Update app.py

app.py

@@ -12,14 +12,19 @@ import glob
 def delete_files_except(folder_path):
   delfile=None
   files = glob.glob(os.path.join(folder_path, '*'))
-  delfile=os.path.basename(files[0])
-  os.remove(folder_path+'/'+delfile)
+  for file in files:
+      if file.endswith('.pdf'):
+          delfile=os.path.basename(file)
+          os.remove(file)
+      else:
+          continue 
   return delfile
+    
 
 
 def MOP(path):
   embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-m3")
-  template_vectors = FAISS.load_local('/content/drive/MyDrive/Innovation Themes /MoP/Template_embeddings', embeddings, allow_dangerous_deserialization=True)
+  template_vectors = FAISS.load_local('vector_database', embeddings, allow_dangerous_deserialization=True)
   text_chunks1=8
   vectors_template=template_vectors
   docs=[]
@@ -29,10 +34,7 @@ def MOP(path):
   text_chunks2 = text_splitter.split_documents(docs)
   vector_Document = FAISS.from_documents(text_chunks2, embedding=embeddings)
   vectors_template.merge_from(vector_Document)
-  # filename_to_keep = 'Dummy_standard MoP_template_new.pdf'
-
   prompt_file = delete_files_except(path)
-   # prompt_file=delete_files_except(filename_to_keep,path)
   repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1"
   llm = HuggingFaceHub(repo_id=repo_id, model_kwargs={"temperature": 0.001, "max_new_tokens": 5000})
   retriever = vectors_template.as_retriever(search_type="similarity",search_kwargs={"k": text_chunks1+len(text_chunks2)})
@@ -40,7 +42,7 @@ def MOP(path):
 
 
   
-  message1= f"""<s> [INST] You have two documents:
+  message1= f """<s> [INST] You have two documents:
 
 Template Document: "Standard_MoP_template.pdf"
 Content Document: "{prompt_file}"
@@ -53,9 +55,9 @@ Assess whether the Content Document ("{prompt_file}") incorporates the necessary
 Provide a "Yes" if the Content Document effectively integrates most of the points listed under the "Pre-Check  Procedures", considering both explicit and implicit coverage.
 If the integration is lacking or incomplete, carefully reconsider whether the points are adequately addressed within the Content Document.
 If you provide a "Yes," provide a detailed explanation highlighting how the Content Document integrates the Pre-Check  Procedures effectively. Discuss both explicit mentions and any implicit coverage that contributes to their implementation.
-Make sure to accurately evaluate the alignment of the "Pre-Check  Procedures" provided in the Template Document with their presence or coverage in the Content Document. [/INST] </s>"""
+Make sure to accurately evaluate the alignment of the "Pre-Check  Procedures" provided in the Template Document with their presence or coverage in the Content Document. [/INST] </s> """
 
-  message2= f"""<s> [INST] You have two documents:
+  message2= f """<s> [INST] You have two documents:
 
 Template Document: "Standard_MoP_template.pdf"
 Content Document: "{prompt_file}"
@@ -68,7 +70,7 @@ Assess whether the Content Document ("{prompt_file}") incorporates the necessary
 Provide a "Yes" if the Content Document effectively integrates most of the points listed under the "Post-Check  Procedures", considering both explicit and implicit coverage.
 If the integration is lacking or incomplete, carefully reconsider whether the points are adequately addressed within the Content Document.
 If you provide a "Yes," provide a detailed explanation highlighting how the Content Document integrates the Post-Check  Procedures effectively. Discuss both explicit mentions and any implicit coverage that contributes to their implementation.
-Make sure to accurately evaluate the alignment of the "Post-Check  Procedures" provided in the Template Document with their presence or coverage in the Content Document. [/INST] </s>"""
+Make sure to accurately evaluate the alignment of the "Post-Check  Procedures" provided in the Template Document with their presence or coverage in the Content Document. [/INST] </s> """
 
   result1=qa.run(message1)
   result2=qa.run(message2)
@@ -96,7 +98,7 @@ Make sure to accurately evaluate the alignment of the "Post-Check  Procedures" p
   
 
 def process_file(fileobj):
-  destination_folder="/content/check/"
+  destination_folder="data"
   file_name = os.path.basename(fileobj)
   destination_path = os.path.join(destination_folder, file_name)
   shutil.copyfile(fileobj.name, destination_path)