Reduce redundancy in rulemaking single page document listings

fec/data/templatetags/filters.py

@@ -4,6 +4,7 @@
 import re
 
 from dateutil.parser import parse as parse_date
+from difflib import SequenceMatcher
 
 from django.conf import settings
 from django_jinja import library
@@ -164,6 +165,43 @@ def filesize(value):
     return '%d %s' % (value, units[unit])
 
 
+@library.filter
+def compare(string, string_1):
+    """Compares two strings to determine if they are nearly identical or semantically similar. 
+    Returns False if `string` is >= than 50% similar to `string_1`. Else returns True.
+    Uses difflib.SequenceMatcher: https://docs.python.org/3/library/difflib.html
+    """
+
+    # Calculate the similarity ratio
+    similarity_ratio = SequenceMatcher(None, string, string_1).ratio()
+
+    # Define a threshold for redundancy (e.g., 50% similar)
+    threshold = .5
+
+    if similarity_ratio >= threshold:
+        return False
+
+    else:
+        return True
+
+
+@library.filter
+def compare_commenter(string, string_1):
+    """Convert `string_1` from "LastName, FirstName" format to "FirstName LastName" 
+    Returns False if `converted_string_1` is in `string`.
+    Returns False if `string` == 'Comment'.
+    Else returns True.
+    """
+
+    # Convert name from "LastName, FirstName" format to "FirstName LastName"
+    converted_string_1 = ' '.join(reversed(string_1.split(', ')))
+
+    if converted_string_1 in string or string == 'Comment':
+        return False
+    else:
+        return True
+
+
 @library.global_function
 def path_for_css(key):
     """Looks up the hashed asset key in rev-manifest-css.json

fec/legal/templates/rulemaking.jinja

@@ -71,16 +71,15 @@
       <ul>
         <li>
           {%- if docL1.url -%}
-          <a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ docL1.url }}" data-doc_id="{{ docL1.doc_id }}" target="_blank">{{ docL1.label }}</a>
+           <a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ docL1.url }}" data-doc_id="{{ docL1.doc_id }}" target="_blank">{{ docL1.label }}</a>
           {%- else -%}
-          {{ docL1.label }}
+           {{ docL1.label }}
           {%- endif -%}
-          {%- if docL1.doc_description and 'pdf' not in docL1.doc_description.lower() -%}
-            <br>{{ docL1.doc_description }}
+          {%- if docL1.doc_description and 'pdf' not in docL1.doc_description.lower() and docL1.doc_description | compare(docL1.label) -%}
+           <br>{{ docL1.doc_description }}
           {%- endif -%}
         </li>
           {%- if docL1.doc_entities -%}
-
                 {%- for entity in docL1.doc_entities -%}
                   <li class="indent">{{ entity.name }}, {{ entity.role }}</li>
                 {%- endfor -%}
@@ -105,7 +104,7 @@
           <ul>
             {% if (doc.label == 'Comments' or doc.label == 'Hearing Schedule') and doc.doc_entities %}
               {% if doc.label == 'Hearing Schedule' %}
-                <li><a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">{{ doc.label }}</a></li>
+               <li><a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">{{ doc.label }}</a></li>
                 {% for entity in doc.doc_entities %}
                   <li class="indent">{{ entity.name }}, {{ entity.role }}</li>
                 {% endfor %}
@@ -114,25 +113,30 @@
                 {% if doc.commenters|length == 1 %}
                   {# Single commenter: show commenter name #}
                   <li>
-                  <a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">{{ doc.commenters[0].name }} | {{ doc.commenters[0].role }}</a>
-                  {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() %}<br>{{ doc.doc_description }}{% endif %}
+                    <a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">{{ doc.commenters[0].name }} | {{ doc.commenters[0].role }}</a>
+                    {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() and doc.doc_description | compare_commenter(doc.commenters[0].name)  %}
+                     <br>{{ doc.doc_description }} 
+                    {% endif %}
                   </li>
-                  {% for entity in doc.doc_entities %}
+                  {% for entity in doc.doc_entities if entity.role != 'Commenter' %}
                     <li class="indent">{{ entity.name }}, {{ entity.role }}</li>
                   {% endfor %}
                 {% else %}
                   {# Multiple commenters: show "Joint comment" #}
                   <li><a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">Joint comment</a>
-                  {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() %}<br>{{ doc.doc_description }}{% endif %}</li>
+                    {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() %}
+                      <br>{{ doc.doc_description }} 
+                    {% endif %}
+                  </li>
                   {% for entity in doc.doc_entities %}
                     <li class="indent">{{ entity.name }}, {{ entity.role }}</li>
                   {% endfor %}
                 {% endif %}
               {% endif %}
             {% else %}
               <a class="embed-pdf" href="{{ CANONICAL_BASE }}{{ doc.url }}" data-doc_id="{{ doc.doc_id }}" target="_blank">{{ doc.label }}</a>
-              {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() %}
-                <br>{{ doc.doc_description }}
+              {% if doc.doc_description and 'pdf' not in doc.doc_description.lower() and doc.doc_description | compare(doc.label)  %}
+               <br>{{ doc.doc_description }} 
               {% endif %}
             {% endif %}
           </ul>