|
| 1 | +import dev.langchain4j.community.store.embedding.yugabytedb.DefaultMetadataStorageConfig; |
| 2 | +import dev.langchain4j.community.store.embedding.yugabytedb.MetadataStorageConfig; |
| 3 | +import dev.langchain4j.community.store.embedding.yugabytedb.MetadataStorageMode; |
| 4 | +import dev.langchain4j.community.store.embedding.yugabytedb.YugabyteDBEmbeddingStore; |
| 5 | +import dev.langchain4j.community.store.embedding.yugabytedb.YugabyteDBEngine; |
| 6 | +import dev.langchain4j.data.document.Metadata; |
| 7 | +import dev.langchain4j.data.embedding.Embedding; |
| 8 | +import dev.langchain4j.data.segment.TextSegment; |
| 9 | +import dev.langchain4j.model.embedding.EmbeddingModel; |
| 10 | +import dev.langchain4j.model.embedding.onnx.allminilml6v2.AllMiniLmL6V2EmbeddingModel; |
| 11 | +import dev.langchain4j.store.embedding.EmbeddingMatch; |
| 12 | +import dev.langchain4j.store.embedding.EmbeddingSearchRequest; |
| 13 | +import dev.langchain4j.store.embedding.EmbeddingStore; |
| 14 | +import dev.langchain4j.store.embedding.filter.Filter; |
| 15 | +import dev.langchain4j.store.embedding.filter.comparison.IsEqualTo; |
| 16 | +import org.testcontainers.containers.GenericContainer; |
| 17 | +import org.testcontainers.containers.wait.strategy.Wait; |
| 18 | +import org.testcontainers.utility.DockerImageName; |
| 19 | + |
| 20 | +import java.time.Duration; |
| 21 | +import java.util.List; |
| 22 | + |
| 23 | +public class YugabyteDBEmbeddingStoreWithMetadataExample { |
| 24 | + |
| 25 | + public static void main(String[] args) { |
| 26 | + GenericContainer<?> yugabyteContainer = null; |
| 27 | + YugabyteDBEngine engine = null; |
| 28 | + |
| 29 | + try { |
| 30 | + DockerImageName dockerImageName = DockerImageName.parse("yugabytedb/yugabyte:2025.1.0.1-b3"); |
| 31 | + yugabyteContainer = new GenericContainer<>(dockerImageName) |
| 32 | + .withExposedPorts(5433, 7000, 9000, 15433, 9042) |
| 33 | + .withCommand("bin/yugabyted", "start", "--background=false") |
| 34 | + .waitingFor(Wait.forListeningPorts(5433).withStartupTimeout(Duration.ofMinutes(5))); |
| 35 | + |
| 36 | + yugabyteContainer.start(); |
| 37 | + |
| 38 | + EmbeddingModel embeddingModel = new AllMiniLmL6V2EmbeddingModel(); |
| 39 | + |
| 40 | + // Create YugabyteDB engine with PostgreSQL driver |
| 41 | + engine = YugabyteDBEngine.builder() |
| 42 | + .host(yugabyteContainer.getHost()) |
| 43 | + .port(yugabyteContainer.getMappedPort(5433)) |
| 44 | + .database("yugabyte") |
| 45 | + .username("yugabyte") |
| 46 | + .password("yugabyte") |
| 47 | + .usePostgreSQLDriver(true) |
| 48 | + .build(); |
| 49 | + |
| 50 | + // Configure metadata storage (JSONB format) |
| 51 | + MetadataStorageConfig metadataConfig = DefaultMetadataStorageConfig.builder() |
| 52 | + .storageMode(MetadataStorageMode.COMBINED_JSONB) |
| 53 | + .build(); |
| 54 | + |
| 55 | + EmbeddingStore<TextSegment> embeddingStore = YugabyteDBEmbeddingStore.builder() |
| 56 | + .engine(engine) |
| 57 | + .tableName("test_embeddings_with_metadata") |
| 58 | + .dimension(embeddingModel.dimension()) |
| 59 | + .metadataStorageConfig(metadataConfig) |
| 60 | + .createTableIfNotExists(true) |
| 61 | + .build(); |
| 62 | + |
| 63 | + // Add embeddings with metadata |
| 64 | + TextSegment segment1 = TextSegment.from("I like football.", |
| 65 | + Metadata.from("category", "sports").put("user", "john")); |
| 66 | + Embedding embedding1 = embeddingModel.embed(segment1).content(); |
| 67 | + embeddingStore.add(embedding1, segment1); |
| 68 | + |
| 69 | + TextSegment segment2 = TextSegment.from("The weather is good today.", |
| 70 | + Metadata.from("category", "weather").put("user", "alice")); |
| 71 | + Embedding embedding2 = embeddingModel.embed(segment2).content(); |
| 72 | + embeddingStore.add(embedding2, segment2); |
| 73 | + |
| 74 | + TextSegment segment3 = TextSegment.from("I love basketball.", |
| 75 | + Metadata.from("category", "sports").put("user", "bob")); |
| 76 | + Embedding embedding3 = embeddingModel.embed(segment3).content(); |
| 77 | + embeddingStore.add(embedding3, segment3); |
| 78 | + |
| 79 | + // Search with metadata filter |
| 80 | + Embedding queryEmbedding = embeddingModel.embed("What sport do you like?").content(); |
| 81 | + |
| 82 | + Filter categoryFilter = new IsEqualTo("category", "sports"); |
| 83 | + |
| 84 | + EmbeddingSearchRequest searchRequest = EmbeddingSearchRequest.builder() |
| 85 | + .queryEmbedding(queryEmbedding) |
| 86 | + .maxResults(5) |
| 87 | + .filter(categoryFilter) |
| 88 | + .build(); |
| 89 | + |
| 90 | + List<EmbeddingMatch<TextSegment>> relevant = embeddingStore.search(searchRequest).matches(); |
| 91 | + |
| 92 | + System.out.println("Found " + relevant.size() + " sports-related results:"); |
| 93 | + for (EmbeddingMatch<TextSegment> match : relevant) { |
| 94 | + System.out.println("Score: " + match.score()); |
| 95 | + System.out.println("Text: " + match.embedded().text()); |
| 96 | + System.out.println("Metadata: " + match.embedded().metadata()); |
| 97 | + System.out.println("---"); |
| 98 | + } |
| 99 | + |
| 100 | + System.out.println("\n✅ Example completed successfully!"); |
| 101 | + |
| 102 | + } catch (Exception e) { |
| 103 | + System.err.println("❌ Error running example: " + e.getMessage()); |
| 104 | + e.printStackTrace(); |
| 105 | + } finally { |
| 106 | + // Give Testcontainers time to cleanup gracefully |
| 107 | + try { |
| 108 | + Thread.sleep(2000); |
| 109 | + } catch (InterruptedException ignored) { |
| 110 | + } |
| 111 | + |
| 112 | + // Cleanup resources |
| 113 | + System.out.println("🧹 Cleaning up resources..."); |
| 114 | + if (engine != null) { |
| 115 | + try { |
| 116 | + engine.close(); |
| 117 | + } catch (Exception e) { |
| 118 | + System.err.println("Error closing engine: " + e.getMessage()); |
| 119 | + } |
| 120 | + } |
| 121 | + if (yugabyteContainer != null) { |
| 122 | + try { |
| 123 | + yugabyteContainer.stop(); |
| 124 | + } catch (Exception e) { |
| 125 | + System.err.println("Error stopping container: " + e.getMessage()); |
| 126 | + } |
| 127 | + } |
| 128 | + } |
| 129 | + } |
| 130 | +} |
| 131 | + |
0 commit comments