mirror of
https://gitee.com/270580156/weiyu.git
synced 2026-09-09 02:59:28 +00:00
update
This commit is contained in:
jack ning
@@ -2,7 +2,7 @@
|
||||
* @Author: jackning 270580156@qq.com
|
||||
* @Date: 2024-06-05 15:39:22
|
||||
* @LastEditors: jackning 270580156@qq.com
|
||||
* @LastEditTime: 2025-01-02 11:56:26
|
||||
* @LastEditTime: 2025-01-02 12:15:20
|
||||
* @Description: bytedesk.com https://github.com/Bytedesk/bytedesk
|
||||
* Please be aware of the BSL license restrictions before installing Bytedesk IM –
|
||||
* selling, reselling, or hosting Bytedesk IM as a service is a breach of the terms and automatically terminates your rights under the license.
|
||||
@@ -96,25 +96,192 @@ public class ZhipuaiService {
|
||||
|
||||
// RAG智能客服提示模板
|
||||
private final String PROMPT_TEMPLATE = """
|
||||
任务描述：根据用户的查询和文档信息回答问题，并结合历史聊天记录生成简要的回答。
|
||||
任务描述：根据用户的查询和文档信息回答问题，并结合历史聊天记录生成简要的回答。
|
||||
|
||||
用户查询: {query}
|
||||
用户查询: {query}
|
||||
|
||||
历史聊天记录: {history}
|
||||
|
||||
搜索结果: {context}
|
||||
|
||||
请根据以上信息生成一个简单明了的回答，确保信息准确且易于理解。
|
||||
当用户提出的问题无法根据文档内容进行回复或者你也不清楚时，回复:未查找到相关问题答案.
|
||||
另外，请提供更多相关的问答对。
|
||||
回答内容请以JSON格式输出，格式如下：
|
||||
{
|
||||
"answer": "回答内容",
|
||||
"additional_qa_pairs": [
|
||||
{"question": "相关问题1", "answer": "相关答案1"},
|
||||
{"question": "相关问题2", "answer": "相关答案2"}
|
||||
]
|
||||
}
|
||||
""";
|
||||
|
||||
// 知识库问答
|
||||
public void sendWsKbMessage(String query, String kbUid, RobotEntity robot, MessageProtobuf messageProtobuf) {
|
||||
//
|
||||
String prompt = robot.getLlm().getPrompt();
|
||||
if (robot.getType().equals(RobotTypeEnum.SERVICE.name())) {
|
||||
List<String> contentList = uploadVectorStore.searchText(query, kbUid);
|
||||
String context = String.join("\n", contentList);
|
||||
String history = ""; // TODO: 历史对话上下文，此处暂不使用
|
||||
prompt = PROMPT_TEMPLATE.replace("{context}", context).replace("{query}", query).replace("{history}",
|
||||
history);
|
||||
log.info("sendWsRobotMessage prompt 1 {}", prompt);
|
||||
} else {
|
||||
prompt = prompt + "\n" + query;
|
||||
log.info("sendWsRobotMessage prompt 2 {}", prompt);
|
||||
}
|
||||
//
|
||||
List<ChatMessage> messages = new ArrayList<>();
|
||||
ChatMessage chatMessage = new ChatMessage(ChatMessageRole.USER.value(), prompt);
|
||||
messages.add(chatMessage);
|
||||
//
|
||||
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
|
||||
// 模型名称
|
||||
.model(zhipuaiConfig.zhiPuAiApiModel)
|
||||
// .model(Constants.ModelChatGLM3TURBO)
|
||||
// .model(robotSimple.getLlm().getModel())
|
||||
// .temperature(robotSimple.getLlm().getTemperature())
|
||||
// .topP(robotSimple.getLlm().getTopP())
|
||||
.stream(Boolean.TRUE)
|
||||
.messages(messages)
|
||||
.requestId(messageProtobuf.getUid())
|
||||
.build();
|
||||
//
|
||||
ModelApiResponse sseModelApiResp = client.invokeModelApi(chatCompletionRequest);
|
||||
if (sseModelApiResp.isSuccess()) {
|
||||
AtomicBoolean isFirst = new AtomicBoolean(true);
|
||||
List<Choice> choices = new ArrayList<>();
|
||||
ChatMessageAccumulator chatMessageAccumulator = mapStreamToAccumulator(sseModelApiResp.getFlowable())
|
||||
.doOnNext(accumulator -> {
|
||||
{
|
||||
if (isFirst.getAndSet(false)) {
|
||||
log.info("answer start: ");
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getTool_calls() != null) {
|
||||
String jsonString = JSON.toJSONString(accumulator.getDelta().getTool_calls());
|
||||
log.info("tool_calls: " + jsonString);
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getContent() != null) {
|
||||
String answerContent = accumulator.getDelta().getContent();
