Merge pull request #17 from uktrade/feature/orpd-52-sort

orpd-52 - feat:add sorting by relevance

orp/orp_search/config.py

@@ -56,4 +56,9 @@ def validate(self):
         if self.limit < 0:
             logger.error("limit must be a positive integer")
             return False
+
+        if self.sort_by:
+            if self.sort_by not in ["recently", "relevance"]:
+                logger.error("sort_by must be 'recently' or 'relevance'")
+                return False
         return True

orp/orp_search/public_gateway.py

@@ -1,5 +1,6 @@
 import logging
 
+import pandas as pd
 import requests  # type: ignore
 
 from jinja2 import Template
@@ -130,17 +131,45 @@ def search(self, config: SearchDocumentConfig):
 
             sorted_df = None
 
-            if config.sort_by == "recently_updated":
+            if config.sort_by == "recently":
                 # Sort the DataFrame by 'date_modified' in descending order
-                sorted_df = filtered_df.sort_values(
-                    by="date_modified", ascending=False
+                # Ensure 'date_issued' is in datetime format
+                filtered_df["date_issued"] = pd.to_datetime(
+                    filtered_df["date_issued"], format="%d/%m/%Y"
                 )
 
-            elif config.sort_by == "recently_published":
                 # Sort the DataFrame by 'date_issued' in descending order
                 sorted_df = filtered_df.sort_values(
                     by="date_issued", ascending=False
                 )
+            elif config.sort_by == "relevance":
+                # Calculate relevance score
+                # (based on the number of keywords found)
+                def calculate_relevance(row, search_terms):
+                    def score_text(text, terms):
+                        text_processed = text.replace(" ", "").lower()
+                        return sum(
+                            1
+                            for term in terms
+                            if term.replace(" ", "").lower() in text_processed
+                        )
+
+                    title_score = score_text(row["title"], search_terms)
+                    description_score = score_text(
+                        row["description"], search_terms
+                    )
+                    return title_score + description_score
+
+                filtered_df["relevance_score"] = filtered_df.apply(
+                    calculate_relevance,
+                    axis=1,
+                    search_terms=config.search_terms,
+                )
+
+                # Sort the DataFrame by 'relevance_score' in descending order
+                sorted_df = filtered_df.sort_values(
+                    by="relevance_score", ascending=False
+                )
 
             if sorted_df is not None:
                 results = sorted_df.to_dict(orient="records")

orp/orp_search/templates/orp.html

@@ -228,7 +228,7 @@ <h2 class="govuk-fieldset__heading">
             </label>
             <select class="govuk-select" id="sort" name="sort">
               <option value="date">Recently updated</option>
-              <option value="sort">???</option>
+              <option value="sort">Relevance</option>
             </select>
           </div>
         </div>

orp/orp_search/views.py

@@ -99,6 +99,10 @@ def search(request: HttpRequest) -> HttpResponse:
     else:
         logger.info("search query: %s", search_query)
 
+    sort_by = request.GET.get("sort", None)
+    if sort_by:
+        logger.info("sort by: %s", sort_by)
+
     logger.info("document types: %s", document_types)
     logger.info("page: %s", offset)
 
@@ -110,6 +114,9 @@ def search(request: HttpRequest) -> HttpResponse:
     if publisher:
         config.publisher_terms = publisher
 
+    if sort_by:
+        config.sort_by = sort_by
+
     # Check if the response is cached
     public_gateway = PublicGateway()
     search_results = public_gateway.search(config)