Update app.py

app.py

@@ -267,47 +267,44 @@ from langchain_core.runnables import RunnablePassthrough
 import chromadb
 chromadb.api.client.SharedSystemClient.clear_system_cache()
 os.environ['MISTRAL_API_KEY'] = 'XuyOObDE7trMbpAeI7OXYr3dnmoWy3L0'
+
 class VectorData():
     def __init__(self):
         embedding_model_name = 'l3cube-pune/punjabi-sentence-similarity-sbert'
-        model_kwargs = {'device':'cpu', "trust_remote_code": True}
-        
+
+        model_kwargs = {'device':'cpu',"trust_remote_code": True}
+
         self.embeddings = HuggingFaceEmbeddings(
             model_name=embedding_model_name,
             model_kwargs=model_kwargs
         )
 
-        # Initialize ChromaDB with new configuration
-        self.client = chromadb.PersistentClient(path="chroma_db")
-        
-        # Create or get collection with embeddings
-        self.collection = self.client.get_or_create_collection(
-            name="my_collection",
-            metadata={"hnsw:space": "cosine"}
-        )
-
-        # Initialize vectorstore
-        self.vectorstore = Chroma(
-            client=self.client,
-            collection_name="my_collection",
-            embedding_function=self.embeddings
-        )
-        
+        self.vectorstore = Chroma(persist_directory="chroma_db", embedding_function=self.embeddings)
         self.retriever = self.vectorstore.as_retriever()
         self.ingested_files = []
-        
-        self.prompt = ChatPromptTemplate.from_messages([...])  # Your existing prompt
+        self.prompt = ChatPromptTemplate.from_messages(
+            [
+                (
+                    "system",
+                    """ਦਿੱਤੇ ਗਏ ਸੰਦਰਭ ਦੇ ਆਧਾਰ 'ਤੇ ਸਵਾਲ ਦਾ ਜਵਾਬ ਦਿਓ। ਜੇਕਰ ਸਵਾਲ ਦਾ ਪ੍ਰਸੰਗ ਵੈਧ ਨਹੀਂ ਹੈ ਤਾਂ ਕੋਈ ਜਵਾਬ ਨਾ ਦਿਓ। ਹਮੇਸ਼ਾ ਜਵਾਬ ਦੇ ਅੰਤ ਵਿੱਚ ਸੰਦਰਭ ਦਾ ਸਰੋਤ ਦਿਓ: 
+                    {context}
+                    """,
+                ),
+                ("human", "{question}"),
+            ]
+        )
         self.llm = ChatMistralAI(model="mistral-large-latest")
         self.rag_chain = (
-            {"context": self.retriever, "question": RunnablePassthrough()}
-            | self.prompt
-            | self.llm
-            | StrOutputParser()
-        )
-    def add_file(self, file):
+                {"context": self.retriever, "question": RunnablePassthrough()}
+                | self.prompt
+                | self.llm
+                | StrOutputParser()
+            )
+
+    def add_file(self,file):
         if file is not None:
             self.ingested_files.append(file.name.split('/')[-1])
-            self.retriever, self.vectorstore = utils.add_doc(file, self.vectorstore)
+            self.retriever, self.vectorstore = utils.add_doc(file,self.vectorstore)
             self.rag_chain = (
                 {"context": self.retriever, "question": RunnablePassthrough()}
                 | self.prompt
@@ -316,9 +313,9 @@ class VectorData():
             )
         return [[name] for name in self.ingested_files]
 
-    def delete_file_by_name(self, file_name):
+    def delete_file_by_name(self,file_name):
         if file_name in self.ingested_files:
-            self.retriever, self.vectorstore = utils.delete_doc(file_name, self.vectorstore)
+            self.retriever, self.vectorstore = utils.delete_doc(file_name,self.vectorstore)
             self.ingested_files.remove(file_name)
         return [[name] for name in self.ingested_files]
 
@@ -333,9 +330,7 @@ class VectorData():
             "ਆਰਗਨ ਆਪਣੇ ਸਾਥੀ ਦੇ ਆਪਣੀ ਪਤਨੀ ਪ੍ਰਤੀ ਸਤਿਕਾਰ ਅਤੇ ਸੇਵਾ ਨੂੰ ਕਿਵੇਂ ਵੇਖਾਉਂਦਾ ਹੈ?",
             "ਜਦੋਂ ਲਕਸ਼ਮਣ ਨੇ ਭਗਵਾਨ ਰਾਮ ਨੂੰ ਜੰਗਲ ਵਿੱਚ ਜਾਣ ਦਾ ਫੈਸਲਾ ਕੀਤਾ ਤਾਂ ਇਹ ਬਿਰਤਾਂਤ ਉਸ ਦੀਆਂ ਭਾਵਨਾਵਾਂ ਨੂੰ ਕਿਵੇਂ ਬਿਆਨ ਕਰਦਾ ਹੈ?"
         ]
-
-
-
+    
 data_obj = VectorData()
 
 # Function to handle question answering
@@ -345,52 +340,67 @@ def answer_question(question):
     return "Please enter a question."
 
 with gr.Blocks() as rag_interface:
+    # Title and Description
     gr.Markdown("# RAG Interface")
     gr.Markdown("Manage documents and ask questions with a Retrieval-Augmented Generation (RAG) system.")
 
