From 2b251b15b1abc452fcf041b79e8d4c1b322d14b9 Mon Sep 17 00:00:00 2001
From: elliVM <126466762+elliVM@users.noreply.github.com>
Date: Fri, 15 Nov 2024 13:31:25 +0200
Subject: [PATCH] Lowercase search term tokens (#117)

* lowercase search term tokens

* apply spotless
---
 .../planner/bloomfilter/SearchTermBloomFilter.java       | 2 +-
 .../planner/bloomfilter/SearchTermBloomFilterTest.java   | 9 +++++----
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/src/main/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilter.java b/src/main/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilter.java
index a30dc70f..4bd37c0f 100644
--- a/src/main/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilter.java
+++ b/src/main/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilter.java
@@ -86,7 +86,7 @@ public byte[] bytes() {
         }
         final BloomFilter filter = BloomFilter.create(expected, fpp);
         for (final String token : stringTokens) {
-            filter.put(token);
+            filter.put(token.toLowerCase());
         }
         try (final ByteArrayOutputStream filterBAOS = new ByteArrayOutputStream()) {
             filter.writeTo(filterBAOS);
diff --git a/src/test/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilterTest.java b/src/test/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilterTest.java
index ba9f6fc3..a46d6ede 100644
--- a/src/test/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilterTest.java
+++ b/src/test/java/com/teragrep/pth_06/planner/bloomfilter/SearchTermBloomFilterTest.java
@@ -100,7 +100,7 @@ public void testRegexExtractedTokens() {
         byte[] bytes = Assertions.assertDoesNotThrow(filter::bytes);
         BloomFilter resultFilter = Assertions
                 .assertDoesNotThrow(() -> BloomFilter.readFrom(new ByteArrayInputStream(bytes)));
-        Assertions.assertTrue(resultFilter.mightContain("Pattern"));
+        Assertions.assertTrue(resultFilter.mightContain("pattern"));
     }
 
     @Test
@@ -110,9 +110,10 @@ public void testTokenizerTokens() {
         byte[] bytes = Assertions.assertDoesNotThrow(filter::bytes);
         BloomFilter resultFilter = Assertions
                 .assertDoesNotThrow(() -> BloomFilter.readFrom(new ByteArrayInputStream(bytes)));
-        Assertions.assertFalse(resultFilter.mightContain("Pattern"));
-        Assertions.assertTrue(resultFilter.mightContain("Without"));
-        Assertions.assertTrue(resultFilter.mightContain("SearchValuePatternInThisString"));
+        // test that tokens present and in lower case
+        Assertions.assertFalse(resultFilter.mightContain("pattern"));
+        Assertions.assertTrue(resultFilter.mightContain("without"));
+        Assertions.assertTrue(resultFilter.mightContain("searchvaluepatterninthisstring"));
     }
 
     @Test