mirror of
https://github.com/tencentmusic/supersonic.git
synced 2025-12-13 13:07:32 +00:00
Merge a number of fixes and improvements (#1896)
This commit is contained in:
@@ -0,0 +1,133 @@
|
||||
package com.tencent.supersonic.headless.server.modeller;
|
||||
|
||||
import com.fasterxml.jackson.databind.ObjectMapper;
|
||||
import com.tencent.supersonic.common.config.ChatModel;
|
||||
import com.tencent.supersonic.common.pojo.ChatApp;
|
||||
import com.tencent.supersonic.common.pojo.ChatModelConfig;
|
||||
import com.tencent.supersonic.common.pojo.enums.AppModule;
|
||||
import com.tencent.supersonic.common.service.ChatModelService;
|
||||
import com.tencent.supersonic.common.util.ChatAppManager;
|
||||
import com.tencent.supersonic.common.util.ContextUtils;
|
||||
import com.tencent.supersonic.common.util.JsonUtil;
|
||||
import com.tencent.supersonic.headless.api.pojo.DbSchema;
|
||||
import com.tencent.supersonic.headless.api.pojo.ModelSchema;
|
||||
import com.tencent.supersonic.headless.api.pojo.request.ModelBuildReq;
|
||||
import dev.langchain4j.model.chat.ChatLanguageModel;
|
||||
import dev.langchain4j.model.input.Prompt;
|
||||
import dev.langchain4j.model.input.PromptTemplate;
|
||||
import dev.langchain4j.provider.ModelProvider;
|
||||
import dev.langchain4j.service.AiServices;
|
||||
import lombok.extern.slf4j.Slf4j;
|
||||
import org.springframework.core.env.Environment;
|
||||
import org.springframework.core.io.ClassPathResource;
|
||||
|
||||
import java.io.InputStream;
|
||||
import java.util.HashMap;
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
import java.util.Optional;
|
||||
import java.util.stream.Collectors;
|
||||
|
||||
@Slf4j
|
||||
public class LLMSemanticModeller implements SemanticModeller {
|
||||
|
||||
public static final String APP_KEY = "BUILD_DATA_MODEL";
|
||||
|
||||
private static final String SYS_EXEMPLAR_FILE = "s2-buildModel-exemplar.json";
|
||||
|
||||
public static final String INSTRUCTION = ""
|
||||
+ "Role: As an experienced data analyst with extensive modeling experience, "
|
||||
+ " you are expected to have a deep understanding of data analysis and data modeling concepts."
|
||||
+ "\nJob: You will be given a database table structure, which includes the database table name, field name,"
|
||||
+ " field type, and field comments. Your task is to utilize this information for data modeling."
|
||||
+ "\nTask:"
|
||||
+ "\n1. Generate a name and description for the model. Please note, 'bizName' refers to the English name, while 'name' is the Chinese name."
|
||||
+ "\n2. Create a Chinese name for the field and categorize the field into one of the following five types:"
|
||||
+ "\n primary_key: This is a unique identifier for a record row in a database."
|
||||
+ "\n foreign_key: This is a key in a database whose value is derived from the primary key of another table."
|
||||
+ "\n data_time: This represents the time when data is generated in the data warehouse."
|
||||
+ "\n dimension: Usually a string type, used for grouping and filtering data. No need to generate aggregate functions"
|
||||
+ "\n measure: Usually a numeric type, used to quantify data from a certain evaluative perspective. "
|
||||
+ " Also, you need to generate aggregate functions(Eg: MAX, MIN, AVG, SUM, COUNT) for the measure type. "
|
||||
+ "\nTip: I will also give you other related dbSchemas. If you determine that different dbSchemas have the same fields, "
|
||||
+ " they can be primary and foreign key relationships."
