add flow to app

app.py

@@ -1,15 +1,52 @@
 import gradio as gr
+import os
+from utils.document_parsing import DocParsing
+from utils.retrieval import Retrieval
+from utils.llm_generation import LLMGeneration
+import json
 
-def process_inputs(api_key, pdf_file, questions):
 
-    # In this placeholder, we'll simply echo the inputs.
-    if pdf_file is not None:
-        pdf_name = pdf_file.name
-    else:
-        pdf_name = "No file uploaded"
-        
+embedding_model_name = "sentence-transformers/all-MiniLM-L6-v2"
+#  Setting up Retriever
+retriever = Retrieval(model_name=embedding_model_name)
+
+
+llm_model_name = "gpt-4o-mini"
+# Settting up LLMGenerator
+llm_generator = LLMGeneration(llm_model_name=llm_model_name)
+
+def set_api_key(api_key):
+    os.environ['OPENAI_API_KEY'] = api_key
+
+def process_inputs(api_key:str, pdf_file, questions: str):
+
+    # Setup Api KEY
+    set_api_key(api_key)
+
+    if pdf_file is None:
+        raise Exception("Blaf")
+    
+    
+    # Parsing the pdf
+    doc_handler = DocParsing(file_path=pdf_file.name,model_name=embedding_model_name)
+    docs = doc_handler.process_pdf()
+
+    # Create vector store
+    retriever.create_vector_store(chunks=docs)
+
+
+    output_dict = {}
     questions_list = questions.strip().split('\n')
-    response = f"API Key: {api_key}\nUploaded PDF: {pdf_name}\nQuestions: {questions_list}"
+    for question in questions_list:
+        
+        # Retrieve top similar chunks
+        similar_chunks = retriever.search(query=question, k=10)
+
+        # Generate the answer
+        output_dict[question] = llm_generator.generate_answer(question, similar_chunks)
+        
+    
+    response = json.dumps(output_dict)
     return response
 
 with gr.Blocks() as demo: