introduce URI::Generic#path_with_query

lib/anemone/cli/url_list.rb

@@ -32,7 +32,7 @@
 
   anemone.on_every_page do |page|
     if options.relative
-      puts page.url.path
+      puts page.url.path_with_query
     else
       puts page.url
     end

lib/anemone/core.rb

@@ -1,4 +1,4 @@
-require 'net/http'
+require 'uri'
 require 'thread'
 require 'anemone/tentacle'
 require 'anemone/page_hash'
@@ -13,15 +13,19 @@ class Core
     # and optional *block*
     #
     def initialize(urls, &block)
-      @urls = [urls].flatten.map{ |url| URI(url) if url.is_a?(String) }
-      @urls.each{ |url| url.path = '/' if url.path.empty? }
+      @urls = Array(urls).map do |url|
+        url = URI(url) if String === url
+        url.path = '/' if url.path.empty?
+        url
+      end
 
       @tentacles = []
       @pages = PageHash.new
       @on_every_page_blocks = []
       @on_pages_like_blocks = Hash.new { |hash,key| hash[key] = [] }
       @skip_link_patterns = []
       @after_crawl_blocks = []
+      @focus_crawl_block = nil
 
       if Anemone.options.obey_robots_txt
         @robots = Robots.new(Anemone.options.user_agent)
@@ -218,3 +222,9 @@ def skip_link?(link)
 
   end
 end
+
+URI::Generic.class_eval do
+  def path_with_query
+    self.path + (self.query ? '?' + self.query : '')
+  end
+end

lib/anemone/http.rb

@@ -32,9 +32,7 @@ def self.get(url, referer = nil)
     #
     def self.get_response(url, headers = {})
       Net::HTTP.start(url.host, url.port) do |http|
-        path = url.path
-        path << '?' << url.query if url.query
-        http.get(path, headers)
+        http.get(url.path_with_query, headers)
       end
     end
   end

lib/anemone/page_hash.rb

@@ -67,7 +67,7 @@ def uniq
         page_added = page.aliases.inject(false) { |r, a| r ||= results.has_key? a}
         if !page.redirect? and !page_added
           results[url] = page.clone 
-          results[url].aliases = []
+          results[url].aliases.clear
         end
       end
 

spec/page_spec.rb

@@ -88,5 +88,12 @@ module Anemone
         ]
     end
 
+    it "should include query string when fetching" do
+      url = URI(FakePage.new('foo?bar=baz').url)
+      page = Page.fetch(url)
+      page.url.should == url
+      page.url.path_with_query.should == '/foo?bar=baz'
+    end
+
   end
 end