Update app.py

app.py

@@ -1,3 +1,4 @@
+#Backup
 import os
 import time
 import threading
@@ -22,74 +23,79 @@ APP_START_TIME = datetime.datetime.now(datetime.timezone.utc)
 
 os.environ["PYTORCH_JIT"] = "0"
 
-# ---------------- PDF & DOCX & JSON Extraction ----------------
+# --- PDF Extraction ---
 def _extract_tables_from_page(page):
+    """Extracts tables from a single page of a PDF."""
+
     tables = page.extract_tables()
+    if not tables:
+        return []
+
     formatted_tables = []
     for table in tables:
         formatted_table = []
         for row in table:
-            formatted_row = [cell if cell is not None else "" for cell in row]
-            formatted_table.append(formatted_row)
+            if row:  # Filter out empty rows
+                formatted_row = [cell if cell is not None else "" for cell in row]  # Replace None with ""
+                formatted_table.append(formatted_row)
+            else:
+                formatted_table.append([""])  # Append an empty row if the row is None
         formatted_tables.append(formatted_table)
     return formatted_tables
-
+    
 def extract_text_from_pdf(pdf_path):
     text_output = StringIO()
     all_tables = []
     try:
         with pdfplumber.open(pdf_path) as pdf:
             for page in pdf.pages:
-                all_tables.extend(_extract_tables_from_page(page))
+                # Extract tables
+                page_tables = _extract_tables_from_page(page)
+                if page_tables:
+                    all_tables.extend(page_tables)
+                # Extract text
                 text = page.extract_text()
                 if text:
                     text_output.write(text + "\n\n")
     except Exception as e:
-        print(f"pdfplumber error: {e}")
+        print(f"Error extracting with pdfplumber: {e}")
+        # Fallback to pdfminer if pdfplumber fails
         with open(pdf_path, 'rb') as file:
-            extract_text_to_fp(file, text_output, laparams=LAParams(), output_type='text')
-    return text_output.getvalue(), all_tables
+            extract_text_to_fp(file, text_output, laparams=LAParams(), output_type='text', codec=None)
+    extracted_text = text_output.getvalue()
+    return extracted_text, all_tables  # Return text and list of tables
+
+def clean_extracted_text(text):
+    lines = text.splitlines()
+    cleaned = []
+    for line in lines:
+        line = line.strip()
+        if line:
+            line = ' '.join(line.split())
+            cleaned.append(line)
+    return '\n'.join(cleaned)
 
 def _format_tables_internal(tables):
+    """Formats extracted tables into a string representation."""
+
     formatted_tables_str = []
     for table in tables:
+        # Use csv writer to handle commas and quotes correctly
         with StringIO() as csvfile:
-            writer = csv.writer(csvfile)
-            writer.writerows(table)
+            csvwriter = csv.writer(csvfile)
+            csvwriter.writerows(table)
             formatted_tables_str.append(csvfile.getvalue())
     return "\n\n".join(formatted_tables_str)
 
-def clean_extracted_text(text):
-    return '\n'.join(' '.join(line.strip().split()) for line in text.splitlines() if line.strip())
-
+# --- DOCX Extraction ---
 def extract_text_from_docx(docx_path):
     try:
         doc = docx.Document(docx_path)
         return '\n'.join(para.text for para in doc.paragraphs)
-    except:
-        return ""
-
-def load_json_data(json_path):
-    try:
-        with open(json_path, 'r', encoding='utf-8') as f:
-            data = json.load(f)
-        if isinstance(data, dict):
-            # Flatten dictionary values (avoiding nested structures as strings)
-            return "\n".join(f"{key}: {value}" for key, value in data.items() if not isinstance(value, (dict, list)))
-        elif isinstance(data, list):
-            # Flatten list of dictionaries
-            all_items = []
-            for item in data:
-                if isinstance(item, dict):
-                    all_items.append("\n".join(f"{key}: {value}" for key, value in item.items() if not isinstance(value, (dict, list))))
-            return "\n\n".join(all_items)
-        else:
-            return json.dumps(data, ensure_ascii=False, indent=2)
-    except Exception as e:
-        print(f"JSON read error: {e}")
+    except Exception:
         return ""
 
-# ---------------- Chunking ----------------
+# --- Chunking ---
 def chunk_text(text, tokenizer, chunk_size=128, chunk_overlap=32, max_tokens=512):
     tokens = tokenizer.tokenize(text)
     chunks = []
@@ -104,13 +110,12 @@ def chunk_text(text, tokenizer, chunk_size=128, chunk_overlap=32, max_tokens=512
         start += chunk_size - chunk_overlap
     return chunks
 
-
 def retrieve_chunks(question, index, embed_model, text_chunks, k=3):
-    q_embedding = embed_model.encode(question)
-    D, I = index.search(np.array([q_embedding]), k)
+    question_embedding = embed_model.encode(question)
+    D, I = index.search(np.array([question_embedding]), k)
     return [text_chunks[i] for i in I[0]]
 
