'commit'

dsLightRag/ShiTi/T2_ImageToMd.py

@@ -29,5 +29,5 @@ completion = client.chat.completions.create(
 ocr_text = completion.choices[0].message.content
 
 # 二、调用格式化函数处理内容
-format_exam_content(client=client, raw_text=ocr_text, output_path="../output/数学OCR整理后的结果.md")
+format_exam_content(raw_text=ocr_text, output_path="../output/数学OCR整理后的结果.md")
 print("保存成功！")

dsLightRag/ShiTi/T3_DocxToMd.py

@@ -90,11 +90,7 @@ async def main():
             f.write(content)
 
     # 新增：使用大模型整理内容
-    format_exam_content(
+    format_exam_content(raw_text=content, output_path=os.path.join(output_dir, "物理Docx整理后的结果.md"))
-        client=llm_model_func(),
-        raw_text=content,
-        output_path=os.path.join(output_dir, "物理Docx整理后的结果.md")
-    )
     logger.info(f"内容整理完成，保存至: {os.path.join(output_dir, '物理Docx整理后的结果.md')}")
 
 

dsLightRag/Util/LightRagUtil.py

@@ -8,6 +8,8 @@ from lightrag import LightRAG
 from lightrag.kg.shared_storage import initialize_pipeline_status
 from lightrag.llm.openai import openai_complete_if_cache, openai_embed
 from lightrag.utils import EmbeddingFunc, logger, set_verbose_debug
+from openai import OpenAI
+
 from Config.Config import *
 
 
@@ -181,7 +183,11 @@ def create_vision_model_func(llm_model_func):
     return vision_model_func
 
 
-def format_exam_content(client, raw_text, output_path):
+def format_exam_content(raw_text, output_path):
+    client = OpenAI(
+        api_key=LLM_API_KEY,
+        base_url=LLM_BASE_URL,
+    )
     """
     将OCR识别的原始试卷内容格式化为标准试题格式