From a0a6de150ad55071b732bc8871f9e70f9e56c9b4 Mon Sep 17 00:00:00 2001
From: Robert Glaser <mail@robert-glaser.de>
Date: Wed, 9 Feb 2011 17:32:14 +0100
Subject: [PATCH] Parser should ignore anchor tags without an href attribute.

---
 lib/rawler/crawler.rb           | 2 +-
 spec/lib/rawler/crawler_spec.rb | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/rawler/crawler.rb b/lib/rawler/crawler.rb
index 155656b..bdb4c97 100644
--- a/lib/rawler/crawler.rb
+++ b/lib/rawler/crawler.rb
@@ -16,7 +16,7 @@ def links
       response = Rawler::Request.get(url)
       
       doc = Nokogiri::HTML(response.body)
-      doc.css('a').map { |a| a['href'] }.map { |url| absolute_url(url) }.select { |url| valid_url?(url) }
+      doc.css('a').map { |a| a['href'] }.select { |url| !url.nil? }.map { |url| absolute_url(url) }.select { |url| valid_url?(url) }
     rescue Errno::ECONNREFUSED
       write("Couldn't connect to #{url}")
       []
diff --git a/spec/lib/rawler/crawler_spec.rb b/spec/lib/rawler/crawler_spec.rb
index 7668e4a..53d1069 100644
--- a/spec/lib/rawler/crawler_spec.rb
+++ b/spec/lib/rawler/crawler_spec.rb
@@ -85,7 +85,7 @@
     let(:url)     { 'http://example.com/path' }
     let(:crawler) { Rawler::Crawler.new(url) }
     let(:js_url)  { "javascript:fn('nbjmup;jhfs.esf{fio/dpn');" }
-    let(:content) { "<a href=\"#{js_url}\">foo</a>" }
+    let(:content) { "<a href=\"#{js_url}\">foo</a><a name=\"foo\">" }
     
     before(:each) do
       register(url, content)