Modified original PR to have nicer looking output in the HTML and so that href targets point to the site

breenmachine · breenmachine · commit ab8c54f48d9b · 2015-06-16T08:46:51.000-04:00
Merge branch 'master' of https://github.com/hlein/httpscreenshot into hlein-master Conflicts: screenshotClustering/cluster.py
diff --git a/httpscreenshot.py b/httpscreenshot.py
@@ -28,6 +28,10 @@
 import shutil
 import hashlib
 
+try:
+    from urllib.parse import quote
+except:
+    from urllib import quote
 
 reload(sys)
 sys.setdefaultencoding("utf8")
@@ -181,9 +185,10 @@ def worker(urlQueue, tout, debug, headless, doProfile, vhosts, subs, extraHosts,
 			except Queue.Empty:
 				continue
 			print '[+] '+str(urlQueue.qsize())+' URLs remaining'
-			screenshotName = urlparse(curUrl[0]).netloc.replace(":", "-")
+			screenshotName = quote(curUrl[0], safe='')
 			if(debug):
 				print '[+] Got URL: '+curUrl[0]
+				print '[+] screenshotName: '+screenshotName
 			if(os.path.exists(screenshotName+".png")):
 				if(debug):
 			 		print "[-] Screenshot already exists, skipping"
diff --git a/screenshotClustering/cluster.py b/screenshotClustering/cluster.py
@@ -8,6 +8,11 @@
 import time
 from bs4 import BeautifulSoup
 
+try:
+    from urllib.parse import quote,unquote
+except:
+    from urllib import quote,unquote
+
 def addAttrToBag(attrName,url,link,wordBags,soup):
 	for tag in soup.findAll('',{attrName:True}):
 		if(isinstance(tag[attrName],str) or isinstance(tag[attrName],unicode)):
@@ -141,9 +146,11 @@ def renderClusterHtml(clust,width,height,scopeFile=None):
     
     for cluster,siteList in clust.iteritems():
         html=html+'<TR>'
-    	html=html+'<TR><TD><img src="'+siteList[0][0:-4]+'png" width='+str(width)+' height='+str(height)+'/></TD></TR>'
+        screenshotName = quote(siteList[0][0:-4], safe='')
+        html=html+'<TR><TD><img src="'+screenshotName+'png" width='+str(width)+' height='+str(height)+'/></TD></TR>'
         for site in siteList:
-            html=html+'<TD onmouseout="clearPopup()" onmouseover="popUp(event,\''+site[0:-4]+'png\');"><a href="http://'+site[site.rfind('/')+1:site.rfind('-')]+':'+site[site.rfind('-')+1:site.rfind('.')]+'">'+site[site.rfind('/')+1:site.rfind('-')]+':'+site[site.rfind('-')+1:site.rfind('.')]+'</a></TD>'
+            screenshotName = quote(site[0:-5], safe='')
+            html=html+'<TD onmouseout="clearPopup()" onmouseover="popUp(event,\''+screenshotName+'.png\');"><a href="'+unquote(unquote(screenshotName[4:]).decode("utf-8")).decode("utf-8")+'">'+unquote(unquote(screenshotName[4:]).decode("utf-8")).decode("utf-8")+'</a></TD>'
         html=html+'</TR>'
     html=html+'</table>'
     footer = '</BODY></HTML>'
