wireproto: compress data from a generator
Currently, the "getbundle" wire protocol command obtains a generator of
data, converts it to a util.chunkbuffer, then converts it back to a
generator via the protocol's groupchunks() implementation. For the SSH
protocol, groupchunks() simply reads 4kb chunks then write()s the
data to a file descriptor. For the HTTP protocol, groupchunks() reads
32kb chunks, feeds those into a zlib compressor, emits compressed data
as it is available, and that is sent to the WSGI layer, where it is
likely turned into HTTP chunked transfer chunks as is or further
buffered and turned into a larger chunk.
For both the SSH and HTTP protocols, there is inefficiency from using
util.chunkbuffer.
For SSH, emitting consistent 4kb chunks sounds nice. However, the file
descriptor it is writing to is almost certainly buffered. That means
that a Python .write() probably doesn't translate into exactly what is
written to the I/O layer.
For HTTP, we're going through an intermediate layer to zlib compress
data. So all util.chunkbuffer is doing is ensuring that the chunks we
feed into the zlib compressor are of uniform size. This means more CPU
time in Python buffering and emitting chunks in util.chunkbuffer but
fewer function calls to zlib.
This patch introduces and implements a new wire protocol abstract
method: compresschunks(). It is like groupchunks() except it operates
on a generator instead of something with a .read(). The SSH
implementation simply proxies chunks. The HTTP implementation uses
zlib compression.
To avoid duplicate code, the HTTP groupchunks() has been reimplemented
in terms of compresschunks().
To prove this all works, the "getbundle" wire protocol command has been
switched to compresschunks(). This removes the util.chunkbuffer from
that command. Now, data essentially streams straight from the
changegroup emitter to the wire, possibly through a zlib compressor.
Generators all the way, baby.
There were slim to no performance changes on the server as measured
with the mozilla-central repository. This is likely because CPU
time is dominated by reading revlogs, producing the changegroup, and
zlib compressing the output stream. Still, this brings us a little
closer to our ideal of using generators everywhere.
#!/usr/bin/env python
from __future__ import absolute_import, print_function
__doc__ = """Tiny HTTP Proxy.
This module implements GET, HEAD, POST, PUT and DELETE methods
on BaseHTTPServer, and behaves as an HTTP proxy. The CONNECT
method is also implemented experimentally, but has not been
tested yet.
Any help will be greatly appreciated. SUZUKI Hisao
"""
__version__ = "0.2.1"
import optparse
import os
import select
import socket
import sys
from mercurial import util
httpserver = util.httpserver
urlparse = util.urlparse
socketserver = util.socketserver
class ProxyHandler (httpserver.basehttprequesthandler):
__base = httpserver.basehttprequesthandler
__base_handle = __base.handle
server_version = "TinyHTTPProxy/" + __version__
rbufsize = 0 # self.rfile Be unbuffered
def handle(self):
(ip, port) = self.client_address
allowed = getattr(self, 'allowed_clients', None)
if allowed is not None and ip not in allowed:
self.raw_requestline = self.rfile.readline()
if self.parse_request():
self.send_error(403)
else:
self.__base_handle()
def log_request(self, code='-', size='-'):
xheaders = [h for h in self.headers.items() if h[0].startswith('x-')]
self.log_message('"%s" %s %s%s',
self.requestline, str(code), str(size),
''.join([' %s:%s' % h for h in sorted(xheaders)]))
def _connect_to(self, netloc, soc):
i = netloc.find(':')
if i >= 0:
host_port = netloc[:i], int(netloc[i + 1:])
else:
host_port = netloc, 80
print("\t" "connect to %s:%d" % host_port)
try: soc.connect(host_port)
except socket.error as arg:
try: msg = arg[1]
except (IndexError, TypeError): msg = arg
self.send_error(404, msg)
return 0
return 1
def do_CONNECT(self):
soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
try:
if self._connect_to(self.path, soc):
self.log_request(200)
self.wfile.write(self.protocol_version +
" 200 Connection established\r\n")
self.wfile.write("Proxy-agent: %s\r\n" % self.version_string())
self.wfile.write("\r\n")
self._read_write(soc, 300)
finally:
print("\t" "bye")
soc.close()
self.connection.close()
def do_GET(self):
(scm, netloc, path, params, query, fragment) = urlparse.urlparse(
self.path, 'http')
if scm != 'http' or fragment or not netloc:
self.send_error(400, "bad url %s" % self.path)
return
soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
try:
if self._connect_to(netloc, soc):
self.log_request()
soc.send("%s %s %s\r\n" % (
self.command,
urlparse.urlunparse(('', '', path, params, query, '')),
self.request_version))
self.headers['Connection'] = 'close'
del self.headers['Proxy-Connection']
for key_val in self.headers.items():
soc.send("%s: %s\r\n" % key_val)
soc.send("\r\n")
self._read_write(soc)
finally:
print("\t" "bye")
soc.close()
self.connection.close()
def _read_write(self, soc, max_idling=20):
iw = [self.connection, soc]
ow = []
count = 0
while True:
count += 1
(ins, _, exs) = select.select(iw, ow, iw, 3)
if exs:
break
if ins:
for i in ins:
if i is soc:
out = self.connection
else:
out = soc
try:
data = i.recv(8192)
except socket.error:
break
if data:
out.send(data)
count = 0
else:
print("\t" "idle", count)
if count == max_idling:
break
do_HEAD = do_GET
do_POST = do_GET
do_PUT = do_GET
do_DELETE = do_GET
class ThreadingHTTPServer (socketserver.ThreadingMixIn,
httpserver.httpserver):
def __init__(self, *args, **kwargs):
httpserver.httpserver.__init__(self, *args, **kwargs)
a = open("proxy.pid", "w")
a.write(str(os.getpid()) + "\n")
a.close()
def runserver(port=8000, bind=""):
server_address = (bind, port)
ProxyHandler.protocol_version = "HTTP/1.0"
httpd = ThreadingHTTPServer(server_address, ProxyHandler)
sa = httpd.socket.getsockname()
print("Serving HTTP on", sa[0], "port", sa[1], "...")
try:
httpd.serve_forever()
except KeyboardInterrupt:
print("\nKeyboard interrupt received, exiting.")
httpd.server_close()
sys.exit(0)
if __name__ == '__main__':
argv = sys.argv
if argv[1:] and argv[1] in ('-h', '--help'):
print(argv[0], "[port [allowed_client_name ...]]")
else:
if argv[2:]:
allowed = []
for name in argv[2:]:
client = socket.gethostbyname(name)
allowed.append(client)
print("Accept: %s (%s)" % (client, name))
ProxyHandler.allowed_clients = allowed
del argv[2:]
else:
print("Any clients will be served...")
parser = optparse.OptionParser()
parser.add_option('-b', '--bind', metavar='ADDRESS',
help='Specify alternate bind address '
'[default: all interfaces]', default='')
(options, args) = parser.parse_args()
port = 8000
if len(args) == 1:
port = int(args[0])
runserver(port, options.bind)