chg: use cache instead of get_info everywhere.

lookyloo/lookyloo.py

@@ -427,7 +427,12 @@ class Lookyloo():
     def capture_cache(self, capture_uuid: str, /) -> Optional[CaptureCache]:
         """Get the cache from redis, rebuild the tree if the internal UUID changed => slow"""
         try:
-            return self._captures_index[capture_uuid]
+            cache = self._captures_index[capture_uuid]
+            # 2022-12-07: New cache format, store the user agent and referers. Re-cache if needed
+            if cache and not cache.user_agent:
+                self._captures_index.reload_cache(capture_uuid)
+                cache = self._captures_index[capture_uuid]
+            return cache
         except NoValidHarFile:
             self.logger.debug('No HAR files, {capture_uuid} is a broken capture.')
             return None

website/web/__init__.py

@@ -662,7 +662,6 @@ def tree(tree_uuid: str, node_uuid: Optional[str]=None):
         ct = lookyloo.get_crawled_tree(tree_uuid)
         b64_thumbnail = lookyloo.get_screenshot_thumbnail(tree_uuid, for_datauri=True)
         screenshot_size = lookyloo.get_screenshot(tree_uuid).getbuffer().nbytes
-        info = lookyloo.get_info(tree_uuid)
         meta = lookyloo.get_meta(tree_uuid)
         hostnode_to_highlight = None
         if node_uuid:
@@ -682,7 +681,7 @@ def tree(tree_uuid: str, node_uuid: Optional[str]=None):
         if cache.error:
             flash(cache.error, 'warning')
         return render_template('tree.html', tree_json=ct.to_json(),
-                               info=info,
+                               info=cache,
                                tree_uuid=tree_uuid, public_domain=lookyloo.public_domain,
                                screenshot_thumbnail=b64_thumbnail, page_title=cache.title,
                                screenshot_size=screenshot_size,

website/web/templates/tree.html

@@ -3,14 +3,14 @@
 {% from 'bootstrap5/utils.html' import render_messages %}
 {% from "macros.html" import shorten_string %}
 
-{% block title %}Capture of {{info['url']}}{% endblock %}
+{% block title %}Capture of {{info.url}}{% endblock %}
 
 {% block card %}
 <meta property="og:title" content="Lookyloo capture" />
 <meta property="og:type" content="website"/>
 <meta
   property="og:description"
-  content="URL captured: {{info['url']}}"
+  content="URL captured: {{info.url}}"
 />
 <meta
   property="og:image"
@@ -148,7 +148,7 @@
   var enable_bookmark = {{ enable_bookmark|tojson }};
   var treeData = {{ tree_json | safe }};
   var parent_uuid = {{ parent_uuid|tojson }};
-  var capture_starttime = new Date(Date.parse("{{ info['capture_time'] }}"));
+  var capture_starttime = new Date(Date.parse("{{ info.timestamp.isoformat() }}"));
   window.addEventListener('DOMContentLoaded', (event) => {
       document.getElementById("start_time").innerHTML =
           `${capture_starttime.getFullYear()}-${("0" + (capture_starttime.getMonth() + 1)).slice(-2)}-${("0" + capture_starttime.getDate()).slice(-2)} ${capture_starttime.toLocaleTimeString()}`;
@@ -413,20 +413,20 @@
     <div class="modal-body">
       <dl class="row">
         <dt class="col-sm-2">URL captured</dt>
-        <dd class="col-sm-10">{{ shorten_string(info['url'], 1000) }}</dd>
+        <dd class="col-sm-10">{{ shorten_string(info.url, 1000) }}</dd>
 
         <dt class="col-sm-2">Page title</dt>
-        <dd class="col-sm-10">{{ info['title'] }}</dd>
+        <dd class="col-sm-10">{{ info.title }}</dd>
 
         <dt class="col-sm-2">Capture time</dt>
         <dd class="col-sm-10" id="start_time"></dd>
 
         <dt class="col-sm-2">User Agent</dt>
-        <dd class="col-sm-10">{{ info['user_agent'] }}</dd>
+        <dd class="col-sm-10">{{ info.user_agent }}</dd>
 
-        {% if 'referer' in info and info['referer'] %}
+        {% if info.referer %}
         <dt class="col-sm-2">Referer</dt>
-        <dd class="col-sm-10">{{ info['referer'] }}</dd>
+        <dd class="col-sm-10">{{ info.referer }}</dd>
         {%endif%}