     with gr.Row():
+        # Left Column: File Management
         with gr.Column():
             gr.Markdown("### File Management")
+
+            # File upload and ingest
             file_input = gr.File(label="Upload File to Ingest")
             add_file_button = gr.Button("Ingest File")
             
+            # Add examples for file upload with proper function and outputs
+
+            # Scrollable list for ingested files
             ingested_files_box = gr.Dataframe(
                 headers=["Files"], 
                 datatype="str",
-                row_count=4,
+                row_count=4,  # Limits the visible rows to create a scrollable view
                 interactive=False
             )
-
-            delete_option = gr.Radio(
-                choices=["Delete by File Name", "Delete All Files"],
-                label="Delete Option"
-            )
-            file_name_input = gr.Textbox(
-                label="Enter File Name to Delete",
-                visible=False
+            gr.Examples(
+                examples=[
+                    ["Examples/RESULT_OCR.txt"],
+                    ["Examples/RESULT_OCR_2.txt"],
+                    ["Examples/RESULT_OCR_3.txt"]
+                ],
+                inputs=file_input,
+                outputs=ingested_files_box,
+                fn=data_obj.add_file,
+                cache_examples=True,
+                label="Example Files"
             )
+
+            # Radio buttons to choose delete option
+            delete_option = gr.Radio(choices=["Delete by File Name", "Delete All Files"], label="Delete Option")
+            file_name_input = gr.Textbox(label="Enter File Name to Delete", visible=False)
             delete_button = gr.Button("Delete Selected")
 
+            # Show or hide file name input based on delete option selection
             def toggle_file_input(option):
                 return gr.update(visible=(option == "Delete by File Name"))
 
-            delete_option.change(
-                fn=toggle_file_input,
-                inputs=delete_option,
-                outputs=file_name_input
-            )
+            delete_option.change(fn=toggle_file_input, inputs=delete_option, outputs=file_name_input)
+
+            # Handle file ingestion
             add_file_button.click(
                 fn=data_obj.add_file,
                 inputs=file_input,
                 outputs=ingested_files_box
             )
 
+            # Handle delete based on selected option
             def delete_action(delete_option, file_name):
                 if delete_option == "Delete by File Name" and file_name:
                     return data_obj.delete_file_by_name(file_name)
                 elif delete_option == "Delete All Files":
                     return data_obj.delete_all_files()
-                return [[name] for name in data_obj.ingested_files]
+                else:
+                    return [[name] for name in data_obj.ingested_files]
 
             delete_button.click(
                 fn=delete_action,
@@ -398,12 +408,21 @@ with gr.Blocks() as rag_interface:
                 outputs=ingested_files_box
             )
 
+        # Right Column: Question Answering
         with gr.Column():
+            # gr.Markdown("### Ask a Question")
+
+            # Question input
+            # question_input = gr.Textbox(label="Enter your question")
+
+            # # Get answer button and answer output
+            # ask_button = gr.Button("Get Answer")
+            # answer_output = gr.Textbox(label="Answer", interactive=False)
+
+            # ask_button.click(fn=answer_question, inputs=question_input, outputs=answer_output)
+
             gr.Markdown("### Ask a Question")
-            example_questions = gr.Radio(
-                choices=data_obj.get_example_questions(),
-                label="Example Questions"
-            )
+            example_questions = gr.Radio(choices=data_obj.get_example_questions(), label="Example Questions")
             question_input = gr.Textbox(label="Enter your question")
             ask_button = gr.Button("Get Answer")
             answer_output = gr.Textbox(label="Answer", interactive=False)
@@ -411,42 +430,8 @@ with gr.Blocks() as rag_interface:
             def set_example_question(example):
                 return gr.update(value=example)
 
-            example_questions.change(
-                fn=set_example_question,
-                inputs=example_questions,
-                outputs=question_input
-            )
-            ask_button.click(
-                fn=answer_question,
-                inputs=question_input,
-                outputs=answer_output
-            )
-
-
-        # Right Column: Question Answering
-        # with gr.Column():
-        #     # gr.Markdown("### Ask a Question")
-
-        #     # Question input
-        #     # question_input = gr.Textbox(label="Enter your question")
-
-        #     # # Get answer button and answer output
-        #     # ask_button = gr.Button("Get Answer")
-        #     # answer_output = gr.Textbox(label="Answer", interactive=False)
-
-        #     # ask_button.click(fn=answer_question, inputs=question_input, outputs=answer_output)
-
-        #     gr.Markdown("### Ask a Question")
-        #     example_questions = gr.Radio(choices=data_obj.get_example_questions(), label="Example Questions")
-        #     question_input = gr.Textbox(label="Enter your question")
-        #     ask_button = gr.Button("Get Answer")
-        #     answer_output = gr.Textbox(label="Answer", interactive=False)
-
-        #     def set_example_question(example):
-        #         return gr.update(value=example)
-
-        #     example_questions.change(fn=set_example_question, inputs=example_questions, outputs=question_input)
-        #     ask_button.click(fn=answer_question, inputs=question_input, outputs=answer_output)
+            example_questions.change(fn=set_example_question, inputs=example_questions, outputs=question_input)
+            ask_button.click(fn=answer_question, inputs=question_input, outputs=answer_output)
 
 
 with gr.Blocks() as demo: