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

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)
+            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/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/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

@@ -126,6 +126,26 @@
                       </td>
                     </tr>
                   {% 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>