##// END OF EJS Templates
wireproto: add streams to frame-based protocol...
wireproto: add streams to frame-based protocol Previously, the frame-based protocol was just a series of frames, with each frame associated with a request ID. In order to scale the protocol, we'll want to enable the use of compression. While it is possible to enable compression at the socket/pipe level, this has its disadvantages. The big one is it undermines the point of frames being standalone, atomic units that can be read and written: if you add compression above the framing protocol, you are back to having a stream-based protocol as opposed to something frame-based. So in order to preserve frames, compression needs to occur at the frame payload level. Compressing each frame's payload individually will limit compression ratios because the window size of the compressor will be limited by the max frame size, which is 32-64kb as currently defined. It will also add CPU overhead, as it is more efficient for compressors to operate on fewer, larger blocks of data than more, smaller blocks. So compressing each frame independently is out. This means we need to compress each frame's payload as if it is part of a larger stream. The simplest approach is to have 1 stream per connection. This could certainly work. However, it has disadvantages (documented below). We could also have 1 stream per RPC/command invocation. (This is the model HTTP/2 goes with.) This also has disadvantages. The main disadvantage to one global stream is that it has the very real potential to create CPU bottlenecks doing compression. Networks are only getting faster and the performance of single CPU cores has been relatively flat. Newer compression formats like zstandard offer better CPU cycle efficiency than predecessors like zlib. But it still all too common to saturate your CPU with compression overhead long before you saturate the network pipe. The main disadvantage with streams per request is that you can't reap the benefits of the compression context for multiple requests. For example, if you send 1000 RPC requests (or HTTP/2 requests for that matter), the response to each would have its own compression context. The overall size of the raw responses would be larger because compression contexts wouldn't be able to reference data from another request or response. The approach for streams as implemented in this commit is to support N streams per connection and for streams to potentially span requests and responses. As explained by the added internals docs, this facilitates servers and clients delegating independent streams and compression to independent threads / CPU cores. This helps alleviate the CPU bottleneck of compression. This design also allows compression contexts to be reused across requests/responses. This can result in improved compression ratios and less overhead for compressors and decompressors having to build new contexts. Another feature that was defined was the ability for individual frames within a stream to declare whether that individual frame's payload uses the content encoding (read: compression) defined by the stream. The idea here is that some servers may serve data from a combination of caches and dynamic resolution. Data coming from caches may be pre-compressed. We want to facilitate servers being able to essentially stream bytes from caches to the wire with minimal overhead. Being able to mix and match with frames are compressed within a stream enables these types of advanced server functionality. This commit defines the new streams mechanism. Basic code for supporting streams in frames has been added. But that code is seriously lacking and doesn't fully conform to the defined protocol. For example, we don't close any streams. And support for content encoding within streams is not yet implemented. The change was rather invasive and I didn't think it would be reasonable to implement the entire feature in a single commit. For the record, I would have loved to reuse an existing multiplexing protocol to build the new wire protocol on top of. However, I couldn't find a protocol that offers the performance and scaling characteristics that I desired. Namely, it should support multiple compression contexts to facilitate scaling out to multiple CPU cores and compression contexts should be able to live longer than single RPC requests. HTTP/2 *almost* fits the bill. But the semantics of HTTP message exchange state that streams can only live for a single request-response. We /could/ tunnel on top of HTTP/2 streams and frames with HEADER and DATA frames. But there's no guarantee that HTTP/2 libraries and proxies would allow us to use HTTP/2 streams and frames without the HTTP message exchange semantics defined in RFC 7540 Section 8. Other RPC protocols like gRPC tunnel are built on top of HTTP/2 and thus preserve its semantics of stream per RPC invocation. Even QUIC does this. We could attempt to invent a higher-level stream that spans HTTP/2 streams. But this would be violating HTTP/2 because there is no guarantee that HTTP/2 streams are routed to the same server. The best we can do - which is what this protocol does - is shoehorn all request and response data into a single HTTP message and create streams within. At that point, we've defined a Content-Type in HTTP parlance. It just so happens our media type can also work as a standalone, stream-based protocol, without leaning on HTTP or similar protocol. Differential Revision: https://phab.mercurial-scm.org/D2907

File last commit:

r37138:a8a902d7 default
r37304:9bfcbe4f default
Show More
mail.py
341 lines | 11.9 KiB | text/x-python | PythonLexer
# mail.py - mail sending bits for mercurial
#
# Copyright 2006 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
import email
import email.charset
import email.header
import email.message
import os
import smtplib
import socket
import time
from .i18n import _
from . import (
encoding,
error,
pycompat,
sslutil,
util,
)
from .utils import (
procutil,
stringutil,
)
class STARTTLS(smtplib.SMTP):
'''Derived class to verify the peer certificate for STARTTLS.