-# ---------------- Groq Answer Generator ----------------
+# --- Groq Answer Generator ---
 def generate_answer_with_groq(question, context):
     url = "https://api.groq.com/openai/v1/chat/completions"
     api_key = os.environ.get("GROQ_API_KEY")
@@ -120,9 +125,8 @@ def generate_answer_with_groq(question, context):
     }
     prompt = (
         f"Customer asked: '{question}'\n\n"
-        f"Here is the relevant information to help:\n{context}\n\n"
-        f"Respond in a friendly and helpful tone as a toy shop support agent, "
-        f"addressing the customer by their name if it's available in the context."
+        f"Here is the relevant product or policy info to help:\n{context}\n\n"
+        f"Respond in a friendly and helpful tone as a toy shop support agent."
     )
     payload = {
         "model": "llama3-8b-8192",
@@ -130,11 +134,9 @@ def generate_answer_with_groq(question, context):
             {
                 "role": "system",
                 "content": (
-                    "You are ToyBot, a friendly WhatsApp assistant for an online toy shop. "
-                    "Help customers with toys, delivery, and returns in a helpful tone. "
-                    "When responding, try to find the customer's name in the provided context "
-                    "and address them directly. If the context contains order details and status, "
-                    "include that information in your response."
+                    "You are ToyBot, a friendly and helpful WhatsApp assistant for an online toy shop. "
+                    "Your goal is to politely answer customer questions, help them choose the right toys, "
+                    "provide order or delivery information, explain return policies, and guide them through purchases."
                 )
             },
             {"role": "user", "content": prompt},
@@ -174,37 +176,46 @@ def send_twilio_message(client, conversation_sid, body):
         author="system", body=body
     )
 
-# ---------------- Knowledge Base Setup ----------------
+# --- Load Knowledge Base ---
 def setup_knowledge_base():
     folder_path = "docs"
     all_text = ""
 
-    for filename in os.listdir(folder_path):
-        file_path = os.path.join(folder_path, filename)
-        if filename.endswith(".pdf"):
-            text, tables = extract_text_from_pdf(file_path)
-            all_text += clean_extracted_text(text) + "\n"
-            all_text += _format_tables_internal(tables) + "\n"
-        elif filename.endswith(".docx"):
-            text = extract_text_from_docx(file_path)
-            all_text += clean_extracted_text(text) + "\n"
-        elif filename.endswith(".json"):
-            text = load_json_data(file_path)
-            all_text += text + "\n"
-        elif filename.endswith(".csv"):
-            try:
-                with open(file_path, newline='', encoding='utf-8') as csvfile:
-                    reader = csv.DictReader(csvfile)
-                    for row in reader:
-                        line = ' | '.join(f"{k}: {v}" for k, v in row.items())
-                        all_text += line + "\n"
-            except Exception as e:
-                print(f"CSV read error: {e}")
+    # Process PDFs
+    for filename in ["FAQ.pdf", "ProductReturnPolicy.pdf"]:
+        pdf_path = os.path.join(folder_path, filename)
+        text, tables = extract_text_from_pdf(pdf_path)
+        all_text += clean_extracted_text(text) + "\n"
+        all_text += _format_tables_internal(tables) + "\n"
+
+    # Process CSVs
+    for filename in ["CustomerOrders.csv"]:
+        csv_path = os.path.join(folder_path, filename)
+        try:
+            with open(csv_path, newline='', encoding='utf-8') as csvfile:
+                reader = csv.DictReader(csvfile)
+                for row in reader:
+                    line = f"Order ID: {row.get('OrderID')} | Customer Name: {row.get('CustomerName')} | Order Date: {row.get('OrderDate')} | ProductID: {row.get('ProductID')} | Date: {row.get('OrderDate')} | Quantity: {row.get('Quantity')} | UnitPrice(USD): {row.get('UnitPrice(USD)')} | TotalPrice(USD): {row.get('TotalPrice(USD)')} | ShippingAddress: {row.get('ShippingAddress')} | OrderStatus: {row.get('OrderStatus')}"
+                    all_text += line + "\n"
+        except Exception as e:
+            print(f"❌ Error reading {filename}: {e}")
+
+    for filename in ["Products.csv"]:
+        csv_path = os.path.join(folder_path, filename)
+        try:
+            with open(csv_path, newline='', encoding='utf-8') as csvfile:
+                reader = csv.DictReader(csvfile)
+                for row in reader:
+                    line = f"Product ID: {row.get('ProductID')} | Toy Name: {row.get('ToyName')} | Category: {row.get('Category')} | Price(USD): {row.get('Price(USD)')} | Stock Quantity: {row.get('StockQuantity')} | Description: {row.get('Description')}"
+                    all_text += line + "\n"
+        except Exception as e:
+            print(f"❌ Error reading {filename}: {e}")
 
-    tokenizer = AutoTokenizer.from_pretrained('bert-base-uncased', model_max_length=512)
+    # Tokenization & chunking
+    tokenizer = AutoTokenizer.from_pretrained('bert-base-uncased')
     chunks = chunk_text(all_text, tokenizer)
     model = SentenceTransformer('all-mpnet-base-v2')
-    embeddings = model.encode(chunks, show_progress_bar=False)
+    embeddings = model.encode(chunks, show_progress_bar=False, truncation=True, max_length=512)
     dim = embeddings[0].shape[0]
     index = faiss.IndexFlatL2(dim)
     index.add(np.array(embeddings).astype('float32'))
@@ -264,10 +275,7 @@ def start_conversation_monitor(client, index, embed_model, text_chunks):
 # --- Streamlit UI ---
 st.set_page_config(page_title="Quasa – Al-Powered WhatsApp Chatbot", layout="wide")
 st.title("📱 Quasa – Al-Powered WhatsApp Chatbot")
-index, model, chunks = setup_knowledge_base()
 
-st.success("Knowledge base loaded.")
-#st.write("Waiting for WhatsApp messages...")
 account_sid = st.secrets.get("TWILIO_SID")
 auth_token = st.secrets.get("TWILIO_TOKEN")
 GROQ_API_KEY = st.secrets.get("GROQ_API_KEY")