.

src/main/kotlin/network/RagService.kt

@@ -1,23 +1,36 @@
-// src/main/kotlin/network/RagService.kt
+package network// src/main/kotlin/network/RagService.kt
 
 import dev.langchain4j.community.rag.content.retriever.lucene.LuceneEmbeddingStore
 import dev.langchain4j.data.document.Metadata
-import dev.langchain4j.data.message.UserMessage
 import dev.langchain4j.data.segment.TextSegment
+import dev.langchain4j.exception.InternalServerException
 import dev.langchain4j.model.openai.OpenAiChatModel
 import dev.langchain4j.model.openai.OpenAiEmbeddingModel
+import dev.langchain4j.service.AiServices
+import dev.langchain4j.service.SystemMessage
 import dev.langchain4j.store.embedding.EmbeddingSearchRequest
 import dev.langchain4j.store.embedding.filter.MetadataFilterBuilder
+import kotlinx.coroutines.Dispatchers
 import kotlinx.coroutines.async
 import kotlinx.coroutines.coroutineScope
+import kotlinx.coroutines.withContext
 import kotlinx.serialization.Serializable
 import kotlinx.serialization.json.Json
-import model.CandleData
-import network.DartCodeManager
-import network.FinancialMapper
-import network.FinancialStatement
-import network.NewsService
+import kotlinx.serialization.json.add
+import kotlinx.serialization.json.addJsonObject
+import kotlinx.serialization.json.buildJsonObject
+import kotlinx.serialization.json.jsonArray
+import kotlinx.serialization.json.jsonObject
+import kotlinx.serialization.json.jsonPrimitive
+import kotlinx.serialization.json.put
+import kotlinx.serialization.json.putJsonArray
+import kotlinx.serialization.json.putJsonObject
+import okhttp3.MediaType.Companion.toMediaType
+import okhttp3.OkHttpClient
+import okhttp3.Request
+import okhttp3.RequestBody.Companion.toRequestBody
 import org.apache.lucene.store.MMapDirectory
+import org.slf4j.MDC.put
 import service.FinancialAnalyzer
 import service.InvestmentScores
 import service.TechnicalAnalyzer
@@ -25,9 +38,10 @@ import service.TradingDecisionCallback
 import service.UrlCacheManager
 import java.nio.file.Paths
 import java.time.Duration
+import java.util.concurrent.TimeUnit
 
 interface TradingAnalyst {
-    @dev.langchain4j.service.SystemMessage("""
+    @SystemMessage("""
         You are a Senior Stock Analyst. 
         Analyze the data and provide a decision in JSON format.
         You must respond ONLY with a valid JSON object.
@@ -48,13 +62,13 @@ object RagService {
         .apiKey("unused")
         .temperature(0.0) // [중요] 0.0으로 설정하여 결정론적 응답 유도
         .timeout(Duration.ofSeconds(60))
-        .frequencyPenalty(1.1)
-        .maxTokens(500) // 👈 루프 방지를 위해 반드시 짧게 제한!
+//        .frequencyPenalty(1.1)
+        .maxTokens(400) // 👈 루프 방지를 위해 반드시 짧게 제한!
         // 1.x 버전에서는 responseFormat이 아래처럼 바뀔 수 있으니 체크하세요
         .responseFormat("json_object")
         .build()
 
-    private val analyst = dev.langchain4j.service.AiServices.builder(TradingAnalyst::class.java)
+    private val analyst = AiServices.builder(TradingAnalyst::class.java)
         .chatModel(chatModel)
         .build()
 
@@ -139,8 +153,18 @@ object RagService {
 
     object JsonSanitizer {
         fun formatJson(raw: String): String {
+            // 실제 응답 로그 출력 (디버깅용)
+            println("📥 [AI Raw Response]:\n$raw")
+
             val regex = Regex("""\{.*\}""", RegexOption.DOT_MATCHES_ALL)
-            return raw.trim()
+            val match = regex.find(raw)?.value
+
+            if (match == null) {
+                println("⚠️ [JsonSanitizer] JSON 형식을 찾을 수 없습니다.")
+                return "{}" // 빈 객체라도 반환하여 EOF 방지
+            }
+
+            return match.trim()
                 .removePrefix("```json")
                 .removePrefix("```")
                 .removeSuffix("```")
@@ -264,13 +288,80 @@ object RagService {
 ////        println(response)
 //        return response.aiMessage().text()
 //    }
+    private const val LLM_API_URL = "http://127.0.0.1:8080/v1/chat/completions"
 
