py3: convert the mode argument of os.fdopen to unicodes (2 of 2)

hgext/convert/subversion.py

@@ -1,1353 +1,1353 b''
 # Subversion 1.4/1.5 Python API backend
 #
 # Copyright(C) 2007 Daniel Holth et al
 from __future__ import absolute_import
 
 import os
 import re
 import tempfile
 import xml.dom.minidom
 
 from mercurial.i18n import _
 from mercurial import (
     encoding,
     error,
     pycompat,
     scmutil,
     util,
 )
 
 from . import common
 
 pickle = util.pickle
 stringio = util.stringio
 propertycache = util.propertycache
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 commandline = common.commandline
 commit = common.commit
 converter_sink = common.converter_sink
 converter_source = common.converter_source
 decodeargs = common.decodeargs
 encodeargs = common.encodeargs
 makedatetimestamp = common.makedatetimestamp
 mapfile = common.mapfile
 MissingTool = common.MissingTool
 NoRepo = common.NoRepo
 
 # Subversion stuff. Works best with very recent Python SVN bindings
 # e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing
 # these bindings.
 
 try:
     import svn
     import svn.client
     import svn.core
     import svn.ra
     import svn.delta
     from . import transport
     import warnings
     warnings.filterwarnings('ignore',
             module='svn.core',
             category=DeprecationWarning)
     svn.core.SubversionException # trigger import to catch error
 
 except ImportError:
     svn = None
 
 class SvnPathNotFound(Exception):
     pass
 
 def revsplit(rev):
     """Parse a revision string and return (uuid, path, revnum).
     >>> revsplit('svn:a2147622-4a9f-4db4-a8d3-13562ff547b2'
     ...          '/proj%20B/mytrunk/mytrunk@1')
     ('a2147622-4a9f-4db4-a8d3-13562ff547b2', '/proj%20B/mytrunk/mytrunk', 1)
     >>> revsplit('svn:8af66a51-67f5-4354-b62c-98d67cc7be1d@1')
     ('', '', 1)
     >>> revsplit('@7')
     ('', '', 7)
     >>> revsplit('7')
     ('', '', 0)
     >>> revsplit('bad')
     ('', '', 0)
     """
     parts = rev.rsplit('@', 1)
     revnum = 0
     if len(parts) > 1:
         revnum = int(parts[1])
     parts = parts[0].split('/', 1)
     uuid = ''
     mod = ''
     if len(parts) > 1 and parts[0].startswith('svn:'):
         uuid = parts[0][4:]
         mod = '/' + parts[1]
     return uuid, mod, revnum
 
 def quote(s):
     # As of svn 1.7, many svn calls expect "canonical" paths. In
     # theory, we should call svn.core.*canonicalize() on all paths
     # before passing them to the API.  Instead, we assume the base url
     # is canonical and copy the behaviour of svn URL encoding function
     # so we can extend it safely with new components. The "safe"
     # characters were taken from the "svn_uri__char_validity" table in
     # libsvn_subr/path.c.
     return urlreq.quote(s, "!$&'()*+,-./:=@_~")
 
 def geturl(path):
     try:
         return svn.client.url_from_path(svn.core.svn_path_canonicalize(path))
     except svn.core.SubversionException:
         # svn.client.url_from_path() fails with local repositories
         pass
     if os.path.isdir(path):
         path = os.path.normpath(os.path.abspath(path))
         if pycompat.osname == 'nt':
             path = '/' + util.normpath(path)
         # Module URL is later compared with the repository URL returned
         # by svn API, which is UTF-8.
         path = encoding.tolocal(path)
         path = 'file://%s' % quote(path)
     return svn.core.svn_path_canonicalize(path)
 
 def optrev(number):
     optrev = svn.core.svn_opt_revision_t()
     optrev.kind = svn.core.svn_opt_revision_number
     optrev.value.number = number
     return optrev
 
 class changedpath(object):
     def __init__(self, p):
         self.copyfrom_path = p.copyfrom_path
         self.copyfrom_rev = p.copyfrom_rev
         self.action = p.action
 
 def get_log_child(fp, url, paths, start, end, limit=0,
                   discover_changed_paths=True, strict_node_history=False):
     protocol = -1
     def receiver(orig_paths, revnum, author, date, message, pool):
         paths = {}
         if orig_paths is not None:
             for k, v in orig_paths.iteritems():
                 paths[k] = changedpath(v)
         pickle.dump((paths, revnum, author, date, message),
                     fp, protocol)
 
     try:
         # Use an ra of our own so that our parent can consume
         # our results without confusing the server.
         t = transport.SvnRaTransport(url=url)
         svn.ra.get_log(t.ra, paths, start, end, limit,
                        discover_changed_paths,
                        strict_node_history,
                        receiver)
     except IOError:
         # Caller may interrupt the iteration
         pickle.dump(None, fp, protocol)
     except Exception as inst:
         pickle.dump(str(inst), fp, protocol)
     else:
         pickle.dump(None, fp, protocol)
     fp.close()
     # With large history, cleanup process goes crazy and suddenly
     # consumes *huge* amount of memory. The output file being closed,
     # there is no need for clean termination.
     os._exit(0)
 
 def debugsvnlog(ui, **opts):
     """Fetch SVN log in a subprocess and channel them back to parent to
     avoid memory collection issues.
     """
     if svn is None:
         raise error.Abort(_('debugsvnlog could not load Subversion python '
                            'bindings'))
 
     args = decodeargs(ui.fin.read())
     get_log_child(ui.fout, *args)
 
 class logstream(object):
     """Interruptible revision log iterator."""
     def __init__(self, stdout):
         self._stdout = stdout
 
     def __iter__(self):
         while True:
             try:
                 entry = pickle.load(self._stdout)
             except EOFError:
                 raise error.Abort(_('Mercurial failed to run itself, check'
                                    ' hg executable is in PATH'))
             try:
                 orig_paths, revnum, author, date, message = entry
             except (TypeError, ValueError):
                 if entry is None:
                     break
                 raise error.Abort(_("log stream exception '%s'") % entry)
             yield entry
 
     def close(self):
         if self._stdout:
             self._stdout.close()
             self._stdout = None
 
 class directlogstream(list):
     """Direct revision log iterator.
     This can be used for debugging and development but it will probably leak
     memory and is not suitable for real conversions."""
     def __init__(self, url, paths, start, end, limit=0,
                   discover_changed_paths=True, strict_node_history=False):
 
         def receiver(orig_paths, revnum, author, date, message, pool):
             paths = {}
             if orig_paths is not None:
                 for k, v in orig_paths.iteritems():
                     paths[k] = changedpath(v)
             self.append((paths, revnum, author, date, message))
 
         # Use an ra of our own so that our parent can consume
         # our results without confusing the server.
         t = transport.SvnRaTransport(url=url)
         svn.ra.get_log(t.ra, paths, start, end, limit,
                        discover_changed_paths,
                        strict_node_history,
                        receiver)
 
     def close(self):
         pass
 
 # Check to see if the given path is a local Subversion repo. Verify this by
 # looking for several svn-specific files and directories in the given
 # directory.
 def filecheck(ui, path, proto):
     for x in ('locks', 'hooks', 'format', 'db'):
         if not os.path.exists(os.path.join(path, x)):
             return False
     return True
 
 # Check to see if a given path is the root of an svn repo over http. We verify
 # this by requesting a version-controlled URL we know can't exist and looking
 # for the svn-specific "not found" XML.
 def httpcheck(ui, path, proto):
     try:
         opener = urlreq.buildopener()
         rsp = opener.open('%s://%s/!svn/ver/0/.svn' % (proto, path))
         data = rsp.read()
     except urlerr.httperror as inst:
         if inst.code != 404:
             # Except for 404 we cannot know for sure this is not an svn repo
             ui.warn(_('svn: cannot probe remote repository, assume it could '
                       'be a subversion repository. Use --source-type if you '
                       'know better.\n'))
             return True
         data = inst.fp.read()
     except Exception:
         # Could be urlerr.urlerror if the URL is invalid or anything else.
         return False
     return '<m:human-readable errcode="160013">' in data
 
 protomap = {'http': httpcheck,
             'https': httpcheck,
             'file': filecheck,
             }
 def issvnurl(ui, url):
     try:
         proto, path = url.split('://', 1)
         if proto == 'file':
             if (pycompat.osname == 'nt' and path[:1] == '/'
                   and path[1:2].isalpha() and path[2:6].lower() == '%3a/'):
                 path = path[:2] + ':/' + path[6:]
             path = urlreq.url2pathname(path)
     except ValueError:
         proto = 'file'
         path = os.path.abspath(url)
     if proto == 'file':
         path = util.pconvert(path)
     check = protomap.get(proto, lambda *args: False)
     while '/' in path:
         if check(ui, path, proto):
             return True
         path = path.rsplit('/', 1)[0]
     return False
 
 # SVN conversion code stolen from bzr-svn and tailor
 #
 # Subversion looks like a versioned filesystem, branches structures
 # are defined by conventions and not enforced by the tool. First,
 # we define the potential branches (modules) as "trunk" and "branches"
 # children directories. Revisions are then identified by their
 # module and revision number (and a repository identifier).
 #
 # The revision graph is really a tree (or a forest). By default, a
 # revision parent is the previous revision in the same module. If the
 # module directory is copied/moved from another module then the
 # revision is the module root and its parent the source revision in
 # the parent module. A revision has at most one parent.
 #
 class svn_source(converter_source):
     def __init__(self, ui, url, revs=None):
         super(svn_source, self).__init__(ui, url, revs=revs)
 
         if not (url.startswith('svn://') or url.startswith('svn+ssh://') or
                 (os.path.exists(url) and
                  os.path.exists(os.path.join(url, '.svn'))) or
                 issvnurl(ui, url)):
             raise NoRepo(_("%s does not look like a Subversion repository")
                          % url)
         if svn is None:
             raise MissingTool(_('could not load Subversion python bindings'))
 
         try:
             version = svn.core.SVN_VER_MAJOR, svn.core.SVN_VER_MINOR
             if version < (1, 4):
                 raise MissingTool(_('Subversion python bindings %d.%d found, '
                                     '1.4 or later required') % version)
         except AttributeError:
             raise MissingTool(_('Subversion python bindings are too old, 1.4 '
                                 'or later required'))
 
         self.lastrevs = {}
 
         latest = None
         try:
             # Support file://path@rev syntax. Useful e.g. to convert
             # deleted branches.
             at = url.rfind('@')
             if at >= 0:
                 latest = int(url[at + 1:])
                 url = url[:at]
         except ValueError:
             pass
         self.url = geturl(url)
         self.encoding = 'UTF-8' # Subversion is always nominal UTF-8
         try:
             self.transport = transport.SvnRaTransport(url=self.url)
             self.ra = self.transport.ra
             self.ctx = self.transport.client
             self.baseurl = svn.ra.get_repos_root(self.ra)
             # Module is either empty or a repository path starting with
             # a slash and not ending with a slash.
             self.module = urlreq.unquote(self.url[len(self.baseurl):])
             self.prevmodule = None
             self.rootmodule = self.module
             self.commits = {}
             self.paths = {}
             self.uuid = svn.ra.get_uuid(self.ra)
         except svn.core.SubversionException:
             ui.traceback()
             svnversion = '%d.%d.%d' % (svn.core.SVN_VER_MAJOR,
                                        svn.core.SVN_VER_MINOR,
                                        svn.core.SVN_VER_MICRO)
             raise NoRepo(_("%s does not look like a Subversion repository "
                            "to libsvn version %s")
                          % (self.url, svnversion))
 
         if revs:
             if len(revs) > 1:
                 raise error.Abort(_('subversion source does not support '
                                    'specifying multiple revisions'))
             try:
                 latest = int(revs[0])
             except ValueError:
                 raise error.Abort(_('svn: revision %s is not an integer') %
                                  revs[0])
 
         self.trunkname = self.ui.config('convert', 'svn.trunk',
                                         'trunk').strip('/')
         self.startrev = self.ui.config('convert', 'svn.startrev', default=0)
         try:
             self.startrev = int(self.startrev)
             if self.startrev < 0:
                 self.startrev = 0
         except ValueError:
             raise error.Abort(_('svn: start revision %s is not an integer')
                              % self.startrev)
 
         try:
             self.head = self.latest(self.module, latest)
         except SvnPathNotFound:
             self.head = None
         if not self.head:
             raise error.Abort(_('no revision found in module %s')
                              % self.module)
         self.last_changed = self.revnum(self.head)
 
         self._changescache = (None, None)
 
         if os.path.exists(os.path.join(url, '.svn/entries')):
             self.wc = url
         else:
             self.wc = None
         self.convertfp = None
 
     def setrevmap(self, revmap):
         lastrevs = {}
         for revid in revmap.iterkeys():
             uuid, module, revnum = revsplit(revid)
             lastrevnum = lastrevs.setdefault(module, revnum)
             if revnum > lastrevnum:
                 lastrevs[module] = revnum
         self.lastrevs = lastrevs
 
     def exists(self, path, optrev):
         try:
             svn.client.ls(self.url.rstrip('/') + '/' + quote(path),
                                  optrev, False, self.ctx)
             return True
         except svn.core.SubversionException:
             return False
 
     def getheads(self):
 
         def isdir(path, revnum):
             kind = self._checkpath(path, revnum)
             return kind == svn.core.svn_node_dir
 
         def getcfgpath(name, rev):
             cfgpath = self.ui.config('convert', 'svn.' + name)
             if cfgpath is not None and cfgpath.strip() == '':
                 return None
             path = (cfgpath or name).strip('/')
             if not self.exists(path, rev):
                 if self.module.endswith(path) and name == 'trunk':
                     # we are converting from inside this directory
                     return None
                 if cfgpath:
                     raise error.Abort(_('expected %s to be at %r, but not found'
                                        ) % (name, path))
                 return None
             self.ui.note(_('found %s at %r\n') % (name, path))
             return path
 
         rev = optrev(self.last_changed)
         oldmodule = ''
         trunk = getcfgpath('trunk', rev)
         self.tags = getcfgpath('tags', rev)
         branches = getcfgpath('branches', rev)
 
         # If the project has a trunk or branches, we will extract heads
         # from them. We keep the project root otherwise.
         if trunk:
             oldmodule = self.module or ''
             self.module += '/' + trunk
             self.head = self.latest(self.module, self.last_changed)
             if not self.head:
                 raise error.Abort(_('no revision found in module %s')
                                  % self.module)
 
         # First head in the list is the module's head
         self.heads = [self.head]
         if self.tags is not None:
             self.tags = '%s/%s' % (oldmodule , (self.tags or 'tags'))
 
         # Check if branches bring a few more heads to the list
         if branches:
             rpath = self.url.strip('/')
             branchnames = svn.client.ls(rpath + '/' + quote(branches),
                                         rev, False, self.ctx)
             for branch in sorted(branchnames):
                 module = '%s/%s/%s' % (oldmodule, branches, branch)
                 if not isdir(module, self.last_changed):
                     continue
                 brevid = self.latest(module, self.last_changed)
                 if not brevid:
                     self.ui.note(_('ignoring empty branch %s\n') % branch)
                     continue
                 self.ui.note(_('found branch %s at %d\n') %
                              (branch, self.revnum(brevid)))
                 self.heads.append(brevid)
 
         if self.startrev and self.heads:
             if len(self.heads) > 1:
                 raise error.Abort(_('svn: start revision is not supported '
                                    'with more than one branch'))
             revnum = self.revnum(self.heads[0])
             if revnum < self.startrev:
                 raise error.Abort(
                     _('svn: no revision found after start revision %d')
                                  % self.startrev)
 
         return self.heads
 
     def _getchanges(self, rev, full):
         (paths, parents) = self.paths[rev]
         copies = {}
         if parents:
             files, self.removed, copies = self.expandpaths(rev, paths, parents)
         if full or not parents:
             # Perform a full checkout on roots
             uuid, module, revnum = revsplit(rev)
             entries = svn.client.ls(self.baseurl + quote(module),
                                     optrev(revnum), True, self.ctx)
             files = [n for n, e in entries.iteritems()
                      if e.kind == svn.core.svn_node_file]
             self.removed = set()
 
         files.sort()
         files = zip(files, [rev] * len(files))
         return (files, copies)
 
     def getchanges(self, rev, full):
         # reuse cache from getchangedfiles
         if self._changescache[0] == rev and not full:
             (files, copies) = self._changescache[1]
         else:
             (files, copies) = self._getchanges(rev, full)
             # caller caches the result, so free it here to release memory
             del self.paths[rev]
         return (files, copies, set())
 
     def getchangedfiles(self, rev, i):
         # called from filemap - cache computed values for reuse in getchanges
         (files, copies) = self._getchanges(rev, False)
         self._changescache = (rev, (files, copies))
         return [f[0] for f in files]
 
     def getcommit(self, rev):
         if rev not in self.commits:
             uuid, module, revnum = revsplit(rev)
             self.module = module
             self.reparent(module)
             # We assume that:
             # - requests for revisions after "stop" come from the
             # revision graph backward traversal. Cache all of them
             # down to stop, they will be used eventually.
             # - requests for revisions before "stop" come to get
             # isolated branches parents. Just fetch what is needed.
             stop = self.lastrevs.get(module, 0)
             if revnum < stop:
                 stop = revnum + 1
             self._fetch_revisions(revnum, stop)
             if rev not in self.commits:
                 raise error.Abort(_('svn: revision %s not found') % revnum)
         revcommit = self.commits[rev]
         # caller caches the result, so free it here to release memory
         del self.commits[rev]
         return revcommit
 
     def checkrevformat(self, revstr, mapname='splicemap'):
         """ fails if revision format does not match the correct format"""
         if not re.match(r'svn:[0-9a-f]{8,8}-[0-9a-f]{4,4}-'
                               r'[0-9a-f]{4,4}-[0-9a-f]{4,4}-[0-9a-f]'
                               r'{12,12}(.*)\@[0-9]+$',revstr):
             raise error.Abort(_('%s entry %s is not a valid revision'
                                ' identifier') % (mapname, revstr))
 
     def numcommits(self):
         return int(self.head.rsplit('@', 1)[1]) - self.startrev
 
     def gettags(self):
         tags = {}
         if self.tags is None:
             return tags
 
         # svn tags are just a convention, project branches left in a
         # 'tags' directory. There is no other relationship than
         # ancestry, which is expensive to discover and makes them hard
         # to update incrementally.  Worse, past revisions may be
         # referenced by tags far away in the future, requiring a deep
         # history traversal on every calculation.  Current code
         # performs a single backward traversal, tracking moves within
         # the tags directory (tag renaming) and recording a new tag
         # everytime a project is copied from outside the tags
         # directory. It also lists deleted tags, this behaviour may
         # change in the future.
         pendings = []
         tagspath = self.tags
         start = svn.ra.get_latest_revnum(self.ra)
         stream = self._getlog([self.tags], start, self.startrev)
         try:
             for entry in stream:
                 origpaths, revnum, author, date, message = entry
                 if not origpaths:
                     origpaths = []
                 copies = [(e.copyfrom_path, e.copyfrom_rev, p) for p, e
                           in origpaths.iteritems() if e.copyfrom_path]
                 # Apply moves/copies from more specific to general
                 copies.sort(reverse=True)
 
                 srctagspath = tagspath
                 if copies and copies[-1][2] == tagspath:
                     # Track tags directory moves
                     srctagspath = copies.pop()[0]
 
                 for source, sourcerev, dest in copies:
                     if not dest.startswith(tagspath + '/'):
                         continue
                     for tag in pendings:
                         if tag[0].startswith(dest):
                             tagpath = source + tag[0][len(dest):]
                             tag[:2] = [tagpath, sourcerev]
                             break
                     else:
                         pendings.append([source, sourcerev, dest])
 
                 # Filter out tags with children coming from different
                 # parts of the repository like:
                 # /tags/tag.1 (from /trunk:10)
                 # /tags/tag.1/foo (from /branches/foo:12)
                 # Here/tags/tag.1 discarded as well as its children.
                 # It happens with tools like cvs2svn. Such tags cannot
                 # be represented in mercurial.
                 addeds = dict((p, e.copyfrom_path) for p, e
                               in origpaths.iteritems()
                               if e.action == 'A' and e.copyfrom_path)
                 badroots = set()
                 for destroot in addeds:
                     for source, sourcerev, dest in pendings:
                         if (not dest.startswith(destroot + '/')
                             or source.startswith(addeds[destroot] + '/')):
                             continue
                         badroots.add(destroot)
                         break
 
                 for badroot in badroots:
                     pendings = [p for p in pendings if p[2] != badroot
                                 and not p[2].startswith(badroot + '/')]
 
                 # Tell tag renamings from tag creations
                 renamings = []
                 for source, sourcerev, dest in pendings:
                     tagname = dest.split('/')[-1]
                     if source.startswith(srctagspath):
                         renamings.append([source, sourcerev, tagname])
                         continue
                     if tagname in tags:
                         # Keep the latest tag value
                         continue
                     # From revision may be fake, get one with changes
                     try:
                         tagid = self.latest(source, sourcerev)
                         if tagid and tagname not in tags:
                             tags[tagname] = tagid
                     except SvnPathNotFound:
                         # It happens when we are following directories
                         # we assumed were copied with their parents
                         # but were really created in the tag
                         # directory.
                         pass
                 pendings = renamings
                 tagspath = srctagspath
         finally:
             stream.close()
         return tags
 
     def converted(self, rev, destrev):
         if not self.wc:
             return
         if self.convertfp is None:
             self.convertfp = open(os.path.join(self.wc, '.svn', 'hg-shamap'),
                                   'a')
         self.convertfp.write('%s %d\n' % (destrev, self.revnum(rev)))
         self.convertfp.flush()
 
     def revid(self, revnum, module=None):
         return 'svn:%s%s@%s' % (self.uuid, module or self.module, revnum)
 
     def revnum(self, rev):
         return int(rev.split('@')[-1])
 
     def latest(self, path, stop=None):
         """Find the latest revid affecting path, up to stop revision
         number. If stop is None, default to repository latest
         revision. It may return a revision in a different module,
         since a branch may be moved without a change being
         reported. Return None if computed module does not belong to
         rootmodule subtree.
         """
         def findchanges(path, start, stop=None):
             stream = self._getlog([path], start, stop or 1)
             try:
                 for entry in stream:
                     paths, revnum, author, date, message = entry
                     if stop is None and paths:
                         # We do not know the latest changed revision,
                         # keep the first one with changed paths.
                         break
                     if revnum <= stop:
                         break
 
                     for p in paths:
                         if (not path.startswith(p) or
                             not paths[p].copyfrom_path):
                             continue
                         newpath = paths[p].copyfrom_path + path[len(p):]
                         self.ui.debug("branch renamed from %s to %s at %d\n" %
                                       (path, newpath, revnum))
                         path = newpath
                         break
                 if not paths:
                     revnum = None
                 return revnum, path
             finally:
                 stream.close()
 
         if not path.startswith(self.rootmodule):
             # Requests on foreign branches may be forbidden at server level
             self.ui.debug('ignoring foreign branch %r\n' % path)
             return None
 
         if stop is None:
             stop = svn.ra.get_latest_revnum(self.ra)
         try:
             prevmodule = self.reparent('')
             dirent = svn.ra.stat(self.ra, path.strip('/'), stop)
             self.reparent(prevmodule)
         except svn.core.SubversionException:
             dirent = None
         if not dirent:
             raise SvnPathNotFound(_('%s not found up to revision %d')
                                   % (path, stop))
 
