Ankur Tyagi 8331a444fd
python3-aiohttp: patch CVE-2025-53643
Details: https://nvd.nist.gov/vuln/detail/CVE-2025-53643

Dropped changes to the test and changelog from the original commit.

Signed-off-by: Ankur Tyagi <ankur.tyagi85@gmail.com>
Signed-off-by: Anuj Mittal <anuj.mittal@oss.qualcomm.com>
2026-01-26 10:04:48 +05:30

193 lines
8.4 KiB
Diff

From 2b45c0cc5f94a4aab25e80580db73c5da1152030 Mon Sep 17 00:00:00 2001
From: Sam Bull <git@sambull.org>
Date: Wed, 9 Jul 2025 19:55:22 +0100
Subject: [PATCH] Add trailer parsing logic (#11269) (#11287)
CVE: CVE-2025-53643
Upstream-Status: Backport [https://github.com/aio-libs/aiohttp/commit/e8d774f635dc6d1cd3174d0e38891da5de0e2b6a]
Dropped changes to the test and changelog from the original commit.
Signed-off-by: Ankur Tyagi <ankur.tyagi85@gmail.com>
---
aiohttp/http_parser.py | 70 ++++++++++++++++++++++--------------------
aiohttp/multipart.py | 2 +-
2 files changed, 38 insertions(+), 34 deletions(-)
diff --git a/aiohttp/http_parser.py b/aiohttp/http_parser.py
index 7a552458e..0a80c5c6d 100644
--- a/aiohttp/http_parser.py
+++ b/aiohttp/http_parser.py
@@ -142,8 +142,8 @@ class HeadersParser:
# note: "raw" does not mean inclusion of OWS before/after the field value
raw_headers = []
- lines_idx = 1
- line = lines[1]
+ lines_idx = 0
+ line = lines[lines_idx]
line_count = len(lines)
while line:
@@ -397,6 +397,7 @@ class HttpParser(abc.ABC, Generic[_MsgT]):
response_with_body=self.response_with_body,
auto_decompress=self._auto_decompress,
lax=self.lax,
+ headers_parser=self._headers_parser,
)
if not payload_parser.done:
self._payload_parser = payload_parser
@@ -416,6 +417,7 @@ class HttpParser(abc.ABC, Generic[_MsgT]):
readall=True,
auto_decompress=self._auto_decompress,
lax=self.lax,
+ headers_parser=self._headers_parser,
)
elif not empty_body and length is None and self.read_until_eof:
payload = StreamReader(
@@ -435,6 +437,7 @@ class HttpParser(abc.ABC, Generic[_MsgT]):
response_with_body=self.response_with_body,
auto_decompress=self._auto_decompress,
lax=self.lax,
+ headers_parser=self._headers_parser,
)
if not payload_parser.done:
self._payload_parser = payload_parser
@@ -471,6 +474,10 @@ class HttpParser(abc.ABC, Generic[_MsgT]):
eof = True
data = b""
+ if isinstance(
+ underlying_exc, (InvalidHeader, TransferEncodingError)
+ ):
+ raise
if eof:
start_pos = 0
@@ -635,7 +642,7 @@ class HttpRequestParser(HttpParser[RawRequestMessage]):
compression,
upgrade,
chunked,
- ) = self.parse_headers(lines)
+ ) = self.parse_headers(lines[1:])
if close is None: # then the headers weren't set in the request
if version_o <= HttpVersion10: # HTTP 1.0 must asks to not close
@@ -715,7 +722,7 @@ class HttpResponseParser(HttpParser[RawResponseMessage]):
compression,
upgrade,
chunked,
- ) = self.parse_headers(lines)
+ ) = self.parse_headers(lines[1:])
if close is None:
if version_o <= HttpVersion10:
@@ -755,6 +762,8 @@ class HttpPayloadParser:
response_with_body: bool = True,
auto_decompress: bool = True,
lax: bool = False,
+ *,
+ headers_parser: HeadersParser,
) -> None:
self._length = 0
self._type = ParseState.PARSE_NONE
@@ -763,6 +772,8 @@ class HttpPayloadParser:
self._chunk_tail = b""
self._auto_decompress = auto_decompress
self._lax = lax
+ self._headers_parser = headers_parser
+ self._trailer_lines: list[bytes] = []
self.done = False
# payload decompression wrapper
@@ -850,7 +861,7 @@ class HttpPayloadParser:
size_b = chunk[:i] # strip chunk-extensions
# Verify no LF in the chunk-extension
if b"\n" in (ext := chunk[i:pos]):
- exc = BadHttpMessage(
+ exc = TransferEncodingError(
f"Unexpected LF in chunk-extension: {ext!r}"
)
set_exception(self.payload, exc)
@@ -871,7 +882,7 @@ class HttpPayloadParser:
chunk = chunk[pos + len(SEP) :]
if size == 0: # eof marker
- self._chunk = ChunkState.PARSE_MAYBE_TRAILERS
+ self._chunk = ChunkState.PARSE_TRAILERS
if self._lax and chunk.startswith(b"\r"):
chunk = chunk[1:]
else:
@@ -909,38 +920,31 @@ class HttpPayloadParser:
self._chunk_tail = chunk
return False, b""
- # if stream does not contain trailer, after 0\r\n
- # we should get another \r\n otherwise
- # trailers needs to be skipped until \r\n\r\n
- if self._chunk == ChunkState.PARSE_MAYBE_TRAILERS:
- head = chunk[: len(SEP)]
- if head == SEP:
- # end of stream
- self.payload.feed_eof()
- return True, chunk[len(SEP) :]
- # Both CR and LF, or only LF may not be received yet. It is
- # expected that CRLF or LF will be shown at the very first
- # byte next time, otherwise trailers should come. The last
- # CRLF which marks the end of response might not be
- # contained in the same TCP segment which delivered the
- # size indicator.
- if not head:
- return False, b""
- if head == SEP[:1]:
- self._chunk_tail = head
- return False, b""
- self._chunk = ChunkState.PARSE_TRAILERS
-
- # read and discard trailer up to the CRLF terminator
if self._chunk == ChunkState.PARSE_TRAILERS:
pos = chunk.find(SEP)
- if pos >= 0:
- chunk = chunk[pos + len(SEP) :]
- self._chunk = ChunkState.PARSE_MAYBE_TRAILERS
- else:
+ if pos < 0: # No line found
self._chunk_tail = chunk
return False, b""
+ line = chunk[:pos]
+ chunk = chunk[pos + len(SEP) :]
+ if SEP == b"\n": # For lax response parsing
+ line = line.rstrip(b"\r")
+ self._trailer_lines.append(line)
+
+ # \r\n\r\n found, end of stream
+ if self._trailer_lines[-1] == b"":
+ # Headers and trailers are defined the same way,
+ # so we reuse the HeadersParser here.
+ try:
+ trailers, raw_trailers = self._headers_parser.parse_headers(
+ self._trailer_lines
+ )
+ finally:
+ self._trailer_lines.clear()
+ self.payload.feed_eof()
+ return True, chunk
+
# Read all bytes until eof
elif self._type == ParseState.PARSE_UNTIL_EOF:
self.payload.feed_data(chunk, len(chunk))
diff --git a/aiohttp/multipart.py b/aiohttp/multipart.py
index 71fc2654a..520ee539e 100644
--- a/aiohttp/multipart.py
+++ b/aiohttp/multipart.py
@@ -723,7 +723,7 @@ class MultipartReader:
raise ValueError(f"Invalid boundary {chunk!r}, expected {self._boundary!r}")
async def _read_headers(self) -> "CIMultiDictProxy[str]":
- lines = [b""]
+ lines = []
while True:
chunk = await self._content.readline()
chunk = chunk.strip()