Update app.py

app.py

@@ -7,16 +7,13 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 import faiss
 from simple_salesforce import Salesforce
 from dotenv import load_dotenv
-import json
-import zipfile
-from pathlib import Path
 
 # Setup logging
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
 # Load environment variables from .env file
-load_dotenv()  # Load the .env file
+load_dotenv()
 
 # Get the Salesforce credentials from environment variables
 sf_username = os.getenv("SF_USERNAME")
@@ -42,23 +39,15 @@ except Exception as e:
     logger.error(f"❌ Salesforce connection failed: {str(e)}")
     raise
 
-# --- Extract zip files and read documents ---
-def extract_zip(zip_path, extract_to):
-    try:
-        with zipfile.ZipFile(zip_path, 'r') as zip_ref:
-            zip_ref.extractall(extract_to)
-        logger.info(f"Extracted {zip_path} to {extract_to}")
-    except Exception as e:
-        logger.error(f"Failed to extract {zip_path}: {str(e)}")
-        raise
-
-def load_documents(folder_path):
-    documents = []
-    sources = []
-    for file in Path(folder_path).rglob("*.txt"):
-        text = file.read_text(encoding="utf-8", errors="ignore")
-        documents.append(text)
-        sources.append(file.name)
+# --- Simulate document loading (replace with actual document loading in Hugging Face) ---
+def load_documents():
+    # Simulate documents for Hugging Face compatibility (replace with actual data)
+    documents = [
+        "Permanent employment status is granted after 6 months of continuous employment with satisfactory performance.",
+        "HR policies include 20 days of paid leave annually and mandatory diversity training.",
+        "Contract clauses require a 30-day notice period for termination."
+    ]
+    sources = ["policy1.txt", "hr1.txt", "contract1.txt"]
     return documents, sources
 
 # --- Chunking ---
@@ -68,70 +57,52 @@ text_splitter = RecursiveCharacterTextSplitter(chunk_size=300, chunk_overlap=50)
 model = SentenceTransformer("all-MiniLM-L6-v2")
 
 # --- Preprocessing ---
-data_dir = Path("./data")
-data_dir.mkdir(exist_ok=True)
-
-doc_folders = [
-    ("Company_Policies.zip", "Company_Policies"),
-    ("HR_Policies.zip", "Hr_Policies"),
-    ("Contract_Clauses.zip", "Contract_Clauses")
-]
-
 all_chunks = []
 metadata = []
 
-for zip_name, folder in doc_folders:
-    zip_path = Path(zip_name)
-    if not zip_path.exists():
-        logger.error(f"Zip file {zip_name} not found")
-        raise FileNotFoundError(f"Zip file {zip_name} not found")
-    extract_path = data_dir / folder
-    extract_path.mkdir(exist_ok=True)
-    extract_zip(zip_path, extract_path)
-    docs, sources = load_documents(extract_path)
+try:
+    docs, sources = load_documents()
     if not docs:
-        logger.error(f"No documents found in {extract_path}")
-        raise ValueError(f"No documents found in {extract_path}")
+        logger.error("No documents found")
+        raise ValueError("No documents found")
     for doc, src in zip(docs, sources):
         chunks = text_splitter.split_text(doc)
         all_chunks.extend(chunks)
-        src_url = f"https://company.com/{folder}/{src}"
+        src_url = f"https://company.com/documents/{src}"
         metadata.extend([src_url] * len(chunks))
+except Exception as e:
+    logger.error(f"Error loading documents: {str(e)}")
+    raise
 
 # --- Embeddings + FAISS index ---
-embeddings = model.encode(all_chunks)
-index = faiss.IndexFlatL2(embeddings.shape[1])
-index.add(np.array(embeddings))
-logger.info("FAISS index built successfully")
+try:
+    embeddings = model.encode(all_chunks)
+    index = faiss.IndexFlatL2(embeddings.shape[1])
+    index.add(np.array(embeddings))
+    logger.info("FAISS index built successfully")
+except Exception as e:
+    logger.error(f"Error building FAISS index: {str(e)}")
+    raise
 
 # --- Create Record in Salesforce ---
 def create_salesforce_record(query, answer, confidence_percentage, source_link):
     try:
-        # Convert the confidence_percentage to Python float (to avoid numpy float32)
         confidence_percentage = float(confidence_percentage)
-
-        # Data with correctly mapped field names
         data = {
-            "Query__c": query,  # Field for User Query
-            "Answer__c": answer,  # Field for Answer
-            "Confidence_Percentage__c": confidence_percentage,  # Field for Confidence Score
-            "Document_link__c": source_link,  # Field for Document Link
+            "Query__c": query,
+            "Answer__c": answer,
+            "Confidence_Percentage__c": confidence_percentage,
+            "Document_link__c": source_link,
         }
-
-        # Creating the record in Salesforce
         response = sf.chat_query_log__c.create(data)
-        
-        # Check if record was created successfully
-        if 'id' in response:  # If the response contains an 'id', the record is created successfully
+        if 'id' in response:
             record_id = response['id']
             logger.info(f"✅ Record created successfully in Salesforce with ID: {record_id}")
-            return record_id  # Return the Salesforce record ID
+            return record_id
         else:
-            # Log the failure response
             logger.error(f"❌ Failed to create Salesforce record. Response: {response}")
             return None
     except Exception as e:
-        # Log any error during record creation
         logger.error(f"Error creating Salesforce record: {str(e)}")
         return None
 
@@ -145,81 +116,61 @@ def answer_query(query):
         top_sources = [metadata[i] for i in I[0]]
         distances = D[0]
 
-        relevant_chunks = [
-            chunk for chunk, dist in zip(top_chunks, distances) if dist < 0.8
-        ]
-        relevant_sources = [
-            src for src, dist in zip(top_sources, distances) if dist < 0.8
-        ]
+        relevant_chunks = [chunk for chunk, dist in zip(top_chunks, distances) if dist < 0.8]
+        relevant_sources = [src for src, dist in zip(top_sources, distances) if dist < 0.8]
 
         if not relevant_chunks:
-            return "No relevant information found.", "Confidence: 0%", "Source Link: None"
+            return "No relevant information found.", "Confidence: 0%", "Source Link: None", None
 
         answer = relevant_chunks[0].strip()
         min_distance = min(distances)
         confidence_percentage = max(0, 100 - (min_distance * 100))
         source_link = relevant_sources[0] if relevant_sources else "None"
 
-        # Create Salesforce record for the query response
         record_id = create_salesforce_record(query, answer, confidence_percentage, source_link)
 
-        if record_id:
-            return (
-                answer,
-                f"Confidence: {confidence_percentage:.2f}%",
-                f"Source Link: {source_link}",
-                f"Salesforce Record ID: {record_id}"  # Display the Salesforce record ID
-            )
-        else:
-            return (
-                answer,
-                f"Confidence: {confidence_percentage:.2f}%",
-                f"Source Link: {source_link}",
-                "Failed to create record in Salesforce"
-            )
+        return (
+            answer,
+            f"Confidence: {confidence_percentage:.2f}%",
+            f"Source Link: {source_link}",
+            record_id
+        )
     except Exception as e:
         logger.error(f"Error in answer_query: {str(e)}")
-        return f"Error: {str(e)}", "", "", ""
+        return f"Error: {str(e)}", "", "", None
 
-# --- Gradio Chatbot UI ---
-def process_question(q):
-    if not q.strip():
-        return "Please enter a question.", "", ""
-    
-    answer, confidence, source, record_id = answer_query(q)
-    return answer, confidence, source, record_id
+# --- Gradio Chatbot Function ---
+def process_question(user_input, history):
+    if not user_input.strip():
+        return history + [[user_input, "Please enter a valid question."]]
+
+    answer, confidence, source, record_id = answer_query(user_input)
+    response = f"{answer}\n\n{confidence}\n{source}"
+    if record_id:
+        response += f"\nSalesforce Record ID: {record_id}"
+    else:
+        response += "\nFailed to create record in Salesforce"
+
+    return history + [[user_input, response]]
 
-# --- Chatbot UI Design (Chat bubble style) ---
+# --- Gradio Chatbot UI ---
 with gr.Blocks(title="Company Documents Q&A Chatbot", theme=gr.themes.Soft()) as demo:
     gr.Markdown("## 📚 Company Documents Q&A Chatbot")
-
+    chatbot = gr.Chatbot(label="Chat", height=400)
     with gr.Row():
-        with gr.Column(scale=3):
-            question = gr.Textbox(
-                label="Ask a Question",
-                placeholder="What are the conditions for permanent employment status?",
-                lines=1,
-                interactive=True
-            )
-        with gr.Column(scale=1):
-            submit_btn = gr.Button("Submit", variant="primary")
+        question = gr.Textbox(
+            label="Ask a Question",
+            placeholder="What are the conditions for permanent employment status?",
+            lines=1,
+            interactive=True
+        )
+        submit_btn = gr.Button("Submit", variant="primary")
+    
+    submit_btn.click(
+        fn=process_question,
+        inputs=[question, chatbot],
+        outputs=chatbot,
+        _js="() => {const txt = document.querySelector('input[type=text]'); txt.value=''; txt.focus(); return []}"
+    )
 
-    with gr.Row():
-        with gr.Column():
-            # Chatbot styled for a more modern chat look with bubbles
-            output_area = gr.HTML(
-                label="Chat", 
-                elem_id="chatbox",
-                value="""
-                    <div style="padding: 10px; background-color: #f5f5f5; border-radius: 10px;">
-                        <div style="padding: 5px 10px; background-color: #dfe1e6; border-radius: 10px; margin-bottom: 10px;">
-                            <b>User:</b> <span id="user-message"> </span>
-                        </div>
-                        <div style="padding: 5px 10px; background-color: #007bff; color: white; border-radius: 10px;">
-                            <b>Bot:</b> <span id="bot-message"> </span>
-                        </div>
-                    </div>""")
-
-    submit_btn.click(fn=process_question, inputs=question, outputs=[output_area])
-
-demo.launch(server_name="0.0.0.0", server_port=7860, share=True)
+demo.launch(server_name="0.0.0.0", server_port=7860)