This class allows to pass any keyword arguments to SSL socket creation.
'''
def __init__(self, ui, host=None, **kwargs):
smtplib.SMTP.__init__(self, **kwargs)
self._ui = ui
self._host = host
def starttls(self, keyfile=None, certfile=None):
if not self.has_extn("starttls"):
msg = "STARTTLS extension not supported by server"
raise smtplib.SMTPException(msg)
(resp, reply) = self.docmd("STARTTLS")
if resp == 220:
self.sock = sslutil.wrapsocket(self.sock, keyfile, certfile,
ui=self._ui,
serverhostname=self._host)
self.file = smtplib.SSLFakeFile(self.sock)
self.helo_resp = None
self.ehlo_resp = None
self.esmtp_features = {}
self.does_esmtp = 0
return (resp, reply)
class SMTPS(smtplib.SMTP):
'''Derived class to verify the peer certificate for SMTPS.
This class allows to pass any keyword arguments to SSL socket creation.
'''
def __init__(self, ui, keyfile=None, certfile=None, host=None,
**kwargs):
self.keyfile = keyfile
self.certfile = certfile
smtplib.SMTP.__init__(self, **kwargs)
self._host = host
self.default_port = smtplib.SMTP_SSL_PORT
self._ui = ui
def _get_socket(self, host, port, timeout):
if self.debuglevel > 0:
self._ui.debug('connect: %r\n' % (host, port))
new_socket = socket.create_connection((host, port), timeout)
new_socket = sslutil.wrapsocket(new_socket,
self.keyfile, self.certfile,
ui=self._ui,
serverhostname=self._host)
self.file = smtplib.SSLFakeFile(new_socket)
return new_socket
def _smtp(ui):
'''build an smtp connection and return a function to send mail'''
local_hostname = ui.config('smtp', 'local_hostname')
tls = ui.config('smtp', 'tls')
# backward compatible: when tls = true, we use starttls.
starttls = tls == 'starttls' or stringutil.parsebool(tls)
smtps = tls == 'smtps'
if (starttls or smtps) and not util.safehasattr(socket, 'ssl'):
raise error.Abort(_("can't use TLS: Python SSL support not installed"))
mailhost = ui.config('smtp', 'host')
if not mailhost:
raise error.Abort(_('smtp.host not configured - cannot send mail'))
if smtps:
ui.note(_('(using smtps)\n'))
s = SMTPS(ui, local_hostname=local_hostname, host=mailhost)
elif starttls:
s = STARTTLS(ui, local_hostname=local_hostname, host=mailhost)
else:
s = smtplib.SMTP(local_hostname=local_hostname)
if smtps:
defaultport = 465
else:
defaultport = 25
mailport = util.getport(ui.config('smtp', 'port', defaultport))
ui.note(_('sending mail: smtp host %s, port %d\n') %
(mailhost, mailport))
s.connect(host=mailhost, port=mailport)
if starttls:
ui.note(_('(using starttls)\n'))
s.ehlo()
s.starttls()
s.ehlo()
if starttls or smtps:
ui.note(_('(verifying remote certificate)\n'))
sslutil.validatesocket(s.sock)
username = ui.config('smtp', 'username')
password = ui.config('smtp', 'password')
if username and not password:
password = ui.getpass()
if username and password:
ui.note(_('(authenticating to mail server as %s)\n') %
(username))
try:
s.login(username, password)
except smtplib.SMTPException as inst:
raise error.Abort(inst)
def send(sender, recipients, msg):
try:
return s.sendmail(sender, recipients, msg)
except smtplib.SMTPRecipientsRefused as inst:
recipients = [r[1] for r in inst.recipients.values()]
raise error.Abort('\n' + '\n'.join(recipients))
except smtplib.SMTPException as inst:
raise error.Abort(inst)
return send
def _sendmail(ui, sender, recipients, msg):
'''send mail using sendmail.'''
program = ui.config('email', 'method')
cmdline = '%s -f %s %s' % (program, stringutil.email(sender),
' '.join(map(stringutil.email, recipients)))
ui.note(_('sending mail: %s\n') % cmdline)
fp = procutil.popen(cmdline, 'w')
fp.write(msg)
ret = fp.close()
if ret:
raise error.Abort('%s %s' % (
os.path.basename(program.split(None, 1)[0]),
procutil.explainexit(ret)[0]))
def _mbox(mbox, sender, recipients, msg):
'''write mails to mbox'''
fp = open(mbox, 'ab+')
# Should be time.asctime(), but Windows prints 2-characters day
# of month instead of one. Make them print the same thing.
date = time.strftime(r'%a %b %d %H:%M:%S %Y', time.localtime())
fp.write('From %s %s\n' % (sender, date))
fp.write(msg)
fp.write('\n\n')
fp.close()
def connect(ui, mbox=None):
'''make a mail connection. return a function to send mail.
call as sendmail(sender, list-of-recipients, msg).'''
if mbox:
open(mbox, 'wb').close()
return lambda s, r, m: _mbox(mbox, s, r, m)
if ui.config('email', 'method') == 'smtp':
return _smtp(ui)
return lambda s, r, m: _sendmail(ui, s, r, m)
def sendmail(ui, sender, recipients, msg, mbox=None):
send = connect(ui, mbox=mbox)
return send(sender, recipients, msg)
def validateconfig(ui):
'''determine if we have enough config data to try sending email.'''
method = ui.config('email', 'method')
if method == 'smtp':
if not ui.config('smtp', 'host'):
raise error.Abort(_('smtp specified as email transport, '
'but no smtp host configured'))
else:
if not procutil.findexe(method):
raise error.Abort(_('%r specified as email transport, '
'but not in PATH') % method)
def codec2iana(cs):
''''''
cs = pycompat.sysbytes(email.charset.Charset(cs).input_charset.lower())
# "latin1" normalizes to "iso8859-1", standard calls for "iso-8859-1"
if cs.startswith("iso") and not cs.startswith("iso-"):
return "iso-" + cs[3:]
return cs
def mimetextpatch(s, subtype='plain', display=False):
'''Return MIME message suitable for a patch.
Charset will be detected by first trying to decode as us-ascii, then utf-8,
and finally the global encodings. If all those fail, fall back to
ISO-8859-1, an encoding with that allows all byte sequences.
Transfer encodings will be used if necessary.'''
cs = ['us-ascii', 'utf-8', encoding.encoding, encoding.fallbackencoding]
if display:
return mimetextqp(s, subtype, 'us-ascii')
for charset in cs:
try:
s.decode(pycompat.sysstr(charset))
return mimetextqp(s, subtype, codec2iana(charset))
except UnicodeDecodeError:
pass
return mimetextqp(s, subtype, "iso-8859-1")
def mimetextqp(body, subtype, charset):
'''Return MIME message.
Quoted-printable transfer encoding will be used if necessary.
'''
cs = email.charset.Charset(charset)
msg = email.message.Message()
msg.set_type(pycompat.sysstr('text/' + subtype))
for line in body.splitlines():
if len(line) > 950:
cs.body_encoding = email.charset.QP
break
msg.set_payload(body, cs)
return msg
def _charsets(ui):
'''Obtains charsets to send mail parts not containing patches.'''
charsets = [cs.lower() for cs in ui.configlist('email', 'charsets')]
fallbacks = [encoding.fallbackencoding.lower(),
encoding.encoding.lower(), 'utf-8']
for cs in fallbacks: # find unique charsets while keeping order
if cs not in charsets:
charsets.append(cs)
return [cs for cs in charsets if not cs.endswith('ascii')]
def _encode(ui, s, charsets):
'''Returns (converted) string, charset tuple.
Finds out best charset by cycling through sendcharsets in descending
order. Tries both encoding and fallbackencoding for input. Only as
last resort send as is in fake ascii.
Caveat: Do not use for mail parts containing patches!'''
try:
s.decode('ascii')
except UnicodeDecodeError:
sendcharsets = charsets or _charsets(ui)
for ics in (encoding.encoding, encoding.fallbackencoding):
try:
u = s.decode(ics)
except UnicodeDecodeError:
continue
for ocs in sendcharsets:
try:
return u.encode(ocs), ocs
except UnicodeEncodeError:
pass
except LookupError:
ui.warn(_('ignoring invalid sendcharset: %s\n') % ocs)
# if ascii, or all conversion attempts fail, send (broken) ascii
return s, 'us-ascii'
def headencode(ui, s, charsets=None, display=False):
'''Returns RFC-2047 compliant header from given string.'''
if not display:
# split into words?
s, cs = _encode(ui, s, charsets)
return str(email.header.Header(s, cs))
return s
def _addressencode(ui, name, addr, charsets=None):
name = headencode(ui, name, charsets)
try:
acc, dom = addr.split('@')
acc = acc.encode('ascii')
dom = dom.decode(encoding.encoding).encode('idna')
addr = '%s@%s' % (acc, dom)
except UnicodeDecodeError:
raise error.Abort(_('invalid email address: %s') % addr)
except ValueError:
try:
# too strict?
addr = addr.encode('ascii')
except UnicodeDecodeError:
raise error.Abort(_('invalid local address: %s') % addr)
return email.utils.formataddr((name, addr))
def addressencode(ui, address, charsets=None, display=False):
'''Turns address into RFC-2047 compliant header.'''
if display or not address:
return address or ''
name, addr = email.utils.parseaddr(address)
return _addressencode(ui, name, addr, charsets)
def addrlistencode(ui, addrs, charsets=None, display=False):
'''Turns a list of addresses into a list of RFC-2047 compliant headers.
A single element of input list may contain multiple addresses, but output
always has one address per item'''
if display:
return [a.strip() for a in addrs if a.strip()]
result = []
for name, addr in email.utils.getaddresses(addrs):
if name or addr:
result.append(_addressencode(ui, name, addr, charsets))
return result
def mimeencode(ui, s, charsets=None, display=False):
'''creates mime text object, encodes it if needed, and sets
charset and transfer-encoding accordingly.'''
cs = 'us-ascii'
if not display:
s, cs = _encode(ui, s, charsets)
return mimetextqp(s, 'plain', cs)
def headdecode(s):
'''Decodes RFC-2047 header'''
uparts = []
for part, charset in email.header.decode_header(s):
if charset is not None:
try:
uparts.append(part.decode(charset))
continue
except UnicodeDecodeError:
pass
try:
uparts.append(part.decode('UTF-8'))
continue
except UnicodeDecodeError:
pass
uparts.append(part.decode('ISO-8859-1'))
return encoding.unitolocal(u' '.join(uparts))