Extract sanitization logic into an extensions + add tests (#827)

Co-authored-by: jack <212554440+jackjackbits@users.noreply.github.com>

localPackages/BitLogger/Package.swift

@@ -18,6 +18,10 @@ let package = Package(
         .target(
             name: "BitLogger",
             path: "Sources"
+        ),
+        .testTarget(
+            name: "BitLoggerTests",
+            dependencies: ["BitLogger"]
         )
     ]
 )

localPackages/BitLogger/Sources/SecureLogger.swift

@@ -68,22 +68,6 @@ public final class SecureLogger {
         return formatter
     }()
     
-    // MARK: - Cached Regex Patterns
-    
-    private static let fingerprintPattern = #/[a-fA-F0-9]{64}/#
-    private static let base64Pattern = #/[A-Za-z0-9+/]{40,}={0,2}/#
-    private static let passwordPattern = #/password["\s:=]+["']?[^"'\s]+["']?/#
-    private static let peerIDPattern = #/peerID: ([a-zA-Z0-9]{8})[a-zA-Z0-9]+/#
-    
-    // MARK: - Sanitization Cache
-    
-    private static let sanitizationCache: NSCache<NSString, NSString> = {
-        let cache = NSCache<NSString, NSString>()
-        cache.countLimit = 100 // Keep last 100 sanitized strings
-        return cache
-    }()
-    private static let cacheQueue = DispatchQueue(label: "chat.bitchat.securelogger.cache", attributes: .concurrent)
-    
     // MARK: - Log Levels
     
     enum LogLevel {
@@ -161,8 +145,8 @@ public extension SecureLogger {
     static func error(_ error: Error, context: @autoclosure () -> String, category: OSLog = .noise,
                       file: String = #file, line: Int = #line, function: String = #function) {
         let location = formatLocation(file: file, line: line, function: function)
-        let sanitized = sanitize(context())
+        let sanitized = context().sanitized()
-        let errorDesc = sanitize(error.localizedDescription)
+        let errorDesc = error.localizedDescription.sanitized()
         
         #if DEBUG
         os_log("%{public}@ Error in %{public}@: %{public}@", log: category, type: .error, location, sanitized, errorDesc)
@@ -186,15 +170,15 @@ public extension SecureLogger {
         var message: String {
             switch self {
             case .handshakeStarted(let peerID):
-                return "Handshake started with peer: \(sanitize(peerID))"
+                return "Handshake started with peer: \(peerID.sanitized())"
             case .handshakeCompleted(let peerID):
-                return "Handshake completed with peer: \(sanitize(peerID))"
+                return "Handshake completed with peer: \(peerID.sanitized())"
             case .handshakeFailed(let peerID, let error):
-                return "Handshake failed with peer: \(sanitize(peerID)), error: \(error)"
+                return "Handshake failed with peer: \(peerID.sanitized()), error: \(error)"
             case .sessionExpired(let peerID):
-                return "Session expired for peer: \(sanitize(peerID))"
+                return "Session expired for peer: \(peerID.sanitized())"
             case .authenticationFailed(let peerID):
-                return "Authentication failed for peer: \(sanitize(peerID))"
+                return "Authentication failed for peer: \(peerID.sanitized())"
             }
         }
     }
@@ -249,7 +233,7 @@ private extension SecureLogger {
                     file: String, line: Int, function: String) {
         guard shouldLog(level) else { return }
         let location = formatLocation(file: file, line: line, function: function)
-        let sanitized = sanitize("\(location) \(message())")
+        let sanitized = "\(location) \(message())".sanitized()
         
