Sakura: Compatible with new SakuraAPI. (#492)

* Feat(Sakura): Compatible with new SakuraAPI by using openai package; Update info.

* Fix bugs and change degen detecting method.

* Fix bug.

* Sakura: add context num option, fix bugs.

* Sakura: Fix bugs.

* restore

* Sakura: restore using requests to interact with API.

LunaTranslator/LunaTranslator/translator/sakura.py

@@ -1,165 +1,151 @@
 from traceback import print_exc
 from translator.basetranslator import basetrans
 import requests
+# OpenAI
+# from openai import OpenAI
 
 class TS(basetrans):
     def langmap(self):
         return {"zh": "zh-CN"}
     def __init__(self, typename) :
-        self.api_type = ""
         self.api_url = ""
-        self.model_type = ""
+        self.history = {
-        self.history = []
+            "ja": [],
+            "zh": []
+        }
         self.session = requests.Session()
         super( ).__init__(typename)
-    def sliding_window(self, query):
+    def sliding_window(self, text_ja, text_zh):
-        self.history.append(query)
+        if text_ja == "" or text_zh == "":
-        if len(self.history) > 4:
+            return
-            del self.history[0]
+        self.history['ja'].append(text_ja)
-        return self.history
+        self.history['zh'].append(text_zh)
-    def list_to_prompt(self, query_list):
+        if len(self.history['ja']) > int(self.config['附带上下文个数（必须打开利用上文翻译）']) + 1:
+            del self.history['ja'][0]
+            del self.history['zh'][0]
+    def get_history(self, key):
         prompt = ""
-        for q in query_list:
+        for q in self.history[key]:
             prompt += q + "\n"
         prompt = prompt.strip()
         return prompt
-    def make_prompt(self, context):
+    def get_client(self, api_url):
-        if self.model_type == "baichuan":
+        if api_url[-4:] == "/v1/":
-            prompt = f"<reserved_106>将下面的日文文本翻译成中文：{context}<reserved_107>"
+            api_url = api_url[:-1]
-        elif self.model_type == "qwen":
+        elif api_url[-3:] == "/v1":
-            prompt = f"<|im_start|>system\n你是一个轻小说翻译模型，可以流畅通顺地以日本轻小说的风格将日文翻译成简体中文，并联系上下文正确使用人称代词，不擅自添加原文中没有的代词。<|im_end|>\n<|im_start|>user\n将下面的日文文本翻译成中文：{context}<|im_end|>\n<|im_start|>assistant\n"
+            pass
+        elif api_url[-1] == '/':
+            api_url += "v1"
         else:
-            prompt = f"<reserved_106>将下面的日文文本翻译成中文：{context}<reserved_107>"
+            api_url += "/v1"
-            
+        self.api_url = api_url
-        return prompt
+        # OpenAI
-        
+        # self.client = OpenAI(api_key="114514", base_url=api_url)
-    def make_request(self, prompt, is_test=False):
+    def make_messages(self, query, history_ja=None, history_zh=None, **kwargs):
-        if self.api_type == "llama.cpp":
+        messages = [
-            request = {
+            {
-                "prompt": prompt,
+                "role": "system",
-                "n_predict": 1 if is_test else int(self.config['max_new_token']),
+                "content": "你是一个轻小说翻译模型，可以流畅通顺地以日本轻小说的风格将日文翻译成简体中文，并联系上下文正确使用人称代词，不擅自添加原文中没有的代词。"
-                "temperature": float(self.config['temperature']),
-                "top_p": float(self.config['top_p']),
-                "repeat_penalty": float(self.config['repetition_penalty']),
-                "frequency_penalty": float(self.config['frequency_penalty']),
-                "top_k": 40,
-                "seed": -1
             }
-            return request
+        ]
-        elif self.api_type == "dev_server" or is_test:
+        if history_ja:
-            request = {
+            messages.append({
-                "prompt": prompt,
+                "role": "user",
-                "max_new_tokens": 1 if is_test else int(self.config['max_new_token']),
+                "content": f"将下面的日文文本翻译成中文：{history_ja}"
-                "do_sample": bool(self.config['do_sample']),
+            })
-                "temperature": float(self.config['temperature']),
+        if history_zh:
-                "top_p": float(self.config['top_p']),
+            messages.append({
-                "repetition_penalty": float(self.config['repetition_penalty']),
+                "role": "assistant",
-                "num_beams": int(self.config['num_beams']),
+                "content": history_zh
-                "frequency_penalty": float(self.config['frequency_penalty']),
+            })
-                "top_k": 40,
+
-                "seed": -1
+        messages.append(
+            {
+                "role": "user",
+                "content": f"将下面的日文文本翻译成中文：{query}"
