basicxman · May 27, 2011 03:51 · May 27, 2011 · May 27, 2011 · May 27, 2011 · May 27, 2011
diff --git a/run.rb b/run.rb
@@ -76,23 +76,13 @@ def get_pages_first_link(url)
     valid_links.first.attr("href")
   end
 
-  # Using .include? is a little inefficient as going in reverse order would be
-  # faster in this case.
-  def has_recursed?
-    last = @breadcrumbs.last
-    (@breadcrumbs.length - 2).downto(0) do |i|
-      return true if @breadcrumbs[i] == last
-    end
-    return false
-  end
-
   def has_reached_end?
     return false if @breadcrumbs.length <= 2
     if @breadcrumbs.last == "/wiki/Philosophy"
       @reached_end = true
       return true
     end
-    return has_recursed?
+    return @breadcrumbs[0..-2].include? @breadcrumbs.last
   end
 
   def add_crumb(url)

diff --git a/run.rb b/run.rb
@@ -76,13 +76,23 @@ def get_pages_first_link(url)
     valid_links.first.attr("href")
   end
 
+  # Using .include? is a little inefficient as going in reverse order would be
+  # faster in this case.
+  def has_recursed?
+    last = @breadcrumbs.last
+    (@breadcrumbs.length - 2).downto(0) do |i|
+      return true if @breadcrumbs[i] == last
+    end
+    return false
+  end
+
   def has_reached_end?
     return false if @breadcrumbs.length <= 2
     if @breadcrumbs.last == "/wiki/Philosophy"
       @reached_end = true
       return true
     end
-    return @breadcrumbs[0..-2].include? @breadcrumbs.last
+    return has_recursed?
   end
 
   def add_crumb(url)

diff --git a/run.rb b/run.rb
@@ -49,8 +49,9 @@ def remove_bracket_links(text)
         end
       end
     end
+    return temp if temp[last_bracket + 1] == '"' # First brackets might be within a valid link.
     temp[first_bracket..last_bracket] = ""
-    temp
+    return temp
   end
 
   def print_results

diff --git a/run.rb b/run.rb
@@ -67,7 +67,8 @@ def get_pages_first_link(url)
     page = Nokogiri::HTML(open(url))
     content = page.css("#bodyContent")
 
-    content.css(".dablink", ".navbox", ".tocolours", ".image").each(&:remove)   # Anchors with a .dablink parent are exlucded.
+    # Remove invalid links.
+    content.css(".dablink", ".navbox", ".tocolours", ".image").each(&:remove)
     content.css("p").first.inner_html = remove_bracket_links content.css("p").first.inner_html
 
     valid_links = content.css("p > a")

diff --git a/run.rb b/run.rb
@@ -67,10 +67,7 @@ def get_pages_first_link(url)
     page = Nokogiri::HTML(open(url))
     content = page.css("#bodyContent")
 
-    content.css(".dablink").each(&:remove)   # Anchors with a .dablink parent are exlucded.
-    content.css(".navbox").each(&:remove)    # Remove navigation box.
-    content.css(".tocolours").each(&:remove) # Remove table of contents side box.
-    content.css(".image").each(&:remove)     # Image links :/
+    content.css(".dablink", ".navbox", ".tocolours", ".image").each(&:remove)   # Anchors with a .dablink parent are exlucded.
     content.css("p").first.inner_html = remove_bracket_links content.css("p").first.inner_html
 
     valid_links = content.css("p > a")

diff --git a/run.rb b/run.rb
@@ -32,24 +32,24 @@ def unwikify(page)
   def remove_bracket_links(text)
     temp = text
     first_bracket = temp.index "("
+    last_bracket = temp.length - 1
 
     return temp if first_bracket.nil?
 
     done = false
     opening_brackets = 0
-    index = first_bracket
-    until done
+    (first_bracket..last_bracket).each do |index|
       if temp[index] == "("
         opening_brackets += 1
       elsif temp[index] == ")"
         opening_brackets -= 1
         if opening_brackets == 0
+          last_bracket = index
           break
         end
       end
-      index += 1
     end
-    temp[first_bracket..index] = ""
+    temp[first_bracket..last_bracket] = ""
     temp
   end
 

diff --git a/run.rb b/run.rb
@@ -31,7 +31,7 @@ def unwikify(page)
 
   def remove_bracket_links(text)
     temp = text
-    first_bracket    = temp.index "("
+    first_bracket = temp.index "("
 
     return temp if first_bracket.nil?
 

