Merge branch 'master' into crawler_v2

bin/export/MispExport.py

@@ -75,23 +75,23 @@ def export_ail_item(item_id):
         tag_misp_object_attributes(l_obj_attr, dict_metadata['tags'])
     return obj
 
-# # TODO: create domain-port-history object
 def export_domain(domain):
     domain_obj = Domain.Domain(domain)
     dict_metadata = domain_obj.get_domain_metadata(tags=True)
-    dict_metadata['ports'] = ['80', '223', '443']
 
     # create domain-ip obj
-    obj = MISPObject('domain-ip', standalone=True)
+    obj = MISPObject('domain-crawled', standalone=True)
     obj.first_seen = dict_metadata['first_seen']
     obj.last_seen = dict_metadata['last_check']
 
     l_obj_attr = []
-    l_obj_attr.append( obj.add_attribute('first-seen', value=dict_metadata['first_seen']) )
-    l_obj_attr.append( obj.add_attribute('last-seen', value=dict_metadata['last_check']) )
     l_obj_attr.append( obj.add_attribute('domain', value=domain) )
-    for port in dict_metadata['ports']:
+    dict_all_url = Domain.get_domain_all_url(domain, domain_obj.get_domain_type())
-        l_obj_attr.append( obj.add_attribute('port', value=port) )
+    for crawled_url in dict_all_url:
+        attribute = obj.add_attribute('url', value=crawled_url)
+        attribute.first_seen = str(dict_all_url[crawled_url]['first_seen'])
+        attribute.last_seen = str(dict_all_url[crawled_url]['last_seen'])
+        l_obj_attr.append( attribute )
 
     # add tags
     if dict_metadata['tags']:

bin/export/MispImport.py

@@ -190,7 +190,7 @@ def unpack_file(map_uuid_global_id, misp_obj):
 def get_misp_import_fct(map_uuid_global_id, misp_obj):
     if misp_obj.name == 'ail-leak':
         unpack_item_obj(map_uuid_global_id, misp_obj)
-    elif misp_obj.name == 'domain-ip':
+    elif misp_obj.name == 'domain-crawled':
         pass
     elif misp_obj.name == 'pgp-meta':
         unpack_obj_pgp(map_uuid_global_id, misp_obj)

bin/lib/Domain.py

@@ -327,6 +327,28 @@ def get_all_domain_up_by_type(domain_type):
     else:
         return ({"status": "error", "reason": "Invalid domain type"}, 400)
 
+def get_domain_all_url(domain, domain_type, domain_ports=None):
+    if not domain_ports:
+        domain_ports = get_domain_all_ports(domain, domain_type)
+    all_url = {}
+    for port in domain_ports:
+        for dict_history in get_domain_history_with_status(domain, domain_type, port, add_root_item=True):
+            if dict_history['status']: # domain UP
+                crawled_items = get_domain_items(domain, dict_history['root_item'])
+                for item_id in crawled_items:
+                    item_url = Item.get_item_link(item_id)
+                    item_date = int(Item.get_item_date(item_id))
+                    if item_url:
+                        if item_url not in all_url:
+                            all_url[item_url] = {'first_seen': item_date,'last_seen': item_date}
+                        else: # update first_seen / last_seen
+                            if item_date < all_url[item_url]['first_seen']:
+                                all_url[item_url]['first_seen'] = item_date
+                            if item_date > all_url[item_url]['last_seen']:
+                                all_url[item_url]['last_seen'] = item_date
+    return all_url
+
+
 def get_domain_items(domain, root_item_id):
     dom_item =  get_domain_item_children(domain, root_item_id)
     dom_item.append(root_item_id)
@@ -605,7 +627,7 @@ def get_domain_history(domain, domain_type, port): # TODO: add date_range: from
     '''
     return r_serv_onion.zrange('crawler_history_{}:{}:{}'.format(domain_type, domain, port), 0, -1, withscores=True)
 
-def get_domain_history_with_status(domain, domain_type, port): # TODO: add date_range: from to + nb_elem
+def get_domain_history_with_status(domain, domain_type, port, add_root_item=False): # TODO: add date_range: from to + nb_elem
     '''
     Retun .
 
@@ -619,14 +641,17 @@ def get_domain_history_with_status(domain, domain_type, port): # TODO: add date_
     history = get_domain_history(domain, domain_type, port)
     for root_item, epoch_val in history:
         epoch_val = int(epoch_val) # force int
+        dict_history = {"epoch": epoch_val, "date": time.strftime('%Y/%m/%d - %H:%M.%S', time.gmtime(epoch_val))}
         # domain down, root_item==epoch_val
         try:
             int(root_item)
-            status = False
+            dict_history['status'] = False
         # domain up, root_item=str
         except ValueError:
-            status = True
+            dict_history['status'] = True
-        l_history.append({"epoch": epoch_val, "date": time.strftime('%Y/%m/%d - %H:%M.%S', time.gmtime(epoch_val)), "status": status})
+            if add_root_item:
+                dict_history['root_item'] = root_item
+        l_history.append(dict_history)
     return l_history
 
 def verify_if_domain_exist(domain):