         # stat() gives us the previous revision on this line of
         # development, but it might be in *another module*. Fetch the
         # log and detect renames down to the latest revision.
         revnum, realpath = findchanges(path, stop, dirent.created_rev)
         if revnum is None:
             # Tools like svnsync can create empty revision, when
             # synchronizing only a subtree for instance. These empty
             # revisions created_rev still have their original values
             # despite all changes having disappeared and can be
             # returned by ra.stat(), at least when stating the root
             # module. In that case, do not trust created_rev and scan
             # the whole history.
             revnum, realpath = findchanges(path, stop)
             if revnum is None:
                 self.ui.debug('ignoring empty branch %r\n' % realpath)
                 return None
 
         if not realpath.startswith(self.rootmodule):
             self.ui.debug('ignoring foreign branch %r\n' % realpath)
             return None
         return self.revid(revnum, realpath)
 
     def reparent(self, module):
         """Reparent the svn transport and return the previous parent."""
         if self.prevmodule == module:
             return module
         svnurl = self.baseurl + quote(module)
         prevmodule = self.prevmodule
         if prevmodule is None:
             prevmodule = ''
         self.ui.debug("reparent to %s\n" % svnurl)
         svn.ra.reparent(self.ra, svnurl)
         self.prevmodule = module
         return prevmodule
 
     def expandpaths(self, rev, paths, parents):
         changed, removed = set(), set()
         copies = {}
 
         new_module, revnum = revsplit(rev)[1:]
         if new_module != self.module:
             self.module = new_module
             self.reparent(self.module)
 
         for i, (path, ent) in enumerate(paths):
             self.ui.progress(_('scanning paths'), i, item=path,
                              total=len(paths), unit=_('paths'))
             entrypath = self.getrelpath(path)
 
             kind = self._checkpath(entrypath, revnum)
             if kind == svn.core.svn_node_file:
                 changed.add(self.recode(entrypath))
                 if not ent.copyfrom_path or not parents:
                     continue
                 # Copy sources not in parent revisions cannot be
                 # represented, ignore their origin for now
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 if ent.copyfrom_rev < prevnum:
                     continue
                 copyfrom_path = self.getrelpath(ent.copyfrom_path, pmodule)
                 if not copyfrom_path:
                     continue
                 self.ui.debug("copied to %s from %s@%s\n" %
                               (entrypath, copyfrom_path, ent.copyfrom_rev))
                 copies[self.recode(entrypath)] = self.recode(copyfrom_path)
             elif kind == 0: # gone, but had better be a deleted *file*
                 self.ui.debug("gone from %s\n" % ent.copyfrom_rev)
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 parentpath = pmodule + "/" + entrypath
                 fromkind = self._checkpath(entrypath, prevnum, pmodule)
 
                 if fromkind == svn.core.svn_node_file:
                     removed.add(self.recode(entrypath))
                 elif fromkind == svn.core.svn_node_dir:
                     oroot = parentpath.strip('/')
                     nroot = path.strip('/')
                     children = self._iterfiles(oroot, prevnum)
                     for childpath in children:
                         childpath = childpath.replace(oroot, nroot)
                         childpath = self.getrelpath("/" + childpath, pmodule)
                         if childpath:
                             removed.add(self.recode(childpath))
                 else:
                     self.ui.debug('unknown path in revision %d: %s\n' % \
                                   (revnum, path))
             elif kind == svn.core.svn_node_dir:
                 if ent.action == 'M':
                     # If the directory just had a prop change,
                     # then we shouldn't need to look for its children.
                     continue
                 if ent.action == 'R' and parents:
                     # If a directory is replacing a file, mark the previous
                     # file as deleted
                     pmodule, prevnum = revsplit(parents[0])[1:]
                     pkind = self._checkpath(entrypath, prevnum, pmodule)
                     if pkind == svn.core.svn_node_file:
                         removed.add(self.recode(entrypath))
                     elif pkind == svn.core.svn_node_dir:
                         # We do not know what files were kept or removed,
                         # mark them all as changed.
                         for childpath in self._iterfiles(pmodule, prevnum):
                             childpath = self.getrelpath("/" + childpath)
                             if childpath:
                                 changed.add(self.recode(childpath))
 
                 for childpath in self._iterfiles(path, revnum):
                     childpath = self.getrelpath("/" + childpath)
                     if childpath:
                         changed.add(self.recode(childpath))
 
                 # Handle directory copies
                 if not ent.copyfrom_path or not parents:
                     continue
                 # Copy sources not in parent revisions cannot be
                 # represented, ignore their origin for now
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 if ent.copyfrom_rev < prevnum:
                     continue
                 copyfrompath = self.getrelpath(ent.copyfrom_path, pmodule)
                 if not copyfrompath:
                     continue
                 self.ui.debug("mark %s came from %s:%d\n"
                               % (path, copyfrompath, ent.copyfrom_rev))
                 children = self._iterfiles(ent.copyfrom_path, ent.copyfrom_rev)
                 for childpath in children:
                     childpath = self.getrelpath("/" + childpath, pmodule)
                     if not childpath:
                         continue
                     copytopath = path + childpath[len(copyfrompath):]
                     copytopath = self.getrelpath(copytopath)
                     copies[self.recode(copytopath)] = self.recode(childpath)
 
         self.ui.progress(_('scanning paths'), None)
         changed.update(removed)
         return (list(changed), removed, copies)
 
     def _fetch_revisions(self, from_revnum, to_revnum):
         if from_revnum < to_revnum:
             from_revnum, to_revnum = to_revnum, from_revnum
 
         self.child_cset = None
 
         def parselogentry(orig_paths, revnum, author, date, message):
             """Return the parsed commit object or None, and True if
             the revision is a branch root.
             """
             self.ui.debug("parsing revision %d (%d changes)\n" %
                           (revnum, len(orig_paths)))
 
             branched = False
             rev = self.revid(revnum)
             # branch log might return entries for a parent we already have
 
             if rev in self.commits or revnum < to_revnum:
                 return None, branched
 
             parents = []
             # check whether this revision is the start of a branch or part
             # of a branch renaming
             orig_paths = sorted(orig_paths.iteritems())
             root_paths = [(p, e) for p, e in orig_paths
                           if self.module.startswith(p)]
             if root_paths:
                 path, ent = root_paths[-1]
                 if ent.copyfrom_path:
                     branched = True
                     newpath = ent.copyfrom_path + self.module[len(path):]
                     # ent.copyfrom_rev may not be the actual last revision
                     previd = self.latest(newpath, ent.copyfrom_rev)
                     if previd is not None:
                         prevmodule, prevnum = revsplit(previd)[1:]
                         if prevnum >= self.startrev:
                             parents = [previd]
                             self.ui.note(
                                 _('found parent of branch %s at %d: %s\n') %
                                 (self.module, prevnum, prevmodule))
                 else:
                     self.ui.debug("no copyfrom path, don't know what to do.\n")
 
             paths = []
             # filter out unrelated paths
             for path, ent in orig_paths:
                 if self.getrelpath(path) is None:
                     continue
                 paths.append((path, ent))
 
             # Example SVN datetime. Includes microseconds.
             # ISO-8601 conformant
             # '2007-01-04T17:35:00.902377Z'
             date = util.parsedate(date[:19] + " UTC", ["%Y-%m-%dT%H:%M:%S"])
             if self.ui.configbool('convert', 'localtimezone'):
                 date = makedatetimestamp(date[0])
 
             if message:
                 log = self.recode(message)
             else:
                 log = ''
 
             if author:
                 author = self.recode(author)
             else:
                 author = ''
 
             try:
                 branch = self.module.split("/")[-1]
                 if branch == self.trunkname:
                     branch = None
             except IndexError:
                 branch = None
 
             cset = commit(author=author,
                           date=util.datestr(date, '%Y-%m-%d %H:%M:%S %1%2'),
                           desc=log,
                           parents=parents,
                           branch=branch,
                           rev=rev)
 
             self.commits[rev] = cset
             # The parents list is *shared* among self.paths and the
             # commit object. Both will be updated below.
             self.paths[rev] = (paths, cset.parents)
             if self.child_cset and not self.child_cset.parents:
                 self.child_cset.parents[:] = [rev]
             self.child_cset = cset
             return cset, branched
 
         self.ui.note(_('fetching revision log for "%s" from %d to %d\n') %
                      (self.module, from_revnum, to_revnum))
 
         try:
             firstcset = None
             lastonbranch = False
             stream = self._getlog([self.module], from_revnum, to_revnum)
             try:
                 for entry in stream:
                     paths, revnum, author, date, message = entry
                     if revnum < self.startrev:
                         lastonbranch = True
                         break
                     if not paths:
                         self.ui.debug('revision %d has no entries\n' % revnum)
                         # If we ever leave the loop on an empty
                         # revision, do not try to get a parent branch
                         lastonbranch = lastonbranch or revnum == 0
                         continue
                     cset, lastonbranch = parselogentry(paths, revnum, author,
                                                        date, message)
                     if cset:
                         firstcset = cset
                     if lastonbranch:
                         break
             finally:
                 stream.close()
 
             if not lastonbranch and firstcset and not firstcset.parents:
                 # The first revision of the sequence (the last fetched one)
                 # has invalid parents if not a branch root. Find the parent
                 # revision now, if any.
                 try:
                     firstrevnum = self.revnum(firstcset.rev)
                     if firstrevnum > 1:
                         latest = self.latest(self.module, firstrevnum - 1)
                         if latest:
                             firstcset.parents.append(latest)
                 except SvnPathNotFound:
                     pass
         except svn.core.SubversionException as xxx_todo_changeme:
             (inst, num) = xxx_todo_changeme.args
             if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION:
                 raise error.Abort(_('svn: branch has no revision %s')
                                  % to_revnum)
             raise
 
     def getfile(self, file, rev):
         # TODO: ra.get_file transmits the whole file instead of diffs.
         if file in self.removed:
             return None, None
         mode = ''
         try:
             new_module, revnum = revsplit(rev)[1:]
             if self.module != new_module:
                 self.module = new_module
                 self.reparent(self.module)
             io = stringio()
             info = svn.ra.get_file(self.ra, file, revnum, io)
             data = io.getvalue()
             # ra.get_file() seems to keep a reference on the input buffer
             # preventing collection. Release it explicitly.
             io.close()
             if isinstance(info, list):
                 info = info[-1]
             mode = ("svn:executable" in info) and 'x' or ''
             mode = ("svn:special" in info) and 'l' or mode
         except svn.core.SubversionException as e:
             notfound = (svn.core.SVN_ERR_FS_NOT_FOUND,
                 svn.core.SVN_ERR_RA_DAV_PATH_NOT_FOUND)
             if e.apr_err in notfound: # File not found
                 return None, None
             raise
         if mode == 'l':
             link_prefix = "link "
             if data.startswith(link_prefix):
                 data = data[len(link_prefix):]
         return data, mode
 
     def _iterfiles(self, path, revnum):
         """Enumerate all files in path at revnum, recursively."""
         path = path.strip('/')
         pool = svn.core.Pool()
         rpath = '/'.join([self.baseurl, quote(path)]).strip('/')
         entries = svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool)
         if path:
             path += '/'
         return ((path + p) for p, e in entries.iteritems()
                 if e.kind == svn.core.svn_node_file)
 
     def getrelpath(self, path, module=None):
         if module is None:
             module = self.module
         # Given the repository url of this wc, say
         #   "http://server/plone/CMFPlone/branches/Plone-2_0-branch"
         # extract the "entry" portion (a relative path) from what
         # svn log --xml says, i.e.
         #   "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py"
         # that is to say "tests/PloneTestCase.py"
         if path.startswith(module):
             relative = path.rstrip('/')[len(module):]
             if relative.startswith('/'):
                 return relative[1:]
             elif relative == '':
                 return relative
 
         # The path is outside our tracked tree...
         self.ui.debug('%r is not under %r, ignoring\n' % (path, module))
         return None
 
     def _checkpath(self, path, revnum, module=None):
         if module is not None:
             prevmodule = self.reparent('')
             path = module + '/' + path
         try:
             # ra.check_path does not like leading slashes very much, it leads
             # to PROPFIND subversion errors
             return svn.ra.check_path(self.ra, path.strip('/'), revnum)
         finally:
             if module is not None:
                 self.reparent(prevmodule)
 
     def _getlog(self, paths, start, end, limit=0, discover_changed_paths=True,
                 strict_node_history=False):
         # Normalize path names, svn >= 1.5 only wants paths relative to
         # supplied URL
         relpaths = []
         for p in paths:
             if not p.startswith('/'):
                 p = self.module + '/' + p
             relpaths.append(p.strip('/'))
         args = [self.baseurl, relpaths, start, end, limit,
                 discover_changed_paths, strict_node_history]
         # developer config: convert.svn.debugsvnlog
         if not self.ui.configbool('convert', 'svn.debugsvnlog', True):
             return directlogstream(*args)
         arg = encodeargs(args)
         hgexe = util.hgexecutable()
         cmd = '%s debugsvnlog' % util.shellquote(hgexe)
         stdin, stdout = util.popen2(util.quotecommand(cmd))
         stdin.write(arg)
         try:
             stdin.close()
         except IOError:
             raise error.Abort(_('Mercurial failed to run itself, check'
                                ' hg executable is in PATH'))
         return logstream(stdout)
 
 pre_revprop_change = '''#!/bin/sh
 
 REPOS="$1"
 REV="$2"
 USER="$3"
 PROPNAME="$4"
 ACTION="$5"
 
 if [ "$ACTION" = "M" -a "$PROPNAME" = "svn:log" ]; then exit 0; fi
 if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-branch" ]; then exit 0; fi
 if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-rev" ]; then exit 0; fi
 
 echo "Changing prohibited revision property" >&2
 exit 1
 '''
 
 class svn_sink(converter_sink, commandline):
     commit_re = re.compile(r'Committed revision (\d+).', re.M)
     uuid_re = re.compile(r'Repository UUID:\s*(\S+)', re.M)
 
     def prerun(self):
         if self.wc:
             os.chdir(self.wc)
 
     def postrun(self):
         if self.wc:
             os.chdir(self.cwd)
 
     def join(self, name):
         return os.path.join(self.wc, '.svn', name)
 
     def revmapfile(self):
         return self.join('hg-shamap')
 
     def authorfile(self):
         return self.join('hg-authormap')
 
     def __init__(self, ui, path):
 
         converter_sink.__init__(self, ui, path)
         commandline.__init__(self, ui, 'svn')
         self.delete = []
         self.setexec = []
         self.delexec = []
         self.copies = []
         self.wc = None
         self.cwd = pycompat.getcwd()
 
         created = False
         if os.path.isfile(os.path.join(path, '.svn', 'entries')):
             self.wc = os.path.realpath(path)
             self.run0('update')
         else:
             if not re.search(r'^(file|http|https|svn|svn\+ssh)\://', path):
                 path = os.path.realpath(path)
                 if os.path.isdir(os.path.dirname(path)):
                     if not os.path.exists(os.path.join(path, 'db', 'fs-type')):
                         ui.status(_('initializing svn repository %r\n') %
                                   os.path.basename(path))
                         commandline(ui, 'svnadmin').run0('create', path)
                         created = path
                     path = util.normpath(path)
                     if not path.startswith('/'):
                         path = '/' + path
                     path = 'file://' + path
 
             wcpath = os.path.join(pycompat.getcwd(), os.path.basename(path) +
                                 '-wc')
             ui.status(_('initializing svn working copy %r\n')
                       % os.path.basename(wcpath))
             self.run0('checkout', path, wcpath)
 
             self.wc = wcpath
         self.opener = scmutil.opener(self.wc)
         self.wopener = scmutil.opener(self.wc)
         self.childmap = mapfile(ui, self.join('hg-childmap'))
         if util.checkexec(self.wc):
             self.is_exec = util.isexec
         else:
             self.is_exec = None
 
         if created:
             hook = os.path.join(created, 'hooks', 'pre-revprop-change')
             fp = open(hook, 'w')
             fp.write(pre_revprop_change)
             fp.close()
             util.setflags(hook, False, True)
 
         output = self.run0('info')
         self.uuid = self.uuid_re.search(output).group(1).strip()
 
     def wjoin(self, *names):
         return os.path.join(self.wc, *names)
 
     @propertycache
     def manifest(self):
         # As of svn 1.7, the "add" command fails when receiving
         # already tracked entries, so we have to track and filter them
         # ourselves.
         m = set()
         output = self.run0('ls', recursive=True, xml=True)
         doc = xml.dom.minidom.parseString(output)
         for e in doc.getElementsByTagName('entry'):
             for n in e.childNodes:
                 if n.nodeType != n.ELEMENT_NODE or n.tagName != 'name':
                     continue
                 name = ''.join(c.data for c in n.childNodes
                                if c.nodeType == c.TEXT_NODE)
                 # Entries are compared with names coming from
                 # mercurial, so bytes with undefined encoding. Our
                 # best bet is to assume they are in local
                 # encoding. They will be passed to command line calls
                 # later anyway, so they better be.
                 m.add(encoding.tolocal(name.encode('utf-8')))
                 break
         return m
 
     def putfile(self, filename, flags, data):
         if 'l' in flags:
             self.wopener.symlink(data, filename)
         else:
             try:
                 if os.path.islink(self.wjoin(filename)):
                     os.unlink(filename)
             except OSError:
                 pass
             self.wopener.write(filename, data)
 
             if self.is_exec:
                 if self.is_exec(self.wjoin(filename)):
                     if 'x' not in flags:
                         self.delexec.append(filename)
                 else:
                     if 'x' in flags:
                         self.setexec.append(filename)
                 util.setflags(self.wjoin(filename), False, 'x' in flags)
 
     def _copyfile(self, source, dest):
         # SVN's copy command pukes if the destination file exists, but
         # our copyfile method expects to record a copy that has
         # already occurred.  Cross the semantic gap.
         wdest = self.wjoin(dest)
         exists = os.path.lexists(wdest)
         if exists:
             fd, tempname = tempfile.mkstemp(
                 prefix='hg-copy-', dir=os.path.dirname(wdest))
             os.close(fd)
             os.unlink(tempname)
             os.rename(wdest, tempname)
         try:
             self.run0('copy', source, dest)
         finally:
             self.manifest.add(dest)
             if exists:
                 try:
                     os.unlink(wdest)
                 except OSError:
                     pass
                 os.rename(tempname, wdest)
 
     def dirs_of(self, files):
         dirs = set()
         for f in files:
             if os.path.isdir(self.wjoin(f)):
                 dirs.add(f)
             i = len(f)
             for i in iter(lambda: f.rfind('/', 0, i), -1):
                 dirs.add(f[:i])
         return dirs
 
     def add_dirs(self, files):
         add_dirs = [d for d in sorted(self.dirs_of(files))
                     if d not in self.manifest]
         if add_dirs:
             self.manifest.update(add_dirs)
             self.xargs(add_dirs, 'add', non_recursive=True, quiet=True)
         return add_dirs
 
     def add_files(self, files):
         files = [f for f in files if f not in self.manifest]
         if files:
             self.manifest.update(files)
             self.xargs(files, 'add', quiet=True)
         return files
 
     def addchild(self, parent, child):
         self.childmap[parent] = child
 
     def revid(self, rev):
         return u"svn:%s@%s" % (self.uuid, rev)
 
     def putcommit(self, files, copies, parents, commit, source, revmap, full,
                   cleanp2):
         for parent in parents:
             try:
                 return self.revid(self.childmap[parent])
             except KeyError:
                 pass
 
         # Apply changes to working copy
         for f, v in files:
             data, mode = source.getfile(f, v)
             if data is None:
                 self.delete.append(f)
             else:
                 self.putfile(f, mode, data)
                 if f in copies:
                     self.copies.append([copies[f], f])
         if full:
             self.delete.extend(sorted(self.manifest.difference(files)))
         files = [f[0] for f in files]
 
         entries = set(self.delete)
         files = frozenset(files)
         entries.update(self.add_dirs(files.difference(entries)))
         if self.copies:
             for s, d in self.copies:
                 self._copyfile(s, d)
             self.copies = []
         if self.delete:
             self.xargs(self.delete, 'delete')
             for f in self.delete:
                 self.manifest.remove(f)
             self.delete = []
         entries.update(self.add_files(files.difference(entries)))
         if self.delexec:
             self.xargs(self.delexec, 'propdel', 'svn:executable')
             self.delexec = []
         if self.setexec:
             self.xargs(self.setexec, 'propset', 'svn:executable', '*')
             self.setexec = []
 
         fd, messagefile = tempfile.mkstemp(prefix='hg-convert-')
-        fp = os.fdopen(fd, 'w')
+        fp = os.fdopen(fd, pycompat.sysstr('w'))
         fp.write(commit.desc)
         fp.close()
         try:
             output = self.run0('commit',
                                username=util.shortuser(commit.author),
                                file=messagefile,
                                encoding='utf-8')
             try:
                 rev = self.commit_re.search(output).group(1)
             except AttributeError:
                 if parents and not files:
                     return parents[0]
                 self.ui.warn(_('unexpected svn output:\n'))
                 self.ui.warn(output)
                 raise error.Abort(_('unable to cope with svn output'))
             if commit.rev:
                 self.run('propset', 'hg:convert-rev', commit.rev,
                          revprop=True, revision=rev)
             if commit.branch and commit.branch != 'default':
                 self.run('propset', 'hg:convert-branch', commit.branch,
                          revprop=True, revision=rev)
             for parent in parents:
                 self.addchild(parent, rev)
             return self.revid(rev)
         finally:
             os.unlink(messagefile)
 
     def puttags(self, tags):
         self.ui.warn(_('writing Subversion tags is not yet implemented\n'))
         return None, None
 
     def hascommitfrommap(self, rev):
         # We trust that revisions referenced in a map still is present
         # TODO: implement something better if necessary and feasible
         return True
 
     def hascommitforsplicemap(self, rev):
         # This is not correct as one can convert to an existing subversion
         # repository and childmap would not list all revisions. Too bad.
         if rev in self.childmap:
             return True
         raise error.Abort(_('splice map revision %s not found in subversion '
                            'child map (revision lookups are not implemented)')
                          % rev)

hgext/gpg.py

@@ -1,318 +1,319 b''
 # Copyright 2005, 2006 Benoit Boissinot <[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.
 
 '''commands to sign and verify changesets'''
 
 from __future__ import absolute_import
 
 import binascii
 import os
 import tempfile
 
 from mercurial.i18n import _
 from mercurial import (
     cmdutil,
     commands,
     error,
     match,
     node as hgnode,
+    pycompat,
     util,
 )
 
 cmdtable = {}
 command = cmdutil.command(cmdtable)
 # 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 = 'ships-with-hg-core'
 
 class gpg(object):
     def __init__(self, path, key=None):
         self.path = path
         self.key = (key and " --local-user \"%s\"" % key) or ""
 
     def sign(self, data):
         gpgcmd = "%s --sign --detach-sign%s" % (self.path, self.key)
         return util.filter(data, gpgcmd)
 
     def verify(self, data, sig):
         """ returns of the good and bad signatures"""
         sigfile = datafile = None
         try:
             # create temporary files
             fd, sigfile = tempfile.mkstemp(prefix="hg-gpg-", suffix=".sig")
-            fp = os.fdopen(fd, 'wb')
+            fp = os.fdopen(fd, pycompat.sysstr('wb'))
             fp.write(sig)
             fp.close()
             fd, datafile = tempfile.mkstemp(prefix="hg-gpg-", suffix=".txt")
-            fp = os.fdopen(fd, 'wb')
+            fp = os.fdopen(fd, pycompat.sysstr('wb'))
             fp.write(data)
             fp.close()
             gpgcmd = ("%s --logger-fd 1 --status-fd 1 --verify "
                       "\"%s\" \"%s\"" % (self.path, sigfile, datafile))
             ret = util.filter("", gpgcmd)
         finally:
             for f in (sigfile, datafile):
                 try:
                     if f:
                         os.unlink(f)
                 except OSError:
                     pass
         keys = []
         key, fingerprint = None, None
         for l in ret.splitlines():
             # see DETAILS in the gnupg documentation
             # filter the logger output
             if not l.startswith("[GNUPG:]"):
                 continue
             l = l[9:]
             if l.startswith("VALIDSIG"):
                 # fingerprint of the primary key
                 fingerprint = l.split()[10]
             elif l.startswith("ERRSIG"):
                 key = l.split(" ", 3)[:2]
                 key.append("")
                 fingerprint = None
             elif (l.startswith("GOODSIG") or
                   l.startswith("EXPSIG") or
                   l.startswith("EXPKEYSIG") or
                   l.startswith("BADSIG")):
                 if key is not None:
                     keys.append(key + [fingerprint])
                 key = l.split(" ", 2)
                 fingerprint = None
         if key is not None:
             keys.append(key + [fingerprint])
         return keys
 
 def newgpg(ui, **opts):
     """create a new gpg instance"""
     gpgpath = ui.config("gpg", "cmd", "gpg")
     gpgkey = opts.get('key')
     if not gpgkey:
         gpgkey = ui.config("gpg", "key", None)
     return gpg(gpgpath, gpgkey)
 
 def sigwalk(repo):
     """
     walk over every sigs, yields a couple
     ((node, version, sig), (filename, linenumber))
     """
     def parsefile(fileiter, context):
         ln = 1
         for l in fileiter:
             if not l:
                 continue
             yield (l.split(" ", 2), (context, ln))
             ln += 1
 
     # read the heads
     fl = repo.file(".hgsigs")
     for r in reversed(fl.heads()):
         fn = ".hgsigs|%s" % hgnode.short(r)
         for item in parsefile(fl.read(r).splitlines(), fn):
             yield item
     try:
         # read local signatures
         fn = "localsigs"
         for item in parsefile(repo.vfs(fn), fn):
             yield item
     except IOError:
         pass
 
 def getkeys(ui, repo, mygpg, sigdata, context):
     """get the keys who signed a data"""
     fn, ln = context
     node, version, sig = sigdata
     prefix = "%s:%d" % (fn, ln)
     node = hgnode.bin(node)
 
     data = node2txt(repo, node, version)
     sig = binascii.a2b_base64(sig)
     keys = mygpg.verify(data, sig)
 
     validkeys = []
     # warn for expired key and/or sigs
     for key in keys:
         if key[0] == "ERRSIG":
             ui.write(_("%s Unknown key ID \"%s\"\n")
                      % (prefix, shortkey(ui, key[1][:15])))
             continue
         if key[0] == "BADSIG":
             ui.write(_("%s Bad signature from \"%s\"\n") % (prefix, key[2]))
             continue
         if key[0] == "EXPSIG":
             ui.write(_("%s Note: Signature has expired"
                        " (signed by: \"%s\")\n") % (prefix, key[2]))
         elif key[0] == "EXPKEYSIG":
             ui.write(_("%s Note: This key has expired"
                        " (signed by: \"%s\")\n") % (prefix, key[2]))
         validkeys.append((key[1], key[2], key[3]))
     return validkeys
 
 @command("sigs", [], _('hg sigs'))
 def sigs(ui, repo):
     """list signed changesets"""
     mygpg = newgpg(ui)
     revs = {}
 
     for data, context in sigwalk(repo):
         node, version, sig = data
         fn, ln = context
         try:
             n = repo.lookup(node)
         except KeyError:
             ui.warn(_("%s:%d node does not exist\n") % (fn, ln))
             continue
         r = repo.changelog.rev(n)
         keys = getkeys(ui, repo, mygpg, data, context)
         if not keys:
             continue
         revs.setdefault(r, [])
         revs[r].extend(keys)
     for rev in sorted(revs, reverse=True):
         for k in revs[rev]:
             r = "%5d:%s" % (rev, hgnode.hex(repo.changelog.node(rev)))
             ui.write("%-30s %s\n" % (keystr(ui, k), r))
 
 @command("sigcheck", [], _('hg sigcheck REV'))
 def sigcheck(ui, repo, rev):
     """verify all the signatures there may be for a particular revision"""
     mygpg = newgpg(ui)
     rev = repo.lookup(rev)
     hexrev = hgnode.hex(rev)
     keys = []
 
     for data, context in sigwalk(repo):
         node, version, sig = data
         if node == hexrev:
             k = getkeys(ui, repo, mygpg, data, context)
             if k:
                 keys.extend(k)
 
     if not keys:
         ui.write(_("no valid signature for %s\n") % hgnode.short(rev))
         return
 
     # print summary
     ui.write(_("%s is signed by:\n") % hgnode.short(rev))
     for key in keys:
         ui.write(" %s\n" % keystr(ui, key))
 
 def keystr(ui, key):
     """associate a string to a key (username, comment)"""
     keyid, user, fingerprint = key
     comment = ui.config("gpg", fingerprint, None)
     if comment:
         return "%s (%s)" % (user, comment)
     else:
         return user
 
 @command("sign",
          [('l', 'local', None, _('make the signature local')),
           ('f', 'force', None, _('sign even if the sigfile is modified')),
           ('', 'no-commit', None, _('do not commit the sigfile after signing')),
           ('k', 'key', '',
            _('the key id to sign with'), _('ID')),
           ('m', 'message', '',
            _('use text as commit message'), _('TEXT')),
           ('e', 'edit', False, _('invoke editor on commit messages')),
          ] + commands.commitopts2,
          _('hg sign [OPTION]... [REV]...'))
 def sign(ui, repo, *revs, **opts):
     """add a signature for the current or given revision
 
     If no revision is given, the parent of the working directory is used,
     or tip if no revision is checked out.
 
