added fraud check

data.json

@@ -0,0 +1,334 @@
+[
+    {
+        "merchant": "TESCO SHREWSBURY CATTLE MARKET",
+        "date": "17/10/22",
+        "total_amount": 55.3,
+        "items": [
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.0
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.0
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 7.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 8.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 8.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.0
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.0
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 1.1
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 1.0
+            }
+        ],
+        "fraud_check": false
+    },
+    {
+        "merchant": "TESCO",
+        "date": "17/10/22",
+        "total_amount": 55.3,
+        "items": [
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            },
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            }
+        ],
+        "fraud_check": false
+    },
+    {
+        "merchant": "TESCO SHREWSBURY CATTLE MARKET",
+        "date": "17/10/22",
+        "total_amount": 55.3,
+        "items": [
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            },
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            }
+        ],
+        "fraud_check": false
+    },
+    {
+        "fraud_check": false,
+        "merchant": "TESCO SHREWSBURY CATTLE MARKET",
+        "date": "17/10/22",
+        "total_amount": 55.3,
+        "items": [
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            }
+        ]
+    },
+    {
+        "fraud_check": false,
+        "merchant": "Tesco",
+        "date": "17/10/22",
+        "total_amount": 55.3,
+        "items": [
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "PROCECCO",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            },
+            {
+                "description": "TAPE MEASURE",
+                "amount": 2.1
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "PROTEIN COOKIE",
+                "amount": 1.8
+            },
+            {
+                "description": "CONDOMS",
+                "amount": 10.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "SAUVIGNON BLNC",
+                "amount": 7.0
+            },
+            {
+                "description": "KRONENBOURG",
+                "amount": 4.6
+            }
+        ]
+    }
+]

fraud.py

@@ -0,0 +1,118 @@
+import json
+import os
+
+DATA_FILE = 'data.json'
+
+def load_receipts():
+    """Load all stored receipts from data.json."""
+    if not os.path.exists(DATA_FILE):
+        print("Data file doesn't exist, returning empty list")
+        return []
+    
+    try:
+        with open(DATA_FILE, 'r') as f:
+            data = json.load(f)
+            print(f"Loaded data type: {type(data)}")
+            
+            # Handle case where data.json contains a single object instead of array
+            if isinstance(data, dict):
+                print("Data is a single dict, wrapping in array")
+                return [data]
+            elif isinstance(data, list):
+                print(f"Data is a list with {len(data)} items")
+                return data
+            else:
+                print("Data is neither dict nor list, returning empty list")
+                return []
+    except Exception as e:
+        print(f"Error loading receipts: {e}")
+        return []
+
+def save_receipts(receipts):
+    """Save receipts back to data.json."""
+    with open(DATA_FILE, 'w') as f:
+        json.dump(receipts, f, indent=4)
+
+def receipts_are_equal(receipt1, receipt2):
+    """Check if two receipts are the same, comparing only basic fields."""
+    print(f"\n=== COMPARING RECEIPTS ===")
+    print(f"Receipt1 merchant: {receipt1.get('merchant')}")
+    print(f"Receipt2 merchant: {receipt2.get('merchant')}")
+    
+    # Check if both receipts exist
+    if not receipt1 or not receipt2:
+        print("One or both receipts are empty")
+        return False
+    
+    # Check basic fields with special handling for merchant
+    merchant1 = receipt1.get('merchant', '').lower()
+    merchant2 = receipt2.get('merchant', '').lower()
+    date1 = receipt1.get('date')
+    date2 = receipt2.get('date')
+    amount1 = receipt1.get('total_amount')
+    amount2 = receipt2.get('total_amount')
+    
+    print(f"Comparing merchant (lowercase): '{merchant1}' vs '{merchant2}'")
+    print(f"Comparing date: '{date1}' vs '{date2}'")
+    print(f"Comparing total_amount: '{amount1}' vs '{amount2}'")
+    
+    if merchant1 != merchant2:
+        print("Merchant field doesn't match")
+        return False
+    
+    if date1 != date2:
+        print("Date field doesn't match")
+        return False
+    
+    if amount1 != amount2:
+        print("Total amount field doesn't match")
+        return False
+    
+    print("All basic fields match - receipts are equal")
+    return True
+
+def is_duplicate(new_receipt, receipts):
+    """Return True if new_receipt is already in receipts."""
+    print(f"\n=== CHECKING FOR DUPLICATES ===")
+    print(f"Checking new receipt against {len(receipts)} existing receipts")
+    
+    for i, old_receipt in enumerate(receipts):
+        print(f"\nChecking against receipt {i}:")
+        print(f"  Stored: {old_receipt.get('merchant')} - {old_receipt.get('date')} - {old_receipt.get('total_amount')}")
+        print(f"  New: {new_receipt.get('merchant')} - {new_receipt.get('date')} - {new_receipt.get('total_amount')}")
+        
+        if receipts_are_equal(old_receipt, new_receipt):
+            print(f"DUPLICATE FOUND at index {i}!")
+            return True
+    
+    print("No duplicates found")
+    return False
+
+def process_receipt(new_receipt):
+    print(f"\n=== PROCESSING RECEIPT ===")
+    # Ensure we're working with a dict, not a string
+    if isinstance(new_receipt, str):
+        try:
+            new_receipt = json.loads(new_receipt)
+        except:
+            return {"error": "Invalid receipt format"}
+    
+    receipts = load_receipts()
+    print(f"Loaded {len(receipts)} existing receipts")
+    
+    if is_duplicate(new_receipt, receipts):
+        print("SETTING FRAUD CHECK TO TRUE")
+        new_receipt['fraud_check'] = True
+        # Do not save, just return
+    else:
+        print("SETTING FRAUD CHECK TO FALSE - SAVING RECEIPT")
+        new_receipt['fraud_check'] = False
+        receipts.append(new_receipt)
+        save_receipts(receipts)
+    
+    return new_receipt
+
+# ---- Usage ----
+# new_receipt = { ... }  # Your receipt JSON here
+# result = process_receipt(new_receipt)
+# print(result)