bin/torcrawler/launch_splash_crawler.sh

@@ -62,7 +62,7 @@ sleep 0.1
 
 for ((i=0;i<=$((${n} - 1));i++)); do
     port_number=$((${p} + $i))
-    screen -S "Docker_Splash" -X screen -t "docker_splash:$port_number" bash -c 'sudo docker run -d -p '$port_number':8050 --restart=always --cpus=1 --memory=4.5G -v '$f':/etc/splash/proxy-profiles/ --net="bridge" scrapinghub/splash --maxrss '$u'; read x'
+    screen -S "Docker_Splash" -X screen -t "docker_splash:$port_number" bash -c 'sudo docker run -d -p '$port_number':8050 --restart=always --cpus=1 --memory=2G -v '$f':/etc/splash/proxy-profiles/ --net="bridge" scrapinghub/splash --maxrss '$u'; read x'
     sleep 0.1
     printf "$GREEN    Splash server launched on port $port_number$DEFAULT\n"
 done

var/www/blueprints/import_export.py

@@ -137,14 +137,10 @@ def export_object_file():
         dict_misp_event_export = None
 
     if l_obj_invalid:
-        for obj_dict in l_obj_to_export:
-            obj_dict['uuid'] = str(uuid.uuid4())
-            obj_dict['type'] = Correlate_object.get_obj_str_type_subtype(obj_dict['type'], obj_dict.get('subtype', None))
-
         # get user saved obj to export # # TODO: # performance
         l_obj_to_export = AILObjects.get_user_list_of_obj_to_export(user_id)
 
-        for obj_dict in l_obj_invalid:
+        for obj_dict in l_obj_invalid: # set uuid input
             obj_dict['uuid'] = str(uuid.uuid4())
             obj_dict['type'] = Correlate_object.get_obj_str_type_subtype(obj_dict['type'], obj_dict.get('subtype', None))
 
@@ -152,7 +148,6 @@ def export_object_file():
                                 l_obj_invalid=l_obj_invalid, dict_misp_event_export=dict_misp_event_export)
     else:
         if export_to_misp and MispExport.ping_misp():
-            l_obj_to_export = AILObjects.get_user_list_of_obj_to_export(user_id)
             event = MispExport.create_list_of_objs_to_export(l_obj_to_export, r_type='event')
 
             event_metadata = MispExport.create_misp_event(event, distribution=dict_misp_event_export.get('export_to_misp', None),
@@ -167,7 +162,6 @@ def export_object_file():
                                     l_obj_invalid=[], dict_misp_event_export=[])
         else:
             # get user saved obj to export # # TODO: # performance
-            l_obj_to_export = AILObjects.get_user_list_of_obj_to_export(user_id)
             json_export = MispExport.create_list_of_objs_to_export(l_obj_to_export)
             export_filename = MispExport.get_export_filename(json_export)
             json_export = MispExport.create_in_memory_file(json_export.to_json())

var/www/templates/domains/card_img_domain.html

@@ -8,7 +8,7 @@
         <div class="text-center">
           <canvas id="canvas_{{loop.index0}}" style="max-height: 400px; max-width: 100%;"></canvas>
         </div>
-        <div class="card-body">
+        <div class="card-body pb-0">
           <h5 class="card-title">
             <a target="_blank" href="{{ url_for('crawler_splash.showDomain') }}?domain={{dict_domain["id"]}}">
               {{dict_domain["id"]}}
@@ -39,6 +39,9 @@
             {%  endfor %}
           </div>
         </div>
+        {% with obj_type='domain', obj_id=dict_domain["id"], obj_lvl=0%}
+            {% include 'import_export/block_add_user_object_to_export_small.html' %}
+        {% endwith %}
       </div>
 
   {% if loop.index0 % 4 == 3 %}

var/www/templates/import_export/block_add_user_object_to_export_small.html

@@ -0,0 +1,12 @@
+<div class="d-flex flex-row-reverse bd-highlight">
+  <div>
+    <a class="btn" target="_blank" href="{{ url_for('import_export.add_object_id_to_export')}}?obj_type={{obj_type}}&obj_id={{obj_id}}&obj_lvl={{obj_lvl}}{%if obj_subtype%}&obj_subtype={{obj_subtype}}{%endif%}">
+      <img id="misp-logo" src="{{ url_for('static', filename='image/misp-logo.png')}}" height="25">
+    </a>
+  </div>
+  <div>
+    <a class="btn btn-outline-light" href="{{ url_for('correlation.show_correlation')}}?object_type={{obj_type}}&correlation_id={{ obj_id }}" target="_blank" style="font-size: 15px">
+      <i class="fas fa-project-diagram text-secondary"></i>
+    </a>
+  </div>
+</div>

var/www/templates/import_export/export_object.html

@@ -73,7 +73,7 @@
 									          <span class="btn btn-info input-group-addon add-field col-1"><i class="fas fa-plus"></i></span>
 									        </div>
 													{% for obj_dict in l_obj_to_export %}
-														{% with obj_type=obj_dict['type'], obj_id=obj_dict['id'], obj_lvl=obj_dict['lvl'], input_uuid=obj_dict['uuid'], obj_error=False%}
+														{% with obj_type=obj_dict['type'], obj_id=obj_dict['id'], obj_lvl=obj_dict['lvl'], input_uuid=obj_dict, obj_error=False%}
 															{% include 'import_export/block_to_export_input.html' %}
 														{% endwith %}
 													{% endfor %}

var/www/update_thirdparty.sh

@@ -112,7 +112,7 @@ then
 fi
 
 #Update MISP Taxonomies and Galaxies
-python3 -m pip install git+https://github.com/MISP/PyTaxonomies
+python3 -m pip install git+https://github.com/MISP/PyTaxonomies --upgrade
 python3 -m pip install git+https://github.com/MISP/PyMISPGalaxies --upgrade
 
 #Update PyMISP