + def wolfram(m, params) + what = params[:words].to_s + terms = CGI.escape what + url = WOLFRAM_API_SEARCH % { + :terms => terms, :key => WOLFRAM_API_KEY + } + + begin + feed = @bot.httputil.get(url) + raise unless feed + rescue => e + m.reply "error asking WolframAlfa about #{what}" + return + end + debug feed + + xml = REXML::Document.new feed + if xml.elements['/queryresult'].attributes['error'] == "true" + m.reply xml.elements['/queryresult/error/text()'].to_s + return + end + unless xml.elements['/queryresult'].attributes['success'] == "true" + m.reply "no data available" + return + end + answer_type, answer = [], [] + xml.elements.each("//pod") { |element| + answer_type << element.attributes['title'] + answer << element.elements['subpod/plaintext'].text + } + # find the first answer that isn't nil, + # starting on the second pod in the array + n = 1 + answer[1..-1].each { |a| + break unless a.nil? + n += 1 + } + if answer[n].nil? + m.reply "no results" + return + end + # strip spaces, pipes, and line breaks + sep = Bold + ' :: ' + Bold + chars = [ [/\n/, sep], [/\t/, " "], [/\s+/, " "], ["|", "-"] ] + chars.each { |c| answer[n].gsub!(c[0], c[1]) } + m.reply answer_type[n] + sep + answer[n] end def wikipedia(m, params) @@ -160,12 +387,27 @@ class SearchPlugin < Plugin params[:firstpar] = @bot.config['wikipedia.first_par'] return google(m, params) end + + def unpedia(m, params) + site = "uncyclopedia.org" + debug "Looking up things on #{site}" + params[:site] = site + params[:filter] = / - Uncyclopedia.*$/ + params[:hits] = @bot.config['wikipedia.hits'] + params[:firstpar] = @bot.config['wikipedia.first_par'] + return google(m, params) + end end plugin = SearchPlugin.new -plugin.map "search *words", :action => 'google' -plugin.map "google *words", :action => 'google' -plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ } -plugin.map "wp *words", :action => 'wikipedia' - +plugin.map "ddg *words", :action => 'duckduckgo', :threaded => true +plugin.map "search *words", :action => 'google', :threaded => true +plugin.map "google *words", :action => 'google', :threaded => true +plugin.map "lucky *words", :action => 'lucky', :threaded => true +plugin.map "gcalc *words", :action => 'gcalc', :threaded => true +plugin.map "gdef *words", :action => 'gdef', :threaded => true +plugin.map "wa *words", :action => 'wolfram', :threaded => true +plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ }, :threaded => true +plugin.map "wp *words", :action => 'wikipedia', :threaded => true +plugin.map "unpedia *words", :action => 'unpedia', :threaded => true