sanitizer: sanitize vin number (#31090)

old-commit-hash: 2c86c023fe2d3072910ed33ff97cd6d3ddf2915d

tools/lib/logreader.py

@@ -12,7 +12,7 @@ import sys
 import urllib.parse
 import warnings
 
-from typing import Iterable, Iterator, List
+from typing import Iterable, Iterator, List, Type
 from urllib.parse import parse_qs, urlparse
 
 from cereal import log as capnp_log
@@ -21,7 +21,8 @@ from openpilot.tools.lib.filereader import FileReader, file_exists
 from openpilot.tools.lib.helpers import RE
 from openpilot.tools.lib.route import Route, SegmentRange
 
-LogIterable = Iterable[capnp._DynamicStructReader]
+LogMessage = Type[capnp._DynamicStructReader]
+LogIterable = Iterable[LogMessage]
 
 
 class _LogFileReader:

tools/lib/sanitizer.py

@@ -0,0 +1,26 @@
+# Utilities for sanitizing routes of only essential data for testing car ports and doing validation.
+
+from openpilot.tools.lib.logreader import LogIterable, LogMessage
+
+
+def sanitize_vin(vin: str):
+  # (last 6 digits of vin are serial number https://en.wikipedia.org/wiki/Vehicle_identification_number)
+  VIN_SENSITIVE = 6
+  return vin[:-VIN_SENSITIVE] + "X" * VIN_SENSITIVE
+
+
+def sanitize_msg(msg: LogMessage) -> LogMessage:
+  if msg.which() == "carParams":
+    msg = msg.as_builder()
+    msg.carParams.carVin = sanitize_vin(msg.carParams.carVin)
+    msg = msg.as_reader()
+  return msg
+
+
+PRESERVE_SERVICES = ["can", "carParams", "pandaStates", "pandaStateDEPRECATED"]
+
+
+def sanitize(lr: LogIterable) -> LogIterable:
+  filtered = filter(lambda msg: msg.which() in PRESERVE_SERVICES, lr)
+  sanitized = map(sanitize_msg, filtered)
+  return sanitized