init sources

src/DoresA/scripts/preprocess_benign.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+# how much to take
+COUNT=1000;
+
+cd res;
+rm alexa.zip;
+rm top-1m.csv;
+
+cd raw;
+
+curl -o alexa.zip http://s3.amazonaws.com/alexa-static/top-1m.csv.zip;
+unzip alexa.zip;
+
+head -n $COUNT top-1m.csv | cut -f2 -d"," >> res/benign_domains.txt;
+

src/DoresA/scripts/preprocess_malicious.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+
+# cleanup
+cd res;
+echo "" > malicious_domains.txt;
+rm malwaredomains.zip;
+rm domains.txt;
+rm phishtank.csv;
+rm zeus.txt;
+
+cd raw;
+
+# malwaredomains.com
+curl -o malwarecomains.zip http://malware-domains.com/files/domains.zip;
+unzip malwaredomains.zip;
+
+tail -n +5 domains.txt | cut -f3 >> ../malicious_domains.txt;
+
+# Phishtank
+curl -o phishtank.csv http://data.phishtank.com/data/online-valid.csv
+
+tail -n +1 phishtank.csv | cut -f2 -d"," >> ../malicious_domains.txt
+
+# ZeuS Tracker
+curl -o zeus.txt https://zeustracker.abuse.ch/blocklist.php?download=baddomains;
+
+tail -n +7 zeus.txt >> ../malicious_domains.txt;
+
+# remove empty lines
+sed -i.bak '/^$/d' ../malicious_domains.txt