memorisdk 1.0.2__py3-none-any.whl → 2.0.0__py3-none-any.whl

memori/utils/security_audit.py

@@ -0,0 +1,594 @@
+"""
+Security audit service for Memori database operations
+Provides comprehensive security validation and monitoring
+"""
+
+import re
+import time
+from dataclasses import dataclass
+from enum import Enum
+from typing import Any, Dict, List, Optional, Tuple
+
+from .exceptions import SecurityError
+from .input_validator import InputValidator
+
+
+class SecurityLevel(str, Enum):
+    """Security severity levels"""
+
+    CRITICAL = "critical"
+    HIGH = "high"
+    MEDIUM = "medium"
+    LOW = "low"
+    INFO = "info"
+
+
+class VulnerabilityType(str, Enum):
+    """Types of security vulnerabilities"""
+
+    SQL_INJECTION = "sql_injection"
+    XSS = "xss"
+    COMMAND_INJECTION = "command_injection"
+    PATH_TRAVERSAL = "path_traversal"
+    PRIVILEGE_ESCALATION = "privilege_escalation"
+    DATA_EXPOSURE = "data_exposure"
+    WEAK_VALIDATION = "weak_validation"
+
+
+@dataclass
+class SecurityFinding:
+    """Represents a security finding"""
+
+    vulnerability_type: VulnerabilityType
+    severity: SecurityLevel
+    description: str
+    location: str
+    recommendation: str
+    evidence: Optional[str] = None
+    remediation_code: Optional[str] = None
+
+
+@dataclass
+class SecurityAuditReport:
+    """Security audit report"""
+
+    findings: List[SecurityFinding]
+    total_queries_audited: int
+    critical_count: int
+    high_count: int
+    medium_count: int
+    low_count: int
+    audit_timestamp: float
+    overall_risk_score: float
+
+
+class DatabaseSecurityAuditor:
+    """Comprehensive security auditor for database operations"""
+
+    # Advanced SQL injection patterns
+    SQL_INJECTION_PATTERNS = [
+        # Union-based injection
+        r"\bunion\s+select\b",
+        r"\bunion\s+all\s+select\b",
+        # Boolean-based injection
+        r"\b(and|or)\s+\d+\s*=\s*\d+",
+        r"\b(and|or)\s+.+\s*(like|=)\s*.+",
+        # Time-based injection
+        r"\bwaitfor\s+delay\b",
+        r"\bsleep\s*\(",
+        r"\bbenchmark\s*\(",
+        r"\bpg_sleep\s*\(",
+        # Comment-based injection
+        r"--\s*[^\r\n]*",
+        r"/\*.*?\*/",
+        r"\#[^\r\n]*",
+        # Stacked queries
+        r";\s*(select|insert|update|delete|drop|create|alter)",
+        # Information gathering
+        r"\binformation_schema\b",
+        r"\bsys\.|sysobjects|syscolumns",
+        r"\bmysql\.user\b",
+        r"\bpg_tables\b",
+        # Dangerous functions
+        r"\bxp_cmdshell\b",
+        r"\bsp_executesql\b",
+        r"\bexec\s*\(",
+        r"\beval\s*\(",
+        # File operations
+        r"\binto\s+outfile\b",
+        r"\binto\s+dumpfile\b",
+        r"\bload_file\s*\(",
+        r"\bcopy\s+.*\bfrom\b",
+    ]
+
+    # XSS patterns
+    XSS_PATTERNS = [
+        r"<script[^>]*>.*?</script>",
+        r"<iframe[^>]*>.*?</iframe>",
+        r"<object[^>]*>.*?</object>",
+        r"<embed[^>]*>",
+        r"<applet[^>]*>.*?</applet>",
+        r"javascript\s*:",
+        r"on\w+\s*=",
+        r"expression\s*\(",
+        r"url\s*\(",
+        r"@import",
+    ]
+
+    # Command injection patterns
+    COMMAND_INJECTION_PATTERNS = [
+        r"[;&|`$]",
+        r"\$\([^)]+\)",
+        r"`[^`]+`",
+        r"\|\s*(cat|ls|pwd|whoami|id|uname)",
+        r"(^|\s)(wget|curl|nc|netcat|telnet|ssh)\s",
+        r"\b(rm|mv|cp|chmod|chown|kill)\s",
+    ]
+
+    # Path traversal patterns
+    PATH_TRAVERSAL_PATTERNS = [
+        r"\.\./",
+        r"\.\.\\",
+        r"%2e%2e%2f",
+        r"%2e%2e\\",
+        r"..%252f",
+        r"..%255c",
+    ]
+
+    def __init__(self):
+        self.findings = []
+        self.queries_audited = 0
+        self.blocked_operations = set()
+
+    def audit_query(
+        self,
+        query: str,
+        params: Optional[List[Any]] = None,
+        context: Optional[str] = None,
+    ) -> List[SecurityFinding]:
+        """Audit a single database query for security vulnerabilities"""
+        findings = []
+        self.queries_audited += 1
+
+        # Audit SQL injection
+        sql_findings = self._audit_sql_injection(query, params, context)
+        findings.extend(sql_findings)
+
+        # Audit parameter validation
+        param_findings = self._audit_parameter_validation(query, params, context)
+        findings.extend(param_findings)
+
+        # Audit privilege operations
+        privilege_findings = self._audit_privilege_operations(query, context)
+        findings.extend(privilege_findings)