     The ``gpg.cmd`` config setting can be used to specify the command
     to run. A default key can be specified with ``gpg.key``.
 
     See :hg:`help dates` for a list of formats valid for -d/--date.
     """
     with repo.wlock():
         return _dosign(ui, repo, *revs, **opts)
 
 def _dosign(ui, repo, *revs, **opts):
     mygpg = newgpg(ui, **opts)
     sigver = "0"
     sigmessage = ""
 
     date = opts.get('date')
     if date:
         opts['date'] = util.parsedate(date)
 
     if revs:
         nodes = [repo.lookup(n) for n in revs]
     else:
         nodes = [node for node in repo.dirstate.parents()
                  if node != hgnode.nullid]
         if len(nodes) > 1:
             raise error.Abort(_('uncommitted merge - please provide a '
                                'specific revision'))
         if not nodes:
             nodes = [repo.changelog.tip()]
 
     for n in nodes:
         hexnode = hgnode.hex(n)
         ui.write(_("signing %d:%s\n") % (repo.changelog.rev(n),
                                          hgnode.short(n)))
         # build data
         data = node2txt(repo, n, sigver)
         sig = mygpg.sign(data)
         if not sig:
             raise error.Abort(_("error while signing"))
         sig = binascii.b2a_base64(sig)
         sig = sig.replace("\n", "")
         sigmessage += "%s %s %s\n" % (hexnode, sigver, sig)
 
     # write it
     if opts['local']:
         repo.vfs.append("localsigs", sigmessage)
         return
 
     if not opts["force"]:
         msigs = match.exact(repo.root, '', ['.hgsigs'])
         if any(repo.status(match=msigs, unknown=True, ignored=True)):
             raise error.Abort(_("working copy of .hgsigs is changed "),
                              hint=_("please commit .hgsigs manually"))
 
     sigsfile = repo.wfile(".hgsigs", "ab")
     sigsfile.write(sigmessage)
     sigsfile.close()
 
     if '.hgsigs' not in repo.dirstate:
         repo[None].add([".hgsigs"])
 
     if opts["no_commit"]:
         return
 
     message = opts['message']
     if not message:
         # we don't translate commit messages
         message = "\n".join(["Added signature for changeset %s"
                              % hgnode.short(n)
                              for n in nodes])
     try:
         editor = cmdutil.getcommiteditor(editform='gpg.sign', **opts)
         repo.commit(message, opts['user'], opts['date'], match=msigs,
                     editor=editor)
     except ValueError as inst:
         raise error.Abort(str(inst))
 
 def shortkey(ui, key):
     if len(key) != 16:
         ui.debug("key ID \"%s\" format error\n" % key)
         return key
 
     return key[-8:]
 
 def node2txt(repo, node, ver):
     """map a manifest into some text"""
     if ver == "0":
         return "%s\n" % hgnode.hex(node)
     else:
         raise error.Abort(_("unknown signature version"))

hgext/transplant.py

@@ -1,743 +1,744 b''
 # Patch transplanting extension for Mercurial
 #
 # Copyright 2006, 2007 Brendan Cully <[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.
 
 '''command to transplant changesets from another branch
 
 This extension allows you to transplant changes to another parent revision,
 possibly in another repository. The transplant is done using 'diff' patches.
 
 Transplanted patches are recorded in .hg/transplant/transplants, as a
 map from a changeset hash to its hash in the source repository.
 '''
 from __future__ import absolute_import
 
 import os
 import tempfile
 from mercurial.i18n import _
 from mercurial import (
     bundlerepo,
     cmdutil,
     error,
     exchange,
     hg,
     match,
     merge,
     node as nodemod,
     patch,
+    pycompat,
     registrar,
     revlog,
     revset,
     scmutil,
     util,
 )
 
 class TransplantError(error.Abort):
     pass
 
 cmdtable = {}
 command = cmdutil.command(cmdtable)
 # 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 = 'ships-with-hg-core'
 
 class transplantentry(object):
     def __init__(self, lnode, rnode):
         self.lnode = lnode
         self.rnode = rnode
 
 class transplants(object):
     def __init__(self, path=None, transplantfile=None, opener=None):
         self.path = path
         self.transplantfile = transplantfile
         self.opener = opener
 
         if not opener:
             self.opener = scmutil.opener(self.path)
         self.transplants = {}
         self.dirty = False
         self.read()
 
     def read(self):
         abspath = os.path.join(self.path, self.transplantfile)
         if self.transplantfile and os.path.exists(abspath):
             for line in self.opener.read(self.transplantfile).splitlines():
                 lnode, rnode = map(revlog.bin, line.split(':'))
                 list = self.transplants.setdefault(rnode, [])
                 list.append(transplantentry(lnode, rnode))
 
     def write(self):
         if self.dirty and self.transplantfile:
             if not os.path.isdir(self.path):
                 os.mkdir(self.path)
             fp = self.opener(self.transplantfile, 'w')
             for list in self.transplants.itervalues():
                 for t in list:
                     l, r = map(nodemod.hex, (t.lnode, t.rnode))
                     fp.write(l + ':' + r + '\n')
             fp.close()
         self.dirty = False
 
     def get(self, rnode):
         return self.transplants.get(rnode) or []
 
     def set(self, lnode, rnode):
         list = self.transplants.setdefault(rnode, [])
         list.append(transplantentry(lnode, rnode))
         self.dirty = True
 
     def remove(self, transplant):
         list = self.transplants.get(transplant.rnode)
         if list:
             del list[list.index(transplant)]
             self.dirty = True
 
 class transplanter(object):
     def __init__(self, ui, repo, opts):
         self.ui = ui
         self.path = repo.join('transplant')
         self.opener = scmutil.opener(self.path)
         self.transplants = transplants(self.path, 'transplants',
                                        opener=self.opener)
         def getcommiteditor():
             editform = cmdutil.mergeeditform(repo[None], 'transplant')
             return cmdutil.getcommiteditor(editform=editform, **opts)
         self.getcommiteditor = getcommiteditor
 
     def applied(self, repo, node, parent):
         '''returns True if a node is already an ancestor of parent
         or is parent or has already been transplanted'''
         if hasnode(repo, parent):
             parentrev = repo.changelog.rev(parent)
         if hasnode(repo, node):
             rev = repo.changelog.rev(node)
             reachable = repo.changelog.ancestors([parentrev], rev,
                                                  inclusive=True)
             if rev in reachable:
                 return True
         for t in self.transplants.get(node):
             # it might have been stripped
             if not hasnode(repo, t.lnode):
                 self.transplants.remove(t)
                 return False
             lnoderev = repo.changelog.rev(t.lnode)
             if lnoderev in repo.changelog.ancestors([parentrev], lnoderev,
                                                     inclusive=True):
                 return True
         return False
 
     def apply(self, repo, source, revmap, merges, opts=None):
         '''apply the revisions in revmap one by one in revision order'''
         if opts is None:
             opts = {}
         revs = sorted(revmap)
         p1, p2 = repo.dirstate.parents()
         pulls = []
         diffopts = patch.difffeatureopts(self.ui, opts)
         diffopts.git = True
 
         lock = tr = None
         try:
             lock = repo.lock()
             tr = repo.transaction('transplant')
             for rev in revs:
                 node = revmap[rev]
                 revstr = '%s:%s' % (rev, nodemod.short(node))
 
                 if self.applied(repo, node, p1):
                     self.ui.warn(_('skipping already applied revision %s\n') %
                                  revstr)
                     continue
 
                 parents = source.changelog.parents(node)
                 if not (opts.get('filter') or opts.get('log')):
                     # If the changeset parent is the same as the
                     # wdir's parent, just pull it.
                     if parents[0] == p1:
                         pulls.append(node)
                         p1 = node
                         continue
                     if pulls:
                         if source != repo:
                             exchange.pull(repo, source.peer(), heads=pulls)
                         merge.update(repo, pulls[-1], False, False)
                         p1, p2 = repo.dirstate.parents()
                         pulls = []
 
                 domerge = False
                 if node in merges:
                     # pulling all the merge revs at once would mean we
                     # couldn't transplant after the latest even if
                     # transplants before them fail.
                     domerge = True
                     if not hasnode(repo, node):
                         exchange.pull(repo, source.peer(), heads=[node])
 
                 skipmerge = False
                 if parents[1] != revlog.nullid:
                     if not opts.get('parent'):
                         self.ui.note(_('skipping merge changeset %s:%s\n')
                                      % (rev, nodemod.short(node)))
                         skipmerge = True
                     else:
                         parent = source.lookup(opts['parent'])
                         if parent not in parents:
                             raise error.Abort(_('%s is not a parent of %s') %
                                               (nodemod.short(parent),
                                                nodemod.short(node)))
                 else:
                     parent = parents[0]
 
                 if skipmerge:
                     patchfile = None
                 else:
                     fd, patchfile = tempfile.mkstemp(prefix='hg-transplant-')
-                    fp = os.fdopen(fd, 'w')
+                    fp = os.fdopen(fd, pycompat.sysstr('w'))
                     gen = patch.diff(source, parent, node, opts=diffopts)
                     for chunk in gen:
                         fp.write(chunk)
                     fp.close()
 
                 del revmap[rev]
                 if patchfile or domerge:
                     try:
                         try:
                             n = self.applyone(repo, node,
                                               source.changelog.read(node),
                                               patchfile, merge=domerge,
                                               log=opts.get('log'),
                                               filter=opts.get('filter'))
                         except TransplantError:
                             # Do not rollback, it is up to the user to
                             # fix the merge or cancel everything
                             tr.close()
                             raise
                         if n and domerge:
                             self.ui.status(_('%s merged at %s\n') % (revstr,
                                       nodemod.short(n)))
                         elif n:
                             self.ui.status(_('%s transplanted to %s\n')
                                            % (nodemod.short(node),
                                               nodemod.short(n)))
                     finally:
                         if patchfile:
                             os.unlink(patchfile)
             tr.close()
             if pulls:
                 exchange.pull(repo, source.peer(), heads=pulls)
                 merge.update(repo, pulls[-1], False, False)
         finally:
             self.saveseries(revmap, merges)
             self.transplants.write()
             if tr:
                 tr.release()
             if lock:
                 lock.release()
 
     def filter(self, filter, node, changelog, patchfile):
         '''arbitrarily rewrite changeset before applying it'''
 
         self.ui.status(_('filtering %s\n') % patchfile)
         user, date, msg = (changelog[1], changelog[2], changelog[4])
         fd, headerfile = tempfile.mkstemp(prefix='hg-transplant-')
-        fp = os.fdopen(fd, 'w')
+        fp = os.fdopen(fd, pycompat.sysstr('w'))
         fp.write("# HG changeset patch\n")
         fp.write("# User %s\n" % user)
         fp.write("# Date %d %d\n" % date)
         fp.write(msg + '\n')
         fp.close()
 
         try:
             self.ui.system('%s %s %s' % (filter, util.shellquote(headerfile),
                                          util.shellquote(patchfile)),
                            environ={'HGUSER': changelog[1],
                                     'HGREVISION': nodemod.hex(node),
                                     },
                            onerr=error.Abort, errprefix=_('filter failed'))
             user, date, msg = self.parselog(file(headerfile))[1:4]
         finally:
             os.unlink(headerfile)
 
         return (user, date, msg)
 
     def applyone(self, repo, node, cl, patchfile, merge=False, log=False,
                  filter=None):
         '''apply the patch in patchfile to the repository as a transplant'''
         (manifest, user, (time, timezone), files, message) = cl[:5]
         date = "%d %d" % (time, timezone)
         extra = {'transplant_source': node}
         if filter:
             (user, date, message) = self.filter(filter, node, cl, patchfile)
 
         if log:
             # we don't translate messages inserted into commits
             message += '\n(transplanted from %s)' % nodemod.hex(node)
 
         self.ui.status(_('applying %s\n') % nodemod.short(node))
         self.ui.note('%s %s\n%s\n' % (user, date, message))
 
         if not patchfile and not merge:
             raise error.Abort(_('can only omit patchfile if merging'))
         if patchfile:
             try:
                 files = set()
                 patch.patch(self.ui, repo, patchfile, files=files, eolmode=None)
                 files = list(files)
             except Exception as inst:
                 seriespath = os.path.join(self.path, 'series')
                 if os.path.exists(seriespath):
                     os.unlink(seriespath)
                 p1 = repo.dirstate.p1()
                 p2 = node
                 self.log(user, date, message, p1, p2, merge=merge)
                 self.ui.write(str(inst) + '\n')
                 raise TransplantError(_('fix up the working directory and run '
                                         'hg transplant --continue'))
         else:
             files = None
         if merge:
             p1, p2 = repo.dirstate.parents()
             repo.setparents(p1, node)
             m = match.always(repo.root, '')
         else:
             m = match.exact(repo.root, '', files)
 
         n = repo.commit(message, user, date, extra=extra, match=m,
                         editor=self.getcommiteditor())
         if not n:
             self.ui.warn(_('skipping emptied changeset %s\n') %
                            nodemod.short(node))
             return None
         if not merge:
             self.transplants.set(n, node)
 
         return n
 
     def canresume(self):
         return os.path.exists(os.path.join(self.path, 'journal'))
 
     def resume(self, repo, source, opts):
         '''recover last transaction and apply remaining changesets'''
         if os.path.exists(os.path.join(self.path, 'journal')):
             n, node = self.recover(repo, source, opts)
             if n:
                 self.ui.status(_('%s transplanted as %s\n') %
                                  (nodemod.short(node),
                                   nodemod.short(n)))
             else:
                 self.ui.status(_('%s skipped due to empty diff\n')
                                % (nodemod.short(node),))
         seriespath = os.path.join(self.path, 'series')
         if not os.path.exists(seriespath):
             self.transplants.write()
             return
         nodes, merges = self.readseries()
         revmap = {}
         for n in nodes:
             revmap[source.changelog.rev(n)] = n
         os.unlink(seriespath)
 
         self.apply(repo, source, revmap, merges, opts)
 
     def recover(self, repo, source, opts):
         '''commit working directory using journal metadata'''
         node, user, date, message, parents = self.readlog()
         merge = False
 
         if not user or not date or not message or not parents[0]:
             raise error.Abort(_('transplant log file is corrupt'))
 
         parent = parents[0]
         if len(parents) > 1:
             if opts.get('parent'):
                 parent = source.lookup(opts['parent'])
                 if parent not in parents:
                     raise error.Abort(_('%s is not a parent of %s') %
                                       (nodemod.short(parent),
                                        nodemod.short(node)))
             else:
                 merge = True
 
         extra = {'transplant_source': node}
         try:
             p1, p2 = repo.dirstate.parents()
             if p1 != parent:
                 raise error.Abort(_('working directory not at transplant '
                                    'parent %s') % nodemod.hex(parent))
             if merge:
                 repo.setparents(p1, parents[1])
             modified, added, removed, deleted = repo.status()[:4]
             if merge or modified or added or removed or deleted:
                 n = repo.commit(message, user, date, extra=extra,
                                 editor=self.getcommiteditor())
                 if not n:
                     raise error.Abort(_('commit failed'))
                 if not merge:
                     self.transplants.set(n, node)
             else:
                 n = None
             self.unlog()
 
             return n, node
         finally:
             # TODO: get rid of this meaningless try/finally enclosing.
             # this is kept only to reduce changes in a patch.
             pass
 
     def readseries(self):
         nodes = []
         merges = []
         cur = nodes
         for line in self.opener.read('series').splitlines():
             if line.startswith('# Merges'):
                 cur = merges
                 continue
             cur.append(revlog.bin(line))
 
         return (nodes, merges)
 
     def saveseries(self, revmap, merges):
         if not revmap:
             return
 
         if not os.path.isdir(self.path):
             os.mkdir(self.path)
         series = self.opener('series', 'w')
         for rev in sorted(revmap):
             series.write(nodemod.hex(revmap[rev]) + '\n')
         if merges:
             series.write('# Merges\n')
             for m in merges:
                 series.write(nodemod.hex(m) + '\n')
         series.close()
 
     def parselog(self, fp):
         parents = []
         message = []
         node = revlog.nullid
         inmsg = False
         user = None
         date = None
         for line in fp.read().splitlines():
             if inmsg:
                 message.append(line)
             elif line.startswith('# User '):
                 user = line[7:]
             elif line.startswith('# Date '):
                 date = line[7:]
             elif line.startswith('# Node ID '):
                 node = revlog.bin(line[10:])
             elif line.startswith('# Parent '):
                 parents.append(revlog.bin(line[9:]))
             elif not line.startswith('# '):
                 inmsg = True
                 message.append(line)
         if None in (user, date):
             raise error.Abort(_("filter corrupted changeset (no user or date)"))
         return (node, user, date, '\n'.join(message), parents)
 
     def log(self, user, date, message, p1, p2, merge=False):
         '''journal changelog metadata for later recover'''
 
         if not os.path.isdir(self.path):
             os.mkdir(self.path)
         fp = self.opener('journal', 'w')
         fp.write('# User %s\n' % user)
         fp.write('# Date %s\n' % date)
         fp.write('# Node ID %s\n' % nodemod.hex(p2))
         fp.write('# Parent ' + nodemod.hex(p1) + '\n')
         if merge:
             fp.write('# Parent ' + nodemod.hex(p2) + '\n')
         fp.write(message.rstrip() + '\n')
         fp.close()
 
     def readlog(self):
         return self.parselog(self.opener('journal'))
 
     def unlog(self):
         '''remove changelog journal'''
         absdst = os.path.join(self.path, 'journal')
         if os.path.exists(absdst):
             os.unlink(absdst)
 
     def transplantfilter(self, repo, source, root):
         def matchfn(node):
             if self.applied(repo, node, root):
                 return False
             if source.changelog.parents(node)[1] != revlog.nullid:
                 return False
             extra = source.changelog.read(node)[5]
             cnode = extra.get('transplant_source')
             if cnode and self.applied(repo, cnode, root):
                 return False
             return True
 
         return matchfn
 
 def hasnode(repo, node):
     try:
         return repo.changelog.rev(node) is not None
     except error.RevlogError:
         return False
 
 def browserevs(ui, repo, nodes, opts):
     '''interactively transplant changesets'''
     displayer = cmdutil.show_changeset(ui, repo, opts)
     transplants = []
     merges = []
     prompt = _('apply changeset? [ynmpcq?]:'
                '$$ &yes, transplant this changeset'
                '$$ &no, skip this changeset'
                '$$ &merge at this changeset'
                '$$ show &patch'
                '$$ &commit selected changesets'
                '$$ &quit and cancel transplant'
                '$$ &? (show this help)')
     for node in nodes:
         displayer.show(repo[node])
         action = None
         while not action:
             action = 'ynmpcq?'[ui.promptchoice(prompt)]
             if action == '?':
                 for c, t in ui.extractchoices(prompt)[1]:
                     ui.write('%s: %s\n' % (c, t))
                 action = None
             elif action == 'p':
                 parent = repo.changelog.parents(node)[0]
                 for chunk in patch.diff(repo, parent, node):
                     ui.write(chunk)
                 action = None
         if action == 'y':
             transplants.append(node)
         elif action == 'm':
             merges.append(node)
         elif action == 'c':
             break
         elif action == 'q':
             transplants = ()
             merges = ()
             break
     displayer.close()
     return (transplants, merges)
 
 @command('transplant',
     [('s', 'source', '', _('transplant changesets from REPO'), _('REPO')),
     ('b', 'branch', [], _('use this source changeset as head'), _('REV')),
     ('a', 'all', None, _('pull all changesets up to the --branch revisions')),
     ('p', 'prune', [], _('skip over REV'), _('REV')),
     ('m', 'merge', [], _('merge at REV'), _('REV')),
     ('', 'parent', '',
      _('parent to choose when transplanting merge'), _('REV')),
     ('e', 'edit', False, _('invoke editor on commit messages')),
     ('', 'log', None, _('append transplant info to log message')),
     ('c', 'continue', None, _('continue last transplant session '
                               'after fixing conflicts')),
     ('', 'filter', '',
      _('filter changesets through command'), _('CMD'))],
     _('hg transplant [-s REPO] [-b BRANCH [-a]] [-p REV] '
       '[-m REV] [REV]...'))
 def transplant(ui, repo, *revs, **opts):
     '''transplant changesets from another branch
 
     Selected changesets will be applied on top of the current working
     directory with the log of the original changeset. The changesets
     are copied and will thus appear twice in the history with different
     identities.
 
     Consider using the graft command if everything is inside the same
     repository - it will use merges and will usually give a better result.
     Use the rebase extension if the changesets are unpublished and you want
     to move them instead of copying them.
 
     If --log is specified, log messages will have a comment appended
     of the form::
 
       (transplanted from CHANGESETHASH)
 
     You can rewrite the changelog message with the --filter option.
     Its argument will be invoked with the current changelog message as
     $1 and the patch as $2.
 
     --source/-s specifies another repository to use for selecting changesets,
     just as if it temporarily had been pulled.
     If --branch/-b is specified, these revisions will be used as
     heads when deciding which changesets to transplant, just as if only
     these revisions had been pulled.
     If --all/-a is specified, all the revisions up to the heads specified
     with --branch will be transplanted.
 
     Example:
 
     - transplant all changes up to REV on top of your current revision::
 
         hg transplant --branch REV --all
 
     You can optionally mark selected transplanted changesets as merge
     changesets. You will not be prompted to transplant any ancestors
     of a merged transplant, and you can merge descendants of them
     normally instead of transplanting them.
 
     Merge changesets may be transplanted directly by specifying the
     proper parent changeset by calling :hg:`transplant --parent`.
 
     If no merges or revisions are provided, :hg:`transplant` will
     start an interactive changeset browser.
 
     If a changeset application fails, you can fix the merge by hand
     and then resume where you left off by calling :hg:`transplant
     --continue/-c`.
     '''
     with repo.wlock():
         return _dotransplant(ui, repo, *revs, **opts)
 
 def _dotransplant(ui, repo, *revs, **opts):
     def incwalk(repo, csets, match=util.always):
         for node in csets:
             if match(node):
                 yield node
 
     def transplantwalk(repo, dest, heads, match=util.always):
         '''Yield all nodes that are ancestors of a head but not ancestors
         of dest.
         If no heads are specified, the heads of repo will be used.'''
         if not heads:
             heads = repo.heads()
         ancestors = []
         ctx = repo[dest]
         for head in heads:
             ancestors.append(ctx.ancestor(repo[head]).node())
         for node in repo.changelog.nodesbetween(ancestors, heads)[0]:
             if match(node):
                 yield node
 
     def checkopts(opts, revs):
         if opts.get('continue'):
             if opts.get('branch') or opts.get('all') or opts.get('merge'):
                 raise error.Abort(_('--continue is incompatible with '
                                    '--branch, --all and --merge'))
             return
         if not (opts.get('source') or revs or
                 opts.get('merge') or opts.get('branch')):
             raise error.Abort(_('no source URL, branch revision, or revision '
                                'list provided'))
         if opts.get('all'):
             if not opts.get('branch'):
                 raise error.Abort(_('--all requires a branch revision'))
             if revs:
                 raise error.Abort(_('--all is incompatible with a '
                                    'revision list'))
 
     checkopts(opts, revs)
 
     if not opts.get('log'):
         # deprecated config: transplant.log
         opts['log'] = ui.config('transplant', 'log')
     if not opts.get('filter'):
         # deprecated config: transplant.filter
         opts['filter'] = ui.config('transplant', 'filter')
 
     tp = transplanter(ui, repo, opts)
 
     p1, p2 = repo.dirstate.parents()
     if len(repo) > 0 and p1 == revlog.nullid:
         raise error.Abort(_('no revision checked out'))
     if opts.get('continue'):
         if not tp.canresume():
             raise error.Abort(_('no transplant to continue'))
     else:
         cmdutil.checkunfinished(repo)
         if p2 != revlog.nullid:
             raise error.Abort(_('outstanding uncommitted merges'))
         m, a, r, d = repo.status()[:4]
         if m or a or r or d:
             raise error.Abort(_('outstanding local changes'))
 
     sourcerepo = opts.get('source')
     if sourcerepo:
         peer = hg.peer(repo, opts, ui.expandpath(sourcerepo))
         heads = map(peer.lookup, opts.get('branch', ()))
         target = set(heads)
         for r in revs:
             try:
                 target.add(peer.lookup(r))
             except error.RepoError:
                 pass
         source, csets, cleanupfn = bundlerepo.getremotechanges(ui, repo, peer,
                                     onlyheads=sorted(target), force=True)
     else:
         source = repo
         heads = map(source.lookup, opts.get('branch', ()))
         cleanupfn = None
 
     try:
         if opts.get('continue'):
             tp.resume(repo, source, opts)
             return
 
         tf = tp.transplantfilter(repo, source, p1)
         if opts.get('prune'):
             prune = set(source.lookup(r)
                         for r in scmutil.revrange(source, opts.get('prune')))
             matchfn = lambda x: tf(x) and x not in prune
         else:
             matchfn = tf
         merges = map(source.lookup, opts.get('merge', ()))
         revmap = {}
         if revs:
             for r in scmutil.revrange(source, revs):
                 revmap[int(r)] = source.lookup(r)
         elif opts.get('all') or not merges:
             if source != repo:
                 alltransplants = incwalk(source, csets, match=matchfn)
             else:
                 alltransplants = transplantwalk(source, p1, heads,
                                                 match=matchfn)
             if opts.get('all'):
                 revs = alltransplants
             else:
                 revs, newmerges = browserevs(ui, source, alltransplants, opts)
                 merges.extend(newmerges)
             for r in revs:
                 revmap[source.changelog.rev(r)] = r
         for r in merges:
             revmap[source.changelog.rev(r)] = r
 
         tp.apply(repo, source, revmap, merges, opts)
     finally:
         if cleanupfn:
             cleanupfn()
 
 revsetpredicate = registrar.revsetpredicate()
 
 @revsetpredicate('transplanted([set])')
 def revsettransplanted(repo, subset, x):
     """Transplanted changesets in set, or all transplanted changesets.
     """
     if x:
         s = revset.getset(repo, subset, x)
     else:
         s = subset
     return revset.baseset([r for r in s if
         repo[r].extra().get('transplant_source')])
 
 templatekeyword = registrar.templatekeyword()
 
 @templatekeyword('transplanted')
 def kwtransplanted(repo, ctx, **args):
     """String. The node identifier of the transplanted
     changeset if any."""
     n = ctx.extra().get('transplant_source')
     return n and nodemod.hex(n) or ''
 
 def extsetup(ui):
     cmdutil.unfinishedstates.append(
         ['transplant/journal', True, False, _('transplant in progress'),
          _("use 'hg transplant --continue' or 'hg update' to abort")])
 
 # tell hggettext to extract docstrings from these functions:
 i18nfunctions = [revsettransplanted, kwtransplanted]

mercurial/changegroup.py

@@ -1,1044 +1,1045 b''
 # changegroup.py - Mercurial changegroup manipulation functions
 #
 #  Copyright 2006 Matt Mackall <[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.
 
 from __future__ import absolute_import
 
 import os
 import struct
 import tempfile
 import weakref
 
 from .i18n import _
 from .node import (
     hex,
     nullrev,
     short,
 )
 
 from . import (
     branchmap,
     dagutil,
     discovery,
     error,
     mdiff,
     phases,
+    pycompat,
     util,
 )
 
 _CHANGEGROUPV1_DELTA_HEADER = "20s20s20s20s"
 _CHANGEGROUPV2_DELTA_HEADER = "20s20s20s20s20s"
 _CHANGEGROUPV3_DELTA_HEADER = ">20s20s20s20s20sH"
 
 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(_("stream ended unexpectedly"
                            " (got %d bytes, expected %d)")
                           % (len(s), n))
     return s
 
 def getchunk(stream):
     """return the next chunk from stream as a string"""
     d = readexactly(stream, 4)
     l = struct.unpack(">l", d)[0]
     if l <= 4:
         if l:
             raise error.Abort(_("invalid chunk length %d") % l)
         return ""
     return readexactly(stream, l - 4)
 
 def chunkheader(length):
     """return a changegroup chunk header (string)"""
     return struct.pack(">l", length + 4)
 
 def closechunk():
     """return a changegroup chunk header (string) for a zero-length chunk"""
     return struct.pack(">l", 0)
 
 def combineresults(results):
     """logic to combine 0 or more addchangegroup results into one"""
     changedheads = 0
     result = 1
     for ret in results:
         # If any changegroup result is 0, return 0
         if ret == 0:
             result = 0
             break
         if ret < -1:
             changedheads += ret + 1
         elif ret > 1:
             changedheads += ret - 1
     if changedheads > 0:
         result = 1 + changedheads
     elif changedheads < 0:
         result = -1 + changedheads
     return result
 
 def writechunks(ui, chunks, filename, vfs=None):
     """Write chunks to a file and return its filename.
 
     The stream is assumed to be a bundle file.
     Existing files will not be overwritten.
     If no filename is specified, a temporary file is created.
     """
     fh = None
     cleanup = None
     try:
         if filename:
             if vfs:
                 fh = vfs.open(filename, "wb")
             else:
                 # Increase default buffer size because default is usually
                 # small (4k is common on Linux).
                 fh = open(filename, "wb", 131072)
         else:
             fd, filename = tempfile.mkstemp(prefix="hg-bundle-", suffix=".hg")
-            fh = os.fdopen(fd, "wb")
+            fh = os.fdopen(fd, pycompat.sysstr("wb"))
         cleanup = filename
         for c in chunks:
             fh.write(c)
         cleanup = None
         return filename
     finally:
         if fh is not None:
             fh.close()
         if cleanup is not None:
             if filename and vfs:
                 vfs.unlink(cleanup)
             else:
                 os.unlink(cleanup)
 
 class cg1unpacker(object):
     """Unpacker for cg1 changegroup streams.
 
     A changegroup unpacker handles the framing of the revision data in
     the wire format. Most consumers will want to use the apply()
     method to add the changes from the changegroup to a repository.
 
     If you're forwarding a changegroup unmodified to another consumer,
     use getchunks(), which returns an iterator of changegroup
     chunks. This is mostly useful for cases where you need to know the
     data stream has ended by observing the end of the changegroup.
 
     deltachunk() is useful only if you're applying delta data. Most
     consumers should prefer apply() instead.
 
     A few other public methods exist. Those are used only for
     bundlerepo and some debug commands - their use is discouraged.
     """
     deltaheader = _CHANGEGROUPV1_DELTA_HEADER
     deltaheadersize = struct.calcsize(deltaheader)
     version = '01'
     _grouplistcount = 1 # One list of files after the manifests
 
     def __init__(self, fh, alg, extras=None):
         if alg is None:
             alg = 'UN'
         if alg not in util.compengines.supportedbundletypes:
             raise error.Abort(_('unknown stream compression type: %s')
                              % alg)
         if alg == 'BZ':
             alg = '_truncatedBZ'
 
         compengine = util.compengines.forbundletype(alg)
         self._stream = compengine.decompressorreader(fh)
         self._type = alg
         self.extras = extras or {}
         self.callback = None
 
     # These methods (compressed, read, seek, tell) all appear to only
     # be used by bundlerepo, but it's a little hard to tell.
     def compressed(self):
         return self._type is not None and self._type != 'UN'
     def read(self, l):
         return self._stream.read(l)
     def seek(self, pos):
         return self._stream.seek(pos)
     def tell(self):
         return self._stream.tell()
     def close(self):
         return self._stream.close()
 
     def _chunklength(self):
         d = readexactly(self._stream, 4)
         l = struct.unpack(">l", d)[0]
         if l <= 4:
             if l:
                 raise error.Abort(_("invalid chunk length %d") % l)
             return 0
         if self.callback:
             self.callback()
         return l - 4
 
     def changelogheader(self):
         """v10 does not have a changelog header chunk"""
         return {}
 
     def manifestheader(self):
         """v10 does not have a manifest header chunk"""
         return {}
 
     def filelogheader(self):
         """return the header of the filelogs chunk, v10 only has the filename"""
         l = self._chunklength()
         if not l:
             return {}
         fname = readexactly(self._stream, l)
         return {'filename': fname}
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, cs = headertuple
         if prevnode is None:
             deltabase = p1
         else:
             deltabase = prevnode
         flags = 0
         return node, p1, p2, deltabase, cs, flags
 
     def deltachunk(self, prevnode):
         l = self._chunklength()
         if not l:
             return {}
         headerdata = readexactly(self._stream, self.deltaheadersize)
         header = struct.unpack(self.deltaheader, headerdata)
         delta = readexactly(self._stream, l - self.deltaheadersize)
         node, p1, p2, deltabase, cs, flags = self._deltaheader(header, prevnode)
         return {'node': node, 'p1': p1, 'p2': p2, 'cs': cs,
                 'deltabase': deltabase, 'delta': delta, 'flags': flags}
 
     def getchunks(self):
         """returns all the chunks contains in the bundle
 
         Used when you need to forward the binary stream to a file or another
         network API. To do so, it parse the changegroup data, otherwise it will
         block in case of sshrepo because it don't know the end of the stream.
         """
         # an empty chunkgroup is the end of the changegroup
         # a changegroup has at least 2 chunkgroups (changelog and manifest).
         # after that, changegroup versions 1 and 2 have a series of groups
         # with one group per file. changegroup 3 has a series of directory
         # manifests before the files.
         count = 0
         emptycount = 0
         while emptycount < self._grouplistcount:
             empty = True
             count += 1
             while True:
                 chunk = getchunk(self)
                 if not chunk:
                     if empty and count > 2:
                         emptycount += 1
                     break
                 empty = False
                 yield chunkheader(len(chunk))
                 pos = 0
                 while pos < len(chunk):
                     next = pos + 2**20
                     yield chunk[pos:next]
                     pos = next
             yield closechunk()
 
     def _unpackmanifests(self, repo, revmap, trp, prog, numchanges):
         # We know that we'll never have more manifests than we had
         # changesets.
         self.callback = prog(_('manifests'), numchanges)
         # no need to check for empty manifest group here:
         # if the result of the merge of 1 and 2 is the same in 3 and 4,
         # no new manifest will be created and the manifest group will
         # be empty during the pull
         self.manifestheader()
         repo.manifestlog._revlog.addgroup(self, revmap, trp)
         repo.ui.progress(_('manifests'), None)
         self.callback = None
 
     def apply(self, repo, srctype, url, emptyok=False,
               targetphase=phases.draft, expectedtotal=None):
         """Add the changegroup returned by source.read() to this repo.
         srctype is a string like 'push', 'pull', or 'unbundle'.  url is
         the URL of the repo where this changegroup is coming from.
 
         Return an integer summarizing the change to this repo:
         - nothing changed or no source: 0
         - more heads than before: 1+added heads (2..n)
         - fewer heads than before: -1-removed heads (-2..-n)
         - number of heads stays the same: 1
         """
         repo = repo.unfiltered()
         def csmap(x):
             repo.ui.debug("add changeset %s\n" % short(x))
             return len(cl)
 
         def revmap(x):
             return cl.rev(x)
 
         changesets = files = revisions = 0
 
         try:
             with repo.transaction("\n".join([srctype,
                                              util.hidepassword(url)])) as tr:
                 # The transaction could have been created before and already
                 # carries source information. In this case we use the top
                 # level data. We overwrite the argument because we need to use
                 # the top level value (if they exist) in this function.
                 srctype = tr.hookargs.setdefault('source', srctype)
                 url = tr.hookargs.setdefault('url', url)
                 repo.hook('prechangegroup', throw=True, **tr.hookargs)
 
                 # write changelog data to temp files so concurrent readers
                 # will not see an inconsistent view
                 cl = repo.changelog
                 cl.delayupdate(tr)
                 oldheads = cl.heads()
 
                 trp = weakref.proxy(tr)
                 # pull off the changeset group
                 repo.ui.status(_("adding changesets\n"))
                 clstart = len(cl)
                 class prog(object):
                     def __init__(self, step, total):
                         self._step = step
                         self._total = total
                         self._count = 1
                     def __call__(self):
                         repo.ui.progress(self._step, self._count,
                                          unit=_('chunks'), total=self._total)
                         self._count += 1
                 self.callback = prog(_('changesets'), expectedtotal)
 
                 efiles = set()
                 def onchangelog(cl, node):
                     efiles.update(cl.readfiles(node))
 
                 self.changelogheader()
                 srccontent = cl.addgroup(self, csmap, trp,
                                          addrevisioncb=onchangelog)
                 efiles = len(efiles)
 
                 if not (srccontent or emptyok):
                     raise error.Abort(_("received changelog group is empty"))
                 clend = len(cl)
                 changesets = clend - clstart
                 repo.ui.progress(_('changesets'), None)
                 self.callback = None
 
                 # pull off the manifest group
                 repo.ui.status(_("adding manifests\n"))
                 self._unpackmanifests(repo, revmap, trp, prog, changesets)
 
                 needfiles = {}
                 if repo.ui.configbool('server', 'validate', default=False):
                     cl = repo.changelog
                     ml = repo.manifestlog
                     # validate incoming csets have their manifests
                     for cset in xrange(clstart, clend):
                         mfnode = cl.changelogrevision(cset).manifest
                         mfest = ml[mfnode].readdelta()
                         # store file nodes we must see
                         for f, n in mfest.iteritems():
                             needfiles.setdefault(f, set()).add(n)
 
                 # process the files
                 repo.ui.status(_("adding file changes\n"))
                 newrevs, newfiles = _addchangegroupfiles(
                     repo, self, revmap, trp, efiles, needfiles)
                 revisions += newrevs
                 files += newfiles
 
                 dh = 0
                 if oldheads:
                     heads = cl.heads()
                     dh = len(heads) - len(oldheads)
                     for h in heads:
                         if h not in oldheads and repo[h].closesbranch():
                             dh -= 1
                 htext = ""
                 if dh:
                     htext = _(" (%+d heads)") % dh
 
                 repo.ui.status(_("added %d changesets"
                                  " with %d changes to %d files%s\n")
                                  % (changesets, revisions, files, htext))
                 repo.invalidatevolatilesets()
 
                 if changesets > 0:
                     if 'node' not in tr.hookargs:
                         tr.hookargs['node'] = hex(cl.node(clstart))
                         tr.hookargs['node_last'] = hex(cl.node(clend - 1))
                         hookargs = dict(tr.hookargs)
                     else:
                         hookargs = dict(tr.hookargs)
                         hookargs['node'] = hex(cl.node(clstart))
                         hookargs['node_last'] = hex(cl.node(clend - 1))
                     repo.hook('pretxnchangegroup', throw=True, **hookargs)
 
                 added = [cl.node(r) for r in xrange(clstart, clend)]
                 publishing = repo.publishing()
                 if srctype in ('push', 'serve'):
                     # Old servers can not push the boundary themselves.
                     # New servers won't push the boundary if changeset already
                     # exists locally as secret
                     #
                     # We should not use added here but the list of all change in
                     # the bundle
                     if publishing:
                         phases.advanceboundary(repo, tr, phases.public,
                                                srccontent)
                     else:
                         # Those changesets have been pushed from the
                         # outside, their phases are going to be pushed
                         # alongside. Therefor `targetphase` is
                         # ignored.
                         phases.advanceboundary(repo, tr, phases.draft,
                                                srccontent)
                         phases.retractboundary(repo, tr, phases.draft, added)
                 elif srctype != 'strip':
                     # publishing only alter behavior during push
                     #
                     # strip should not touch boundary at all
                     phases.retractboundary(repo, tr, targetphase, added)
 
                 if changesets > 0:
                     if srctype != 'strip':
                         # During strip, branchcache is invalid but
                         # coming call to `destroyed` will repair it.
                         # In other case we can safely update cache on
                         # disk.
                         repo.ui.debug('updating the branch cache\n')
                         branchmap.updatecache(repo.filtered('served'))
 
                     def runhooks():
                         # These hooks run when the lock releases, not when the
                         # transaction closes. So it's possible for the changelog
                         # to have changed since we last saw it.
                         if clstart >= len(repo):
                             return
 
                         repo.hook("changegroup", **hookargs)
 
                         for n in added:
                             args = hookargs.copy()
                             args['node'] = hex(n)
                             del args['node_last']
                             repo.hook("incoming", **args)
 
                         newheads = [h for h in repo.heads()
                                     if h not in oldheads]
                         repo.ui.log("incoming",
                                     "%s incoming changes - new heads: %s\n",
                                     len(added),
                                     ', '.join([hex(c[:6]) for c in newheads]))
 
                     tr.addpostclose('changegroup-runhooks-%020i' % clstart,
                                     lambda tr: repo._afterlock(runhooks))
         finally:
             repo.ui.flush()
         # never return 0 here:
         if dh < 0:
             return dh - 1
         else:
             return dh + 1
 
 class cg2unpacker(cg1unpacker):
     """Unpacker for cg2 streams.
 
     cg2 streams add support for generaldelta, so the delta header
     format is slightly different. All other features about the data
     remain the same.
     """
     deltaheader = _CHANGEGROUPV2_DELTA_HEADER
     deltaheadersize = struct.calcsize(deltaheader)
     version = '02'
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, deltabase, cs = headertuple
         flags = 0
         return node, p1, p2, deltabase, cs, flags
 
 class cg3unpacker(cg2unpacker):
     """Unpacker for cg3 streams.
 
     cg3 streams add support for exchanging treemanifests and revlog
     flags. It adds the revlog flags to the delta header and an empty chunk
     separating manifests and files.
     """
     deltaheader = _CHANGEGROUPV3_DELTA_HEADER
     deltaheadersize = struct.calcsize(deltaheader)
     version = '03'
     _grouplistcount = 2 # One list of manifests and one list of files
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, deltabase, cs, flags = headertuple
         return node, p1, p2, deltabase, cs, flags
 
     def _unpackmanifests(self, repo, revmap, trp, prog, numchanges):
         super(cg3unpacker, self)._unpackmanifests(repo, revmap, trp, prog,
                                                   numchanges)
         for chunkdata in iter(self.filelogheader, {}):
             # If we get here, there are directory manifests in the changegroup
             d = chunkdata["filename"]
             repo.ui.debug("adding %s revisions\n" % d)
             dirlog = repo.manifestlog._revlog.dirlog(d)
             if not dirlog.addgroup(self, revmap, trp):
                 raise error.Abort(_("received dir revlog group is empty"))
 
 class headerlessfixup(object):
     def __init__(self, fh, h):
         self._h = h
         self._fh = fh
     def read(self, n):
         if self._h:
             d, self._h = self._h[:n], self._h[n:]
             if len(d) < n:
                 d += readexactly(self._fh, n - len(d))
             return d
         return readexactly(self._fh, n)
 
 class cg1packer(object):
     deltaheader = _CHANGEGROUPV1_DELTA_HEADER
     version = '01'
     def __init__(self, repo, bundlecaps=None):
         """Given a source repo, construct a bundler.
 
         bundlecaps is optional and can be used to specify the set of
         capabilities which can be used to build the bundle.
         """
         # Set of capabilities we can use to build the bundle.
         if bundlecaps is None:
             bundlecaps = set()
         self._bundlecaps = bundlecaps
         # experimental config: bundle.reorder
         reorder = repo.ui.config('bundle', 'reorder', 'auto')
         if reorder == 'auto':
             reorder = None
         else:
             reorder = util.parsebool(reorder)
         self._repo = repo
         self._reorder = reorder
         self._progress = repo.ui.progress
         if self._repo.ui.verbose and not self._repo.ui.debugflag:
             self._verbosenote = self._repo.ui.note
         else:
             self._verbosenote = lambda s: None
 
     def close(self):
         return closechunk()
 
     def fileheader(self, fname):
         return chunkheader(len(fname)) + fname
 
     # Extracted both for clarity and for overriding in extensions.
     def _sortgroup(self, revlog, nodelist, lookup):
         """Sort nodes for change group and turn them into revnums."""
         # for generaldelta revlogs, we linearize the revs; this will both be
         # much quicker and generate a much smaller bundle
         if (revlog._generaldelta and self._reorder is None) or self._reorder:
             dag = dagutil.revlogdag(revlog)
             return dag.linearize(set(revlog.rev(n) for n in nodelist))
         else:
             return sorted([revlog.rev(n) for n in nodelist])
 
     def group(self, nodelist, revlog, lookup, units=None):
         """Calculate a delta group, yielding a sequence of changegroup chunks
         (strings).
 
         Given a list of changeset revs, return a set of deltas and
         metadata corresponding to nodes. The first delta is
         first parent(nodelist[0]) -> nodelist[0], the receiver is
         guaranteed to have this parent as it has all history before
         these changesets. In the case firstparent is nullrev the
         changegroup starts with a full revision.
 
         If units is not None, progress detail will be generated, units specifies
         the type of revlog that is touched (changelog, manifest, etc.).
         """
         # if we don't have any revisions touched by these changesets, bail
         if len(nodelist) == 0:
             yield self.close()
             return
 
         revs = self._sortgroup(revlog, nodelist, lookup)
 
         # add the parent of the first rev
         p = revlog.parentrevs(revs[0])[0]
         revs.insert(0, p)
 
         # build deltas
         total = len(revs) - 1
         msgbundling = _('bundling')
         for r in xrange(len(revs) - 1):
             if units is not None:
                 self._progress(msgbundling, r + 1, unit=units, total=total)
             prev, curr = revs[r], revs[r + 1]
             linknode = lookup(revlog.node(curr))
             for c in self.revchunk(revlog, curr, prev, linknode):
                 yield c
 
         if units is not None:
             self._progress(msgbundling, None)
         yield self.close()
 
     # filter any nodes that claim to be part of the known set
     def prune(self, revlog, missing, commonrevs):
         rr, rl = revlog.rev, revlog.linkrev
         return [n for n in missing if rl(rr(n)) not in commonrevs]
 
     def _packmanifests(self, dir, mfnodes, lookuplinknode):
         """Pack flat manifests into a changegroup stream."""
         assert not dir
         for chunk in self.group(mfnodes, self._repo.manifestlog._revlog,
                                 lookuplinknode, units=_('manifests')):
             yield chunk
 
     def _manifestsdone(self):
         return ''
 
     def generate(self, commonrevs, clnodes, fastpathlinkrev, source):
         '''yield a sequence of changegroup chunks (strings)'''
         repo = self._repo
         cl = repo.changelog
 
         clrevorder = {}
         mfs = {} # needed manifests
         fnodes = {} # needed file nodes
         changedfiles = set()
 
         # Callback for the changelog, used to collect changed files and manifest
         # nodes.
         # Returns the linkrev node (identity in the changelog case).
         def lookupcl(x):
             c = cl.read(x)
             clrevorder[x] = len(clrevorder)
             n = c[0]
             # record the first changeset introducing this manifest version
             mfs.setdefault(n, x)
             # Record a complete list of potentially-changed files in
             # this manifest.
             changedfiles.update(c[3])
             return x
 
         self._verbosenote(_('uncompressed size of bundle content:\n'))
         size = 0
         for chunk in self.group(clnodes, cl, lookupcl, units=_('changesets')):
             size += len(chunk)
             yield chunk
         self._verbosenote(_('%8.i (changelog)\n') % size)
 
         # We need to make sure that the linkrev in the changegroup refers to
         # the first changeset that introduced the manifest or file revision.
         # The fastpath is usually safer than the slowpath, because the filelogs
         # are walked in revlog order.
         #
         # When taking the slowpath with reorder=None and the manifest revlog
         # uses generaldelta, the manifest may be walked in the "wrong" order.
         # Without 'clrevorder', we would get an incorrect linkrev (see fix in
         # cc0ff93d0c0c).
         #
         # When taking the fastpath, we are only vulnerable to reordering
         # of the changelog itself. The changelog never uses generaldelta, so
         # it is only reordered when reorder=True. To handle this case, we
         # simply take the slowpath, which already has the 'clrevorder' logic.
         # This was also fixed in cc0ff93d0c0c.
         fastpathlinkrev = fastpathlinkrev and not self._reorder
         # Treemanifests don't work correctly with fastpathlinkrev
         # either, because we don't discover which directory nodes to
         # send along with files. This could probably be fixed.
         fastpathlinkrev = fastpathlinkrev and (
             'treemanifest' not in repo.requirements)
 
         for chunk in self.generatemanifests(commonrevs, clrevorder,
                 fastpathlinkrev, mfs, fnodes):
             yield chunk
         mfs.clear()
         clrevs = set(cl.rev(x) for x in clnodes)
 
         if not fastpathlinkrev:
             def linknodes(unused, fname):
                 return fnodes.get(fname, {})
         else:
             cln = cl.node
             def linknodes(filerevlog, fname):
                 llr = filerevlog.linkrev
                 fln = filerevlog.node
                 revs = ((r, llr(r)) for r in filerevlog)
                 return dict((fln(r), cln(lr)) for r, lr in revs if lr in clrevs)
 
         for chunk in self.generatefiles(changedfiles, linknodes, commonrevs,
                                         source):
             yield chunk
 
         yield self.close()
 
         if clnodes:
             repo.hook('outgoing', node=hex(clnodes[0]), source=source)
 
     def generatemanifests(self, commonrevs, clrevorder, fastpathlinkrev, mfs,
                           fnodes):
         repo = self._repo
         mfl = repo.manifestlog
         dirlog = mfl._revlog.dirlog
         tmfnodes = {'': mfs}
 
         # Callback for the manifest, used to collect linkrevs for filelog
         # revisions.
         # Returns the linkrev node (collected in lookupcl).
         def makelookupmflinknode(dir):
             if fastpathlinkrev:
                 assert not dir
                 return mfs.__getitem__
 
             def lookupmflinknode(x):
                 """Callback for looking up the linknode for manifests.
 
