From 9df00a132a6ce64e97c455dc26076345e79f68c7 Mon Sep 17 00:00:00 2001
From: "patchback[bot]" <45432694+patchback[bot]@users.noreply.github.com>
Date: Sun, 1 Dec 2024 21:55:51 +0000
Subject: [PATCH] [PR #10073/349b7565 backport][3.11] Improve performance of
 parsing headers (#10083)

Co-authored-by: J. Nick Koston <nick@koston.org>
---
 CHANGES/10073.misc.rst   |  1 +
 aiohttp/_http_parser.pyx | 56 ++++++++++++++++++----------------------
 2 files changed, 26 insertions(+), 31 deletions(-)
 create mode 100644 CHANGES/10073.misc.rst

diff --git a/CHANGES/10073.misc.rst b/CHANGES/10073.misc.rst
new file mode 100644
index 00000000000..2a7e3514ea3
--- /dev/null
+++ b/CHANGES/10073.misc.rst
@@ -0,0 +1 @@
+Improved performance of parsing headers when using the C parser -- by :user:`bdraco`.
diff --git a/aiohttp/_http_parser.pyx b/aiohttp/_http_parser.pyx
index 988e4247f93..19dc3e63b74 100644
--- a/aiohttp/_http_parser.pyx
+++ b/aiohttp/_http_parser.pyx
@@ -71,7 +71,7 @@ cdef object CONTENT_ENCODING = hdrs.CONTENT_ENCODING
 cdef object EMPTY_PAYLOAD = _EMPTY_PAYLOAD
 cdef object StreamReader = _StreamReader
 cdef object DeflateBuffer = _DeflateBuffer
-
+cdef bytes EMPTY_BYTES = b""
 
 cdef inline object extend(object buf, const char* at, size_t length):
     cdef Py_ssize_t s
@@ -277,8 +277,9 @@ cdef class HttpParser:
         cparser.llhttp_t* _cparser
         cparser.llhttp_settings_t* _csettings
 
-        bytearray _raw_name
-        bytearray _raw_value
+        bytes _raw_name
+        object _name
+        bytes _raw_value
         bint      _has_value
 
         object _protocol
@@ -296,7 +297,7 @@ cdef class HttpParser:
         bytearray   _buf
         str     _path
         str     _reason
-        object  _headers
+        list    _headers
         list    _raw_headers
         bint    _upgraded
         list    _messages
@@ -350,8 +351,8 @@ cdef class HttpParser:
         self._payload_exception = payload_exception
         self._messages = []
 
-        self._raw_name = bytearray()
-        self._raw_value = bytearray()
+        self._raw_name = EMPTY_BYTES
+        self._raw_value = EMPTY_BYTES
         self._has_value = False
 
         self._max_line_size = max_line_size
@@ -378,42 +379,35 @@ cdef class HttpParser:
         self._limit = limit
 
     cdef _process_header(self):
-        if self._raw_name:
-            raw_name = bytes(self._raw_name)
-            raw_value = bytes(self._raw_value)
-
-            name = find_header(raw_name)
-            value = raw_value.decode('utf-8', 'surrogateescape')
+        cdef str value
+        if self._raw_name is not EMPTY_BYTES:
+            name = find_header(self._raw_name)
+            value = self._raw_value.decode('utf-8', 'surrogateescape')
 
-            self._headers.add(name, value)
+            self._headers.append((name, value))
 
             if name is CONTENT_ENCODING:
                 self._content_encoding = value
 
-            PyByteArray_Resize(self._raw_name, 0)
-            PyByteArray_Resize(self._raw_value, 0)
             self._has_value = False
-            self._raw_headers.append((raw_name, raw_value))
+            self._raw_headers.append((self._raw_name, self._raw_value))
+            self._raw_name = EMPTY_BYTES
+            self._raw_value = EMPTY_BYTES
 
     cdef _on_header_field(self, char* at, size_t length):
-        cdef Py_ssize_t size
-        cdef char *buf
         if self._has_value:
             self._process_header()
 
-        size = PyByteArray_Size(self._raw_name)
-        PyByteArray_Resize(self._raw_name, size + length)
-        buf = PyByteArray_AsString(self._raw_name)
-        memcpy(buf + size, at, length)
+        if self._raw_name is EMPTY_BYTES:
+            self._raw_name = at[:length]
+        else:
+            self._raw_name += at[:length]
 
     cdef _on_header_value(self, char* at, size_t length):
-        cdef Py_ssize_t size
-        cdef char *buf
-
-        size = PyByteArray_Size(self._raw_value)
-        PyByteArray_Resize(self._raw_value, size + length)
-        buf = PyByteArray_AsString(self._raw_value)
-        memcpy(buf + size, at, length)
+        if self._raw_value is EMPTY_BYTES:
+            self._raw_value = at[:length]
+        else:
+            self._raw_value += at[:length]
         self._has_value = True
 
     cdef _on_headers_complete(self):
@@ -424,7 +418,7 @@ cdef class HttpParser:
         chunked = self._cparser.flags & cparser.F_CHUNKED
 
         raw_headers = tuple(self._raw_headers)
-        headers = CIMultiDictProxy(self._headers)
+        headers = CIMultiDictProxy(CIMultiDict(self._headers))
 
         if self._cparser.type == cparser.HTTP_REQUEST:
             allowed = upgrade and headers.get("upgrade", "").lower() in ALLOWED_UPGRADES
@@ -672,7 +666,7 @@ cdef int cb_on_message_begin(cparser.llhttp_t* parser) except -1:
     cdef HttpParser pyparser = <HttpParser>parser.data
 
     pyparser._started = True
-    pyparser._headers = CIMultiDict()
+    pyparser._headers = []
     pyparser._raw_headers = []
     PyByteArray_Resize(pyparser._buf, 0)
     pyparser._path = None