lib/krawler.rb in krawler-1.0.12 vs lib/krawler.rb in krawler-1.0.13
- old
+ new
@@ -125,10 +125,10 @@
new_url = URI(new_link)
new_link = new_url.to_s
rescue ArgumentError # junk link
next
end
-
+
if @domain || (new_link =~ /^#{Regexp.escape(@host)}/) || (new_link =~ /^\//) # don't crawl external domains
next if @crawled_links.include?(new_link) || @links_to_crawl.include?(new_link) # don't crawl what we've alread crawled
next if @exclude && new_link =~ /#{@exclude}/ # don't crawl excluded matched paths