                 Returns the linkrev node for the specified manifest.
 
                 SIDE EFFECT:
 
                 1) fclnodes gets populated with the list of relevant
                    file nodes if we're not using fastpathlinkrev
                 2) When treemanifests are in use, collects treemanifest nodes
                    to send
 
                 Note that this means manifests must be completely sent to
                 the client before you can trust the list of files and
                 treemanifests to send.
                 """
                 clnode = tmfnodes[dir][x]
                 mdata = mfl.get(dir, x).readfast(shallow=True)
                 for p, n, fl in mdata.iterentries():
                     if fl == 't': # subdirectory manifest
                         subdir = dir + p + '/'
                         tmfclnodes = tmfnodes.setdefault(subdir, {})
                         tmfclnode = tmfclnodes.setdefault(n, clnode)
                         if clrevorder[clnode] < clrevorder[tmfclnode]:
                             tmfclnodes[n] = clnode
                     else:
                         f = dir + p
                         fclnodes = fnodes.setdefault(f, {})
                         fclnode = fclnodes.setdefault(n, clnode)
                         if clrevorder[clnode] < clrevorder[fclnode]:
                             fclnodes[n] = clnode
                 return clnode
             return lookupmflinknode
 
         size = 0
         while tmfnodes:
             dir = min(tmfnodes)
             nodes = tmfnodes[dir]
             prunednodes = self.prune(dirlog(dir), nodes, commonrevs)
             if not dir or prunednodes:
                 for x in self._packmanifests(dir, prunednodes,
                                              makelookupmflinknode(dir)):
                     size += len(x)
                     yield x
             del tmfnodes[dir]
         self._verbosenote(_('%8.i (manifests)\n') % size)
         yield self._manifestsdone()
 
     # The 'source' parameter is useful for extensions
     def generatefiles(self, changedfiles, linknodes, commonrevs, source):
         repo = self._repo
         progress = self._progress
         msgbundling = _('bundling')
 
         total = len(changedfiles)
         # for progress output
         msgfiles = _('files')
         for i, fname in enumerate(sorted(changedfiles)):
             filerevlog = repo.file(fname)
             if not filerevlog:
                 raise error.Abort(_("empty or missing revlog for %s") % fname)
 
             linkrevnodes = linknodes(filerevlog, fname)
             # Lookup for filenodes, we collected the linkrev nodes above in the
             # fastpath case and with lookupmf in the slowpath case.
             def lookupfilelog(x):
                 return linkrevnodes[x]
 
             filenodes = self.prune(filerevlog, linkrevnodes, commonrevs)
             if filenodes:
                 progress(msgbundling, i + 1, item=fname, unit=msgfiles,
                          total=total)
                 h = self.fileheader(fname)
                 size = len(h)
                 yield h
                 for chunk in self.group(filenodes, filerevlog, lookupfilelog):
                     size += len(chunk)
                     yield chunk
                 self._verbosenote(_('%8.i  %s\n') % (size, fname))
         progress(msgbundling, None)
 
     def deltaparent(self, revlog, rev, p1, p2, prev):
         return prev
 
     def revchunk(self, revlog, rev, prev, linknode):
         node = revlog.node(rev)
         p1, p2 = revlog.parentrevs(rev)
         base = self.deltaparent(revlog, rev, p1, p2, prev)
 
         prefix = ''
         if revlog.iscensored(base) or revlog.iscensored(rev):
             try:
                 delta = revlog.revision(node, raw=True)
             except error.CensoredNodeError as e:
                 delta = e.tombstone
             if base == nullrev:
                 prefix = mdiff.trivialdiffheader(len(delta))
             else:
                 baselen = revlog.rawsize(base)
                 prefix = mdiff.replacediffheader(baselen, len(delta))
         elif base == nullrev:
             delta = revlog.revision(node, raw=True)
             prefix = mdiff.trivialdiffheader(len(delta))
         else:
             delta = revlog.revdiff(base, rev)
         p1n, p2n = revlog.parents(node)
         basenode = revlog.node(base)
         flags = revlog.flags(rev)
         meta = self.builddeltaheader(node, p1n, p2n, basenode, linknode, flags)
         meta += prefix
         l = len(meta) + len(delta)
         yield chunkheader(l)
         yield meta
         yield delta
     def builddeltaheader(self, node, p1n, p2n, basenode, linknode, flags):
         # do nothing with basenode, it is implicitly the previous one in HG10
         # do nothing with flags, it is implicitly 0 for cg1 and cg2
         return struct.pack(self.deltaheader, node, p1n, p2n, linknode)
 
 class cg2packer(cg1packer):
     version = '02'
     deltaheader = _CHANGEGROUPV2_DELTA_HEADER
 
     def __init__(self, repo, bundlecaps=None):
         super(cg2packer, self).__init__(repo, bundlecaps)
         if self._reorder is None:
             # Since generaldelta is directly supported by cg2, reordering
             # generally doesn't help, so we disable it by default (treating
             # bundle.reorder=auto just like bundle.reorder=False).
             self._reorder = False
 
     def deltaparent(self, revlog, rev, p1, p2, prev):
         dp = revlog.deltaparent(rev)
         if dp == nullrev and revlog.storedeltachains:
             # Avoid sending full revisions when delta parent is null. Pick prev
             # in that case. It's tempting to pick p1 in this case, as p1 will
             # be smaller in the common case. However, computing a delta against
             # p1 may require resolving the raw text of p1, which could be
             # expensive. The revlog caches should have prev cached, meaning
             # less CPU for changegroup generation. There is likely room to add
             # a flag and/or config option to control this behavior.
             return prev
         elif dp == nullrev:
             # revlog is configured to use full snapshot for a reason,
             # stick to full snapshot.
             return nullrev
         elif dp not in (p1, p2, prev):
             # Pick prev when we can't be sure remote has the base revision.
             return prev
         else:
             return dp
 
     def builddeltaheader(self, node, p1n, p2n, basenode, linknode, flags):
         # Do nothing with flags, it is implicitly 0 in cg1 and cg2
         return struct.pack(self.deltaheader, node, p1n, p2n, basenode, linknode)
 
 class cg3packer(cg2packer):
     version = '03'
     deltaheader = _CHANGEGROUPV3_DELTA_HEADER
 
     def _packmanifests(self, dir, mfnodes, lookuplinknode):
         if dir:
             yield self.fileheader(dir)
 
         dirlog = self._repo.manifestlog._revlog.dirlog(dir)
         for chunk in self.group(mfnodes, dirlog, lookuplinknode,
                                 units=_('manifests')):
             yield chunk
 
     def _manifestsdone(self):
         return self.close()
 
     def builddeltaheader(self, node, p1n, p2n, basenode, linknode, flags):
         return struct.pack(
             self.deltaheader, node, p1n, p2n, basenode, linknode, flags)
 
 _packermap = {'01': (cg1packer, cg1unpacker),
              # cg2 adds support for exchanging generaldelta
              '02': (cg2packer, cg2unpacker),
              # cg3 adds support for exchanging revlog flags and treemanifests
              '03': (cg3packer, cg3unpacker),
 }
 
 def allsupportedversions(repo):
     versions = set(_packermap.keys())
     if not (repo.ui.configbool('experimental', 'changegroup3') or
             repo.ui.configbool('experimental', 'treemanifest') or
             'treemanifest' in repo.requirements):
         versions.discard('03')
     return versions
 
 # Changegroup versions that can be applied to the repo
 def supportedincomingversions(repo):
     return allsupportedversions(repo)
 
 # Changegroup versions that can be created from the repo
 def supportedoutgoingversions(repo):
     versions = allsupportedversions(repo)
     if 'treemanifest' in repo.requirements:
         # Versions 01 and 02 support only flat manifests and it's just too
         # expensive to convert between the flat manifest and tree manifest on
         # the fly. Since tree manifests are hashed differently, all of history
         # would have to be converted. Instead, we simply don't even pretend to
         # support versions 01 and 02.
         versions.discard('01')
         versions.discard('02')
     return versions
 
 def safeversion(repo):
     # Finds the smallest version that it's safe to assume clients of the repo
     # will support. For example, all hg versions that support generaldelta also
     # support changegroup 02.
     versions = supportedoutgoingversions(repo)
     if 'generaldelta' in repo.requirements:
         versions.discard('01')
     assert versions
     return min(versions)
 
 def getbundler(version, repo, bundlecaps=None):
     assert version in supportedoutgoingversions(repo)
     return _packermap[version][0](repo, bundlecaps)
 
 def getunbundler(version, fh, alg, extras=None):
     return _packermap[version][1](fh, alg, extras=extras)
 
 def _changegroupinfo(repo, nodes, source):
     if repo.ui.verbose or source == 'bundle':
         repo.ui.status(_("%d changesets found\n") % len(nodes))
     if repo.ui.debugflag:
         repo.ui.debug("list of changesets:\n")
         for node in nodes:
             repo.ui.debug("%s\n" % hex(node))
 
 def getsubsetraw(repo, outgoing, bundler, source, fastpath=False):
     repo = repo.unfiltered()
     commonrevs = outgoing.common
     csets = outgoing.missing
     heads = outgoing.missingheads
     # We go through the fast path if we get told to, or if all (unfiltered
     # heads have been requested (since we then know there all linkrevs will
     # be pulled by the client).
     heads.sort()
     fastpathlinkrev = fastpath or (
             repo.filtername is None and heads == sorted(repo.heads()))
 
     repo.hook('preoutgoing', throw=True, source=source)
     _changegroupinfo(repo, csets, source)
     return bundler.generate(commonrevs, csets, fastpathlinkrev, source)
 
 def getsubset(repo, outgoing, bundler, source, fastpath=False):
     gengroup = getsubsetraw(repo, outgoing, bundler, source, fastpath)
     return getunbundler(bundler.version, util.chunkbuffer(gengroup), None,
                         {'clcount': len(outgoing.missing)})
 
 def changegroupsubset(repo, roots, heads, source, version='01'):
     """Compute a changegroup consisting of all the nodes that are
     descendants of any of the roots and ancestors of any of the heads.
     Return a chunkbuffer object whose read() method will return
     successive changegroup chunks.
 
     It is fairly complex as determining which filenodes and which
     manifest nodes need to be included for the changeset to be complete
     is non-trivial.
 
     Another wrinkle is doing the reverse, figuring out which changeset in
     the changegroup a particular filenode or manifestnode belongs to.
     """
     outgoing = discovery.outgoing(repo, missingroots=roots, missingheads=heads)
     bundler = getbundler(version, repo)
     return getsubset(repo, outgoing, bundler, source)
 
 def getlocalchangegroupraw(repo, source, outgoing, bundlecaps=None,
                            version='01'):
     """Like getbundle, but taking a discovery.outgoing as an argument.
 
     This is only implemented for local repos and reuses potentially
     precomputed sets in outgoing. Returns a raw changegroup generator."""
     if not outgoing.missing:
         return None
     bundler = getbundler(version, repo, bundlecaps)
     return getsubsetraw(repo, outgoing, bundler, source)
 
 def getlocalchangegroup(repo, source, outgoing, bundlecaps=None,
                         version='01'):
     """Like getbundle, but taking a discovery.outgoing as an argument.
 
     This is only implemented for local repos and reuses potentially
     precomputed sets in outgoing."""
     if not outgoing.missing:
         return None
     bundler = getbundler(version, repo, bundlecaps)
     return getsubset(repo, outgoing, bundler, source)
 
 def getchangegroup(repo, source, outgoing, bundlecaps=None,
                    version='01'):
     """Like changegroupsubset, but returns the set difference between the
     ancestors of heads and the ancestors common.
 
     If heads is None, use the local heads. If common is None, use [nullid].
 
     The nodes in common might not all be known locally due to the way the
     current discovery protocol works.
     """
     return getlocalchangegroup(repo, source, outgoing, bundlecaps=bundlecaps,
                                version=version)
 
 def changegroup(repo, basenodes, source):
     # to avoid a race we use changegroupsubset() (issue1320)
     return changegroupsubset(repo, basenodes, repo.heads(), source)
 
 def _addchangegroupfiles(repo, source, revmap, trp, expectedfiles, needfiles):
     revisions = 0
     files = 0
     for chunkdata in iter(source.filelogheader, {}):
         files += 1
         f = chunkdata["filename"]
         repo.ui.debug("adding %s revisions\n" % f)
         repo.ui.progress(_('files'), files, unit=_('files'),
                          total=expectedfiles)
         fl = repo.file(f)
         o = len(fl)
         try:
             if not fl.addgroup(source, revmap, trp):
                 raise error.Abort(_("received file revlog group is empty"))
         except error.CensoredBaseError as e:
             raise error.Abort(_("received delta base is censored: %s") % e)
         revisions += len(fl) - o
         if f in needfiles:
             needs = needfiles[f]
             for new in xrange(o, len(fl)):
                 n = fl.node(new)
                 if n in needs:
                     needs.remove(n)
                 else:
                     raise error.Abort(
                         _("received spurious file revlog entry"))
             if not needs:
                 del needfiles[f]
     repo.ui.progress(_('files'), None)
 
     for f, needs in needfiles.iteritems():
         fl = repo.file(f)
         for n in needs:
             try:
                 fl.rev(n)
             except error.LookupError:
                 raise error.Abort(
                     _('missing file data for %s:%s - run hg verify') %
                     (f, hex(n)))
 
     return revisions, files

mercurial/pure/osutil.py

@@ -1,365 +1,365 b''
 # osutil.py - pure Python version of osutil.c
 #
 #  Copyright 2009 Matt 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 ctypes
 import ctypes.util
 import os
 import socket
 import stat as statmod
 
 from . import (
     policy,
     pycompat,
 )
 
 modulepolicy = policy.policy
 policynocffi = policy.policynocffi
 
 def _mode_to_kind(mode):
     if statmod.S_ISREG(mode):
         return statmod.S_IFREG
     if statmod.S_ISDIR(mode):
         return statmod.S_IFDIR
     if statmod.S_ISLNK(mode):
         return statmod.S_IFLNK
     if statmod.S_ISBLK(mode):
         return statmod.S_IFBLK
     if statmod.S_ISCHR(mode):
         return statmod.S_IFCHR
     if statmod.S_ISFIFO(mode):
         return statmod.S_IFIFO
     if statmod.S_ISSOCK(mode):
         return statmod.S_IFSOCK
     return mode
 
 def listdirpure(path, stat=False, skip=None):
     '''listdir(path, stat=False) -> list_of_tuples
 
     Return a sorted list containing information about the entries
     in the directory.
 
     If stat is True, each element is a 3-tuple:
 
       (name, type, stat object)
 
     Otherwise, each element is a 2-tuple:
 
       (name, type)
     '''
     result = []
     prefix = path
     if not prefix.endswith(pycompat.ossep):
         prefix += pycompat.ossep
     names = os.listdir(path)
     names.sort()
     for fn in names:
         st = os.lstat(prefix + fn)
         if fn == skip and statmod.S_ISDIR(st.st_mode):
             return []
         if stat:
             result.append((fn, _mode_to_kind(st.st_mode), st))
         else:
             result.append((fn, _mode_to_kind(st.st_mode)))
     return result
 
 ffi = None
 if modulepolicy not in policynocffi and pycompat.sysplatform == 'darwin':
     try:
         from _osutil_cffi import ffi, lib
     except ImportError:
         if modulepolicy == 'cffi': # strict cffi import
             raise
 
 if pycompat.sysplatform == 'darwin' and ffi is not None:
     listdir_batch_size = 4096
     # tweakable number, only affects performance, which chunks
     # of bytes do we get back from getattrlistbulk
 
     attrkinds = [None] * 20 # we need the max no for enum VXXX, 20 is plenty
 
     attrkinds[lib.VREG] = statmod.S_IFREG
     attrkinds[lib.VDIR] = statmod.S_IFDIR
     attrkinds[lib.VLNK] = statmod.S_IFLNK
     attrkinds[lib.VBLK] = statmod.S_IFBLK
     attrkinds[lib.VCHR] = statmod.S_IFCHR
     attrkinds[lib.VFIFO] = statmod.S_IFIFO
     attrkinds[lib.VSOCK] = statmod.S_IFSOCK
 
     class stat_res(object):
         def __init__(self, st_mode, st_mtime, st_size):
             self.st_mode = st_mode
             self.st_mtime = st_mtime
             self.st_size = st_size
 
     tv_sec_ofs = ffi.offsetof("struct timespec", "tv_sec")
     buf = ffi.new("char[]", listdir_batch_size)
 
     def listdirinternal(dfd, req, stat, skip):
         ret = []
         while True:
             r = lib.getattrlistbulk(dfd, req, buf, listdir_batch_size, 0)
             if r == 0:
                 break
             if r == -1:
                 raise OSError(ffi.errno, os.strerror(ffi.errno))
             cur = ffi.cast("val_attrs_t*", buf)
             for i in range(r):
                 lgt = cur.length
                 assert lgt == ffi.cast('uint32_t*', cur)[0]
                 ofs = cur.name_info.attr_dataoffset
                 str_lgt = cur.name_info.attr_length
                 base_ofs = ffi.offsetof('val_attrs_t', 'name_info')
                 name = str(ffi.buffer(ffi.cast("char*", cur) + base_ofs + ofs,
                            str_lgt - 1))
                 tp = attrkinds[cur.obj_type]
                 if name == "." or name == "..":
                     continue
                 if skip == name and tp == statmod.S_ISDIR:
                     return []
                 if stat:
                     mtime = cur.mtime.tv_sec
                     mode = (cur.accessmask & ~lib.S_IFMT)| tp
                     ret.append((name, tp, stat_res(st_mode=mode, st_mtime=mtime,
                                 st_size=cur.datalength)))
                 else:
                     ret.append((name, tp))
                 cur = ffi.cast("val_attrs_t*", int(ffi.cast("intptr_t", cur))
                     + lgt)
         return ret
 
     def listdir(path, stat=False, skip=None):
         req = ffi.new("struct attrlist*")
         req.bitmapcount = lib.ATTR_BIT_MAP_COUNT
         req.commonattr = (lib.ATTR_CMN_RETURNED_ATTRS |
                           lib.ATTR_CMN_NAME |
                           lib.ATTR_CMN_OBJTYPE |
                           lib.ATTR_CMN_ACCESSMASK |
                           lib.ATTR_CMN_MODTIME)
         req.fileattr = lib.ATTR_FILE_DATALENGTH
         dfd = lib.open(path, lib.O_RDONLY, 0)
         if dfd == -1:
             raise OSError(ffi.errno, os.strerror(ffi.errno))
 
         try:
             ret = listdirinternal(dfd, req, stat, skip)
         finally:
             try:
                 lib.close(dfd)
             except BaseException:
                 pass # we ignore all the errors from closing, not
                 # much we can do about that
         return ret
 else:
     listdir = listdirpure
 
 if pycompat.osname != 'nt':
     posixfile = open
 
     _SCM_RIGHTS = 0x01
     _socklen_t = ctypes.c_uint
 
     if pycompat.sysplatform.startswith('linux'):
         # socket.h says "the type should be socklen_t but the definition of
         # the kernel is incompatible with this."
         _cmsg_len_t = ctypes.c_size_t
         _msg_controllen_t = ctypes.c_size_t
         _msg_iovlen_t = ctypes.c_size_t
     else:
         _cmsg_len_t = _socklen_t
         _msg_controllen_t = _socklen_t
         _msg_iovlen_t = ctypes.c_int
 
     class _iovec(ctypes.Structure):
         _fields_ = [
             (u'iov_base', ctypes.c_void_p),
             (u'iov_len', ctypes.c_size_t),
         ]
 
     class _msghdr(ctypes.Structure):
         _fields_ = [
             (u'msg_name', ctypes.c_void_p),
             (u'msg_namelen', _socklen_t),
             (u'msg_iov', ctypes.POINTER(_iovec)),
             (u'msg_iovlen', _msg_iovlen_t),
             (u'msg_control', ctypes.c_void_p),
             (u'msg_controllen', _msg_controllen_t),
             (u'msg_flags', ctypes.c_int),
         ]
 
     class _cmsghdr(ctypes.Structure):
         _fields_ = [
             (u'cmsg_len', _cmsg_len_t),
             (u'cmsg_level', ctypes.c_int),
             (u'cmsg_type', ctypes.c_int),
             (u'cmsg_data', ctypes.c_ubyte * 0),
         ]
 
     _libc = ctypes.CDLL(ctypes.util.find_library(u'c'), use_errno=True)
     _recvmsg = getattr(_libc, 'recvmsg', None)
     if _recvmsg:
         _recvmsg.restype = getattr(ctypes, 'c_ssize_t', ctypes.c_long)
         _recvmsg.argtypes = (ctypes.c_int, ctypes.POINTER(_msghdr),
                              ctypes.c_int)
     else:
         # recvmsg isn't always provided by libc; such systems are unsupported
         def _recvmsg(sockfd, msg, flags):
             raise NotImplementedError('unsupported platform')
 
     def _CMSG_FIRSTHDR(msgh):
         if msgh.msg_controllen < ctypes.sizeof(_cmsghdr):
             return
         cmsgptr = ctypes.cast(msgh.msg_control, ctypes.POINTER(_cmsghdr))
         return cmsgptr.contents
 
     # The pure version is less portable than the native version because the
     # handling of socket ancillary data heavily depends on C preprocessor.
     # Also, some length fields are wrongly typed in Linux kernel.
     def recvfds(sockfd):
         """receive list of file descriptors via socket"""
         dummy = (ctypes.c_ubyte * 1)()
         iov = _iovec(ctypes.cast(dummy, ctypes.c_void_p), ctypes.sizeof(dummy))
         cbuf = ctypes.create_string_buffer(256)
         msgh = _msghdr(None, 0,
                        ctypes.pointer(iov), 1,
                        ctypes.cast(cbuf, ctypes.c_void_p), ctypes.sizeof(cbuf),
                        0)
         r = _recvmsg(sockfd, ctypes.byref(msgh), 0)
         if r < 0:
             e = ctypes.get_errno()
             raise OSError(e, os.strerror(e))
         # assumes that the first cmsg has fds because it isn't easy to write
         # portable CMSG_NXTHDR() with ctypes.
         cmsg = _CMSG_FIRSTHDR(msgh)
         if not cmsg:
             return []
         if (cmsg.cmsg_level != socket.SOL_SOCKET or
             cmsg.cmsg_type != _SCM_RIGHTS):
             return []
         rfds = ctypes.cast(cmsg.cmsg_data, ctypes.POINTER(ctypes.c_int))
         rfdscount = ((cmsg.cmsg_len - _cmsghdr.cmsg_data.offset) /
                      ctypes.sizeof(ctypes.c_int))
         return [rfds[i] for i in xrange(rfdscount)]
 
 else:
     import msvcrt
 
     _kernel32 = ctypes.windll.kernel32
 
     _DWORD = ctypes.c_ulong
     _LPCSTR = _LPSTR = ctypes.c_char_p
     _HANDLE = ctypes.c_void_p
 
     _INVALID_HANDLE_VALUE = _HANDLE(-1).value
 
     # CreateFile
     _FILE_SHARE_READ = 0x00000001
     _FILE_SHARE_WRITE = 0x00000002
     _FILE_SHARE_DELETE = 0x00000004
 
     _CREATE_ALWAYS = 2
     _OPEN_EXISTING = 3
     _OPEN_ALWAYS = 4
 
     _GENERIC_READ = 0x80000000
     _GENERIC_WRITE = 0x40000000
 
     _FILE_ATTRIBUTE_NORMAL = 0x80
 
     # open_osfhandle flags
     _O_RDONLY = 0x0000
     _O_RDWR = 0x0002
     _O_APPEND = 0x0008
 
     _O_TEXT = 0x4000
     _O_BINARY = 0x8000
 
     # types of parameters of C functions used (required by pypy)
 
     _kernel32.CreateFileA.argtypes = [_LPCSTR, _DWORD, _DWORD, ctypes.c_void_p,
         _DWORD, _DWORD, _HANDLE]
     _kernel32.CreateFileA.restype = _HANDLE
 
     def _raiseioerror(name):
         err = ctypes.WinError()
         raise IOError(err.errno, '%s: %s' % (name, err.strerror))
 
     class posixfile(object):
         '''a file object aiming for POSIX-like semantics
 
         CPython's open() returns a file that was opened *without* setting the
         _FILE_SHARE_DELETE flag, which causes rename and unlink to abort.
         This even happens if any hardlinked copy of the file is in open state.
         We set _FILE_SHARE_DELETE here, so files opened with posixfile can be
         renamed and deleted while they are held open.
         Note that if a file opened with posixfile is unlinked, the file
         remains but cannot be opened again or be recreated under the same name,
         until all reading processes have closed the file.'''
 
         def __init__(self, name, mode='r', bufsize=-1):
             if 'b' in mode:
                 flags = _O_BINARY
             else:
                 flags = _O_TEXT
 
             m0 = mode[0]
             if m0 == 'r' and '+' not in mode:
                 flags |= _O_RDONLY
                 access = _GENERIC_READ
             else:
                 # work around http://support.microsoft.com/kb/899149 and
                 # set _O_RDWR for 'w' and 'a', even if mode has no '+'
                 flags |= _O_RDWR
                 access = _GENERIC_READ | _GENERIC_WRITE
 
             if m0 == 'r':
                 creation = _OPEN_EXISTING
             elif m0 == 'w':
                 creation = _CREATE_ALWAYS
             elif m0 == 'a':
                 creation = _OPEN_ALWAYS
                 flags |= _O_APPEND
             else:
                 raise ValueError("invalid mode: %s" % mode)
 
             fh = _kernel32.CreateFileA(name, access,
                     _FILE_SHARE_READ | _FILE_SHARE_WRITE | _FILE_SHARE_DELETE,
                     None, creation, _FILE_ATTRIBUTE_NORMAL, None)
             if fh == _INVALID_HANDLE_VALUE:
                 _raiseioerror(name)
 
             fd = msvcrt.open_osfhandle(fh, flags)
             if fd == -1:
                 _kernel32.CloseHandle(fh)
                 _raiseioerror(name)
 
-            f = os.fdopen(fd, mode, bufsize)
+            f = os.fdopen(fd, pycompat.sysstr(mode), bufsize)
             # unfortunately, f.name is '<fdopen>' at this point -- so we store
             # the name on this wrapper. We cannot just assign to f.name,
             # because that attribute is read-only.
             object.__setattr__(self, 'name', name)
             object.__setattr__(self, '_file', f)
 
         def __iter__(self):
             return self._file
 
         def __getattr__(self, name):
             return getattr(self._file, name)
 
         def __setattr__(self, name, value):
             '''mimics the read-only attributes of Python file objects
             by raising 'TypeError: readonly attribute' if someone tries:
               f = posixfile('foo.txt')
               f.name = 'bla'  '''
             return self._file.__setattr__(name, value)
 
         def __enter__(self):
             return self._file.__enter__()
 
         def __exit__(self, exc_type, exc_value, exc_tb):
             return self._file.__exit__(exc_type, exc_value, exc_tb)

mercurial/ui.py

@@ -1,1431 +1,1431 b''
 # ui.py - user interface bits for mercurial
 #
 # Copyright 2005-2007 Matt Mackall <[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.
 
 from __future__ import absolute_import
 
 import contextlib
 import errno
 import getpass
 import inspect
 import os
 import re
 import socket
 import sys
 import tempfile
 import traceback
 
 from .i18n import _
 from .node import hex
 
 from . import (
     config,
     encoding,
     error,
     formatter,
     progress,
     pycompat,
     scmutil,
     util,
 )
 
 urlreq = util.urlreq
 
 samplehgrcs = {
     'user':
 """# example user config (see 'hg help config' for more info)
 [ui]
 # name and email, e.g.
 # username = Jane Doe <[e-mail]>
 username =
 
 [extensions]
 # uncomment these lines to enable some popular extensions
 # (see 'hg help extensions' for more info)
 #
 # pager =
 # color =""",
 
     'cloned':
 """# example repository config (see 'hg help config' for more info)
 [paths]
 default = %s
 
 # path aliases to other clones of this repo in URLs or filesystem paths
 # (see 'hg help config.paths' for more info)
 #
 # default-push = ssh://[e-mail]/hg/jdoes-fork
 # my-fork      = ssh://[e-mail]/hg/jdoes-fork
 # my-clone     = /home/jdoe/jdoes-clone
 
 [ui]
 # name and email (local to this repository, optional), e.g.
 # username = Jane Doe <[e-mail]>
 """,
 
     'local':
 """# example repository config (see 'hg help config' for more info)
 [paths]
 # path aliases to other clones of this repo in URLs or filesystem paths
 # (see 'hg help config.paths' for more info)
 #
 # default      = http://example.com/hg/example-repo
 # default-push = ssh://[e-mail]/hg/jdoes-fork
 # my-fork      = ssh://[e-mail]/hg/jdoes-fork
 # my-clone     = /home/jdoe/jdoes-clone
 
 [ui]
 # name and email (local to this repository, optional), e.g.
 # username = Jane Doe <[e-mail]>
 """,
 
     'global':
 """# example system-wide hg config (see 'hg help config' for more info)
 
 [extensions]
 # uncomment these lines to enable some popular extensions
 # (see 'hg help extensions' for more info)
 #
 # blackbox =
 # color =
 # pager =""",
 }
 
 class ui(object):
     def __init__(self, src=None):
         """Create a fresh new ui object if no src given
 
         Use uimod.ui.load() to create a ui which knows global and user configs.
         In most cases, you should use ui.copy() to create a copy of an existing
         ui object.
         """
         # _buffers: used for temporary capture of output
         self._buffers = []
         # 3-tuple describing how each buffer in the stack behaves.
         # Values are (capture stderr, capture subprocesses, apply labels).
         self._bufferstates = []
         # When a buffer is active, defines whether we are expanding labels.
         # This exists to prevent an extra list lookup.
         self._bufferapplylabels = None
         self.quiet = self.verbose = self.debugflag = self.tracebackflag = False
         self._reportuntrusted = True
         self._ocfg = config.config() # overlay
         self._tcfg = config.config() # trusted
         self._ucfg = config.config() # untrusted
         self._trustusers = set()
         self._trustgroups = set()
         self.callhooks = True
         # Insecure server connections requested.
         self.insecureconnections = False
 
         if src:
             self.fout = src.fout
             self.ferr = src.ferr
             self.fin = src.fin
 
             self._tcfg = src._tcfg.copy()
             self._ucfg = src._ucfg.copy()
             self._ocfg = src._ocfg.copy()
             self._trustusers = src._trustusers.copy()
             self._trustgroups = src._trustgroups.copy()
             self.environ = src.environ
             self.callhooks = src.callhooks
             self.insecureconnections = src.insecureconnections
             self.fixconfig()
 
             self.httppasswordmgrdb = src.httppasswordmgrdb
         else:
             self.fout = util.stdout
             self.ferr = util.stderr
             self.fin = util.stdin
 
             # shared read-only environment
             self.environ = encoding.environ
 
             self.httppasswordmgrdb = urlreq.httppasswordmgrwithdefaultrealm()
 
         allowed = self.configlist('experimental', 'exportableenviron')
         if '*' in allowed:
             self._exportableenviron = self.environ
         else:
             self._exportableenviron = {}
             for k in allowed:
                 if k in self.environ:
                     self._exportableenviron[k] = self.environ[k]
 
     @classmethod
     def load(cls):
         """Create a ui and load global and user configs"""
         u = cls()
         # we always trust global config files
         for f in scmutil.rcpath():
             u.readconfig(f, trust=True)
         return u
 
     def copy(self):
         return self.__class__(self)
 
     def resetstate(self):
         """Clear internal state that shouldn't persist across commands"""
         if self._progbar:
             self._progbar.resetstate()  # reset last-print time of progress bar
         self.httppasswordmgrdb = urlreq.httppasswordmgrwithdefaultrealm()
 
     def formatter(self, topic, opts):
         return formatter.formatter(self, topic, opts)
 
     def _trusted(self, fp, f):
         st = util.fstat(fp)
         if util.isowner(st):
             return True
 
         tusers, tgroups = self._trustusers, self._trustgroups
         if '*' in tusers or '*' in tgroups:
             return True
 
         user = util.username(st.st_uid)
         group = util.groupname(st.st_gid)
         if user in tusers or group in tgroups or user == util.username():
             return True
 
         if self._reportuntrusted:
             self.warn(_('not trusting file %s from untrusted '
                         'user %s, group %s\n') % (f, user, group))
         return False
 
     def readconfig(self, filename, root=None, trust=False,
                    sections=None, remap=None):
         try:
             fp = open(filename, u'rb')
         except IOError:
             if not sections: # ignore unless we were looking for something
                 return
             raise
 
         cfg = config.config()
         trusted = sections or trust or self._trusted(fp, filename)
 
         try:
             cfg.read(filename, fp, sections=sections, remap=remap)
             fp.close()
         except error.ConfigError as inst:
             if trusted:
                 raise
             self.warn(_("ignored: %s\n") % str(inst))
 
         if self.plain():
             for k in ('debug', 'fallbackencoding', 'quiet', 'slash',
                       'logtemplate', 'statuscopies', 'style',
                       'traceback', 'verbose'):
                 if k in cfg['ui']:
                     del cfg['ui'][k]
             for k, v in cfg.items('defaults'):
                 del cfg['defaults'][k]
         # Don't remove aliases from the configuration if in the exceptionlist
         if self.plain('alias'):
             for k, v in cfg.items('alias'):
                 del cfg['alias'][k]
         if self.plain('revsetalias'):
             for k, v in cfg.items('revsetalias'):
                 del cfg['revsetalias'][k]
         if self.plain('templatealias'):
             for k, v in cfg.items('templatealias'):
                 del cfg['templatealias'][k]
 
         if trusted:
             self._tcfg.update(cfg)
             self._tcfg.update(self._ocfg)
         self._ucfg.update(cfg)
         self._ucfg.update(self._ocfg)
 
         if root is None:
             root = os.path.expanduser('~')
         self.fixconfig(root=root)
 
     def fixconfig(self, root=None, section=None):
         if section in (None, 'paths'):
             # expand vars and ~
             # translate paths relative to root (or home) into absolute paths
             root = root or pycompat.getcwd()
             for c in self._tcfg, self._ucfg, self._ocfg:
                 for n, p in c.items('paths'):
                     # Ignore sub-options.
                     if ':' in n:
                         continue
                     if not p:
                         continue
                     if '%%' in p:
                         s = self.configsource('paths', n) or 'none'
                         self.warn(_("(deprecated '%%' in path %s=%s from %s)\n")
                                   % (n, p, s))
                         p = p.replace('%%', '%')
                     p = util.expandpath(p)
                     if not util.hasscheme(p) and not os.path.isabs(p):
                         p = os.path.normpath(os.path.join(root, p))
                     c.set("paths", n, p)
 
         if section in (None, 'ui'):
             # update ui options
             self.debugflag = self.configbool('ui', 'debug')
             self.verbose = self.debugflag or self.configbool('ui', 'verbose')
             self.quiet = not self.debugflag and self.configbool('ui', 'quiet')
             if self.verbose and self.quiet:
                 self.quiet = self.verbose = False
             self._reportuntrusted = self.debugflag or self.configbool("ui",
                 "report_untrusted", True)
             self.tracebackflag = self.configbool('ui', 'traceback', False)
 
         if section in (None, 'trusted'):
             # update trust information
             self._trustusers.update(self.configlist('trusted', 'users'))
             self._trustgroups.update(self.configlist('trusted', 'groups'))
 
     def backupconfig(self, section, item):
         return (self._ocfg.backup(section, item),
                 self._tcfg.backup(section, item),
                 self._ucfg.backup(section, item),)
     def restoreconfig(self, data):
         self._ocfg.restore(data[0])
         self._tcfg.restore(data[1])
         self._ucfg.restore(data[2])
 
     def setconfig(self, section, name, value, source=''):
         for cfg in (self._ocfg, self._tcfg, self._ucfg):
             cfg.set(section, name, value, source)
         self.fixconfig(section=section)
 
     def _data(self, untrusted):
         return untrusted and self._ucfg or self._tcfg
 
     def configsource(self, section, name, untrusted=False):
         return self._data(untrusted).source(section, name)
 
     def config(self, section, name, default=None, untrusted=False):
         if isinstance(name, list):
             alternates = name
         else:
             alternates = [name]
 
         for n in alternates:
             value = self._data(untrusted).get(section, n, None)
             if value is not None:
                 name = n
                 break
         else:
             value = default
 
         if self.debugflag and not untrusted and self._reportuntrusted:
             for n in alternates:
                 uvalue = self._ucfg.get(section, n)
                 if uvalue is not None and uvalue != value:
                     self.debug("ignoring untrusted configuration option "
                                "%s.%s = %s\n" % (section, n, uvalue))
         return value
 
     def configsuboptions(self, section, name, default=None, untrusted=False):
         """Get a config option and all sub-options.
 
         Some config options have sub-options that are declared with the
         format "key:opt = value". This method is used to return the main
         option and all its declared sub-options.
 
         Returns a 2-tuple of ``(option, sub-options)``, where `sub-options``
         is a dict of defined sub-options where keys and values are strings.
         """
         data = self._data(untrusted)
         main = data.get(section, name, default)
         if self.debugflag and not untrusted and self._reportuntrusted:
             uvalue = self._ucfg.get(section, name)
             if uvalue is not None and uvalue != main:
                 self.debug('ignoring untrusted configuration option '
                            '%s.%s = %s\n' % (section, name, uvalue))
 
         sub = {}
         prefix = '%s:' % name
         for k, v in data.items(section):
             if k.startswith(prefix):
                 sub[k[len(prefix):]] = v
 
         if self.debugflag and not untrusted and self._reportuntrusted:
             for k, v in sub.items():
                 uvalue = self._ucfg.get(section, '%s:%s' % (name, k))
                 if uvalue is not None and uvalue != v:
                     self.debug('ignoring untrusted configuration option '
                                '%s:%s.%s = %s\n' % (section, name, k, uvalue))
 
         return main, sub
 
     def configpath(self, section, name, default=None, untrusted=False):
         'get a path config item, expanded relative to repo root or config file'
         v = self.config(section, name, default, untrusted)
         if v is None:
             return None
         if not os.path.isabs(v) or "://" not in v:
             src = self.configsource(section, name, untrusted)
             if ':' in src:
                 base = os.path.dirname(src.rsplit(':')[0])
                 v = os.path.join(base, os.path.expanduser(v))
         return v
 
     def configbool(self, section, name, default=False, untrusted=False):
         """parse a configuration element as a boolean
 
         >>> u = ui(); s = 'foo'
         >>> u.setconfig(s, 'true', 'yes')
         >>> u.configbool(s, 'true')
         True
         >>> u.setconfig(s, 'false', 'no')
         >>> u.configbool(s, 'false')
         False
         >>> u.configbool(s, 'unknown')
         False
         >>> u.configbool(s, 'unknown', True)
         True
         >>> u.setconfig(s, 'invalid', 'somevalue')
         >>> u.configbool(s, 'invalid')
         Traceback (most recent call last):
             ...
         ConfigError: foo.invalid is not a boolean ('somevalue')
         """
 
         v = self.config(section, name, None, untrusted)
         if v is None:
             return default
         if isinstance(v, bool):
             return v
         b = util.parsebool(v)
         if b is None:
             raise error.ConfigError(_("%s.%s is not a boolean ('%s')")
                                     % (section, name, v))
         return b
 
     def configint(self, section, name, default=None, untrusted=False):
         """parse a configuration element as an integer
 
         >>> u = ui(); s = 'foo'
         >>> u.setconfig(s, 'int1', '42')
         >>> u.configint(s, 'int1')
         42
         >>> u.setconfig(s, 'int2', '-42')
         >>> u.configint(s, 'int2')
         -42
         >>> u.configint(s, 'unknown', 7)
         7
         >>> u.setconfig(s, 'invalid', 'somevalue')
         >>> u.configint(s, 'invalid')
         Traceback (most recent call last):
             ...
         ConfigError: foo.invalid is not an integer ('somevalue')
         """
 
         v = self.config(section, name, None, untrusted)
         if v is None:
             return default
         try:
             return int(v)
         except ValueError:
             raise error.ConfigError(_("%s.%s is not an integer ('%s')")
                                     % (section, name, v))
 
     def configbytes(self, section, name, default=0, untrusted=False):
         """parse a configuration element as a quantity in bytes
 
         Units can be specified as b (bytes), k or kb (kilobytes), m or
         mb (megabytes), g or gb (gigabytes).
 
         >>> u = ui(); s = 'foo'
         >>> u.setconfig(s, 'val1', '42')
         >>> u.configbytes(s, 'val1')
         42
         >>> u.setconfig(s, 'val2', '42.5 kb')
         >>> u.configbytes(s, 'val2')
         43520
         >>> u.configbytes(s, 'unknown', '7 MB')
         7340032
         >>> u.setconfig(s, 'invalid', 'somevalue')
         >>> u.configbytes(s, 'invalid')
         Traceback (most recent call last):
             ...
         ConfigError: foo.invalid is not a byte quantity ('somevalue')
         """
 
         value = self.config(section, name)
         if value is None:
             if not isinstance(default, str):
                 return default
             value = default
         try:
             return util.sizetoint(value)
         except error.ParseError:
             raise error.ConfigError(_("%s.%s is not a byte quantity ('%s')")
                                     % (section, name, value))
 
     def configlist(self, section, name, default=None, untrusted=False):
         """parse a configuration element as a list of comma/space separated
         strings
 
         >>> u = ui(); s = 'foo'
         >>> u.setconfig(s, 'list1', 'this,is "a small" ,test')
         >>> u.configlist(s, 'list1')
         ['this', 'is', 'a small', 'test']
         """
 
         def _parse_plain(parts, s, offset):
             whitespace = False
             while offset < len(s) and (s[offset].isspace() or s[offset] == ','):
                 whitespace = True
                 offset += 1
             if offset >= len(s):
                 return None, parts, offset
             if whitespace:
                 parts.append('')
             if s[offset] == '"' and not parts[-1]:
                 return _parse_quote, parts, offset + 1
             elif s[offset] == '"' and parts[-1][-1] == '\\':
                 parts[-1] = parts[-1][:-1] + s[offset]
                 return _parse_plain, parts, offset + 1
             parts[-1] += s[offset]
             return _parse_plain, parts, offset + 1
 
         def _parse_quote(parts, s, offset):
             if offset < len(s) and s[offset] == '"': # ""
                 parts.append('')
                 offset += 1
                 while offset < len(s) and (s[offset].isspace() or
                         s[offset] == ','):
                     offset += 1
                 return _parse_plain, parts, offset
 
             while offset < len(s) and s[offset] != '"':
                 if (s[offset] == '\\' and offset + 1 < len(s)
                         and s[offset + 1] == '"'):
                     offset += 1
                     parts[-1] += '"'
                 else:
                     parts[-1] += s[offset]
                 offset += 1
 
             if offset >= len(s):
                 real_parts = _configlist(parts[-1])
                 if not real_parts:
                     parts[-1] = '"'
                 else:
                     real_parts[0] = '"' + real_parts[0]
                     parts = parts[:-1]
                     parts.extend(real_parts)
                 return None, parts, offset
 
             offset += 1
             while offset < len(s) and s[offset] in [' ', ',']:
                 offset += 1
 
             if offset < len(s):
                 if offset + 1 == len(s) and s[offset] == '"':
                     parts[-1] += '"'
                     offset += 1
                 else:
                     parts.append('')
             else:
                 return None, parts, offset
 
             return _parse_plain, parts, offset
 
         def _configlist(s):
             s = s.rstrip(' ,')
             if not s:
                 return []
             parser, parts, offset = _parse_plain, [''], 0
             while parser:
                 parser, parts, offset = parser(parts, s, offset)
             return parts
 
         result = self.config(section, name, untrusted=untrusted)
         if result is None:
             result = default or []
         if isinstance(result, bytes):
             result = _configlist(result.lstrip(' ,\n'))
             if result is None:
                 result = default or []
         return result
 
     def hasconfig(self, section, name, untrusted=False):
         return self._data(untrusted).hasitem(section, name)
 
     def has_section(self, section, untrusted=False):
         '''tell whether section exists in config.'''
         return section in self._data(untrusted)
 
     def configitems(self, section, untrusted=False, ignoresub=False):
         items = self._data(untrusted).items(section)
         if ignoresub:
             newitems = {}
             for k, v in items:
                 if ':' not in k:
                     newitems[k] = v
             items = newitems.items()
         if self.debugflag and not untrusted and self._reportuntrusted:
             for k, v in self._ucfg.items(section):
                 if self._tcfg.get(section, k) != v:
                     self.debug("ignoring untrusted configuration option "
                                "%s.%s = %s\n" % (section, k, v))
         return items
 
     def walkconfig(self, untrusted=False):
         cfg = self._data(untrusted)
         for section in cfg.sections():
             for name, value in self.configitems(section, untrusted):
                 yield section, name, value
 
     def plain(self, feature=None):
         '''is plain mode active?
 
         Plain mode means that all configuration variables which affect
         the behavior and output of Mercurial should be
         ignored. Additionally, the output should be stable,
         reproducible and suitable for use in scripts or applications.
 
         The only way to trigger plain mode is by setting either the
         `HGPLAIN' or `HGPLAINEXCEPT' environment variables.
 
         The return value can either be
         - False if HGPLAIN is not set, or feature is in HGPLAINEXCEPT
         - True otherwise
         '''
         if ('HGPLAIN' not in encoding.environ and
                 'HGPLAINEXCEPT' not in encoding.environ):
             return False
         exceptions = encoding.environ.get('HGPLAINEXCEPT',
                 '').strip().split(',')
         if feature and exceptions:
             return feature not in exceptions
         return True
 
     def username(self):
         """Return default username to be used in commits.
 
         Searched in this order: $HGUSER, [ui] section of hgrcs, $EMAIL
         and stop searching if one of these is set.
         If not found and ui.askusername is True, ask the user, else use
         ($LOGNAME or $USER or $LNAME or $USERNAME) + "@full.hostname".
         """
         user = encoding.environ.get("HGUSER")
         if user is None:
             user = self.config("ui", ["username", "user"])
             if user is not None:
                 user = os.path.expandvars(user)
         if user is None:
             user = encoding.environ.get("EMAIL")
         if user is None and self.configbool("ui", "askusername"):
             user = self.prompt(_("enter a commit username:"), default=None)
         if user is None and not self.interactive():
             try:
                 user = '%s@%s' % (util.getuser(), socket.getfqdn())
                 self.warn(_("no username found, using '%s' instead\n") % user)
             except KeyError:
                 pass
         if not user:
             raise error.Abort(_('no username supplied'),
                              hint=_("use 'hg config --edit' "
                                     'to set your username'))
         if "\n" in user:
             raise error.Abort(_("username %s contains a newline\n")
                               % repr(user))
         return user
 
     def shortuser(self, user):
         """Return a short representation of a user name or email address."""
         if not self.verbose:
             user = util.shortuser(user)
         return user
 
     def expandpath(self, loc, default=None):
         """Return repository location relative to cwd or from [paths]"""
         try:
             p = self.paths.getpath(loc)
             if p:
                 return p.rawloc
         except error.RepoError:
             pass
 
         if default:
             try:
                 p = self.paths.getpath(default)
                 if p:
                     return p.rawloc
             except error.RepoError:
                 pass
 
         return loc
 
     @util.propertycache
     def paths(self):
         return paths(self)
 
     def pushbuffer(self, error=False, subproc=False, labeled=False):
         """install a buffer to capture standard output of the ui object
 
         If error is True, the error output will be captured too.
 
         If subproc is True, output from subprocesses (typically hooks) will be
         captured too.
 
         If labeled is True, any labels associated with buffered
         output will be handled. By default, this has no effect
         on the output returned, but extensions and GUI tools may
         handle this argument and returned styled output. If output
         is being buffered so it can be captured and parsed or
         processed, labeled should not be set to True.
         """
         self._buffers.append([])
         self._bufferstates.append((error, subproc, labeled))
         self._bufferapplylabels = labeled
 
     def popbuffer(self):
         '''pop the last buffer and return the buffered output'''
         self._bufferstates.pop()
         if self._bufferstates:
             self._bufferapplylabels = self._bufferstates[-1][2]
         else:
             self._bufferapplylabels = None
 
         return "".join(self._buffers.pop())
 
     def write(self, *args, **opts):
         '''write args to output
 
         By default, this method simply writes to the buffer or stdout,
         but extensions or GUI tools may override this method,
         write_err(), popbuffer(), and label() to style output from
         various parts of hg.
 
         An optional keyword argument, "label", can be passed in.
         This should be a string containing label names separated by
         space. Label names take the form of "topic.type". For example,
         ui.debug() issues a label of "ui.debug".
 
         When labeling output for a specific command, a label of
         "cmdname.type" is recommended. For example, status issues
         a label of "status.modified" for modified files.
         '''
         if self._buffers and not opts.get('prompt', False):
             self._buffers[-1].extend(a for a in args)
         else:
             self._progclear()
             for a in args:
                 self.fout.write(a)
 
     def write_err(self, *args, **opts):
         self._progclear()
         try:
             if self._bufferstates and self._bufferstates[-1][0]:
                 return self.write(*args, **opts)
             if not getattr(self.fout, 'closed', False):
                 self.fout.flush()
             for a in args:
                 self.ferr.write(a)
             # stderr may be buffered under win32 when redirected to files,
             # including stdout.
             if not getattr(self.ferr, 'closed', False):
                 self.ferr.flush()
         except IOError as inst:
             if inst.errno not in (errno.EPIPE, errno.EIO, errno.EBADF):
                 raise
 
     def flush(self):
         try: self.fout.flush()
         except (IOError, ValueError): pass
         try: self.ferr.flush()
         except (IOError, ValueError): pass
 
     def _isatty(self, fh):
         if self.configbool('ui', 'nontty', False):
             return False
         return util.isatty(fh)
 
     def interface(self, feature):
         """what interface to use for interactive console features?
 
         The interface is controlled by the value of `ui.interface` but also by
         the value of feature-specific configuration. For example:
 
         ui.interface.histedit = text
         ui.interface.chunkselector = curses
 
         Here the features are "histedit" and "chunkselector".
 
         The configuration above means that the default interfaces for commands
         is curses, the interface for histedit is text and the interface for
         selecting chunk is crecord (the best curses interface available).
 
         Consider the following example:
         ui.interface = curses
         ui.interface.histedit = text
 
         Then histedit will use the text interface and chunkselector will use
         the default curses interface (crecord at the moment).
         """
         alldefaults = frozenset(["text", "curses"])
 
         featureinterfaces = {
             "chunkselector": [
                 "text",
                 "curses",
             ]
         }
 
         # Feature-specific interface
         if feature not in featureinterfaces.keys():
             # Programming error, not user error
             raise ValueError("Unknown feature requested %s" % feature)
 
         availableinterfaces = frozenset(featureinterfaces[feature])
         if alldefaults > availableinterfaces:
             # Programming error, not user error. We need a use case to
             # define the right thing to do here.
             raise ValueError(
                 "Feature %s does not handle all default interfaces" %
                 feature)
 
         if self.plain():
             return "text"
 
         # Default interface for all the features
         defaultinterface = "text"
         i = self.config("ui", "interface", None)
         if i in alldefaults:
             defaultinterface = i
 
         choseninterface = defaultinterface
         f = self.config("ui", "interface.%s" % feature, None)
         if f in availableinterfaces:
             choseninterface = f
 
         if i is not None and defaultinterface != i:
             if f is not None:
                 self.warn(_("invalid value for ui.interface: %s\n") %
                           (i,))
             else:
                 self.warn(_("invalid value for ui.interface: %s (using %s)\n") %
                          (i, choseninterface))
         if f is not None and choseninterface != f:
             self.warn(_("invalid value for ui.interface.%s: %s (using %s)\n") %
                       (feature, f, choseninterface))
 
         return choseninterface
 
     def interactive(self):
         '''is interactive input allowed?
 
         An interactive session is a session where input can be reasonably read
         from `sys.stdin'. If this function returns false, any attempt to read
         from stdin should fail with an error, unless a sensible default has been
         specified.
 
         Interactiveness is triggered by the value of the `ui.interactive'
         configuration variable or - if it is unset - when `sys.stdin' points
         to a terminal device.
 
         This function refers to input only; for output, see `ui.formatted()'.
         '''
         i = self.configbool("ui", "interactive", None)
         if i is None:
             # some environments replace stdin without implementing isatty
             # usually those are non-interactive
             return self._isatty(self.fin)
 
         return i
 
     def termwidth(self):
         '''how wide is the terminal in columns?
         '''
         if 'COLUMNS' in encoding.environ:
             try:
                 return int(encoding.environ['COLUMNS'])
             except ValueError:
                 pass
         return scmutil.termsize(self)[0]
 
     def formatted(self):
         '''should formatted output be used?
 
         It is often desirable to format the output to suite the output medium.
         Examples of this are truncating long lines or colorizing messages.
         However, this is not often not desirable when piping output into other
         utilities, e.g. `grep'.
 
         Formatted output is triggered by the value of the `ui.formatted'
         configuration variable or - if it is unset - when `sys.stdout' points
         to a terminal device. Please note that `ui.formatted' should be
         considered an implementation detail; it is not intended for use outside
         Mercurial or its extensions.
 
         This function refers to output only; for input, see `ui.interactive()'.
         This function always returns false when in plain mode, see `ui.plain()'.
         '''
         if self.plain():
             return False
 
         i = self.configbool("ui", "formatted", None)
         if i is None:
             # some environments replace stdout without implementing isatty
             # usually those are non-interactive
             return self._isatty(self.fout)
 
         return i
 
     def _readline(self, prompt=''):
         if self._isatty(self.fin):
             try:
                 # magically add command line editing support, where
                 # available
                 import readline
                 # force demandimport to really load the module
                 readline.read_history_file
                 # windows sometimes raises something other than ImportError
             except Exception:
                 pass
 
         # call write() so output goes through subclassed implementation
         # e.g. color extension on Windows
         self.write(prompt, prompt=True)
 
         # instead of trying to emulate raw_input, swap (self.fin,
         # self.fout) with (sys.stdin, sys.stdout)
         oldin = sys.stdin
         oldout = sys.stdout
         sys.stdin = self.fin
         sys.stdout = self.fout
         # prompt ' ' must exist; otherwise readline may delete entire line
         # - http://bugs.python.org/issue12833
         line = raw_input(' ')
         sys.stdin = oldin
         sys.stdout = oldout
 
         # When stdin is in binary mode on Windows, it can cause
         # raw_input() to emit an extra trailing carriage return
         if os.linesep == '\r\n' and line and line[-1] == '\r':
             line = line[:-1]
         return line
 
     def prompt(self, msg, default="y"):
         """Prompt user with msg, read response.
         If ui is not interactive, the default is returned.
         """
         if not self.interactive():
             self.write(msg, ' ', default or '', "\n")
             return default
         try:
             r = self._readline(self.label(msg, 'ui.prompt'))
             if not r:
                 r = default
             if self.configbool('ui', 'promptecho'):
                 self.write(r, "\n")
             return r
         except EOFError:
             raise error.ResponseExpected()
 
     @staticmethod
     def extractchoices(prompt):
         """Extract prompt message and list of choices from specified prompt.
 
         This returns tuple "(message, choices)", and "choices" is the
         list of tuple "(response character, text without &)".
 
         >>> ui.extractchoices("awake? $$ &Yes $$ &No")
         ('awake? ', [('y', 'Yes'), ('n', 'No')])
         >>> ui.extractchoices("line\\nbreak? $$ &Yes $$ &No")
         ('line\\nbreak? ', [('y', 'Yes'), ('n', 'No')])
         >>> ui.extractchoices("want lots of $$money$$?$$Ye&s$$N&o")
         ('want lots of $$money$$?', [('s', 'Yes'), ('o', 'No')])
         """
 
         # Sadly, the prompt string may have been built with a filename
         # containing "$$" so let's try to find the first valid-looking
         # prompt to start parsing. Sadly, we also can't rely on
         # choices containing spaces, ASCII, or basically anything
         # except an ampersand followed by a character.
         m = re.match(r'(?s)(.+?)\$\$([^\$]*&[^ \$].*)', prompt)
         msg = m.group(1)
         choices = [p.strip(' ') for p in m.group(2).split('$$')]
         return (msg,
                 [(s[s.index('&') + 1].lower(), s.replace('&', '', 1))
                  for s in choices])
 
     def promptchoice(self, prompt, default=0):
         """Prompt user with a message, read response, and ensure it matches
         one of the provided choices. The prompt is formatted as follows:
 
            "would you like fries with that (Yn)? $$ &Yes $$ &No"
 
         The index of the choice is returned. Responses are case
         insensitive. If ui is not interactive, the default is
         returned.
         """
 
         msg, choices = self.extractchoices(prompt)
         resps = [r for r, t in choices]
         while True:
             r = self.prompt(msg, resps[default])
             if r.lower() in resps:
                 return resps.index(r.lower())
             self.write(_("unrecognized response\n"))
 
     def getpass(self, prompt=None, default=None):
         if not self.interactive():
             return default
         try:
             self.write_err(self.label(prompt or _('password: '), 'ui.prompt'))
             # disable getpass() only if explicitly specified. it's still valid
             # to interact with tty even if fin is not a tty.
             if self.configbool('ui', 'nontty'):
                 l = self.fin.readline()
                 if not l:
                     raise EOFError
                 return l.rstrip('\n')
             else:
                 return getpass.getpass('')
         except EOFError:
             raise error.ResponseExpected()
     def status(self, *msg, **opts):
         '''write status message to output (if ui.quiet is False)
 
         This adds an output label of "ui.status".
         '''
         if not self.quiet:
             opts['label'] = opts.get('label', '') + ' ui.status'
             self.write(*msg, **opts)
     def warn(self, *msg, **opts):
         '''write warning message to output (stderr)
 
         This adds an output label of "ui.warning".
         '''
         opts['label'] = opts.get('label', '') + ' ui.warning'
         self.write_err(*msg, **opts)
     def note(self, *msg, **opts):
         '''write note to output (if ui.verbose is True)
 
         This adds an output label of "ui.note".
         '''
         if self.verbose:
             opts['label'] = opts.get('label', '') + ' ui.note'
             self.write(*msg, **opts)
     def debug(self, *msg, **opts):
         '''write debug message to output (if ui.debugflag is True)
 
         This adds an output label of "ui.debug".
         '''
         if self.debugflag:
             opts['label'] = opts.get('label', '') + ' ui.debug'
             self.write(*msg, **opts)
 
     def edit(self, text, user, extra=None, editform=None, pending=None,
              repopath=None):
         extra_defaults = {
             'prefix': 'editor',
             'suffix': '.txt',
         }
         if extra is not None:
             extra_defaults.update(extra)
         extra = extra_defaults
 
         rdir = None
         if self.configbool('experimental', 'editortmpinhg'):
             rdir = repopath
         (fd, name) = tempfile.mkstemp(prefix='hg-' + extra['prefix'] + '-',
                                       suffix=extra['suffix'], text=True,
                                       dir=rdir)
         try:
-            f = os.fdopen(fd, "w")
+            f = os.fdopen(fd, pycompat.sysstr("w"))
             f.write(text)
             f.close()
 
             environ = {'HGUSER': user}
             if 'transplant_source' in extra:
                 environ.update({'HGREVISION': hex(extra['transplant_source'])})
             for label in ('intermediate-source', 'source', 'rebase_source'):
                 if label in extra:
                     environ.update({'HGREVISION': extra[label]})
                     break
             if editform:
                 environ.update({'HGEDITFORM': editform})
             if pending:
                 environ.update({'HG_PENDING': pending})
 
             editor = self.geteditor()
 
             self.system("%s \"%s\"" % (editor, name),
                         environ=environ,
                         onerr=error.Abort, errprefix=_("edit failed"))
 
             f = open(name)
             t = f.read()
             f.close()
         finally:
             os.unlink(name)
 
         return t
 
     def system(self, cmd, environ=None, cwd=None, onerr=None, errprefix=None):
         '''execute shell command with appropriate output stream. command
         output will be redirected if fout is not stdout.
         '''
         out = self.fout
         if any(s[1] for s in self._bufferstates):
             out = self
         return util.system(cmd, environ=environ, cwd=cwd, onerr=onerr,
                            errprefix=errprefix, out=out)
 
     def traceback(self, exc=None, force=False):
         '''print exception traceback if traceback printing enabled or forced.
         only to call in exception handler. returns true if traceback
         printed.'''
         if self.tracebackflag or force:
             if exc is None:
                 exc = sys.exc_info()
             cause = getattr(exc[1], 'cause', None)
 
             if cause is not None:
                 causetb = traceback.format_tb(cause[2])
                 exctb = traceback.format_tb(exc[2])
                 exconly = traceback.format_exception_only(cause[0], cause[1])
 
                 # exclude frame where 'exc' was chained and rethrown from exctb
                 self.write_err('Traceback (most recent call last):\n',
                                ''.join(exctb[:-1]),
                                ''.join(causetb),
                                ''.join(exconly))
             else:
                 output = traceback.format_exception(exc[0], exc[1], exc[2])
                 self.write_err(''.join(output))
         return self.tracebackflag or force
 
     def geteditor(self):
         '''return editor to use'''
         if pycompat.sysplatform == 'plan9':
             # vi is the MIPS instruction simulator on Plan 9. We
             # instead default to E to plumb commit messages to
             # avoid confusion.
             editor = 'E'
         else:
             editor = 'vi'
         return (encoding.environ.get("HGEDITOR") or
                 self.config("ui", "editor") or
                 encoding.environ.get("VISUAL") or
                 encoding.environ.get("EDITOR", editor))
 
     @util.propertycache
     def _progbar(self):
         """setup the progbar singleton to the ui object"""
         if (self.quiet or self.debugflag
                 or self.configbool('progress', 'disable', False)
                 or not progress.shouldprint(self)):
             return None
         return getprogbar(self)
 
     def _progclear(self):
         """clear progress bar output if any. use it before any output"""
         if '_progbar' not in vars(self): # nothing loaded yet
             return
         if self._progbar is not None and self._progbar.printed:
             self._progbar.clear()
 
     def progress(self, topic, pos, item="", unit="", total=None):
         '''show a progress message
 
         By default a textual progress bar will be displayed if an operation
         takes too long. 'topic' is the current operation, 'item' is a
         non-numeric marker of the current position (i.e. the currently
         in-process file), 'pos' is the current numeric position (i.e.
         revision, bytes, etc.), unit is a corresponding unit label,
         and total is the highest expected pos.
 
         Multiple nested topics may be active at a time.
 
         All topics should be marked closed by setting pos to None at
         termination.
         '''
         if self._progbar is not None:
             self._progbar.progress(topic, pos, item=item, unit=unit,
                                    total=total)
         if pos is None or not self.configbool('progress', 'debug'):
             return
 
         if unit:
             unit = ' ' + unit
         if item:
             item = ' ' + item
 
         if total:
             pct = 100.0 * pos / total
             self.debug('%s:%s %s/%s%s (%4.2f%%)\n'
                      % (topic, item, pos, total, unit, pct))
         else:
             self.debug('%s:%s %s%s\n' % (topic, item, pos, unit))
 
     def log(self, service, *msg, **opts):
         '''hook for logging facility extensions
 
         service should be a readily-identifiable subsystem, which will
         allow filtering.
 
         *msg should be a newline-terminated format string to log, and
         then any values to %-format into that format string.
 
         **opts currently has no defined meanings.
         '''
 
     def label(self, msg, label):
         '''style msg based on supplied label
 
         Like ui.write(), this just returns msg unchanged, but extensions
         and GUI tools can override it to allow styling output without
         writing it.
 
         ui.write(s, 'label') is equivalent to
         ui.write(ui.label(s, 'label')).
         '''
         return msg
 
     def develwarn(self, msg, stacklevel=1, config=None):
         """issue a developer warning message
 
         Use 'stacklevel' to report the offender some layers further up in the
         stack.
         """
         if not self.configbool('devel', 'all-warnings'):
             if config is not None and not self.configbool('devel', config):
                 return
         msg = 'devel-warn: ' + msg
         stacklevel += 1 # get in develwarn
         if self.tracebackflag:
             util.debugstacktrace(msg, stacklevel, self.ferr, self.fout)
             self.log('develwarn', '%s at:\n%s' %
                      (msg, ''.join(util.getstackframes(stacklevel))))
         else:
             curframe = inspect.currentframe()
             calframe = inspect.getouterframes(curframe, 2)
             self.write_err('%s at: %s:%s (%s)\n'
                            % ((msg,) + calframe[stacklevel][1:4]))
             self.log('develwarn', '%s at: %s:%s (%s)\n',
                      msg, *calframe[stacklevel][1:4])
             curframe = calframe = None  # avoid cycles
 
     def deprecwarn(self, msg, version):
         """issue a deprecation warning
 
         - msg: message explaining what is deprecated and how to upgrade,
         - version: last version where the API will be supported,
         """
         if not (self.configbool('devel', 'all-warnings')
                 or self.configbool('devel', 'deprec-warn')):
             return
         msg += ("\n(compatibility will be dropped after Mercurial-%s,"
                 " update your code.)") % version
         self.develwarn(msg, stacklevel=2, config='deprec-warn')
 
     def exportableenviron(self):
         """The environment variables that are safe to export, e.g. through
         hgweb.
         """
         return self._exportableenviron
 
     @contextlib.contextmanager
     def configoverride(self, overrides, source=""):
         """Context manager for temporary config overrides
         `overrides` must be a dict of the following structure:
         {(section, name) : value}"""
         backups = {}
         try:
             for (section, name), value in overrides.items():
                 backups[(section, name)] = self.backupconfig(section, name)
                 self.setconfig(section, name, value, source)
             yield
         finally:
             for __, backup in backups.items():
                 self.restoreconfig(backup)
             # just restoring ui.quiet config to the previous value is not enough
             # as it does not update ui.quiet class member
             if ('ui', 'quiet') in overrides:
                 self.fixconfig(section='ui')
 
 class paths(dict):
     """Represents a collection of paths and their configs.
 
     Data is initially derived from ui instances and the config files they have
     loaded.
     """
     def __init__(self, ui):
         dict.__init__(self)
 
         for name, loc in ui.configitems('paths', ignoresub=True):
             # No location is the same as not existing.
             if not loc:
                 continue
             loc, sub = ui.configsuboptions('paths', name)
             self[name] = path(ui, name, rawloc=loc, suboptions=sub)
 
     def getpath(self, name, default=None):
         """Return a ``path`` from a string, falling back to default.
 
         ``name`` can be a named path or locations. Locations are filesystem
         paths or URIs.
 
         Returns None if ``name`` is not a registered path, a URI, or a local
         path to a repo.
         """
         # Only fall back to default if no path was requested.
         if name is None:
             if not default:
                 default = ()
             elif not isinstance(default, (tuple, list)):
                 default = (default,)
             for k in default:
                 try:
                     return self[k]
                 except KeyError:
                     continue
             return None
 
         # Most likely empty string.
         # This may need to raise in the future.
         if not name:
             return None
 
         try:
             return self[name]
         except KeyError:
             # Try to resolve as a local path or URI.
             try:
                 # We don't pass sub-options in, so no need to pass ui instance.
                 return path(None, None, rawloc=name)
             except ValueError:
                 raise error.RepoError(_('repository %s does not exist') %
                                         name)
 
 _pathsuboptions = {}
 
 def pathsuboption(option, attr):
     """Decorator used to declare a path sub-option.
 
     Arguments are the sub-option name and the attribute it should set on
     ``path`` instances.
 
     The decorated function will receive as arguments a ``ui`` instance,
     ``path`` instance, and the string value of this option from the config.
     The function should return the value that will be set on the ``path``
     instance.
 
     This decorator can be used to perform additional verification of
     sub-options and to change the type of sub-options.
     """
     def register(func):
         _pathsuboptions[option] = (attr, func)
         return func
     return register
 
 @pathsuboption('pushurl', 'pushloc')
 def pushurlpathoption(ui, path, value):
     u = util.url(value)
     # Actually require a URL.
     if not u.scheme:
         ui.warn(_('(paths.%s:pushurl not a URL; ignoring)\n') % path.name)
         return None
 
     # Don't support the #foo syntax in the push URL to declare branch to
     # push.
     if u.fragment:
         ui.warn(_('("#fragment" in paths.%s:pushurl not supported; '
                   'ignoring)\n') % path.name)
         u.fragment = None
 
     return str(u)
 
 @pathsuboption('pushrev', 'pushrev')
 def pushrevpathoption(ui, path, value):
     return value
 
 class path(object):
     """Represents an individual path and its configuration."""
 
     def __init__(self, ui, name, rawloc=None, suboptions=None):
         """Construct a path from its config options.
 
         ``ui`` is the ``ui`` instance the path is coming from.
         ``name`` is the symbolic name of the path.
         ``rawloc`` is the raw location, as defined in the config.
         ``pushloc`` is the raw locations pushes should be made to.
 
         If ``name`` is not defined, we require that the location be a) a local
         filesystem path with a .hg directory or b) a URL. If not,
         ``ValueError`` is raised.
         """
         if not rawloc:
             raise ValueError('rawloc must be defined')
 
         # Locations may define branches via syntax <base>#<branch>.
         u = util.url(rawloc)
         branch = None
         if u.fragment:
             branch = u.fragment
             u.fragment = None
 
         self.url = u
         self.branch = branch
 
         self.name = name
         self.rawloc = rawloc
         self.loc = str(u)
 
         # When given a raw location but not a symbolic name, validate the
         # location is valid.
         if not name and not u.scheme and not self._isvalidlocalpath(self.loc):
             raise ValueError('location is not a URL or path to a local '
                              'repo: %s' % rawloc)
 
         suboptions = suboptions or {}
 
         # Now process the sub-options. If a sub-option is registered, its
         # attribute will always be present. The value will be None if there
         # was no valid sub-option.
         for suboption, (attr, func) in _pathsuboptions.iteritems():
             if suboption not in suboptions:
                 setattr(self, attr, None)
                 continue
 
             value = func(ui, self, suboptions[suboption])
             setattr(self, attr, value)
 
     def _isvalidlocalpath(self, path):
         """Returns True if the given path is a potentially valid repository.
         This is its own function so that extensions can change the definition of
         'valid' in this case (like when pulling from a git repo into a hg
         one)."""
         return os.path.isdir(os.path.join(path, '.hg'))
 
     @property
     def suboptions(self):
         """Return sub-options and their values for this path.
 