models.py

@@ -6,6 +6,7 @@ class ReceiptItem(BaseModel):
     amount: float
 
 class ReceiptData(BaseModel):
+    fraud_check: Optional[bool] = False  # Optional field for fraud detection
     merchant: str
     date: str  
     total_amount: float

pipeline.py

@@ -7,6 +7,7 @@ import base64
 import json
 from models import ReceiptData, ChildFeeForm
 from form_fill import fill_child_fee_pdf
+from fraud import process_receipt
 
 load_dotenv()
 openai.api_key = os.getenv("OPENAI_API_KEY", "").strip()
@@ -20,9 +21,10 @@ reciept_system_prompt = (
     "    description: str\n"
     "    amount: float\n\n"
     "class ReceiptData(BaseModel):\n"
-    "    merchant: str\n"
+    "    fraud_check: Optional[bool] = False  # Optional field for fraud detection, always set to false"
+    "    merchant: str #Only extract the brand name, not the branch name - Only the brand\n"
     "    date: str\n"
-    "    total_amount: float\n"
+    "    total_amount: float\n #Try your hardest to find the accurate total amount\n"
     "    items: Optional[List[ReceiptItem]] = None\n"
     "- Extract only the above given information.\n"
     "- If a value is missing, set it to null, \"\", or an empty list as appropriate.\n"
@@ -54,14 +56,14 @@ fee_bill_system_prompt = (
 
 
 
-def pil_to_bytes(pil_img, quality=60):
+def pil_to_bytes(pil_img, quality=70):
     buf = BytesIO()
     pil_img.save(buf, format='JPEG', quality=quality)
     buf.seek(0)
     return buf
 
 
-def preprocess_image(pil_img, max_size=512):
+def preprocess_image(pil_img, max_size=812):
     return pil_img.resize((max_size, max_size), Image.LANCZOS)
 
 
@@ -95,10 +97,19 @@ def extract_info(pil_img):
             if raw_output.startswith("json"):
                 raw_output = raw_output[4:].strip()
         data = json.loads(raw_output)
-        print(data)
+        # print(data)
         validated = ReceiptData(**data)
-        json_block = json.dumps(validated.dict(), indent=2, ensure_ascii=False)
-        return f"```json\n{json_block}\n```"
+        # json_block = json.dumps(validated.dict(), indent=2, ensure_ascii=False)
+
+        validated_dict = validated.dict()  # This is a Python dict, perfect for fraud check
+        print(validated_dict)
+        result = process_receipt(validated_dict)  # This expects a dict!
+
+
+        result_json = json.dumps(result, indent=2, ensure_ascii=True)  # For display
+        print(result_json)
+        return f"```json\n{result_json}\n```"
+
     except Exception as e:
         return f"```json\n{json.dumps({'error': str(e), 'raw_output': raw_output}, indent=2)}\n```"