make archivebox server work for urls, hashes, and timestamps

archivebox/core/urls.py

@@ -1,7 +1,10 @@
 from django.contrib import admin
 
 from django.urls import path, include
+from django.views import static
 from django.conf import settings
+from django.contrib.staticfiles import views
+from django.views.generic.base import RedirectView
 
 from core.views import MainIndex, AddLinks, LinkDetails
 
@@ -9,17 +12,19 @@ admin.site.site_header = 'ArchiveBox Admin'
 admin.site.index_title = 'Archive Administration'
 
 urlpatterns = [
-    path('archive/<timestamp>/', LinkDetails.as_view(), name='LinkDetails'),
+    path('index.html', RedirectView.as_view(url='/')),
+    path('index.json', static.serve, {'document_root': settings.OUTPUT_DIR, 'path': 'index.json'}),
+    path('robots.txt', static.serve, {'document_root': settings.OUTPUT_DIR, 'path': 'robots.txt'}),
+    path('favicon.ico', static.serve, {'document_root': settings.OUTPUT_DIR, 'path': 'favicon.ico'}),
+
+    path('archive/', RedirectView.as_view(url='/')),
+    path('archive/<path:path>', LinkDetails.as_view(), name='LinkAssets'),
+    path('add/', AddLinks.as_view(), name='AddLinks'),
+    
+    path('static/<path>', views.serve),
     path('accounts/', include('django.contrib.auth.urls')),
     path('admin/', admin.site.urls),
-    path('add/', AddLinks.as_view(), name='AddLinks'),
     path('', MainIndex.as_view(), name='Home'),
 ]
 
 
-if settings.SERVE_STATIC:
-    # serve staticfiles via runserver
-    from django.contrib.staticfiles import views
-    urlpatterns += [
-        path('static/<path>', views.serve),
-    ]

archivebox/core/views.py

@@ -1,9 +1,15 @@
-from django.shortcuts import render
+__package__ = 'archivebox.core'
 
-from django.views import View
+from django.shortcuts import render, redirect
 
-from .index import load_main_index, load_main_index_meta
+from django.http import HttpResponse
-from .config import OUTPUT_DIR
+from django.views import View, static
+
+from core.models import Page
+
+from ..index import load_main_index, load_main_index_meta
+from ..config import OUTPUT_DIR, VERSION, FOOTER_INFO
+from ..util import base_url
 
 
 class MainIndex(View):
@@ -17,6 +23,8 @@ class MainIndex(View):
             'updated': meta_info['updated'],
             'num_links': meta_info['num_links'],
             'links': all_links,
+            'VERSION': VERSION,
+            'FOOTER_INFO': FOOTER_INFO,
         }
 
         return render(template_name=self.template, request=request, context=context)
@@ -41,7 +49,45 @@ class AddLinks(View):
 
 
 class LinkDetails(View):
-    template = 'link_details.html'
+    def get(self, request, path):
+        # missing trailing slash -> redirect to index
+        if '/' not in path:
+            return redirect(f'{path}/index.html')
 
-    def get(self, request):
+        try:
-        return render(template_name=self.template, request=request, context={})
+            slug, archivefile = path.split('/', 1)
+        except (IndexError, ValueError):
+            slug, archivefile = path.split('/', 1)[0], 'index.html'
+
+        all_pages = list(Page.objects.all())
+
+        # slug is a timestamp
+        by_ts = {page.timestamp: page for page in all_pages}
+        try:
+            return static.serve(request, archivefile, by_ts[slug].link_dir, show_indexes=True)
+        except KeyError:
+            pass
+
+        # slug is a hash
+        by_hash = {page.url_hash: page for page in all_pages}
+        try:
+            timestamp = by_hash[slug].timestamp
+            return redirect(f'/archive/{timestamp}/{archivefile}')
+        except KeyError:
+            pass
+
+        # slug is a URL
+        by_url = {page.base_url: page for page in all_pages}
+        try:
+            # TODO: add multiple snapshot support by showing index of all snapshots
+            # for given url instead of redirecting to timestamp index
+            timestamp = by_url[base_url(path)].timestamp
+            return redirect(f'/archive/{timestamp}/index.html')
+        except KeyError:
+            pass
+
+        return HttpResponse(
+            'No archived link matches the given timestamp or hash.',
+            content_type="text/plain",
+            status=404,
+        )

archivebox/themes/default/main_index.html

@@ -208,20 +208,24 @@
             <tbody>
                 {% for link in links %}
                     <tr>
-                        <td title="$timestamp">{{link.bookmarked_date}}</td>
+                        <td title="{{link.timestamp}}">{{link.bookmarked_date}}</td>
                         <td class="title-col">
-                            <a href="$archive_path/index.html"><img src="$favicon_url" class="link-favicon" decoding="async"></a>
+                            {% if link.is_archived %}
-                            <a href="$archive_path/$wget_url" title="{{link.title}}">
+                                <a href="archive/{{link.timestamp}}/index.html"><img src="archive/{{link.timestamp}}/favicon.ico" class="link-favicon" decoding="async"></a>
-                                <span data-title-for="{{link.url}}" data-archived="$is_archived">{{link.title}}</span>
+                            {% else %}
+                                <a href="archive/{{link.timestamp}}/index.html"><img src="{% static 'spinner.gif' %}" class="link-favicon" decoding="async"></a>
+                            {% endif %}
+                            <a href="archive/{{link.timestamp}}/{{link.canonical_outputs.wget_path}}" title="{{link.title}}">
+                                <span data-title-for="{{link.url}}" data-archived="{{link.is_archived}}">{{link.title}}</span>
                                 <small style="float:right">{{link.tags|default:''}}</small>
                             </a>
                         </td>
                         <td>
-                            <a href="$archive_path/index.html">📄 
+                            <a href="archive/{{link.timestamp}}/index.html">📄 
                                 <span data-number-for="{{link.url}}" title="Fetching any missing files...">{{link.num_outputs}} <img src="{% static 'spinner.gif' %}" class="files-spinner" decoding="async"/></span>
                             </a>
                         </td>
-                        <td style="text-align:left"><a href="$url">{{link.url}}</a></td>
+                        <td style="text-align:left"><a href="{{link.url}}">{{link.url}}</a></td>
                     </tr>
                 {% endfor %}
             </tbody>
@@ -231,10 +235,10 @@
             <center>
                 <small>
                     Archive created using <a href="https://github.com/pirate/ArchiveBox" title="Github">ArchiveBox</a>
-                    version <a href="https://github.com/pirate/ArchiveBox/tree/v$version" title="Git commit">v$version</a> &nbsp; | &nbsp; 
+                    version <a href="https://github.com/pirate/ArchiveBox/tree/v{{VERSION}}" title="Git commit">v{{VERSION}}</a> &nbsp; | &nbsp; 
                     Download index as <a href="index.json" title="JSON summary of archived links.">JSON</a>
                     <br/><br/>
-                    $footer_info
+                    {{FOOTER_INFO}}
                 </small>
             </center>
             <br/>

archivebox/themes/legacy/link_details.html

@@ -246,7 +246,7 @@
                         </a>
                     </div>
                     <div class="col-lg-8">
-                        <img src="$link_dir/favicon.ico" alt="Favicon">
+                        <img src="favicon.ico" alt="Favicon">
                         &nbsp;&nbsp;
                         $title
                         &nbsp;&nbsp;