From d9b0a0dab646bc4a1a5b57cea74506c199a7b7b3 Mon Sep 17 00:00:00 2001
From: Fedele Mantuano <mantuano.fedele@gmail.com>
Date: Thu, 26 Jul 2018 22:11:36 +0200
Subject: [PATCH] Bugfix

---
 requirements_optional.txt                  |  2 +-
 src/modules/attachments/post_processing.py | 23 +++++++++++++++++-----
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/requirements_optional.txt b/requirements_optional.txt
index 60f266c..45b1d5c 100644
--- a/requirements_optional.txt
+++ b/requirements_optional.txt
@@ -1,5 +1,5 @@
 elasticsearch>=6.0.0, <7
 redis>=2.10.5, <3
 shodan
-tika-app
+tika-app>=1.4.0
 virustotal-api
diff --git a/src/modules/attachments/post_processing.py b/src/modules/attachments/post_processing.py
index 5ce1716..b7fbdcd 100644
--- a/src/modules/attachments/post_processing.py
+++ b/src/modules/attachments/post_processing.py
@@ -84,13 +84,20 @@ def tika(conf, attachments):
 
     if conf["enabled"]:
         from tikapp import TikaApp
-        tika = TikaApp(file_jar=conf["path_jar"],
-                       memory_allocation=conf["memory_allocation"])
+
+        tika = TikaApp(
+            file_jar=conf["path_jar"],
+            memory_allocation=conf["memory_allocation"])
+
+        wtlist = conf.get("whitelist_content_types", [])
+        if not wtlist:
+            log.warning(
+                "Apache Tika analysis setted, without whitelist content types")
+            return
 
         for a in attachments:
             if not a.get("is_filtered", False):
-                if a["Content-Type"] in \
-                        conf.get("whitelist_content_types", []):
+                if a["Content-Type"] in wtlist:
                     payload = a["payload"]
 
                     if a["content_transfer_encoding"] != "base64":
@@ -132,7 +139,13 @@ def virustotal(conf, attachments):
         from .utils import reformat_virustotal
 
         vt = VirusTotalPublicApi(conf["api_key"])
-        wtlist = conf["whitelist_content_types"]
+        wtlist = conf.get("whitelist_content_types", [])
+
+        # I don't have content types to analyze
+        if not wtlist:
+            log.warning(
+                "Virustotal analysis setted, without whitelist content types")
+            return
 
         for a in attachments:
             if not a.get("is_filtered", False) and a["Content-Type"] in wtlist: