py3: use io.BytesIO directly

Previously, pycompat.bytesio and pycompat.stringio referred to
io.BytesIO. And util.bytesio and util.stringio aliased the pycompat
symbols.

This commit switches everything to use io.BytesIO directly. util.bytesio
and util.stringio still exist to provide backwards compatibility, as
they were the preferred symbols.

Differential Revision: https://phab.mercurial-scm.org/D12252

hgext/phabricator.py

@@ -1,2401 +1,2402 b''
 # phabricator.py - simple Phabricator integration
 #
 # Copyright 2017 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 """simple Phabricator integration (EXPERIMENTAL)
 
 This extension provides a ``phabsend`` command which sends a stack of
 changesets to Phabricator, and a ``phabread`` command which prints a stack of
 revisions in a format suitable for :hg:`import`, and a ``phabupdate`` command
 to update statuses in batch.
 
 A "phabstatus" view for :hg:`show` is also provided; it displays status
 information of Phabricator differentials associated with unfinished
 changesets.
 
 By default, Phabricator requires ``Test Plan`` which might prevent some
 changeset from being sent. The requirement could be disabled by changing
 ``differential.require-test-plan-field`` config server side.
 
 Config::
 
     [phabricator]
     # Phabricator URL
     url = https://phab.example.com/
 
     # Repo callsign. If a repo has a URL https://$HOST/diffusion/FOO, then its
     # callsign is "FOO".
     callsign = FOO
 
     # curl command to use. If not set (default), use builtin HTTP library to
     # communicate. If set, use the specified curl command. This could be useful
     # if you need to specify advanced options that is not easily supported by
     # the internal library.
     curlcmd = curl --connect-timeout 2 --retry 3 --silent
 
     # retry failed command N time (default 0). Useful when using the extension
     # over flakly connection.
     #
     # We wait `retry.interval` between each retry, in seconds.
     # (default 1 second).
     retry = 3
     retry.interval = 10
 
     # the retry option can combine well with the http.timeout one.
     #
     # For example to give up on http request after 20 seconds:
     [http]
     timeout=20
 
     [auth]
     example.schemes = https
     example.prefix = phab.example.com
 
     # API token. Get it from https://$HOST/conduit/login/
     example.phabtoken = cli-xxxxxxxxxxxxxxxxxxxxxxxxxxxx
 """
 
 from __future__ import absolute_import
 
 import base64
 import contextlib
 import hashlib
+import io
 import itertools
 import json
 import mimetypes
 import operator
 import re
 import time
 
 from mercurial.node import bin, short
 from mercurial.i18n import _
 from mercurial.pycompat import getattr
 from mercurial.thirdparty import attr
 from mercurial import (
     cmdutil,
     context,
     copies,
     encoding,
     error,
     exthelper,
     graphmod,
     httpconnection as httpconnectionmod,
     localrepo,
     logcmdutil,
     match,
     mdiff,
     obsutil,
     parser,
     patch,
     phases,
     pycompat,
     rewriteutil,
     scmutil,
     smartset,
     tags,
     templatefilters,
     templateutil,
     url as urlmod,
     util,
 )
 from mercurial.utils import (
     procutil,
     stringutil,
     urlutil,
 )
 from . import show
 
 
 # Note for extension authors: ONLY specify testedwith = 'ships-with-hg-core' for
 # extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
 # be specifying the version(s) of Mercurial they are tested with, or
 # leave the attribute unspecified.
 testedwith = b'ships-with-hg-core'
 
 eh = exthelper.exthelper()
 
 cmdtable = eh.cmdtable
 command = eh.command
 configtable = eh.configtable
 templatekeyword = eh.templatekeyword
 uisetup = eh.finaluisetup
 
 # developer config: phabricator.batchsize
 eh.configitem(
     b'phabricator',
     b'batchsize',
     default=12,
 )
 eh.configitem(
     b'phabricator',
     b'callsign',
     default=None,
 )
 eh.configitem(
     b'phabricator',
     b'curlcmd',
     default=None,
 )
 # developer config: phabricator.debug
 eh.configitem(
     b'phabricator',
     b'debug',
     default=False,
 )
 # developer config: phabricator.repophid
 eh.configitem(
     b'phabricator',
     b'repophid',
     default=None,
 )
 eh.configitem(
     b'phabricator',
     b'retry',
     default=0,
 )
 eh.configitem(
     b'phabricator',
     b'retry.interval',
     default=1,
 )
 eh.configitem(
     b'phabricator',
     b'url',
     default=None,
 )
 eh.configitem(
     b'phabsend',
     b'confirm',
     default=False,
 )
 eh.configitem(
     b'phabimport',
     b'secret',
     default=False,
 )
 eh.configitem(
     b'phabimport',
     b'obsolete',
     default=False,
 )
 
 colortable = {
     b'phabricator.action.created': b'green',
     b'phabricator.action.skipped': b'magenta',
     b'phabricator.action.updated': b'magenta',
     b'phabricator.drev': b'bold',
     b'phabricator.status.abandoned': b'magenta dim',
     b'phabricator.status.accepted': b'green bold',
     b'phabricator.status.closed': b'green',
     b'phabricator.status.needsreview': b'yellow',
     b'phabricator.status.needsrevision': b'red',
     b'phabricator.status.changesplanned': b'red',
 }
 
 _VCR_FLAGS = [
     (
         b'',
         b'test-vcr',
         b'',
         _(
             b'Path to a vcr file. If nonexistent, will record a new vcr transcript'
             b', otherwise will mock all http requests using the specified vcr file.'
             b' (ADVANCED)'
         ),
     ),
 ]
 
 
 @eh.wrapfunction(localrepo, "loadhgrc")
 def _loadhgrc(orig, ui, wdirvfs, hgvfs, requirements, *args, **opts):
     """Load ``.arcconfig`` content into a ui instance on repository open."""
     result = False
     arcconfig = {}
 
     try:
         # json.loads only accepts bytes from 3.6+
         rawparams = encoding.unifromlocal(wdirvfs.read(b".arcconfig"))
         # json.loads only returns unicode strings
         arcconfig = pycompat.rapply(
             lambda x: encoding.unitolocal(x)
             if isinstance(x, pycompat.unicode)
             else x,
             pycompat.json_loads(rawparams),
         )
 
         result = True
     except ValueError:
         ui.warn(_(b"invalid JSON in %s\n") % wdirvfs.join(b".arcconfig"))
     except IOError:
         pass
 
     cfg = util.sortdict()
 
     if b"repository.callsign" in arcconfig:
         cfg[(b"phabricator", b"callsign")] = arcconfig[b"repository.callsign"]
 
     if b"phabricator.uri" in arcconfig:
         cfg[(b"phabricator", b"url")] = arcconfig[b"phabricator.uri"]
 
     if cfg:
         ui.applyconfig(cfg, source=wdirvfs.join(b".arcconfig"))
 
     return (
         orig(ui, wdirvfs, hgvfs, requirements, *args, **opts) or result
     )  # Load .hg/hgrc
 
 
 def vcrcommand(name, flags, spec, helpcategory=None, optionalrepo=False):
     fullflags = flags + _VCR_FLAGS
 
     def hgmatcher(r1, r2):
         if r1.uri != r2.uri or r1.method != r2.method:
             return False
         r1params = util.urlreq.parseqs(r1.body)
         r2params = util.urlreq.parseqs(r2.body)
         for key in r1params:
             if key not in r2params:
                 return False
             value = r1params[key][0]
             # we want to compare json payloads without worrying about ordering
             if value.startswith(b'{') and value.endswith(b'}'):
                 r1json = pycompat.json_loads(value)
                 r2json = pycompat.json_loads(r2params[key][0])
                 if r1json != r2json:
                     return False
             elif r2params[key][0] != value:
                 return False
         return True
 
     def sanitiserequest(request):
         request.body = re.sub(
             br'cli-[a-z0-9]+', br'cli-hahayouwish', request.body
         )
         return request
 
     def sanitiseresponse(response):
         if 'set-cookie' in response['headers']:
             del response['headers']['set-cookie']
         return response
 
     def decorate(fn):
         def inner(*args, **kwargs):
             vcr = kwargs.pop('test_vcr')
             if vcr:
                 cassette = pycompat.fsdecode(vcr)
                 import hgdemandimport
 
                 with hgdemandimport.deactivated():
                     import vcr as vcrmod
                     import vcr.stubs as stubs
 
                     vcr = vcrmod.VCR(
                         serializer='json',
                         before_record_request=sanitiserequest,
                         before_record_response=sanitiseresponse,
                         custom_patches=[
                             (
                                 urlmod,
                                 'httpconnection',
                                 stubs.VCRHTTPConnection,
                             ),
                             (
                                 urlmod,
                                 'httpsconnection',
                                 stubs.VCRHTTPSConnection,
                             ),
                         ],
                     )
                     vcr.register_matcher('hgmatcher', hgmatcher)
                     with vcr.use_cassette(cassette, match_on=['hgmatcher']):
                         return fn(*args, **kwargs)
             return fn(*args, **kwargs)
 
         cmd = util.checksignature(inner, depth=2)
         cmd.__name__ = fn.__name__
         cmd.__doc__ = fn.__doc__
 
         return command(
             name,
             fullflags,
             spec,
             helpcategory=helpcategory,
             optionalrepo=optionalrepo,
         )(cmd)
 
     return decorate
 
 
 def _debug(ui, *msg, **opts):
     """write debug output for Phabricator if ``phabricator.debug`` is set
 
     Specifically, this avoids dumping Conduit and HTTP auth chatter that is
     printed with the --debug argument.
     """
     if ui.configbool(b"phabricator", b"debug"):
         flag = ui.debugflag
         try:
             ui.debugflag = True
             ui.write(*msg, **opts)
         finally:
             ui.debugflag = flag
 
 
 def urlencodenested(params):
     """like urlencode, but works with nested parameters.
 
     For example, if params is {'a': ['b', 'c'], 'd': {'e': 'f'}}, it will be
     flattened to {'a[0]': 'b', 'a[1]': 'c', 'd[e]': 'f'} and then passed to
     urlencode. Note: the encoding is consistent with PHP's http_build_query.
     """
     flatparams = util.sortdict()
 
     def process(prefix, obj):
         if isinstance(obj, bool):
             obj = {True: b'true', False: b'false'}[obj]  # Python -> PHP form
         lister = lambda l: [(b'%d' % k, v) for k, v in enumerate(l)]
         items = {list: lister, dict: lambda x: x.items()}.get(type(obj))
         if items is None:
             flatparams[prefix] = obj
         else:
             for k, v in items(obj):
                 if prefix:
                     process(b'%s[%s]' % (prefix, k), v)
                 else:
                     process(k, v)
 
     process(b'', params)
     return urlutil.urlreq.urlencode(flatparams)
 
 
 def readurltoken(ui):
     """return conduit url, token and make sure they exist
 
     Currently read from [auth] config section. In the future, it might
     make sense to read from .arcconfig and .arcrc as well.
     """
     url = ui.config(b'phabricator', b'url')
     if not url:
         raise error.Abort(
             _(b'config %s.%s is required') % (b'phabricator', b'url')
         )
 
     res = httpconnectionmod.readauthforuri(ui, url, urlutil.url(url).user)
     token = None
 
     if res:
         group, auth = res
 
         ui.debug(b"using auth.%s.* for authentication\n" % group)
 
         token = auth.get(b'phabtoken')
 
     if not token:
         raise error.Abort(
             _(b'Can\'t find conduit token associated to %s') % (url,)
         )
 
     return url, token
 
 
 def callconduit(ui, name, params):
     """call Conduit API, params is a dict. return json.loads result, or None"""
     host, token = readurltoken(ui)
     url, authinfo = urlutil.url(b'/'.join([host, b'api', name])).authinfo()
     ui.debug(b'Conduit Call: %s %s\n' % (url, pycompat.byterepr(params)))
     params = params.copy()
     params[b'__conduit__'] = {
         b'token': token,
     }
     rawdata = {
         b'params': templatefilters.json(params),
         b'output': b'json',
         b'__conduit__': 1,
     }
     data = urlencodenested(rawdata)
     curlcmd = ui.config(b'phabricator', b'curlcmd')
     if curlcmd:
         sin, sout = procutil.popen2(
             b'%s -d @- %s' % (curlcmd, procutil.shellquote(url))
         )
         sin.write(data)
         sin.close()
         body = sout.read()
     else:
         urlopener = urlmod.opener(ui, authinfo)
         request = util.urlreq.request(pycompat.strurl(url), data=data)
         max_try = ui.configint(b'phabricator', b'retry') + 1
         timeout = ui.configwith(float, b'http', b'timeout')
         for try_count in range(max_try):
             try:
                 with contextlib.closing(
                     urlopener.open(request, timeout=timeout)
                 ) as rsp:
                     body = rsp.read()
                 break
             except util.urlerr.urlerror as err:
                 if try_count == max_try - 1:
                     raise
                 ui.debug(
                     b'Conduit Request failed (try %d/%d): %r\n'
                     % (try_count + 1, max_try, err)
                 )
                 # failing request might come from overloaded server
                 retry_interval = ui.configint(b'phabricator', b'retry.interval')
                 time.sleep(retry_interval)
     ui.debug(b'Conduit Response: %s\n' % body)
     parsed = pycompat.rapply(
         lambda x: encoding.unitolocal(x)
         if isinstance(x, pycompat.unicode)
         else x,
         # json.loads only accepts bytes from py3.6+
         pycompat.json_loads(encoding.unifromlocal(body)),
     )
     if parsed.get(b'error_code'):
         msg = _(b'Conduit Error (%s): %s') % (
             parsed[b'error_code'],
             parsed[b'error_info'],
         )
         raise error.Abort(msg)
     return parsed[b'result']
 
 
 @vcrcommand(b'debugcallconduit', [], _(b'METHOD'), optionalrepo=True)
 def debugcallconduit(ui, repo, name):
     """call Conduit API
 
     Call parameters are read from stdin as a JSON blob. Result will be written
     to stdout as a JSON blob.
     """
     # json.loads only accepts bytes from 3.6+
     rawparams = encoding.unifromlocal(ui.fin.read())
     # json.loads only returns unicode strings
     params = pycompat.rapply(
         lambda x: encoding.unitolocal(x)
         if isinstance(x, pycompat.unicode)
         else x,
         pycompat.json_loads(rawparams),
     )
     # json.dumps only accepts unicode strings
     result = pycompat.rapply(
         lambda x: encoding.unifromlocal(x) if isinstance(x, bytes) else x,
         callconduit(ui, name, params),
     )
     s = json.dumps(result, sort_keys=True, indent=2, separators=(u',', u': '))
     ui.write(b'%s\n' % encoding.unitolocal(s))
 
 
 def getrepophid(repo):
     """given callsign, return repository PHID or None"""
     # developer config: phabricator.repophid
     repophid = repo.ui.config(b'phabricator', b'repophid')
     if repophid:
         return repophid
     callsign = repo.ui.config(b'phabricator', b'callsign')
     if not callsign:
         return None
     query = callconduit(
         repo.ui,
         b'diffusion.repository.search',
         {b'constraints': {b'callsigns': [callsign]}},
     )
     if len(query[b'data']) == 0:
         return None
     repophid = query[b'data'][0][b'phid']
     repo.ui.setconfig(b'phabricator', b'repophid', repophid)
     return repophid
 
 
 _differentialrevisiontagre = re.compile(br'\AD([1-9][0-9]*)\Z')
 _differentialrevisiondescre = re.compile(
     br'^Differential Revision:\s*(?P<url>(?:.*)D(?P<id>[1-9][0-9]*))$', re.M
 )
 
 
 def getoldnodedrevmap(repo, nodelist):
     """find previous nodes that has been sent to Phabricator
 
     return {node: (oldnode, Differential diff, Differential Revision ID)}
     for node in nodelist with known previous sent versions, or associated
     Differential Revision IDs. ``oldnode`` and ``Differential diff`` could
     be ``None``.
 
     Examines commit messages like "Differential Revision:" to get the
     association information.
 
     If such commit message line is not found, examines all precursors and their
     tags. Tags with format like "D1234" are considered a match and the node
     with that tag, and the number after "D" (ex. 1234) will be returned.
 
     The ``old node``, if not None, is guaranteed to be the last diff of
     corresponding Differential Revision, and exist in the repo.
     """
     unfi = repo.unfiltered()
     has_node = unfi.changelog.index.has_node
 
     result = {}  # {node: (oldnode?, lastdiff?, drev)}
     # ordered for test stability when printing new -> old mapping below
     toconfirm = util.sortdict()  # {node: (force, {precnode}, drev)}
     for node in nodelist:
         ctx = unfi[node]
         # For tags like "D123", put them into "toconfirm" to verify later
         precnodes = list(obsutil.allpredecessors(unfi.obsstore, [node]))
         for n in precnodes:
             if has_node(n):
                 for tag in unfi.nodetags(n):
                     m = _differentialrevisiontagre.match(tag)
                     if m:
                         toconfirm[node] = (0, set(precnodes), int(m.group(1)))
                         break
                 else:
                     continue  # move to next predecessor
                 break  # found a tag, stop
         else:
             # Check commit message
             m = _differentialrevisiondescre.search(ctx.description())
             if m:
                 toconfirm[node] = (1, set(precnodes), int(m.group('id')))
 
     # Double check if tags are genuine by collecting all old nodes from
     # Phabricator, and expect precursors overlap with it.
     if toconfirm:
         drevs = [drev for force, precs, drev in toconfirm.values()]
         alldiffs = callconduit(
             unfi.ui, b'differential.querydiffs', {b'revisionIDs': drevs}
         )
 
         def getnodes(d, precset):
             # Ignore other nodes that were combined into the Differential
             # that aren't predecessors of the current local node.
             return [n for n in getlocalcommits(d) if n in precset]
 
         for newnode, (force, precset, drev) in toconfirm.items():
             diffs = [
                 d for d in alldiffs.values() if int(d[b'revisionID']) == drev
             ]
 
             # local predecessors known by Phabricator
             phprecset = {n for d in diffs for n in getnodes(d, precset)}
 
             # Ignore if precursors (Phabricator and local repo) do not overlap,
             # and force is not set (when commit message says nothing)
             if not force and not phprecset:
                 tagname = b'D%d' % drev
                 tags.tag(
                     repo,
                     tagname,
                     repo.nullid,
                     message=None,
                     user=None,
                     date=None,
                     local=True,
                 )
                 unfi.ui.warn(
                     _(
                         b'D%d: local tag removed - does not match '
                         b'Differential history\n'
                     )
                     % drev
                 )
                 continue
 
             # Find the last node using Phabricator metadata, and make sure it
             # exists in the repo
             oldnode = lastdiff = None
             if diffs:
                 lastdiff = max(diffs, key=lambda d: int(d[b'id']))
                 oldnodes = getnodes(lastdiff, precset)
 
                 _debug(
                     unfi.ui,
                     b"%s mapped to old nodes %s\n"
                     % (
                         short(newnode),
                         stringutil.pprint([short(n) for n in sorted(oldnodes)]),
                     ),
                 )
 
                 # If this commit was the result of `hg fold` after submission,
                 # and now resubmitted with --fold, the easiest thing to do is
                 # to leave the node clear.  This only results in creating a new
                 # diff for the _same_ Differential Revision if this commit is
                 # the first or last in the selected range.  If we picked a node
                 # from the list instead, it would have to be the lowest if at
                 # the beginning of the --fold range, or the highest at the end.
                 # Otherwise, one or more of the nodes wouldn't be considered in
                 # the diff, and the Differential wouldn't be properly updated.
                 # If this commit is the result of `hg split` in the same
                 # scenario, there is a single oldnode here (and multiple
                 # newnodes mapped to it).  That makes it the same as the normal
                 # case, as the edges of the newnode range cleanly maps to one
                 # oldnode each.
                 if len(oldnodes) == 1:
                     oldnode = oldnodes[0]
                 if oldnode and not has_node(oldnode):
                     oldnode = None
 
             result[newnode] = (oldnode, lastdiff, drev)
 
     return result
 
 
 def getdrevmap(repo, revs):
     """Return a dict mapping each rev in `revs` to their Differential Revision
     ID or None.
     """
     result = {}
     for rev in revs:
         result[rev] = None
         ctx = repo[rev]
         # Check commit message
         m = _differentialrevisiondescre.search(ctx.description())
         if m:
             result[rev] = int(m.group('id'))
             continue
         # Check tags
         for tag in repo.nodetags(ctx.node()):
             m = _differentialrevisiontagre.match(tag)
             if m:
                 result[rev] = int(m.group(1))
                 break
 
     return result
 
 
 def getdiff(basectx, ctx, diffopts):
     """plain-text diff without header (user, commit message, etc)"""
     output = util.stringio()
     for chunk, _label in patch.diffui(
         ctx.repo(), basectx.p1().node(), ctx.node(), None, opts=diffopts
     ):
         output.write(chunk)
     return output.getvalue()
 
 
 class DiffChangeType(object):
     ADD = 1
     CHANGE = 2
     DELETE = 3
     MOVE_AWAY = 4
     COPY_AWAY = 5
     MOVE_HERE = 6
     COPY_HERE = 7
     MULTICOPY = 8
 
 
 class DiffFileType(object):
     TEXT = 1
     IMAGE = 2
     BINARY = 3
 
 
 @attr.s
 class phabhunk(dict):
     """Represents a Differential hunk, which is owned by a Differential change"""
 
     oldOffset = attr.ib(default=0)  # camelcase-required
     oldLength = attr.ib(default=0)  # camelcase-required
     newOffset = attr.ib(default=0)  # camelcase-required
     newLength = attr.ib(default=0)  # camelcase-required
     corpus = attr.ib(default='')
     # These get added to the phabchange's equivalents
     addLines = attr.ib(default=0)  # camelcase-required
     delLines = attr.ib(default=0)  # camelcase-required
 
 
 @attr.s
 class phabchange(object):
     """Represents a Differential change, owns Differential hunks and owned by a
     Differential diff.  Each one represents one file in a diff.
     """
 
     currentPath = attr.ib(default=None)  # camelcase-required
     oldPath = attr.ib(default=None)  # camelcase-required
     awayPaths = attr.ib(default=attr.Factory(list))  # camelcase-required
     metadata = attr.ib(default=attr.Factory(dict))
     oldProperties = attr.ib(default=attr.Factory(dict))  # camelcase-required
     newProperties = attr.ib(default=attr.Factory(dict))  # camelcase-required
     type = attr.ib(default=DiffChangeType.CHANGE)
     fileType = attr.ib(default=DiffFileType.TEXT)  # camelcase-required
     commitHash = attr.ib(default=None)  # camelcase-required
     addLines = attr.ib(default=0)  # camelcase-required
     delLines = attr.ib(default=0)  # camelcase-required
     hunks = attr.ib(default=attr.Factory(list))
 
     def copynewmetadatatoold(self):
         for key in list(self.metadata.keys()):
             newkey = key.replace(b'new:', b'old:')
             self.metadata[newkey] = self.metadata[key]
 
     def addoldmode(self, value):
         self.oldProperties[b'unix:filemode'] = value
 
     def addnewmode(self, value):
         self.newProperties[b'unix:filemode'] = value
 
     def addhunk(self, hunk):
         if not isinstance(hunk, phabhunk):
             raise error.Abort(b'phabchange.addhunk only takes phabhunks')
         self.hunks.append(pycompat.byteskwargs(attr.asdict(hunk)))
         # It's useful to include these stats since the Phab web UI shows them,
         # and uses them to estimate how large a change a Revision is. Also used
         # in email subjects for the [+++--] bit.
         self.addLines += hunk.addLines
         self.delLines += hunk.delLines
 
 
 @attr.s
 class phabdiff(object):
     """Represents a Differential diff, owns Differential changes.  Corresponds
     to a commit.
     """
 
     # Doesn't seem to be any reason to send this (output of uname -n)
     sourceMachine = attr.ib(default=b'')  # camelcase-required
     sourcePath = attr.ib(default=b'/')  # camelcase-required
     sourceControlBaseRevision = attr.ib(default=b'0' * 40)  # camelcase-required
     sourceControlPath = attr.ib(default=b'/')  # camelcase-required
     sourceControlSystem = attr.ib(default=b'hg')  # camelcase-required
     branch = attr.ib(default=b'default')
     bookmark = attr.ib(default=None)
     creationMethod = attr.ib(default=b'phabsend')  # camelcase-required
     lintStatus = attr.ib(default=b'none')  # camelcase-required
     unitStatus = attr.ib(default=b'none')  # camelcase-required
     changes = attr.ib(default=attr.Factory(dict))
     repositoryPHID = attr.ib(default=None)  # camelcase-required
 
     def addchange(self, change):
         if not isinstance(change, phabchange):
             raise error.Abort(b'phabdiff.addchange only takes phabchanges')
         self.changes[change.currentPath] = pycompat.byteskwargs(
             attr.asdict(change)
         )
 
 
 def maketext(pchange, basectx, ctx, fname):
     """populate the phabchange for a text file"""
     repo = ctx.repo()
     fmatcher = match.exact([fname])
     diffopts = mdiff.diffopts(git=True, context=32767)
     _pfctx, _fctx, header, fhunks = next(
         patch.diffhunks(repo, basectx.p1(), ctx, fmatcher, opts=diffopts)
     )
 
     for fhunk in fhunks:
         (oldOffset, oldLength, newOffset, newLength), lines = fhunk
         corpus = b''.join(lines[1:])
         shunk = list(header)
         shunk.extend(lines)
         _mf, _mt, addLines, delLines, _hb = patch.diffstatsum(
             patch.diffstatdata(util.iterlines(shunk))
         )
         pchange.addhunk(
             phabhunk(
                 oldOffset,
                 oldLength,
                 newOffset,
                 newLength,
                 corpus,
                 addLines,
                 delLines,
             )
         )
 
 
 def uploadchunks(fctx, fphid):
     """upload large binary files as separate chunks.
     Phab requests chunking over 8MiB, and splits into 4MiB chunks
     """
     ui = fctx.repo().ui
     chunks = callconduit(ui, b'file.querychunks', {b'filePHID': fphid})
     with ui.makeprogress(
         _(b'uploading file chunks'), unit=_(b'chunks'), total=len(chunks)
     ) as progress:
         for chunk in chunks:
             progress.increment()
             if chunk[b'complete']:
                 continue
             bstart = int(chunk[b'byteStart'])
             bend = int(chunk[b'byteEnd'])
             callconduit(
                 ui,
                 b'file.uploadchunk',
                 {
                     b'filePHID': fphid,
                     b'byteStart': bstart,
                     b'data': base64.b64encode(fctx.data()[bstart:bend]),
                     b'dataEncoding': b'base64',
                 },
             )
 
 
 def uploadfile(fctx):
     """upload binary files to Phabricator"""
     repo = fctx.repo()
     ui = repo.ui
     fname = fctx.path()
     size = fctx.size()
     fhash = pycompat.bytestr(hashlib.sha256(fctx.data()).hexdigest())
 
     # an allocate call is required first to see if an upload is even required
     # (Phab might already have it) and to determine if chunking is needed
     allocateparams = {
         b'name': fname,
         b'contentLength': size,
         b'contentHash': fhash,
     }
     filealloc = callconduit(ui, b'file.allocate', allocateparams)
     fphid = filealloc[b'filePHID']
 
     if filealloc[b'upload']:
         ui.write(_(b'uploading %s\n') % bytes(fctx))
         if not fphid:
             uploadparams = {
                 b'name': fname,
                 b'data_base64': base64.b64encode(fctx.data()),
             }
             fphid = callconduit(ui, b'file.upload', uploadparams)
         else:
             uploadchunks(fctx, fphid)
     else:
         ui.debug(b'server already has %s\n' % bytes(fctx))
 
     if not fphid:
         raise error.Abort(b'Upload of %s failed.' % bytes(fctx))
 
     return fphid
 
 
 def addoldbinary(pchange, oldfctx, fctx):
     """add the metadata for the previous version of a binary file to the
     phabchange for the new version
 
