feat #565: add vector knowledge ingestion demo

src/test/java/com/cloudhandson/vpdbackoffice/service/VectorChunkerTest.java

@@ -0,0 +1,36 @@
+package com.cloudhandson.vpdbackoffice.service;
+
+import static org.assertj.core.api.Assertions.assertThat;
+import static org.assertj.core.api.Assertions.assertThatThrownBy;
+
+import org.junit.jupiter.api.Test;
+
+class VectorChunkerTest {
+
+  @Test
+  void keepsParagraphsTogetherUntilTheConfiguredLimit() {
+    var chunks = VectorChunker.chunk(
+        "첫 번째 문단입니다.\n\n두 번째 문단은 같은 지식자료입니다.", 80);
+
+    assertThat(chunks).hasSize(1);
+    assertThat(chunks.get(0).chunkNo()).isEqualTo(1);
+    assertThat(chunks.get(0).text()).contains("첫 번째", "두 번째");
+  }
+
+  @Test
+  void splitsLongTextAndNumbersChunksFromOne() {
+    var chunks = VectorChunker.chunk("문장 하나입니다. ".repeat(30), 100);
+
+    assertThat(chunks).hasSizeGreaterThan(1);
+    assertThat(chunks).extracting("chunkNo").containsExactlyElementsOf(
+        java.util.stream.IntStream.rangeClosed(1, chunks.size()).boxed().toList());
+    assertThat(chunks).allSatisfy(chunk -> assertThat(chunk.text()).isNotBlank());
+  }
+
+  @Test
+  void rejectsBlankContent() {
+    assertThatThrownBy(() -> VectorChunker.chunk("  ", 600))
+        .isInstanceOf(AppException.class)
+        .hasMessageContaining("청킹할 문서 본문");
+  }
+}