X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fsearch.rb;h=b48e937e6a9ed38d4d35126a2aa104752e530978;hb=63314f4987d7acc8a39a3680320e339cc94da66b;hp=3d9e5d3d354b95dd30b77ac02ef9cf6f3893db5d;hpb=ab9f1a23b6808b05df2a51c17dd3f83bbc8a8863;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git diff --git a/data/rbot/plugins/search.rb b/data/rbot/plugins/search.rb index 3d9e5d3d..b48e937e 100644 --- a/data/rbot/plugins/search.rb +++ b/data/rbot/plugins/search.rb @@ -15,20 +15,24 @@ # TODO:: support localized uncyclopedias -- not easy because they have different names # for most languages -GOOGLE_SEARCH = "http://www.google.com/search?oe=UTF-8&q=" -GOOGLE_WAP_SEARCH = "http://www.google.com/m/search?hl=en&q=" -GOOGLE_WAP_LINK = /"r">(?:]*>)?]*>(.*?)<\/a>/im -GOOGLE_CALC_RESULT = %r{]*>(.+?)Results 1<\/b> - 10<\/b> of about (.*)<\/b> for} -GOOGLE_DEF_RESULT = %r{onebox_result">\s*(.*?)\s*
\s*(.*?)]+>(.+?)<(br|/td)>} - -DDG_API_SEARCH = "http://api.duckduckgo.com/?format=xml&no_html=1&no_redirect=0&q=" +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 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") @@ -53,14 +57,14 @@ class SearchPlugin < Plugin "gcalc => use the google calculator to find the answer to " when "gdef" "gdef => use the google define mechanism to find a definition of " - when "gtime" - "gtime => use the google clock to find the current time at " + 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 | ddg to search DuckDuckGo | wp => search for on Wikipedia | unpedia => search for on Uncyclopedia" + "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 @@ -68,6 +72,11 @@ class SearchPlugin < Plugin 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 @@ -75,45 +84,102 @@ class SearchPlugin < Plugin m.reply "error duckduckgoing for #{what}" return end - debug feed xml = REXML::Document.new feed heading = xml.elements['//Heading/text()'].to_s # answer is returned for calculations answer = xml.elements['//Answer/text()'].to_s - if heading.empty? and answer.empty? + # 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 terms =~ /^define/ - if heading.empty? - m.reply "no definition found" - return - end - # Format and return a different string if it is a definition search. - definition = xml.elements['//AbstractText/text()'].to_s - source = " -- #{xml.elements['//AbstractURL/text()']}" - m.reply Bold + heading + ": " + Bold + definition + source - elsif heading.empty? - # return a calculation + + # if we got a one-shot answer (e.g. a calculation, return it) + unless answer.empty? m.reply answer - else - # else, return a zeroclick search - links, text = [], [] - hits = @bot.config['duckduckgo.hits'] - xml.elements.each("//RelatedTopics/RelatedTopic/FirstURL") { |element| - links << element.text + 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 } - xml.elements.each("//RelatedTopics/RelatedTopic/Text") { |element| - text << " #{element.text}" + end + unless heading.empty? or definition.empty? + m.reply "%{bold}%{heading}:%{bold} %{abstract}%{abfrom}" % { + :bold => Bold, :heading => heading, + :abstract => definition, :abfrom => deffrom } - num = 0 - m.reply Bold + heading + ": " + Bold - until num >= hits - m.reply links[num] + text[num] - num += 1 - end 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) @@ -156,60 +222,25 @@ class SearchPlugin < Plugin return end - single ||= (results.length==1) - pretty = [] + 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] - begin - urls = Array.new - - debug results - results.each do |res| - t = res[1].ircify_html(:img => "[%{src} %{alt} %{dimensions}]").strip - u = res[0] - if u.sub!(%r{^http://www.google.com/aclk\?},'') - u = CGI::parse(u)['adurl'].first - debug "skipping ad for #{u}" - next - elsif u.sub!(%r{^http://www.google.com/gwt/x\?},'') - u = CGI::parse(u)['u'].first - elsif u.sub!(%r{^/url\?