+
+        # Audit data exposure
+        exposure_findings = self._audit_data_exposure(query, context)
+        findings.extend(exposure_findings)
+
+        return findings
+
+    def _audit_sql_injection(
+        self, query: str, params: Optional[List[Any]], context: Optional[str]
+    ) -> List[SecurityFinding]:
+        """Audit for SQL injection vulnerabilities"""
+        findings = []
+        query_lower = query.lower()
+
+        # Check for dangerous patterns
+        for pattern in self.SQL_INJECTION_PATTERNS:
+            matches = re.finditer(pattern, query_lower, re.IGNORECASE | re.MULTILINE)
+            for match in matches:
+                findings.append(
+                    SecurityFinding(
+                        vulnerability_type=VulnerabilityType.SQL_INJECTION,
+                        severity=SecurityLevel.CRITICAL,
+                        description=f"Potential SQL injection pattern detected: {pattern}",
+                        location=context or "unknown",
+                        recommendation="Use parameterized queries and input validation",
+                        evidence=f"Match: '{match.group()}' at position {match.start()}",
+                        remediation_code="Use ? or %s placeholders with parameter binding",
+                    )
+                )
+
+        # Check for string concatenation in queries
+        if any(char in query for char in ["+", "||"]) and "VALUES" in query.upper():
+            findings.append(
+                SecurityFinding(
+                    vulnerability_type=VulnerabilityType.SQL_INJECTION,
+                    severity=SecurityLevel.HIGH,
+                    description="String concatenation detected in SQL query",
+                    location=context or "unknown",
+                    recommendation="Use parameterized queries instead of string concatenation",
+                    evidence="Query contains string concatenation operators",
+                )
+            )
+
+        # Check if parameters are properly used
+        if params is None and any(var in query_lower for var in ["%s", "?"]):
+            findings.append(
+                SecurityFinding(
+                    vulnerability_type=VulnerabilityType.WEAK_VALIDATION,
+                    severity=SecurityLevel.MEDIUM,
+                    description="Query has parameter placeholders but no parameters provided",
+                    location=context or "unknown",
+                    recommendation="Ensure all parameter placeholders have corresponding values",
+                )
+            )
+
+        return findings
+
+    def _audit_parameter_validation(
+        self, query: str, params: Optional[List[Any]], context: Optional[str]
+    ) -> List[SecurityFinding]:
+        """Audit parameter validation"""
+        findings = []
+
+        if params:
+            for i, param in enumerate(params):
+                # Check for potential XSS in parameters
+                if isinstance(param, str):
+                    for pattern in self.XSS_PATTERNS:
+                        if re.search(pattern, param, re.IGNORECASE):
+                            findings.append(
+                                SecurityFinding(
+                                    vulnerability_type=VulnerabilityType.XSS,
+                                    severity=SecurityLevel.HIGH,
+                                    description=f"XSS pattern in parameter {i}",
+                                    location=context or "unknown",
+                                    recommendation="Sanitize and validate input parameters",
+                                    evidence=f"Parameter value: {param[:100]}...",
+                                )
+                            )
+
+                    # Check for command injection
+                    for pattern in self.COMMAND_INJECTION_PATTERNS:
+                        if re.search(pattern, param):
+                            findings.append(
+                                SecurityFinding(
+                                    vulnerability_type=VulnerabilityType.COMMAND_INJECTION,
+                                    severity=SecurityLevel.CRITICAL,
+                                    description=f"Command injection pattern in parameter {i}",
+                                    location=context or "unknown",
+                                    recommendation="Validate and sanitize all user input",
+                                    evidence=f"Parameter value: {param[:100]}...",
+                                )
+                            )
+
+                    # Check for path traversal
+                    for pattern in self.PATH_TRAVERSAL_PATTERNS:
+                        if re.search(pattern, param):
+                            findings.append(
+                                SecurityFinding(
+                                    vulnerability_type=VulnerabilityType.PATH_TRAVERSAL,
+                                    severity=SecurityLevel.HIGH,
+                                    description=f"Path traversal pattern in parameter {i}",
+                                    location=context or "unknown",
+                                    recommendation="Validate file paths and restrict access",
+                                    evidence=f"Parameter value: {param[:100]}...",
+                                )
+                            )
+
+                # Check parameter size
+                if isinstance(param, str) and len(param) > 10000:
+                    findings.append(
+                        SecurityFinding(
+                            vulnerability_type=VulnerabilityType.DATA_EXPOSURE,
+                            severity=SecurityLevel.MEDIUM,
+                            description=f"Unusually large parameter {i} ({len(param)} chars)",
+                            location=context or "unknown",
+                            recommendation="Implement input length limits",
+                            evidence=f"Parameter length: {len(param)} characters",
+                        )
+                    )
+
+        return findings
+
+    def _audit_privilege_operations(
+        self, query: str, context: Optional[str]
+    ) -> List[SecurityFinding]:
+        """Audit for privilege escalation attempts"""
+        findings = []
+        query_upper = query.upper().strip()
+
+        # Dangerous DDL operations
+        dangerous_ddl = ["DROP", "CREATE USER", "ALTER USER", "GRANT", "REVOKE"]
+        for operation in dangerous_ddl:
+            if query_upper.startswith(operation):
+                findings.append(
+                    SecurityFinding(
+                        vulnerability_type=VulnerabilityType.PRIVILEGE_ESCALATION,
+                        severity=SecurityLevel.CRITICAL,
+                        description=f"Dangerous DDL operation: {operation}",
+                        location=context or "unknown",
+                        recommendation="Restrict DDL operations to administrative contexts",
+                        evidence=f"Query starts with: {operation}",
+                    )
+                )
+
+        # System function calls
+        system_functions = [
+            "SYSTEM",
+            "SHELL",
+            "EXEC",
+            "EXECUTE",
+            "XP_CMDSHELL",
+            "SP_EXECUTESQL",
+            "OPENROWSET",
+            "OPENDATASOURCE",
+        ]
+
+        for func in system_functions:
+            if func in query_upper:
+                findings.append(
+                    SecurityFinding(
+                        vulnerability_type=VulnerabilityType.PRIVILEGE_ESCALATION,
+                        severity=SecurityLevel.CRITICAL,
+                        description=f"System function call detected: {func}",
+                        location=context or "unknown",
+                        recommendation="Avoid system function calls in application queries",
+                        evidence=f"Function: {func}",
+                    )
+                )
+
+        return findings
+
+    def _audit_data_exposure(
+        self, query: str, context: Optional[str]
+    ) -> List[SecurityFinding]:
+        """Audit for potential data exposure issues"""
+        findings = []
+        query_upper = query.upper()
+
+        # SELECT * queries (potential over-exposure)
+        if re.search(r"SELECT\s+\*", query_upper):
+            findings.append(
+                SecurityFinding(
+                    vulnerability_type=VulnerabilityType.DATA_EXPOSURE,
+                    severity=SecurityLevel.MEDIUM,
+                    description="SELECT * query may expose unnecessary data",
+                    location=context or "unknown",
+                    recommendation="Specify exact columns needed instead of using *",
+                    evidence="Query uses SELECT *",
+                )
+            )
+
+        # Missing WHERE clauses in UPDATE/DELETE
+        if query_upper.startswith(("UPDATE", "DELETE")) and "WHERE" not in query_upper:
+            findings.append(
+                SecurityFinding(
+                    vulnerability_type=VulnerabilityType.DATA_EXPOSURE,
+                    severity=SecurityLevel.CRITICAL,
+                    description="UPDATE/DELETE without WHERE clause",
+                    location=context or "unknown",
+                    recommendation="Always include WHERE clause in UPDATE/DELETE operations",
+                    evidence="Missing WHERE clause",
+                )
+            )
+
+        # Potential sensitive data in queries
+        sensitive_keywords = [
+            "PASSWORD",
+            "TOKEN",
+            "SECRET",
+            "KEY",
+            "PRIVATE",
+            "SSN",
+            "SOCIAL_SECURITY",
+            "CREDIT_CARD",
+            "CVV",
+        ]
+
+        for keyword in sensitive_keywords:
+            if keyword in query_upper:
+                findings.append(
+                    SecurityFinding(
+                        vulnerability_type=VulnerabilityType.DATA_EXPOSURE,
+                        severity=SecurityLevel.HIGH,
+                        description=f"Query references potentially sensitive data: {keyword}",
+                        location=context or "unknown",
+                        recommendation="Ensure sensitive data is properly protected and encrypted",
+                        evidence=f"Keyword: {keyword}",
+                    )
+                )
+
+        return findings
+
+    def validate_query_safety(
+        self,
+        query: str,
+        params: Optional[List[Any]] = None,
+        context: Optional[str] = None,
+        strict_mode: bool = True,
+    ) -> Tuple[bool, List[SecurityFinding]]:
+        """Validate if a query is safe to execute"""
+        findings = self.audit_query(query, params, context)
+
+        # Determine if query should be blocked
+        has_critical = any(f.severity == SecurityLevel.CRITICAL for f in findings)
+        has_high = any(f.severity == SecurityLevel.HIGH for f in findings)
+
+        is_safe = True
+        if strict_mode:
+            is_safe = not (has_critical or has_high)
+        else:
+            is_safe = not has_critical
+
+        return is_safe, findings
+
+    def generate_audit_report(self) -> SecurityAuditReport:
+        """Generate comprehensive security audit report"""
+        critical_count = sum(
+            1 for f in self.findings if f.severity == SecurityLevel.CRITICAL
+        )
+        high_count = sum(1 for f in self.findings if f.severity == SecurityLevel.HIGH)
+        medium_count = sum(
+            1 for f in self.findings if f.severity == SecurityLevel.MEDIUM
+        )
+        low_count = sum(1 for f in self.findings if f.severity == SecurityLevel.LOW)
+
+        # Calculate risk score (0-100)
+        risk_score = min(
+            100,
+            (critical_count * 25 + high_count * 10 + medium_count * 5 + low_count * 1),
+        )
+
+        return SecurityAuditReport(
+            findings=self.findings.copy(),
+            total_queries_audited=self.queries_audited,
+            critical_count=critical_count,
+            high_count=high_count,
+            medium_count=medium_count,
+            low_count=low_count,
+            audit_timestamp=time.time(),
+            overall_risk_score=risk_score,
+        )
+
+    def get_remediation_suggestions(self) -> Dict[VulnerabilityType, List[str]]:
+        """Get remediation suggestions grouped by vulnerability type"""
+        suggestions = {
+            VulnerabilityType.SQL_INJECTION: [
+                "Use parameterized queries with ? or %s placeholders",
+                "Validate and sanitize all user inputs",
+                "Implement input length limits",
+                "Use ORM frameworks when possible",
+                "Escape special SQL characters",
+                "Implement allowlist validation for dynamic queries",
+            ],
+            VulnerabilityType.XSS: [
+                "HTML encode all user input before display",
+                "Use Content Security Policy (CSP) headers",
+                "Validate input against expected patterns",
+                "Sanitize data before database storage",
+                "Implement output encoding",
+            ],
+            VulnerabilityType.COMMAND_INJECTION: [
+                "Never pass user input directly to system commands",
+                "Use allowlist validation for system operations",
+                "Implement proper input validation",
+                "Use safe APIs instead of shell commands",
