增加使用模型实现本地翻译

README.md

@@ -307,6 +307,17 @@ PyPDF2使用相对简单，但只支持英文，对中文支持不太友好；
 
 	    return ret
 
+#### 本地翻译 ####
+需要使用huggingface的opus-mt-en-zh模型进行翻译  
+也可以下载到本地，将文件夹放在
+src/TranslateTool/TranslateTool/model/opus-mt-en-zhs
+```
+--- src
+--- --- TranslateTool
+--- --- --- TranslateTool
+--- --- --- --- model
+```
+
 ### 写入文档 ###
 #### 写TXT文档 ####
 TXT文档的写比较简单，代码如下所示：

src/TranslateTool/TranslateTool/T_Docx.py

@@ -54,7 +54,8 @@ def translate(self):
             content = para.text.strip()    # 去掉多余空格
 
             if content != '':
-                ret = translate_func(content)
+                # ret = translate_func(content)
+                ret = translate_local(content)
                 trans = ret if ret else '翻译失败'
                 # 写入新文件
                 new_doc.add_paragraph(content)

src/TranslateTool/TranslateTool/T_Pdf_PyPDF2.py

@@ -56,7 +56,8 @@ def translate(self):
             for line in content_list:
                 line = line.strip()
                 if line:
-                    ret = translate_func(line)
+                    # ret = translate_func(line)
+                    ret = translate_local(line)
                     trans = ret if ret else '翻译失败'
                     self.write(line + '\n')
                     self.write(trans)

src/TranslateTool/TranslateTool/T_Pdf_pdfminer.py

@@ -90,7 +90,8 @@ def translate(self):
                         content = out.get_text().strip()
                         if content:
                             to_trans_content = content.replace("\r\n","")
-                            ret = translate_func(to_trans_content)
+                            # ret = translate_func(to_trans_content)
+                            ret = translate_local(to_trans_content)
                             trans = ret if ret else '翻译失败'
 
                             self.write(content)

src/TranslateTool/TranslateTool/T_Txt.py

@@ -53,7 +53,8 @@ def translate(self):
 
                 temp_line = temp_line.strip()
                 if temp_line:
-                    ret = translate_func(temp_line)
+                    # ret = translate_func(temp_line)
+                    ret = translate_local(temp_line)
                     trans = ret if ret else '翻译失败'
                     self.write(temp_line)
                     self.write(trans + '\n')
@@ -67,7 +68,8 @@ def translate(self):
             line = f.readline()
 
         if temp_line:  # 防止拼接之后，有一部分未执行翻译
-            ret = translate_func(temp_line)
+            # ret = translate_func(temp_line)
+            ret = translate_local(temp_line)
             trans = ret if ret else '翻译失败'
             self.write(temp_line)
             self.write(trans + '\n')
@@ -78,6 +80,7 @@ def translate(self):
         Logger().write(self.fileName + '翻译完成，新文档：' + self.new_path)
 
 
+
     def prepare(self):
         '''准备：生成的文件名和路径'''
 

src/TranslateTool/TranslateTool/TranslateFunc.py

@@ -24,7 +24,7 @@
 import requests   # pip intasll requests
 from Py4Js import *
 from Logger import *
-
+from translate import *
 
 # 百度翻译方法
 def baidu_translate(content,type=1):
@@ -187,13 +187,21 @@ def google_translate(content):
     else:
         return (False,ret)
 
+def translate_local(content):
+    '''本地模型翻译'''
+    # 模型下载到本地可实现离线操作
+
+    tokenizer,model = load_model()
+    ret = translate(content,tokenizer,model)
+    return ret
 
 def translate_func(content):
     '''集成百度、谷歌、有道多合一的翻译'''
 
-    funcs = [google_translate, youdao_translate]    # baidu_translate,google_translate,youdao_translate
+    funcs = [translate_local,google_translate, youdao_translate]    # baidu_translate,google_translate,youdao_translate
     count = 0
 
+
     # 循环调用百度、谷歌、有道API，其中如果谁调成功就返回，或者大于等于9次没有成功也返回。
     while True:
         for i in range(len(funcs)):
@@ -212,3 +220,6 @@ def translate_func(content):
                     return ''
                 else:
                     continue
+
+
+

src/TranslateTool/TranslateTool/translate.py

@@ -0,0 +1,48 @@
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+
+class Config():
+    path = "src/TranslateTool/TranslateTool/model/opus-mt-en-zh"
+
+
+def load_model():
+    config = Config()
+
+    tokenizer = AutoTokenizer.from_pretrained(config.path)
+
+    model = AutoModelForSeq2SeqLM.from_pretrained(config.path)
+
+    return tokenizer,model 
+
+
+def IN():
+    sentence=input("请输入:")
+    return sentence
+
+def translate(text,tokenizer,model):
+    input_ids = tokenizer.encode(text, return_tensors="pt")
+    outputs = model.generate(input_ids)
+    decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return decoded
+
+def run():
+    print("---load_model---")
+    tokenizer,model = load_model()
+    print("---英译中---")
+    print("输入exit结束")
+    while (True):
+        text = IN()
+        if text == 'exit':
+
+            break
+
+        decoded = translate(text,tokenizer,model)
+
+        print(decoded)
+
+
+if __name__ == '__main__': 
+    run()
+
+
+