Update app.py

app.py

@@ -1,16 +1,34 @@
 import sqlite3
 import json
-from transformers import AutoModelForCausalLM, AutoTokenizer
 import uuid
 import datetime
+import logging
 
-# Initialize AI model (Mistral-7B placeholder; replace with actual model)
-model_name = "mistralai/Mixtral-8x7B-Instruct-v0.1"
+# Setup logging
+logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
+
+# Attempt to import transformers
 try:
-    tokenizer = AutoTokenizer.from_pretrained(model_name)
-    model = AutoModelForCausalLM.from_pretrained(model_name)
-except:
-    # Fallback for demo (mock AI response)
+    from transformers import AutoModelForCausalLM, AutoTokenizer
+    TRANSFORMERS_AVAILABLE = True
+except ImportError:
+    logging.warning("Transformers library not found. Using fallback parser.")
+    TRANSFORMERS_AVAILABLE = False
+    AutoModelForCausalLM = None
+    AutoTokenizer = None
+
+# Initialize AI model (distilbert as placeholder; replace with fine-tuned model or Mistral-7B)
+model_name = "distilbert-base-uncased"  # Lightweight model for demo
+if TRANSFORMERS_AVAILABLE:
+    try:
+        tokenizer = AutoTokenizer.from_pretrained(model_name)
+        model = AutoModelForCausalLM.from_pretrained(model_name)
+        logging.info(f"Loaded model: {model_name}")
+    except Exception as e:
+        logging.error(f"Failed to load model {model_name}: {e}")
+        tokenizer = None
+        model = None
+else:
     tokenizer = None
     model = None
 
@@ -79,24 +97,25 @@ def initialize_chart_of_accounts():
     VALUES (?, ?, ?, ?, ?, ?)
     """, accounts)
     conn.commit()
+    logging.info("Chart of accounts initialized.")
 
 # Parse prompt using AI model (or fallback)
 def parse_prompt(prompt):
     if model and tokenizer:
-        input_text = f"""
-        Parse the following accounting prompt into a JSON object with:
-        - debit: {account, type, amount}
-        - credit: {account, type, amount}
-        - payment_method: 'cash' or 'credit' or null
-        Prompt: {prompt}
-        """
-        inputs = tokenizer(input_text, return_tensors="pt")
-        outputs = model.generate(**inputs, max_length=300)
-        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
         try:
+            input_text = f"""
+            Parse the following accounting prompt into a JSON object with:
+            - debit: {{account, type, amount}}
+            - credit: {{account, type, amount}}
+            - payment_method: 'cash' or 'credit' or null
+            Prompt: {prompt}
+            """
+            inputs = tokenizer(input_text, return_tensors="pt")
+            outputs = model.generate(**inputs, max_length=300)
+            response = tokenizer.decode(outputs[0], skip_special_tokens=True)
             return json.loads(response)
-        except:
-            pass
+        except Exception as e:
+            logging.warning(f"Model parsing failed: {e}. Using fallback parser.")
 
     # Fallback parsing for common scenarios
     prompt_lower = prompt.lower()
@@ -110,6 +129,7 @@ def parse_prompt(prompt):
                 pass
     
     if not amount:
+        logging.error("No amount found in prompt.")
         return None
 
     if "laptop" in prompt_lower:
@@ -130,6 +150,7 @@ def parse_prompt(prompt):
             "credit": {"account": credit_account, "type": credit_type, "amount": amount},
             "payment_method": payment_method
         }
+    logging.error("Prompt not recognized.")
     return None
 
 # Generate journal entry
@@ -184,6 +205,7 @@ def generate_journal_entry(prompt, follow_up_response=None):
     VALUES (?, ?, ?, ?, ?, ?)
     """, (entry_id, date, debit_result[0], credit_result[0], amount, prompt))
     conn.commit()
+    logging.info(f"Journal entry created: Debit {debit_account} ${amount}, Credit {credit_account} ${amount}")
 
     return f"Journal Entry Created: Debit {debit_account} ${amount}, Credit {credit_account} ${amount}"