|
||||
+ "\nDBSchema: {{DBSchema}}" + "\nOtherRelatedDBSchema: {{otherRelatedDBSchema}}"
|
||||
+ "\nExemplar: {{exemplar}}";
|
||||
|
||||
private final ObjectMapper objectMapper = JsonUtil.INSTANCE.getObjectMapper();
|
||||
|
||||
public LLMSemanticModeller() {
|
||||
ChatAppManager.register(APP_KEY, ChatApp.builder().prompt(INSTRUCTION).name("构造数据语义模型")
|
||||
.appModule(AppModule.HEADLESS).description("通过大模型来构造数据语义模型").enable(true).build());
|
||||
}
|
||||
|
||||
interface ModelSchemaExtractor {
|
||||
ModelSchema generateModelSchema(String text);
|
||||
}
|
||||
|
||||
@Override
|
||||
public ModelSchema build(DbSchema dbSchema, List<DbSchema> dbSchemas,
|
||||
ModelBuildReq modelBuildReq) {
|
||||
Optional<ChatApp> chatApp = ChatAppManager.getApp(APP_KEY);
|
||||
if (!chatApp.isPresent() || !chatApp.get().isEnable()) {
|
||||
return null;
|
||||
}
|
||||
List<DbSchema> otherDbSchema = getOtherDbSchema(dbSchema, dbSchemas);
|
||||
ModelSchemaExtractor extractor =
|
||||
AiServices.create(ModelSchemaExtractor.class, getChatModel(modelBuildReq));
|
||||
Prompt prompt = generatePrompt(dbSchema, otherDbSchema, chatApp.get());
|
||||
ModelSchema modelSchema =
|
||||
extractor.generateModelSchema(prompt.toUserMessage().singleText());
|
||||
log.info("dbSchema: {}\n otherRelatedDBSchema:{}\n modelSchema: {}",
|
||||
JsonUtil.toString(dbSchema), JsonUtil.toString(otherDbSchema),
|
||||
JsonUtil.toString(modelSchema));
|
||||
return modelSchema;
|
||||
}
|
||||
|
||||
private List<DbSchema> getOtherDbSchema(DbSchema curSchema, List<DbSchema> dbSchemas) {
|
||||
return dbSchemas.stream()
|
||||
.filter(dbSchema -> !dbSchema.getTable().equals(curSchema.getTable()))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
private ChatLanguageModel getChatModel(ModelBuildReq modelBuildReq) {
|
||||
ChatModelConfig chatModelConfig = modelBuildReq.getChatModelConfig();
|
||||
if (chatModelConfig == null) {
|
||||
ChatModelService chatModelService = ContextUtils.getBean(ChatModelService.class);
|
||||
ChatModel chatModel = chatModelService.getChatModel(modelBuildReq.getChatModelId());
|
||||
chatModelConfig = chatModel.getConfig();
|
||||
}
|
||||
return ModelProvider.getChatModel(chatModelConfig);
|
||||
}
|
||||
|
||||
private Prompt generatePrompt(DbSchema dbSchema, List<DbSchema> otherDbSchema,
|
||||
ChatApp chatApp) {
|
||||
Map<String, Object> variable = new HashMap<>();
|
||||
variable.put("exemplar", loadExemplars());
|
||||
variable.put("DBSchema", JsonUtil.toString(dbSchema));
|
||||
variable.put("otherRelatedDBSchema", JsonUtil.toString(otherDbSchema));
|
||||
return PromptTemplate.from(chatApp.getPrompt()).apply(variable);
|
||||
}
|
||||
|
||||
private String loadExemplars() {
|
||||
Environment environment = ContextUtils.getBean(Environment.class);
|
||||
String enableExemplarLoading =
|
||||
environment.getProperty("s2.model.building.exemplars.enabled");
|
||||
if (Boolean.TRUE.equals(Boolean.parseBoolean(enableExemplarLoading))) {
|
||||
log.info("Not enable load model-building exemplars");
|
||||
return "";
|
||||
}
|
||||
try {
|
||||
ClassPathResource resource = new ClassPathResource(SYS_EXEMPLAR_FILE);
|
||||
if (resource.exists()) {
|
||||
InputStream inputStream = resource.getInputStream();
|
||||
return objectMapper
|
||||
.writeValueAsString(objectMapper.readValue(inputStream, Object.class));
|
||||
}
|
||||
} catch (Exception e) {
|
||||
log.error("Failed to load model-building system exemplars", e);
|
||||
}
|
||||
return "";
|
||||
}
|
||||
|
||||
}
|
||||
@@ -8,6 +8,7 @@ import com.tencent.supersonic.headless.chat.knowledge.DictWord;
|
||||
import com.tencent.supersonic.headless.chat.knowledge.KnowledgeBaseService;
|
||||
import com.tencent.supersonic.headless.chat.knowledge.builder.WordBuilderFactory;
|
||||
import com.tencent.supersonic.headless.server.service.SchemaService;
|
||||
import lombok.Data;
|
||||
import lombok.extern.slf4j.Slf4j;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.stereotype.Service;
|
||||
@@ -21,6 +22,7 @@ import java.util.stream.Collectors;
|
||||
|
||||
@Service
|
||||
@Slf4j
|
||||
@Data
|
||||
public class DictWordService {
|
||||
|
||||
@Autowired
|
||||
@@ -80,14 +82,6 @@ public class DictWordService {
|
||||
natures.addAll(natureList);
|
||||
}
|
||||
|
||||
public List<DictWord> getPreDictWords() {
|
||||
return preDictWords;
|
||||
}
|
||||
|
||||
public void setPreDictWords(List<DictWord> preDictWords) {
|
||||
this.preDictWords = preDictWords;
|
||||
}
|
||||
|
||||
private List<SchemaElement> distinct(List<SchemaElement> metas) {
|
||||
if (CollectionUtils.isEmpty(metas)) {
|
||||
return metas;
|
||||
|
||||
Reference in New Issue
Block a user