From ce471c9bb1e145c1d74038299949ddf525ca52f6 Mon Sep 17 00:00:00 2001
From: Kishore Nallan <kishore@kishorelive.com>
Date: Mon, 9 Jan 2017 19:17:53 +0530
Subject: [PATCH] Keep the hashset bounded by deleting the element to be
 replaced in heap from the set.

---
 include/topster.h  | 21 +++++++++++++++------
 src/collection.cpp |  4 ++--
 2 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/include/topster.h b/include/topster.h
index b541667a..a15fcc3c 100644
--- a/include/topster.h
+++ b/include/topster.h
@@ -33,18 +33,20 @@ struct Topster {
     }
 
     void add(const uint64_t &key, const uint64_t &match_score, const int64_t &primary_attr, const int64_t &secondary_attr){
-        if(dedup_keys.count(key) != 0) {
-            return ;
-        }
-
-        dedup_keys.insert(key);
-
         if (size >= MAX_SIZE) {
             if(!is_greater(data[0], match_score, primary_attr, secondary_attr)) {
                 // when incoming value is less than the smallest in the heap, ignore
                 return;
             }
 
+            if(dedup_keys.count(key) != 0) {
+                // when the key already exists, ignore
+                return ;
+            }
+
+            dedup_keys.erase(data[0].key);
+            dedup_keys.insert(key);
+
             data[0].key = key;
             data[0].match_score = match_score;
             data[0].primary_attr = primary_attr;
@@ -67,6 +69,13 @@ struct Topster {
                 i = next;
             }
         } else {
+            if(dedup_keys.count(key) != 0) {
+                // when the key already exists, ignore
+                return ;
+            }
+
+            dedup_keys.insert(key);
+
             data[size].key = key;
             data[size].match_score = match_score;
             data[size].primary_attr = primary_attr;
diff --git a/src/collection.cpp b/src/collection.cpp
index 8f4d5274..69b94549 100644
--- a/src/collection.cpp
+++ b/src/collection.cpp
@@ -537,9 +537,9 @@ void Collection::remove(std::string id) {
 }
 
 std::string Collection::get_seq_id_key(uint32_t seq_id) {
-    return collection_id + "_" + SEQ_ID_PREFIX + std::to_string(seq_id);
+    return std::to_string(collection_id) + "_" + SEQ_ID_PREFIX + std::to_string(seq_id);
 }
 
 std::string Collection::get_doc_id_key(std::string doc_id) {
-    return collection_id + "_" + DOC_ID_PREFIX + doc_id;
+    return std::to_string(collection_id) + "_" + DOC_ID_PREFIX + doc_id;
 }
\ No newline at end of file