     ``oldfctx`` is the previous version of the file; ``fctx`` is the new
     version of the file, or None if the file is being removed.
     """
     if not fctx or fctx.cmp(oldfctx):
         # Files differ, add the old one
         pchange.metadata[b'old:file:size'] = oldfctx.size()
         mimeguess, _enc = mimetypes.guess_type(
             encoding.unifromlocal(oldfctx.path())
         )
         if mimeguess:
             pchange.metadata[b'old:file:mime-type'] = pycompat.bytestr(
                 mimeguess
             )
         fphid = uploadfile(oldfctx)
         pchange.metadata[b'old:binary-phid'] = fphid
     else:
         # If it's left as IMAGE/BINARY web UI might try to display it
         pchange.fileType = DiffFileType.TEXT
         pchange.copynewmetadatatoold()
 
 
 def makebinary(pchange, fctx):
     """populate the phabchange for a binary file"""
     pchange.fileType = DiffFileType.BINARY
     fphid = uploadfile(fctx)
     pchange.metadata[b'new:binary-phid'] = fphid
     pchange.metadata[b'new:file:size'] = fctx.size()
     mimeguess, _enc = mimetypes.guess_type(encoding.unifromlocal(fctx.path()))
     if mimeguess:
         mimeguess = pycompat.bytestr(mimeguess)
         pchange.metadata[b'new:file:mime-type'] = mimeguess
         if mimeguess.startswith(b'image/'):
             pchange.fileType = DiffFileType.IMAGE
 
 
 # Copied from mercurial/patch.py
 gitmode = {b'l': b'120000', b'x': b'100755', b'': b'100644'}
 
 
 def notutf8(fctx):
     """detect non-UTF-8 text files since Phabricator requires them to be marked
     as binary
     """
     try:
         fctx.data().decode('utf-8')
         return False
     except UnicodeDecodeError:
         fctx.repo().ui.write(
             _(b'file %s detected as non-UTF-8, marked as binary\n')
             % fctx.path()
         )
         return True
 
 
 def addremoved(pdiff, basectx, ctx, removed):
     """add removed files to the phabdiff. Shouldn't include moves"""
     for fname in removed:
         pchange = phabchange(
             currentPath=fname, oldPath=fname, type=DiffChangeType.DELETE
         )
         oldfctx = basectx.p1()[fname]
         pchange.addoldmode(gitmode[oldfctx.flags()])
         if not (oldfctx.isbinary() or notutf8(oldfctx)):
             maketext(pchange, basectx, ctx, fname)
 
         pdiff.addchange(pchange)
 
 
 def addmodified(pdiff, basectx, ctx, modified):
     """add modified files to the phabdiff"""
     for fname in modified:
         fctx = ctx[fname]
         oldfctx = basectx.p1()[fname]
         pchange = phabchange(currentPath=fname, oldPath=fname)
         filemode = gitmode[fctx.flags()]
         originalmode = gitmode[oldfctx.flags()]
         if filemode != originalmode:
             pchange.addoldmode(originalmode)
             pchange.addnewmode(filemode)
 
         if (
             fctx.isbinary()
             or notutf8(fctx)
             or oldfctx.isbinary()
             or notutf8(oldfctx)
         ):
             makebinary(pchange, fctx)
             addoldbinary(pchange, oldfctx, fctx)
         else:
             maketext(pchange, basectx, ctx, fname)
 
         pdiff.addchange(pchange)
 
 
 def addadded(pdiff, basectx, ctx, added, removed):
     """add file adds to the phabdiff, both new files and copies/moves"""
     # Keep track of files that've been recorded as moved/copied, so if there are
     # additional copies we can mark them (moves get removed from removed)
     copiedchanges = {}
     movedchanges = {}
 
     copy = {}
     if basectx != ctx:
         copy = copies.pathcopies(basectx.p1(), ctx)
 
     for fname in added:
         fctx = ctx[fname]
         oldfctx = None
         pchange = phabchange(currentPath=fname)
 
         filemode = gitmode[fctx.flags()]
 
         if copy:
             originalfname = copy.get(fname, fname)
         else:
             originalfname = fname
             if fctx.renamed():
                 originalfname = fctx.renamed()[0]
 
         renamed = fname != originalfname
 
         if renamed:
             oldfctx = basectx.p1()[originalfname]
             originalmode = gitmode[oldfctx.flags()]
             pchange.oldPath = originalfname
 
             if originalfname in removed:
                 origpchange = phabchange(
                     currentPath=originalfname,
                     oldPath=originalfname,
                     type=DiffChangeType.MOVE_AWAY,
                     awayPaths=[fname],
                 )
                 movedchanges[originalfname] = origpchange
                 removed.remove(originalfname)
                 pchange.type = DiffChangeType.MOVE_HERE
             elif originalfname in movedchanges:
                 movedchanges[originalfname].type = DiffChangeType.MULTICOPY
                 movedchanges[originalfname].awayPaths.append(fname)
                 pchange.type = DiffChangeType.COPY_HERE
             else:  # pure copy
                 if originalfname not in copiedchanges:
                     origpchange = phabchange(
                         currentPath=originalfname, type=DiffChangeType.COPY_AWAY
                     )
                     copiedchanges[originalfname] = origpchange
                 else:
                     origpchange = copiedchanges[originalfname]
                 origpchange.awayPaths.append(fname)
                 pchange.type = DiffChangeType.COPY_HERE
 
             if filemode != originalmode:
                 pchange.addoldmode(originalmode)
                 pchange.addnewmode(filemode)
         else:  # Brand-new file
             pchange.addnewmode(gitmode[fctx.flags()])
             pchange.type = DiffChangeType.ADD
 
         if (
             fctx.isbinary()
             or notutf8(fctx)
             or (oldfctx and (oldfctx.isbinary() or notutf8(oldfctx)))
         ):
             makebinary(pchange, fctx)
             if renamed:
                 addoldbinary(pchange, oldfctx, fctx)
         else:
             maketext(pchange, basectx, ctx, fname)
 
         pdiff.addchange(pchange)
 
     for _path, copiedchange in copiedchanges.items():
         pdiff.addchange(copiedchange)
     for _path, movedchange in movedchanges.items():
         pdiff.addchange(movedchange)
 
 
 def creatediff(basectx, ctx):
     """create a Differential Diff"""
     repo = ctx.repo()
     repophid = getrepophid(repo)
     # Create a "Differential Diff" via "differential.creatediff" API
     pdiff = phabdiff(
         sourceControlBaseRevision=b'%s' % basectx.p1().hex(),
         branch=b'%s' % ctx.branch(),
     )
     modified, added, removed, _d, _u, _i, _c = basectx.p1().status(ctx)
     # addadded will remove moved files from removed, so addremoved won't get
     # them
     addadded(pdiff, basectx, ctx, added, removed)
     addmodified(pdiff, basectx, ctx, modified)
     addremoved(pdiff, basectx, ctx, removed)
     if repophid:
         pdiff.repositoryPHID = repophid
     diff = callconduit(
         repo.ui,
         b'differential.creatediff',
         pycompat.byteskwargs(attr.asdict(pdiff)),
     )
     if not diff:
         if basectx != ctx:
             msg = _(b'cannot create diff for %s::%s') % (basectx, ctx)
         else:
             msg = _(b'cannot create diff for %s') % ctx
         raise error.Abort(msg)
     return diff
 
 
 def writediffproperties(ctxs, diff):
     """write metadata to diff so patches could be applied losslessly
 
     ``ctxs`` is the list of commits that created the diff, in ascending order.
     The list is generally a single commit, but may be several when using
     ``phabsend --fold``.
     """
     # creatediff returns with a diffid but query returns with an id
     diffid = diff.get(b'diffid', diff.get(b'id'))
     basectx = ctxs[0]
     tipctx = ctxs[-1]
 
     params = {
         b'diff_id': diffid,
         b'name': b'hg:meta',
         b'data': templatefilters.json(
             {
                 b'user': tipctx.user(),
                 b'date': b'%d %d' % tipctx.date(),
                 b'branch': tipctx.branch(),
                 b'node': tipctx.hex(),
                 b'parent': basectx.p1().hex(),
             }
         ),
     }
     callconduit(basectx.repo().ui, b'differential.setdiffproperty', params)
 
     commits = {}
     for ctx in ctxs:
         commits[ctx.hex()] = {
             b'author': stringutil.person(ctx.user()),
             b'authorEmail': stringutil.email(ctx.user()),
             b'time': int(ctx.date()[0]),
             b'commit': ctx.hex(),
             b'parents': [ctx.p1().hex()],
             b'branch': ctx.branch(),
         }
     params = {
         b'diff_id': diffid,
         b'name': b'local:commits',
         b'data': templatefilters.json(commits),
     }
     callconduit(basectx.repo().ui, b'differential.setdiffproperty', params)
 
 
 def createdifferentialrevision(
     ctxs,
     revid=None,
     parentrevphid=None,
     oldbasenode=None,
     oldnode=None,
     olddiff=None,
     actions=None,
     comment=None,
 ):
     """create or update a Differential Revision
 
     If revid is None, create a new Differential Revision, otherwise update
     revid. If parentrevphid is not None, set it as a dependency.
 
     If there is a single commit for the new Differential Revision, ``ctxs`` will
     be a list of that single context.  Otherwise, it is a list that covers the
     range of changes for the differential, where ``ctxs[0]`` is the first change
     to include and ``ctxs[-1]`` is the last.
 
     If oldnode is not None, check if the patch content (without commit message
     and metadata) has changed before creating another diff.  For a Revision with
     a single commit, ``oldbasenode`` and ``oldnode`` have the same value.  For a
     Revision covering multiple commits, ``oldbasenode`` corresponds to
     ``ctxs[0]`` the previous time this Revision was posted, and ``oldnode``
     corresponds to ``ctxs[-1]``.
 
     If actions is not None, they will be appended to the transaction.
     """
     ctx = ctxs[-1]
     basectx = ctxs[0]
 
     repo = ctx.repo()
     if oldnode:
         diffopts = mdiff.diffopts(git=True, context=32767)
         unfi = repo.unfiltered()
         oldctx = unfi[oldnode]
         oldbasectx = unfi[oldbasenode]
         neednewdiff = getdiff(basectx, ctx, diffopts) != getdiff(
             oldbasectx, oldctx, diffopts
         )
     else:
         neednewdiff = True
 
     transactions = []
     if neednewdiff:
         diff = creatediff(basectx, ctx)
         transactions.append({b'type': b'update', b'value': diff[b'phid']})
         if comment:
             transactions.append({b'type': b'comment', b'value': comment})
     else:
         # Even if we don't need to upload a new diff because the patch content
         # does not change. We might still need to update its metadata so
         # pushers could know the correct node metadata.
         assert olddiff
         diff = olddiff
     writediffproperties(ctxs, diff)
 
     # Set the parent Revision every time, so commit re-ordering is picked-up
     if parentrevphid:
         transactions.append(
             {b'type': b'parents.set', b'value': [parentrevphid]}
         )
 
     if actions:
         transactions += actions
 
     # When folding multiple local commits into a single review, arcanist will
     # take the summary line of the first commit as the title, and then
     # concatenate the rest of the remaining messages (including each of their
     # first lines) to the rest of the first commit message (each separated by
     # an empty line), and use that as the summary field.  Do the same here.
     # For commits with only a one line message, there is no summary field, as
     # this gets assigned to the title.
     fields = util.sortdict()  # sorted for stable wire protocol in tests
 
     for i, _ctx in enumerate(ctxs):
         # Parse commit message and update related fields.
         desc = _ctx.description()
         info = callconduit(
             repo.ui, b'differential.parsecommitmessage', {b'corpus': desc}
         )
 
         for k in [b'title', b'summary', b'testPlan']:
             v = info[b'fields'].get(k)
             if not v:
                 continue
 
             if i == 0:
                 # Title, summary and test plan (if present) are taken verbatim
                 # for the first commit.
                 fields[k] = v.rstrip()
                 continue
             elif k == b'title':
                 # Add subsequent titles (i.e. the first line of the commit
                 # message) back to the summary.
                 k = b'summary'
 
             # Append any current field to the existing composite field
             fields[k] = b'\n\n'.join(filter(None, [fields.get(k), v.rstrip()]))
 
     for k, v in fields.items():
         transactions.append({b'type': k, b'value': v})
 
     params = {b'transactions': transactions}
     if revid is not None:
         # Update an existing Differential Revision
         params[b'objectIdentifier'] = revid
 
     revision = callconduit(repo.ui, b'differential.revision.edit', params)
     if not revision:
         if len(ctxs) == 1:
             msg = _(b'cannot create revision for %s') % ctx
         else:
             msg = _(b'cannot create revision for %s::%s') % (basectx, ctx)
         raise error.Abort(msg)
 
     return revision, diff
 
 
 def userphids(ui, names):
     """convert user names to PHIDs"""
     names = [name.lower() for name in names]
     query = {b'constraints': {b'usernames': names}}
     result = callconduit(ui, b'user.search', query)
     # username not found is not an error of the API. So check if we have missed
     # some names here.
     data = result[b'data']
     resolved = {entry[b'fields'][b'username'].lower() for entry in data}
     unresolved = set(names) - resolved
     if unresolved:
         raise error.Abort(
             _(b'unknown username: %s') % b' '.join(sorted(unresolved))
         )
     return [entry[b'phid'] for entry in data]
 
 
 def _print_phabsend_action(ui, ctx, newrevid, action):
     """print the ``action`` that occurred when posting ``ctx`` for review
 
     This is a utility function for the sending phase of ``phabsend``, which
     makes it easier to show a status for all local commits with `--fold``.
     """
     actiondesc = ui.label(
         {
             b'created': _(b'created'),
             b'skipped': _(b'skipped'),
             b'updated': _(b'updated'),
         }[action],
         b'phabricator.action.%s' % action,
     )
     drevdesc = ui.label(b'D%d' % newrevid, b'phabricator.drev')
     summary = cmdutil.format_changeset_summary(ui, ctx, b'phabsend')
     ui.write(_(b'%s - %s - %s\n') % (drevdesc, actiondesc, summary))
 
 
 def _amend_diff_properties(unfi, drevid, newnodes, diff):
     """update the local commit list for the ``diff`` associated with ``drevid``
 
     This is a utility function for the amend phase of ``phabsend``, which
     converts failures to warning messages.
     """
     _debug(
         unfi.ui,
         b"new commits: %s\n" % stringutil.pprint([short(n) for n in newnodes]),
     )
 
     try:
         writediffproperties([unfi[newnode] for newnode in newnodes], diff)
     except util.urlerr.urlerror:
         # If it fails just warn and keep going, otherwise the DREV
         # associations will be lost
         unfi.ui.warnnoi18n(b'Failed to update metadata for D%d\n' % drevid)
 
 
 @vcrcommand(
     b'phabsend',
     [
         (b'r', b'rev', [], _(b'revisions to send'), _(b'REV')),
         (b'', b'amend', True, _(b'update commit messages')),
         (b'', b'reviewer', [], _(b'specify reviewers')),
         (b'', b'blocker', [], _(b'specify blocking reviewers')),
         (
             b'm',
             b'comment',
             b'',
             _(b'add a comment to Revisions with new/updated Diffs'),
         ),
         (b'', b'confirm', None, _(b'ask for confirmation before sending')),
         (b'', b'fold', False, _(b'combine the revisions into one review')),
     ],
     _(b'REV [OPTIONS]'),
     helpcategory=command.CATEGORY_IMPORT_EXPORT,
 )
 def phabsend(ui, repo, *revs, **opts):
     """upload changesets to Phabricator
 
     If there are multiple revisions specified, they will be send as a stack
     with a linear dependencies relationship using the order specified by the
     revset.
 
     For the first time uploading changesets, local tags will be created to
     maintain the association. After the first time, phabsend will check
     obsstore and tags information so it can figure out whether to update an
     existing Differential Revision, or create a new one.
 
     If --amend is set, update commit messages so they have the
     ``Differential Revision`` URL, remove related tags. This is similar to what
     arcanist will do, and is more desired in author-push workflows. Otherwise,
     use local tags to record the ``Differential Revision`` association.
 
     The --confirm option lets you confirm changesets before sending them. You
     can also add following to your configuration file to make it default
     behaviour::
 
         [phabsend]
         confirm = true
 
     By default, a separate review will be created for each commit that is
     selected, and will have the same parent/child relationship in Phabricator.
     If ``--fold`` is set, multiple commits are rolled up into a single review
     as if diffed from the parent of the first revision to the last.  The commit
     messages are concatenated in the summary field on Phabricator.
 
     phabsend will check obsstore and the above association to decide whether to
     update an existing Differential Revision, or create a new one.
     """
     opts = pycompat.byteskwargs(opts)
     revs = list(revs) + opts.get(b'rev', [])
     revs = logcmdutil.revrange(repo, revs)
     revs.sort()  # ascending order to preserve topological parent/child in phab
 
     if not revs:
         raise error.Abort(_(b'phabsend requires at least one changeset'))
     if opts.get(b'amend'):
         cmdutil.checkunfinished(repo)
 
     ctxs = [repo[rev] for rev in revs]
 
     if any(c for c in ctxs if c.obsolete()):
         raise error.Abort(_(b"obsolete commits cannot be posted for review"))
 
     # Ensure the local commits are an unbroken range.  The semantics of the
     # --fold option implies this, and the auto restacking of orphans requires
     # it.  Otherwise A+C in A->B->C will cause B to be orphaned, and C' to
     # get A' as a parent.
     def _fail_nonlinear_revs(revs, revtype):
         badnodes = [repo[r].node() for r in revs]
         raise error.Abort(
             _(b"cannot phabsend multiple %s revisions: %s")
             % (revtype, scmutil.nodesummaries(repo, badnodes)),
             hint=_(b"the revisions must form a linear chain"),
         )
 
     heads = repo.revs(b'heads(%ld)', revs)
     if len(heads) > 1:
         _fail_nonlinear_revs(heads, b"head")
 
     roots = repo.revs(b'roots(%ld)', revs)
     if len(roots) > 1:
         _fail_nonlinear_revs(roots, b"root")
 
     fold = opts.get(b'fold')
     if fold:
         if len(revs) == 1:
             # TODO: just switch to --no-fold instead?
             raise error.Abort(_(b"cannot fold a single revision"))
 
         # There's no clear way to manage multiple commits with a Dxxx tag, so
         # require the amend option.  (We could append "_nnn", but then it
         # becomes jumbled if earlier commits are added to an update.)  It should
         # lock the repo and ensure that the range is editable, but that would
         # make the code pretty convoluted.  The default behavior of `arc` is to
         # create a new review anyway.
         if not opts.get(b"amend"):
             raise error.Abort(_(b"cannot fold with --no-amend"))
 
         # It might be possible to bucketize the revisions by the DREV value, and
         # iterate over those groups when posting, and then again when amending.
         # But for simplicity, require all selected revisions to be for the same
         # DREV (if present).  Adding local revisions to an existing DREV is
         # acceptable.
         drevmatchers = [
             _differentialrevisiondescre.search(ctx.description())
             for ctx in ctxs
         ]
         if len({m.group('url') for m in drevmatchers if m}) > 1:
             raise error.Abort(
                 _(b"cannot fold revisions with different DREV values")
             )
 
     # {newnode: (oldnode, olddiff, olddrev}
     oldmap = getoldnodedrevmap(repo, [repo[r].node() for r in revs])
 
     confirm = ui.configbool(b'phabsend', b'confirm')
     confirm |= bool(opts.get(b'confirm'))
     if confirm:
         confirmed = _confirmbeforesend(repo, revs, oldmap)
         if not confirmed:
             raise error.Abort(_(b'phabsend cancelled'))
 
     actions = []
     reviewers = opts.get(b'reviewer', [])
     blockers = opts.get(b'blocker', [])
     phids = []
     if reviewers:
         phids.extend(userphids(repo.ui, reviewers))
     if blockers:
         phids.extend(
             map(
                 lambda phid: b'blocking(%s)' % phid,
                 userphids(repo.ui, blockers),
             )
         )
     if phids:
         actions.append({b'type': b'reviewers.add', b'value': phids})
 
     drevids = []  # [int]
     diffmap = {}  # {newnode: diff}
 
     # Send patches one by one so we know their Differential Revision PHIDs and
     # can provide dependency relationship
     lastrevphid = None
     for ctx in ctxs:
         if fold:
             ui.debug(b'sending rev %d::%d\n' % (ctx.rev(), ctxs[-1].rev()))
         else:
             ui.debug(b'sending rev %d\n' % ctx.rev())
 
         # Get Differential Revision ID
         oldnode, olddiff, revid = oldmap.get(ctx.node(), (None, None, None))
         oldbasenode, oldbasediff, oldbaserevid = oldnode, olddiff, revid
 
         if fold:
             oldbasenode, oldbasediff, oldbaserevid = oldmap.get(
                 ctxs[-1].node(), (None, None, None)
             )
 
         if oldnode != ctx.node() or opts.get(b'amend'):
             # Create or update Differential Revision
             revision, diff = createdifferentialrevision(
                 ctxs if fold else [ctx],
                 revid,
                 lastrevphid,
                 oldbasenode,
                 oldnode,
                 olddiff,
                 actions,
                 opts.get(b'comment'),
             )
 
             if fold:
                 for ctx in ctxs:
                     diffmap[ctx.node()] = diff
             else:
                 diffmap[ctx.node()] = diff
 
             newrevid = int(revision[b'object'][b'id'])
             newrevphid = revision[b'object'][b'phid']
             if revid:
                 action = b'updated'
             else:
                 action = b'created'
 
             # Create a local tag to note the association, if commit message
             # does not have it already
             if not fold:
                 m = _differentialrevisiondescre.search(ctx.description())
                 if not m or int(m.group('id')) != newrevid:
                     tagname = b'D%d' % newrevid
                     tags.tag(
                         repo,
                         tagname,
                         ctx.node(),
                         message=None,
                         user=None,
                         date=None,
                         local=True,
                     )
         else:
             # Nothing changed. But still set "newrevphid" so the next revision
             # could depend on this one and "newrevid" for the summary line.
             newrevphid = querydrev(repo.ui, b'%d' % revid)[0][b'phid']
             newrevid = revid
             action = b'skipped'
 
         drevids.append(newrevid)
         lastrevphid = newrevphid
 
         if fold:
             for c in ctxs:
                 if oldmap.get(c.node(), (None, None, None))[2]:
                     action = b'updated'
                 else:
                     action = b'created'
                 _print_phabsend_action(ui, c, newrevid, action)
             break
 
         _print_phabsend_action(ui, ctx, newrevid, action)
 
     # Update commit messages and remove tags
     if opts.get(b'amend'):
         unfi = repo.unfiltered()
         drevs = callconduit(ui, b'differential.query', {b'ids': drevids})
         with repo.wlock(), repo.lock(), repo.transaction(b'phabsend'):
             # Eagerly evaluate commits to restabilize before creating new
             # commits.  The selected revisions are excluded because they are
             # automatically restacked as part of the submission process.
             restack = [
                 c
                 for c in repo.set(
                     b"(%ld::) - (%ld) - unstable() - obsolete() - public()",
                     revs,
                     revs,
                 )
             ]
             wnode = unfi[b'.'].node()
             mapping = {}  # {oldnode: [newnode]}
             newnodes = []
 
             drevid = drevids[0]
 
             for i, rev in enumerate(revs):
                 old = unfi[rev]
                 if not fold:
                     drevid = drevids[i]
                 drev = [d for d in drevs if int(d[b'id']) == drevid][0]
 
                 newdesc = get_amended_desc(drev, old, fold)
                 # Make sure commit message contain "Differential Revision"
                 if (
                     old.description() != newdesc
                     or old.p1().node() in mapping
                     or old.p2().node() in mapping
                 ):
                     if old.phase() == phases.public:
                         ui.warn(
                             _(b"warning: not updating public commit %s\n")
                             % scmutil.formatchangeid(old)
                         )
                         continue
                     parents = [
                         mapping.get(old.p1().node(), (old.p1(),))[0],
                         mapping.get(old.p2().node(), (old.p2(),))[0],
                     ]
                     newdesc = rewriteutil.update_hash_refs(
                         repo,
                         newdesc,
                         mapping,
                     )
                     new = context.metadataonlyctx(
                         repo,
                         old,
                         parents=parents,
                         text=newdesc,
                         user=old.user(),
                         date=old.date(),
                         extra=old.extra(),
                     )
 
                     newnode = new.commit()
 
                     mapping[old.node()] = [newnode]
 
                     if fold:
                         # Defer updating the (single) Diff until all nodes are
                         # collected.  No tags were created, so none need to be
                         # removed.
                         newnodes.append(newnode)
                         continue
 
                     _amend_diff_properties(
                         unfi, drevid, [newnode], diffmap[old.node()]
                     )
 
                     # Remove local tags since it's no longer necessary
                     tagname = b'D%d' % drevid
                     if tagname in repo.tags():
                         tags.tag(
                             repo,
                             tagname,
                             repo.nullid,
                             message=None,
                             user=None,
                             date=None,
                             local=True,
                         )
                 elif fold:
                     # When folding multiple commits into one review with
                     # --fold, track even the commits that weren't amended, so
                     # that their association isn't lost if the properties are
                     # rewritten below.
                     newnodes.append(old.node())
 
             # If the submitted commits are public, no amend takes place so
             # there are no newnodes and therefore no diff update to do.
             if fold and newnodes:
                 diff = diffmap[old.node()]
 
                 # The diff object in diffmap doesn't have the local commits
                 # because that could be returned from differential.creatediff,
                 # not differential.querydiffs.  So use the queried diff (if
                 # present), or force the amend (a new revision is being posted.)
                 if not olddiff or set(newnodes) != getlocalcommits(olddiff):
                     _debug(ui, b"updating local commit list for D%d\n" % drevid)
                     _amend_diff_properties(unfi, drevid, newnodes, diff)
                 else:
                     _debug(
                         ui,
                         b"local commit list for D%d is already up-to-date\n"
                         % drevid,
                     )
             elif fold:
                 _debug(ui, b"no newnodes to update\n")
 
             # Restack any children of first-time submissions that were orphaned
             # in the process.  The ctx won't report that it is an orphan until
             # the cleanup takes place below.
             for old in restack:
                 parents = [
                     mapping.get(old.p1().node(), (old.p1(),))[0],
                     mapping.get(old.p2().node(), (old.p2(),))[0],
                 ]
                 new = context.metadataonlyctx(
                     repo,
                     old,
                     parents=parents,
                     text=rewriteutil.update_hash_refs(
                         repo, old.description(), mapping
                     ),
                     user=old.user(),
                     date=old.date(),
                     extra=old.extra(),
                 )
 
                 newnode = new.commit()
 
                 # Don't obsolete unselected descendants of nodes that have not
                 # been changed in this transaction- that results in an error.
                 if newnode != old.node():
                     mapping[old.node()] = [newnode]
                     _debug(
                         ui,
                         b"restabilizing %s as %s\n"
                         % (short(old.node()), short(newnode)),
                     )
                 else:
                     _debug(
                         ui,
                         b"not restabilizing unchanged %s\n" % short(old.node()),
                     )
 
             scmutil.cleanupnodes(repo, mapping, b'phabsend', fixphase=True)
             if wnode in mapping:
                 unfi.setparents(mapping[wnode][0])
 
 
 # Map from "hg:meta" keys to header understood by "hg import". The order is
 # consistent with "hg export" output.
 _metanamemap = util.sortdict(
     [
         (b'user', b'User'),
         (b'date', b'Date'),
         (b'branch', b'Branch'),
         (b'node', b'Node ID'),
         (b'parent', b'Parent '),
     ]
 )
 
 
 def _confirmbeforesend(repo, revs, oldmap):
     url, token = readurltoken(repo.ui)
     ui = repo.ui
     for rev in revs:
         ctx = repo[rev]
         oldnode, olddiff, drevid = oldmap.get(ctx.node(), (None, None, None))
         if drevid:
             drevdesc = ui.label(b'D%d' % drevid, b'phabricator.drev')
         else:
             drevdesc = ui.label(_(b'NEW'), b'phabricator.drev')
 
         ui.write(
             _(b'%s - %s\n')
             % (
                 drevdesc,
                 cmdutil.format_changeset_summary(ui, ctx, b'phabsend'),
             )
         )
 
     if ui.promptchoice(
         _(b'Send the above changes to %s (Y/n)?$$ &Yes $$ &No') % url
     ):
         return False
 
     return True
 
 
 _knownstatusnames = {
     b'accepted',
     b'needsreview',
     b'needsrevision',
     b'closed',
     b'abandoned',
     b'changesplanned',
 }
 
 
 def _getstatusname(drev):
     """get normalized status name from a Differential Revision"""
     return drev[b'statusName'].replace(b' ', b'').lower()
 
 
 # Small language to specify differential revisions. Support symbols: (), :X,
 # +, and -.
 
 _elements = {
     # token-type: binding-strength, primary, prefix, infix, suffix
     b'(': (12, None, (b'group', 1, b')'), None, None),
     b':': (8, None, (b'ancestors', 8), None, None),
     b'&': (5, None, None, (b'and_', 5), None),
     b'+': (4, None, None, (b'add', 4), None),
     b'-': (4, None, None, (b'sub', 4), None),
     b')': (0, None, None, None, None),
     b'symbol': (0, b'symbol', None, None, None),
     b'end': (0, None, None, None, None),
 }
 
 
 def _tokenize(text):
     view = memoryview(text)  # zero-copy slice
     special = b'():+-& '
     pos = 0
     length = len(text)
     while pos < length:
         symbol = b''.join(
             itertools.takewhile(
                 lambda ch: ch not in special, pycompat.iterbytestr(view[pos:])
             )
         )
         if symbol:
             yield (b'symbol', symbol, pos)
             pos += len(symbol)
         else:  # special char, ignore space
             if text[pos : pos + 1] != b' ':
                 yield (text[pos : pos + 1], None, pos)
             pos += 1
     yield (b'end', None, pos)
 
 
 def _parse(text):
     tree, pos = parser.parser(_elements).parse(_tokenize(text))
     if pos != len(text):
         raise error.ParseError(b'invalid token', pos)
     return tree
 
 
 def _parsedrev(symbol):
     """str -> int or None, ex. 'D45' -> 45; '12' -> 12; 'x' -> None"""
     if symbol.startswith(b'D') and symbol[1:].isdigit():
         return int(symbol[1:])
     if symbol.isdigit():
         return int(symbol)
 
 
 def _prefetchdrevs(tree):
     """return ({single-drev-id}, {ancestor-drev-id}) to prefetch"""
     drevs = set()
     ancestordrevs = set()
     op = tree[0]
     if op == b'symbol':
         r = _parsedrev(tree[1])
         if r:
             drevs.add(r)
     elif op == b'ancestors':
         r, a = _prefetchdrevs(tree[1])
         drevs.update(r)
         ancestordrevs.update(r)
         ancestordrevs.update(a)
     else:
         for t in tree[1:]:
             r, a = _prefetchdrevs(t)
             drevs.update(r)
             ancestordrevs.update(a)
     return drevs, ancestordrevs
 
 
 def querydrev(ui, spec):
     """return a list of "Differential Revision" dicts
 
     spec is a string using a simple query language, see docstring in phabread
     for details.
 
