From f3d4bb49249765a352b98080c33d4f8f497de06d Mon Sep 17 00:00:00 2001
From: Kishore Nallan <kishorenc@gmail.com>
Date: Fri, 27 Jan 2023 15:21:56 +0530
Subject: [PATCH] Limit text indexed for string sorting to prevent abuse.

---
 src/adi_tree.cpp | 3 ---
 src/index.cpp    | 5 +++--
 2 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/src/adi_tree.cpp b/src/adi_tree.cpp
index 056a994e..7deb9fdb 100644
--- a/src/adi_tree.cpp
+++ b/src/adi_tree.cpp
@@ -2,9 +2,6 @@
 #include <vector>
 #include "adi_tree.h"
 #include "logger.h"
-#include <set>
-
-//std::set<adi_node_t*> nodes;
 
 struct adi_node_t {
     uint16_t num_children;
diff --git a/src/index.cpp b/src/index.cpp
index f3101f0d..94f084cd 100644
--- a/src/index.cpp
+++ b/src/index.cpp
@@ -1069,10 +1069,11 @@ void Index::index_field_in_memory(const field& afield, std::vector<index_record>
 
             std::string raw_str = document[afield.name].get<std::string>();
             Tokenizer str_tokenizer("", true, false, "", {' '});
-            std::string processed_str;
             str_tokenizer.tokenize(raw_str);
 
-            str_tree->index(seq_id, raw_str);
+            if(!raw_str.empty()) {
+                str_tree->index(seq_id, raw_str.substr(0, 2000));
+            }
         }
     }
 }