bug-1764633: remove non-printable/non-ascii characters from signature (…

…#6782)

The crash signature contains data from stackwalker output as well as
JavaException annotations and other places, too. Because of this, it's
possible for the signature to contain non-ascii and non-printable
characters. This rule restricts signatures to printable ASCII
characters.

socorro/signature/generator.py

@@ -18,6 +18,7 @@
     SignatureIPCMessageName,
     SignatureRunWatchDog,
     SignatureShutdownTimeout,
+    SigPrintableCharsOnly,
     SigTruncate,
     StackOverflowSignature,
     StackwalkerErrorSignatureRule,
@@ -37,6 +38,7 @@
     StackOverflowSignature,
     HungProcess,
     # NOTE(willkg): These should always come last and in this order
+    SigPrintableCharsOnly,
     SigFixWhitespace,
     SigTruncate,
 ]

socorro/signature/rules.py

@@ -918,6 +918,20 @@ def action(self, crash_data, result):
         return True
 
 
+class SigPrintableCharsOnly(Rule):
+    """Remove non-printable characters from signature."""
+
+    def action(self, crash_data, result):
+        original_sig = result.signature
+        sig = "".join(
+            [c for c in original_sig.strip() if c.isascii() and c.isprintable()]
+        )
+        if sig != original_sig:
+            result.set_signature(self.name, sig)
+            result.info(self.name, "unprintable characters removed")
+        return True
+
+
 class SigFixWhitespace(Rule):
     """Fix whitespace in signatures.
 

socorro/signature/tests/test_rules.py

@@ -1718,6 +1718,30 @@ def test_action_non_ascii_abort_message(self):
         assert result.signature == "Abort | unknown | hello"
 
 
+class TestSigPrintableCharsOnly:
+    @pytest.mark.parametrize(
+        "signature, expected",
+        [
+            ("everything | fine", "everything | fine"),
+            # Non-printable null character
+            ("libxul.so\x00 | frame2", "libxul.so | frame2"),
+            # Non-ascii emoji
+            ("libxul.so\U0001f600 | frame2", "libxul.so | frame2"),
+        ],
+    )
+    def test_whitespace_fixing(self, signature, expected):
+        rule = rules.SigPrintableCharsOnly()
+        result = generator.Result()
+        result.signature = signature
+        action_result = rule.action({}, result)
+        assert action_result is True
+        assert result.signature == expected
+        if signature != expected:
+            assert result.notes == [
+                "SigPrintableCharsOnly: unprintable characters removed"
+            ]
+
+
 class TestSigFixWhitespace:
     @pytest.mark.parametrize(
         "signature, expected",