fix the tests

davidkyle · Dec 11, 2024 · 71c1012 · 71c1012
1 parent 78ea3e7
commit 71c1012
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 23 deletions.
diff --git a/...ference/src/main/java/org/elasticsearch/xpack/inference/chunking/WordBoundaryChunker.java b/...ference/src/main/java/org/elasticsearch/xpack/inference/chunking/WordBoundaryChunker.java
@@ -96,10 +96,6 @@ List<ChunkPosition> chunkPositions(String input, int chunkSize, int overlap) {
             throw new IllegalArgumentException("Invalid chunking parameters, overlap [" + overlap + "] must be >= 0");
         }
 
-        if (input.isEmpty()) {
-            return List.of();
-        }
-
         var chunkPositions = new ArrayList<ChunkPosition>();
 
         // This position in the chunk is where the next overlapping chunk will start

diff --git a/...rc/test/java/org/elasticsearch/xpack/inference/chunking/SentenceBoundaryChunkerTests.java b/...rc/test/java/org/elasticsearch/xpack/inference/chunking/SentenceBoundaryChunkerTests.java
@@ -44,37 +44,37 @@ private List<String> textChunks(
     }
 
     public void testEmptyString() {
-        var chunks = new SentenceBoundaryChunker().chunk("", 100, randomBoolean());
+        var chunks = textChunks(new SentenceBoundaryChunker(), "", 100, randomBoolean());
         assertThat(chunks, hasSize(1));
         assertThat(chunks.get(0), Matchers.is(""));
     }
 
     public void testBlankString() {
-        var chunks = new SentenceBoundaryChunker().chunk("   ", 100, randomBoolean());
+        var chunks = textChunks(new SentenceBoundaryChunker(), "   ", 100, randomBoolean());
         assertThat(chunks, hasSize(1));
         assertThat(chunks.get(0), Matchers.is("   "));
     }
 
     public void testSingleChar() {
-        var chunks = new SentenceBoundaryChunker().chunk("   b", 100, randomBoolean());
+        var chunks = textChunks(new SentenceBoundaryChunker(), "   b", 100, randomBoolean());
         assertThat(chunks, Matchers.contains("   b"));
 
-        chunks = new SentenceBoundaryChunker().chunk("b", 100, randomBoolean());
+        chunks = textChunks(new SentenceBoundaryChunker(), "b", 100, randomBoolean());
         assertThat(chunks, Matchers.contains("b"));
 
-        chunks = new SentenceBoundaryChunker().chunk(". ", 100, randomBoolean());
+        chunks = textChunks(new SentenceBoundaryChunker(), ". ", 100, randomBoolean());
         assertThat(chunks, Matchers.contains(". "));
 
-        chunks = new SentenceBoundaryChunker().chunk(" , ", 100, randomBoolean());
+        chunks = textChunks(new SentenceBoundaryChunker(), " , ", 100, randomBoolean());
         assertThat(chunks, Matchers.contains(" , "));
 
-        chunks = new SentenceBoundaryChunker().chunk(" ,", 100, randomBoolean());
+        chunks = textChunks(new SentenceBoundaryChunker(), " ,", 100, randomBoolean());
         assertThat(chunks, Matchers.contains(" ,"));
     }
 
     public void testSingleCharRepeated() {
         var input = "a".repeat(32_000);
-        var chunks = new SentenceBoundaryChunker().chunk(input, 100, randomBoolean());
+        var chunks = textChunks(new SentenceBoundaryChunker(), input, 100, randomBoolean());
         assertThat(chunks, Matchers.contains(input));
     }
 

diff --git a/...ce/src/test/java/org/elasticsearch/xpack/inference/chunking/WordBoundaryChunkerTests.java b/...ce/src/test/java/org/elasticsearch/xpack/inference/chunking/WordBoundaryChunkerTests.java
@@ -72,10 +72,6 @@ public class WordBoundaryChunkerTests extends ESTestCase {
      * Use the chunk functions that return offsets where possible
      */
     List<String> textChunks(WordBoundaryChunker chunker, String input, int chunkSize, int overlap) {
-        if (input.isEmpty()) {
-            return List.of("");
-        }
-
         var chunkPositions = chunker.chunk(input, chunkSize, overlap);
         return chunkPositions.stream().map(p -> input.substring(p.start(), p.end())).collect(Collectors.toList());
     }
@@ -242,31 +238,31 @@ public void testWhitespace() {
     }
 
     public void testBlankString() {
-        var chunks = new WordBoundaryChunker().chunk("   ", 100, 10);
+        var chunks = textChunks(new WordBoundaryChunker(), "   ", 100, 10);
         assertThat(chunks, hasSize(1));
         assertThat(chunks.get(0), Matchers.is("   "));
     }
 
     public void testSingleChar() {
-        var chunks = new WordBoundaryChunker().chunk("   b", 100, 10);
+        var chunks = textChunks(new WordBoundaryChunker(), "   b", 100, 10);
         assertThat(chunks, Matchers.contains("   b"));
 
-        chunks = new WordBoundaryChunker().chunk("b", 100, 10);
+        chunks = textChunks(new WordBoundaryChunker(), "b", 100, 10);
         assertThat(chunks, Matchers.contains("b"));
 
-        chunks = new WordBoundaryChunker().chunk(". ", 100, 10);
+        chunks = textChunks(new WordBoundaryChunker(), ". ", 100, 10);
         assertThat(chunks, Matchers.contains(". "));
 
-        chunks = new WordBoundaryChunker().chunk(" , ", 100, 10);
+        chunks = textChunks(new WordBoundaryChunker(), " , ", 100, 10);
         assertThat(chunks, Matchers.contains(" , "));
 
-        chunks = new WordBoundaryChunker().chunk(" ,", 100, 10);
+        chunks = textChunks(new WordBoundaryChunker(), " ,", 100, 10);
         assertThat(chunks, Matchers.contains(" ,"));
     }
 
     public void testSingleCharRepeated() {
         var input = "a".repeat(32_000);
-        var chunks = new WordBoundaryChunker().chunk(input, 100, 10);
+        var chunks = textChunks(new WordBoundaryChunker(), input, 100, 10);
         assertThat(chunks, Matchers.contains(input));
     }