         #if DEBUG
         os_log("%{public}@", log: category, type: level.osLogType, sanitized)
@@ -282,58 +266,6 @@ private extension SecureLogger {
         let timestamp = timestampFormatter.string(from: Date())
         return "[\(timestamp)] [\(fileName):\(line) \(function)]"
     }
-    
-    /// Sanitize strings to remove potentially sensitive data
-    static func sanitize(_ input: String) -> String {
-        let key = input as NSString
-        
-        // Check cache first
-        var cachedValue: String?
-        cacheQueue.sync {
-            cachedValue = sanitizationCache.object(forKey: key) as String?
-        }
-        
-        if let cached = cachedValue {
-            return cached
-        }
-        
-        // Perform sanitization
-        var sanitized = input
-        
-        // Remove full fingerprints (keep first 8 chars for debugging)
-        sanitized = sanitized.replacing(fingerprintPattern) { match in
-            let fingerprint = String(match.output)
-            return String(fingerprint.prefix(8)) + "..."
-        }
-        
-        // Remove base64 encoded data that might be keys
-        sanitized = sanitized.replacing(base64Pattern) { _ in
-            "<base64-data>"
-        }
-        
-        // Remove potential passwords (assuming they're in quotes or after "password:")
-        sanitized = sanitized.replacing(passwordPattern) { _ in
-            "password: <redacted>"
-        }
-        
-        // Truncate peer IDs to first 8 characters
-        sanitized = sanitized.replacing(peerIDPattern) { match in
-            "peerID: \(match.1)..."
-        }
-        
-        // Cache the result
-        cacheQueue.sync {
-            sanitizationCache.setObject(sanitized as NSString, forKey: key)
-        }
-        
-        return sanitized
-    }
-    
-    /// Sanitize individual values
-    static func sanitize<T>(_ value: T) -> String {
-        let stringValue = String(describing: value)
-        return sanitize(stringValue)
-    }
 }
 
 // MARK: - Migration Helper

localPackages/BitLogger/Sources/String+Sanitization.swift

@@ -0,0 +1,67 @@
+//
+// String+Sanitization.swift
+// BitLogger
+//
+// This is free and unencumbered software released into the public domain.
+// For more information, see <https://unlicense.org>
+//
+
+import Foundation
+
+extension String {
+    /// Sanitize strings to remove potentially sensitive data
+    func sanitized() -> String {
+        let key = self as NSString
+        
+        // Check cache first
+        if let cached = Self.queue.sync(execute: { Self.cache.object(forKey: key) }) {
+            return cached as String
+        }
+        
+        var sanitized = self
+        
+        // Remove full fingerprints (keep first 8 chars for debugging)
+        let fingerprintPattern = #/[a-fA-F0-9]{64}/#
+        sanitized = sanitized.replacing(fingerprintPattern) { match in
+            let fingerprint = String(match.output)
+            return String(fingerprint.prefix(8)) + "..."
+        }
+        
+        // Remove base64 encoded data that might be keys
+        let base64Pattern = #/[A-Za-z0-9+/]{40,}={0,2}/#
+        sanitized = sanitized.replacing(base64Pattern) { _ in
+            "<base64-data>"
+        }
+        
+        // Remove potential passwords (assuming they're in quotes or after "password:")
+        let passwordPattern = #/password["\s:=]+["']?[^"'\s]+["']?/#
+        sanitized = sanitized.replacing(passwordPattern) { _ in
+            "password: <redacted>"
+        }
+        
+        // Truncate peer IDs to first 8 characters
+        let peerIDPattern = #/peerID: ([a-zA-Z0-9]{8})[a-zA-Z0-9]+/#
+        sanitized = sanitized.replacing(peerIDPattern) { match in
+            "peerID: \(match.1)..."
+        }
+        
+        // Cache the result
+        Self.queue.sync {
+            Self.cache.setObject(sanitized as NSString, forKey: key)
+        }
+        
+        return sanitized
+    }
+}
+
+// MARK: - Cache Helpers
+
+private extension String {
+    static let queue = DispatchQueue(label: "chat.bitchat.securelogger.cache", attributes: .concurrent)
+
+    static let cache: NSCache<NSString, NSString> = {
+        let cache = NSCache<NSString, NSString>()
+        cache.countLimit = 100 // Keep last 100 sanitized strings
+        return cache
+    }()
+}