     A "Differential Revision dict" looks like:
 
         {
             "activeDiffPHID": "PHID-DIFF-xoqnjkobbm6k4dk6hi72",
             "authorPHID": "PHID-USER-tv3ohwc4v4jeu34otlye",
             "auxiliary": {
               "phabricator:depends-on": [
                 "PHID-DREV-gbapp366kutjebt7agcd"
               ]
               "phabricator:projects": [],
             },
             "branch": "default",
             "ccs": [],
             "commits": [],
             "dateCreated": "1499181406",
             "dateModified": "1499182103",
             "diffs": [
               "3",
               "4",
             ],
             "hashes": [],
             "id": "2",
             "lineCount": "2",
             "phid": "PHID-DREV-672qvysjcczopag46qty",
             "properties": {},
             "repositoryPHID": "PHID-REPO-hub2hx62ieuqeheznasv",
             "reviewers": [],
             "sourcePath": null
             "status": "0",
             "statusName": "Needs Review",
             "summary": "",
             "testPlan": "",
             "title": "example",
             "uri": "https://phab.example.com/D2",
         }
     """
     # TODO: replace differential.query and differential.querydiffs with
     # differential.diff.search because the former (and their output) are
     # frozen, and planned to be deprecated and removed.
 
     def fetch(params):
         """params -> single drev or None"""
         key = (params.get(b'ids') or params.get(b'phids') or [None])[0]
         if key in prefetched:
             return prefetched[key]
         drevs = callconduit(ui, b'differential.query', params)
         # Fill prefetched with the result
         for drev in drevs:
             prefetched[drev[b'phid']] = drev
             prefetched[int(drev[b'id'])] = drev
         if key not in prefetched:
             raise error.Abort(
                 _(b'cannot get Differential Revision %r') % params
             )
         return prefetched[key]
 
     def getstack(topdrevids):
         """given a top, get a stack from the bottom, [id] -> [id]"""
         visited = set()
         result = []
         queue = [{b'ids': [i]} for i in topdrevids]
         while queue:
             params = queue.pop()
             drev = fetch(params)
             if drev[b'id'] in visited:
                 continue
             visited.add(drev[b'id'])
             result.append(int(drev[b'id']))
             auxiliary = drev.get(b'auxiliary', {})
             depends = auxiliary.get(b'phabricator:depends-on', [])
             for phid in depends:
                 queue.append({b'phids': [phid]})
         result.reverse()
         return smartset.baseset(result)
 
     # Initialize prefetch cache
     prefetched = {}  # {id or phid: drev}
 
     tree = _parse(spec)
     drevs, ancestordrevs = _prefetchdrevs(tree)
 
     # developer config: phabricator.batchsize
     batchsize = ui.configint(b'phabricator', b'batchsize')
 
     # Prefetch Differential Revisions in batch
     tofetch = set(drevs)
     for r in ancestordrevs:
         tofetch.update(range(max(1, r - batchsize), r + 1))
     if drevs:
         fetch({b'ids': list(tofetch)})
     validids = sorted(set(getstack(list(ancestordrevs))) | set(drevs))
 
     # Walk through the tree, return smartsets
     def walk(tree):
         op = tree[0]
         if op == b'symbol':
             drev = _parsedrev(tree[1])
             if drev:
                 return smartset.baseset([drev])
             elif tree[1] in _knownstatusnames:
                 drevs = [
                     r
                     for r in validids
                     if _getstatusname(prefetched[r]) == tree[1]
                 ]
                 return smartset.baseset(drevs)
             else:
                 raise error.Abort(_(b'unknown symbol: %s') % tree[1])
         elif op in {b'and_', b'add', b'sub'}:
             assert len(tree) == 3
             return getattr(operator, op)(walk(tree[1]), walk(tree[2]))
         elif op == b'group':
             return walk(tree[1])
         elif op == b'ancestors':
             return getstack(walk(tree[1]))
         else:
             raise error.ProgrammingError(b'illegal tree: %r' % tree)
 
     return [prefetched[r] for r in walk(tree)]
 
 
 def getdescfromdrev(drev):
     """get description (commit message) from "Differential Revision"
 
     This is similar to differential.getcommitmessage API. But we only care
     about limited fields: title, summary, test plan, and URL.
     """
     title = drev[b'title']
     summary = drev[b'summary'].rstrip()
     testplan = drev[b'testPlan'].rstrip()
     if testplan:
         testplan = b'Test Plan:\n%s' % testplan
     uri = b'Differential Revision: %s' % drev[b'uri']
     return b'\n\n'.join(filter(None, [title, summary, testplan, uri]))
 
 
 def get_amended_desc(drev, ctx, folded):
     """similar to ``getdescfromdrev``, but supports a folded series of commits
 
     This is used when determining if an individual commit needs to have its
     message amended after posting it for review.  The determination is made for
     each individual commit, even when they were folded into one review.
     """
     if not folded:
         return getdescfromdrev(drev)
 
     uri = b'Differential Revision: %s' % drev[b'uri']
 
     # Since the commit messages were combined when posting multiple commits
     # with --fold, the fields can't be read from Phabricator here, or *all*
     # affected local revisions will end up with the same commit message after
     # the URI is amended in.  Append in the DREV line, or update it if it
     # exists.  At worst, this means commit message or test plan updates on
     # Phabricator aren't propagated back to the repository, but that seems
     # reasonable for the case where local commits are effectively combined
     # in Phabricator.
     m = _differentialrevisiondescre.search(ctx.description())
     if not m:
         return b'\n\n'.join([ctx.description(), uri])
 
     return _differentialrevisiondescre.sub(uri, ctx.description())
 
 
 def getlocalcommits(diff):
     """get the set of local commits from a diff object
 
     See ``getdiffmeta()`` for an example diff object.
     """
     props = diff.get(b'properties') or {}
     commits = props.get(b'local:commits') or {}
     if len(commits) > 1:
         return {bin(c) for c in commits.keys()}
 
     # Storing the diff metadata predates storing `local:commits`, so continue
     # to use that in the --no-fold case.
     return {bin(getdiffmeta(diff).get(b'node', b'')) or None}
 
 
 def getdiffmeta(diff):
     """get commit metadata (date, node, user, p1) from a diff object
 
     The metadata could be "hg:meta", sent by phabsend, like:
 
         "properties": {
           "hg:meta": {
             "branch": "default",
             "date": "1499571514 25200",
             "node": "98c08acae292b2faf60a279b4189beb6cff1414d",
             "user": "Foo Bar <[e-mail]>",
             "parent": "6d0abad76b30e4724a37ab8721d630394070fe16"
           }
         }
 
     Or converted from "local:commits", sent by "arc", like:
 
         "properties": {
           "local:commits": {
             "98c08acae292b2faf60a279b4189beb6cff1414d": {
               "author": "Foo Bar",
               "authorEmail": "[e-mail]"
               "branch": "default",
               "commit": "98c08acae292b2faf60a279b4189beb6cff1414d",
               "local": "1000",
               "message": "...",
               "parents": ["6d0abad76b30e4724a37ab8721d630394070fe16"],
               "rev": "98c08acae292b2faf60a279b4189beb6cff1414d",
               "summary": "...",
               "tag": "",
               "time": 1499546314,
             }
           }
         }
 
     Note: metadata extracted from "local:commits" will lose time zone
     information.
     """
     props = diff.get(b'properties') or {}
     meta = props.get(b'hg:meta')
     if not meta:
         if props.get(b'local:commits'):
             commit = sorted(props[b'local:commits'].values())[0]
             meta = {}
             if b'author' in commit and b'authorEmail' in commit:
                 meta[b'user'] = b'%s <%s>' % (
                     commit[b'author'],
                     commit[b'authorEmail'],
                 )
             if b'time' in commit:
                 meta[b'date'] = b'%d 0' % int(commit[b'time'])
             if b'branch' in commit:
                 meta[b'branch'] = commit[b'branch']
             node = commit.get(b'commit', commit.get(b'rev'))
             if node:
                 meta[b'node'] = node
             if len(commit.get(b'parents', ())) >= 1:
                 meta[b'parent'] = commit[b'parents'][0]
         else:
             meta = {}
     if b'date' not in meta and b'dateCreated' in diff:
         meta[b'date'] = b'%s 0' % diff[b'dateCreated']
     if b'branch' not in meta and diff.get(b'branch'):
         meta[b'branch'] = diff[b'branch']
     if b'parent' not in meta and diff.get(b'sourceControlBaseRevision'):
         meta[b'parent'] = diff[b'sourceControlBaseRevision']
     return meta
 
 
 def _getdrevs(ui, stack, specs):
     """convert user supplied DREVSPECs into "Differential Revision" dicts
 
     See ``hg help phabread`` for how to specify each DREVSPEC.
     """
     if len(specs) > 0:
 
         def _formatspec(s):
             if stack:
                 s = b':(%s)' % s
             return b'(%s)' % s
 
         spec = b'+'.join(pycompat.maplist(_formatspec, specs))
 
         drevs = querydrev(ui, spec)
         if drevs:
             return drevs
 
     raise error.Abort(_(b"empty DREVSPEC set"))
 
 
 def readpatch(ui, drevs, write):
     """generate plain-text patch readable by 'hg import'
 
     write takes a list of (DREV, bytes), where DREV is the differential number
     (as bytes, without the "D" prefix) and the bytes are the text of a patch
     to be imported. drevs is what "querydrev" returns, results of
     "differential.query".
     """
     # Prefetch hg:meta property for all diffs
     diffids = sorted({max(int(v) for v in drev[b'diffs']) for drev in drevs})
     diffs = callconduit(ui, b'differential.querydiffs', {b'ids': diffids})
 
     patches = []
 
     # Generate patch for each drev
     for drev in drevs:
         ui.note(_(b'reading D%s\n') % drev[b'id'])
 
         diffid = max(int(v) for v in drev[b'diffs'])
         body = callconduit(ui, b'differential.getrawdiff', {b'diffID': diffid})
         desc = getdescfromdrev(drev)
         header = b'# HG changeset patch\n'
 
         # Try to preserve metadata from hg:meta property. Write hg patch
         # headers that can be read by the "import" command. See patchheadermap
         # and extract in mercurial/patch.py for supported headers.
         meta = getdiffmeta(diffs[b'%d' % diffid])
         for k in _metanamemap.keys():
             if k in meta:
                 header += b'# %s %s\n' % (_metanamemap[k], meta[k])
 
         content = b'%s%s\n%s' % (header, desc, body)
         patches.append((drev[b'id'], content))
 
     # Write patches to the supplied callback
     write(patches)
 
 
 @vcrcommand(
     b'phabread',
     [(b'', b'stack', False, _(b'read dependencies'))],
     _(b'DREVSPEC... [OPTIONS]'),
     helpcategory=command.CATEGORY_IMPORT_EXPORT,
     optionalrepo=True,
 )
 def phabread(ui, repo, *specs, **opts):
     """print patches from Phabricator suitable for importing
 
     DREVSPEC could be a Differential Revision identity, like ``D123``, or just
     the number ``123``. It could also have common operators like ``+``, ``-``,
     ``&``, ``(``, ``)`` for complex queries. Prefix ``:`` could be used to
     select a stack.  If multiple DREVSPEC values are given, the result is the
     union of each individually evaluated value.  No attempt is currently made
     to reorder the values to run from parent to child.
 
     ``abandoned``, ``accepted``, ``closed``, ``needsreview``, ``needsrevision``
     could be used to filter patches by status. For performance reason, they
     only represent a subset of non-status selections and cannot be used alone.
 
     For example, ``:D6+8-(2+D4)`` selects a stack up to D6, plus D8 and exclude
     D2 and D4. ``:D9 & needsreview`` selects "Needs Review" revisions in a
     stack up to D9.
 
     If --stack is given, follow dependencies information and read all patches.
     It is equivalent to the ``:`` operator.
     """
     opts = pycompat.byteskwargs(opts)
     drevs = _getdrevs(ui, opts.get(b'stack'), specs)
 
     def _write(patches):
         for drev, content in patches:
             ui.write(content)
 
     readpatch(ui, drevs, _write)
 
 
 @vcrcommand(
     b'phabimport',
     [(b'', b'stack', False, _(b'import dependencies as well'))],
     _(b'DREVSPEC... [OPTIONS]'),
     helpcategory=command.CATEGORY_IMPORT_EXPORT,
 )
 def phabimport(ui, repo, *specs, **opts):
     """import patches from Phabricator for the specified Differential Revisions
 
     The patches are read and applied starting at the parent of the working
     directory.
 
     See ``hg help phabread`` for how to specify DREVSPEC.
     """
     opts = pycompat.byteskwargs(opts)
 
     # --bypass avoids losing exec and symlink bits when importing on Windows,
     # and allows importing with a dirty wdir.  It also aborts instead of leaving
     # rejects.
     opts[b'bypass'] = True
 
     # Mandatory default values, synced with commands.import
     opts[b'strip'] = 1
     opts[b'prefix'] = b''
     # Evolve 9.3.0 assumes this key is present in cmdutil.tryimportone()
     opts[b'obsolete'] = False
 
     if ui.configbool(b'phabimport', b'secret'):
         opts[b'secret'] = True
     if ui.configbool(b'phabimport', b'obsolete'):
         opts[b'obsolete'] = True  # Handled by evolve wrapping tryimportone()
 
     def _write(patches):
         parents = repo[None].parents()
 
         with repo.wlock(), repo.lock(), repo.transaction(b'phabimport'):
             for drev, contents in patches:
                 ui.status(_(b'applying patch from D%s\n') % drev)
 
-                with patch.extract(ui, pycompat.bytesio(contents)) as patchdata:
+                with patch.extract(ui, io.BytesIO(contents)) as patchdata:
                     msg, node, rej = cmdutil.tryimportone(
                         ui,
                         repo,
                         patchdata,
                         parents,
                         opts,
                         [],
                         None,  # Never update wdir to another revision
                     )
 
                     if not node:
                         raise error.Abort(_(b'D%s: no diffs found') % drev)
 
                     ui.note(msg + b'\n')
                     parents = [repo[node]]
 
     drevs = _getdrevs(ui, opts.get(b'stack'), specs)
 
     readpatch(repo.ui, drevs, _write)
 
 
 @vcrcommand(
     b'phabupdate',
     [
         (b'', b'accept', False, _(b'accept revisions')),
         (b'', b'reject', False, _(b'reject revisions')),
         (b'', b'request-review', False, _(b'request review on revisions')),
         (b'', b'abandon', False, _(b'abandon revisions')),
         (b'', b'reclaim', False, _(b'reclaim revisions')),
         (b'', b'close', False, _(b'close revisions')),
         (b'', b'reopen', False, _(b'reopen revisions')),
         (b'', b'plan-changes', False, _(b'plan changes for revisions')),
         (b'', b'resign', False, _(b'resign as a reviewer from revisions')),
         (b'', b'commandeer', False, _(b'commandeer revisions')),
         (b'm', b'comment', b'', _(b'comment on the last revision')),
         (b'r', b'rev', b'', _(b'local revision to update'), _(b'REV')),
     ],
     _(b'[DREVSPEC...| -r REV...] [OPTIONS]'),
     helpcategory=command.CATEGORY_IMPORT_EXPORT,
     optionalrepo=True,
 )
 def phabupdate(ui, repo, *specs, **opts):
     """update Differential Revision in batch
 
     DREVSPEC selects revisions. See :hg:`help phabread` for its usage.
     """
     opts = pycompat.byteskwargs(opts)
     transactions = [
         b'abandon',
         b'accept',
         b'close',
         b'commandeer',
         b'plan-changes',
         b'reclaim',
         b'reject',
         b'reopen',
         b'request-review',
         b'resign',
     ]
     flags = [n for n in transactions if opts.get(n.replace(b'-', b'_'))]
     if len(flags) > 1:
         raise error.Abort(_(b'%s cannot be used together') % b', '.join(flags))
 
     actions = []
     for f in flags:
         actions.append({b'type': f, b'value': True})
 
     revs = opts.get(b'rev')
     if revs:
         if not repo:
             raise error.InputError(_(b'--rev requires a repository'))
 
         if specs:
             raise error.InputError(_(b'cannot specify both DREVSPEC and --rev'))
 
         drevmap = getdrevmap(repo, logcmdutil.revrange(repo, [revs]))
         specs = []
         unknown = []
         for r, d in pycompat.iteritems(drevmap):
             if d is None:
                 unknown.append(repo[r])
             else:
                 specs.append(b'D%d' % d)
         if unknown:
             raise error.InputError(
                 _(b'selected revisions without a Differential: %s')
                 % scmutil.nodesummaries(repo, unknown)
             )
 
     drevs = _getdrevs(ui, opts.get(b'stack'), specs)
     for i, drev in enumerate(drevs):
         if i + 1 == len(drevs) and opts.get(b'comment'):
             actions.append({b'type': b'comment', b'value': opts[b'comment']})
         if actions:
             params = {
                 b'objectIdentifier': drev[b'phid'],
                 b'transactions': actions,
             }
             callconduit(ui, b'differential.revision.edit', params)
 
 
 @eh.templatekeyword(b'phabreview', requires={b'ctx'})
 def template_review(context, mapping):
     """:phabreview: Object describing the review for this changeset.
     Has attributes `url` and `id`.
     """
     ctx = context.resource(mapping, b'ctx')
     m = _differentialrevisiondescre.search(ctx.description())
     if m:
         return templateutil.hybriddict(
             {
                 b'url': m.group('url'),
                 b'id': b"D%s" % m.group('id'),
             }
         )
     else:
         tags = ctx.repo().nodetags(ctx.node())
         for t in tags:
             if _differentialrevisiontagre.match(t):
                 url = ctx.repo().ui.config(b'phabricator', b'url')
                 if not url.endswith(b'/'):
                     url += b'/'
                 url += t
 
                 return templateutil.hybriddict(
                     {
                         b'url': url,
                         b'id': t,
                     }
                 )
     return None
 
 
 @eh.templatekeyword(b'phabstatus', requires={b'ctx', b'repo', b'ui'})
 def template_status(context, mapping):
     """:phabstatus: String. Status of Phabricator differential."""
     ctx = context.resource(mapping, b'ctx')
     repo = context.resource(mapping, b'repo')
     ui = context.resource(mapping, b'ui')
 
     rev = ctx.rev()
     try:
         drevid = getdrevmap(repo, [rev])[rev]
     except KeyError:
         return None
     drevs = callconduit(ui, b'differential.query', {b'ids': [drevid]})
     for drev in drevs:
         if int(drev[b'id']) == drevid:
             return templateutil.hybriddict(
                 {
                     b'url': drev[b'uri'],
                     b'status': drev[b'statusName'],
                 }
             )
     return None
 
 
 @show.showview(b'phabstatus', csettopic=b'work')
 def phabstatusshowview(ui, repo, displayer):
     """Phabricator differiential status"""
     revs = repo.revs('sort(_underway(), topo)')
     drevmap = getdrevmap(repo, revs)
     unknownrevs, drevids, revsbydrevid = [], set(), {}
     for rev, drevid in pycompat.iteritems(drevmap):
         if drevid is not None:
             drevids.add(drevid)
             revsbydrevid.setdefault(drevid, set()).add(rev)
         else:
             unknownrevs.append(rev)
 
     drevs = callconduit(ui, b'differential.query', {b'ids': list(drevids)})
     drevsbyrev = {}
     for drev in drevs:
         for rev in revsbydrevid[int(drev[b'id'])]:
             drevsbyrev[rev] = drev
 
     def phabstatus(ctx):
         drev = drevsbyrev[ctx.rev()]
         status = ui.label(
             b'%(statusName)s' % drev,
             b'phabricator.status.%s' % _getstatusname(drev),
         )
         ui.write(b"\n%s %s\n" % (drev[b'uri'], status))
 
     revs -= smartset.baseset(unknownrevs)
     revdag = graphmod.dagwalker(repo, revs)
 
     ui.setconfig(b'experimental', b'graphshorten', True)
     displayer._exthook = phabstatus
     nodelen = show.longestshortest(repo, revs)
     logcmdutil.displaygraph(
         ui,
         repo,
         revdag,
         displayer,
         graphmod.asciiedges,
         props={b'nodelen': nodelen},
     )

mercurial/pure/mpatch.py

@@ -1,135 +1,135 b''
 # mpatch.py - Python implementation of mpatch.c
 #
 # Copyright 2009 Olivia Mackall <[e-mail]> 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
 
+import io
 import struct
 
-from .. import pycompat
 
-stringio = pycompat.bytesio
+stringio = io.BytesIO
 
 
 class mpatchError(Exception):
     """error raised when a delta cannot be decoded"""
 
 
 # This attempts to apply a series of patches in time proportional to
 # the total size of the patches, rather than patches * len(text). This
 # means rather than shuffling strings around, we shuffle around
 # pointers to fragments with fragment lists.
 #
 # When the fragment lists get too long, we collapse them. To do this
 # efficiently, we do all our operations inside a buffer created by
 # mmap and simply use memmove. This avoids creating a bunch of large
 # temporary string buffers.
 
 
 def _pull(dst, src, l):  # pull l bytes from src
     while l:
         f = src.pop()
         if f[0] > l:  # do we need to split?
             src.append((f[0] - l, f[1] + l))
             dst.append((l, f[1]))
             return
         dst.append(f)
         l -= f[0]
 
 
 def _move(m, dest, src, count):
     """move count bytes from src to dest
 
     The file pointer is left at the end of dest.
     """
     m.seek(src)
     buf = m.read(count)
     m.seek(dest)
     m.write(buf)
 
 
 def _collect(m, buf, list):
     start = buf
     for l, p in reversed(list):
         _move(m, buf, p, l)
         buf += l
     return (buf - start, start)
 
 
 def patches(a, bins):
     if not bins:
         return a
 
     plens = [len(x) for x in bins]
     pl = sum(plens)
     bl = len(a) + pl
     tl = bl + bl + pl  # enough for the patches and two working texts
     b1, b2 = 0, bl
 
     if not tl:
         return a
 
     m = stringio()
 
     # load our original text
     m.write(a)
     frags = [(len(a), b1)]
 
     # copy all the patches into our segment so we can memmove from them
     pos = b2 + bl
     m.seek(pos)
     for p in bins:
         m.write(p)
 
     for plen in plens:
         # if our list gets too long, execute it
         if len(frags) > 128:
             b2, b1 = b1, b2
             frags = [_collect(m, b1, frags)]
 
         new = []
         end = pos + plen
         last = 0
         while pos < end:
             m.seek(pos)
             try:
                 p1, p2, l = struct.unpack(b">lll", m.read(12))
             except struct.error:
                 raise mpatchError(b"patch cannot be decoded")
             _pull(new, frags, p1 - last)  # what didn't change
             _pull([], frags, p2 - p1)  # what got deleted
             new.append((l, pos + 12))  # what got added
             pos += l + 12
             last = p2
         frags.extend(reversed(new))  # what was left at the end
 
     t = _collect(m, b2, frags)
 
     m.seek(t[1])
     return m.read(t[0])
 
 
 def patchedsize(orig, delta):
     outlen, last, bin = 0, 0, 0
     binend = len(delta)
     data = 12
 
     while data <= binend:
         decode = delta[bin : bin + 12]
         start, end, length = struct.unpack(b">lll", decode)
         if start > end:
             break
         bin = data + length
         data = bin + 12
         outlen += start - last
         last = end
         outlen += length
 
     if bin != binend:
         raise mpatchError(b"patch cannot be decoded")
 
     outlen += orig - last
     return outlen

mercurial/pure/parsers.py

@@ -1,975 +1,976 b''
 # parsers.py - Python implementation of parsers.c
 #
 # Copyright 2009 Olivia Mackall <[e-mail]> 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
 
+import io
 import stat
 import struct
 import zlib
 
 from ..node import (
     nullrev,
     sha1nodeconstants,
 )
 from ..thirdparty import attr
 from .. import (
     error,
     pycompat,
     revlogutils,
     util,
 )
 
 from ..revlogutils import nodemap as nodemaputil
 from ..revlogutils import constants as revlog_constants
 
-stringio = pycompat.bytesio
+stringio = io.BytesIO
 
 
 _pack = struct.pack
 _unpack = struct.unpack
 _compress = zlib.compress
 _decompress = zlib.decompress
 
 
 # a special value used internally for `size` if the file come from the other parent
 FROM_P2 = -2
 
 # a special value used internally for `size` if the file is modified/merged/added
 NONNORMAL = -1
 
 # a special value used internally for `time` if the time is ambigeous
 AMBIGUOUS_TIME = -1
 
 # Bits of the `flags` byte inside a node in the file format
 DIRSTATE_V2_WDIR_TRACKED = 1 << 0
 DIRSTATE_V2_P1_TRACKED = 1 << 1
 DIRSTATE_V2_P2_INFO = 1 << 2
 DIRSTATE_V2_MODE_EXEC_PERM = 1 << 3
 DIRSTATE_V2_MODE_IS_SYMLINK = 1 << 4
 DIRSTATE_V2_HAS_FALLBACK_EXEC = 1 << 5
 DIRSTATE_V2_FALLBACK_EXEC = 1 << 6
 DIRSTATE_V2_HAS_FALLBACK_SYMLINK = 1 << 7
 DIRSTATE_V2_FALLBACK_SYMLINK = 1 << 8
 DIRSTATE_V2_EXPECTED_STATE_IS_MODIFIED = 1 << 9
 DIRSTATE_V2_HAS_MODE_AND_SIZE = 1 << 10
 DIRSTATE_V2_HAS_MTIME = 1 << 11
 DIRSTATE_V2_MTIME_SECOND_AMBIGUOUS = 1 << 12
 DIRSTATE_V2_DIRECTORY = 1 << 13
 DIRSTATE_V2_ALL_UNKNOWN_RECORDED = 1 << 14
 DIRSTATE_V2_ALL_IGNORED_RECORDED = 1 << 15
 
 
 @attr.s(slots=True, init=False)
 class DirstateItem(object):
     """represent a dirstate entry
 
     It hold multiple attributes
 
     # about file tracking
     - wc_tracked: is the file tracked by the working copy
     - p1_tracked: is the file tracked in working copy first parent
     - p2_info: the file has been involved in some merge operation. Either
                because it was actually merged, or because the p2 version was
                ahead, or because some rename moved it there. In either case
                `hg status` will want it displayed as modified.
 
     # about the file state expected from p1 manifest:
     - mode: the file mode in p1
     - size: the file size in p1
 
     These value can be set to None, which mean we don't have a meaningful value
     to compare with. Either because we don't really care about them as there
     `status` is known without having to look at the disk or because we don't
     know these right now and a full comparison will be needed to find out if
     the file is clean.
 
     # about the file state on disk last time we saw it:
     - mtime: the last known clean mtime for the file.
 
     This value can be set to None if no cachable state exist. Either because we
     do not care (see previous section) or because we could not cache something
     yet.
     """
 
     _wc_tracked = attr.ib()
     _p1_tracked = attr.ib()
     _p2_info = attr.ib()
     _mode = attr.ib()
     _size = attr.ib()
     _mtime_s = attr.ib()
     _mtime_ns = attr.ib()
     _fallback_exec = attr.ib()
     _fallback_symlink = attr.ib()
     _mtime_second_ambiguous = attr.ib()
 
     def __init__(
         self,
         wc_tracked=False,
         p1_tracked=False,
         p2_info=False,
         has_meaningful_data=True,
         has_meaningful_mtime=True,
         parentfiledata=None,
         fallback_exec=None,
         fallback_symlink=None,
     ):
         self._wc_tracked = wc_tracked
         self._p1_tracked = p1_tracked
         self._p2_info = p2_info
 
         self._fallback_exec = fallback_exec
         self._fallback_symlink = fallback_symlink
 
         self._mode = None
         self._size = None
         self._mtime_s = None
         self._mtime_ns = None
         self._mtime_second_ambiguous = False
         if parentfiledata is None:
             has_meaningful_mtime = False
             has_meaningful_data = False
         elif parentfiledata[2] is None:
             has_meaningful_mtime = False
         if has_meaningful_data:
             self._mode = parentfiledata[0]
             self._size = parentfiledata[1]
         if has_meaningful_mtime:
             (
                 self._mtime_s,
                 self._mtime_ns,
                 self._mtime_second_ambiguous,
             ) = parentfiledata[2]
 
     @classmethod
     def from_v2_data(cls, flags, size, mtime_s, mtime_ns):
         """Build a new DirstateItem object from V2 data"""
         has_mode_size = bool(flags & DIRSTATE_V2_HAS_MODE_AND_SIZE)
         has_meaningful_mtime = bool(flags & DIRSTATE_V2_HAS_MTIME)
         mode = None
 
         if flags & +DIRSTATE_V2_EXPECTED_STATE_IS_MODIFIED:
             # we do not have support for this flag in the code yet,
             # force a lookup for this file.
             has_mode_size = False
             has_meaningful_mtime = False
 
         fallback_exec = None
         if flags & DIRSTATE_V2_HAS_FALLBACK_EXEC:
             fallback_exec = flags & DIRSTATE_V2_FALLBACK_EXEC
 
         fallback_symlink = None
         if flags & DIRSTATE_V2_HAS_FALLBACK_SYMLINK:
             fallback_symlink = flags & DIRSTATE_V2_FALLBACK_SYMLINK
 
         if has_mode_size:
             assert stat.S_IXUSR == 0o100
             if flags & DIRSTATE_V2_MODE_EXEC_PERM:
                 mode = 0o755
             else:
                 mode = 0o644
             if flags & DIRSTATE_V2_MODE_IS_SYMLINK:
                 mode |= stat.S_IFLNK
             else:
                 mode |= stat.S_IFREG
 
         second_ambiguous = flags & DIRSTATE_V2_MTIME_SECOND_AMBIGUOUS
         return cls(
             wc_tracked=bool(flags & DIRSTATE_V2_WDIR_TRACKED),
             p1_tracked=bool(flags & DIRSTATE_V2_P1_TRACKED),
             p2_info=bool(flags & DIRSTATE_V2_P2_INFO),
             has_meaningful_data=has_mode_size,
             has_meaningful_mtime=has_meaningful_mtime,
             parentfiledata=(mode, size, (mtime_s, mtime_ns, second_ambiguous)),
             fallback_exec=fallback_exec,
             fallback_symlink=fallback_symlink,
         )
 
     @classmethod
     def from_v1_data(cls, state, mode, size, mtime):
         """Build a new DirstateItem object from V1 data
 
         Since the dirstate-v1 format is frozen, the signature of this function
         is not expected to change, unlike the __init__ one.
         """
         if state == b'm':
             return cls(wc_tracked=True, p1_tracked=True, p2_info=True)
         elif state == b'a':
             return cls(wc_tracked=True)
         elif state == b'r':
             if size == NONNORMAL:
                 p1_tracked = True
                 p2_info = True
             elif size == FROM_P2:
                 p1_tracked = False
                 p2_info = True
             else:
                 p1_tracked = True
                 p2_info = False
             return cls(p1_tracked=p1_tracked, p2_info=p2_info)
         elif state == b'n':
             if size == FROM_P2:
                 return cls(wc_tracked=True, p2_info=True)
             elif size == NONNORMAL:
                 return cls(wc_tracked=True, p1_tracked=True)
             elif mtime == AMBIGUOUS_TIME:
                 return cls(
                     wc_tracked=True,
                     p1_tracked=True,
                     has_meaningful_mtime=False,
                     parentfiledata=(mode, size, (42, 0, False)),
                 )
             else:
                 return cls(
                     wc_tracked=True,
                     p1_tracked=True,
                     parentfiledata=(mode, size, (mtime, 0, False)),
                 )
         else:
             raise RuntimeError(b'unknown state: %s' % state)
 
     def set_possibly_dirty(self):
         """Mark a file as "possibly dirty"
 
         This means the next status call will have to actually check its content
         to make sure it is correct.
         """
         self._mtime_s = None
         self._mtime_ns = None
 
     def set_clean(self, mode, size, mtime):
         """mark a file as "clean" cancelling potential "possibly dirty call"
 
         Note: this function is a descendant of `dirstate.normal` and is
         currently expected to be call on "normal" entry only. There are not
         reason for this to not change in the future as long as the ccode is
         updated to preserve the proper state of the non-normal files.
         """
         self._wc_tracked = True
         self._p1_tracked = True
         self._mode = mode
         self._size = size
         self._mtime_s, self._mtime_ns, self._mtime_second_ambiguous = mtime
 
     def set_tracked(self):
         """mark a file as tracked in the working copy
 
         This will ultimately be called by command like `hg add`.
         """
         self._wc_tracked = True
         # `set_tracked` is replacing various `normallookup` call. So we mark
         # the files as needing lookup
         #
         # Consider dropping this in the future in favor of something less broad.
         self._mtime_s = None
         self._mtime_ns = None
 
     def set_untracked(self):
         """mark a file as untracked in the working copy
 
         This will ultimately be called by command like `hg remove`.
         """
         self._wc_tracked = False
         self._mode = None
         self._size = None
         self._mtime_s = None
         self._mtime_ns = None
 
     def drop_merge_data(self):
         """remove all "merge-only" from a DirstateItem
 
         This is to be call by the dirstatemap code when the second parent is dropped
         """
         if self._p2_info:
             self._p2_info = False
             self._mode = None
             self._size = None
             self._mtime_s = None
             self._mtime_ns = None
 
     @property
     def mode(self):
         return self.v1_mode()
 
     @property
     def size(self):
         return self.v1_size()
 
     @property
     def mtime(self):
         return self.v1_mtime()
 
     def mtime_likely_equal_to(self, other_mtime):
         self_sec = self._mtime_s
         if self_sec is None:
             return False
         self_ns = self._mtime_ns
         other_sec, other_ns, second_ambiguous = other_mtime
         if self_sec != other_sec:
             # seconds are different theses mtime are definitly not equal
             return False
         elif other_ns == 0 or self_ns == 0:
             # at least one side as no nano-seconds information
 
             if self._mtime_second_ambiguous:
                 # We cannot trust the mtime in this case
                 return False
             else:
                 # the "seconds" value was reliable on its own. We are good to go.
                 return True
         else:
             # We have nano second information, let us use them !
             return self_ns == other_ns
 
     @property
     def state(self):
         """
         States are:
           n  normal
           m  needs merging
           r  marked for removal
           a  marked for addition
 
         XXX This "state" is a bit obscure and mostly a direct expression of the
         dirstatev1 format. It would make sense to ultimately deprecate it in
         favor of the more "semantic" attributes.
         """
         if not self.any_tracked:
             return b'?'
         return self.v1_state()
 
     @property
     def has_fallback_exec(self):
         """True if "fallback" information are available for the "exec" bit
 
         Fallback information can be stored in the dirstate to keep track of
         filesystem attribute tracked by Mercurial when the underlying file
         system or operating system does not support that property, (e.g.
         Windows).
 
         Not all version of the dirstate on-disk storage support preserving this
         information.
         """
         return self._fallback_exec is not None
 
     @property
     def fallback_exec(self):
         """ "fallback" information for the executable bit
 
         True if the file should be considered executable when we cannot get
         this information from the files system. False if it should be
         considered non-executable.
 
         See has_fallback_exec for details."""
         return self._fallback_exec
 
     @fallback_exec.setter
     def set_fallback_exec(self, value):
         """control "fallback" executable bit
 
         Set to:
         - True if the file should be considered executable,
         - False if the file should be considered non-executable,
         - None if we do not have valid fallback data.
 
         See has_fallback_exec for details."""
         if value is None:
             self._fallback_exec = None
         else:
             self._fallback_exec = bool(value)
 
     @property
     def has_fallback_symlink(self):
         """True if "fallback" information are available for symlink status
 
         Fallback information can be stored in the dirstate to keep track of
         filesystem attribute tracked by Mercurial when the underlying file
         system or operating system does not support that property, (e.g.
         Windows).
 
         Not all version of the dirstate on-disk storage support preserving this
         information."""
         return self._fallback_symlink is not None
 
     @property
     def fallback_symlink(self):
         """ "fallback" information for symlink status
 
         True if the file should be considered executable when we cannot get
         this information from the files system. False if it should be
         considered non-executable.
 
         See has_fallback_exec for details."""
         return self._fallback_symlink
 
     @fallback_symlink.setter
     def set_fallback_symlink(self, value):
         """control "fallback" symlink status
 
         Set to:
         - True if the file should be considered a symlink,
         - False if the file should be considered not a symlink,
         - None if we do not have valid fallback data.
 
         See has_fallback_symlink for details."""
         if value is None:
             self._fallback_symlink = None
         else:
             self._fallback_symlink = bool(value)
 
     @property
     def tracked(self):
         """True is the file is tracked in the working copy"""
         return self._wc_tracked
 
     @property
     def any_tracked(self):
         """True is the file is tracked anywhere (wc or parents)"""
         return self._wc_tracked or self._p1_tracked or self._p2_info
 
     @property
     def added(self):
         """True if the file has been added"""
         return self._wc_tracked and not (self._p1_tracked or self._p2_info)
 
     @property
     def maybe_clean(self):
         """True if the file has a chance to be in the "clean" state"""
         if not self._wc_tracked:
             return False
         elif not self._p1_tracked:
             return False
         elif self._p2_info:
             return False
         return True
 
     @property
     def p1_tracked(self):
         """True if the file is tracked in the first parent manifest"""
         return self._p1_tracked
 
     @property
     def p2_info(self):
         """True if the file needed to merge or apply any input from p2
 
         See the class documentation for details.
         """
         return self._wc_tracked and self._p2_info
 
     @property
     def removed(self):
         """True if the file has been removed"""
         return not self._wc_tracked and (self._p1_tracked or self._p2_info)
 
     def v2_data(self):
         """Returns (flags, mode, size, mtime) for v2 serialization"""
         flags = 0
         if self._wc_tracked:
             flags |= DIRSTATE_V2_WDIR_TRACKED
         if self._p1_tracked:
             flags |= DIRSTATE_V2_P1_TRACKED
         if self._p2_info:
             flags |= DIRSTATE_V2_P2_INFO
         if self._mode is not None and self._size is not None:
             flags |= DIRSTATE_V2_HAS_MODE_AND_SIZE
             if self.mode & stat.S_IXUSR:
                 flags |= DIRSTATE_V2_MODE_EXEC_PERM
             if stat.S_ISLNK(self.mode):
                 flags |= DIRSTATE_V2_MODE_IS_SYMLINK
         if self._mtime_s is not None:
             flags |= DIRSTATE_V2_HAS_MTIME
         if self._mtime_second_ambiguous:
             flags |= DIRSTATE_V2_MTIME_SECOND_AMBIGUOUS
 
         if self._fallback_exec is not None:
             flags |= DIRSTATE_V2_HAS_FALLBACK_EXEC
             if self._fallback_exec:
                 flags |= DIRSTATE_V2_FALLBACK_EXEC
 
         if self._fallback_symlink is not None:
             flags |= DIRSTATE_V2_HAS_FALLBACK_SYMLINK
             if self._fallback_symlink:
                 flags |= DIRSTATE_V2_FALLBACK_SYMLINK
 
         # Note: we do not need to do anything regarding
         # DIRSTATE_V2_ALL_UNKNOWN_RECORDED and DIRSTATE_V2_ALL_IGNORED_RECORDED
         # since we never set _DIRSTATE_V2_HAS_DIRCTORY_MTIME
         return (flags, self._size or 0, self._mtime_s or 0, self._mtime_ns or 0)
 
     def v1_state(self):
         """return a "state" suitable for v1 serialization"""
         if not self.any_tracked:
             # the object has no state to record, this is -currently-
             # unsupported
             raise RuntimeError('untracked item')
         elif self.removed:
             return b'r'
         elif self._p1_tracked and self._p2_info:
             return b'm'
         elif self.added:
             return b'a'
         else:
             return b'n'
 
     def v1_mode(self):
         """return a "mode" suitable for v1 serialization"""
         return self._mode if self._mode is not None else 0
 
     def v1_size(self):
         """return a "size" suitable for v1 serialization"""
         if not self.any_tracked:
             # the object has no state to record, this is -currently-
             # unsupported
             raise RuntimeError('untracked item')
         elif self.removed and self._p1_tracked and self._p2_info:
             return NONNORMAL
         elif self._p2_info:
             return FROM_P2
         elif self.removed:
             return 0
         elif self.added:
             return NONNORMAL
         elif self._size is None:
             return NONNORMAL
         else:
             return self._size
 
     def v1_mtime(self):
         """return a "mtime" suitable for v1 serialization"""
         if not self.any_tracked:
             # the object has no state to record, this is -currently-
             # unsupported
             raise RuntimeError('untracked item')
         elif self.removed:
             return 0
         elif self._mtime_s is None:
             return AMBIGUOUS_TIME
         elif self._p2_info:
             return AMBIGUOUS_TIME
         elif not self._p1_tracked:
             return AMBIGUOUS_TIME
         elif self._mtime_second_ambiguous:
             return AMBIGUOUS_TIME
         else:
             return self._mtime_s
 
 
 def gettype(q):
     return int(q & 0xFFFF)
 
 
 class BaseIndexObject(object):
     # Can I be passed to an algorithme implemented in Rust ?
     rust_ext_compat = 0
     # Format of an index entry according to Python's `struct` language
     index_format = revlog_constants.INDEX_ENTRY_V1
     # Size of a C unsigned long long int, platform independent
     big_int_size = struct.calcsize(b'>Q')
     # Size of a C long int, platform independent
     int_size = struct.calcsize(b'>i')
     # An empty index entry, used as a default value to be overridden, or nullrev
     null_item = (
         0,
         0,
         0,
         -1,
         -1,
         -1,
         -1,
         sha1nodeconstants.nullid,
         0,
         0,
         revlog_constants.COMP_MODE_INLINE,
         revlog_constants.COMP_MODE_INLINE,
         revlog_constants.RANK_UNKNOWN,
     )
 
     @util.propertycache
     def entry_size(self):
         return self.index_format.size
 
     @util.propertycache
     def _nodemap(self):
         nodemap = nodemaputil.NodeMap({sha1nodeconstants.nullid: nullrev})
         for r in range(0, len(self)):
             n = self[r][7]
             nodemap[n] = r
         return nodemap
 
     def has_node(self, node):
         """return True if the node exist in the index"""
         return node in self._nodemap
 
     def rev(self, node):
         """return a revision for a node
 
         If the node is unknown, raise a RevlogError"""
         return self._nodemap[node]
 
     def get_rev(self, node):
         """return a revision for a node
 
         If the node is unknown, return None"""
         return self._nodemap.get(node)
 
     def _stripnodes(self, start):
         if '_nodemap' in vars(self):
             for r in range(start, len(self)):
                 n = self[r][7]
                 del self._nodemap[n]
 
     def clearcaches(self):
         self.__dict__.pop('_nodemap', None)
 
     def __len__(self):
         return self._lgt + len(self._extra)
 
     def append(self, tup):
         if '_nodemap' in vars(self):
             self._nodemap[tup[7]] = len(self)
         data = self._pack_entry(len(self), tup)
         self._extra.append(data)
 
     def _pack_entry(self, rev, entry):
         assert entry[8] == 0
         assert entry[9] == 0
         return self.index_format.pack(*entry[:8])
 
     def _check_index(self, i):
         if not isinstance(i, int):
             raise TypeError(b"expecting int indexes")
         if i < 0 or i >= len(self):
             raise IndexError(i)
 
     def __getitem__(self, i):
         if i == -1:
             return self.null_item
         self._check_index(i)
         if i >= self._lgt:
             data = self._extra[i - self._lgt]
         else:
             index = self._calculate_index(i)
             data = self._data[index : index + self.entry_size]
         r = self._unpack_entry(i, data)
         if self._lgt and i == 0:
             offset = revlogutils.offset_type(0, gettype(r[0]))
             r = (offset,) + r[1:]
         return r
 
     def _unpack_entry(self, rev, data):
         r = self.index_format.unpack(data)
         r = r + (
             0,
             0,
             revlog_constants.COMP_MODE_INLINE,
             revlog_constants.COMP_MODE_INLINE,
             revlog_constants.RANK_UNKNOWN,
         )
         return r
 
     def pack_header(self, header):
         """pack header information as binary"""
         v_fmt = revlog_constants.INDEX_HEADER
         return v_fmt.pack(header)
 
     def entry_binary(self, rev):
         """return the raw binary string representing a revision"""
         entry = self[rev]
         p = revlog_constants.INDEX_ENTRY_V1.pack(*entry[:8])
         if rev == 0:
             p = p[revlog_constants.INDEX_HEADER.size :]
         return p
 
 
 class IndexObject(BaseIndexObject):
     def __init__(self, data):
         assert len(data) % self.entry_size == 0, (
             len(data),
             self.entry_size,
             len(data) % self.entry_size,
         )
         self._data = data
         self._lgt = len(data) // self.entry_size
         self._extra = []
 
     def _calculate_index(self, i):
         return i * self.entry_size
 
     def __delitem__(self, i):
         if not isinstance(i, slice) or not i.stop == -1 or i.step is not None:
             raise ValueError(b"deleting slices only supports a:-1 with step 1")
         i = i.start
         self._check_index(i)
         self._stripnodes(i)
         if i < self._lgt:
             self._data = self._data[: i * self.entry_size]
             self._lgt = i
             self._extra = []
         else:
             self._extra = self._extra[: i - self._lgt]
 
 
 class PersistentNodeMapIndexObject(IndexObject):
     """a Debug oriented class to test persistent nodemap
 
     We need a simple python object to test API and higher level behavior. See
     the Rust implementation for  more serious usage. This should be used only
     through the dedicated `devel.persistent-nodemap` config.
     """
 
     def nodemap_data_all(self):
         """Return bytes containing a full serialization of a nodemap
 
         The nodemap should be valid for the full set of revisions in the
         index."""
         return nodemaputil.persistent_data(self)
 
     def nodemap_data_incremental(self):
         """Return bytes containing a incremental update to persistent nodemap
 
         This containst the data for an append-only update of the data provided
         in the last call to `update_nodemap_data`.
         """
         if self._nm_root is None:
             return None
         docket = self._nm_docket
         changed, data = nodemaputil.update_persistent_data(
             self, self._nm_root, self._nm_max_idx, self._nm_docket.tip_rev
         )
 
         self._nm_root = self._nm_max_idx = self._nm_docket = None
         return docket, changed, data
 
     def update_nodemap_data(self, docket, nm_data):
         """provide full block of persisted binary data for a nodemap
 
         The data are expected to come from disk. See `nodemap_data_all` for a
         produceur of such data."""
         if nm_data is not None:
             self._nm_root, self._nm_max_idx = nodemaputil.parse_data(nm_data)
             if self._nm_root:
                 self._nm_docket = docket
             else:
                 self._nm_root = self._nm_max_idx = self._nm_docket = None
 
 
 class InlinedIndexObject(BaseIndexObject):
     def __init__(self, data, inline=0):
         self._data = data
         self._lgt = self._inline_scan(None)
         self._inline_scan(self._lgt)
         self._extra = []
 
     def _inline_scan(self, lgt):
         off = 0
         if lgt is not None:
             self._offsets = [0] * lgt
         count = 0
         while off <= len(self._data) - self.entry_size:
             start = off + self.big_int_size
             (s,) = struct.unpack(
                 b'>i',
                 self._data[start : start + self.int_size],
             )
             if lgt is not None:
                 self._offsets[count] = off
             count += 1
             off += self.entry_size + s
         if off != len(self._data):
             raise ValueError(b"corrupted data")
         return count
 
     def __delitem__(self, i):
         if not isinstance(i, slice) or not i.stop == -1 or i.step is not None:
             raise ValueError(b"deleting slices only supports a:-1 with step 1")
         i = i.start
         self._check_index(i)
         self._stripnodes(i)
         if i < self._lgt:
             self._offsets = self._offsets[:i]
             self._lgt = i
             self._extra = []
         else:
             self._extra = self._extra[: i - self._lgt]
 
     def _calculate_index(self, i):
         return self._offsets[i]
 
 
 def parse_index2(data, inline, format=revlog_constants.REVLOGV1):
     if format == revlog_constants.CHANGELOGV2:
         return parse_index_cl_v2(data)
     if not inline:
         if format == revlog_constants.REVLOGV2:
             cls = IndexObject2
         else:
             cls = IndexObject
         return cls(data), None
     cls = InlinedIndexObject
     return cls(data, inline), (0, data)
 
 
 def parse_index_cl_v2(data):
     return IndexChangelogV2(data), None
 
 
 class IndexObject2(IndexObject):
     index_format = revlog_constants.INDEX_ENTRY_V2
 
     def replace_sidedata_info(
         self,
         rev,
         sidedata_offset,
         sidedata_length,
         offset_flags,
         compression_mode,
     ):
         """
         Replace an existing index entry's sidedata offset and length with new
         ones.
         This cannot be used outside of the context of sidedata rewriting,
         inside the transaction that creates the revision `rev`.
         """
         if rev < 0:
             raise KeyError
         self._check_index(rev)
         if rev < self._lgt:
             msg = b"cannot rewrite entries outside of this transaction"
             raise KeyError(msg)
         else:
             entry = list(self[rev])
             entry[0] = offset_flags
             entry[8] = sidedata_offset
             entry[9] = sidedata_length
             entry[11] = compression_mode
             entry = tuple(entry)
             new = self._pack_entry(rev, entry)
             self._extra[rev - self._lgt] = new
 
     def _unpack_entry(self, rev, data):
         data = self.index_format.unpack(data)
         entry = data[:10]
         data_comp = data[10] & 3
         sidedata_comp = (data[10] & (3 << 2)) >> 2
         return entry + (data_comp, sidedata_comp, revlog_constants.RANK_UNKNOWN)
 
     def _pack_entry(self, rev, entry):
         data = entry[:10]
         data_comp = entry[10] & 3
         sidedata_comp = (entry[11] & 3) << 2
         data += (data_comp | sidedata_comp,)
 
         return self.index_format.pack(*data)
 
     def entry_binary(self, rev):
         """return the raw binary string representing a revision"""
         entry = self[rev]
         return self._pack_entry(rev, entry)
 
     def pack_header(self, header):
         """pack header information as binary"""
         msg = 'version header should go in the docket, not the index: %d'
         msg %= header
         raise error.ProgrammingError(msg)
 
 
 class IndexChangelogV2(IndexObject2):
     index_format = revlog_constants.INDEX_ENTRY_CL_V2
 
     null_item = (
         IndexObject2.null_item[: revlog_constants.ENTRY_RANK]
         + (0,)  # rank of null is 0
         + IndexObject2.null_item[revlog_constants.ENTRY_RANK :]
     )
 
     def _unpack_entry(self, rev, data, r=True):
         items = self.index_format.unpack(data)
         return (
             items[revlog_constants.INDEX_ENTRY_V2_IDX_OFFSET],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_COMPRESSED_LENGTH],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_UNCOMPRESSED_LENGTH],
             rev,
             rev,
             items[revlog_constants.INDEX_ENTRY_V2_IDX_PARENT_1],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_PARENT_2],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_NODEID],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_SIDEDATA_OFFSET],
             items[
                 revlog_constants.INDEX_ENTRY_V2_IDX_SIDEDATA_COMPRESSED_LENGTH
             ],
             items[revlog_constants.INDEX_ENTRY_V2_IDX_COMPRESSION_MODE] & 3,
             (items[revlog_constants.INDEX_ENTRY_V2_IDX_COMPRESSION_MODE] >> 2)
             & 3,
             items[revlog_constants.INDEX_ENTRY_V2_IDX_RANK],
         )
 
     def _pack_entry(self, rev, entry):
 
         base = entry[revlog_constants.ENTRY_DELTA_BASE]
         link_rev = entry[revlog_constants.ENTRY_LINK_REV]
         assert base == rev, (base, rev)
         assert link_rev == rev, (link_rev, rev)
         data = (
             entry[revlog_constants.ENTRY_DATA_OFFSET],
             entry[revlog_constants.ENTRY_DATA_COMPRESSED_LENGTH],
             entry[revlog_constants.ENTRY_DATA_UNCOMPRESSED_LENGTH],
             entry[revlog_constants.ENTRY_PARENT_1],
             entry[revlog_constants.ENTRY_PARENT_2],
             entry[revlog_constants.ENTRY_NODE_ID],
             entry[revlog_constants.ENTRY_SIDEDATA_OFFSET],
             entry[revlog_constants.ENTRY_SIDEDATA_COMPRESSED_LENGTH],
             entry[revlog_constants.ENTRY_DATA_COMPRESSION_MODE] & 3
             | (entry[revlog_constants.ENTRY_SIDEDATA_COMPRESSION_MODE] & 3)
             << 2,
             entry[revlog_constants.ENTRY_RANK],
         )
         return self.index_format.pack(*data)
 
 
 def parse_index_devel_nodemap(data, inline):
     """like parse_index2, but alway return a PersistentNodeMapIndexObject"""
     return PersistentNodeMapIndexObject(data), None
 
 
 def parse_dirstate(dmap, copymap, st):
     parents = [st[:20], st[20:40]]
     # dereference fields so they will be local in loop
     format = b">cllll"
     e_size = struct.calcsize(format)
     pos1 = 40
     l = len(st)
 
     # the inner loop
     while pos1 < l:
         pos2 = pos1 + e_size
         e = _unpack(b">cllll", st[pos1:pos2])  # a literal here is faster
         pos1 = pos2 + e[4]
         f = st[pos2:pos1]
         if b'\0' in f:
             f, c = f.split(b'\0')
             copymap[f] = c
         dmap[f] = DirstateItem.from_v1_data(*e[:4])
     return parents
 
 
 def pack_dirstate(dmap, copymap, pl):
     cs = stringio()
     write = cs.write
     write(b"".join(pl))
     for f, e in pycompat.iteritems(dmap):
         if f in copymap:
             f = b"%s\0%s" % (f, copymap[f])
         e = _pack(
             b">cllll",
             e.v1_state(),
             e.v1_mode(),
             e.v1_size(),
             e.v1_mtime(),
             len(f),
         )
         write(e)
         write(f)
     return cs.getvalue()

mercurial/pycompat.py

@@ -1,471 +1,468 b''
 # pycompat.py - portability shim for python 3
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Mercurial portability shim for python 3.
 
 This contains aliases to hide python version-specific details from the core.
 """
 
 from __future__ import absolute_import
 
 import builtins
 import codecs
 import concurrent.futures as futures
 import functools
 import getopt
 import http.client as httplib
 import http.cookiejar as cookielib
 import inspect
 import io
 import json
 import os
 import pickle
 import queue
 import shlex
 import socketserver
 import struct
 import sys
 import tempfile
 import xmlrpc.client as xmlrpclib
 
 
 ispy3 = sys.version_info[0] >= 3
 ispypy = '__pypy__' in sys.builtin_module_names
 TYPE_CHECKING = False
 
 if not globals():  # hide this from non-pytype users
     import typing
 
     TYPE_CHECKING = typing.TYPE_CHECKING
 
 
 def future_set_exception_info(f, exc_info):
     f.set_exception(exc_info[0])
 
 
 FileNotFoundError = builtins.FileNotFoundError
 
 
 def identity(a):
     return a
 
 
 def _rapply(f, xs):
     if xs is None:
         # assume None means non-value of optional data
         return xs
     if isinstance(xs, (list, set, tuple)):
         return type(xs)(_rapply(f, x) for x in xs)
     if isinstance(xs, dict):
         return type(xs)((_rapply(f, k), _rapply(f, v)) for k, v in xs.items())
     return f(xs)
 
 
 def rapply(f, xs):
     """Apply function recursively to every item preserving the data structure
 
     >>> def f(x):
     ...     return 'f(%s)' % x
     >>> rapply(f, None) is None
     True
     >>> rapply(f, 'a')
     'f(a)'
     >>> rapply(f, {'a'}) == {'f(a)'}
     True
     >>> rapply(f, ['a', 'b', None, {'c': 'd'}, []])
     ['f(a)', 'f(b)', None, {'f(c)': 'f(d)'}, []]
 
     >>> xs = [object()]
     >>> rapply(identity, xs) is xs
     True
     """
     if f is identity:
         # fast path mainly for py2
         return xs
     return _rapply(f, xs)
 
 
 if os.name == r'nt' and sys.version_info >= (3, 6):
     # MBCS (or ANSI) filesystem encoding must be used as before.
     # Otherwise non-ASCII filenames in existing repositories would be
     # corrupted.
     # This must be set once prior to any fsencode/fsdecode calls.
     sys._enablelegacywindowsfsencoding()  # pytype: disable=module-attr
 
 fsencode = os.fsencode
 fsdecode = os.fsdecode
 oscurdir = os.curdir.encode('ascii')
 oslinesep = os.linesep.encode('ascii')
 osname = os.name.encode('ascii')
 ospathsep = os.pathsep.encode('ascii')
 ospardir = os.pardir.encode('ascii')
 ossep = os.sep.encode('ascii')
 osaltsep = os.altsep
 if osaltsep:
     osaltsep = osaltsep.encode('ascii')
 osdevnull = os.devnull.encode('ascii')
 
 sysplatform = sys.platform.encode('ascii')
 sysexecutable = sys.executable
 if sysexecutable:
     sysexecutable = os.fsencode(sysexecutable)
-bytesio = io.BytesIO
-# TODO deprecate stringio name, as it is a lie on Python 3.
-stringio = bytesio
 
 
 def maplist(*args):
     return list(map(*args))
 
 
 def rangelist(*args):
     return list(range(*args))
 
 
 def ziplist(*args):
     return list(zip(*args))
 
 
 rawinput = input
 getargspec = inspect.getfullargspec
 
 long = int
 
 if getattr(sys, 'argv', None) is not None:
     # On POSIX, the char** argv array is converted to Python str using
     # Py_DecodeLocale(). The inverse of this is Py_EncodeLocale(), which
     # isn't directly callable from Python code. In practice, os.fsencode()
     # can be used instead (this is recommended by Python's documentation
     # for sys.argv).
     #
     # On Windows, the wchar_t **argv is passed into the interpreter as-is.
     # Like POSIX, we need to emulate what Py_EncodeLocale() would do. But
     # there's an additional wrinkle. What we really want to access is the
     # ANSI codepage representation of the arguments, as this is what
     # `int main()` would receive if Python 3 didn't define `int wmain()`
     # (this is how Python 2 worked). To get that, we encode with the mbcs
     # encoding, which will pass CP_ACP to the underlying Windows API to
     # produce bytes.
     if os.name == r'nt':
         sysargv = [a.encode("mbcs", "ignore") for a in sys.argv]
     else:
         sysargv = [fsencode(a) for a in sys.argv]
 
 bytechr = struct.Struct('>B').pack
 byterepr = b'%r'.__mod__
 
 
 class bytestr(bytes):
     """A bytes which mostly acts as a Python 2 str
 
     >>> bytestr(), bytestr(bytearray(b'foo')), bytestr(u'ascii'), bytestr(1)
     ('', 'foo', 'ascii', '1')
     >>> s = bytestr(b'foo')
     >>> assert s is bytestr(s)
 
     __bytes__() should be called if provided:
 
     >>> class bytesable(object):
     ...     def __bytes__(self):
     ...         return b'bytes'
     >>> bytestr(bytesable())
     'bytes'
 
     There's no implicit conversion from non-ascii str as its encoding is
     unknown:
 
     >>> bytestr(chr(0x80)) # doctest: +ELLIPSIS
     Traceback (most recent call last):
       ...
     UnicodeEncodeError: ...
 
     Comparison between bytestr and bytes should work:
 
     >>> assert bytestr(b'foo') == b'foo'
     >>> assert b'foo' == bytestr(b'foo')
     >>> assert b'f' in bytestr(b'foo')
     >>> assert bytestr(b'f') in b'foo'
 
     Sliced elements should be bytes, not integer:
 
     >>> s[1], s[:2]
     (b'o', b'fo')
     >>> list(s), list(reversed(s))
     ([b'f', b'o', b'o'], [b'o', b'o', b'f'])
 
     As bytestr type isn't propagated across operations, you need to cast
     bytes to bytestr explicitly:
 
     >>> s = bytestr(b'foo').upper()
     >>> t = bytestr(s)
     >>> s[0], t[0]
     (70, b'F')
 
     Be careful to not pass a bytestr object to a function which expects
     bytearray-like behavior.
 
     >>> t = bytes(t)  # cast to bytes
     >>> assert type(t) is bytes
     """
 
     # Trick pytype into not demanding Iterable[int] be passed to __new__(),
     # since the appropriate bytes format is done internally.
     #
     # https://github.com/google/pytype/issues/500
     if TYPE_CHECKING:
 
         def __init__(self, s=b''):
             pass
 
     def __new__(cls, s=b''):
         if isinstance(s, bytestr):
             return s
         if not isinstance(
             s, (bytes, bytearray)
         ) and not hasattr(  # hasattr-py3-only
             s, u'__bytes__'
         ):
             s = str(s).encode('ascii')
         return bytes.__new__(cls, s)
 
     def __getitem__(self, key):
         s = bytes.__getitem__(self, key)
         if not isinstance(s, bytes):
             s = bytechr(s)
         return s
 
     def __iter__(self):
         return iterbytestr(bytes.__iter__(self))
 
     def __repr__(self):
         return bytes.__repr__(self)[1:]  # drop b''
 
 
 def iterbytestr(s):
     """Iterate bytes as if it were a str object of Python 2"""
     return map(bytechr, s)
 
 
 def maybebytestr(s):
     """Promote bytes to bytestr"""
     if isinstance(s, bytes):
         return bytestr(s)
     return s
 
 
 def sysbytes(s):
     """Convert an internal str (e.g. keyword, __doc__) back to bytes
 
     This never raises UnicodeEncodeError, but only ASCII characters
     can be round-trip by sysstr(sysbytes(s)).
     """
     if isinstance(s, bytes):
         return s
     return s.encode('utf-8')
 
 
 def sysstr(s):
     """Return a keyword str to be passed to Python functions such as
     getattr() and str.encode()
 
     This never raises UnicodeDecodeError. Non-ascii characters are
     considered invalid and mapped to arbitrary but unique code points
     such that 'sysstr(a) != sysstr(b)' for all 'a != b'.
     """
     if isinstance(s, builtins.str):
         return s
     return s.decode('latin-1')
 
 
 def strurl(url):
     """Converts a bytes url back to str"""
     if isinstance(url, bytes):
         return url.decode('ascii')
     return url
 
 
 def bytesurl(url):
     """Converts a str url to bytes by encoding in ascii"""
     if isinstance(url, str):
         return url.encode('ascii')
     return url
 
 
 def raisewithtb(exc, tb):
     """Raise exception with the given traceback"""
     raise exc.with_traceback(tb)
 
 
 def getdoc(obj):
     """Get docstring as bytes; may be None so gettext() won't confuse it
     with _('')"""
     doc = getattr(obj, '__doc__', None)
     if doc is None:
         return doc
     return sysbytes(doc)
 
 
 def _wrapattrfunc(f):
     @functools.wraps(f)
     def w(object, name, *args):
         return f(object, sysstr(name), *args)
 
     return w
 
 
 # these wrappers are automagically imported by hgloader
 delattr = _wrapattrfunc(builtins.delattr)
 getattr = _wrapattrfunc(builtins.getattr)
 hasattr = _wrapattrfunc(builtins.hasattr)
 setattr = _wrapattrfunc(builtins.setattr)
 xrange = builtins.range
 unicode = str
 
 
 def open(name, mode=b'r', buffering=-1, encoding=None):
     return builtins.open(name, sysstr(mode), buffering, encoding)
 
 
 safehasattr = _wrapattrfunc(builtins.hasattr)
 
 
 def _getoptbwrapper(orig, args, shortlist, namelist):
     """
     Takes bytes arguments, converts them to unicode, pass them to
     getopt.getopt(), convert the returned values back to bytes and then
     return them for Python 3 compatibility as getopt.getopt() don't accepts
     bytes on Python 3.
     """
     args = [a.decode('latin-1') for a in args]
     shortlist = shortlist.decode('latin-1')
     namelist = [a.decode('latin-1') for a in namelist]
     opts, args = orig(args, shortlist, namelist)
     opts = [(a[0].encode('latin-1'), a[1].encode('latin-1')) for a in opts]
     args = [a.encode('latin-1') for a in args]
     return opts, args
 
 
 def strkwargs(dic):
     """
     Converts the keys of a python dictonary to str i.e. unicodes so that
     they can be passed as keyword arguments as dictionaries with bytes keys
     can't be passed as keyword arguments to functions on Python 3.
     """
     dic = {k.decode('latin-1'): v for k, v in dic.items()}
     return dic
 
 
 def byteskwargs(dic):
     """
     Converts keys of python dictionaries to bytes as they were converted to
     str to pass that dictonary as a keyword argument on Python 3.
     """
     dic = {k.encode('latin-1'): v for k, v in dic.items()}
     return dic
 
 
 # TODO: handle shlex.shlex().
 def shlexsplit(s, comments=False, posix=True):
     """
     Takes bytes argument, convert it to str i.e. unicodes, pass that into
     shlex.split(), convert the returned value to bytes and return that for
     Python 3 compatibility as shelx.split() don't accept bytes on Python 3.
     """
     ret = shlex.split(s.decode('latin-1'), comments, posix)
     return [a.encode('latin-1') for a in ret]
 
 
 iteritems = lambda x: x.items()
 itervalues = lambda x: x.values()
 
 # Python 3.5's json.load and json.loads require str. We polyfill its
 # code for detecting encoding from bytes.
 if sys.version_info[0:2] < (3, 6):
 
     def _detect_encoding(b):
         bstartswith = b.startswith
         if bstartswith((codecs.BOM_UTF32_BE, codecs.BOM_UTF32_LE)):
             return 'utf-32'
         if bstartswith((codecs.BOM_UTF16_BE, codecs.BOM_UTF16_LE)):
             return 'utf-16'
         if bstartswith(codecs.BOM_UTF8):
             return 'utf-8-sig'
 
         if len(b) >= 4:
             if not b[0]:
                 # 00 00 -- -- - utf-32-be
                 # 00 XX -- -- - utf-16-be
                 return 'utf-16-be' if b[1] else 'utf-32-be'
             if not b[1]:
                 # XX 00 00 00 - utf-32-le
                 # XX 00 00 XX - utf-16-le
                 # XX 00 XX -- - utf-16-le
                 return 'utf-16-le' if b[2] or b[3] else 'utf-32-le'
         elif len(b) == 2:
             if not b[0]:
                 # 00 XX - utf-16-be
                 return 'utf-16-be'
             if not b[1]:
                 # XX 00 - utf-16-le
                 return 'utf-16-le'
         # default
         return 'utf-8'
 
     def json_loads(s, *args, **kwargs):
         if isinstance(s, (bytes, bytearray)):
             s = s.decode(_detect_encoding(s), 'surrogatepass')
 
         return json.loads(s, *args, **kwargs)
 
 
 else:
     json_loads = json.loads
 
 isjython = sysplatform.startswith(b'java')
 
 isdarwin = sysplatform.startswith(b'darwin')
 islinux = sysplatform.startswith(b'linux')
 isposix = osname == b'posix'
 iswindows = osname == b'nt'
 
 
 def getoptb(args, shortlist, namelist):
     return _getoptbwrapper(getopt.getopt, args, shortlist, namelist)
 
 
 def gnugetoptb(args, shortlist, namelist):
     return _getoptbwrapper(getopt.gnu_getopt, args, shortlist, namelist)
 
 
 def mkdtemp(suffix=b'', prefix=b'tmp', dir=None):
     return tempfile.mkdtemp(suffix, prefix, dir)
 
 
 # text=True is not supported; use util.from/tonativeeol() instead
 def mkstemp(suffix=b'', prefix=b'tmp', dir=None):
     return tempfile.mkstemp(suffix, prefix, dir)
 
 
 # TemporaryFile does not support an "encoding=" argument on python2.
 # This wrapper file are always open in byte mode.
 def unnamedtempfile(mode=None, *args, **kwargs):
     if mode is None:
         mode = 'w+b'
     else:
         mode = sysstr(mode)
     assert 'b' in mode
     return tempfile.TemporaryFile(mode, *args, **kwargs)
 
 
 # NamedTemporaryFile does not support an "encoding=" argument on python2.
 # This wrapper file are always open in byte mode.
 def namedtempfile(
     mode=b'w+b', bufsize=-1, suffix=b'', prefix=b'tmp', dir=None, delete=True
 ):
     mode = sysstr(mode)
     assert 'b' in mode
     return tempfile.NamedTemporaryFile(
         mode, bufsize, suffix=suffix, prefix=prefix, dir=dir, delete=delete
     )

mercurial/util.py

@@ -1,3360 +1,3361 b''
 # util.py - Mercurial utility functions and platform specific implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Olivia Mackall <[e-mail]>
 #  Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Mercurial utility functions and platform specific implementations.
 
 This contains helper routines that are independent of the SCM core and
 hide platform-specific details from the core.
 """
 
 from __future__ import absolute_import, print_function
 
 import abc
 import collections
 import contextlib
 import errno
 import gc
 import hashlib
+import io
 import itertools
 import locale
 import mmap
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import stat
 import sys
 import time
 import traceback
 import warnings
 
 from .node import hex
 from .thirdparty import attr
 from .pycompat import (
     delattr,
     getattr,
     open,
     setattr,
 )
 from .node import hex
 from hgdemandimport import tracing
 from . import (
     encoding,
     error,
     i18n,
     policy,
     pycompat,
     urllibcompat,
 )
 from .utils import (
     compression,
     hashutil,
     procutil,
     stringutil,
 )
 
 if pycompat.TYPE_CHECKING:
     from typing import (
         Iterator,
         List,
         Optional,
         Tuple,
     )
 
 
 base85 = policy.importmod('base85')
 osutil = policy.importmod('osutil')
 
 b85decode = base85.b85decode
 b85encode = base85.b85encode
 
 cookielib = pycompat.cookielib
 httplib = pycompat.httplib
 safehasattr = pycompat.safehasattr
 socketserver = pycompat.socketserver
-bytesio = pycompat.bytesio
+bytesio = io.BytesIO
 # TODO deprecate stringio name, as it is a lie on Python 3.
 stringio = bytesio
 xmlrpclib = pycompat.xmlrpclib
 
 httpserver = urllibcompat.httpserver
 urlerr = urllibcompat.urlerr
 urlreq = urllibcompat.urlreq
 
 # workaround for win32mbcs
 _filenamebytestr = pycompat.bytestr
 
 if pycompat.iswindows:
     from . import windows as platform
 else:
     from . import posix as platform
 
 _ = i18n._
 
 abspath = platform.abspath
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 expandglobs = platform.expandglobs
 getfsmountpoint = platform.getfsmountpoint
 getfstype = platform.getfstype
 get_password = platform.get_password
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 isexec = platform.isexec
 isowner = platform.isowner
 listdir = osutil.listdir
 localpath = platform.localpath
 lookupreg = platform.lookupreg
 makedir = platform.makedir
 nlinks = platform.nlinks
 normpath = platform.normpath
 normcase = platform.normcase
 normcasespec = platform.normcasespec
 normcasefallback = platform.normcasefallback
 openhardlinks = platform.openhardlinks
 oslink = platform.oslink
 parsepatchoutput = platform.parsepatchoutput
 pconvert = platform.pconvert
 poll = platform.poll
 posixfile = platform.posixfile
 readlink = platform.readlink
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setflags = platform.setflags
 split = platform.split
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 umask = platform.umask
 unlink = platform.unlink
 username = platform.username
 
 
 def setumask(val):
     # type: (int) -> None
     '''updates the umask. used by chg server'''
     if pycompat.iswindows:
         return
     os.umask(val)
     global umask
     platform.umask = umask = val & 0o777
 
 
 # small compat layer
 compengines = compression.compengines
 SERVERROLE = compression.SERVERROLE
 CLIENTROLE = compression.CLIENTROLE
 
 try:
     recvfds = osutil.recvfds
 except AttributeError:
     pass
 
 # Python compatibility
 
 _notset = object()
 
 
 def bitsfrom(container):
     bits = 0
     for bit in container:
         bits |= bit
     return bits
 
 
 # python 2.6 still have deprecation warning enabled by default. We do not want
 # to display anything to standard user so detect if we are running test and
 # only use python deprecation warning in this case.
 _dowarn = bool(encoding.environ.get(b'HGEMITWARNINGS'))
 if _dowarn:
     # explicitly unfilter our warning for python 2.7
     #
     # The option of setting PYTHONWARNINGS in the test runner was investigated.
     # However, module name set through PYTHONWARNINGS was exactly matched, so
     # we cannot set 'mercurial' and have it match eg: 'mercurial.scmutil'. This
     # makes the whole PYTHONWARNINGS thing useless for our usecase.
     warnings.filterwarnings('default', '', DeprecationWarning, 'mercurial')
     warnings.filterwarnings('default', '', DeprecationWarning, 'hgext')
     warnings.filterwarnings('default', '', DeprecationWarning, 'hgext3rd')
 if _dowarn and pycompat.ispy3:
     # silence warning emitted by passing user string to re.sub()
     warnings.filterwarnings(
         'ignore', 'bad escape', DeprecationWarning, 'mercurial'
     )
     warnings.filterwarnings(
         'ignore', 'invalid escape sequence', DeprecationWarning, 'mercurial'
     )
     # TODO: reinvent imp.is_frozen()
     warnings.filterwarnings(
         'ignore',
         'the imp module is deprecated',
         DeprecationWarning,
         'mercurial',
     )
 
 
 def nouideprecwarn(msg, version, stacklevel=1):
     """Issue an python native deprecation warning
 
     This is a noop outside of tests, use 'ui.deprecwarn' when possible.
     """
     if _dowarn:
         msg += (
             b"\n(compatibility will be dropped after Mercurial-%s,"
             b" update your code.)"
         ) % version
         warnings.warn(pycompat.sysstr(msg), DeprecationWarning, stacklevel + 1)
         # on python 3 with chg, we will need to explicitly flush the output
         sys.stderr.flush()
 
 
 DIGESTS = {
     b'md5': hashlib.md5,
     b'sha1': hashutil.sha1,
     b'sha512': hashlib.sha512,
 }
 # List of digest types from strongest to weakest
 DIGESTS_BY_STRENGTH = [b'sha512', b'sha1', b'md5']
 
 for k in DIGESTS_BY_STRENGTH:
     assert k in DIGESTS
 
 
 class digester(object):
     """helper to compute digests.
 
     This helper can be used to compute one or more digests given their name.
 
     >>> d = digester([b'md5', b'sha1'])
     >>> d.update(b'foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d[b'md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d[b'sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred([b'md5', b'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=b''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise error.Abort(_(b'unknown digest type: %s') % k)
             self._hashes[k] = DIGESTS[k]()
         if s:
             self.update(s)
 
     def update(self, data):
         for h in self._hashes.values():
             h.update(data)
 
     def __getitem__(self, key):
         if key not in DIGESTS:
             raise error.Abort(_(b'unknown digest type: %s') % k)
         return hex(self._hashes[key].digest())
 
     def __iter__(self):
         return iter(self._hashes)
 
     @staticmethod
     def preferred(supported):
         """returns the strongest digest type in both supported and DIGESTS."""
 
         for k in DIGESTS_BY_STRENGTH:
             if k in supported:
                 return k
         return None
 
 
 class digestchecker(object):
     """file handle wrapper that additionally checks content against a given
     size and digests.
 
         d = digestchecker(fh, size, {'md5': '...'})
 
     When multiple digests are given, all of them are validated.
     """
 
     def __init__(self, fh, size, digests):
         self._fh = fh
         self._size = size
         self._got = 0
         self._digests = dict(digests)
         self._digester = digester(self._digests.keys())
 
     def read(self, length=-1):
         content = self._fh.read(length)
         self._digester.update(content)
         self._got += len(content)
         return content
 
     def validate(self):
         if self._size != self._got:
             raise error.Abort(
                 _(b'size mismatch: expected %d, got %d')
                 % (self._size, self._got)
             )
         for k, v in self._digests.items():
             if v != self._digester[k]:
                 # i18n: first parameter is a digest name
                 raise error.Abort(
                     _(b'%s mismatch: expected %s, got %s')
                     % (k, v, self._digester[k])
                 )
 
 
 try:
     buffer = buffer  # pytype: disable=name-error
 except NameError:
 
     def buffer(sliceable, offset=0, length=None):
         if length is not None:
             return memoryview(sliceable)[offset : offset + length]
         return memoryview(sliceable)[offset:]
 
 
 _chunksize = 4096
 
 
 class bufferedinputpipe(object):
     """a manually buffered input pipe
 
     Python will not let us use buffered IO and lazy reading with 'polling' at
     the same time. We cannot probe the buffer state and select will not detect
     that data are ready to read if they are already buffered.
 
     This class let us work around that by implementing its own buffering
     (allowing efficient readline) while offering a way to know if the buffer is
     empty from the output (allowing collaboration of the buffer with polling).
 
     This class lives in the 'util' module because it makes use of the 'os'
     module from the python stdlib.
     """
 
     def __new__(cls, fh):
         # If we receive a fileobjectproxy, we need to use a variation of this
         # class that notifies observers about activity.
         if isinstance(fh, fileobjectproxy):
             cls = observedbufferedinputpipe
 
         return super(bufferedinputpipe, cls).__new__(cls)
 
     def __init__(self, input):
         self._input = input
         self._buffer = []
         self._eof = False
         self._lenbuf = 0
 
     @property
     def hasbuffer(self):
         """True is any data is currently buffered
 
         This will be used externally a pre-step for polling IO. If there is
         already data then no polling should be set in place."""
         return bool(self._buffer)
 
     @property
     def closed(self):
         return self._input.closed
 
     def fileno(self):
         return self._input.fileno()
 
     def close(self):
         return self._input.close()
 
     def read(self, size):
         while (not self._eof) and (self._lenbuf < size):
             self._fillbuffer()
         return self._frombuffer(size)
 
     def unbufferedread(self, size):
         if not self._eof and self._lenbuf == 0:
             self._fillbuffer(max(size, _chunksize))
         return self._frombuffer(min(self._lenbuf, size))
 
     def readline(self, *args, **kwargs):
         if len(self._buffer) > 1:
             # this should not happen because both read and readline end with a
             # _frombuffer call that collapse it.
             self._buffer = [b''.join(self._buffer)]
             self._lenbuf = len(self._buffer[0])
         lfi = -1
         if self._buffer:
             lfi = self._buffer[-1].find(b'\n')
         while (not self._eof) and lfi < 0:
             self._fillbuffer()
             if self._buffer:
                 lfi = self._buffer[-1].find(b'\n')
         size = lfi + 1
         if lfi < 0:  # end of file
             size = self._lenbuf
         elif len(self._buffer) > 1:
             # we need to take previous chunks into account
             size += self._lenbuf - len(self._buffer[-1])
         return self._frombuffer(size)
 
     def _frombuffer(self, size):
         """return at most 'size' data from the buffer
 
         The data are removed from the buffer."""
         if size == 0 or not self._buffer:
             return b''
         buf = self._buffer[0]
         if len(self._buffer) > 1:
             buf = b''.join(self._buffer)
 
         data = buf[:size]
         buf = buf[len(data) :]
         if buf:
             self._buffer = [buf]
             self._lenbuf = len(buf)
         else:
             self._buffer = []
             self._lenbuf = 0
         return data
 
     def _fillbuffer(self, size=_chunksize):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), size)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
         return data
 
 
 def mmapread(fp, size=None):
     if size == 0:
         # size of 0 to mmap.mmap() means "all data"
         # rather than "zero bytes", so special case that.
         return b''
     elif size is None:
         size = 0
     fd = getattr(fp, 'fileno', lambda: fp)()
     try:
         return mmap.mmap(fd, size, access=mmap.ACCESS_READ)
     except ValueError:
         # Empty files cannot be mmapped, but mmapread should still work.  Check
         # if the file is empty, and if so, return an empty buffer.
         if os.fstat(fd).st_size == 0:
             return b''
         raise
 
 
 class fileobjectproxy(object):
     """A proxy around file objects that tells a watcher when events occur.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
 
     __slots__ = (
         '_orig',
         '_observer',
     )
 
     def __init__(self, fh, observer):
         object.__setattr__(self, '_orig', fh)
         object.__setattr__(self, '_observer', observer)
 
     def __getattribute__(self, name):
         ours = {
             '_observer',
             # IOBase
             'close',
             # closed if a property
             'fileno',
             'flush',
             'isatty',
             'readable',
             'readline',
             'readlines',
             'seek',
             'seekable',
             'tell',
             'truncate',
             'writable',
             'writelines',
             # RawIOBase
             'read',
             'readall',
             'readinto',
             'write',
             # BufferedIOBase
             # raw is a property
             'detach',
             # read defined above
             'read1',
             # readinto defined above
             # write defined above
         }
 
         # We only observe some methods.
         if name in ours:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, '_orig'), name)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, '_orig'))
 
     __bool__ = __nonzero__
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, '_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, '_orig'), name, value)
 
     def __iter__(self):
         return object.__getattribute__(self, '_orig').__iter__()
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, '_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, '_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def close(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'close', *args, **kwargs
         )
 
     def fileno(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'fileno', *args, **kwargs
         )
 
     def flush(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'flush', *args, **kwargs
         )
 
     def isatty(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'isatty', *args, **kwargs
         )
 
     def readable(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'readable', *args, **kwargs
         )
 
     def readline(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'readline', *args, **kwargs
         )
 
     def readlines(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'readlines', *args, **kwargs
         )
 
     def seek(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'seek', *args, **kwargs
         )
 
     def seekable(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'seekable', *args, **kwargs
         )
 
     def tell(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'tell', *args, **kwargs
         )
 
     def truncate(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'truncate', *args, **kwargs
         )
 
     def writable(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'writable', *args, **kwargs
         )
 
     def writelines(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'writelines', *args, **kwargs
         )
 
     def read(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'read', *args, **kwargs
         )
 
     def readall(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'readall', *args, **kwargs
         )
 
     def readinto(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'readinto', *args, **kwargs
         )
 
     def write(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'write', *args, **kwargs
         )
 
     def detach(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'detach', *args, **kwargs
         )
 
     def read1(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'read1', *args, **kwargs
         )
 
 
 class observedbufferedinputpipe(bufferedinputpipe):
     """A variation of bufferedinputpipe that is aware of fileobjectproxy.
 
     ``bufferedinputpipe`` makes low-level calls to ``os.read()`` that
     bypass ``fileobjectproxy``. Because of this, we need to make
     ``bufferedinputpipe`` aware of these operations.
 
     This variation of ``bufferedinputpipe`` can notify observers about
     ``os.read()`` events. It also re-publishes other events, such as
     ``read()`` and ``readline()``.
     """
 
     def _fillbuffer(self):
         res = super(observedbufferedinputpipe, self)._fillbuffer()
 
         fn = getattr(self._input._observer, 'osread', None)
         if fn:
             fn(res, _chunksize)
 
         return res
 
     # We use different observer methods because the operation isn't
     # performed on the actual file object but on us.
     def read(self, size):
         res = super(observedbufferedinputpipe, self).read(size)
 
         fn = getattr(self._input._observer, 'bufferedread', None)
         if fn:
             fn(res, size)
 
         return res
 
     def readline(self, *args, **kwargs):
         res = super(observedbufferedinputpipe, self).readline(*args, **kwargs)
 
         fn = getattr(self._input._observer, 'bufferedreadline', None)
         if fn:
             fn(res)
 
         return res
 
 
 PROXIED_SOCKET_METHODS = {
     'makefile',
     'recv',
     'recvfrom',
     'recvfrom_into',
     'recv_into',
     'send',
     'sendall',
     'sendto',
     'setblocking',
     'settimeout',
     'gettimeout',
     'setsockopt',
 }
 
 
 class socketproxy(object):
     """A proxy around a socket that tells a watcher when events occur.
 
     This is like ``fileobjectproxy`` except for sockets.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
 
     __slots__ = (
         '_orig',
         '_observer',
     )
 
     def __init__(self, sock, observer):
         object.__setattr__(self, '_orig', sock)
         object.__setattr__(self, '_observer', observer)
 
     def __getattribute__(self, name):
         if name in PROXIED_SOCKET_METHODS:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, '_orig'), name)
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, '_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, '_orig'), name, value)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, '_orig'))
 
     __bool__ = __nonzero__
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, '_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, '_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def makefile(self, *args, **kwargs):
         res = object.__getattribute__(self, '_observedcall')(
             'makefile', *args, **kwargs
         )
 
         # The file object may be used for I/O. So we turn it into a
         # proxy using our observer.
         observer = object.__getattribute__(self, '_observer')
         return makeloggingfileobject(
             observer.fh,
             res,
             observer.name,
             reads=observer.reads,
             writes=observer.writes,
             logdata=observer.logdata,
             logdataapis=observer.logdataapis,
         )
 
     def recv(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'recv', *args, **kwargs
         )
 
     def recvfrom(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'recvfrom', *args, **kwargs
         )
 
     def recvfrom_into(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'recvfrom_into', *args, **kwargs
         )
 
     def recv_into(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'recv_info', *args, **kwargs
         )
 
     def send(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'send', *args, **kwargs
         )
 
     def sendall(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'sendall', *args, **kwargs
         )
 
     def sendto(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'sendto', *args, **kwargs
         )
 
     def setblocking(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'setblocking', *args, **kwargs
         )
 
     def settimeout(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'settimeout', *args, **kwargs
         )
 
     def gettimeout(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'gettimeout', *args, **kwargs
         )
 
     def setsockopt(self, *args, **kwargs):
         return object.__getattribute__(self, '_observedcall')(
             'setsockopt', *args, **kwargs
         )
 
 
 class baseproxyobserver(object):
     def __init__(self, fh, name, logdata, logdataapis):
         self.fh = fh
         self.name = name
         self.logdata = logdata
         self.logdataapis = logdataapis
 
     def _writedata(self, data):
         if not self.logdata:
             if self.logdataapis:
                 self.fh.write(b'\n')
                 self.fh.flush()
             return
 
         # Simple case writes all data on a single line.
         if b'\n' not in data:
             if self.logdataapis:
                 self.fh.write(b': %s\n' % stringutil.escapestr(data))
             else:
                 self.fh.write(
                     b'%s>     %s\n' % (self.name, stringutil.escapestr(data))
                 )
             self.fh.flush()
             return
 
         # Data with newlines is written to multiple lines.
         if self.logdataapis:
             self.fh.write(b':\n')
 
         lines = data.splitlines(True)
         for line in lines:
             self.fh.write(
                 b'%s>     %s\n' % (self.name, stringutil.escapestr(line))
             )
         self.fh.flush()
 
 
 class fileobjectobserver(baseproxyobserver):
     """Logs file object activity."""
 
     def __init__(
         self, fh, name, reads=True, writes=True, logdata=False, logdataapis=True
     ):
         super(fileobjectobserver, self).__init__(fh, name, logdata, logdataapis)
         self.reads = reads
         self.writes = writes
 
     def read(self, res, size=-1):
         if not self.reads:
             return
         # Python 3 can return None from reads at EOF instead of empty strings.
         if res is None:
             res = b''
 
         if size == -1 and res == b'':
             # Suppress pointless read(-1) calls that return
             # nothing. These happen _a lot_ on Python 3, and there
             # doesn't seem to be a better workaround to have matching
             # Python 2 and 3 behavior. :(
             return
 
         if self.logdataapis:
             self.fh.write(b'%s> read(%d) -> %d' % (self.name, size, len(res)))
 
         self._writedata(res)
 
     def readline(self, res, limit=-1):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(b'%s> readline() -> %d' % (self.name, len(res)))
 
         self._writedata(res)
 
     def readinto(self, res, dest):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> readinto(%d) -> %r' % (self.name, len(dest), res)
             )
 
         data = dest[0:res] if res is not None else b''
 
         # _writedata() uses "in" operator and is confused by memoryview because
         # characters are ints on Python 3.
         if isinstance(data, memoryview):
             data = data.tobytes()
 
         self._writedata(data)
 
     def write(self, res, data):
         if not self.writes:
             return
 
         # Python 2 returns None from some write() calls. Python 3 (reasonably)
         # returns the integer bytes written.
         if res is None and data:
             res = len(data)
 
         if self.logdataapis:
             self.fh.write(b'%s> write(%d) -> %r' % (self.name, len(data), res))
 
         self._writedata(data)
 
     def flush(self, res):
         if not self.writes:
             return
 
         self.fh.write(b'%s> flush() -> %r\n' % (self.name, res))
 
     # For observedbufferedinputpipe.
     def bufferedread(self, res, size):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> bufferedread(%d) -> %d' % (self.name, size, len(res))
             )
 
         self._writedata(res)
 
     def bufferedreadline(self, res):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> bufferedreadline() -> %d' % (self.name, len(res))
             )
 
         self._writedata(res)
 
 
 def makeloggingfileobject(
     logh, fh, name, reads=True, writes=True, logdata=False, logdataapis=True
 ):
     """Turn a file object into a logging file object."""
 
     observer = fileobjectobserver(
         logh,
         name,
         reads=reads,
         writes=writes,
         logdata=logdata,
         logdataapis=logdataapis,
     )
     return fileobjectproxy(fh, observer)
 
 
 class socketobserver(baseproxyobserver):
     """Logs socket activity."""
 
     def __init__(
         self,
         fh,
         name,
         reads=True,
         writes=True,
         states=True,
         logdata=False,
         logdataapis=True,
     ):
         super(socketobserver, self).__init__(fh, name, logdata, logdataapis)
         self.reads = reads
         self.writes = writes
         self.states = states
 
     def makefile(self, res, mode=None, bufsize=None):
         if not self.states:
             return
 
         self.fh.write(b'%s> makefile(%r, %r)\n' % (self.name, mode, bufsize))
 
     def recv(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recv(%d, %d) -> %d' % (self.name, size, flags, len(res))
             )
         self._writedata(res)
 
     def recvfrom(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recvfrom(%d, %d) -> %d'
                 % (self.name, size, flags, len(res[0]))
             )
 
         self._writedata(res[0])
 
     def recvfrom_into(self, res, buf, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recvfrom_into(%d, %d) -> %d'
                 % (self.name, size, flags, res[0])
             )
 
         self._writedata(buf[0 : res[0]])
 
     def recv_into(self, res, buf, size=0, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recv_into(%d, %d) -> %d' % (self.name, size, flags, res)
             )
 
         self._writedata(buf[0:res])
 
     def send(self, res, data, flags=0):
         if not self.writes:
             return
 
         self.fh.write(
             b'%s> send(%d, %d) -> %d' % (self.name, len(data), flags, len(res))
         )
         self._writedata(data)
 
     def sendall(self, res, data, flags=0):
         if not self.writes:
             return
 
         if self.logdataapis:
             # Returns None on success. So don't bother reporting return value.
             self.fh.write(
                 b'%s> sendall(%d, %d)' % (self.name, len(data), flags)
             )
 
         self._writedata(data)
 
     def sendto(self, res, data, flagsoraddress, address=None):
         if not self.writes:
             return
 
         if address:
             flags = flagsoraddress
         else:
             flags = 0
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> sendto(%d, %d, %r) -> %d'
                 % (self.name, len(data), flags, address, res)
             )
 
         self._writedata(data)
 
     def setblocking(self, res, flag):
         if not self.states:
             return
 
         self.fh.write(b'%s> setblocking(%r)\n' % (self.name, flag))
 
     def settimeout(self, res, value):
         if not self.states:
             return
 
         self.fh.write(b'%s> settimeout(%r)\n' % (self.name, value))
 
     def gettimeout(self, res):
         if not self.states:
             return
 
         self.fh.write(b'%s> gettimeout() -> %f\n' % (self.name, res))
 
     def setsockopt(self, res, level, optname, value):
         if not self.states:
             return
 
         self.fh.write(
             b'%s> setsockopt(%r, %r, %r) -> %r\n'
             % (self.name, level, optname, value, res)
         )
 
 
 def makeloggingsocket(
     logh,
     fh,
     name,
     reads=True,
     writes=True,
     states=True,
     logdata=False,
     logdataapis=True,
 ):
     """Turn a socket into a logging socket."""
 
     observer = socketobserver(
         logh,
         name,
         reads=reads,
         writes=writes,
         states=states,
         logdata=logdata,
         logdataapis=logdataapis,
     )
     return socketproxy(fh, observer)
 
 
 def version():
     """Return version information if available."""
     try:
         from . import __version__
 
         return __version__.version
     except ImportError:
         return b'unknown'
 
 
 def versiontuple(v=None, n=4):
     """Parses a Mercurial version string into an N-tuple.
 
     The version string to be parsed is specified with the ``v`` argument.
     If it isn't defined, the current Mercurial version string will be parsed.
 
     ``n`` can be 2, 3, or 4. Here is how some version strings map to
     returned values:
 
     >>> v = b'3.6.1+190-df9b73d2d444'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, 1)
     >>> versiontuple(v, 4)
     (3, 6, 1, '190-df9b73d2d444')
 
     >>> versiontuple(b'3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = b'3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = b'3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = b'3.9-rc+2-02a8fea4289b'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc+2-02a8fea4289b')
 
     >>> versiontuple(b'4.6rc0')
     (4, 6, None, 'rc0')
     >>> versiontuple(b'4.6rc0+12-425d55e54f98')
     (4, 6, None, 'rc0+12-425d55e54f98')
     >>> versiontuple(b'.1.2.3')
     (None, None, None, '.1.2.3')
     >>> versiontuple(b'12.34..5')
     (12, 34, None, '..5')
     >>> versiontuple(b'1.2.3.4.5.6')
     (1, 2, 3, '.4.5.6')
     """
     if not v:
         v = version()
     m = remod.match(br'(\d+(?:\.\d+){,2})[+-]?(.*)', v)
     if not m:
         vparts, extra = b'', v
     elif m.group(2):
         vparts, extra = m.groups()
     else:
         vparts, extra = m.group(1), None
 
     assert vparts is not None  # help pytype
 
     vints = []
     for i in vparts.split(b'.'):
         try:
             vints.append(int(i))
         except ValueError:
             break
     # (3, 6) -> (3, 6, None)
     while len(vints) < 3:
         vints.append(None)
 
     if n == 2:
         return (vints[0], vints[1])
     if n == 3:
         return (vints[0], vints[1], vints[2])
     if n == 4:
         return (vints[0], vints[1], vints[2], extra)
 
     raise error.ProgrammingError(b"invalid version part request: %d" % n)
 
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     if func.__code__.co_argcount == 0:
         listcache = []
 
         def f():
             if len(listcache) == 0:
                 listcache.append(func())
             return listcache[0]
 
         return f
     cache = {}
     if func.__code__.co_argcount == 1:
         # we gain a small amount of time because
         # we don't need to pack/unpack the list
         def f(arg):
             if arg not in cache:
                 cache[arg] = func(arg)
             return cache[arg]
 
     else:
 
         def f(*args):
             if args not in cache:
                 cache[args] = func(*args)
             return cache[args]
 
     return f
 
 
 class cow(object):
     """helper class to make copy-on-write easier
 
     Call preparewrite before doing any writes.
     """
 
     def preparewrite(self):
         """call this before writes, return self or a copied new object"""
         if getattr(self, '_copied', 0):
             self._copied -= 1
             # Function cow.__init__ expects 1 arg(s), got 2 [wrong-arg-count]
             return self.__class__(self)  # pytype: disable=wrong-arg-count
         return self
 
     def copy(self):
         """always do a cheap copy"""
         self._copied = getattr(self, '_copied', 0) + 1
         return self
 
 
 class sortdict(collections.OrderedDict):
     """a simple sorted dictionary
 
     >>> d1 = sortdict([(b'a', 0), (b'b', 1)])
     >>> d2 = d1.copy()
     >>> d2
     sortdict([('a', 0), ('b', 1)])
     >>> d2.update([(b'a', 2)])
     >>> list(d2.keys()) # should still be in last-set order
     ['b', 'a']
     >>> d1.insert(1, b'a.5', 0.5)
     >>> d1
     sortdict([('a', 0), ('a.5', 0.5), ('b', 1)])
     """
 
     def __setitem__(self, key, value):
         if key in self:
             del self[key]
         super(sortdict, self).__setitem__(key, value)
 
     if pycompat.ispypy:
         # __setitem__() isn't called as of PyPy 5.8.0
         def update(self, src, **f):
             if isinstance(src, dict):
                 src = pycompat.iteritems(src)
             for k, v in src:
                 self[k] = v
             for k in f:
                 self[k] = f[k]
 
     def insert(self, position, key, value):
         for (i, (k, v)) in enumerate(list(self.items())):
             if i == position:
                 self[key] = value
             if i >= position:
                 del self[k]
                 self[k] = v
 
 
 class cowdict(cow, dict):
     """copy-on-write dict
 
     Be sure to call d = d.preparewrite() before writing to d.
 
     >>> a = cowdict()
     >>> a is a.preparewrite()
     True
     >>> b = a.copy()
     >>> b is a
     True
     >>> c = b.copy()
     >>> c is a
     True
     >>> a = a.preparewrite()
     >>> b is a
     False
     >>> a is a.preparewrite()
     True
     >>> c = c.preparewrite()
     >>> b is c
     False
     >>> b is b.preparewrite()
     True
     """
 
 
 class cowsortdict(cow, sortdict):
     """copy-on-write sortdict
 
     Be sure to call d = d.preparewrite() before writing to d.
     """
 
 
 class transactional(object):  # pytype: disable=ignored-metaclass
     """Base class for making a transactional type into a context manager."""
 
     __metaclass__ = abc.ABCMeta
 
     @abc.abstractmethod
     def close(self):
         """Successfully closes the transaction."""
 
     @abc.abstractmethod
     def release(self):
         """Marks the end of the transaction.
 
         If the transaction has not been closed, it will be aborted.
         """
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         try:
             if exc_type is None:
                 self.close()
         finally:
             self.release()
 
 
 @contextlib.contextmanager
 def acceptintervention(tr=None):
     """A context manager that closes the transaction on InterventionRequired
 
     If no transaction was provided, this simply runs the body and returns
     """
     if not tr:
         yield
         return
     try:
         yield
         tr.close()
     except error.InterventionRequired:
         tr.close()
         raise
     finally:
         tr.release()
 
 
 @contextlib.contextmanager
 def nullcontextmanager(enter_result=None):
     yield enter_result
 
 
 class _lrucachenode(object):
     """A node in a doubly linked list.
 
     Holds a reference to nodes on either side as well as a key-value
     pair for the dictionary entry.
     """
 
     __slots__ = ('next', 'prev', 'key', 'value', 'cost')
 
     def __init__(self):
         self.next = self
         self.prev = self
 
         self.key = _notset
         self.value = None
         self.cost = 0
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
         self.value = None
         self.cost = 0
 
 
 class lrucachedict(object):
     """Dict that caches most recent accesses and sets.
 
     The dict consists of an actual backing dict - indexed by original
     key - and a doubly linked circular list defining the order of entries in
     the cache.
 
     The head node is the newest entry in the cache. If the cache is full,
     we recycle head.prev and make it the new head. Cache accesses result in
     the node being moved to before the existing head and being marked as the
     new head node.
 
     Items in the cache can be inserted with an optional "cost" value. This is
     simply an integer that is specified by the caller. The cache can be queried
     for the total cost of all items presently in the cache.
 
     The cache can also define a maximum cost. If a cache insertion would
     cause the total cost of the cache to go beyond the maximum cost limit,
     nodes will be evicted to make room for the new code. This can be used
     to e.g. set a max memory limit and associate an estimated bytes size
     cost to each item in the cache. By default, no maximum cost is enforced.
     """
 
     def __init__(self, max, maxcost=0):
         self._cache = {}
 
         self._head = _lrucachenode()
         self._size = 1
         self.capacity = max
         self.totalcost = 0
         self.maxcost = maxcost
 
     def __len__(self):
         return len(self._cache)
 
     def __contains__(self, k):
         return k in self._cache
 
     def __iter__(self):
         # We don't have to iterate in cache order, but why not.
         n = self._head
         for i in range(len(self._cache)):
             yield n.key
             n = n.next
 
     def __getitem__(self, k):
         node = self._cache[k]
         self._movetohead(node)
         return node.value
 
     def insert(self, k, v, cost=0):
         """Insert a new item in the cache with optional cost value."""
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             self.totalcost -= node.cost
             node.value = v
             node.cost = cost
             self.totalcost += cost
             self._movetohead(node)
 
             if self.maxcost:
                 self._enforcecostlimit()
 
             return
 
         if self._size < self.capacity:
             node = self._addcapacity()
         else:
             # Grab the last/oldest item.
             node = self._head.prev
 
         # At capacity. Kill the old entry.
         if node.key is not _notset:
             self.totalcost -= node.cost
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         node.cost = cost
         self.totalcost += cost
         self._cache[k] = node
         # And mark it as newest entry. No need to adjust order since it
         # is already self._head.prev.
         self._head = node
 
         if self.maxcost:
             self._enforcecostlimit()
 
     def __setitem__(self, k, v):
         self.insert(k, v)
 
     def __delitem__(self, k):
         self.pop(k)
 
     def pop(self, k, default=_notset):
         try:
             node = self._cache.pop(k)
         except KeyError:
             if default is _notset:
                 raise
             return default
 
         assert node is not None  # help pytype
         value = node.value
         self.totalcost -= node.cost
         node.markempty()
 
         # Temporarily mark as newest item before re-adjusting head to make
         # this node the oldest item.
         self._movetohead(node)
         self._head = node.next
 
         return value
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self.__getitem__(k)
         except KeyError:
             return default
 
     def peek(self, k, default=_notset):
         """Get the specified item without moving it to the head
 
         Unlike get(), this doesn't mutate the internal state. But be aware
         that it doesn't mean peek() is thread safe.
         """
         try:
             node = self._cache[k]
             assert node is not None  # help pytype
             return node.value
         except KeyError:
             if default is _notset:
                 raise
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             self.totalcost -= n.cost
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self, capacity=None, maxcost=0):
         """Create a new cache as a copy of the current one.
 
         By default, the new cache has the same capacity as the existing one.
         But, the cache capacity can be changed as part of performing the
         copy.
 
         Items in the copy have an insertion/access order matching this
         instance.
         """
 
         capacity = capacity or self.capacity
         maxcost = maxcost or self.maxcost
         result = lrucachedict(capacity, maxcost=maxcost)
 
         # We copy entries by iterating in oldest-to-newest order so the copy
         # has the correct ordering.
 
         # Find the first non-empty entry.
         n = self._head.prev
         while n.key is _notset and n is not self._head:
             n = n.prev
 
         # We could potentially skip the first N items when decreasing capacity.
         # But let's keep it simple unless it is a performance problem.
         for i in range(len(self._cache)):
             result.insert(n.key, n.value, cost=n.cost)
             n = n.prev
 
         return result
 
     def popoldest(self):
         """Remove the oldest item from the cache.
 
         Returns the (key, value) describing the removed cache entry.
         """
         if not self._cache:
             return
 
         # Walk the linked list backwards starting at tail node until we hit
         # a non-empty node.
         n = self._head.prev
 
         assert n is not None  # help pytype
 
         while n.key is _notset:
             n = n.prev
 
         assert n is not None  # help pytype
 
         key, value = n.key, n.value
 
         # And remove it from the cache and mark it as empty.
         del self._cache[n.key]
         self.totalcost -= n.cost
         n.markempty()
 
         return key, value
 
     def _movetohead(self, node):
         """Mark a node as the newest, making it the new head.
 
         When a node is accessed, it becomes the freshest entry in the LRU
         list, which is denoted by self._head.
 
         Visually, let's make ``N`` the new head node (* denotes head):
 
             previous/oldest <-> head <-> next/next newest
 
             ----<->--- A* ---<->-----
             |                       |
             E <-> D <-> N <-> C <-> B
 
         To:
 
             ----<->--- N* ---<->-----
             |                       |
             E <-> D <-> C <-> B <-> A
 
         This requires the following moves:
 
            C.next = D  (node.prev.next = node.next)
            D.prev = C  (node.next.prev = node.prev)
            E.next = N  (head.prev.next = node)
            N.prev = E  (node.prev = head.prev)
            N.next = A  (node.next = head)
            A.prev = N  (head.prev = node)
         """
         head = self._head
         # C.next = D
         node.prev.next = node.next
         # D.prev = C
         node.next.prev = node.prev
         # N.prev = E
         node.prev = head.prev
         # N.next = A
         # It is tempting to do just "head" here, however if node is
         # adjacent to head, this will do bad things.
         node.next = head.prev.next
         # E.next = N
         node.next.prev = node
         # A.prev = N
         node.prev.next = node
 
         self._head = node
 
     def _addcapacity(self):
         """Add a node to the circular linked list.
 
         The new node is inserted before the head node.
         """
         head = self._head
         node = _lrucachenode()
         head.prev.next = node
         node.prev = head.prev
         node.next = head
         head.prev = node
         self._size += 1
         return node
 
     def _enforcecostlimit(self):
         # This should run after an insertion. It should only be called if total
         # cost limits are being enforced.
         # The most recently inserted node is never evicted.
         if len(self) <= 1 or self.totalcost <= self.maxcost:
             return
 
         # This is logically equivalent to calling popoldest() until we
         # free up enough cost. We don't do that since popoldest() needs
         # to walk the linked list and doing this in a loop would be
         # quadratic. So we find the first non-empty node and then
         # walk nodes until we free up enough capacity.
         #
         # If we only removed the minimum number of nodes to free enough
         # cost at insert time, chances are high that the next insert would
         # also require pruning. This would effectively constitute quadratic
         # behavior for insert-heavy workloads. To mitigate this, we set a
         # target cost that is a percentage of the max cost. This will tend
         # to free more nodes when the high water mark is reached, which
         # lowers the chances of needing to prune on the subsequent insert.
         targetcost = int(self.maxcost * 0.75)
 
         n = self._head.prev
         while n.key is _notset:
             n = n.prev
 
         while len(self) > 1 and self.totalcost > targetcost:
             del self._cache[n.key]
             self.totalcost -= n.cost
             n.markempty()
             n = n.prev
 
 
 def lrucachefunc(func):
     '''cache most recent results of function calls'''
     cache = {}
     order = collections.deque()
     if func.__code__.co_argcount == 1:
 
         def f(arg):
             if arg not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[arg] = func(arg)
             else:
                 order.remove(arg)
             order.append(arg)
             return cache[arg]
 
     else:
 
         def f(*args):
             if args not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[args] = func(*args)
             else:
                 order.remove(args)
             order.append(args)
             return cache[args]
 
     return f
 
 
 class propertycache(object):
     def __init__(self, func):
         self.func = func
         self.name = func.__name__
 
     def __get__(self, obj, type=None):
         result = self.func(obj)
         self.cachevalue(obj, result)
         return result
 
     def cachevalue(self, obj, value):
         # __dict__ assignment required to bypass __setattr__ (eg: repoview)
         obj.__dict__[self.name] = value
 
 
 def clearcachedproperty(obj, prop):
     '''clear a cached property value, if one has been set'''
     prop = pycompat.sysstr(prop)
     if prop in obj.__dict__:
         del obj.__dict__[prop]
 
 
 def increasingchunks(source, min=1024, max=65536):
     """return no less than min bytes per chunk while data remains,
     doubling min after each chunk until it reaches max"""
 
     def log2(x):
         if not x:
             return 0
         i = 0
         while x:
             x >>= 1
             i += 1
         return i - 1
 
     buf = []
     blen = 0
     for chunk in source:
         buf.append(chunk)
         blen += len(chunk)
         if blen >= min:
             if min < max:
                 min = min << 1
                 nmin = 1 << log2(blen)
                 if nmin > min:
                     min = nmin
                 if min > max:
                     min = max
             yield b''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield b''.join(buf)
 
 
 def always(fn):
     return True
 
 
 def never(fn):
     return False
 
 
 def nogc(func):
     """disable garbage collector
 
     Python's garbage collector triggers a GC each time a certain number of
     container objects (the number being defined by gc.get_threshold()) are
     allocated even when marked not to be tracked by the collector. Tracking has
     no effect on when GCs are triggered, only on what objects the GC looks
     into. As a workaround, disable GC while building complex (huge)
     containers.
 
     This garbage collector issue have been fixed in 2.7. But it still affect
     CPython's performance.
     """
 
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
 
     return wrapper
 
 
 if pycompat.ispypy:
     # PyPy runs slower with gc disabled
     nogc = lambda x: x
 
 
 def pathto(root, n1, n2):
     # type: (bytes, bytes, bytes) -> bytes
     """return the relative path from one place to another.
     root should use os.sep to separate directories
     n1 should use os.sep to separate directories
     n2 should use "/" to separate directories
     returns an os.sep-separated path.
 
     If n1 is a relative path, it's assumed it's
     relative to root.
     n2 should always be relative to root.
     """
     if not n1:
         return localpath(n2)
     if os.path.isabs(n1):
         if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]:
             return os.path.join(root, localpath(n2))
         n2 = b'/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split(b'/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return pycompat.ossep.join(([b'..'] * len(a)) + b) or b'.'
 
 
 def checksignature(func, depth=1):
     '''wrap a function with code to check for calling errors'''
 
     def check(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except TypeError:
             if len(traceback.extract_tb(sys.exc_info()[2])) == depth:
                 raise error.SignatureError
             raise
 
     return check
 
 
 # a whilelist of known filesystems where hardlink works reliably
 _hardlinkfswhitelist = {
     b'apfs',
     b'btrfs',
     b'ext2',
     b'ext3',
     b'ext4',
     b'hfs',
     b'jfs',
     b'NTFS',
     b'reiserfs',
     b'tmpfs',
     b'ufs',
     b'xfs',
     b'zfs',
 }
 
 
 def copyfile(
     src,
     dest,
     hardlink=False,
     copystat=False,
     checkambig=False,
     nb_bytes=None,
     no_hardlink_cb=None,
     check_fs_hardlink=True,
 ):
     """copy a file, preserving mode and optionally other stat info like
     atime/mtime
 
     checkambig argument is used with filestat, and is useful only if
     destination file is guarded by any lock (e.g. repo.lock or
     repo.wlock).
 
     copystat and checkambig should be exclusive.
 
     nb_bytes: if set only copy the first `nb_bytes` of the source file.
     """
     assert not (copystat and checkambig)
     oldstat = None
     if os.path.lexists(dest):
         if checkambig:
             oldstat = checkambig and filestat.frompath(dest)
         unlink(dest)
     if hardlink and check_fs_hardlink:
         # Hardlinks are problematic on CIFS (issue4546), do not allow hardlinks
         # unless we are confident that dest is on a whitelisted filesystem.
         try:
             fstype = getfstype(os.path.dirname(dest))
         except OSError:
             fstype = None
         if fstype not in _hardlinkfswhitelist:
             if no_hardlink_cb is not None:
                 no_hardlink_cb()
             hardlink = False
     if hardlink:
         try:
             oslink(src, dest)
             if nb_bytes is not None:
                 m = "the `nb_bytes` argument is incompatible with `hardlink`"
                 raise error.ProgrammingError(m)
             return
         except (IOError, OSError) as exc:
             if exc.errno != errno.EEXIST and no_hardlink_cb is not None:
                 no_hardlink_cb()
             # fall back to normal copy
     if os.path.islink(src):
         os.symlink(os.readlink(src), dest)
         # copytime is ignored for symlinks, but in general copytime isn't needed
         # for them anyway
         if nb_bytes is not None:
             m = "cannot use `nb_bytes` on a symlink"
             raise error.ProgrammingError(m)
     else:
         try:
             shutil.copyfile(src, dest)
             if copystat:
                 # copystat also copies mode
                 shutil.copystat(src, dest)
             else:
                 shutil.copymode(src, dest)
                 if oldstat and oldstat.stat:
                     newstat = filestat.frompath(dest)
                     if newstat.isambig(oldstat):
                         # stat of copied file is ambiguous to original one
                         advanced = (
                             oldstat.stat[stat.ST_MTIME] + 1
                         ) & 0x7FFFFFFF
                         os.utime(dest, (advanced, advanced))
             # We could do something smarter using `copy_file_range` call or similar
             if nb_bytes is not None:
                 with open(dest, mode='r+') as f:
                     f.truncate(nb_bytes)
         except shutil.Error as inst:
             raise error.Abort(stringutil.forcebytestr(inst))
 
 
 def copyfiles(src, dst, hardlink=None, progress=None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     def settopic():
         if progress:
             progress.topic = _(b'linking') if hardlink else _(b'copying')
 
     if os.path.isdir(src):
         if hardlink is None:
             hardlink = (
                 os.stat(src).st_dev == os.stat(os.path.dirname(dst)).st_dev
             )
         settopic()
         os.mkdir(dst)
         for name, kind in listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress)
             num += n
     else:
         if hardlink is None:
             hardlink = (
                 os.stat(os.path.dirname(src)).st_dev
                 == os.stat(os.path.dirname(dst)).st_dev
             )
         settopic()
 
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError) as exc:
                 if exc.errno != errno.EEXIST:
                     hardlink = False
                 # XXX maybe try to relink if the file exist ?
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         if progress:
             progress.increment()
 
     return hardlink, num
 
 
 _winreservednames = {
     b'con',
     b'prn',
     b'aux',
     b'nul',
     b'com1',
     b'com2',
     b'com3',
     b'com4',
     b'com5',
     b'com6',
     b'com7',
     b'com8',
     b'com9',
     b'lpt1',
     b'lpt2',
     b'lpt3',
     b'lpt4',
     b'lpt5',
     b'lpt6',
     b'lpt7',
     b'lpt8',
     b'lpt9',
 }
 _winreservedchars = b':*?"<>|'
 
 
 def checkwinfilename(path):
     # type: (bytes) -> Optional[bytes]
     r"""Check that the base-relative path is a valid filename on Windows.
     Returns None if the path is ok, or a UI string describing the problem.
 
     >>> checkwinfilename(b"just/a/normal/path")
     >>> checkwinfilename(b"foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/xml.con")
     >>> checkwinfilename(b"foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename(b"foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename(b"../bar")
     >>> checkwinfilename(b"foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename(b"foo\\/bar")
     "directory name ends with '\\', which is invalid on Windows"
     """
     if path.endswith(b'\\'):
         return _(b"filename ends with '\\', which is invalid on Windows")
     if b'\\/' in path:
         return _(b"directory name ends with '\\', which is invalid on Windows")
     for n in path.replace(b'\\', b'/').split(b'/'):
         if not n:
             continue
         for c in _filenamebytestr(n):
             if c in _winreservedchars:
                 return (
                     _(
                         b"filename contains '%s', which is reserved "
                         b"on Windows"
                     )
                     % c
                 )
             if ord(c) <= 31:
                 return _(
                     b"filename contains '%s', which is invalid on Windows"
                 ) % stringutil.escapestr(c)
         base = n.split(b'.')[0]
         if base and base.lower() in _winreservednames:
             return (
                 _(b"filename contains '%s', which is reserved on Windows")
                 % base
             )
         t = n[-1:]
         if t in b'. ' and n not in b'..':
             return (
                 _(
                     b"filename ends with '%s', which is not allowed "
                     b"on Windows"
                 )
                 % t
             )
 
 
 timer = getattr(time, "perf_counter", None)
 
 if pycompat.iswindows:
     checkosfilename = checkwinfilename
     if not timer:
         timer = time.clock
 else:
     # mercurial.windows doesn't have platform.checkosfilename
     checkosfilename = platform.checkosfilename  # pytype: disable=module-attr
     if not timer:
         timer = time.time
 
 
 def makelock(info, pathname):
     """Create a lock file atomically if possible
 
     This may leave a stale lock file if symlink isn't supported and signal
     interrupt is enabled.
     """
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError:  # no symlink in os
         pass
 
     flags = os.O_CREAT | os.O_WRONLY | os.O_EXCL | getattr(os, 'O_BINARY', 0)
     ld = os.open(pathname, flags)
     os.write(ld, info)
     os.close(ld)
 
 
 def readlock(pathname):
     # type: (bytes) -> bytes
     try:
         return readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError:  # no symlink in os
         pass
     with posixfile(pathname, b'rb') as fp:
         return fp.read()
 
 
 def fstat(fp):
     '''stat file object that may not have fileno method.'''
     try:
         return os.fstat(fp.fileno())
     except AttributeError:
         return os.stat(fp.name)
 
 
 # File system features
 
 
 def fscasesensitive(path):
     # type: (bytes) -> bool
     """
     Return true if the given path is on a case-sensitive filesystem
 
     Requires a path (like /foo/.hg) ending with a foldable final
     directory component.
     """
     s1 = os.lstat(path)
     d, b = os.path.split(path)
     b2 = b.upper()
     if b == b2:
         b2 = b.lower()
         if b == b2:
             return True  # no evidence against case sensitivity
     p2 = os.path.join(d, b2)
     try:
         s2 = os.lstat(p2)
         if s2 == s1:
             return False
         return True
     except OSError:
         return True
 
 
 _re2_input = lambda x: x
 try:
     import re2  # pytype: disable=import-error
 
     _re2 = None
 except ImportError:
     _re2 = False
 
 
 class _re(object):
     def _checkre2(self):
         global _re2
         global _re2_input
 
         check_pattern = br'\[([^\[]+)\]'
         check_input = b'[ui]'
         try:
             # check if match works, see issue3964
             _re2 = bool(re2.match(check_pattern, check_input))
         except ImportError:
             _re2 = False
         except TypeError:
             # the `pyre-2` project provides a re2 module that accept bytes
             # the `fb-re2` project provides a re2 module that acccept sysstr
             check_pattern = pycompat.sysstr(check_pattern)
             check_input = pycompat.sysstr(check_input)
             _re2 = bool(re2.match(check_pattern, check_input))
             _re2_input = pycompat.sysstr
 
     def compile(self, pat, flags=0):
         """Compile a regular expression, using re2 if possible
 
         For best performance, use only re2-compatible regexp features. The
         only flags from the re module that are re2-compatible are
         IGNORECASE and MULTILINE."""
         if _re2 is None:
             self._checkre2()
         if _re2 and (flags & ~(remod.IGNORECASE | remod.MULTILINE)) == 0:
             if flags & remod.IGNORECASE:
                 pat = b'(?i)' + pat
             if flags & remod.MULTILINE:
                 pat = b'(?m)' + pat
             try:
                 return re2.compile(_re2_input(pat))
             except re2.error:
                 pass
         return remod.compile(pat, flags)
 
     @propertycache
     def escape(self):
         """Return the version of escape corresponding to self.compile.
 
         This is imperfect because whether re2 or re is used for a particular
         function depends on the flags, etc, but it's the best we can do.
         """
         global _re2
         if _re2 is None:
             self._checkre2()
         if _re2:
             return re2.escape
         else:
             return remod.escape
 
 
 re = _re()
 
 _fspathcache = {}
 
 
 def fspath(name, root):
     # type: (bytes, bytes) -> bytes
     """Get name in the case stored in the filesystem
 
     The name should be relative to root, and be normcase-ed for efficiency.
 
     Note that this function is unnecessary, and should not be
     called, for case-sensitive filesystems (simply because it's expensive).
 
     The root should be normcase-ed, too.
     """
 
     def _makefspathcacheentry(dir):
         return {normcase(n): n for n in os.listdir(dir)}
 
     seps = pycompat.ossep
     if pycompat.osaltsep:
         seps = seps + pycompat.osaltsep
     # Protect backslashes. This gets silly very quickly.
     seps.replace(b'\\', b'\\\\')
     pattern = remod.compile(br'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normpath(root)
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = _makefspathcacheentry(dir)
         contents = _fspathcache[dir]
 
         found = contents.get(part)
         if not found:
             # retry "once per directory" per "dirstate.walk" which
             # may take place for each patches of "hg qpush", for example
             _fspathcache[dir] = contents = _makefspathcacheentry(dir)
             found = contents.get(part)
 
         result.append(found or part)
         dir = os.path.join(dir, part)
 
     return b''.join(result)
 
 
 def checknlink(testfile):
     # type: (bytes) -> bool
     '''check whether hardlink count reporting works properly'''
 
     # testfile may be open, so we need a separate file for checking to
     # work around issue2543 (or testfile may get lost on Samba shares)
     f1, f2, fp = None, None, None
     try:
         fd, f1 = pycompat.mkstemp(
             prefix=b'.%s-' % os.path.basename(testfile),
             suffix=b'1~',
             dir=os.path.dirname(testfile),
         )
         os.close(fd)
         f2 = b'%s2~' % f1[:-2]
 
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fp = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fp is not None:
             fp.close()
         for f in (f1, f2):
             try:
                 if f is not None:
                     os.unlink(f)
             except OSError:
                 pass
 
 
 def endswithsep(path):
     # type: (bytes) -> bool
     '''Check path ends with os.sep or os.altsep.'''
     return bool(  # help pytype
         path.endswith(pycompat.ossep)
         or pycompat.osaltsep
         and path.endswith(pycompat.osaltsep)
     )
 
 
 def splitpath(path):
     # type: (bytes) -> List[bytes]
     """Split path by os.sep.
     Note that this function does not use os.altsep because this is
     an alternative of simple "xxx.split(os.sep)".
     It is recommended to use os.path.normpath() before using this
     function if need."""
     return path.split(pycompat.ossep)
 
 
 def mktempcopy(name, emptyok=False, createmode=None, enforcewritable=False):
     """Create a temporary file with the same contents from name
 
     The permission bits are copied from the original file.
 
     If the temporary file is going to be truncated immediately, you
     can use emptyok=True as an optimization.
 
     Returns the name of the temporary file.
     """
     d, fn = os.path.split(name)
     fd, temp = pycompat.mkstemp(prefix=b'.%s-' % fn, suffix=b'~', dir=d)
     os.close(fd)
     # Temporary files are created with mode 0600, which is usually not
     # what we want.  If the original file already exists, just copy
     # its mode.  Otherwise, manually obey umask.
     copymode(name, temp, createmode, enforcewritable)
 
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, b"rb")
         except IOError as inst:
             if inst.errno == errno.ENOENT:
                 return temp
             if not getattr(inst, 'filename', None):
                 inst.filename = name
             raise
         ofp = posixfile(temp, b"wb")
         for chunk in filechunkiter(ifp):
             ofp.write(chunk)
         ifp.close()
         ofp.close()
     except:  # re-raises
         try:
             os.unlink(temp)
         except OSError:
             pass
         raise
     return temp
 
 
 class filestat(object):
     """help to exactly detect change of a file
 
     'stat' attribute is result of 'os.stat()' if specified 'path'
     exists. Otherwise, it is None. This can avoid preparative
     'exists()' examination on client side of this class.
     """
 
     def __init__(self, stat):
         self.stat = stat
 
     @classmethod
     def frompath(cls, path):
         try:
             stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             stat = None
         return cls(stat)
 
     @classmethod
     def fromfp(cls, fp):
         stat = os.fstat(fp.fileno())
         return cls(stat)
 
     __hash__ = object.__hash__
 
     def __eq__(self, old):
         try:
             # if ambiguity between stat of new and old file is
             # avoided, comparison of size, ctime and mtime is enough
             # to exactly detect change of a file regardless of platform
             return (
                 self.stat.st_size == old.stat.st_size
                 and self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME]
                 and self.stat[stat.ST_MTIME] == old.stat[stat.ST_MTIME]
             )
         except AttributeError:
             pass
         try:
             return self.stat is None and old.stat is None
         except AttributeError:
             return False
 
     def isambig(self, old):
         """Examine whether new (= self) stat is ambiguous against old one
 
         "S[N]" below means stat of a file at N-th change:
 
         - S[n-1].ctime  < S[n].ctime: can detect change of a file
         - S[n-1].ctime == S[n].ctime
           - S[n-1].ctime  < S[n].mtime: means natural advancing (*1)
           - S[n-1].ctime == S[n].mtime: is ambiguous (*2)
           - S[n-1].ctime  > S[n].mtime: never occurs naturally (don't care)
         - S[n-1].ctime  > S[n].ctime: never occurs naturally (don't care)
 
         Case (*2) above means that a file was changed twice or more at
         same time in sec (= S[n-1].ctime), and comparison of timestamp
         is ambiguous.
 
         Base idea to avoid such ambiguity is "advance mtime 1 sec, if
         timestamp is ambiguous".
 
         But advancing mtime only in case (*2) doesn't work as
         expected, because naturally advanced S[n].mtime in case (*1)
         might be equal to manually advanced S[n-1 or earlier].mtime.
 
         Therefore, all "S[n-1].ctime == S[n].ctime" cases should be
         treated as ambiguous regardless of mtime, to avoid overlooking
         by confliction between such mtime.
 
         Advancing mtime "if isambig(oldstat)" ensures "S[n-1].mtime !=
         S[n].mtime", even if size of a file isn't changed.
         """
         try:
             return self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME]
         except AttributeError:
             return False
 
     def avoidambig(self, path, old):
         """Change file stat of specified path to avoid ambiguity
 
         'old' should be previous filestat of 'path'.
 
         This skips avoiding ambiguity, if a process doesn't have
         appropriate privileges for 'path'. This returns False in this
         case.
 
         Otherwise, this returns True, as "ambiguity is avoided".
         """
         advanced = (old.stat[stat.ST_MTIME] + 1) & 0x7FFFFFFF
         try:
             os.utime(path, (advanced, advanced))
         except OSError as inst:
             if inst.errno == errno.EPERM:
                 # utime() on the file created by another user causes EPERM,
                 # if a process doesn't have appropriate privileges
                 return False
             raise
         return True
 
     def __ne__(self, other):
         return not self == other
 
 
 class atomictempfile(object):
     """writable file object that atomically updates a file
 
     All writes will go to a temporary copy of the original file. Call
     close() when you are done writing, and atomictempfile will rename
     the temporary copy to the original name, making the changes
     visible. If the object is destroyed without being closed, all your
     writes are discarded.
 
     checkambig argument of constructor is used with filestat, and is
     useful only if target file is guarded by any lock (e.g. repo.lock
     or repo.wlock).
     """
 
     def __init__(self, name, mode=b'w+b', createmode=None, checkambig=False):
         self.__name = name  # permanent name
         self._tempname = mktempcopy(
             name,
             emptyok=(b'w' in mode),
             createmode=createmode,
             enforcewritable=(b'w' in mode),
         )
 
         self._fp = posixfile(self._tempname, mode)
         self._checkambig = checkambig
 
         # delegated methods
         self.read = self._fp.read
         self.write = self._fp.write
         self.seek = self._fp.seek
         self.tell = self._fp.tell
         self.fileno = self._fp.fileno
 
     def close(self):
         if not self._fp.closed:
             self._fp.close()
             filename = localpath(self.__name)
             oldstat = self._checkambig and filestat.frompath(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat.frompath(filename)
                 if newstat.isambig(oldstat):
                     # stat of changed file is ambiguous to original one
                     advanced = (oldstat.stat[stat.ST_MTIME] + 1) & 0x7FFFFFFF
                     os.utime(filename, (advanced, advanced))
             else:
                 rename(self._tempname, filename)
 
     def discard(self):
         if not self._fp.closed:
             try:
                 os.unlink(self._tempname)
             except OSError:
                 pass
             self._fp.close()
 
     def __del__(self):
         if safehasattr(self, '_fp'):  # constructor actually did something
             self.discard()
 
     def __enter__(self):
         return self
 
     def __exit__(self, exctype, excvalue, traceback):
         if exctype is not None:
             self.discard()
         else:
             self.close()
 
 
 def unlinkpath(f, ignoremissing=False, rmdir=True):
     # type: (bytes, bool, bool) -> None
     """unlink and remove the directory if it is empty"""
     if ignoremissing:
         tryunlink(f)
     else:
         unlink(f)
     if rmdir:
         # try removing directories that might now be empty
         try:
             removedirs(os.path.dirname(f))
         except OSError:
             pass
 
 
 def tryunlink(f):
     # type: (bytes) -> None
     """Attempt to remove a file, ignoring ENOENT errors."""
     try:
         unlink(f)
     except OSError as e:
         if e.errno != errno.ENOENT:
             raise
 
 
 def makedirs(name, mode=None, notindexed=False):
     # type: (bytes, Optional[int], bool) -> None
     """recursive directory creation with parent mode inheritance
 
     Newly created directories are marked as "not to be indexed by
     the content indexing service", if ``notindexed`` is specified
     for "write" mode access.
     """
     try:
         makedir(name, notindexed)
     except OSError as err:
         if err.errno == errno.EEXIST:
             return
         if err.errno != errno.ENOENT or not name:
             raise
         parent = os.path.dirname(abspath(name))
         if parent == name:
             raise
         makedirs(parent, mode, notindexed)
         try:
             makedir(name, notindexed)
         except OSError as err:
             # Catch EEXIST to handle races
             if err.errno == errno.EEXIST:
                 return
             raise
     if mode is not None:
         os.chmod(name, mode)
 
 
 def readfile(path):
     # type: (bytes) -> bytes
     with open(path, b'rb') as fp:
         return fp.read()
 
 
 def writefile(path, text):
     # type: (bytes, bytes) -> None
     with open(path, b'wb') as fp:
         fp.write(text)
 
 
 def appendfile(path, text):
     # type: (bytes, bytes) -> None
     with open(path, b'ab') as fp:
         fp.write(text)
 
 
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter):
         """in_iter is the iterator that's iterating over the input chunks."""
 
         def splitbig(chunks):
             for chunk in chunks:
                 if len(chunk) > 2 ** 20:
                     pos = 0
                     while pos < len(chunk):
                         end = pos + 2 ** 18
                         yield chunk[pos:end]
                         pos = end
                 else:
                     yield chunk
 
         self.iter = splitbig(in_iter)
         self._queue = collections.deque()
         self._chunkoffset = 0
 
     def read(self, l=None):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry.
 
         If size parameter is omitted, read everything"""
         if l is None:
             return b''.join(self.iter)
 
         left = l
         buf = []
         queue = self._queue
         while left > 0:
             # refill the queue
             if not queue:
                 target = 2 ** 18
                 for chunk in self.iter:
                     queue.append(chunk)
                     target -= len(chunk)
                     if target <= 0:
                         break
                 if not queue:
                     break
 
             # The easy way to do this would be to queue.popleft(), modify the
             # chunk (if necessary), then queue.appendleft(). However, for cases
             # where we read partial chunk content, this incurs 2 dequeue
             # mutations and creates a new str for the remaining chunk in the
             # queue. Our code below avoids this overhead.
 
             chunk = queue[0]
             chunkl = len(chunk)
             offset = self._chunkoffset
 
             # Use full chunk.
             if offset == 0 and left >= chunkl:
                 left -= chunkl
                 queue.popleft()
                 buf.append(chunk)
                 # self._chunkoffset remains at 0.
                 continue
 
             chunkremaining = chunkl - offset
 
             # Use all of unconsumed part of chunk.
             if left >= chunkremaining:
                 left -= chunkremaining
                 queue.popleft()
                 # offset == 0 is enabled by block above, so this won't merely
                 # copy via ``chunk[0:]``.
                 buf.append(chunk[offset:])
                 self._chunkoffset = 0
 
             # Partial chunk needed.
             else:
                 buf.append(chunk[offset : offset + left])
                 self._chunkoffset += left
                 left -= chunkremaining
 
         return b''.join(buf)
 
 
 def filechunkiter(f, size=131072, limit=None):
     """Create a generator that produces the data in the file size
     (default 131072) bytes at a time, up to optional limit (default is
     to read all data).  Chunks may be less than size bytes if the
     chunk is the last chunk in the file, or the file is a socket or
     some other type of file that sometimes reads less data than is
     requested."""
     assert size >= 0
     assert limit is None or limit >= 0
     while True:
         if limit is None:
             nbytes = size
         else:
             nbytes = min(limit, size)
         s = nbytes and f.read(nbytes)
         if not s:
             break
         if limit:
             limit -= len(s)
         yield s
 
 
 class cappedreader(object):
     """A file object proxy that allows reading up to N bytes.
 
     Given a source file object, instances of this type allow reading up to
     N bytes from that source file object. Attempts to read past the allowed
     limit are treated as EOF.
 
     It is assumed that I/O is not performed on the original file object
     in addition to I/O that is performed by this instance. If there is,
     state tracking will get out of sync and unexpected results will ensue.
     """
 
     def __init__(self, fh, limit):
         """Allow reading up to <limit> bytes from <fh>."""
         self._fh = fh
         self._left = limit
 
     def read(self, n=-1):
         if not self._left:
             return b''
 
         if n < 0:
             n = self._left
 
         data = self._fh.read(min(n, self._left))
         self._left -= len(data)
         assert self._left >= 0
 
         return data
 
     def readinto(self, b):
         res = self.read(len(b))
         if res is None:
             return None
 
         b[0 : len(res)] = res
         return len(res)
 
 
 def unitcountfn(*unittable):
     '''return a function that renders a readable count of some quantity'''
 
     def go(count):
         for multiplier, divisor, format in unittable:
             if abs(count) >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 
 def processlinerange(fromline, toline):
     # type: (int, int) -> Tuple[int, int]
     """Check that linerange <fromline>:<toline> makes sense and return a
     0-based range.
 
     >>> processlinerange(10, 20)
     (9, 20)
     >>> processlinerange(2, 1)
     Traceback (most recent call last):
         ...
     ParseError: line range must be positive
     >>> processlinerange(0, 5)
     Traceback (most recent call last):
         ...
     ParseError: fromline must be strictly positive
     """
     if toline - fromline < 0:
         raise error.ParseError(_(b"line range must be positive"))
     if fromline < 1:
         raise error.ParseError(_(b"fromline must be strictly positive"))
     return fromline - 1, toline
 
 
 bytecount = unitcountfn(
     (100, 1 << 30, _(b'%.0f GB')),
     (10, 1 << 30, _(b'%.1f GB')),
     (1, 1 << 30, _(b'%.2f GB')),
     (100, 1 << 20, _(b'%.0f MB')),
     (10, 1 << 20, _(b'%.1f MB')),
     (1, 1 << 20, _(b'%.2f MB')),
     (100, 1 << 10, _(b'%.0f KB')),
     (10, 1 << 10, _(b'%.1f KB')),
     (1, 1 << 10, _(b'%.2f KB')),
     (1, 1, _(b'%.0f bytes')),
 )
 
 
 class transformingwriter(object):
     """Writable file wrapper to transform data by function"""
 
     def __init__(self, fp, encode):
         self._fp = fp
         self._encode = encode
 
     def close(self):
         self._fp.close()
 
     def flush(self):
         self._fp.flush()
 
     def write(self, data):
         return self._fp.write(self._encode(data))
 
 
 # Matches a single EOL which can either be a CRLF where repeated CR
 # are removed or a LF. We do not care about old Macintosh files, so a
 # stray CR is an error.
 _eolre = remod.compile(br'\r*\n')
 
 
 def tolf(s):
     # type: (bytes) -> bytes
     return _eolre.sub(b'\n', s)
 
 
 def tocrlf(s):
     # type: (bytes) -> bytes
     return _eolre.sub(b'\r\n', s)
 
 
 def _crlfwriter(fp):
     return transformingwriter(fp, tocrlf)
 
 
 if pycompat.oslinesep == b'\r\n':
     tonativeeol = tocrlf
     fromnativeeol = tolf
     nativeeolwriter = _crlfwriter
 else:
     tonativeeol = pycompat.identity
     fromnativeeol = pycompat.identity
     nativeeolwriter = pycompat.identity
 
 if pyplatform.python_implementation() == b'CPython' and sys.version_info < (
     3,
     0,
 ):
     # There is an issue in CPython that some IO methods do not handle EINTR
     # correctly. The following table shows what CPython version (and functions)
     # are affected (buggy: has the EINTR bug, okay: otherwise):
     #
     #                | < 2.7.4 | 2.7.4 to 2.7.12 | >= 3.0
     #   --------------------------------------------------
     #    fp.__iter__ | buggy   | buggy           | okay
     #    fp.read*    | buggy   | okay [1]        | okay
     #
     # [1]: fixed by changeset 67dc99a989cd in the cpython hg repo.
     #
     # Here we workaround the EINTR issue for fileobj.__iter__. Other methods
     # like "read*" work fine, as we do not support Python < 2.7.4.
     #
     # Although we can workaround the EINTR issue for fp.__iter__, it is slower:
     # "for x in fp" is 4x faster than "for x in iter(fp.readline, '')" in
     # CPython 2, because CPython 2 maintains an internal readahead buffer for
     # fp.__iter__ but not other fp.read* methods.
     #
     # On modern systems like Linux, the "read" syscall cannot be interrupted
     # when reading "fast" files like on-disk files. So the EINTR issue only
     # affects things like pipes, sockets, ttys etc. We treat "normal" (S_ISREG)
     # files approximately as "fast" files and use the fast (unsafe) code path,
     # to minimize the performance impact.
 
     def iterfile(fp):
         fastpath = True
         if type(fp) is file:
             fastpath = stat.S_ISREG(os.fstat(fp.fileno()).st_mode)
         if fastpath:
             return fp
         else:
             # fp.readline deals with EINTR correctly, use it as a workaround.
             return iter(fp.readline, b'')
 
 
 else:
     # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed.
     def iterfile(fp):
         return fp
 
 
 def iterlines(iterator):
     # type: (Iterator[bytes]) -> Iterator[bytes]
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line
 
 
 def expandpath(path):
     # type: (bytes) -> bytes
     return os.path.expanduser(os.path.expandvars(path))
 
 
 def interpolate(prefix, mapping, s, fn=None, escape_prefix=False):
     """Return the result of interpolating items in the mapping into string s.
 
     prefix is a single character string, or a two character string with
     a backslash as the first character if the prefix needs to be escaped in
     a regular expression.
 
     fn is an optional function that will be applied to the replacement text
     just before replacement.
 
     escape_prefix is an optional flag that allows using doubled prefix for
     its escaping.
     """
     fn = fn or (lambda s: s)
     patterns = b'|'.join(mapping.keys())
     if escape_prefix:
         patterns += b'|' + prefix
         if len(prefix) > 1:
             prefix_char = prefix[1:]
         else:
             prefix_char = prefix
         mapping[prefix_char] = prefix_char
     r = remod.compile(br'%s(%s)' % (prefix, patterns))
     return r.sub(lambda x: fn(mapping[x.group()[1:]]), s)
 
 
 timecount = unitcountfn(
     (1, 1e3, _(b'%.0f s')),
     (100, 1, _(b'%.1f s')),
     (10, 1, _(b'%.2f s')),
     (1, 1, _(b'%.3f s')),
     (100, 0.001, _(b'%.1f ms')),
     (10, 0.001, _(b'%.2f ms')),
     (1, 0.001, _(b'%.3f ms')),
     (100, 0.000001, _(b'%.1f us')),
     (10, 0.000001, _(b'%.2f us')),
     (1, 0.000001, _(b'%.3f us')),
     (100, 0.000000001, _(b'%.1f ns')),
     (10, 0.000000001, _(b'%.2f ns')),
     (1, 0.000000001, _(b'%.3f ns')),
 )
 
 
 @attr.s
 class timedcmstats(object):
     """Stats information produced by the timedcm context manager on entering."""
 
     # the starting value of the timer as a float (meaning and resulution is
     # platform dependent, see util.timer)
     start = attr.ib(default=attr.Factory(lambda: timer()))
     # the number of seconds as a floating point value; starts at 0, updated when
     # the context is exited.
     elapsed = attr.ib(default=0)
     # the number of nested timedcm context managers.
     level = attr.ib(default=1)
 
     def __bytes__(self):
         return timecount(self.elapsed) if self.elapsed else b'<unknown>'
 
     __str__ = encoding.strmethod(__bytes__)
 
 
 @contextlib.contextmanager
 def timedcm(whencefmt, *whenceargs):
     """A context manager that produces timing information for a given context.
 
     On entering a timedcmstats instance is produced.
 
     This context manager is reentrant.
 
     """
     # track nested context managers
     timedcm._nested += 1
     timing_stats = timedcmstats(level=timedcm._nested)
     try:
         with tracing.log(whencefmt, *whenceargs):
             yield timing_stats
     finally:
         timing_stats.elapsed = timer() - timing_stats.start
         timedcm._nested -= 1
 
 
 timedcm._nested = 0
 
 
 def timed(func):
     """Report the execution time of a function call to stderr.
 
     During development, use as a decorator when you need to measure
     the cost of a function, e.g. as follows:
 
     @util.timed
     def foo(a, b, c):
         pass
     """
 
     def wrapper(*args, **kwargs):
         with timedcm(pycompat.bytestr(func.__name__)) as time_stats:
             result = func(*args, **kwargs)
         stderr = procutil.stderr
         stderr.write(
             b'%s%s: %s\n'
             % (
                 b' ' * time_stats.level * 2,
                 pycompat.bytestr(func.__name__),
                 time_stats,
             )
         )
         return result
 
     return wrapper
 
 
 _sizeunits = (
     (b'm', 2 ** 20),
     (b'k', 2 ** 10),
     (b'g', 2 ** 30),
     (b'kb', 2 ** 10),
     (b'mb', 2 ** 20),
     (b'gb', 2 ** 30),
     (b'b', 1),
 )
 
 
 def sizetoint(s):
     # type: (bytes) -> int
     """Convert a space specifier to a byte count.
 
     >>> sizetoint(b'30')
     30
     >>> sizetoint(b'2.2kb')
     2252
     >>> sizetoint(b'6M')
     6291456
     """
     t = s.strip().lower()
     try:
         for k, u in _sizeunits:
             if t.endswith(k):
                 return int(float(t[: -len(k)]) * u)
         return int(t)
     except ValueError:
         raise error.ParseError(_(b"couldn't parse size: %s") % s)
 
 
 class hooks(object):
     """A collection of hook functions that can be used to extend a
     function's behavior. Hooks are called in lexicographic order,
     based on the names of their sources."""
 
     def __init__(self):
         self._hooks = []
 
     def add(self, source, hook):
         self._hooks.append((source, hook))
 
     def __call__(self, *args):
         self._hooks.sort(key=lambda x: x[0])
         results = []
         for source, hook in self._hooks:
             results.append(hook(*args))
         return results
 
 
 def getstackframes(skip=0, line=b' %-*s in %s\n', fileline=b'%s:%d', depth=0):
     """Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last entries, then return the last 'depth' entries.
     Each file+linenumber is formatted according to fileline.
     Each line is formatted according to line.
     If line is None, it yields:
       length of longest filepath+line number,
       filepath+linenumber,
       function
 
     Not be used in production code but very convenient while developing.
     """
     entries = [
         (fileline % (pycompat.sysbytes(fn), ln), pycompat.sysbytes(func))
         for fn, ln, func, _text in traceback.extract_stack()[: -skip - 1]
     ][-depth:]
     if entries:
         fnmax = max(len(entry[0]) for entry in entries)
         for fnln, func in entries:
             if line is None:
                 yield (fnmax, fnln, func)
             else:
                 yield line % (fnmax, fnln, func)
 
 
 def debugstacktrace(
     msg=b'stacktrace',
     skip=0,
     f=procutil.stderr,
     otherf=procutil.stdout,
     depth=0,
     prefix=b'',
 ):
     """Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' entries closest to the call, then show 'depth' entries.
     By default it will flush stdout first.
     It can be used everywhere and intentionally does not require an ui object.
     Not be used in production code but very convenient while developing.
     """
     if otherf:
         otherf.flush()
     f.write(b'%s%s at:\n' % (prefix, msg.rstrip()))
     for line in getstackframes(skip + 1, depth=depth):
         f.write(prefix + line)
     f.flush()
 
 
 # convenient shortcut
 dst = debugstacktrace
 
 
 def safename(f, tag, ctx, others=None):
     """
     Generate a name that it is safe to rename f to in the given context.
 
     f:      filename to rename
     tag:    a string tag that will be included in the new name
     ctx:    a context, in which the new name must not exist
     others: a set of other filenames that the new name must not be in
 
     Returns a file name of the form oldname~tag[~number] which does not exist
     in the provided context and is not in the set of other names.
     """
     if others is None:
         others = set()
 
     fn = b'%s~%s' % (f, tag)
     if fn not in ctx and fn not in others:
         return fn
     for n in itertools.count(1):
         fn = b'%s~%s~%s' % (f, tag, n)
         if fn not in ctx and fn not in others:
             return fn
 
 
 def readexactly(stream, n):
     '''read n bytes from stream.read and abort if less was available'''
     s = stream.read(n)
     if len(s) < n:
         raise error.Abort(
             _(b"stream ended unexpectedly (got %d bytes, expected %d)")
             % (len(s), n)
         )
     return s
 
 
 def uvarintencode(value):
     """Encode an unsigned integer value to a varint.
 
     A varint is a variable length integer of 1 or more bytes. Each byte
     except the last has the most significant bit set. The lower 7 bits of
     each byte store the 2's complement representation, least significant group
     first.
 
     >>> uvarintencode(0)
     '\\x00'
     >>> uvarintencode(1)
     '\\x01'
     >>> uvarintencode(127)
     '\\x7f'
     >>> uvarintencode(1337)
     '\\xb9\\n'
     >>> uvarintencode(65536)
     '\\x80\\x80\\x04'
     >>> uvarintencode(-1)
     Traceback (most recent call last):
         ...
     ProgrammingError: negative value for uvarint: -1
     """
     if value < 0:
         raise error.ProgrammingError(b'negative value for uvarint: %d' % value)
     bits = value & 0x7F
     value >>= 7
     bytes = []
     while value:
         bytes.append(pycompat.bytechr(0x80 | bits))
         bits = value & 0x7F
         value >>= 7
     bytes.append(pycompat.bytechr(bits))
 
     return b''.join(bytes)
 
 
 def uvarintdecodestream(fh):
     """Decode an unsigned variable length integer from a stream.
 
     The passed argument is anything that has a ``.read(N)`` method.
 
     >>> try:
     ...     from StringIO import StringIO as BytesIO
     ... except ImportError:
     ...     from io import BytesIO
     >>> uvarintdecodestream(BytesIO(b'\\x00'))
     0
     >>> uvarintdecodestream(BytesIO(b'\\x01'))
     1
     >>> uvarintdecodestream(BytesIO(b'\\x7f'))
     127
     >>> uvarintdecodestream(BytesIO(b'\\xb9\\n'))
     1337
     >>> uvarintdecodestream(BytesIO(b'\\x80\\x80\\x04'))
     65536
     >>> uvarintdecodestream(BytesIO(b'\\x80'))
     Traceback (most recent call last):
         ...
     Abort: stream ended unexpectedly (got 0 bytes, expected 1)
     """
     result = 0
     shift = 0
     while True:
         byte = ord(readexactly(fh, 1))
         result |= (byte & 0x7F) << shift
         if not (byte & 0x80):
             return result
         shift += 7
 
 
 # Passing the '' locale means that the locale should be set according to the
 # user settings (environment variables).
 # Python sometimes avoids setting the global locale settings. When interfacing
 # with C code (e.g. the curses module or the Subversion bindings), the global
 # locale settings must be initialized correctly. Python 2 does not initialize
 # the global locale settings on interpreter startup. Python 3 sometimes
 # initializes LC_CTYPE, but not consistently at least on Windows. Therefore we
 # explicitly initialize it to get consistent behavior if it's not already
 # initialized. Since CPython commit 177d921c8c03d30daa32994362023f777624b10d,
 # LC_CTYPE is always initialized. If we require Python 3.8+, we should re-check
 # if we can remove this code.
 @contextlib.contextmanager
 def with_lc_ctype():
     oldloc = locale.setlocale(locale.LC_CTYPE, None)
     if oldloc == 'C':
         try:
             try:
                 locale.setlocale(locale.LC_CTYPE, '')
             except locale.Error:
                 # The likely case is that the locale from the environment
                 # variables is unknown.
                 pass
             yield
         finally:
             locale.setlocale(locale.LC_CTYPE, oldloc)
     else:
         yield
 
 
 def _estimatememory():
     # type: () -> Optional[int]
     """Provide an estimate for the available system memory in Bytes.
 
     If no estimate can be provided on the platform, returns None.
     """
     if pycompat.sysplatform.startswith(b'win'):
         # On Windows, use the GlobalMemoryStatusEx kernel function directly.
         from ctypes import c_long as DWORD, c_ulonglong as DWORDLONG
         from ctypes.wintypes import (  # pytype: disable=import-error
             Structure,
             byref,
             sizeof,
             windll,
         )
 
         class MEMORYSTATUSEX(Structure):
             _fields_ = [
                 ('dwLength', DWORD),
                 ('dwMemoryLoad', DWORD),
                 ('ullTotalPhys', DWORDLONG),
                 ('ullAvailPhys', DWORDLONG),
                 ('ullTotalPageFile', DWORDLONG),
                 ('ullAvailPageFile', DWORDLONG),
                 ('ullTotalVirtual', DWORDLONG),
                 ('ullAvailVirtual', DWORDLONG),
                 ('ullExtendedVirtual', DWORDLONG),
             ]
 
         x = MEMORYSTATUSEX()
         x.dwLength = sizeof(x)
         windll.kernel32.GlobalMemoryStatusEx(byref(x))
         return x.ullAvailPhys
 
     # On newer Unix-like systems and Mac OSX, the sysconf interface
     # can be used. _SC_PAGE_SIZE is part of POSIX; _SC_PHYS_PAGES
     # seems to be implemented on most systems.
     try:
         pagesize = os.sysconf(os.sysconf_names['SC_PAGE_SIZE'])
         pages = os.sysconf(os.sysconf_names['SC_PHYS_PAGES'])
         return pagesize * pages
     except OSError:  # sysconf can fail
         pass
     except KeyError:  # unknown parameter
         pass

tests/test-basic.t

@@ -1,258 +1,259 b''
 Create a repository:
 
 #if no-extraextensions
   $ hg config
   chgserver.idletimeout=60
   devel.all-warnings=true
   devel.default-date=0 0
   extensions.fsmonitor= (fsmonitor !)
   format.use-dirstate-v2=1 (dirstate-v2 !)
   largefiles.usercache=$TESTTMP/.cache/largefiles
   lfs.usercache=$TESTTMP/.cache/lfs
   ui.slash=True
   ui.interactive=False
   ui.detailed-exit-code=True
   ui.merge=internal:merge
   ui.mergemarkers=detailed
   ui.promptecho=True
   ui.ssh=* (glob)
   ui.timeout.warn=15
   web.address=localhost
   web\.ipv6=(?:True|False) (re)
   web.server-header=testing stub value
 #endif
 
   $ hg init t
   $ cd t
 
 Prepare a changeset:
 
   $ echo a > a
   $ hg add a
 
   $ hg status
   A a
 
 Writes to stdio succeed and fail appropriately
 
 #if devfull
   $ hg status 2>/dev/full
   A a
 
   $ hg status >/dev/full
   abort: No space left on device* (glob)
   [255]
 #endif
 
 #if devfull
   $ hg status >/dev/full 2>&1
   [255]
 
   $ hg status ENOENT 2>/dev/full
   [255]
 #endif
 
 On Python 3, stdio may be None:
 
   $ hg debuguiprompt --config ui.interactive=true 0<&-
    abort: Bad file descriptor (no-rhg !)
    abort: response expected (rhg !)
   [255]
   $ hg version -q 0<&-
   Mercurial Distributed SCM * (glob)
 
 #if py3 no-rhg
   $ hg version -q 1>&-
   abort: Bad file descriptor
   [255]
 #else
   $ hg version -q 1>&-
 #endif
   $ hg unknown -q 1>&-
   hg: unknown command 'unknown'
   (did you mean debugknown?)
   [10]
 
   $ hg version -q 2>&-
   Mercurial Distributed SCM * (glob)
   $ hg unknown -q 2>&-
   [10]
 
   $ hg commit -m test
 
 This command is ancient:
 
   $ hg history
   changeset:   0:acb14030fe0a
   tag:         tip
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     test
   
 
 Verify that updating to revision 0 via commands.update() works properly
 
   $ cat <<EOF > update_to_rev0.py
   > from mercurial import commands, hg, ui as uimod
   > myui = uimod.ui.load()
   > repo = hg.repository(myui, path=b'.')
   > commands.update(myui, repo, rev=b"0")
   > EOF
   $ hg up null
   0 files updated, 0 files merged, 1 files removed, 0 files unresolved
   $ "$PYTHON" ./update_to_rev0.py
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg identify -n
   0
 
 
 Poke around at hashes:
 
   $ hg manifest --debug
   b789fdd96dc2f3bd229c1dd8eedf0fc60e2b68e3 644   a
 
   $ hg cat a
   a
 
 Verify should succeed:
 
   $ hg verify
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   checked 1 changesets with 1 changes to 1 files
 
 Repository root:
 
   $ hg root
   $TESTTMP/t
   $ hg log -l1 -T '{reporoot}\n'
   $TESTTMP/t
   $ hg root -Tjson | sed 's|\\\\|\\|g'
   [
    {
     "hgpath": "$TESTTMP/t/.hg",
     "reporoot": "$TESTTMP/t",
     "storepath": "$TESTTMP/t/.hg/store"
    }
   ]
 
 At the end...
 
   $ cd ..
 
 Status message redirection:
 
   $ hg init empty
 
  status messages are sent to stdout by default:
 
   $ hg outgoing -R t empty -Tjson 2>/dev/null
   comparing with empty
   searching for changes
   [
    {
     "bookmarks": [],
     "branch": "default",
     "date": [0, 0],
     "desc": "test",
     "node": "acb14030fe0a21b60322c440ad2d20cf7685a376",
     "parents": ["0000000000000000000000000000000000000000"],
     "phase": "draft",
     "rev": 0,
     "tags": ["tip"],
     "user": "test"
    }
   ]
 
  which can be configured to send to stderr, so the output wouldn't be
  interleaved:
 
   $ cat <<'EOF' >> "$HGRCPATH"
   > [ui]
   > message-output = stderr
   > EOF
   $ hg outgoing -R t empty -Tjson 2>/dev/null
   [
    {
     "bookmarks": [],
     "branch": "default",
     "date": [0, 0],
     "desc": "test",
     "node": "acb14030fe0a21b60322c440ad2d20cf7685a376",
     "parents": ["0000000000000000000000000000000000000000"],
     "phase": "draft",
     "rev": 0,
     "tags": ["tip"],
     "user": "test"
    }
   ]
   $ hg outgoing -R t empty -Tjson >/dev/null
   comparing with empty
   searching for changes
 
  this option should be turned off by HGPLAIN= since it may break scripting use:
 
   $ HGPLAIN= hg outgoing -R t empty -Tjson 2>/dev/null
   comparing with empty
   searching for changes
   [
    {
     "bookmarks": [],
     "branch": "default",
     "date": [0, 0],
     "desc": "test",
     "node": "acb14030fe0a21b60322c440ad2d20cf7685a376",
     "parents": ["0000000000000000000000000000000000000000"],
     "phase": "draft",
     "rev": 0,
     "tags": ["tip"],
     "user": "test"
    }
   ]
 
  but still overridden by --config:
 
   $ HGPLAIN= hg outgoing -R t empty -Tjson --config ui.message-output=stderr \
   > 2>/dev/null
   [
    {
     "bookmarks": [],
     "branch": "default",
     "date": [0, 0],
     "desc": "test",
     "node": "acb14030fe0a21b60322c440ad2d20cf7685a376",
     "parents": ["0000000000000000000000000000000000000000"],
     "phase": "draft",
     "rev": 0,
     "tags": ["tip"],
     "user": "test"
    }
   ]
 
 Invalid ui.message-output option:
 
   $ hg log -R t --config ui.message-output=bad
   abort: invalid ui.message-output destination: bad
   [255]
 
 Underlying message streams should be updated when ui.fout/ferr are set:
 
   $ cat <<'EOF' > capui.py
-  > from mercurial import pycompat, registrar
+  > import io
+  > from mercurial import registrar
   > cmdtable = {}
   > command = registrar.command(cmdtable)
   > @command(b'capui', norepo=True)
   > def capui(ui):
   >     out = ui.fout
-  >     ui.fout = pycompat.bytesio()
+  >     ui.fout = io.BytesIO()
   >     ui.status(b'status\n')
-  >     ui.ferr = pycompat.bytesio()
+  >     ui.ferr = io.BytesIO()
   >     ui.warn(b'warn\n')
   >     out.write(b'stdout: %s' % ui.fout.getvalue())
   >     out.write(b'stderr: %s' % ui.ferr.getvalue())
   > EOF
   $ hg --config extensions.capui=capui.py --config ui.message-output=stdio capui
   stdout: status
   stderr: warn

tests/test-util.py

@@ -1,143 +1,144 b''
 # unit tests for mercuril.util utilities
 from __future__ import absolute_import
 
 import contextlib
+import io
 import itertools
 import unittest
 
 from mercurial import pycompat, util, utils
 
 
 @contextlib.contextmanager
 def mocktimer(incr=0.1, *additional_targets):
     """Replaces util.timer and additional_targets with a mock
 
     The timer starts at 0. On each call the time incremented by the value
     of incr. If incr is an iterable, then the time is incremented by the
     next value from that iterable, looping in a cycle when reaching the end.
 
     additional_targets must be a sequence of (object, attribute_name) tuples;
     the mock is set with setattr(object, attribute_name, mock).
 
     """
     time = [0]
     try:
         incr = itertools.cycle(incr)
     except TypeError:
         incr = itertools.repeat(incr)
 
     def timer():
         time[0] += next(incr)
         return time[0]
 
     # record original values
     orig = util.timer
     additional_origs = [(o, a, getattr(o, a)) for o, a in additional_targets]
 
     # mock out targets
     util.timer = timer
     for obj, attr in additional_targets:
         setattr(obj, attr, timer)
 
     try:
         yield
     finally:
         # restore originals
         util.timer = orig
         for args in additional_origs:
             setattr(*args)
 
 
 # attr.s default factory for util.timedstats.start binds the timer we
 # need to mock out.
 _start_default = (util.timedcmstats.start.default, 'factory')
 
 
 @contextlib.contextmanager
 def capturestderr():
-    """Replace utils.procutil.stderr with a pycompat.bytesio instance
+    """Replace utils.procutil.stderr with an io.BytesIO instance
 
     The instance is made available as the return value of __enter__.
 
     This contextmanager is reentrant.
 
     """
     orig = utils.procutil.stderr
-    utils.procutil.stderr = pycompat.bytesio()
+    utils.procutil.stderr = io.BytesIO()
     try:
         yield utils.procutil.stderr
     finally:
         utils.procutil.stderr = orig
 
 
 class timedtests(unittest.TestCase):
     def testtimedcmstatsstr(self):
         stats = util.timedcmstats()
         self.assertEqual(str(stats), '<unknown>')
         self.assertEqual(bytes(stats), b'<unknown>')
         stats.elapsed = 12.34
         self.assertEqual(str(stats), pycompat.sysstr(util.timecount(12.34)))
         self.assertEqual(bytes(stats), util.timecount(12.34))
 
     def testtimedcmcleanexit(self):
         # timestamps 1, 4, elapsed time of 4 - 1 = 3
         with mocktimer([1, 3], _start_default):
             with util.timedcm('pass') as stats:
                 # actual context doesn't matter
                 pass
 
         self.assertEqual(stats.start, 1)
         self.assertEqual(stats.elapsed, 3)
         self.assertEqual(stats.level, 1)
 
     def testtimedcmnested(self):
         # timestamps 1, 3, 6, 10, elapsed times of 6 - 3 = 3 and 10 - 1 = 9
         with mocktimer([1, 2, 3, 4], _start_default):
             with util.timedcm('outer') as outer_stats:
                 with util.timedcm('inner') as inner_stats:
                     # actual context doesn't matter
                     pass
 
         self.assertEqual(outer_stats.start, 1)
         self.assertEqual(outer_stats.elapsed, 9)
         self.assertEqual(outer_stats.level, 1)
 
         self.assertEqual(inner_stats.start, 3)
         self.assertEqual(inner_stats.elapsed, 3)
         self.assertEqual(inner_stats.level, 2)
 
     def testtimedcmexception(self):
         # timestamps 1, 4, elapsed time of 4 - 1 = 3
         with mocktimer([1, 3], _start_default):
             try:
                 with util.timedcm('exceptional') as stats:
                     raise ValueError()
             except ValueError:
                 pass
 
         self.assertEqual(stats.start, 1)
         self.assertEqual(stats.elapsed, 3)
         self.assertEqual(stats.level, 1)
 
     def testtimeddecorator(self):
         @util.timed
         def testfunc(callcount=1):
             callcount -= 1
             if callcount:
                 testfunc(callcount)
 
         # timestamps 1, 2, 3, 4, elapsed time of 3 - 2 = 1 and 4 - 1 = 3
         with mocktimer(1, _start_default):
             with capturestderr() as out:
                 testfunc(2)
 
         self.assertEqual(
             out.getvalue(),
             (b'    testfunc: 1.000 s\n' b'  testfunc: 3.000 s\n'),
         )
 
 
 if __name__ == '__main__':
     import silenttestrunner
 
     silenttestrunner.main(__name__)