##// 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:

r36815:d4c760c9 default
r37304:9bfcbe4f default
Show More
encoding.py
582 lines | 18.9 KiB | text/x-python | PythonLexer
# encoding.py - character transcoding support for Mercurial
#
# Copyright 2005-2009 Matt Mackall <mpm@selenic.com> and others
#
# 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, print_function
import locale
import os
import unicodedata
from . import (
error,
policy,
pycompat,
)
from .pure import (
charencode as charencodepure,
)
charencode = policy.importmod(r'charencode')
isasciistr = charencode.isasciistr
asciilower = charencode.asciilower
asciiupper = charencode.asciiupper
_jsonescapeu8fast = charencode.jsonescapeu8fast
_sysstr = pycompat.sysstr
if pycompat.ispy3:
unichr = chr
# These unicode characters are ignored by HFS+ (Apple Technote 1150,
# "Unicode Subtleties"), so we need to ignore them in some places for
# sanity.
_ignore = [unichr(int(x, 16)).encode("utf-8") for x in
"200c 200d 200e 200f 202a 202b 202c 202d 202e "
"206a 206b 206c 206d 206e 206f feff".split()]
# verify the next function will work
assert all(i.startswith(("\xe2", "\xef")) for i in _ignore)
def hfsignoreclean(s):
"""Remove codepoints ignored by HFS+ from s.
>>> hfsignoreclean(u'.h\u200cg'.encode('utf-8'))
'.hg'
>>> hfsignoreclean(u'.h\ufeffg'.encode('utf-8'))
'.hg'
"""
if "\xe2" in s or "\xef" in s:
for c in _ignore:
s = s.replace(c, '')
return s
# encoding.environ is provided read-only, which may not be used to modify
# the process environment
_nativeenviron = (not pycompat.ispy3 or os.supports_bytes_environ)
if not pycompat.ispy3:
environ = os.environ # re-exports
elif _nativeenviron:
environ = os.environb # re-exports
else:
# preferred encoding isn't known yet; use utf-8 to avoid unicode error
# and recreate it once encoding is settled
environ = dict((k.encode(u'utf-8'), v.encode(u'utf-8'))
for k, v in os.environ.items()) # re-exports
_encodingfixers = {
'646': lambda: 'ascii',
'ANSI_X3.4-1968': lambda: 'ascii',
}
try:
encoding = environ.get("HGENCODING")
if not encoding:
encoding = locale.getpreferredencoding().encode('ascii') or 'ascii'
encoding = _encodingfixers.get(encoding, lambda: encoding)()
except locale.Error:
encoding = 'ascii'
encodingmode = environ.get("HGENCODINGMODE", "strict")
fallbackencoding = 'ISO-8859-1'
class localstr(bytes):
'''This class allows strings that are unmodified to be
round-tripped to the local encoding and back'''
def __new__(cls, u, l):
s = bytes.__new__(cls, l)
s._utf8 = u
return s
def __hash__(self):
return hash(self._utf8) # avoid collisions in local string space
def tolocal(s):
"""
Convert a string from internal UTF-8 to local encoding
All internal strings should be UTF-8 but some repos before the
implementation of locale support may contain latin1 or possibly
other character sets. We attempt to decode everything strictly
using UTF-8, then Latin-1, and failing that, we use UTF-8 and
replace unknown characters.
The localstr class is used to cache the known UTF-8 encoding of
strings next to their local representation to allow lossless
round-trip conversion back to UTF-8.
>>> u = b'foo: \\xc3\\xa4' # utf-8
>>> l = tolocal(u)
>>> l
'foo: ?'
>>> fromlocal(l)
'foo: \\xc3\\xa4'
>>> u2 = b'foo: \\xc3\\xa1'
>>> d = { l: 1, tolocal(u2): 2 }
>>> len(d) # no collision
2
>>> b'foo: ?' in d
False
>>> l1 = b'foo: \\xe4' # historical latin1 fallback
>>> l = tolocal(l1)
>>> l
'foo: ?'
>>> fromlocal(l) # magically in utf-8
'foo: \\xc3\\xa4'
"""
if isasciistr(s):
return s
try:
try:
# make sure string is actually stored in UTF-8
u = s.decode('UTF-8')
if encoding == 'UTF-8':
# fast path
return s
r = u.encode(_sysstr(encoding), u"replace")
if u == r.decode(_sysstr(encoding)):
# r is a safe, non-lossy encoding of s
return r
return localstr(s, r)
except UnicodeDecodeError:
# we should only get here if we're looking at an ancient changeset
try:
u = s.decode(_sysstr(fallbackencoding))
r = u.encode(_sysstr(encoding), u"replace")
if u == r.decode(_sysstr(encoding)):
# r is a safe, non-lossy encoding of s
return r
return localstr(u.encode('UTF-8'), r)
except UnicodeDecodeError:
u = s.decode("utf-8", "replace") # last ditch
# can't round-trip
return u.encode(_sysstr(encoding), u"replace")
except LookupError as k:
raise error.Abort(k, hint="please check your locale settings")
def fromlocal(s):
"""
Convert a string from the local character encoding to UTF-8
We attempt to decode strings using the encoding mode set by
HGENCODINGMODE, which defaults to 'strict'. In this mode, unknown
characters will cause an error message. Other modes include
'replace', which replaces unknown characters with a special
Unicode character, and 'ignore', which drops the character.
"""
# can we do a lossless round-trip?
if isinstance(s, localstr):
return s._utf8
if isasciistr(s):
return s
try:
u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
return u.encode("utf-8")
except UnicodeDecodeError as inst:
sub = s[max(0, inst.start - 10):inst.start + 10]
raise error.Abort("decoding near '%s': %s!"
% (sub, pycompat.bytestr(inst)))
except LookupError as k:
raise error.Abort(k, hint="please check your locale settings")
def unitolocal(u):
"""Convert a unicode string to a byte string of local encoding"""
return tolocal(u.encode('utf-8'))
def unifromlocal(s):
"""Convert a byte string of local encoding to a unicode string"""
return fromlocal(s).decode('utf-8')
def unimethod(bytesfunc):
"""Create a proxy method that forwards __unicode__() and __str__() of
Python 3 to __bytes__()"""
def unifunc(obj):
return unifromlocal(bytesfunc(obj))
return unifunc
# converter functions between native str and byte string. use these if the
# character encoding is not aware (e.g. exception message) or is known to
# be locale dependent (e.g. date formatting.)
if pycompat.ispy3:
strtolocal = unitolocal
strfromlocal = unifromlocal
strmethod = unimethod
else:
strtolocal = pycompat.identity
strfromlocal = pycompat.identity
strmethod = pycompat.identity
if not _nativeenviron:
# now encoding and helper functions are available, recreate the environ
# dict to be exported to other modules
environ = dict((tolocal(k.encode(u'utf-8')), tolocal(v.encode(u'utf-8')))
for k, v in os.environ.items()) # re-exports
# How to treat ambiguous-width characters. Set to 'wide' to treat as wide.
_wide = _sysstr(environ.get("HGENCODINGAMBIGUOUS", "narrow") == "wide"
and "WFA" or "WF")
def colwidth(s):
"Find the column width of a string for display in the local encoding"
return ucolwidth(s.decode(_sysstr(encoding), u'replace'))
def ucolwidth(d):
"Find the column width of a Unicode string for display"
eaw = getattr(unicodedata, 'east_asian_width', None)
if eaw is not None:
return sum([eaw(c) in _wide and 2 or 1 for c in d])
return len(d)
def getcols(s, start, c):
'''Use colwidth to find a c-column substring of s starting at byte
index start'''
for x in xrange(start + c, len(s)):
t = s[start:x]
if colwidth(t) == c:
return t
def trim(s, width, ellipsis='', leftside=False):
"""Trim string 's' to at most 'width' columns (including 'ellipsis').
If 'leftside' is True, left side of string 's' is trimmed.
'ellipsis' is always placed at trimmed side.
>>> from .node import bin
>>> def bprint(s):
... print(pycompat.sysstr(s))
>>> ellipsis = b'+++'
>>> from . import encoding
>>> encoding.encoding = b'utf-8'
>>> t = b'1234567890'
>>> bprint(trim(t, 12, ellipsis=ellipsis))
1234567890
>>> bprint(trim(t, 10, ellipsis=ellipsis))
1234567890
>>> bprint(trim(t, 8, ellipsis=ellipsis))
12345+++
>>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))
+++67890
>>> bprint(trim(t, 8))
12345678
>>> bprint(trim(t, 8, leftside=True))
34567890
>>> bprint(trim(t, 3, ellipsis=ellipsis))
+++
>>> bprint(trim(t, 1, ellipsis=ellipsis))
+
>>> u = u'\u3042\u3044\u3046\u3048\u304a' # 2 x 5 = 10 columns
>>> t = u.encode(pycompat.sysstr(encoding.encoding))
>>> bprint(trim(t, 12, ellipsis=ellipsis))
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a
>>> bprint(trim(t, 10, ellipsis=ellipsis))
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a
>>> bprint(trim(t, 8, ellipsis=ellipsis))
\xe3\x81\x82\xe3\x81\x84+++
>>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))
+++\xe3\x81\x88\xe3\x81\x8a
>>> bprint(trim(t, 5))
\xe3\x81\x82\xe3\x81\x84
>>> bprint(trim(t, 5, leftside=True))
\xe3\x81\x88\xe3\x81\x8a
>>> bprint(trim(t, 4, ellipsis=ellipsis))
+++
>>> bprint(trim(t, 4, ellipsis=ellipsis, leftside=True))
+++
>>> t = bin(b'112233445566778899aa') # invalid byte sequence
>>> bprint(trim(t, 12, ellipsis=ellipsis))
\x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa
>>> bprint(trim(t, 10, ellipsis=ellipsis))
\x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa
>>> bprint(trim(t, 8, ellipsis=ellipsis))
\x11\x22\x33\x44\x55+++
>>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))
+++\x66\x77\x88\x99\xaa
>>> bprint(trim(t, 8))
\x11\x22\x33\x44\x55\x66\x77\x88
>>> bprint(trim(t, 8, leftside=True))
\x33\x44\x55\x66\x77\x88\x99\xaa
>>> bprint(trim(t, 3, ellipsis=ellipsis))
+++
>>> bprint(trim(t, 1, ellipsis=ellipsis))
+
"""
try:
u = s.decode(_sysstr(encoding))
except UnicodeDecodeError:
if len(s) <= width: # trimming is not needed
return s
width -= len(ellipsis)
if width <= 0: # no enough room even for ellipsis
return ellipsis[:width + len(ellipsis)]
if leftside:
return ellipsis + s[-width:]
return s[:width] + ellipsis
if ucolwidth(u) <= width: # trimming is not needed
return s
width -= len(ellipsis)
if width <= 0: # no enough room even for ellipsis
return ellipsis[:width + len(ellipsis)]
if leftside:
uslice = lambda i: u[i:]
concat = lambda s: ellipsis + s
else:
uslice = lambda i: u[:-i]
concat = lambda s: s + ellipsis
for i in xrange(1, len(u)):
usub = uslice(i)
if ucolwidth(usub) <= width:
return concat(usub.encode(_sysstr(encoding)))
return ellipsis # no enough room for multi-column characters
def lower(s):
"best-effort encoding-aware case-folding of local string s"
try:
return asciilower(s)
except UnicodeDecodeError:
pass
try:
if isinstance(s, localstr):
u = s._utf8.decode("utf-8")
else:
u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
lu = u.lower()
if u == lu:
return s # preserve localstring
return lu.encode(_sysstr(encoding))
except UnicodeError:
return s.lower() # we don't know how to fold this except in ASCII
except LookupError as k:
raise error.Abort(k, hint="please check your locale settings")
def upper(s):
"best-effort encoding-aware case-folding of local string s"
try:
return asciiupper(s)
except UnicodeDecodeError:
return upperfallback(s)
def upperfallback(s):
try:
if isinstance(s, localstr):
u = s._utf8.decode("utf-8")
else:
u = s.decode(_sysstr(encoding), _sysstr(encodingmode))
uu = u.upper()
if u == uu:
return s # preserve localstring
return uu.encode(_sysstr(encoding))
except UnicodeError:
return s.upper() # we don't know how to fold this except in ASCII
except LookupError as k:
raise error.Abort(k, hint="please check your locale settings")
class normcasespecs(object):
'''what a platform's normcase does to ASCII strings
This is specified per platform, and should be consistent with what normcase
on that platform actually does.
lower: normcase lowercases ASCII strings
upper: normcase uppercases ASCII strings
other: the fallback function should always be called
This should be kept in sync with normcase_spec in util.h.'''
lower = -1
upper = 1
other = 0
def jsonescape(s, paranoid=False):
'''returns a string suitable for JSON
JSON is problematic for us because it doesn't support non-Unicode
bytes. To deal with this, we take the following approach:
- localstr objects are converted back to UTF-8
- valid UTF-8/ASCII strings are passed as-is
- other strings are converted to UTF-8b surrogate encoding
- apply JSON-specified string escaping
(escapes are doubled in these tests)
>>> jsonescape(b'this is a test')
'this is a test'
>>> jsonescape(b'escape characters: \\0 \\x0b \\x7f')
'escape characters: \\\\u0000 \\\\u000b \\\\u007f'
>>> jsonescape(b'escape characters: \\b \\t \\n \\f \\r \\" \\\\')
'escape characters: \\\\b \\\\t \\\\n \\\\f \\\\r \\\\" \\\\\\\\'
>>> jsonescape(b'a weird byte: \\xdd')
'a weird byte: \\xed\\xb3\\x9d'
>>> jsonescape(b'utf-8: caf\\xc3\\xa9')
'utf-8: caf\\xc3\\xa9'
>>> jsonescape(b'')
''
If paranoid, non-ascii and common troublesome characters are also escaped.
This is suitable for web output.
>>> s = b'escape characters: \\0 \\x0b \\x7f'
>>> assert jsonescape(s) == jsonescape(s, paranoid=True)
>>> s = b'escape characters: \\b \\t \\n \\f \\r \\" \\\\'
>>> assert jsonescape(s) == jsonescape(s, paranoid=True)
>>> jsonescape(b'escape boundary: \\x7e \\x7f \\xc2\\x80', paranoid=True)
'escape boundary: ~ \\\\u007f \\\\u0080'
>>> jsonescape(b'a weird byte: \\xdd', paranoid=True)
'a weird byte: \\\\udcdd'
>>> jsonescape(b'utf-8: caf\\xc3\\xa9', paranoid=True)
'utf-8: caf\\\\u00e9'
>>> jsonescape(b'non-BMP: \\xf0\\x9d\\x84\\x9e', paranoid=True)
'non-BMP: \\\\ud834\\\\udd1e'
>>> jsonescape(b'<foo@example.org>', paranoid=True)
'\\\\u003cfoo@example.org\\\\u003e'
'''
u8chars = toutf8b(s)
try:
return _jsonescapeu8fast(u8chars, paranoid)
except ValueError:
pass
return charencodepure.jsonescapeu8fallback(u8chars, paranoid)
# We need to decode/encode U+DCxx codes transparently since invalid UTF-8
# bytes are mapped to that range.
if pycompat.ispy3:
_utf8strict = r'surrogatepass'
else:
_utf8strict = r'strict'
_utf8len = [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 4]
def getutf8char(s, pos):
'''get the next full utf-8 character in the given string, starting at pos
Raises a UnicodeError if the given location does not start a valid
utf-8 character.
'''
# find how many bytes to attempt decoding from first nibble
l = _utf8len[ord(s[pos:pos + 1]) >> 4]
if not l: # ascii
return s[pos:pos + 1]
c = s[pos:pos + l]
# validate with attempted decode
c.decode("utf-8", _utf8strict)
return c
def toutf8b(s):
'''convert a local, possibly-binary string into UTF-8b
This is intended as a generic method to preserve data when working
with schemes like JSON and XML that have no provision for
arbitrary byte strings. As Mercurial often doesn't know
what encoding data is in, we use so-called UTF-8b.
If a string is already valid UTF-8 (or ASCII), it passes unmodified.
Otherwise, unsupported bytes are mapped to UTF-16 surrogate range,
uDC00-uDCFF.
Principles of operation:
- ASCII and UTF-8 data successfully round-trips and is understood
by Unicode-oriented clients
- filenames and file contents in arbitrary other encodings can have
be round-tripped or recovered by clueful clients
- local strings that have a cached known UTF-8 encoding (aka
localstr) get sent as UTF-8 so Unicode-oriented clients get the
Unicode data they want
- because we must preserve UTF-8 bytestring in places such as
filenames, metadata can't be roundtripped without help
(Note: "UTF-8b" often refers to decoding a mix of valid UTF-8 and
arbitrary bytes into an internal Unicode format that can be
re-encoded back into the original. Here we are exposing the
internal surrogate encoding as a UTF-8 string.)
'''
if not isinstance(s, localstr) and isasciistr(s):
return s
if "\xed" not in s:
if isinstance(s, localstr):
return s._utf8
try:
s.decode('utf-8', _utf8strict)
return s
except UnicodeDecodeError:
pass
s = pycompat.bytestr(s)
r = ""
pos = 0
l = len(s)
while pos < l:
try:
c = getutf8char(s, pos)
if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf":
# have to re-escape existing U+DCxx characters
c = unichr(0xdc00 + ord(s[pos])).encode('utf-8', _utf8strict)
pos += 1
else:
pos += len(c)
except UnicodeDecodeError:
c = unichr(0xdc00 + ord(s[pos])).encode('utf-8', _utf8strict)
pos += 1
r += c
return r
def fromutf8b(s):
'''Given a UTF-8b string, return a local, possibly-binary string.
return the original binary string. This
is a round-trip process for strings like filenames, but metadata
that's was passed through tolocal will remain in UTF-8.
>>> roundtrip = lambda x: fromutf8b(toutf8b(x)) == x
>>> m = b"\\xc3\\xa9\\x99abcd"
>>> toutf8b(m)
'\\xc3\\xa9\\xed\\xb2\\x99abcd'
>>> roundtrip(m)
True
>>> roundtrip(b"\\xc2\\xc2\\x80")
True
>>> roundtrip(b"\\xef\\xbf\\xbd")
True
>>> roundtrip(b"\\xef\\xef\\xbf\\xbd")
True
>>> roundtrip(b"\\xf1\\x80\\x80\\x80\\x80")
True
'''
if isasciistr(s):
return s
# fast path - look for uDxxx prefixes in s
if "\xed" not in s:
return s
# We could do this with the unicode type but some Python builds
# use UTF-16 internally (issue5031) which causes non-BMP code
# points to be escaped. Instead, we use our handy getutf8char
# helper again to walk the string without "decoding" it.
s = pycompat.bytestr(s)
r = ""
pos = 0
l = len(s)
while pos < l:
c = getutf8char(s, pos)
pos += len(c)
# unescape U+DCxx characters
if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf":
c = pycompat.bytechr(ord(c.decode("utf-8", _utf8strict)) & 0xff)
r += c
return r