[improvement][chat]Inject schema info into the prompt of LLMSqlCorrector.

headless/chat/src/main/java/com/tencent/supersonic/headless/chat/corrector/LLMSqlCorrector.java

@@ -1,6 +1,7 @@
 package com.tencent.supersonic.headless.chat.corrector;
 
 import com.tencent.supersonic.common.pojo.ChatApp;
+import com.tencent.supersonic.common.pojo.Text2SQLExemplar;
 import com.tencent.supersonic.common.pojo.enums.AppModule;
 import com.tencent.supersonic.common.util.ChatAppManager;
 import com.tencent.supersonic.headless.api.pojo.SemanticParseInfo;
@@ -34,11 +35,8 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
             + "please take a review and help correct it if necessary." + "\n#Rules: "
             + "\n1.ALWAYS follow the output format: `opinion=(POSITIVE|NEGATIVE),sql=(corrected sql if NEGATIVE; empty string if POSITIVE)`."
             + "\n2.NO NEED to check date filters as the junior engineer seldom makes mistakes in this regard."
-            + "\n3.DO NOT miss the AGGREGATE operator of metrics, always add it as needed."
+            + "\n3.SQL columns and values must be mentioned in the `#Schema`."
-            + "\n4.ALWAYS use `with` statement if nested aggregation is needed."
+            + "\n#Question:{{question}} #Schema:{{schema}} #InputSQL:{{sql}} #Response:";
-            + "\n5.ALWAYS enclose alias declared by `AS` command in underscores."
-            + "\n6.Alias created by `AS` command must be in the same language ast the `Question`."
-            + "\n#Question:{{question}} #InputSQL:{{sql}} #Response:";
 
     public LLMSqlCorrector() {
         ChatAppManager.register(APP_KEY, ChatApp.builder().prompt(INSTRUCTION).name("语义SQL修正")
@@ -67,12 +65,15 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
             return;
         }
 
+        Text2SQLExemplar exemplar = (Text2SQLExemplar)semanticParseInfo.getProperties()
+                .get(Text2SQLExemplar.PROPERTY_KEY);
+
         ChatLanguageModel chatLanguageModel =
                 ModelProvider.getChatModel(chatApp.getChatModelConfig());
         SemanticSqlExtractor extractor =
                 AiServices.create(SemanticSqlExtractor.class, chatLanguageModel);
         Prompt prompt = generatePrompt(chatQueryContext.getRequest().getQueryText(),
-                semanticParseInfo, chatApp.getPrompt());
+                semanticParseInfo, chatApp.getPrompt(), exemplar);
         SemanticSql s2Sql = extractor.generateSemanticSql(prompt.toUserMessage().singleText());
         keyPipelineLog.info("LLMSqlCorrector modelReq:\n{} \nmodelResp:\n{}", prompt.text(), s2Sql);
         if ("NEGATIVE".equals(s2Sql.getOpinion()) && StringUtils.isNotBlank(s2Sql.getSql())) {
@@ -81,10 +82,11 @@ public class LLMSqlCorrector extends BaseSemanticCorrector {
     }
 
     private Prompt generatePrompt(String queryText, SemanticParseInfo semanticParseInfo,
-            String promptTemplate) {
+            String promptTemplate, Text2SQLExemplar exemplar) {
         Map<String, Object> variable = new HashMap<>();
         variable.put("question", queryText);
         variable.put("sql", semanticParseInfo.getSqlInfo().getCorrectedS2SQL());
+        variable.put("schema", exemplar.getDbSchema());
 
         return PromptTemplate.from(promptTemplate).apply(variable);
     }

headless/core/src/main/java/com/tencent/supersonic/headless/core/translator/DefaultSemanticTranslator.java

@@ -44,6 +44,8 @@ public class DefaultSemanticTranslator implements SemanticTranslator {
             for (QueryOptimizer queryOptimizer : ComponentFactory.getQueryOptimizers()) {
                 queryOptimizer.rewrite(queryStatement);
             }
+            log.info("translated query SQL: [{}]",
+                    StringUtils.normalizeSpace(queryStatement.getSql()));
         } catch (Exception e) {
             queryStatement.setErrMsg(e.getMessage());
             log.error("Failed to translate query [{}]", e.getMessage(), e);