o [NSE] Added getLimitations function to httpspider that returns any

limitations imposed on the crawler. [Patrik]

CHANGELOG

@@ -1,5 +1,8 @@
 # Nmap Changelog ($Id$); -*-text-*-
 
+o [NSE] Added getLimitations function to httpspider that returns any 
+  limitations imposed on the crawler. [Patrik]
+
 o [NSE] Modified the httpspider library to prefetch links in the queue and
   change how script arguments are processed. Script and library arguments are
   now processed from within the library. [Patrik]

nselib/httpspider.lua

@@ -501,6 +501,8 @@ Crawler = {
 			o:addDefaultBlacklist()
 		end
 		
+		stdnse.print_debug(2, "%s: %s", LIBRARY_NAME, o:getLimitations())
+		
 		return o
 	end,
 	
@@ -642,6 +644,33 @@ Crawler = {
 		self.url = self.url or '/'
 	end,	
 	
+	-- gets a string of limitations imposed on the crawl
+	getLimitations = function(self)
+		local o = self.options
+		local limits = {}
+		
+		if ( o.maxdepth > 0 or o.maxpagecount > 0 or
+			 o.withinhost or o.wihtindomain ) then
+			if ( o.maxdepth > 0 ) then
+				table.insert(limits, ("maxdepth=%d"):format(o.maxdepth))
+			end
+			if ( o.maxpagecount > 0 ) then
+				table.insert(limits, ("maxpagecount=%d"):format(o.maxpagecount))
+			end
+			if ( o.withindomain ) then
+				table.insert(limits, ("withindomain=%s"):format(o.base_url:getDomain()))
+			end
+			if ( o.withinhost ) then
+				table.insert(limits, ("withinhost=%s"):format(o.base_url:getHost()))
+			end
+		end
+	
+		if ( #limits > 0 ) then
+			return ("Spidering limited to: %s"):format(stdnse.strjoin("; ", limits))
+		end
+	end,
+	
+	-- does the crawling
 	crawl = function(self)
 
 		self.response_queue = self.response_queue or {}

scripts/http-email-harvest.nse

@@ -46,13 +46,6 @@ function action(host, port)
 
 	crawler:set_timeout(10000)
 
-	local maxdepth, maxpagecount = crawler.options.maxdepth, crawler.options.maxpagecount
-	if ( maxdepth < 0 ) then maxdepth = nil end
-	if ( maxpagecount < 0 ) then maxpagecount = nil end
-	
-	stdnse.print_debug(2, "%s: Running crawler maxdepth: %s; maxpagecount: %s", 
-		SCRIPT_NAME, maxdepth or "[none]", maxpagecount or "[none]")
-
 	local emails = {}
 	while(true) do
 		local status, r = crawler:crawl()
@@ -80,20 +73,7 @@ function action(host, port)
 		table.insert(results, email)
   	end
 
-	-- Inform the user of the limitations that were used
+	results.name = crawler:getLimitations()
-	if ( maxdepth > 0 or maxpagecount > 0 ) then
+	
-		local limit = "Spidering limited to: "
-		if ( maxdepth > 0 ) then
-			limit = limit .. ("maxdepth=%d; "):format(maxdepth)
-		end
-		if ( maxpagecount > 0 ) then
-			limit = limit .. ("maxpagecount=%d"):format(maxpagecount)
-		end
-		if ( #results == 0 ) then
-			table.insert(results, limit)
-		else
-			results.name = limit
-		end
-	end
 	return stdnse.format_output(true, results)
 end