+    private suspend fun callLlamaWithSchema(prompt: String): String {
+        val jsonMediaType = "application/json; charset=utf-8".toMediaType()
+
+        // 문자열 치환 대신 안전한 JSON 객체 빌더 사용
+        val requestBodyJson = buildJsonObject {
+            put("model", "local-model")
+            put("temperature", 0.1) // 0.1 유지 (결정론적 응답)
+            put("top_p", 0.9)
+            put("max_tokens", 500)
+
+            putJsonArray("messages") {
+                addJsonObject {
+                    put("role", "system")
+                    put("content", "You are a helpful AI financial analyst. You must output responses ONLY in valid JSON format.")
+                }
+                addJsonObject {
+                    put("role", "user")
+                    put("content", prompt)
+                }
+            }
+
+            // 💡 복잡한 json_schema를 지우고, 단순히 JSON 형식으로만 내보내라고 지시합니다.
+            putJsonObject("response_format") {
+                put("type", "json_object")
+            }
+        }.toString()
+        println("requestBodyJson =>> $requestBodyJson")
+        val request = Request.Builder()
+            .url(LLM_API_URL)
+            .post(requestBodyJson.toRequestBody(jsonMediaType))
+            .build()
+
+        return kotlinx.coroutines.Dispatchers.IO.let {
+            kotlinx.coroutines.withContext(it) {
+                httpClient.newCall(request).execute().use { response ->
+                    if (!response.isSuccessful) throw Exception("LLM API Error: ${response.code} ${response.message}")
+
+                    val responseBody = response.body?.string() ?: "{}"
+                    val json = Json.parseToJsonElement(responseBody).jsonObject
+                    json["choices"]?.jsonArray?.get(0)?.jsonObject?.get("message")?.jsonObject?.get("content")?.jsonPrimitive?.content ?: "{}"
+                }
+            }
+        }
+    }
+
+    private val httpClient = OkHttpClient.Builder()
+        .connectTimeout(60, TimeUnit.SECONDS)
+        .readTimeout(120, TimeUnit.SECONDS)
+        .build()
     suspend fun decideTrading(
         stockName: String,
         scores: InvestmentScores, // 직접 계산한 점수 객체
         financialStmt: FinancialStatement, // 매핑된 재무 수치 객체
         tempDecision: TradingDecision
     ): TradingDecision? {
+        // 💡 1. 뉴스 데이터가 유효한지(100자 이상인지) 확인
+        val validNews = tempDecision.newsContext?.takeIf { it.trim().length >= 100 }
+
+        // 💡 2. 동적 데이터 섹션 구성
+        val newsDataSection = if (validNews != null) {
+            "3. News Context: $validNews"
+        } else {
+            "3. News Context: None available. Base your decision ONLY on System Scores and Financials."
+        }
+
+        // 💡 3. 동적 제약 조건 구성
+        val newsConstraint = if (validNews != null) {
+            "- Match Financials with News: If profit is negative but news is hyped, stay CAUTIOUS (HOLD)."
+        } else {
+            "- No news data is available. Rely strictly on Financials and System Scores for your 'decision' and 'reason'."
+        }
+
         val prompt = """
 # Task: Senior AI Investment Analyst
 Analyze the stock '$stockName' and determine the final trading decision based on the data below.
@@ -278,51 +369,52 @@ Analyze the stock '$stockName' and determine the final trading decision based on
 # Data
 1. System Scores: Scalping(${scores.ultraShort}), Short(${scores.shortTerm}), Mid(${scores.midTerm}), Long(${scores.longTerm})
 2. Financials: Operating Profit ${if(financialStmt.isOperatingProfitPositive) "PROFIT" else "LOSS"} (Growth: ${"%.2f".format(financialStmt.operatingProfitGrowth)}%), ROE: ${"%.2f".format(financialStmt.roe)}%, Debt: ${"%.2f".format(financialStmt.debtRatio)}%
-3. News Context: ${tempDecision.newsContext?.take(400)} // 👈 뉴스 길이를 물리적으로 제한
+$newsDataSection
 
