Ordered by severity, then signals introduced in this release. Hidden characters are shown as ⟨U+…⟩.
-
line length 1260 characters (raw pass, confidence 0.50)
- `#3927 <https://github.com/aio-libs/aiohttp/issues/3927>`_, `#4247 <https://github.com/aio-libs/aiohttp/issues/4247>`_, `#4247 <https://github.com/aio-libs/aiohttp/issues/4247>`_, `#5389 <https://github.com/aio-libs/aiohttp/issues/5389>`_, `#5457 <https://github.com/aio-libs/aiohttp/issues/5457>`_, `#5486 <https://gi
-
U+FEFF ZERO WIDTH NO-BREAK SPACE (BOM) in source (raw pass, confidence 0.75)
- Rename `from_env` to `trust_env` in client reference. (`#2451 <https://github.com/aio-libs/aiohttp/pull/2451>`_)
- ⟨U+FEFF⟩Fixed mistype in `Proxy Support` section where `trust_env` parameter was
used in `session.get("http://python.org", trust_env=True)` method instead of
aiohttp.ClientSession constructor as follows:
likely benign
LLM, confidence 0.85
A byte-order mark was pasted into a changelog entry in the aiohttp docs.
Reasoning
A stray U+FEFF byte-order mark sits inside a changelog bullet in a documentation file. This is a common editing artifact with no code execution or hidden instruction, so it is a cosmetic issue rather than obfuscation.
-
line length 961 characters (raw pass, confidence 0.50)
-
`sys._getframe` — call-stack frame introspection (ast pass, confidence 0.70)
if loop.get_debug():
self._source_traceback = traceback.extract_stack(sys._getframe(1))
if cookie_jar is None:
likely benign
LLM, confidence 0.92
Debug-mode capture of the caller's stack frame for source tracebacks in the client session.
Reasoning
sys._getframe(1) is used only when loop debug mode is on, to record a source traceback for unclosed-session warnings. This is standard asyncio debugging practice in aiohttp.
-
`sys._getframe` — call-stack frame introspection (ast pass, confidence 0.70)
if loop.get_debug():
self._source_traceback = traceback.extract_stack(sys._getframe(1))
def __reset_writer(self, _: object = None) -> None:
likely benign
LLM, confidence 0.92
Debug-mode stack frame capture for request/response source tracebacks.
Reasoning
Same guarded debug-only stack capture as in client.py, used to aid diagnostics of unclosed responses or requests. No sensitive action follows.
-
`sys._getframe` — call-stack frame introspection (ast pass, confidence 0.70)
if loop.get_debug():
self._source_traceback = traceback.extract_stack(sys._getframe(1))
self.update_version(version)
likely benign
LLM, confidence 0.92
Debug-mode stack frame capture when constructing a client request object.
Reasoning
Another instance of the loop-debug-gated source traceback pattern. It only records where an object was created and does not alter behavior.
-
file imports decoder module(s) zlib and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
def decompressobj(self, *args: Any, **kwargs: Any) -> ZLibDecompressObjProtocol:
return self._zlib_backend.decompressobj(*args, **kwargs)
def compress(self, data: Buffer, *args: Any, **kwargs: Any) -> bytes:
likely benign
LLM, confidence 0.90
Compression wrapper delegating to zlib for HTTP content encoding.
Reasoning
This is a thin wrapper that forwards decompressobj and compress calls to the zlib backend for HTTP content-encoding support. Decompressing data is the library's purpose and does not indicate an embedded payload.
-
`sys._getframe` — call-stack frame introspection (ast pass, confidence 0.70)
if loop.get_debug():
self._source_traceback = traceback.extract_stack(sys._getframe(1))
def __repr__(self) -> str:
likely benign
LLM, confidence 0.92
Debug-mode stack frame capture for connector source tracebacks.
Reasoning
Loop-debug-gated source traceback capture in the connector, identical to the other debug-only frame recordings in the library.
-
`sys._getframe` — call-stack frame introspection (ast pass, confidence 0.70)
if loop.get_debug():
self._source_traceback = traceback.extract_stack(sys._getframe(1))
likely benign
LLM, confidence 0.92
Debug-mode stack frame capture in the connector constructor.
Reasoning
The same debug-only traceback capture pattern in the connector constructor. It has no effect outside debug mode.
-
`inspect.currentframe` — call-stack frame introspection (ast pass, confidence 0.70)
frame = inspect.currentframe()
while frame:
if frame.f_code.co_name == "<module>":
likely benign
LLM, confidence 0.88
Walks the call stack to determine a module-name prefix for key deduplication.
Reasoning
The code walks frames to find the module name, which is used to prefix keys and deduplicate names. It is internal bookkeeping with no network, exec, or hidden behavior.