|
||||
log.info("answerContent {}", answerContent);
|
||||
if (StringUtils.hasText(answerContent)) {
|
||||
messageProtobuf.setType(MessageTypeEnum.STREAM);
|
||||
messageProtobuf.setContent(answerContent);
|
||||
//
|
||||
// MessageUtils.notifyUser(messageProtobuf);
|
||||
messageSendService.sendProtobufMessage(messageProtobuf);
|
||||
}
|
||||
}
|
||||
}
|
||||
})
|
||||
.doOnComplete(() -> {
|
||||
log.info("answer end");
|
||||
})
|
||||
.lastElement()
|
||||
.blockingGet();
|
||||
|
||||
ModelData data = new ModelData();
|
||||
data.setChoices(choices);
|
||||
data.setUsage(chatMessageAccumulator.getUsage());
|
||||
data.setId(chatMessageAccumulator.getId());
|
||||
data.setCreated(chatMessageAccumulator.getCreated());
|
||||
data.setRequestId(chatCompletionRequest.getRequestId());
|
||||
sseModelApiResp.setFlowable(null);// 打印前置空
|
||||
sseModelApiResp.setData(data);
|
||||
// 存储到数据库
|
||||
// robotMessage.setPromptTokens(chatMessageAccumulator.getUsage().getPromptTokens());
|
||||
// robotMessage.setCompletionTokens(chatMessageAccumulator.getUsage().getCompletionTokens());
|
||||
// robotMessage.setTotalTokens(chatMessageAccumulator.getUsage().getTotalTokens());
|
||||
// 后处理步骤：生成更多的问答对
|
||||
// List<String> additionalQAPairs = generateAdditionalQAPairs(question,
|
||||
// robotMessage.getAnswer());
|
||||
// additionalQAPairs.forEach(qaPair -> {
|
||||
// log.info("Additional QA Pair: {}", qaPair);
|
||||
// // 这里可以将更多的问答对存储到数据库或发送给客户端
|
||||
// });
|
||||
}
|
||||
|
||||
String result = JSON.toJSONString(sseModelApiResp);
|
||||
log.info("websocket output:" + result);
|
||||
}
|
||||
|
||||
public void sendWsKbAutoReply(String query, String kbUid, MessageProtobuf messageProtobuf) {
|
||||
//
|
||||
List<String> contentList = uploadVectorStore.searchText(query, kbUid);
|
||||
String context = String.join("\n", contentList);
|
||||
String prompt = PROMPT_BLUEPRINT.replace("{context}", context).replace("{query}", query);
|
||||
log.info("sendWsAutoReply prompt {}", prompt);
|
||||
//
|
||||
List<ChatMessage> messages = new ArrayList<>();
|
||||
ChatMessage chatMessage = new ChatMessage(ChatMessageRole.USER.value(), prompt);
|
||||
messages.add(chatMessage);
|
||||
//
|
||||
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
|
||||
// 模型名称
|
||||
.model(zhipuaiConfig.zhiPuAiApiModel)
|
||||
// .model(Constants.ModelChatGLM3TURBO)
|
||||
// .model(robotSimple.getLlm().getModel())
|
||||
// .temperature(robotSimple.getLlm().getTemperature())
|
||||
// .topP(robotSimple.getLlm().getTopP())
|
||||
.stream(Boolean.TRUE)
|
||||
.messages(messages)
|
||||
.requestId(messageProtobuf.getUid())
|
||||
.build();
|
||||
//
|
||||
ModelApiResponse sseModelApiResp = client.invokeModelApi(chatCompletionRequest);
|
||||
if (sseModelApiResp.isSuccess()) {
|
||||
AtomicBoolean isFirst = new AtomicBoolean(true);
|
||||
List<Choice> choices = new ArrayList<>();
|
||||
ChatMessageAccumulator chatMessageAccumulator = mapStreamToAccumulator(sseModelApiResp.getFlowable())
|
||||
.doOnNext(accumulator -> {
|
||||
{
|
||||
if (isFirst.getAndSet(false)) {
|
||||
log.info("answer start: ");
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getTool_calls() != null) {
|
||||
String jsonString = JSON.toJSONString(accumulator.getDelta().getTool_calls());
|
||||
log.info("tool_calls: " + jsonString);
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getContent() != null) {
|
||||
String answerContent = accumulator.getDelta().getContent();
|
||||
log.info("answerContent {}", answerContent);
|
||||
