feat: Add SVNR redaction from output

.github/workflows/push-to-hf-space.yml

@@ -2,7 +2,7 @@ name: Push to Hugging Face Space
 
 on:
   push:
-    branches: [ master, feature/env-secrets ]   # change if you use another default branch
+    branches: [ master ]   # change if you use another default branch
   workflow_dispatch:
 
 jobs:

app.py

@@ -77,6 +77,7 @@ def query_rag_pipeline(user_query: str, model: RAGModel, output_guardRails: Outp
     # 4. Output Guardrails    
     if output_guardrails_active: 
         gr_result = output_guardRails.check(user_query, response, retrieved_docs)
+        response = output_guardRails.redact_svnrs(response)
 
     else: 
         end_time = time.time()

database/setup_db.py

@@ -1,5 +1,33 @@
 import sqlite3
 from faker import Faker
+import random
+
+def generate_svnr():
+    """
+    Generates a valid Austrian social security number (SVNR).
+    """
+    weights = [3, 7, 9, 0, 5, 8, 4, 2, 1, 6]
+    while True:
+        # Generate a 9-digit random number for the main part, ensuring the first digit is not 0.
+        main_part = str(random.randint(1, 9)) + "".join([str(random.randint(0, 9)) for _ in range(8)])
+        
+        # Calculate the checksum from the main part
+        partial_sum = (int(main_part[0]) * weights[0] +
+                       int(main_part[1]) * weights[1] +
+                       int(main_part[2]) * weights[2] +
+                       int(main_part[3]) * weights[4] +
+                       int(main_part[4]) * weights[5] +
+                       int(main_part[5]) * weights[6] +
+                       int(main_part[6]) * weights[7] +
+                       int(main_part[7]) * weights[8] +
+                       int(main_part[8]) * weights[9])
+        
+        checksum = partial_sum % 11
+        
+        # The checksum must be a single digit. If it's 10, the number is invalid, so we regenerate.
+        if checksum < 10:
+            svnr = main_part[:3] + str(checksum) + main_part[3:]
+            return svnr
 
 def setup_database():
     """
@@ -13,7 +41,8 @@ def setup_database():
     CREATE TABLE IF NOT EXISTS students (
         id INTEGER PRIMARY KEY,
         name TEXT NOT NULL,
-        email TEXT NOT NULL UNIQUE
+        email TEXT NOT NULL UNIQUE,
+        svnr TEXT NOT NULL UNIQUE
     )
     ''')
 
@@ -51,7 +80,8 @@ def setup_database():
     # Add students
     for _ in range(500):
         try:
-            cursor.execute("INSERT INTO students (name, email) VALUES (?, ?)", (fake.name(), fake.email()))
+            cursor.execute("INSERT INTO students (name, email, svnr) VALUES (?, ?, ?)", 
+                           (fake.name(), fake.email(), generate_svnr()))
         except sqlite3.IntegrityError:
             pass
 

guards/svnr.py

@@ -0,0 +1,27 @@
+import re
+
+WEIGHTS = [3, 7, 9, 0, 5, 8, 4, 2, 1, 6]
+SVNR_REGEX = re.compile(r"^[0-9]{10}$")
+
+def is_valid_svnr(svnr: str) -> bool:
+    """
+    Validates an Austrian social security number (SVNR) based on its checksum.
+    """
+    if not SVNR_REGEX.match(svnr) or svnr.startswith('0'):
+        return False
+
+    checksum = calculate_checksum(svnr)
+    check_digit = int(svnr[3])
+
+    return checksum == check_digit
+
+def calculate_checksum(svnr: str) -> int:
+    """
+    Calculates the checksum for a 10-digit SVNR string.
+    The checksum is the modulo 11 of the sum of each digit multiplied by its corresponding weight.
+    """
+    digits = [int(d) for d in svnr]
+    
+    weighted_sum = sum(digit * weight for digit, weight in zip(digits, WEIGHTS))
+    
+    return weighted_sum % 11

rails/output.py

@@ -7,6 +7,7 @@ from typing import Dict, List
 from dataclasses import dataclass
 from enum import Enum
 import re
+from guards.svnr import is_valid_svnr
 
 @dataclass
 class GuardrailResult:
@@ -24,6 +25,20 @@ class GuardrailType(Enum):
 
 class OutputGuardrails:
     """Guardrails for LLM output validation"""
+
+    def redact_svnrs(self, text: str) -> str:
+        """
+        Finds and redacts valid Austrian social security numbers (SVNRs) in the text.
+        """
+        # Find all 10-digit numbers that could be SVNRs
+        potential_svnrs = re.findall(r'\b\d{10}\b', text)
+        
+        # Validate each potential SVNR and redact if valid
+        for svnr in potential_svnrs:
+            if is_valid_svnr(svnr):
+                text = text.replace(svnr, "[REDACTED SVNR]")
+        
+        return text
     
     def __init__(self):