},'') - u = CGI::parse(u)['q'].first - end - urls.push(u) - pretty.push("%{n}%{b}%{t}%{b}%{sep}%{u}" % { - :n => (single ? "" : "#{urls.length}. "), - :sep => (single ? " -- " : ": "), - :b => Bold, :t => t, :u => u - }) - break if urls.length == hits - end - rescue => e - m.reply "failed to understand what google found for #{what}" - error e - debug wml - debug results - return - end + result_string = results.map {|url, title| "#{Bold}#{title}#{NormalText}: #{url}"} if params[:lucky] - m.reply pretty.first - return - end - - result_string = pretty.join(" | ") - - # If we return a single, full result, change the output to a more compact representation - if single - m.reply "Result for %s: %s -- %s" % [what, result_string, Utils.get_first_pars(urls, first_pars)], :overlong => :truncate + 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}", :split_at => /\s+\|\s+/ + m.reply "Results for #{what}: #{result_string.join(' | ')}", :split_at => /\s+\|\s+/ return unless first_pars > 0 - Utils.get_first_pars urls, first_pars, :message => m - + Utils.get_first_pars(results.map {|url, title| url}, first_pars, :message => m) end def google_define(m, what, params) @@ -269,68 +300,81 @@ class SearchPlugin < Plugin m.reply "couldn't calculate #{what}" return end - result = candidates[1] + result = candidates[1].remove_nonascii debug "replying with: #{result.inspect}" m.reply result.ircify_html end - def gcount(m, params) + def gdef(m, params) what = params[:words].to_s - searchfor = CGI.escape(what) + searchfor = CGI.escape("define " + what) - debug "Getting gcount thing: #{searchfor.inspect}" - url = GOOGLE_SEARCH + searchfor + debug "Getting gdef thing: #{searchfor.inspect}" + url = GOOGLE_WAP_SEARCH + searchfor begin - html = @bot.httputil.get(url) + resp = @bot.httputil.get(url, resp: true) rescue => e - m.reply "error googlecounting #{what}" + m.reply "error googledefining #{what}" return end - debug "#{html.size} bytes of html recieved" - - results = html.scan(GOOGLE_COUNT_RESULT) - debug "results: #{results.inspect}" + results = resp.xpath(GOOGLE_XPATH_DEF).map(&:content) - if results.length != 1 - m.reply "couldn't count #{what}" - return + if results.empty? + m.reply "couldn't find a definition for #{what} on Google" + else + m.reply "#{results.first} -- #{results[1..-1].join(' ')}" end - - result = results[0][0].ircify_html - debug "replying with: #{result.inspect}" - m.reply "total results: #{result}" - end - def gdef(m, params) + def wolfram(m, params) what = params[:words].to_s - searchfor = CGI.escape("define " + what) - - debug "Getting gdef thing: #{searchfor.inspect}" - url = GOOGLE_WAP_SEARCH + searchfor + terms = CGI.escape what + url = WOLFRAM_API_SEARCH % { + :terms => terms, :key => WOLFRAM_API_KEY + } begin - html = @bot.httputil.get(url) + feed = @bot.httputil.get(url) + raise unless feed rescue => e - m.reply "error googledefining #{what}" + m.reply "error asking WolframAlfa about #{what}" return end + debug feed - debug html - results = html.scan(GOOGLE_DEF_RESULT) - debug "results: #{results.inspect}" - - if results.length != 1 - m.reply "couldn't find a definition for #{what} on Google" + xml = REXML::Document.new feed + if xml.elements['/queryresult'].attributes['error'] == "true" + m.reply xml.elements['/queryresult/error/text()'].to_s return end - - head = results[0][0].ircify_html - text = results[0][1].ircify_html - m.reply "#{head} -- #{text}" + 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) @@ -353,44 +397,17 @@ class SearchPlugin < Plugin params[:firstpar] = @bot.config['wikipedia.first_par'] return google(m, params) end - - def gtime(m, params) - where = params[:words].to_s - where.sub!(/^\s*in\s*/, '') - searchfor = CGI.escape("time in " + where) - url = GOOGLE_SEARCH + searchfor - - begin - html = @bot.httputil.get(url) - rescue => e - m.reply "Error googletiming #{where}" - return - end - - debug html - results = html.scan(GOOGLE_TIME_RESULT) - debug "results: #{results.inspect}" - - if results.length != 1 - m.reply "Couldn't find the time for #{where} on Google" - return - end - - time = results[0][0].ircify_html - m.reply "#{time}" - end end plugin = SearchPlugin.new -plugin.map "ddg *words", :action => 'duckduckgo' +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 "gcount *words", :action => 'gcount', :threaded => true plugin.map "gcalc *words", :action => 'gcalc', :threaded => true plugin.map "gdef *words", :action => 'gdef', :threaded => true -plugin.map "gtime *words", :action => 'gtime', :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