Merge pull request #393 from aclist/feat/redaction-filter

feat: redaction filter

dzgui/app_init.py

@@ -4,7 +4,7 @@ import shutil
 from typing import TYPE_CHECKING
 
 
-from dzgui.const.constants import APP_NAME, APP_NAME_LOWER
+from dzgui.const.constants import APP_NAME
 from dzgui.const.enum import Preferences
 from dzgui.config.query import lookup
 from dzgui.config.userprefs import UserPrefs
@@ -21,6 +21,7 @@ from dzgui.strings import boot
 # from dzgui.util.map_count import get_map_count
 from dzgui.util.deck import is_steam_deck, is_game_mode
 from dzgui.util.localize import set_locale
+from dzgui.util.redact import RedactionFilter, REDACTION_PATTERNS
 from dzgui.util.strings import init
 
 from dzgui.views.base import App
@@ -40,13 +41,19 @@ def make_parents(path: "Path") -> None:
 
 
 def setup_logger(log_path: "Path") -> None:
+    # TODO: put in consts?
     _format = (
         "%(asctime)s␞%(levelname)s␞%(filename)s::%(funcName)s::%(lineno)s␞%(message)s"
     )
+
     fh = logging.FileHandler(log_path)
     formatter = logging.Formatter(_format)
     fh.setFormatter(formatter)
     fh.setLevel(logging.DEBUG)
+
+    _filter = RedactionFilter(patterns=REDACTION_PATTERNS)
+    fh.addFilter(_filter)
+
     logger.setLevel(logging.DEBUG)
     logger.addHandler(fh)
 

dzgui/model/model_factory.py

@@ -2,7 +2,6 @@ from dataclasses import dataclass
 from typing import Any, Sequence, TYPE_CHECKING
 
 from dzgui.const.enum import HELP_MENU_ROWS
-from dzgui.util.redact import redact_log
 from dzgui.util.strings import delimiter
 
 import gi
@@ -134,9 +133,7 @@ class ModelFactory:
         with open(path, "r") as f:
             lines = [line.split(delimiter) for line in f.read().splitlines()]
             for record in lines:
-                # NOTE: strips PII and API keys
+                store.append(record)
-                clean = redact_log(record)
-                store.append(clean)
         return store
 
     def new_model_from_class(self, cls: type) -> FastInsertListStore:

dzgui/util/diag.py

@@ -1,3 +1,4 @@
+import logging
 import os
 import platform
 
@@ -9,7 +10,9 @@ from dzgui.const.constants import APP_NAME
 from dzgui.const.enum import Preferences
 from dzgui.config.query import lookup
 from dzgui.init.prefix import get_version
-from dzgui.util.redact import redact
+from dzgui.util.redact import redact_home
+
+logger = logging.getLogger(APP_NAME)
 
 
 def get_cpu_model() -> str:
@@ -41,12 +44,16 @@ def write_diagnostic(config: Path, outfile: Path) -> None:
     # TODO: test availability on other distros
     date = datetime.now().isoformat()
 
-    distro = platform.freedesktop_os_release()["ID_LIKE"]
+    try:
+        distro = platform.freedesktop_os_release()["ID_LIKE"]
+    except Exception as e:
+        logger.warn(e)
+        distro = "Unknown"
+
     kernel = os.uname().release
     cpu = get_cpu_model()
     version = get_version()
 
-    branch = lookup(config, Preferences.BRANCH)
     debug = lookup(config, Preferences.DEBUG)
     install = lookup(config, Preferences.INSTALL)
     default = lookup(config, Preferences.DEFAULT)
@@ -54,8 +61,8 @@ def write_diagnostic(config: Path, outfile: Path) -> None:
     steam_path = Path(default)
     workshop_path = get_local_mod_path(steam_path)
 
-    steam_redacted = redact(default)
+    steam_redacted = redact_home(default)
-    workshop_redacted = redact(str(workshop_path))
+    workshop_redacted = redact_home(str(workshop_path))
 
     mods = get_local_mod_ids(steam_path)
     mods_pretty = print_mods(mods)
@@ -65,7 +72,7 @@ def write_diagnostic(config: Path, outfile: Path) -> None:
 
     # FIXME: extraneous newlines in lists of mods
     template = f"""\
-    {APP_NAME} version {version} ({branch})
+    {APP_NAME} version {version}
     Date: {date}
     ===============================
     Distribution: {distro}

dzgui/util/redact.py

@@ -1,20 +1,29 @@
+import logging
 import re
+from typing import Literal
 
-def redact(text: str) -> str:
+api_filter = r"(.*&key=)([^&]*)(.*)"
-    r = r"(/home/)([^/])*"
+home_filter = r"(/home/)([^\s'\/]*)(.*)"
-    cleaned = re.sub(r, r"/home/REDACTED", text)
+REDACTED = r"\1REDACTED\3"
+REDACTION_PATTERNS = [api_filter, home_filter]
+
+
+def redact_home(text: str) -> str:
+    pat = re.compile(home_filter)
+    cleaned = pat.sub(REDACTED, text)
     return cleaned
 
