Improve gpt performance

Author: Zipper-1

mathtranslate/chatgpt.py

@@ -5,6 +5,7 @@
 import openai
 import sys
 import time
+import re
 
 class GPTTranslator:
     def __init__(self):
@@ -15,23 +16,43 @@ def __init__(self):
 
 
     def format_prompt(self, text, language_to, language_from):
-        PROMPT_PROTOTYPE = "As an academic expert with specialized knowledge in various fields, please provide a proficient and precise translation translation from {} to {} of the academic text enclosed in 🔤. It is crucial to maintaining the original phrase or sentence and ensure accuracy while utilizing the appropriate language. Please provide only the translated result without any additional explanation and remove 🔤. The text is as follows:  🔤 {} 🔤  "
+        PROMPT_PROTOTYPE = 'As an academic expert with specialized knowledge in various fields, please provide a proficient and precise translation translation from {} to {} of the academic text enclosed in 🔤. It is crucial to maintaining the original phrase or sentence and ensure accuracy while utilizing the appropriate language. Please provide only the translated result without any additional explanation and remove 🔤. Do not modify or delete any word contains "/XMATHX_" such as /XMATHX_0, /XMATHX_1, /XMATHX_3_4. The text is as follows: 🔤 {} 🔤  '
         #prompt prototype changed from https://github.com/windingwind/zotero-pdf-translate
-        return PROMPT_PROTOTYPE.format(language_from,language_to,text)
+        SYSTEM_PROMPT_PROTOTYPE = 'You are an academic translator with specialized knowledge in various fields, please provide a proficient and precise translation translation from {} to {} of the academic text enclosed in 🔤.Do not modify or delete any word contains "/XMATHX_" such as /XMATHX_0, /XMATHX_1, /XMATHX_3_4.'
+        return {'system':SYSTEM_PROMPT_PROTOTYPE.format(language_from,language_to),'user':PROMPT_PROTOTYPE.format(language_from,language_to,text)}
 
     def get_server_errormsg(self,error):
         try :
             return error.response.json()['error']['message']
         except Exception :
             return error.message
 
+    def find_all_mathmask(self,text):
+        mask_pattern=re.compile(r'/XMATHX(_[0-9])+')
+        masks = set([i.group() for i in re.finditer(pattern=mask_pattern,string=text)])
+        return masks
+    
+    def is_gpt_output_valid(self,masks,text_translated):
+        masks_translated = self.find_all_mathmask(text_translated)
+        return (masks_translated==masks)
+
+    def is_text_all_mask(self,masks,text):
+        for mask in masks:
+            text = text.replace(mask,'')
+        return text.isspace()
+
+
     def call_openai_api(self,prompt):
         messages= [{
+            "role":"system",
+            "content": prompt['system']
+        },
+            {
 		        "role": "user",
-		        "content": prompt
+		        "content": prompt['user']
 	        }]
         try:
-            return self.client.chat.completions.create(model=self.model,temperature=0.8,messages=messages)
+            return self.client.chat.completions.create(model=self.model,temperature=1,messages=messages)
         except openai.RateLimitError as e:
             print('API rate limit exceeded, retry after 15s')
             time.sleep(15)
@@ -50,10 +71,13 @@ def call_openai_api(self,prompt):
 
 
     def translate(self, text, language_to, language_from):
+        masks = self.find_all_mathmask(text)
+        if self.is_text_all_mask(masks,text):
+            return text
         while True:
             result = self.call_openai_api(self.format_prompt(text, language_to, language_from))
-            content_translated = result.choices[0].message.content
-            if '🔤' not in content_translated:
+            content_translated = result.choices[0].message.content.replace('🔤','')
+            if self.is_gpt_output_valid(masks,content_translated):
                 return content_translated
 
 

mathtranslate/config.py

@@ -29,7 +29,7 @@ class Config:
     openai_api_endpoint_default = 'https://api.openai.com'
     openai_api_key_default = None
 
-    math_code = 'XMATHX'
+    math_code = '/XMATHX' #better for gpt to understand
     log_file = f'{app_dir}/translate_log'
     raw_mularg_command_list = [('textcolor', 2, (1, ))]
     mularg_command_list = [('textcolor', 2, (1, ))]

mathtranslate/utils.py

@@ -116,7 +116,7 @@ def process_options(options):
         config.set_variable(config.openai_api_endpoint_path, config.openai_api_endpoint_default)
         print('OpenAI api key (something like sk-xxx...):')
         config.set_variable(config.openai_api_key_path, config.openai_api_key_default)
-        print('ChatGPT model name: (leave empty for default {}'.format(config.openai_model_name_default))
+        print('ChatGPT model name: (leave empty for default {})'.format(config.openai_model_name_default))
         config.set_variable(config.openai_model_name_path,config.openai_model_name_default)
         print('saved!')
         config.load()
@@ -169,8 +169,6 @@ def process_options(options):
             print('Please setup api info for openAI api first by')
             print('translate_tex --setgpt')
             sys.exit()
-        options.threads = 1
-        print('disable mult-threading for chatGPT api')
 
     if options.threads < 0:
         print('threads must be a non-zero integer number (>=0 where 0 means auto), set to auto')