실습 자바 및 랑크 인 4J 가이드

    dev.langchain4j
    langchain4j
    ${langchain4j.version}



    dev.langchain4j
    langchain4j-open-ai
    ${langchain4j.version}



    dev.langchain4j
    langchain4j-ollama
    ${langchain4j.version}

Component ID: PUMP-001. Type: Centrifugal Pump. Status: Running. Connected to: VALVE-001, PIPE-002. Location: Sector A.

Component ID: VALVE-001. Type: Gate Valve. Status: Open. Connected to: PUMP-001, TANK-A. Location: Sector A.

Component ID: SENSOR-T1. Type: Temperature Sensor. Monitors: PUMP-001 Casing. Reading: 65C. Unit: Celsius. Location: Sector A.

Component ID: SENSOR-P1. Type: Pressure Sensor. Monitors: PIPE-002. Reading: 150. Unit: PSI. Location: Sector B.

Component ID: MOTOR-001. Type: Electric Motor. Status: Running. Drives: PUMP-001. Location: Sector A.

Fault ID: F001. Description: High Temperature on PUMP-001. Possible Causes: Low lubrication, bearing wear, blocked outlet VALVE-001. Recommended Action: Check lubrication levels and bearing condition.

Event ID: E001. Description: Pressure drop in PIPE-002 below 100 PSI. Related Components: PUMP-001, VALVE-001, SENSOR-P1. Possible Causes: Leak in PIPE-002, PUMP-001 failure, VALVE-001 partially closed.

Rule ID: R001. Condition: If SENSOR-T1 reading > 80C. Action: Generate HIGH_TEMP_ALERT for PUMP-001. Priority: High.

Maintenance Note M001: PUMP-001 bearings last replaced 6 months ago. Next inspection due in 1 month.

Safety Procedure S001: Before servicing PUMP-001, ensure MOTOR-001 is locked out and VALVE-001 is closed.

package com.example; // Use your package name

package ca.bazlur.util;

import ca.bazlur.service.KnowledgeBaseService;
import dev.langchain4j.data.document.Document;
import dev.langchain4j.data.document.DocumentParser;
import dev.langchain4j.data.document.DocumentSplitter;
import dev.langchain4j.data.document.parser.TextDocumentParser;
import dev.langchain4j.data.document.splitter.DocumentSplitters;
import dev.langchain4j.data.segment.TextSegment;
import dev.langchain4j.model.embedding.EmbeddingModel;
import dev.langchain4j.model.ollama.OllamaEmbeddingModel;
import dev.langchain4j.model.openai.OpenAiEmbeddingModel; // Choose one
// import dev.langchain4j.model.ollama.OllamaEmbeddingModel; // Choose one
import dev.langchain4j.store.embedding.EmbeddingSearchRequest;
import dev.langchain4j.store.embedding.EmbeddingStore;
import dev.langchain4j.store.embedding.EmbeddingStoreIngestor;
import dev.langchain4j.store.embedding.inmemory.InMemoryEmbeddingStore;

import java.io.IOException;
import java.io.InputStream;
import java.net.URISyntaxException;
import java.net.URL;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.util.List;
import java.util.Objects;

public class KnowledgeBaseIngestor {

    /**
     * Loads documents from resource files, creates embeddings, and stores them in an in-memory store.
     *
     * @return An EmbeddingStore containing the processed knowledge base.
     * @throws URISyntaxException if the resource file paths are invalid.
     */
    public static EmbeddingStore ingestData() throws URISyntaxException, IOException {
        System.out.println("Starting knowledge base ingestion...");

        // --- 1. Load Documents ---
        Document componentsDoc = loadDocumentFromResource("components.txt", new TextDocumentParser());
        Document knowledgeDoc = loadDocumentFromResource("knowledge.txt", new TextDocumentParser());
        List documents = List.of(componentsDoc, knowledgeDoc);
        System.out.println("Documents loaded successfully.");

        // --- 2. Setup Embedding Model ---
        // Choose *one* embedding model provider:

        // Option A: OpenAI (Requires OPENAI_API_KEY environment variable or use "demo")
//      System.out.println("Initializing OpenAI Embedding Model...");
//      EmbeddingModel embeddingModel = OpenAiEmbeddingModel.builder()
//              .apiKey(System.getenv("OPENAI_API_KEY") != null ? System.getenv("OPENAI_API_KEY") : "demo")
//              .logRequests(true) // Optional: Log requests to OpenAI
//              .logResponses(true) // Optional: Log responses from OpenAI
//              .build();

        // Option B: Ollama (Requires Ollama server running locally)
        System.out.println("Initializing Ollama Embedding Model...");
        EmbeddingModel embeddingModel = OllamaEmbeddingModel.builder()
                .baseUrl(" // Default Ollama URL
                .modelName("llama3")
                .build();
        System.out.println("Embedding Model initialized.");

        // --- 3. Setup Embedding Store ---
        // We use a simple in-memory store for this demo.
        // For persistent storage, explore options like Chroma, Pinecone, Weaviate, etc.
        System.out.println("Initializing In-Memory Embedding Store...");
        EmbeddingStore embeddingStore = new InMemoryEmbeddingStore();
        System.out.println("Embedding Store initialized.");

        // --- 4. Setup Ingestion Pipeline ---
        // Define how documents are split into segments (chunking strategy)
        // recursive(maxSegmentSize, maxOverlap) splits text recursively, trying to keep paragraphs/sentences together.
        // 300 characters per segment, 30 characters overlap between segments.
        DocumentSplitter splitter = DocumentSplitters.recursive(300, 30);
        System.out.println("Using recursive document splitter (300 chars, 30 overlap).");

        // EmbeddingStoreIngestor handles splitting, embedding, and storing.
        EmbeddingStoreIngestor ingestor = EmbeddingStoreIngestor.builder()
                .documentSplitter(splitter)
                .embeddingModel(embeddingModel)
                .embeddingStore(embeddingStore)
                .build();

        // --- 5. Ingest Documents ---
        System.out.println("Ingesting documents into the embedding store...");
        ingestor.ingest(documents);
        System.out.println("Ingestion complete. Embedding store contains");

        return embeddingStore;
    }

    /**
     * Helper method to get the Path of a resource file.
     * Handles running from IDE and within a JAR file.
     * @param resourceName The name of the file in src/main/resources
     * @return The Path object for the resource.
     * @throws URISyntaxException If the resource URL is malformed.
     * @throws RuntimeException If the resource is not found.
     */
    private static Document loadDocumentFromResource(String resourceName, DocumentParser parser) throws IOException {
        try (InputStream inputStream = getResourceAsStream(resourceName)) {
            Objects.requireNonNull(inputStream, "Resource not found: " + resourceName);
            return parser.parse(inputStream);
        }
    }

    protected static InputStream getResourceAsStream(String resourceName) {
        return KnowledgeBaseService.class.getClassLoader().getResourceAsStream(resourceName);
    }

    public static void main(String[] args) {
        try {
            EmbeddingStore store = ingestData();
        } catch (URISyntaxException e) {
            System.err.println("Error finding resource files: " + e.getMessage());
            e.printStackTrace();
        } catch (Exception e) {
            System.err.println("An error occurred during ingestion: " + e.getMessage());
            e.printStackTrace();
        }
    }
}