new: Add initial redirects columns

lookyloo/lookyloo.py

@@ -30,7 +30,7 @@ import logging
 
 from pysanejs import SaneJS
 from scrapysplashwrapper import crawl
-from har2tree import CrawledTree, Har2TreeError
+from har2tree import CrawledTree, Har2TreeError, HarFile
 
 
 class Lookyloo():
@@ -72,17 +72,13 @@ class Lookyloo():
             return
         with (report_dir / 'uuid').open() as f:
             uuid = f.read().strip()
-        with har_files[0].open() as f:
-            j = json.load(f)
-            title = j['log']['pages'][0]['title']
-            timestamp = j['log']['pages'][0]['startedDateTime']
-            if j['log']['entries']:
-                first_url = j['log']['entries'][0]['request']['url']
-            else:
-                first_url = '-'
-            if not title:
-                title = '!! No title found !! '
-        cache = {'uuid': uuid, 'title': title, 'timestamp': timestamp, 'url': first_url}
+        har = HarFile(har_files[0])
+
+        cache: Dict[str, Union[str, int]] = {'uuid': uuid,
+                                             'title': har.initial_title,
+                                             'timestamp': har.initial_start_time,
+                                             'url': har.first_url,
+                                             'redirects': json.dumps(har.initial_redirects)}
         if (report_dir / 'no_index').exists():  # If the folders claims anonymity
             cache['no_index'] = 1
         if uuid and not self.redis.exists(str(report_dir)):
@@ -92,7 +88,9 @@ class Lookyloo():
     def report_cache(self, report_dir: Union[str, Path]) -> Dict:
         if isinstance(report_dir, Path):
             report_dir = str(report_dir)
-        return self.redis.hgetall(report_dir)
+        cached = self.redis.hgetall(report_dir)
+        cached['redirects'] = json.loads(cached['redirects'])
+        return cached
 
     def _init_existing_dumps(self) -> None:
         for report_dir in self.report_dirs:

poetry.lock

@@ -208,7 +208,7 @@ lxml = "^4.4.2"
 six = "^1.14.0"
 
 [package.source]
-reference = "12c88d6298e52a8458e220b6910597cb5539daa9"
+reference = "147ac0f014249358af7b54cf02d5c85644a60645"
 type = "git"
 url = "https://github.com/viper-framework/har2tree.git"
 [[package]]

website/web/__init__.py

@@ -175,6 +175,6 @@ def index():
             continue
         date, time = cached['timestamp'].split('T')
         time, _ = time.split('.', 1)
-        titles.append((cached['uuid'], cached['title'], date, time, cached['url']))
+        titles.append((cached['uuid'], cached['title'], date, time, cached['url'], cached['redirects']))
     titles = sorted(titles, key=lambda x: (x[2], x[3]), reverse=True)
     return render_template('index.html', titles=titles)

website/web/templates/index.html

@@ -8,23 +8,35 @@
     </br></br>
   </center>
 
-  <table class="table table-hover" data-toggle="table" data-search="true">
-  <thead>
-        <tr>
-            <th scope="col">Page title</th>
-            <th scope="col">Initial URL</th>
-        <tr>
+  <table class="table" data-toggle="table" data-search="true" data-show-columns="true">
+    <thead>
+     <tr>
+       <th>Page title</th>
+       <th>Initial URL</th>
+       <th>Initial redirects</th>
+     <tr>
     </thead>
     <tbody>
-        {% for uuid, page_title, date, time, url in titles %}
-        <tr>
-            <td>
-                <a href="{{ url_for('tree', tree_uuid=uuid) }}">{{ page_title }}</a>
-                <div id="timestamp">  <b>{{ date }}</b> {{ time }}</div>
-            </td>
-            <td>{{ url }}</td>
-        </tr>
-        {% endfor %}
+      {% for uuid, page_title, date, time, url, redirects in titles %}
+      <tr>
+        <td>
+          <a href="{{ url_for('tree', tree_uuid=uuid) }}">{{ page_title }}</a>
+          <div id="timestamp">  <b>{{ date }}</b> {{ time }}</div>
+        </td>
+        <td>{{ url }}</td>
+        <td>
+          {% if redirects %}
+            <ul>
+            {% for r in redirects %}
+              <li>{{ r }}</li>
+            {% endfor %}
+            </ul>
+          {% else%}
+            No redirect
+          {%endif%}
+        </td>
+      </tr>
+      {% endfor %}
     </tbody>
   </table>
 {% endblock %}

website/web/templates/main.html

@@ -21,9 +21,13 @@
   </head>
   <body>
     <!-- Your page contont -->
+    <!--
     <div class="container">
+    -->
       {% block content %}{% endblock%}
+    <!--
     </div>
+    -->
 
     {% block scripts %}
     <!-- Optional JavaScript -->