 # Constraints
-1. 모든 점수와 confidence는 0에서 100 사이의 **정수(Integer)**로만 작성하십시오.
+- Copy the exact 'System Scores' from the Data section into the output JSON.
 - Match Financials with News: If profit is negative but news is hyped, stay CAUTIOUS (HOLD).
-- Synchronization: High scalping score + positive news momentum = Higher BUY confidence.
-- Output: Response ONLY in valid JSON format. No extra text.
+- The "reason" field MUST be written in KOREAN and MUST NOT exceed 50 characters. Keep it concise.
+- Output ONLY a valid JSON object matching the exact structure below. DO NOT output placeholder text like '(integer)'.
 
-# Output JSON Format (Reason must be in Korean)
+# Example Output JSON Format
 {
-  "ultraShortScore": ${scores.ultraShort},
-  "shortTermScore": ${scores.shortTerm},
-  "midTermScore": ${scores.midTerm},
-  "longTermScore": ${scores.longTerm},
-  "decision": "BUY/SELL/HOLD",
-  "reason": "재무와 뉴스를 대조한 분석 결과 (한국어)",
-  "confidence": 0~100
+  "ultraShortScore": 0,
+  "shortTermScore": 0,
+  "midTermScore": 0,
+  "longTermScore": 0,
+  "decision": "HOLD",
+  "reason": "적자 지속 및 네수파립 임상 대기 중으로 관망 필요",
+  "confidence": 50
 }
 """.trimIndent()
 
-        val response = chatModel.chat(UserMessage.from(prompt))
-        val rawResponse = response.aiMessage().text()
-        val jsonResponse = JsonSanitizer.formatJson(rawResponse)
-//        println("📥 [AI Raw JSON]:\n$jsonResponse")
 
-        // 2. 유연한 파서 설정 (소수점 및 예외 상황 대응)
-        val lenientJson = Json {
-            ignoreUnknownKeys = true
-            isLenient = true
-            coerceInputValues = true
-        }
-
-
-        // JSON 파싱 (Kotlinx Serialization 활용)
         return try {
-//            println(jsonResponse)
-            val decision = lenientJson.decodeFromString<TradingDecision>(jsonResponse)
-            decision.financialData = tempDecision.financialData
-            decision.newsContext = tempDecision.newsContext
-            decision.techSummary = tempDecision.techSummary
-            decision.stockCode = tempDecision.stockCode
-            decision.corpName = tempDecision.corpName
-            decision.stockName = tempDecision.stockName
+            val rawResponse = callLlamaWithSchema(prompt)
+            println("📥 [AI Strict JSON]:\n$rawResponse")
+
+            // 엄격한 스키마가 적용되었으므로 JsonSanitizer 없이 바로 파싱 가능
+            val lenientJson = Json {
+                ignoreUnknownKeys = true
+                isLenient = true
+                coerceInputValues = true
+            }
+
+            val decision = lenientJson.decodeFromString<TradingDecision>(rawResponse)
+
+            // 데이터 매핑
+            decision.apply {
+                financialData = tempDecision.financialData
+                newsContext = tempDecision.newsContext
+                techSummary = tempDecision.techSummary
+                stockCode = tempDecision.stockCode
+                corpName = tempDecision.corpName
+                this.stockName = tempDecision.stockName
+            }
+
             decision
-        } catch (e: dev.langchain4j.exception.InternalServerException) {
+        } catch (e: InternalServerException) {
             // 서버 에러 (컨텍스트 초과 등) 발생 시 로그 남기고 null 반환 혹은 커스텀 에러 처리
             println("🚨 [AI Server Error] ${e.message}")
             if (e.message?.contains("Context size") == true) {