diff --git a/run.rb b/run.rb
@@ -31,8 +31,25 @@ def unwikify(page)
 
   def remove_bracket_links(text)
     temp = text
-    temp.gsub! /\<span.*?\<\/span\>/, ""
-    temp.gsub! /^.*?\)/, "" 
+    first_bracket    = temp.index "("
+
+    return temp if first_bracket.nil?
+
+    done = false
+    opening_brackets = 0
+    index = first_bracket
+    until done
+      if temp[index] == "("
+        opening_brackets += 1
+      elsif temp[index] == ")"
+        opening_brackets -= 1
+        if opening_brackets == 0
+          break
+        end
+      end
+      index += 1
+    end
+    temp[first_bracket..index] = ""
     temp
   end
 

diff --git a/run.rb b/run.rb
@@ -41,15 +41,19 @@ def print_results
     puts "[#{@breadcrumbs.length}] " + @breadcrumbs.map { |crumb| unwikify(crumb) }.join(" -> ")
   end
 
+  def remove(elm)
+    elm.remove
+  end
+
   def get_pages_first_link(url)
     puts url
     page = Nokogiri::HTML(open(url))
     content = page.css("#bodyContent")
 
-    content.css(".dablink").each   { |elm| elm.remove } # Anchors with a .dablink parent are exlucded.
-    content.css(".navbox").each    { |elm| elm.remove } # Remove navigation box.
-    content.css(".tocolours").each { |elm| elm.remove } # Remove table of contents side box.
-    content.css(".image").each     { |elm| elm.remove } # Image links :/
+    content.css(".dablink").each(&:remove)   # Anchors with a .dablink parent are exlucded.
+    content.css(".navbox").each(&:remove)    # Remove navigation box.
+    content.css(".tocolours").each(&:remove) # Remove table of contents side box.
+    content.css(".image").each(&:remove)     # Image links :/
     content.css("p").first.inner_html = remove_bracket_links content.css("p").first.inner_html
 
     valid_links = content.css("p > a")
@@ -85,4 +89,4 @@ def crawl
 
 end
 
-go = ExtendedMind.new
+go = ExtendedMind.new
diff --git a/run.rb b/run.rb
@@ -0,0 +1,88 @@
+#!/usr/bin/env ruby
+
+# Extended Mind
+# Wikipedia checker, the concept is that for any Wikipedia article you can
+# eventually get to the article on Philosophy if you click the first link
+# on the article.
+# http://xkcd.com/903 (see alt text)
+
+require 'open-uri'
+require 'nokogiri'
+
+class ExtendedMind
+
+  def initialize
+    @start_page  = ARGV[0]
+    @breadcrumbs = []
+
+    start
+    crawl
+
+    print_results
+  end
+
+  def wikify(page)
+    "http://wikipedia.org" + page
+  end
+
+  def unwikify(page)
+    page.gsub("/wiki/", "") 
+  end
+
+  def remove_bracket_links(text)
+    temp = text
+    temp.gsub! /\<span.*?\<\/span\>/, ""
+    temp.gsub! /^.*?\)/, "" 
+    temp
+  end
+
+  def print_results
+    puts "\n[-] Unable to check, recurses.\n" if @reached_end.nil?
+    puts "[#{@breadcrumbs.length}] " + @breadcrumbs.map { |crumb| unwikify(crumb) }.join(" -> ")
+  end
+
+  def get_pages_first_link(url)
+    puts url
+    page = Nokogiri::HTML(open(url))
+    content = page.css("#bodyContent")
+
+    content.css(".dablink").each   { |elm| elm.remove } # Anchors with a .dablink parent are exlucded.
+    content.css(".navbox").each    { |elm| elm.remove } # Remove navigation box.
+    content.css(".tocolours").each { |elm| elm.remove } # Remove table of contents side box.
+    content.css(".image").each     { |elm| elm.remove } # Image links :/
+    content.css("p").first.inner_html = remove_bracket_links content.css("p").first.inner_html
+
+    valid_links = content.css("p > a")
+    valid_links.first.attr("href")
+  end
+
+  def has_reached_end?
+    return false if @breadcrumbs.length <= 2
+    if @breadcrumbs.last == "/wiki/Philosophy"
+      @reached_end = true
+      return true
+    end
+    return @breadcrumbs[0..-2].include? @breadcrumbs.last
+  end
+
+  def add_crumb(url)
+    @breadcrumbs << get_pages_first_link(wikify(url))
+  end
+
+  def start
+    begin
+      add_crumb("/wiki/#{@start_page}")
+    rescue
+      abort "Invalid start!"
+    end
+  end
+
+  def crawl
+    until has_reached_end?
+      add_crumb(@breadcrumbs.last)
+    end
+  end
+
+end
+
+go = ExtendedMind.new
No results found