chg: [tags] search ocrs and images by tags + fix ocr, filter invalid image

bin/lib/objects/Ocrs.py

@@ -148,6 +148,7 @@ class Ocr(AbstractDaterangeObject):
         if options is None:
             options = set()
         meta = self._get_meta(options=options)
+        meta['tags'] = self.get_tags()
         meta['content'] = self.get_content()
 
         # optional meta fields

bin/modules/OcrExtractor.py

@@ -101,7 +101,12 @@ class OcrExtractor(AbstractModule):
             languages = get_model_languages(image)
             languages = Ocrs.sanityze_ocr_languages(languages, ocr_languages=self.ocr_languages)
             print(image.id, languages)
-            texts = Ocrs.extract_text(path, languages)
+            try:
+                texts = Ocrs.extract_text(path, languages)
+            except ValueError as e:
+                self.logger.warning(e)
+                self.obj.add_tag('infoleak:confirmed="false-positive"')
+                texts = None
             if texts:
                 print('create')
                 ocr = Ocrs.create(image.id, texts)

var/www/blueprints/objects_ocr.py

@@ -108,6 +108,7 @@ def object_ocr():
         meta = meta[0]
         languages = Language.get_translation_languages()
         return render_template("ShowOcr.html", meta=meta,
+                               bootstrap_label=bootstrap_label,
                                ail_tags=Tag.get_modal_add_tags(meta['id'], meta['type'], meta['subtype']),
                                translation_languages=languages, translation_target=target)
 

var/www/blueprints/tags_ui.py

@@ -293,6 +293,24 @@ def tags_search_messages():
     dict_tagged['date'] = Date.sanitise_date_range('', '', separator='-')
     return render_template("tags/search_obj_by_tags.html", bootstrap_label=bootstrap_label, dict_tagged=dict_tagged)
 
+@tags_ui.route('/tag/search/image')
+@login_required
+@login_read_only
+def tags_search_images():
+    object_type = 'image'
+    dict_tagged = {"object_type": object_type, "object_name": object_type.title() + "s"}
+    dict_tagged['date'] = Date.sanitise_date_range('', '', separator='-')
+    return render_template("tags/search_obj_by_tags.html", bootstrap_label=bootstrap_label, dict_tagged=dict_tagged)
+
+@tags_ui.route('/tag/search/ocr')
+@login_required
+@login_read_only
+def tags_search_ocrs():
+    object_type = 'ocr'
+    dict_tagged = {"object_type": object_type, "object_name": object_type.title() + "s"}
+    dict_tagged['date'] = Date.sanitise_date_range('', '', separator='-')
+    return render_template("tags/search_obj_by_tags.html", bootstrap_label=bootstrap_label, dict_tagged=dict_tagged)
+
 @tags_ui.route('/tag/search/domain')
 @login_required
 @login_read_only

var/www/templates/correlation/show_correlation.html

@@ -733,9 +733,14 @@ if (d.popover) {
 								desc = desc + "fa-times-circle\"></i>DOWN"
 							}
 							desc = desc + "</div></dd>"
-						} else if (key!="tags" && key!="id" && key!="img" && key!="icon" && key!="link" && key!="type") {
+						} else if (key!=="tags" && key!=="id" && key!=="img" && key!=="svg_icon" && key!=="icon" && key!=="link" && key!=="type") {
                             if (data[key]) {
-							    desc = desc + "<dt class=\"col-sm-3 px-0\">" + sanitize_text(key) + "</dt><dd class=\"col-sm-9 px-0\">" + sanitize_text(data[key]) + "</dd>"
+                                if ((key==="first_seen" || key==="last_seen") && data[key].length===8) {
+                                    let date = sanitize_text(data[key])
+                                    desc = desc + "<dt class=\"col-sm-3 px-0\">" + sanitize_text(key) + "</dt><dd class=\"col-sm-9 px-0\">" + date.slice(0, 4) + "-" + date.slice(4, 6) + "-" + date.slice(6, 8) + "</dd>"
+                                } else {
+							        desc = desc + "<dt class=\"col-sm-3 px-0\">" + sanitize_text(key) + "</dt><dd class=\"col-sm-9 px-0\">" + sanitize_text(data[key]) + "</dd>"
+                                }
                             }
 						}
 					});

var/www/templates/tags/menu_sidebar.html

@@ -22,6 +22,18 @@
       	        Search Messages by Tags
       	      </a>
       	    </li>
+            <li class="nav-item">
+      	      <a class="nav-link" href="{{ url_for('tags_ui.tags_search_images') }}" id="nav_tags_search_image">
+      	        <i class="fas fa-image"></i>
+      	        Search Images by Tags
+      	      </a>
+      	    </li>
+            <li class="nav-item">
+      	      <a class="nav-link" href="{{ url_for('tags_ui.tags_search_ocrs') }}" id="nav_tags_search_message">
+      	        <i class="fas fa-expand"></i>
+      	        Search Ocrs by Tags
+      	      </a>
+      	    </li>
       	    <li class="nav-item">
       	      <a class="nav-link" href="{{ url_for('tags_ui.tags_search_domains') }}" id="nav_tags_search_domain">
       					<i class="fab fa-html5"></i>

var/www/templates/tags/search_obj_by_tags.html

@@ -125,7 +125,27 @@
                         </div>
                       </td>
                     </tr>
-                   {% endfor %}
+                  {% endfor %}
+                    {%elif dict_tagged["object_type"]=="screenshot"%}
+
+                {%elif dict_tagged["object_type"]=="image" or dict_tagged["object_type"]=="ocr"%}
+                    {% for dict_obj in dict_tagged["tagged_obj"] %}
+                        <tr>
+                            <td class="pb-0">
+                                <a target="_blank" href="{{ url_for('correlation.show_correlation') }}?type={{dict_tagged['object_type']}}&id={{dict_obj['id']}}" class="text-secondary">
+                                    <div style="line-height:0.9;">{{ dict_obj['id'] }}</div>
+                                </a>
+                                <div class="mb-2">
+                                    {% for tag in dict_obj['tags'] %}
+                                        <a href="{{ url_for('tags_ui.get_obj_by_tags') }}?object_type={{dict_tagged['object_type']}}&ltags={{ tag }}">
+                                            <span class="badge badge-{{ bootstrap_label[loop.index0 % 5] }}">{{ tag }}</span>
+                                        </a>
+                                    {%  endfor %}
+                                </div>
+                            </td>
+                        </tr>
+                    {% endfor %}
+
                 {%elif dict_tagged["object_type"]=="message"%}
                   {% for dict_obj in dict_tagged["tagged_obj"] %}
                     <tr>