fix: always quote control chars

None of the invisible ASCII Control chars 0 - 32 were replaced.
This PR replaces them with a unicode representation, the same as go-logfmt does it.
e.g. `\x00` → `"\u0000"`

This is security critical, as currently `\r` and other things would be part of the raw log.
It would e.g. allow via ANSI escape sequences to set information in a linux terminal, like the window name/title.

Author: spaceone

src/logfmter/formatter.py

@@ -35,6 +35,10 @@
     "threadName",
 )
 
+QUOTE_CHARS = str.maketrans("", "", " =" + "".join(chr(c) for c in range(0x20)))
+REPLACEMENTS = {chr(c): f"\\u{c:04x}" for c in range(0x20)}
+REPLACEMENTS.update({"\n": "\\n", "\t": "\\t", "\r": "\\r"})
+
 
 class _DefaultFormatter(logging.Formatter):
     def format(self, record):
@@ -59,13 +63,7 @@ def format_string(cls, value: str) -> str:
         Process the provided string with any necessary quoting and/or escaping.
         """
         needs_dquote_escaping = '"' in value
-        needs_newline_escaping = "\n" in value
-        needs_quoting = (
-            " " in value
-            or "=" in value
-            or needs_dquote_escaping
-            or needs_newline_escaping
-        )
+        needs_quoting = needs_dquote_escaping or value.translate(QUOTE_CHARS) != value
         needs_backslash_escaping = "\\" in value and needs_quoting
 
         if needs_backslash_escaping:
@@ -74,8 +72,8 @@ def format_string(cls, value: str) -> str:
         if needs_dquote_escaping:
             value = value.replace('"', '\\"')
 
-        if needs_newline_escaping:
-            value = value.replace("\n", "\\n")
+        for char, replacement in REPLACEMENTS.items():
+            value = value.replace(char, replacement)
 
         if needs_quoting:
             value = '"{}"'.format(value)

tests/test_formatter.py

@@ -16,13 +16,24 @@
     ("=", '"="'),
     # All double quotes must be escaped.
     ('"', '"\\""'),
+    # Null bytes must be escaped and quoted
+    ("\x00", r'"\u0000"'),
+    # All whitespace chars must be escaped and quoted
+    ("\n", '"\\n"'),
+    ("\r", '"\\r"'),
+    ("\t", '"\\t"'),
+    # All other control chars must be escaped and quoted
+    ("\x07", r'"\u0007"'),
+    (
+        "".join(chr(c) for c in range(0x20) if chr(c) not in "\t\n\r"),
+        r'"\u0000\u0001\u0002\u0003\u0004\u0005\u0006\u0007\u0008\u000b\u000c\u000e\u000f\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f"',
+    ),
     # If the string requires escaping and quoting, then both
     # operations should be performed.
     (' "', '" \\""'),
     # If the string is empty, then it should be left empty.
     ("", ""),
     # If the string contains a newline, then it should be escaped.
-    ("\n", '"\\n"'),
     ("\n\n", '"\\n\\n"'),
     # If the string contains a backslash and needs to be quoted, then
     # the backslashes need to be escaped.