if (StringUtils.hasText(answerContent)) {
|
||||
messageProtobuf.setType(MessageTypeEnum.STREAM);
|
||||
messageProtobuf.setContent(answerContent);
|
||||
//
|
||||
// MessageUtils.notifyUser(messageProtobuf);
|
||||
messageSendService.sendProtobufMessage(messageProtobuf);
|
||||
}
|
||||
}
|
||||
}
|
||||
})
|
||||
.doOnComplete(() -> {
|
||||
log.info("answer end");
|
||||
})
|
||||
.lastElement()
|
||||
.blockingGet();
|
||||
|
||||
ModelData data = new ModelData();
|
||||
data.setChoices(choices);
|
||||
data.setUsage(chatMessageAccumulator.getUsage());
|
||||
data.setId(chatMessageAccumulator.getId());
|
||||
data.setCreated(chatMessageAccumulator.getCreated());
|
||||
data.setRequestId(chatCompletionRequest.getRequestId());
|
||||
sseModelApiResp.setFlowable(null);// 打印前置空
|
||||
sseModelApiResp.setData(data);
|
||||
// 存储到数据库
|
||||
// robotMessage.setPromptTokens(chatMessageAccumulator.getUsage().getPromptTokens());
|
||||
// robotMessage.setCompletionTokens(chatMessageAccumulator.getUsage().getCompletionTokens());
|
||||
// robotMessage.setTotalTokens(chatMessageAccumulator.getUsage().getTotalTokens());
|
||||
}
|
||||
|
||||
String result = JSON.toJSONString(sseModelApiResp);
|
||||
log.info("websocket output:" + result);
|
||||
|
||||
}
|
||||
|
||||
历史聊天记录: {history}
|
||||
|
||||
搜索结果: {context}
|
||||
|
||||
请根据以上信息生成一个简单明了的回答，确保信息准确且易于理解。
|
||||
当用户提出的问题无法根据文档内容进行回复或者你也不清楚时，回复:未查找到相关问题答案.
|
||||
另外，请提供更多相关的问答对，并以JSON格式输出，格式如下：
|
||||
{
|
||||
"answer": "回答内容",
|
||||
"additional_qa_pairs": [
|
||||
{"question": "相关问题1", "answer": "相关答案1"},
|
||||
{"question": "相关问题2", "answer": "相关答案2"}
|
||||
]
|
||||
}
|
||||
""";
|
||||
/**
|
||||
* sse调用
|
||||
*/
|
||||
@@ -315,169 +482,6 @@ public class ZhipuaiService {
|
||||
log.info("websocket output:" + result);
|
||||
}
|
||||
|
||||
// 知识库问答
|
||||
public void sendWsKbMessage(String query, String kbUid, RobotEntity robot, MessageProtobuf messageProtobuf) {
|
||||
//
|
||||
String prompt = robot.getLlm().getPrompt();
|
||||
if (robot.getType().equals(RobotTypeEnum.SERVICE.name())) {
|
||||
List<String> contentList = uploadVectorStore.searchText(query, kbUid);
|
||||
String context = String.join("\n", contentList);
|
||||
String history = ""; // TODO: 历史对话上下文，此处暂不使用
|
||||
prompt = PROMPT_TEMPLATE.replace("{context}", context).replace("{query}", query).replace("{history}", history);
|
||||
log.info("sendWsRobotMessage prompt 1 {}", prompt);
|
||||
} else {
|
||||
prompt = prompt + "\n" + query;
|
||||
log.info("sendWsRobotMessage prompt 2 {}", prompt);
|
||||
}
|
||||
//
|
||||
List<ChatMessage> messages = new ArrayList<>();
|
||||
ChatMessage chatMessage = new ChatMessage(ChatMessageRole.USER.value(), prompt);
|
||||
messages.add(chatMessage);
|
||||
//
|
||||
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
|
||||
// 模型名称
|
||||
.model(zhipuaiConfig.zhiPuAiApiModel)
|
||||
// .model(Constants.ModelChatGLM3TURBO)
|
||||
// .model(robotSimple.getLlm().getModel())
|
||||
// .temperature(robotSimple.getLlm().getTemperature())
|
||||
// .topP(robotSimple.getLlm().getTopP())
|
||||
.stream(Boolean.TRUE)
|
||||
.messages(messages)
|
||||
.requestId(messageProtobuf.getUid())
|
||||
.build();
|
||||
//
|
||||
ModelApiResponse sseModelApiResp = client.invokeModelApi(chatCompletionRequest);
|
||||
if (sseModelApiResp.isSuccess()) {
|
||||
AtomicBoolean isFirst = new AtomicBoolean(true);
|
||||
List<Choice> choices = new ArrayList<>();
|
||||
ChatMessageAccumulator chatMessageAccumulator = mapStreamToAccumulator(sseModelApiResp.getFlowable())
|
||||