-def redact_log(record: list) -> list[str]:
+
-    """
+class RedactionFilter(logging.Filter):
-    requests library includes Steam API key in URL params
+    def __init__(self, patterns: list[str] | None = None) -> None:
-    """
+        super().__init__()
-    clean = []
+        self._patterns = [re.compile(pat) for pat in (patterns or [])]
-    for item in record:
+
-        if "&key=" in item:
+    def filter(self, record: logging.LogRecord) -> Literal[True]:
-            pat = r"(.*&key=)(\S+)(.*)"
+        for pattern in self._patterns:
-            scrubbed = re.sub(pat, r"\1REDACTED\3", item)
+            try:
-            clean.append(scrubbed)
+                record.msg = pattern.sub(REDACTED, record.msg)
-        else:
+            except TypeError:
-            clean.append(item)
+                exception_text = f"{type(record.msg).__name__}: {record.msg}"
-    return clean
+                record.msg = pattern.sub(REDACTED, exception_text)
+        return True

dzgui/views/pages/options.py

@@ -288,6 +288,7 @@ class Options(Gtk.Box):
 
     def _on_api_change_failed(self, emitter: "Emitter") -> None:
         self.old_entry.set_text(self.old_text)
+        # TODO: use popover
         dialog = ExceptionDialog(self.controller, errors.api_validation_error)
         dialog.run()
 

pyproject.toml

@@ -110,6 +110,7 @@ markers = [
     "mods: tests mod metadata/link creation",
     "pefile: validate PE files",
     "post_install: requires a completed installation",
+    "redact: log redaction mechanisms",
     "slow: long-running tests",
     "webtest: checks remote endpoints"
 ]

tests/test_bare_conf_files.py

@@ -4,7 +4,6 @@ import os
 
 from dzgui.app_init import copy_bare_configs
 from dzgui.config.xdg import get_xdg_paths, parse_filepaths
-from pathlib import Path
 
 
 CONF_STRING = "DZGUI_CONF\n"
@@ -27,8 +26,7 @@ def state_files():
 
 
 @pytest.fixture
-def xdg_paths():
+def xdg_paths(monkeypatch):
-    paths = []
     routes = {
         "XDG_CONFIG_HOME": "",
         "XDG_STATE_HOME": "",
@@ -39,7 +37,7 @@ def xdg_paths():
         tmp = tempfile.TemporaryDirectory(delete=False)
         routes[route] = tmp.name
     for k, v in routes.items():
-        os.environ[k] = v
+        monkeypatch.setenv(k, v)
     env = get_xdg_paths()
     return parse_filepaths(env)
 

tests/test_config.py

@@ -95,8 +95,3 @@ def test_missing_values(unset_values):
     j = convert.rc2json(unset_values)
     j = json.loads(j)
     assert j.keys() == config_boilerplate.keys()
-
-
-
-
-# TODO: test that when a config file is created from scratch, it contains all values

tests/test_log_redaction.py

@@ -0,0 +1,54 @@
+import logging
+import pytest
+
+from dzgui.util.redact import RedactionFilter, REDACTION_PATTERNS
+
+
+class RecordsListHandler(logging.Handler):
+    def __init__(self) -> None:
+        super().__init__()
+        self.records_list = []
+
+    def emit(self, record: logging.LogRecord) -> None:
+        self.records_list.append(record)
+
+    def pop(self) -> None:
+        return self.records_list[-1].msg
+
+
+@pytest.mark.redact
+@pytest.mark.parametrize(
+    "log_error, expect",
+    [
+        ("/home/SENSITIVE_USERNAME/subdir", "/home/REDACTED/subdir"),
+        (
+            "https://url.com/?api&key=SENSITIVE_KEY&results=10",
+            "https://url.com/?api&key=REDACTED&results=10",
+        ),
+        (
+            "https://url.com/?api&key=SENSITIVE_KEY",
+            "https://url.com/?api&key=REDACTED",
+        ),
+        (
+            "Error in directory: '/home/SENSITIVE_USERNAME/'",
+            "Error in directory: '/home/REDACTED/'",
+        ),
+        (
+            "Error in directory: '/home/SENSITIVE_USERNAME'",
+            "Error in directory: '/home/REDACTED'",
+        ),
+    ],
+)
+def test_log_redaction(log_error: str, expect: str) -> None:
+    logger = logging.getLogger("TEST")
+
+    handler = RecordsListHandler()
+    logger.addHandler(handler)
+    logger.setLevel(logging.DEBUG)
+
+    _filter = RedactionFilter(patterns=REDACTION_PATTERNS)
+    logger.addFilter(_filter)
+
+    logger.critical(log_error)
+    redacted = handler.pop()
+    assert expect == redacted

tests/test_remote_api.py

@@ -20,7 +20,6 @@ def test_steam(config):
     key = config["steam_api"]
     assert probe.test_steam_api(key)
 
-
 def test_bm(config):
     key = config["bm_api"]
     assert probe.test_bm_api(key)