From 26fc3bc08c82295f2c0326c20011397cf9615b66 Mon Sep 17 00:00:00 2001
From: Alex Parsons <alex.parsons@mysociety.org>
Date: Mon, 23 Sep 2024 19:17:14 +0000
Subject: [PATCH] Re-enable popular searches

- Explicit number of IP addresses required.
- Reduce number of items listed to 5.
- Easier to add exclusions.
- Support for limiting by section.
---
 classes/Common.php                        |  2 +-
 classes/Homepage.php                      |  2 +-
 www/includes/easyparliament/searchlog.php | 53 +++++++++++++++++++----
 3 files changed, 47 insertions(+), 10 deletions(-)

diff --git a/classes/Common.php b/classes/Common.php
index 8086a3c554..529ce95959 100644
--- a/classes/Common.php
+++ b/classes/Common.php
@@ -6,7 +6,7 @@ class Common {
 
     public function getPopularSearches() {
         global $SEARCHLOG;
-        $popular_searches = $SEARCHLOG->popular_recent(10, 2000);
+        $popular_searches = $SEARCHLOG->popular_recent(5, 2000, 5);
 
         return $popular_searches;
     }
diff --git a/classes/Homepage.php b/classes/Homepage.php
index 13aeda313b..80f24d54e5 100644
--- a/classes/Homepage.php
+++ b/classes/Homepage.php
@@ -41,7 +41,7 @@ public function display() {
         $data["commons_dissolved"] = isset($dissolution[1]);
 
         $data['regional'] = $this->getRegionalList();
-        $data['popular_searches'] = []; #$common->getPopularSearches();
+        $data['popular_searches'] = $common->getPopularSearches();
         $data['urls'] = $this->getURLs();
         $data['calendar'] = $this->getCalendarData();
         $data['featured'] = $this->getEditorialContent();
diff --git a/www/includes/easyparliament/searchlog.php b/www/includes/easyparliament/searchlog.php
index d3dbe568a7..6db8e914c1 100644
--- a/www/includes/easyparliament/searchlog.php
+++ b/www/includes/easyparliament/searchlog.php
@@ -69,17 +69,50 @@ public function add($searchlogdata) {
     }
 
     // Select popular queries
-    public function popular_recent($count, $max_chars = null) {
+    public function popular_recent(int $count, ?int $max_chars = null, ?int $minimum_hits = 5,?string $section_string = null) {
+
+        // Banned terms
+        $banned_terms = ["twat"];
+
+        $query = "SELECT query_string, COUNT(DISTINCT ip_address) AS c FROM search_query_log
+            WHERE count_hits != 0 AND page_number=1
+            AND query_time > date_sub(NOW(), INTERVAL 3 DAY)";
+
+        // allow limiting by section (e.g. to show a Scotland specific search)
+        if (!empty($section_string)) {
+            $query .= " AND query_string LIKE :section_string";
+        }
+
+        $query .= " GROUP BY query_string HAVING c >= :minimum_hits
+            ORDER BY c desc LIMIT :count;";
+
+        $params = [
+            ':minimum_hits' => $minimum_hits,
+            ':count' => $count
+        ];
+
+        if (!empty($section_string)) {
+            $params[':section_string'] = '%' . $section_string . '%';
+        }
+
+        $q = $this->db->query($query, $params);
 
-        $q =  $this->db->query("SELECT query_string, count(*) AS c FROM search_query_log
-                WHERE count_hits != 0 AND query_string != 'twat'
-           AND query_string != 'suffragettes' AND page_number=1
-                AND query_time > date_sub(NOW(), INTERVAL 1 DAY)
-                GROUP BY query_string ORDER BY c desc LIMIT $count;");
 
         $popular_searches = array();
         foreach ($q as $row) {
-            array_push($popular_searches, $this->_db_row_to_array($row));
+            $row_array = $this->_db_row_to_array($row, $section_string);
+        
+            // exclude all queries where a banned term is part of the query
+            $banned = false;
+            foreach ($banned_terms as $term) {
+                if (stripos($row_array['query'], $term) !== false) {
+                    $banned = true;
+                    break;
+                }
+            }
+            if (!$banned) {
+                array_push($popular_searches, $row_array);
+            }
         }
 
         //maximum number of chars?
@@ -101,7 +134,7 @@ public function popular_recent($count, $max_chars = null) {
         return $popular_searches;
     }
 
-    public function _db_row_to_array($row) {
+    public function _db_row_to_array($row, $section_string = null) {
         $query = $row['query_string'];
         $this->SEARCHURL->insert(array('s'=>$query, 'pop'=>1));
         $url = $this->SEARCHURL->generate();
@@ -120,6 +153,10 @@ public function _db_row_to_array($row) {
         }
         $visible_name = preg_replace('/"/', '', $query);
 
+        if (isset($section_string)) {
+            $visible_name = preg_replace('/' . $section_string . '/', '', $visible_name);
+        }
+
         $rowarray = $row;
         $rowarray['query'] = $query;
         $rowarray['visible_name'] = $visible_name;