.doOnNext(accumulator -> {
|
||||
{
|
||||
if (isFirst.getAndSet(false)) {
|
||||
log.info("answer start: ");
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getTool_calls() != null) {
|
||||
String jsonString = JSON.toJSONString(accumulator.getDelta().getTool_calls());
|
||||
log.info("tool_calls: " + jsonString);
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getContent() != null) {
|
||||
String answerContent = accumulator.getDelta().getContent();
|
||||
log.info("answerContent {}", answerContent);
|
||||
if (StringUtils.hasText(answerContent)) {
|
||||
messageProtobuf.setType(MessageTypeEnum.STREAM);
|
||||
messageProtobuf.setContent(answerContent);
|
||||
//
|
||||
// MessageUtils.notifyUser(messageProtobuf);
|
||||
messageSendService.sendProtobufMessage(messageProtobuf);
|
||||
}
|
||||
}
|
||||
}
|
||||
})
|
||||
.doOnComplete(() -> {
|
||||
log.info("answer end");
|
||||
})
|
||||
.lastElement()
|
||||
.blockingGet();
|
||||
|
||||
ModelData data = new ModelData();
|
||||
data.setChoices(choices);
|
||||
data.setUsage(chatMessageAccumulator.getUsage());
|
||||
data.setId(chatMessageAccumulator.getId());
|
||||
data.setCreated(chatMessageAccumulator.getCreated());
|
||||
data.setRequestId(chatCompletionRequest.getRequestId());
|
||||
sseModelApiResp.setFlowable(null);// 打印前置空
|
||||
sseModelApiResp.setData(data);
|
||||
// 存储到数据库
|
||||
// robotMessage.setPromptTokens(chatMessageAccumulator.getUsage().getPromptTokens());
|
||||
// robotMessage.setCompletionTokens(chatMessageAccumulator.getUsage().getCompletionTokens());
|
||||
// robotMessage.setTotalTokens(chatMessageAccumulator.getUsage().getTotalTokens());
|
||||
// 后处理步骤：生成更多的问答对
|
||||
// List<String> additionalQAPairs = generateAdditionalQAPairs(question, robotMessage.getAnswer());
|
||||
// additionalQAPairs.forEach(qaPair -> {
|
||||
// log.info("Additional QA Pair: {}", qaPair);
|
||||
// // 这里可以将更多的问答对存储到数据库或发送给客户端
|
||||
// });
|
||||
}
|
||||
|
||||
String result = JSON.toJSONString(sseModelApiResp);
|
||||
log.info("websocket output:" + result);
|
||||
}
|
||||
|
||||
public void sendWsKbAutoReply(String query, String kbUid, MessageProtobuf messageProtobuf) {
|
||||
//
|
||||
List<String> contentList = uploadVectorStore.searchText(query, kbUid);
|
||||
String context = String.join("\n", contentList);
|
||||
String prompt = PROMPT_BLUEPRINT.replace("{context}", context).replace("{query}", query);
|
||||
log.info("sendWsAutoReply prompt {}", prompt);
|
||||
//
|
||||
List<ChatMessage> messages = new ArrayList<>();
|
||||
ChatMessage chatMessage = new ChatMessage(ChatMessageRole.USER.value(), prompt);
|
||||
messages.add(chatMessage);
|
||||
//
|
||||
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
|
||||
// 模型名称
|
||||
.model(zhipuaiConfig.zhiPuAiApiModel)
|
||||
// .model(Constants.ModelChatGLM3TURBO)
|
||||
// .model(robotSimple.getLlm().getModel())
|
||||
// .temperature(robotSimple.getLlm().getTemperature())
|
||||
// .topP(robotSimple.getLlm().getTopP())
|
||||
.stream(Boolean.TRUE)
|
||||
.messages(messages)
|
||||
.requestId(messageProtobuf.getUid())
|
||||
.build();
|
||||
//
|
||||
ModelApiResponse sseModelApiResp = client.invokeModelApi(chatCompletionRequest);
|
||||
if (sseModelApiResp.isSuccess()) {
|
||||
AtomicBoolean isFirst = new AtomicBoolean(true);
|
||||
List<Choice> choices = new ArrayList<>();
|
||||
ChatMessageAccumulator chatMessageAccumulator = mapStreamToAccumulator(sseModelApiResp.getFlowable())
|
||||
.doOnNext(accumulator -> {
|
||||
{
|
||||
if (isFirst.getAndSet(false)) {
|
||||
log.info("answer start: ");