             }
-            return request
+        )
-        elif self.api_type == "openai_like":
+        return messages
-            raise NotImplementedError(f"1: {self.api_type}")
+
-        else:
+
-            raise NotImplementedError(f"2: {self.api_type}")
+    def send_request(self, query, is_test=False, **kwargs):
-    def parse_output(self, output: str, length):
+        extra_query = {
-        output = output.strip()
+            'do_sample': bool(self.config['do_sample']),
-        output_list = output.split("\n")
+            'num_beams': int(self.config['num_beams']),
-        if len(output_list) != length:
+            'repetition_penalty': float(self.config['repetition_penalty']),
-            # fallback to no history translation
+        }
-            return None
+        messages = self.make_messages(query, **kwargs)
-        else:
-            return output_list[-1]
-    def do_post(self, request):
         try:
-            response = self.session.post(self.api_url, json=request).json()
+            # OpenAI
-            if self.api_type == "dev_server":
+            # output = self.client.chat.completions.create(
-                output = response['results'][0]['text']
+            data = dict(
-                new_token = response['results'][0]['new_token']
+                model="sukinishiro",
-            elif self.api_type == "llama.cpp":
+                messages=messages,
-                output =  response['content']
+                temperature=float(self.config['temperature']),
-                new_token = response['tokens_predicted']
+                top_p=float(self.config['top_p']),
-            else:
+                max_tokens= 1 if is_test else int(self.config['max_new_token']),
-                raise NotImplementedError("3")
+                frequency_penalty=float(kwargs['frequency_penalty']) if "frequency_penalty" in kwargs.keys() else float(self.config['frequency_penalty']),
+                seed=-1,
+                extra_query=extra_query,
+                stream=False,
+            )
+            output = self.session.post(self.api_url + "/chat/completions", json=data).json()
         except Exception as e:
-            raise Exception(str(e) + f"\napi_type: '{self.api_type}', api_url: '{self.api_url}', model_type: '{self.model_type}'\n与API接口通信失败，请检查设置的API服务器监听地址是否正确，或检查API服务器是否正常开启。")
+            raise ValueError(f"无法连接到Sakura API：{self.api_url}，请检查你的API链接是否正确填写，以及API后端是否成功启动。")
-        return output, new_token, response
+        return output
-    def set_model_type(self):
+
-        #TODO: get model type from api
-        self.model_type = "NotImplemented"
-        request = self.make_request("test", is_test=True)
-        _, _, response = self.do_post(request)
-        if self.api_type == "llama.cpp":
-            model_name: str = response['model']
-            model_version = model_name.split("-")[-2]
-            if "0.8" in model_version:
-                self.model_type = "baichuan"
-            elif "0.9" in model_version:
-                self.model_type = "qwen"
-        return
-    def set_api_type(self):
-        endpoint = self.config['API接口地址']
-        if endpoint[-1] != "/":
-            endpoint += "/"
-        api_url = endpoint + "api/v1/generate"
-        test_json = self.make_request("test", is_test=True)
-        try:
-            response = self.session.post(api_url, json=test_json)
-        except Exception as e:
-            raise Exception(str(e) + f"\napi_type: '{self.api_type}', api_url: '{self.api_url}', model_type: '{self.model_type}'\n与API接口通信失败，请检查设置的API服务器监听地址是否正确，或检查API服务器是否正常开启。")
-        try:
-            response = response.json()
-            output = response['results'][0]['text']
-            new_token = response['results'][0]['new_token']
-            self.api_type = "dev_server"
-            self.api_url = api_url
-        except:
-            self.api_type = "llama.cpp"
-            self.api_url = endpoint + "completion"
-        
-        return
     def translate(self, query):
         self.checkempty(['API接口地址'])
-        if self.api_type == "":
+        if self.api_url == "":
-            self.set_api_type()
+            self.get_client(self.config['API接口地址'])
-            self.set_model_type()
+        frequency_penalty = float(self.config['frequency_penalty'])
+        if not bool(self.config['利用上文信息翻译（通常会有一定的效果提升，但会导致变慢）']):
+            output = self.send_request(query)
+            completion_tokens = output["usage"]["completion_tokens"]
+            output_text = output["choices"][0]["message"]["content"]
 
