Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -93,8 +93,8 @@ def detect_anomalies(df):
|
|
| 93 |
df["is_anomaly"] = model.fit_predict(X_scaled)
|
| 94 |
return df
|
| 95 |
|
| 96 |
-
def calculate_fraud_score(amount, is_anomaly,
|
| 97 |
-
"""Calculate fraud score based on amount, anomaly, and
|
| 98 |
score = 0.0
|
| 99 |
reasoning = []
|
| 100 |
|
|
@@ -109,9 +109,9 @@ def calculate_fraud_score(amount, is_anomaly, items_listed):
|
|
| 109 |
score += 30
|
| 110 |
reasoning.append("Invoice flagged as an anomaly.")
|
| 111 |
|
| 112 |
-
if
|
| 113 |
score += 10
|
| 114 |
-
reasoning.append("Excessive
|
| 115 |
|
| 116 |
return min(score, 100), "; ".join(reasoning)
|
| 117 |
|
|
@@ -123,33 +123,31 @@ def process_invoice(pdf_file):
|
|
| 123 |
|
| 124 |
vendor_name, amount = extract_entities(text)
|
| 125 |
invoice_date = datetime.now().date()
|
| 126 |
-
|
| 127 |
|
| 128 |
data = {
|
| 129 |
"invoice_id": str(uuid.uuid4()),
|
| 130 |
"vendor_name": vendor_name,
|
| 131 |
"amount": amount,
|
| 132 |
"invoice_date": invoice_date,
|
| 133 |
-
"
|
| 134 |
}
|
| 135 |
df = pd.DataFrame([data])
|
| 136 |
|
| 137 |
df = detect_anomalies(df)
|
| 138 |
|
| 139 |
fraud_score, fraud_reasoning = calculate_fraud_score(
|
| 140 |
-
df["amount"].iloc[0], df["is_anomaly"].iloc[0],
|
| 141 |
)
|
| 142 |
|
| 143 |
output = {
|
| 144 |
"Invoice_Record__c": {
|
| 145 |
-
"
|
| 146 |
"Invoice_Amount__c": amount,
|
| 147 |
"Invoice_Date__c": str(invoice_date),
|
| 148 |
-
"Items_Listed__c": items_listed,
|
| 149 |
"Fraud_Score__c": fraud_score,
|
| 150 |
"Fraud_Reasoning__c": fraud_reasoning,
|
| 151 |
"Flagged__c": fraud_score > 50,
|
| 152 |
-
"Reviewed_By__c": None,
|
| 153 |
"Status__c": "Flagged" if fraud_score > 50 else "Cleared"
|
| 154 |
},
|
| 155 |
"Entities": {
|
|
@@ -168,14 +166,12 @@ def process_invoice(pdf_file):
|
|
| 168 |
if sf is not None:
|
| 169 |
try:
|
| 170 |
sf.Invoice_Record__c.create({
|
| 171 |
-
"
|
| 172 |
"Invoice_Amount__c": amount,
|
| 173 |
"Invoice_Date__c": str(invoice_date),
|
| 174 |
-
"Items_Listed__c": items_listed,
|
| 175 |
"Fraud_Score__c": fraud_score,
|
| 176 |
"Fraud_Reasoning__c": fraud_reasoning,
|
| 177 |
"Flagged__c": fraud_score > 50,
|
| 178 |
-
"Reviewed_By__c": None,
|
| 179 |
"Status__c": "Flagged" if fraud_score > 50 else "Cleared"
|
| 180 |
})
|
| 181 |
print("Salesforce record created successfully.")
|
|
@@ -203,4 +199,4 @@ iface = gr.Interface(
|
|
| 203 |
)
|
| 204 |
|
| 205 |
if __name__ == "__main__":
|
| 206 |
-
iface.launch()
|
|
|
|
| 93 |
df["is_anomaly"] = model.fit_predict(X_scaled)
|
| 94 |
return df
|
| 95 |
|
| 96 |
+
def calculate_fraud_score(amount, is_anomaly, text_length):
|
| 97 |
+
"""Calculate fraud score based on amount, anomaly, and text length."""
|
| 98 |
score = 0.0
|
| 99 |
reasoning = []
|
| 100 |
|
|
|
|
| 109 |
score += 30
|
| 110 |
reasoning.append("Invoice flagged as an anomaly.")
|
| 111 |
|
| 112 |
+
if text_length > 500:
|
| 113 |
score += 10
|
| 114 |
+
reasoning.append("Excessive text length in invoice.")
|
| 115 |
|
| 116 |
return min(score, 100), "; ".join(reasoning)
|
| 117 |
|
|
|
|
| 123 |
|
| 124 |
vendor_name, amount = extract_entities(text)
|
| 125 |
invoice_date = datetime.now().date()
|
| 126 |
+
text_length = len(text)
|
| 127 |
|
| 128 |
data = {
|
| 129 |
"invoice_id": str(uuid.uuid4()),
|
| 130 |
"vendor_name": vendor_name,
|
| 131 |
"amount": amount,
|
| 132 |
"invoice_date": invoice_date,
|
| 133 |
+
"text_length": text_length
|
| 134 |
}
|
| 135 |
df = pd.DataFrame([data])
|
| 136 |
|
| 137 |
df = detect_anomalies(df)
|
| 138 |
|
| 139 |
fraud_score, fraud_reasoning = calculate_fraud_score(
|
| 140 |
+
df["amount"].iloc[0], df["is_anomaly"].iloc[0], text_length
|
| 141 |
)
|
| 142 |
|
| 143 |
output = {
|
| 144 |
"Invoice_Record__c": {
|
| 145 |
+
"Name": vendor_name, # Map vendor_name to Name field
|
| 146 |
"Invoice_Amount__c": amount,
|
| 147 |
"Invoice_Date__c": str(invoice_date),
|
|
|
|
| 148 |
"Fraud_Score__c": fraud_score,
|
| 149 |
"Fraud_Reasoning__c": fraud_reasoning,
|
| 150 |
"Flagged__c": fraud_score > 50,
|
|
|
|
| 151 |
"Status__c": "Flagged" if fraud_score > 50 else "Cleared"
|
| 152 |
},
|
| 153 |
"Entities": {
|
|
|
|
| 166 |
if sf is not None:
|
| 167 |
try:
|
| 168 |
sf.Invoice_Record__c.create({
|
| 169 |
+
"Name": vendor_name, # Map vendor_name to Name field
|
| 170 |
"Invoice_Amount__c": amount,
|
| 171 |
"Invoice_Date__c": str(invoice_date),
|
|
|
|
| 172 |
"Fraud_Score__c": fraud_score,
|
| 173 |
"Fraud_Reasoning__c": fraud_reasoning,
|
| 174 |
"Flagged__c": fraud_score > 50,
|
|
|
|
| 175 |
"Status__c": "Flagged" if fraud_score > 50 else "Cleared"
|
| 176 |
})
|
| 177 |
print("Salesforce record created successfully.")
|
|
|
|
| 199 |
)
|
| 200 |
|
| 201 |
if __name__ == "__main__":
|
| 202 |
+
iface.launch()
|