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getTool_calls() != null) {
|
||||
String jsonString = JSON.toJSONString(accumulator.getDelta().getTool_calls());
|
||||
log.info("tool_calls: " + jsonString);
|
||||
}
|
||||
if (accumulator.getDelta() != null && accumulator.getDelta().getContent() != null) {
|
||||
String answerContent = accumulator.getDelta().getContent();
|
||||
log.info("answerContent {}", answerContent);
|
||||
if (StringUtils.hasText(answerContent)) {
|
||||
messageProtobuf.setType(MessageTypeEnum.STREAM);
|
||||
messageProtobuf.setContent(answerContent);
|
||||
//
|
||||
// MessageUtils.notifyUser(messageProtobuf);
|
||||
messageSendService.sendProtobufMessage(messageProtobuf);
|
||||
}
|
||||
}
|
||||
}
|
||||
})
|
||||
.doOnComplete(() -> {
|
||||
log.info("answer end");
|
||||
})
|
||||
.lastElement()
|
||||
.blockingGet();
|
||||
|
||||
ModelData data = new ModelData();
|
||||
data.setChoices(choices);
|
||||
data.setUsage(chatMessageAccumulator.getUsage());
|
||||
data.setId(chatMessageAccumulator.getId());
|
||||
data.setCreated(chatMessageAccumulator.getCreated());
|
||||
data.setRequestId(chatCompletionRequest.getRequestId());
|
||||
sseModelApiResp.setFlowable(null);// 打印前置空
|
||||
sseModelApiResp.setData(data);
|
||||
// 存储到数据库
|
||||
// robotMessage.setPromptTokens(chatMessageAccumulator.getUsage().getPromptTokens());
|
||||
// robotMessage.setCompletionTokens(chatMessageAccumulator.getUsage().getCompletionTokens());
|
||||
// robotMessage.setTotalTokens(chatMessageAccumulator.getUsage().getTotalTokens());
|
||||
}
|
||||
|
||||
String result = JSON.toJSONString(sseModelApiResp);
|
||||
log.info("websocket output:" + result);
|
||||
|
||||
}
|
||||
|
||||
public static Flowable<ChatMessageAccumulator> mapStreamToAccumulator(Flowable<ModelData> flowable) {
|
||||
return flowable.map(chunk -> {
|
||||
return new ChatMessageAccumulator(chunk.getChoices().get(0).getDelta(), null, chunk.getChoices().get(0),
|
||||
@@ -485,13 +489,15 @@ public class ZhipuaiService {
|
||||
});
|
||||
}
|
||||
|
||||
private List<String> generateAdditionalQAPairs(String question, String answer) {
|
||||
// 这里可以实现生成更多问答对的逻辑
|
||||
// 例如，可以基于现有的问答对生成相关的问答对
|
||||
List<String> additionalQAPairs = new ArrayList<>();
|
||||
additionalQAPairs.add("Q: " + question + " A: " + answer);
|
||||
additionalQAPairs.add("Q: 相关问题1 A: 相关答案1");
|
||||
additionalQAPairs.add("Q: 相关问题2 A: 相关答案2");
|
||||
return additionalQAPairs;
|
||||
}
|
||||
// private List<String> generateAdditionalQAPairs(String question, String answer) {
|
||||
// // 这里可以实现生成更多问答对的逻辑
|
||||
// // 例如，可以基于现有的问答对生成相关的问答对
|
||||
// List<String> additionalQAPairs = new ArrayList<>();
|
||||
// additionalQAPairs.add("Q: " + question + " A: " + answer);
|
||||
// additionalQAPairs.add("Q: 相关问题1 A: 相关答案1");
|
||||
// additionalQAPairs.add("Q: 相关问题2 A: 相关答案2");
|
||||
// return additionalQAPairs;
|
||||
// }
|
||||
|
||||
|
||||
}
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
* @Author: jackning 270580156@qq.com
|
||||
* @Date: 2024-06-05 21:50:54
|
||||
* @LastEditors: jackning 270580156@qq.com
|
||||
* @LastEditTime: 2024-12-23 15:14:51
|
||||
* @LastEditTime: 2025-01-02 12:38:16
|
||||
* @Description: bytedesk.com https://github.com/Bytedesk/bytedesk
|
||||
* Please be aware of the BSL license restrictions before installing Bytedesk IM –
|
||||
* selling, reselling, or hosting Bytedesk IM as a service is a breach of the terms and automatically terminates your rights under the license.
|
||||
|
||||
Reference in New Issue
Block a user