-        prompt = self.make_prompt(query)
-        request = self.make_request(prompt)
-
-        if not self.config['利用上文信息翻译（通常会有一定的效果提升，但会导致变慢）']:
-            output, new_token, _ = self.do_post(request)
-            
             if bool(self.config['fix_degeneration']):
                 cnt = 0
-                while new_token == self.config['max_new_token']:
+                while completion_tokens == int(self.config['max_new_token']):
                     # detect degeneration, fixing
-                    request['frequency_penalty'] += 0.1
+                    frequency_penalty += 0.1
-                    output, new_token, _ = self.do_post(request)
+                    output = self.send_request(query, frequency_penalty=frequency_penalty)
-                    
+                    completion_tokens = output["usage"]["completion_tokens"]
+                    output_text = output["choices"][0]["message"]["content"]
                     cnt += 1
                     if cnt == 2:
                         break
         else:
-            query_list = self.sliding_window(query)
+            # 实验性功能，测试效果后决定是否加入。
-            request['prompt'] = self.make_prompt(query)
+            # fallback = False
-            output, new_token, _ = self.do_post(request)
+            # if self.config['启用日文上下文模式']:
-            
+            #     history_prompt = self.get_history('ja')
+            #     output = self.send_request(history_prompt + "\n" + query)
+            #     completion_tokens = output.usage.completion_tokens
+            #     output_text = output.choices[0].message.content
+
+            #     if len(output_text.split("\n")) == len(history_prompt.split("\n")) + 1:
+            #         output_text = output_text.split("\n")[-1]
+            #     else:
+            #         fallback = True
+            # 如果日文上下文模式失败，则fallback到中文上下文模式。
+            # if fallback or not self.config['启用日文上下文模式']:
+
+            history_prompt = self.get_history('zh')
+            output = self.send_request(query, history_zh=history_prompt)
+            completion_tokens = output["usage"]["completion_tokens"]
+            output_text = output["choices"][0]["message"]["content"]
+
             if bool(self.config['fix_degeneration']):
                 cnt = 0
-                while new_token == self.config['max_new_token']:
+                while completion_tokens == int(self.config['max_new_token']):
-                    # detect degeneration, fixing
+                    frequency_penalty += 0.1
-                    request['frequency_penalty'] += 0.1
+                    output = self.send_request(query, history_zh=history_prompt, frequency_penalty=frequency_penalty)
-                    output, new_token, _ = self.do_post(request)
+                    completion_tokens = output["usage"]["completion_tokens"]
-                    
+                    output_text = output["choices"][0]["message"]["content"]
                     cnt += 1
-                    if cnt == 2:
+                    if cnt == 3:
+                        output_text = "Error：模型无法完整输出或退化无法解决，请调大设置中的max_new_token！！！原输出：" + output_text
                         break
-                    
+            self.sliding_window(query, output_text)
-            output = self.parse_output(output, len(query_list))
+        return output_text
-            if not output:
-                request['prompt'] = self.make_prompt(query)
-                output, new_token, _ = self.do_post(request)
-        return output

LunaTranslator/files/defaultconfig/translatorsetting.json

@@ -463,10 +463,11 @@
     },
 	"sakura": {
         "args": {
-			"Sakura部署教程": "https://sakura.srpr.moe",
+			"Sakura部署教程": "https://github.com/SakuraLLM/Sakura-13B-Galgame/wiki",
 			"Github仓库": "https://github.com/SakuraLLM/Sakura-13B-Galgame",
-            "API接口地址": "http://127.0.0.1:5000/",
+            "API接口地址": "http://127.0.0.1:8080/",
             "利用上文信息翻译（通常会有一定的效果提升，但会导致变慢）": false,
+			"附带上下文个数（必须打开利用上文翻译）": 3,
             "temperature": 0.1,
 			"top_p": 0.3,
 			"num_beams": 1,
@@ -492,6 +493,12 @@
 			"利用上文信息翻译（通常会有一定的效果提升，但会导致变慢）": {
 				"type": "switch"
 			},
+			"附带上下文个数（必须打开利用上文翻译）":{
+				"type":"intspin",
+				"min":1,
+				"max":32,
+				"step":1
+			},
 			"temperature":{
 				"type":"spin",
 				"min":0,