localPackages/BitLogger/Tests/StringSanitizationTests.swift

@@ -0,0 +1,143 @@
+//
+//  StringSanitizationTests.swift
+//  BitLogger
+//
+//  Created by Islam on 19/10/2025.
+//
+
+import Testing
+@testable import BitLogger
+
+struct StringSanitizationTests {
+    
+    @Test("64-hex fingerprint is truncated to first 8 chars followed by ellipsis")
+    func fingerprintTruncation() async throws {
+        let fingerprint = "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef"
+        #expect(fingerprint.count == 64)
+        
+        let input = "fingerprint=\(fingerprint)"
+        let output = input.sanitized()
+        
+        #expect(output.contains("fingerprint=01234567..."))
+        // Ensure no full fingerprint remains
+        #expect(output.contains(fingerprint) == false)
+    }
+    
+    @Test("Multiple fingerprints in a string are all truncated")
+    func multipleFingerprintTruncation() async throws {
+        let fp1 = String(repeating: "a", count: 64)
+        let fp2 = String(repeating: "b", count: 64)
+        let input = "fp1=\(fp1) fp2=\(fp2)"
+        let output = input.sanitized()
+        #expect(output.contains("fp1=aaaaaaaa..."))
+        #expect(output.contains("fp2=bbbbbbbb..."))
+        #expect(output.contains(fp1) == false)
+        #expect(output.contains(fp2) == false)
+    }
+    
+    @Test("Base64-like long data is replaced with <base64-data>")
+    func base64Replacement() async throws {
+        // 44+ chars of base64 characters
+        let base64ish = "QUJDREVGR0hJSktMTU5PUFFSU1RVVldYWVo5ODc2NTQzMjE="
+        let input = "payload=\(base64ish)"
+        let output = input.sanitized()
+        #expect(output == "payload=<base64-data>")
+    }
+    
+    @Test("Base64-like without padding is replaced with <base64-data>")
+    func base64NoPaddingReplacement() async throws {
+        let base64ish = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"
+        #expect(base64ish.count >= 40)
+        let input = "b64:\(base64ish)"
+        let output = input.sanitized()
+        #expect(output == "b64:<base64-data>")
+    }
+    
+    @Test("Short base64-like strings (below threshold) are not replaced")
+    func shortBase64NotReplaced() async throws {
+        let short = "QUJDREVGR0hJSktMTU5P" // < 40 chars
+        let input = "payload=\(short)"
+        let output = input.sanitized()
+        #expect(output == input)
+    }
+    
+    @Test("Password redaction for key:value formats", arguments: [
+        "password: secret123",
+        "password=secret123",
+        "password = secret123",
+        "password: 'secret123'",
+        "password:\"secret123\"",
+        "password='secret123'"
+    ])
+    func passwordRedactionKeyValue(password: String) async throws {
+        #expect(password.sanitized() == "password: <redacted>")
+    }
+    
+    @Test("Password redaction inside wider messages")
+    func passwordRedactionInContext() async throws {
+        let input = "user=john password: 'p@ssW0rd' attempt=1"
+        let output = input.sanitized()
+        #expect(output == "user=john password: <redacted> attempt=1")
+    }
+    
+    @Test("PeerID is truncated to first 8 chars followed by ellipsis")
+    func peerIDTruncation() async throws {
+        let peer = "ABCDEF12GHIJKL34"
+        let input = "peerID: \(peer)"
+        let output = input.sanitized()
+        #expect(output == "peerID: ABCDEF12...")
+    }
+    
+    @Test("PeerID not truncated when exactly 8 chars")
+    func peerIDExactlyEightNotTruncated() async throws {
+        let peer = "ABCDEF12"
+        let input = "peerID: \(peer)"
+        let output = input.sanitized()
+        // Pattern only matches when there are more than 8 trailing chars, so unchanged
+        #expect(output == input)
+    }
+    
+    @Test("Non-matching content remains unchanged")
+    func nonMatchingUnchanged() async throws {
+        let input = "Hello world 123 - nothing sensitive here."
+        let output = input.sanitized()
+        #expect(output == input)
+    }
+    
+    @Test("Idempotency: sanitizing twice yields same result")
+    func idempotentSanitization() async throws {
+        let input = """
+        fingerprint=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef \
+        password: "superSecret" \
+        peerID: ZYXWVUT987654321 \
+        payload=ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/
+        """
+        let once = input.sanitized()
+        let twice = once.sanitized()
+        #expect(once == twice)
+    }
+    
+    @Test("Mixed content: all rules apply in a single string")
+    func mixedContent() async throws {
+        let fingerprint = "fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210"
+        let base64ish = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"
+        let peer = "PEERID01EXTRA"
+        let input = "fp=\(fingerprint) password='x' peerID: \(peer) data=\(base64ish)"
+        let output = input.sanitized()
+        #expect(output.contains("fp=fedcba98..."))
+        #expect(output.contains("password: <redacted>"))
+        #expect(output.contains("peerID: PEERID01..."))
+        #expect(output.contains("data=<base64-data>"))
+        #expect(output.contains(fingerprint) == false)
+        #expect(output.contains(base64ish) == false)
+    }
+    
+    @Test("Cache returns consistent result for repeated inputs")
+    func cacheHitConsistency() async throws {
+        let input = "password: hunter2"
+        let first = input.sanitized()
+        let second = input.sanitized()
+        #expect(first == "password: <redacted>")
+        #expect(first == second)
+    }
+}