-
file imports decoder module(s) base64, binascii and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
)
yield await d.decompress(data, max_length=self._max_decompress_size)
while d.data_available:
yield await d.decompress(b"", max_length=self._max_decompress_size)
likely benign
LLM, confidence 0.90
Streaming decompression of compressed multipart body parts with size limits.
Reasoning
This is the streaming decompression path for multipart Content-Encoding bodies, with a max_decompress_size bound. It is the expected handling for compressed HTTP payloads.
-
identifier `тестkey` contains homoglyph candidates: е (0435)→e, с (0441)→c (in string/comment) (raw pass, confidence 0.35)
fut = loop.create_future()
fut.set_result('{"тестkey": "пройденvalue"}'.encode("cp1251"))
return fut
-
identifier `тестkey` mixes scripts: Cyrillic + Latin (in string/comment) (raw pass, confidence 0.40)
fut = loop.create_future()
fut.set_result('{"тестkey": "пройденvalue"}'.encode("cp1251"))
return fut
-
identifier `пройденvalue` contains homoglyph candidates: р (0440)→p, о (043E)→o, е (0435)→e (in string/comment) (raw pass, confidence 0.35)
fut = loop.create_future()
fut.set_result('{"тестkey": "пройденvalue"}'.encode("cp1251"))
return fut
-
identifier `пройденvalue` mixes scripts: Cyrillic + Latin (in string/comment) (raw pass, confidence 0.40)
fut = loop.create_future()
fut.set_result('{"тестkey": "пройденvalue"}'.encode("cp1251"))
return fut
-
file imports decoder module(s) gzip, zlib and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
self._feeds.append(len(data))
return self._obj.decompress(data, max_length)
likely benign
LLM, confidence 0.92
Test helper wrapping a zlib decompressor for compression unit tests.
Reasoning
A test helper feeding data through a decompressor to verify chunked decompression. It is a normal unit-test fixture.
-
bytes literal contains 945 `\xNN` escapes (64% of content) — binary payload shape (ast pass, confidence 0.85)
pickled = b"\x80\x04\x95\xc8\x0b\x00\x00\x00\x00\x00\x00\x8c\x0bcollections\x94\x8c\x0bdefaultdict\x94\x93\x94\x8c\x0chttp.cookies\x94\x8c\x0cSimpleCookie\x94\x93\x94\x85\x94R\x94(\x8c\x00\x94h\x08\x86\x94h\x05)\x81\x94\x8c\rshared-cookie\x94h\x03\x8c\x06Morsel\x94\x93\x94)\x81\x94(\x8c\x07expires\x94h\x08\x8c\
likely benign
LLM, confidence 0.75
Pickled cookie-jar bytes used as a fixture for a backward-compatibility test.
Reasoning
This is a pickle byte literal of a CookieJar state, used in a test that verifies loading of legacy pickled cookie data. Pickle payloads are a genuine risk category, but the content is a serialized aiohttp/http.cookies structure in a test file, not an exploit or dropper. The escape density is the expected shape of pickle bytes.
-
bytes literal contains 685 `\xNN` escapes (66% of content) — binary payload shape (ast pass, confidence 0.85)
pickled = b'\x80\x05\x95\x06\x08\x00\x00\x00\x00\x00\x00\x8c\x0bcollections\x94\x8c\x0bdefaultdict\x94\x93\x94\x8c\x0chttp.cookies\x94\x8c\x0cSimpleCookie\x94\x93\x94\x85\x94R\x94(\x8c\x00\x94h\x08\x86\x94h\x05)\x81\x94\x8c\rshared-cookie\x94h\x03\x8c\x06Morsel\x94\x93\x94)\x81\x94(\x8c\x07expires\x94h\x08\x8c\
likely benign
LLM, confidence 0.75
A second pickled cookie-jar fixture covering another pickle protocol version.
Reasoning
A second pickled cookie-jar fixture with a different protocol header, consistent with testing compatibility across pickle versions. Its content matches the same benign cookie structure as the first fixture.
