nmapxml: don't use unnecessary temporary files

Instead of saving data to insecure temporary files created without
mktemp, simply echo them directly to the next stage of a pipeline.

Also reset the $software at the beginning of each round, to prevent
processing and then having to deduplicate many things, twice.

searchsploit

@@ -346,9 +346,6 @@ function searchsploitout()
 ## Read XML file
 function nmapxml()
 {
-  ## Remove any old traces
-  rm -f /tmp/searchsploit.{tmp,out}
-
   ## Feedback to the end user
   echo -e "[i] Reading: '${FILE}'\n"
 
@@ -368,29 +365,24 @@ function nmapxml()
           ## If we have already looped around and got something, save it before moving onto the current value
           if [[ "${software}" ]]; then
             #searchsploitout
-            echo "${software}" >> /tmp/searchsploit.out
+            echo "${software}"
           fi
           ## Something is better than nothing. Will just go on the default service that matches the port.   e.g. domain
           software="${input}"
           ## Might not get any more than this, if -sV failed
-          echo "${software}" > /tmp/searchsploit.tmp
           ;;
         "[PRODUCT]")
           ## We have a name, but no version (yet?)   e.g. dnsmasq
           software="${input}"
-          echo "${software}" > /tmp/searchsploit.tmp
           ;;
         "[VERSION]")
           software="${software} ${input}"
           ## Name & version. There isn't any more information to get, game over.   e.g. dnsmasq 2.72
-          echo "${software}" >> /tmp/searchsploit.out
-          echo "" > /tmp/searchsploit.tmp
+          echo "${software}"
+          software=
           ;;
       esac
-  done
-
-  ## Read in from file (so there are no duplicates - ...but unable to print out IPs)
-  cat /tmp/searchsploit.out /tmp/searchsploit.tmp 2>/dev/null | tr '[:upper:]' '[:lower:]' | awk '!x[$0]++' | while read software; do
+  done | tr '[:upper:]' '[:lower:]' | awk '!x[$0]++' | while read software; do
     searchsploitout
   done
 }