fix: improve file investigation results (#7376)

ietf/doc/templatetags/ietf_filters.py

@@ -906,13 +906,17 @@ def mtime(path):
     """Returns a datetime object representing mtime given a pathlib Path object"""
     return datetime.datetime.fromtimestamp(path.stat().st_mtime).astimezone(ZoneInfo(settings.TIME_ZONE))
 
+@register.filter
+def mtime_is_epoch(path):
+    return path.stat().st_mtime == 0
+
 @register.filter
 def url_for_path(path):
     """Consructs a 'best' URL for web access to the given pathlib Path object.
 
     Assumes that the path is into the Internet-Draft archive or the proceedings.
     """
-    if path.match(f"{settings.AGENDA_PATH}/**/*"):
+    if Path(settings.AGENDA_PATH) in path.parents:
         return (
             f"https://www.ietf.org/proceedings/{path.relative_to(settings.AGENDA_PATH)}"
         )

ietf/doc/utils.py

@@ -1388,14 +1388,18 @@ def investigate_fragment(name_fragment):
     can_verify = set()
     for root in [settings.INTERNET_DRAFT_PATH, settings.INTERNET_DRAFT_ARCHIVE_DIR]:
         can_verify.update(list(Path(root).glob(f"*{name_fragment}*")))
-
+    archive_verifiable_names = set([p.name for p in can_verify])
+    # Can also verify drafts in proceedings directories
     can_verify.update(list(Path(settings.AGENDA_PATH).glob(f"**/*{name_fragment}*")))
 
     # N.B. This reflects the assumption that the internet draft archive dir is in the
     # a directory with other collections (at /a/ietfdata/draft/collections as this is written)
-    unverifiable_collections = set(
+    unverifiable_collections = set([
+        p for p in
         Path(settings.INTERNET_DRAFT_ARCHIVE_DIR).parent.glob(f"**/*{name_fragment}*")
-    )
+        if p.name not in archive_verifiable_names
+    ])
+    
     unverifiable_collections.difference_update(can_verify)
 
     expected_names = set([p.name for p in can_verify.union(unverifiable_collections)])

ietf/templates/doc/investigate.html

@@ -23,12 +23,24 @@
               <th scope="col" data-sort="name">Name</th>
               <th scope="col" data-sort="modified">Last Modified On</th>
               <th scope="col" data-sort="link">Link</th>
+              <th scope="col" data-sort="source">Source</th>
             </tr>
           </thead>
           <tbody>
             {% for path in results.can_verify %}
               {% with url=path|url_for_path %}
-              <tr><td>{{path.name}}</td><td>{{path|mtime|date:"DATETIME_FORMAT"}}</td><td><a href="{{url}}">{{url}}</a></td></tr>
+                <tr>
+                    <td>{{path.name}}</td>
+                    <td>
+                        {% if path|mtime_is_epoch %}
+                          Timestamp has been lost (is Unix Epoch)
+                        {% else %}
+                          {{path|mtime|date:"DATETIME_FORMAT"}}
+                        {% endif %}
+                    </td>
+                    <td><a href="{{url}}">{{url}}</a></td>
+                    <td>{{path}}</td>
+                </tr>
               {% endwith %}
             {% endfor %}
           </tbody>
@@ -53,7 +65,13 @@
               {% with url=path|url_for_path %}
               <tr>
                 <td>{{path.name}}</td>
-                <td>{{path|mtime|date:"DATETIME_FORMAT"}}</td>
+                <td>
+                    {% if path|mtime_is_epoch %}
+                      Timestamp has been lost (is Unix Epoch)
+                    {% else %}
+                      {{path|mtime|date:"DATETIME_FORMAT"}}
+                    {% endif %}
+                </td>
                 <td><a href="{{url}}">{{url}}</a></td>
                 <td>{{path}}</td>
               </tr>
@@ -77,7 +95,13 @@
               {% with url=path|url_for_path %}
               <tr>
                 <td>{{path.name}}</td>
-                <td>{{path|mtime|date:"DATETIME_FORMAT"}}</td>
+                <td>
+                    {% if path|mtime_is_epoch %}
+                      Timestamp has been lost (is Unix Epoch)
+                    {% else %}
+                      {{path|mtime|date:"DATETIME_FORMAT"}}
+                    {% endif %}
+                </td>
                 <td><a href="{{url}}">{{url}}</a></td>
               </tr>
               {% endwith %}