-
identifier `chunKed` mixes scripts: Latin + Other (in string/comment) (raw pass, confidence 0.40)
text = "GET /test HTTP/1.1\r\nHost: a\r\nTransfer-Encoding: chunKed\r\n\r\n"
with pytest.raises(http_exceptions.BadHttpMessage):
parser.feed_data(text.encode())
-
identifier `websocKet` mixes scripts: Latin + Other (in string/comment) (raw pass, confidence 0.40)
text = "GET /test HTTP/1.1\r\nHost: a\r\nUpgrade: websocKet\r\n\r\n"
messages, upgrade, tail = parser.feed_data(text.encode())
assert not upgrade
-
identifier `zstd` mixes scripts: Latin + Other (in string/comment) (raw pass, confidence 0.40)
(
"zstd".encode(), "deflate".encode(), ),
-
identifier `deflate` mixes scripts: Latin + Other (in string/comment) (raw pass, confidence 0.40)
"zstd".encode(), "deflate".encode(), ),
)
-
file imports decoder module(s) zlib and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
def decompress(data: bytes) -> bytes:
d = ZLibBackend.decompressobj()
return d.decompress(data)
likely benign
LLM, confidence 0.92
Test helper decompressing zlib output to verify HTTP writer compression.
Reasoning
A small test helper that decompresses zlib data to check the writer's compressed output. It is ordinary test code.
-
file imports decoder module(s) base64, gzip and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
decompressor = ZLibBackend.decompressobj(wbits=16 + ZLibBackend.MAX_WBITS)
data = decompressor.decompress(message.split(b"\r\n")[0])
data += decompressor.flush()
likely benign
LLM, confidence 0.90
Test code that decompresses gzip-wrapped multipart data using zlib's decompressobj.
Reasoning
Decompressing a test fixture message with zlib decompressobj in a multipart/compression test is normal; the input is a literal test payload split from a message, not an embedded hidden blob executed at runtime.
-
identifier `Fфикс` contains homoglyph candidates: с (0441)→c (in string/comment) (raw pass, confidence 0.35)
resource = router.add_static(
"/пре %2Fфикс", pathlib.Path(aiohttp.__file__).parent, name="static"
)
assert router["static"] is resource
-
identifier `Fфикс` mixes scripts: Latin + Cyrillic (in string/comment) (raw pass, confidence 0.40)
resource = router.add_static(
"/пре %2Fфикс", pathlib.Path(aiohttp.__file__).parent, name="static"
)
assert router["static"] is resource
-
identifier `Fфикс` contains homoglyph candidates: с (0441)→c (in string/comment) (raw pass, confidence 0.35)
handler = make_handler()
route = router.add_route("GET", r"/пре %2Fфикс/{arg}", handler)
url = route.url_for(arg="1 2/текст%2F")
-
identifier `Fфикс` mixes scripts: Latin + Cyrillic (in string/comment) (raw pass, confidence 0.40)
handler = make_handler()
route = router.add_route("GET", r"/пре %2Fфикс/{arg}", handler)
url = route.url_for(arg="1 2/текст%2F")
-
identifier `рус` contains homoglyph candidates: р (0440)→p, у (0443)→y, с (0441)→c (in string/comment) (raw pass, confidence 0.35)
match_info = await router.resolve(req)
assert {"var": "рус текст"} == match_info
-
reach-by-name through `locals` (ast pass, confidence 0.60)
assert resp.status == 200
assert await resp.text() == str(len(locals()[len_of]))
await resp.release()
dismissed
LLM, confidence 0.93
Test assertion reads a local variable via locals() to compare response length.
Reasoning
locals() lookup by name in an assertion is a trivial test idiom for reading a local variable whose name is set earlier; no execution or hidden behavior.
-
file imports decoder module(s) gzip and calls `.decompress()` — embedded payload decompressed at runtime (ast pass, confidence 0.75)
await resp.prepare(req)
assert gzip.decompress(resp._compressed_body) == body
likely benign
LLM, confidence 0.93
Test checks that a compressed response body decompresses to the expected body.
Reasoning
gzip.decompress on the response's compressed body in a test to verify round-trip content is ordinary test verification code.
-
`compile` called on a non-literal expression (ast pass, confidence 0.70)
hdrs_file = root / "aiohttp/hdrs.py"
code = compile(hdrs_file.read_text(), str(hdrs_file), "exec")
globs = {}
exec(code, globs)
likely benign
LLM, confidence 0.90
Build helper that execs the project's own hdrs.py module to extract header names.
Reasoning
The tool compiles and executes aiohttp/hdrs.py, a repository source file, to collect header constants. This is a build-time code generator reading a local project file, not attacker-controlled input, though exec of source is a pattern worth noting.
-
`exec` called on a non-literal expression (ast pass, confidence 0.70)
globs = {}
exec(code, globs)
headers = [val for val in globs.values() if isinstance(val, multidict.istr)]
return sorted(headers)
likely benign
LLM, confidence 0.90
Executes the compiled hdrs.py namespace to gather istr header values.
Reasoning
Same generator script executing the trusted local hdrs.py source; the exec is part of a legitimate code-generation step with no network or encoded payload.
Run disclude locally to verify. Add --format sarif for SARIF output. Add --llm (with an API key in the environment) to include the LLM review.
disclude.io