search penalty

peachjam/models/core_document_model.py

@@ -679,6 +679,11 @@ def get_cited_work_frbr_uris(self):
 
         return work_frbr_uris
 
+    def search_penalty(self):
+        """Optionally provide a penalty for this document in search results. This cannot be zero or None."""
+        # provide a very small number instead of zero
+        return 0.0000001
+
 
 def file_location(instance, filename):
     if not instance.document.pk:

peachjam/models/generic_document.py

@@ -86,6 +86,13 @@ class Meta(CoreDocument.Meta):
     def __str__(self):
         return self.title
 
+    def search_penalty(self):
+        # non-principal (ie. amendment) works get a slight search penalty so that principal works
+        # tend to appear above them in search results
+        if self.metadata_json and self.metadata_json.get("principal", None) is False:
+            return 10.0
+        return super().search_penalty()
+
     def apply_labels(self):
         # label to indicate that this legislation is repealed
         label, _ = Label.objects.get_or_create(

peachjam_search/documents.py

@@ -92,6 +92,9 @@ class SearchableDocument(Document):
     nature_pt = fields.KeywordField()
 
     ranking = RankField(attr="work.ranking")
+    # a negative boost to search results; this must be a positive number, but is treated as a penalty
+    # it is applied linearly, simply reducing the score by this amount
+    penalty = RankField(attr="search_penalty", positive_score_impact=False)
 
     pages = fields.NestedField(
         properties={

peachjam_search/views.py

@@ -62,15 +62,20 @@ def construct_search(self, request, view, search_backend):
 class RankFeatureBackend(BaseSearchQueryBackend):
     @classmethod
     def construct_search(cls, request, view, search_backend):
+        # apply penalty as a linear change to the score
+        # DISABLED until the penalty field is populated
+        # queries = [Q("rank_feature", field="penalty", boost=1.0, linear={})]
+        queries = []
 
         if pj_settings().pagerank_boost_value:
             rank = Q(
                 "rank_feature",
                 field="ranking",
                 boost=pj_settings().pagerank_boost_value,
             )
-            return [rank]
-        return []
+            queries.append(rank)
+
+        return queries
 
 
 class NestedPageQueryBackend(BaseSearchQueryBackend):