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 = /<a accesskey="(\d)" href=".*?u=(.*?)">(.*?)<\/a>/im
-GOOGLE_WAP_LINK = /<a href="(?:.*?u=(.*?)|(http:\/\/.*?))">(.*?)<\/a>/im
-GOOGLE_CALC_RESULT = %r{<img src=/images/calc_img\.gif(?: width=40 height=30 alt="")?>.*?<h2 class=r[^>]*><b>(.+?)</b>}
+GOOGLE_WAP_LINK = /"r">(?:<div[^>]*>)?<a href="([^"]+)"[^>]*>(.*?)<\/a>/im
+GOOGLE_CALC_RESULT = %r{<h[1-6] class="r" [^>]*>(.+?)</h}
GOOGLE_COUNT_RESULT = %r{<font size=-1>Results <b>1<\/b> - <b>10<\/b> of about <b>(.*)<\/b> for}
-GOOGLE_DEF_RESULT = %r{<a href="([^"]*)"[^>]*>(Web definitions for .*?)<br/>(.*?)<br/>(.*?)\s-\s+<a href}
+GOOGLE_DEF_RESULT = %r{onebox_result">\s*(.*?)\s*<br/>\s*(.*?)<table}
GOOGLE_TIME_RESULT = %r{alt="Clock"></td><td valign=[^>]+>(.+?)<(br|/td)>}
+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 = "4EU37Y-TX9WJG3JH3"
+
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")
def help(plugin, topic="")
case topic
+ when "ddg"
+ "Use '#{topic} <string>' to return a search or calculation from " +
+ "DuckDuckGo. Use #{topic} define <string> to return a definition."
when "search", "google"
"#{topic} <string> => search google for <string>"
when "gcalc"
"gdef <term(s)> => use the google define mechanism to find a definition of <term(s)>"
when "gtime"
"gtime <location> => use the google clock to find the current time at <location>"
+ when "wa"
+ "wa <string> => searches WolframAlpha for <string>"
when "wp"
"wp [<code>] <string> => search for <string> on Wikipedia. You can select a national <code> to only search the national Wikipedia"
when "unpedia"
"unpedia <string> => search for <string> on Uncyclopedia"
else
- "search <string> (or: google <string>) => search google for <string> | wp <string> => search for <string> on Wikipedia | unpedia <string> => search for <string> on Uncyclopedia"
+ "search <string> (or: google <string>) => search google for <string> | ddg <string> to search DuckDuckGo | wp <string> => search for <string> on Wikipedia | wa <string> => search for <string> on WolframAlpha | unpedia <string> => search for <string> 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
+ 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
+ # 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
+ 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]
return
end
results = wml.scan(GOOGLE_WAP_LINK)
+
if results.length == 0
m.reply "no results found for #{what}"
return
end
+
single ||= (results.length==1)
- urls = Array.new
- n = 0
- results = results[0...hits].map { |res|
- n += 1
- t = Utils.decode_html_entities res[2].gsub(filter, '').strip
- u = URI.unescape(res[0] || res[1])
- urls.push(u)
- "%{n}%{b}%{t}%{b}%{sep}%{u}" % {
- :n => (single ? "" : "#{n}. "),
- :sep => (single ? " -- " : ": "),
- :b => Bold, :t => t, :u => u
- }
- }.join(" | ")
+ pretty = []
+
+ 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
if params[:lucky]
- m.reply results.first
+ 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, results, Utils.get_first_pars(urls, first_pars)], :overlong => :truncate
+ m.reply "Result for %s: %s -- %s" % [what, result_string, Utils.get_first_pars(urls, first_pars)], :overlong => :truncate
return
end
- m.reply "Results for #{what}: #{results}", :split_at => /\s+\|\s+/
+ m.reply "Results for #{what}: #{result_string}", :split_at => /\s+\|\s+/
return unless first_pars > 0
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 <b>/, related_index)
+ raise unless defs_index
+ defs_end = wml.index(/<input/, defs_index)
+ raise unless defs_end
+ rescue => 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)
searchfor = CGI.escape(what)
debug "Getting gcalc thing: #{searchfor.inspect}"
- url = GOOGLE_SEARCH + searchfor
+ url = GOOGLE_WAP_SEARCH + searchfor
begin
html = @bot.httputil.get(url)
end
debug "#{html.size} bytes of html recieved"
+ debug html
- results = html.scan(GOOGLE_CALC_RESULT)
- debug "results: #{results.inspect}"
+ candidates = html.match(GOOGLE_CALC_RESULT)
+ debug "candidates: #{candidates.inspect}"
- if results.length != 1
+ if candidates.nil?
m.reply "couldn't calculate #{what}"
return
end
+ result = candidates[1]
- result = results[0][0].ircify_html
debug "replying with: #{result.inspect}"
- m.reply "#{result}"
+ m.reply result.ircify_html
end
def gcount(m, params)
return
end
- gdef_link = "http://www.google.com" + CGI.unescapeHTML(results[0][0]) # could be used to extract all defs
- head = results[0][1].ircify_html
- text = results[0][2].ircify_html
- link = results[0][3]
- m.reply "#{head} -- #{link}\n#{text}"
+ head = results[0][0].ircify_html
+ text = results[0][1].ircify_html
+ m.reply "#{head} -- #{text}"
+ end
- ### gdef_link could be used for something like
- # html_defs = @bot.httputil.get(gdef_link)
- # related_index = html_defs.index(/Related phrases:/, 0)
- # defs_index = html_defs.index(/Definitions of <b>/, related_index)
+ def wolfram(m, params)
+ what = params[:words].to_s
+ terms = CGI.escape what
+ url = WOLFRAM_API_SEARCH % {
+ :terms => terms, :key => WOLFRAM_API_KEY
+ }
- # related = html_defs[related_index..defs_index]
- # defs = html_defs[defs_index..-1]
+ begin
+ feed = @bot.httputil.get(url)
+ raise unless feed
+ rescue => e
+ m.reply "error asking WolframAlfa about #{what}"
+ return
+ end
+ debug feed
- # m.reply defs.gsub(' <br/>','<li>').ircify_html
+ 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 = []
+ xml.elements.each("//pod/subpod/plaintext") { |element|
+ answer << element.text
+ }
+ # strip spaces and line breaks
+ answer[1].gsub!(/\n/, Bold + ' :: ' + Bold )
+ answer[1].gsub!(/\t/, ' ')
+ answer[1].gsub!(/\s+/, ' ')
+ m.reply answer[1].to_s
end
def wikipedia(m, params)
plugin = SearchPlugin.new
+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 "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
-
projects / user / henk / code / ruby / rbot.git / blobdiff