+                "Run applications with minimal privileges",
+            ],
+            VulnerabilityType.PATH_TRAVERSAL: [
+                "Validate and sanitize all file paths",
+                "Use absolute paths with validation",
+                "Implement chroot or similar containment",
+                "Restrict file system access permissions",
+                "Use safe file handling libraries",
+            ],
+            VulnerabilityType.PRIVILEGE_ESCALATION: [
+                "Implement proper access controls",
+                "Use principle of least privilege",
+                "Validate user permissions before operations",
+                "Audit administrative operations",
+                "Separate administrative and user interfaces",
+            ],
+            VulnerabilityType.DATA_EXPOSURE: [
+                "Use specific column names instead of SELECT *",
+                "Implement data classification and handling policies",
+                "Use encryption for sensitive data",
+                "Implement proper access logging",
+                "Regular security audits of data access patterns",
+            ],
+        }
+        return suggestions
+
+
+class SecureQueryBuilder:
+    """Helper class for building secure queries"""
+
+    def __init__(self, auditor: DatabaseSecurityAuditor):
+        self.auditor = auditor
+
+    def build_safe_select(
+        self,
+        table: str,
+        columns: List[str],
+        where_conditions: Dict[str, Any],
+        limit: Optional[int] = None,
+    ) -> Tuple[str, List[Any]]:
+        """Build a safe SELECT query"""
+        # Validate inputs
+        table = InputValidator.sanitize_sql_identifier(table)
+        validated_columns = [
+            InputValidator.sanitize_sql_identifier(col) for col in columns
+        ]
+
+        # Build query
+        columns_str = ", ".join(validated_columns)
+        query = f"SELECT {columns_str} FROM {table}"
+        params = []
+
+        if where_conditions:
+            where_parts = []
+            for column, value in where_conditions.items():
+                safe_column = InputValidator.sanitize_sql_identifier(column)
+                where_parts.append(f"{safe_column} = ?")
+                params.append(value)
+            query += f" WHERE {' AND '.join(where_parts)}"
+
+        if limit:
+            query += " LIMIT ?"
+            params.append(int(limit))
+
+        # Audit the query
+        is_safe, findings = self.auditor.validate_query_safety(
+            query, params, "build_safe_select"
+        )
+        if not is_safe:
+            raise SecurityError(f"Generated query failed security audit: {findings}")
+
+        return query, params
+
+    def build_safe_insert(
+        self, table: str, data: Dict[str, Any]
+    ) -> Tuple[str, List[Any]]:
+        """Build a safe INSERT query"""
+        # Validate inputs
+        table = InputValidator.sanitize_sql_identifier(table)
+        validated_data = {}
+
+        for column, value in data.items():
+            safe_column = InputValidator.sanitize_sql_identifier(column)
+            validated_data[safe_column] = value
+
+        columns = list(validated_data.keys())
+        values = list(validated_data.values())
+
+        # Build query
+        columns_str = ", ".join(columns)
+        placeholders = ", ".join(["?"] * len(values))
+        query = f"INSERT INTO {table} ({columns_str}) VALUES ({placeholders})"
+
+        # Audit the query
+        is_safe, findings = self.auditor.validate_query_safety(
+            query, values, "build_safe_insert"
+        )
+        if not is_safe:
+            raise SecurityError(f"Generated query failed security audit: {findings}")
+
+        return query, values
+
+
+# Global security auditor instance
+_global_auditor = DatabaseSecurityAuditor()
+
+
+def get_security_auditor() -> DatabaseSecurityAuditor:
+    """Get the global security auditor instance"""
+    return _global_auditor
+
+
+def audit_query(
+    query: str, params: Optional[List[Any]] = None, context: Optional[str] = None
+) -> List[SecurityFinding]:
+    """Convenience function to audit a single query"""
+    return _global_auditor.audit_query(query, params, context)
+
+
+def validate_query_safety(
+    query: str, params: Optional[List[Any]] = None, context: Optional[str] = None
+) -> bool:
+    """Convenience function to validate query safety"""
+    is_safe, _ = _global_auditor.validate_query_safety(query, params, context)
+    return is_safe