         This is intended to be used for presentation purposes.
         """
         d = {}
         for subopt, (attr, _func) in _pathsuboptions.iteritems():
             value = getattr(self, attr)
             if value is not None:
                 d[subopt] = value
         return d
 
 # we instantiate one globally shared progress bar to avoid
 # competing progress bars when multiple UI objects get created
 _progresssingleton = None
 
 def getprogbar(ui):
     global _progresssingleton
     if _progresssingleton is None:
         # passing 'ui' object to the singleton is fishy,
         # this is how the extension used to work but feel free to rework it.
         _progresssingleton = progress.progbar(ui)
     return _progresssingleton

mercurial/util.py

@@ -1,3551 +1,3551 b''
 # util.py - Mercurial utility functions and platform specific implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Matt 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
 
 import bz2
 import calendar
 import collections
 import datetime
 import errno
 import gc
 import hashlib
 import imp
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import signal
 import socket
 import stat
 import string
 import subprocess
 import sys
 import tempfile
 import textwrap
 import time
 import traceback
 import zlib
 
 from . import (
     encoding,
     error,
     i18n,
     osutil,
     parsers,
     pycompat,
 )
 
 empty = pycompat.empty
 httplib = pycompat.httplib
 httpserver = pycompat.httpserver
 pickle = pycompat.pickle
 queue = pycompat.queue
 socketserver = pycompat.socketserver
 stderr = pycompat.stderr
 stdin = pycompat.stdin
 stdout = pycompat.stdout
 stringio = pycompat.stringio
 urlerr = pycompat.urlerr
 urlparse = pycompat.urlparse
 urlreq = pycompat.urlreq
 xmlrpclib = pycompat.xmlrpclib
 
 def isatty(fp):
     try:
         return fp.isatty()
     except AttributeError:
         return False
 
 # glibc determines buffering on first write to stdout - if we replace a TTY
 # destined stdout with a pipe destined stdout (e.g. pager), we want line
 # buffering
 if isatty(stdout):
     stdout = os.fdopen(stdout.fileno(), 'wb', 1)
 
 if pycompat.osname == 'nt':
     from . import windows as platform
     stdout = platform.winstdout(stdout)
 else:
     from . import posix as platform
 
 _ = i18n._
 
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 executablepath = platform.executablepath
 expandglobs = platform.expandglobs
 explainexit = platform.explainexit
 findexe = platform.findexe
 gethgcmd = platform.gethgcmd
 getuser = platform.getuser
 getpid = os.getpid
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 hidewindow = platform.hidewindow
 isexec = platform.isexec
 isowner = platform.isowner
 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
 popen = platform.popen
 posixfile = platform.posixfile
 quotecommand = platform.quotecommand
 readpipe = platform.readpipe
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setbinary = platform.setbinary
 setflags = platform.setflags
 setsignalhandler = platform.setsignalhandler
 shellquote = platform.shellquote
 spawndetached = platform.spawndetached
 split = platform.split
 sshargs = platform.sshargs
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 testpid = platform.testpid
 umask = platform.umask
 unlink = platform.unlink
 unlinkpath = platform.unlinkpath
 username = platform.username
 
 # Python compatibility
 
 _notset = object()
 
 # disable Python's problematic floating point timestamps (issue4836)
 # (Python hypocritically says you shouldn't change this behavior in
 # libraries, and sure enough Mercurial is not a library.)
 os.stat_float_times(False)
 
 def safehasattr(thing, attr):
     return getattr(thing, attr, _notset) is not _notset
 
 def bitsfrom(container):
     bits = 0
     for bit in container:
         bits |= bit
     return bits
 
 DIGESTS = {
     'md5': hashlib.md5,
     'sha1': hashlib.sha1,
     'sha512': hashlib.sha512,
 }
 # List of digest types from strongest to weakest
 DIGESTS_BY_STRENGTH = ['sha512', 'sha1', '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(['md5', 'sha1'])
     >>> d.update('foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d['md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d['sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred(['md5', 'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise Abort(_('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 Abort(_('unknown digest type: %s') % k)
         return self._hashes[key].hexdigest()
 
     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 Abort(_('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 Abort(_('%s mismatch: expected %s, got %s') %
                     (k, v, self._digester[k]))
 
 try:
     buffer = buffer
 except NameError:
     if not pycompat.ispy3:
         def buffer(sliceable, offset=0, length=None):
             if length is not None:
                 return sliceable[offset:offset + length]
             return sliceable[offset:]
     else:
         def buffer(sliceable, offset=0, length=None):
             if length is not None:
                 return memoryview(sliceable)[offset:offset + length]
             return memoryview(sliceable)[offset:]
 
 closefds = pycompat.osname == 'posix'
 
 _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 __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 readline(self, *args, **kwargs):
         if 1 < len(self._buffer):
             # this should not happen because both read and readline end with a
             # _frombuffer call that collapse it.
             self._buffer = [''.join(self._buffer)]
             self._lenbuf = len(self._buffer[0])
         lfi = -1
         if self._buffer:
             lfi = self._buffer[-1].find('\n')
         while (not self._eof) and lfi < 0:
             self._fillbuffer()
             if self._buffer:
                 lfi = self._buffer[-1].find('\n')
         size = lfi + 1
         if lfi < 0: # end of file
             size = self._lenbuf
         elif 1 < len(self._buffer):
             # 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 ''
         buf = self._buffer[0]
         if 1 < len(self._buffer):
             buf = ''.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):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), _chunksize)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
 def popen2(cmd, env=None, newlines=False):
     # Setting bufsize to -1 lets the system decide the buffer size.
     # The default for bufsize is 0, meaning unbuffered. This leads to
     # poor performance on Mac OS X: http://bugs.python.org/issue4194
     p = subprocess.Popen(cmd, shell=True, bufsize=-1,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout
 
 def popen3(cmd, env=None, newlines=False):
     stdin, stdout, stderr, p = popen4(cmd, env, newlines)
     return stdin, stdout, stderr
 
 def popen4(cmd, env=None, newlines=False, bufsize=-1):
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          stderr=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout, p.stderr, p
 
 def version():
     """Return version information if available."""
     try:
         from . import __version__
         return __version__.version
     except ImportError:
         return '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 = '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('3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = '3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = '3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = '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')
     """
     if not v:
         v = version()
     parts = remod.split('[\+-]', v, 1)
     if len(parts) == 1:
         vparts, extra = parts[0], None
     else:
         vparts, extra = parts
 
     vints = []
     for i in vparts.split('.'):
         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)
 
 # used by parsedate
 defaultdateformats = (
     '%Y-%m-%dT%H:%M:%S', # the 'real' ISO8601
     '%Y-%m-%dT%H:%M',    #   without seconds
     '%Y-%m-%dT%H%M%S',   # another awful but legal variant without :
     '%Y-%m-%dT%H%M',     #   without seconds
     '%Y-%m-%d %H:%M:%S', # our common legal variant
     '%Y-%m-%d %H:%M',    #   without seconds
     '%Y-%m-%d %H%M%S',   # without :
     '%Y-%m-%d %H%M',     #   without seconds
     '%Y-%m-%d %I:%M:%S%p',
     '%Y-%m-%d %H:%M',
     '%Y-%m-%d %I:%M%p',
     '%Y-%m-%d',
     '%m-%d',
     '%m/%d',
     '%m/%d/%y',
     '%m/%d/%Y',
     '%a %b %d %H:%M:%S %Y',
     '%a %b %d %I:%M:%S%p %Y',
     '%a, %d %b %Y %H:%M:%S',        #  GNU coreutils "/bin/date --rfc-2822"
     '%b %d %H:%M:%S %Y',
     '%b %d %I:%M:%S%p %Y',
     '%b %d %H:%M:%S',
     '%b %d %I:%M:%S%p',
     '%b %d %H:%M',
     '%b %d %I:%M%p',
     '%b %d %Y',
     '%b %d',
     '%H:%M:%S',
     '%I:%M:%S%p',
     '%H:%M',
     '%I:%M%p',
 )
 
 extendeddateformats = defaultdateformats + (
     "%Y",
     "%Y-%m",
     "%b",
     "%b %Y",
     )
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     if func.__code__.co_argcount == 0:
         cache = []
         def f():
             if len(cache) == 0:
                 cache.append(func())
             return cache[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 sortdict(dict):
     '''a simple sorted dictionary'''
     def __init__(self, data=None):
         self._list = []
         if data:
             self.update(data)
     def copy(self):
         return sortdict(self)
     def __setitem__(self, key, val):
         if key in self:
             self._list.remove(key)
         self._list.append(key)
         dict.__setitem__(self, key, val)
     def __iter__(self):
         return self._list.__iter__()
     def update(self, src):
         if isinstance(src, dict):
             src = src.iteritems()
         for k, v in src:
             self[k] = v
     def clear(self):
         dict.clear(self)
         self._list = []
     def items(self):
         return [(k, self[k]) for k in self._list]
     def __delitem__(self, key):
         dict.__delitem__(self, key)
         self._list.remove(key)
     def pop(self, key, *args, **kwargs):
         dict.pop(self, key, *args, **kwargs)
         try:
             self._list.remove(key)
         except ValueError:
             pass
     def keys(self):
         return self._list[:]
     def iterkeys(self):
         return self._list.__iter__()
     def iteritems(self):
         for k in self._list:
             yield k, self[k]
     def insert(self, index, key, val):
         self._list.insert(index, key)
         dict.__setitem__(self, key, val)
     def __repr__(self):
         if not self:
             return '%s()' % self.__class__.__name__
         return '%s(%r)' % (self.__class__.__name__, self.items())
 
 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__ = (u'next', u'prev', u'key', u'value')
 
     def __init__(self):
         self.next = None
         self.prev = None
 
         self.key = _notset
         self.value = None
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
 
 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.
     """
     def __init__(self, max):
         self._cache = {}
 
         self._head = head = _lrucachenode()
         head.prev = head
         head.next = head
         self._size = 1
         self._capacity = max
 
     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 __setitem__(self, k, v):
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             node.value = v
             self._movetohead(node)
             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:
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         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
 
     def __delitem__(self, k):
         node = self._cache.pop(k)
         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
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self._cache[k].value
         except KeyError:
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self):
         result = lrucachedict(self._capacity)
         n = self._head.prev
         # Iterate in oldest-to-newest order, so the copy has the right ordering
         for i in range(len(self._cache)):
             result[n.key] = n.value
             n = n.prev
         return result
 
     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 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 pipefilter(s, cmd):
     '''filter string S through command CMD, returning its output'''
     p = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE)
     pout, perr = p.communicate(s)
     return pout
 
 def tempfilter(s, cmd):
     '''filter string S through a pair of temporary files with CMD.
     CMD is used as a template to create the real command to be run,
     with the strings INFILE and OUTFILE replaced by the real names of
     the temporary files generated.'''
     inname, outname = None, None
     try:
         infd, inname = tempfile.mkstemp(prefix='hg-filter-in-')
-        fp = os.fdopen(infd, 'wb')
+        fp = os.fdopen(infd, pycompat.sysstr('wb'))
         fp.write(s)
         fp.close()
         outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-')
         os.close(outfd)
         cmd = cmd.replace('INFILE', inname)
         cmd = cmd.replace('OUTFILE', outname)
         code = os.system(cmd)
         if pycompat.sysplatform == 'OpenVMS' and code & 1:
             code = 0
         if code:
             raise Abort(_("command '%s' failed: %s") %
                         (cmd, explainexit(code)))
         return readfile(outname)
     finally:
         try:
             if inname:
                 os.unlink(inname)
         except OSError:
             pass
         try:
             if outname:
                 os.unlink(outname)
         except OSError:
             pass
 
 filtertable = {
     'tempfile:': tempfilter,
     'pipe:': pipefilter,
     }
 
 def filter(s, cmd):
     "filter a string through a command that transforms its input to its output"
     for name, fn in filtertable.iteritems():
         if cmd.startswith(name):
             return fn(s, cmd[len(name):].lstrip())
     return pipefilter(s, cmd)
 
 def binary(s):
     """return true if a string is binary data"""
     return bool(s and '\0' in s)
 
 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 ''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield ''.join(buf)
 
 Abort = error.Abort
 
 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.
     """
     if sys.version_info >= (2, 7):
         return func
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
     return wrapper
 
 def pathto(root, n1, n2):
     '''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 = '/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split('/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return pycompat.ossep.join((['..'] * len(a)) + b) or '.'
 
 def mainfrozen():
     """return True if we are a frozen executable.
 
     The code supports py2exe (most common, Windows only) and tools/freeze
     (portable, not much used).
     """
     return (safehasattr(sys, "frozen") or # new py2exe
             safehasattr(sys, "importers") or # old py2exe
             imp.is_frozen(u"__main__")) # tools/freeze
 
 # the location of data files matching the source code
 if mainfrozen() and getattr(sys, 'frozen', None) != 'macosx_app':
     # executable version (py2exe) doesn't support __file__
     datapath = os.path.dirname(pycompat.sysexecutable)
 else:
     datapath = os.path.dirname(__file__)
 
 if not isinstance(datapath, bytes):
     datapath = pycompat.fsencode(datapath)
 
 i18n.setdatapath(datapath)
 
 _hgexecutable = None
 
 def hgexecutable():
     """return location of the 'hg' executable.
 
     Defaults to $HG or 'hg' in the search path.
     """
     if _hgexecutable is None:
         hg = encoding.environ.get('HG')
         mainmod = sys.modules['__main__']
         if hg:
             _sethgexecutable(hg)
         elif mainfrozen():
             if getattr(sys, 'frozen', None) == 'macosx_app':
                 # Env variable set by py2app
                 _sethgexecutable(encoding.environ['EXECUTABLEPATH'])
             else:
                 _sethgexecutable(pycompat.sysexecutable)
         elif os.path.basename(getattr(mainmod, '__file__', '')) == 'hg':
             _sethgexecutable(mainmod.__file__)
         else:
             exe = findexe('hg') or os.path.basename(sys.argv[0])
             _sethgexecutable(exe)
     return _hgexecutable
 
 def _sethgexecutable(path):
     """set location of the 'hg' executable"""
     global _hgexecutable
     _hgexecutable = path
 
 def _isstdout(f):
     fileno = getattr(f, 'fileno', None)
     return fileno and fileno() == sys.__stdout__.fileno()
 
 def shellenviron(environ=None):
     """return environ with optional override, useful for shelling out"""
     def py2shell(val):
         'convert python object into string that is useful to shell'
         if val is None or val is False:
             return '0'
         if val is True:
             return '1'
         return str(val)
     env = dict(encoding.environ)
     if environ:
         env.update((k, py2shell(v)) for k, v in environ.iteritems())
     env['HG'] = hgexecutable()
     return env
 
 def system(cmd, environ=None, cwd=None, onerr=None, errprefix=None, out=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if command fails and onerr is None, return status, else raise onerr
     object as exception.
 
     if out is specified, it is assumed to be a file-like object that has a
     write() method. stdout and stderr will be redirected to out.'''
     try:
         stdout.flush()
     except Exception:
         pass
     origcmd = cmd
     cmd = quotecommand(cmd)
     if pycompat.sysplatform == 'plan9' and (sys.version_info[0] == 2
                                     and sys.version_info[1] < 7):
         # subprocess kludge to work around issues in half-baked Python
         # ports, notably bichued/python:
         if not cwd is None:
             os.chdir(cwd)
         rc = os.system(cmd)
     else:
         env = shellenviron(environ)
         if out is None or _isstdout(out):
             rc = subprocess.call(cmd, shell=True, close_fds=closefds,
                                  env=env, cwd=cwd)
         else:
             proc = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                                     env=env, cwd=cwd, stdout=subprocess.PIPE,
                                     stderr=subprocess.STDOUT)
             for line in iter(proc.stdout.readline, ''):
                 out.write(line)
             proc.wait()
             rc = proc.returncode
         if pycompat.sysplatform == 'OpenVMS' and rc & 1:
             rc = 0
     if rc and onerr:
         errmsg = '%s %s' % (os.path.basename(origcmd.split(None, 1)[0]),
                             explainexit(rc)[0])
         if errprefix:
             errmsg = '%s: %s' % (errprefix, errmsg)
         raise onerr(errmsg)
     return rc
 
 def checksignature(func):
     '''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])) == 1:
                 raise error.SignatureError
             raise
 
     return check
 
 def copyfile(src, dest, hardlink=False, copystat=False, checkambig=False):
     '''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.
     '''
     assert not (copystat and checkambig)
     oldstat = None
     if os.path.lexists(dest):
         if checkambig:
             oldstat = checkambig and filestat(dest)
         unlink(dest)
     # hardlinks are problematic on CIFS, quietly ignore this flag
     # until we find a way to work around it cleanly (issue4546)
     if False and hardlink:
         try:
             oslink(src, dest)
             return
         except (IOError, OSError):
             pass # 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
     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(dest)
                     if newstat.isambig(oldstat):
                         # stat of copied file is ambiguous to original one
                         advanced = (oldstat.stat.st_mtime + 1) & 0x7fffffff
                         os.utime(dest, (advanced, advanced))
         except shutil.Error as inst:
             raise Abort(str(inst))
 
 def copyfiles(src, dst, hardlink=None, progress=lambda t, pos: None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     if hardlink is None:
         hardlink = (os.stat(src).st_dev ==
                     os.stat(os.path.dirname(dst)).st_dev)
     if hardlink:
         topic = _('linking')
     else:
         topic = _('copying')
 
     if os.path.isdir(src):
         os.mkdir(dst)
         for name, kind in osutil.listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             def nprog(t, pos):
                 if pos is not None:
                     return progress(t, pos + num)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress=nprog)
             num += n
     else:
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         progress(topic, num)
     progress(topic, None)
 
     return hardlink, num
 
 _winreservednames = '''con prn aux nul
     com1 com2 com3 com4 com5 com6 com7 com8 com9
     lpt1 lpt2 lpt3 lpt4 lpt5 lpt6 lpt7 lpt8 lpt9'''.split()
 _winreservedchars = ':*?"<>|'
 def checkwinfilename(path):
     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("just/a/normal/path")
     >>> checkwinfilename("foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename("foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/xml.con")
     >>> checkwinfilename("foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename("foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename("foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename("../bar")
     >>> checkwinfilename("foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename("foo\\/bar")
     "directory name ends with '\\', which is invalid on Windows"
     '''
     if path.endswith('\\'):
         return _("filename ends with '\\', which is invalid on Windows")
     if '\\/' in path:
         return _("directory name ends with '\\', which is invalid on Windows")
     for n in path.replace('\\', '/').split('/'):
         if not n:
             continue
         for c in n:
             if c in _winreservedchars:
                 return _("filename contains '%s', which is reserved "
                          "on Windows") % c
             if ord(c) <= 31:
                 return _("filename contains %r, which is invalid "
                          "on Windows") % c
         base = n.split('.')[0]
         if base and base.lower() in _winreservednames:
             return _("filename contains '%s', which is reserved "
                      "on Windows") % base
         t = n[-1]
         if t in '. ' and n not in '..':
             return _("filename ends with '%s', which is not allowed "
                      "on Windows") % t
 
 if pycompat.osname == 'nt':
     checkosfilename = checkwinfilename
 else:
     checkosfilename = platform.checkosfilename
 
 def makelock(info, pathname):
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError: # no symlink in os
         pass
 
     ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL)
     os.write(ld, info)
     os.close(ld)
 
 def readlock(pathname):
     try:
         return os.readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError: # no symlink in os
         pass
     fp = posixfile(pathname)
     r = fp.read()
     fp.close()
     return r
 
 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):
     """
     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
 
 try:
     import re2
     _re2 = None
 except ImportError:
     _re2 = False
 
 class _re(object):
     def _checkre2(self):
         global _re2
         try:
             # check if match works, see issue3964
             _re2 = bool(re2.match(r'\[([^\[]+)\]', '[ui]'))
         except ImportError:
             _re2 = False
 
     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 = '(?i)' + pat
             if flags & remod.MULTILINE:
                 pat = '(?m)' + pat
             try:
                 return re2.compile(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):
     '''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 dict((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('\\','\\\\')
     pattern = remod.compile(r'([^%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 ''.join(result)
 
 def checknlink(testfile):
     '''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 = testfile + ".hgtmp1"
     if os.path.lexists(f1):
         return False
     try:
         posixfile(f1, 'w').close()
     except IOError:
         try:
             os.unlink(f1)
         except OSError:
             pass
         return False
 
     f2 = testfile + ".hgtmp2"
     fd = None
     try:
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fd = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fd is not None:
             fd.close()
         for f in (f1, f2):
             try:
                 os.unlink(f)
             except OSError:
                 pass
 
 def endswithsep(path):
     '''Check path ends with os.sep or os.altsep.'''
     return (path.endswith(pycompat.ossep)
             or pycompat.osaltsep and path.endswith(pycompat.osaltsep))
 
 def splitpath(path):
     '''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 gui():
     '''Are we running in a GUI?'''
     if pycompat.sysplatform == 'darwin':
         if 'SSH_CONNECTION' in encoding.environ:
             # handle SSH access to a box where the user is logged in
             return False
         elif getattr(osutil, 'isgui', None):
             # check if a CoreGraphics session is available
             return osutil.isgui()
         else:
             # pure build; use a safe default
             return True
     else:
         return pycompat.osname == "nt" or encoding.environ.get("DISPLAY")
 
 def mktempcopy(name, emptyok=False, createmode=None):
     """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 = tempfile.mkstemp(prefix='.%s-' % fn, 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)
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, "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, "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, path):
         try:
             self.stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             self.stat = None
 
     __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.st_ctime == old.stat.st_ctime and
                     self.stat.st_mtime == old.stat.st_mtime)
         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.st_ctime == old.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'.
         """
         advanced = (old.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
             raise
 
     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='w+b', createmode=None, checkambig=False):
         self.__name = name      # permanent name
         self._tempname = mktempcopy(name, emptyok=('w' in mode),
                                     createmode=createmode)
         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(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat(filename)
                 if newstat.isambig(oldstat):
                     # stat of changed file is ambiguous to original one
                     advanced = (oldstat.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 makedirs(name, mode=None, notindexed=False):
     """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(os.path.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):
     with open(path, 'rb') as fp:
         return fp.read()
 
 def writefile(path, text):
     with open(path, 'wb') as fp:
         fp.write(text)
 
 def appendfile(path, text):
     with open(path, '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.
         targetsize is how big a buffer to try to maintain."""
         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 ''.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 ''.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
 
 def makedate(timestamp=None):
     '''Return a unix timestamp (or the current time) as a (unixtime,
     offset) tuple based off the local timezone.'''
     if timestamp is None:
         timestamp = time.time()
     if timestamp < 0:
         hint = _("check your clock")
         raise Abort(_("negative timestamp: %d") % timestamp, hint=hint)
     delta = (datetime.datetime.utcfromtimestamp(timestamp) -
              datetime.datetime.fromtimestamp(timestamp))
     tz = delta.days * 86400 + delta.seconds
     return timestamp, tz
 
 def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'):
     """represent a (unixtime, offset) tuple as a localized time.
     unixtime is seconds since the epoch, and offset is the time zone's
     number of seconds away from UTC.
 
     >>> datestr((0, 0))
     'Thu Jan 01 00:00:00 1970 +0000'
     >>> datestr((42, 0))
     'Thu Jan 01 00:00:42 1970 +0000'
     >>> datestr((-42, 0))
     'Wed Dec 31 23:59:18 1969 +0000'
     >>> datestr((0x7fffffff, 0))
     'Tue Jan 19 03:14:07 2038 +0000'
     >>> datestr((-0x80000000, 0))
     'Fri Dec 13 20:45:52 1901 +0000'
     """
     t, tz = date or makedate()
     if "%1" in format or "%2" in format or "%z" in format:
         sign = (tz > 0) and "-" or "+"
         minutes = abs(tz) // 60
         q, r = divmod(minutes, 60)
         format = format.replace("%z", "%1%2")
         format = format.replace("%1", "%c%02d" % (sign, q))
         format = format.replace("%2", "%02d" % r)
     d = t - tz
     if d > 0x7fffffff:
         d = 0x7fffffff
     elif d < -0x80000000:
         d = -0x80000000
     # Never use time.gmtime() and datetime.datetime.fromtimestamp()
     # because they use the gmtime() system call which is buggy on Windows
     # for negative values.
     t = datetime.datetime(1970, 1, 1) + datetime.timedelta(seconds=d)
     s = t.strftime(format)
     return s
 
 def shortdate(date=None):
     """turn (timestamp, tzoff) tuple into iso 8631 date."""
     return datestr(date, format='%Y-%m-%d')
 
 def parsetimezone(s):
     """find a trailing timezone, if any, in string, and return a
        (offset, remainder) pair"""
 
     if s.endswith("GMT") or s.endswith("UTC"):
         return 0, s[:-3].rstrip()
 
     # Unix-style timezones [+-]hhmm
     if len(s) >= 5 and s[-5] in "+-" and s[-4:].isdigit():
         sign = (s[-5] == "+") and 1 or -1
         hours = int(s[-4:-2])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-5].rstrip()
 
     # ISO8601 trailing Z
     if s.endswith("Z") and s[-2:-1].isdigit():
         return 0, s[:-1]
 
     # ISO8601-style [+-]hh:mm
     if (len(s) >= 6 and s[-6] in "+-" and s[-3] == ":" and
         s[-5:-3].isdigit() and s[-2:].isdigit()):
         sign = (s[-6] == "+") and 1 or -1
         hours = int(s[-5:-3])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-6]
 
     return None, s
 
 def strdate(string, format, defaults=[]):
     """parse a localized time string and return a (unixtime, offset) tuple.
     if the string cannot be parsed, ValueError is raised."""
     # NOTE: unixtime = localunixtime + offset
     offset, date = parsetimezone(string)
 
     # add missing elements from defaults
     usenow = False # default to using biased defaults
     for part in ("S", "M", "HI", "d", "mb", "yY"): # decreasing specificity
         found = [True for p in part if ("%"+p) in format]
         if not found:
             date += "@" + defaults[part][usenow]
             format += "@%" + part[0]
         else:
             # We've found a specific time element, less specific time
             # elements are relative to today
             usenow = True
 
     timetuple = time.strptime(date, format)
     localunixtime = int(calendar.timegm(timetuple))
     if offset is None:
         # local timezone
         unixtime = int(time.mktime(timetuple))
         offset = unixtime - localunixtime
     else:
         unixtime = localunixtime + offset
     return unixtime, offset
 
 def parsedate(date, formats=None, bias=None):
     """parse a localized date/time and return a (unixtime, offset) tuple.
 
     The date may be a "unixtime offset" string or in one of the specified
     formats. If the date already is a (unixtime, offset) tuple, it is returned.
 
     >>> parsedate(' today ') == parsedate(\
                                   datetime.date.today().strftime('%b %d'))
     True
     >>> parsedate( 'yesterday ') == parsedate((datetime.date.today() -\
                                                datetime.timedelta(days=1)\
                                               ).strftime('%b %d'))
     True
     >>> now, tz = makedate()
     >>> strnow, strtz = parsedate('now')
     >>> (strnow - now) < 1
     True
     >>> tz == strtz
     True
     """
     if bias is None:
         bias = {}
     if not date:
         return 0, 0
     if isinstance(date, tuple) and len(date) == 2:
         return date
     if not formats:
         formats = defaultdateformats
     date = date.strip()
 
     if date == 'now' or date == _('now'):
         return makedate()
     if date == 'today' or date == _('today'):
         date = datetime.date.today().strftime('%b %d')
     elif date == 'yesterday' or date == _('yesterday'):
         date = (datetime.date.today() -
                 datetime.timedelta(days=1)).strftime('%b %d')
 
     try:
         when, offset = map(int, date.split(' '))
     except ValueError:
         # fill out defaults
         now = makedate()
         defaults = {}
         for part in ("d", "mb", "yY", "HI", "M", "S"):
             # this piece is for rounding the specific end of unknowns
             b = bias.get(part)
             if b is None:
                 if part[0] in "HMS":
                     b = "00"
                 else:
                     b = "0"
 
             # this piece is for matching the generic end to today's date
             n = datestr(now, "%" + part[0])
 
             defaults[part] = (b, n)
 
         for format in formats:
             try:
                 when, offset = strdate(date, format, defaults)
             except (ValueError, OverflowError):
                 pass
             else:
                 break
         else:
             raise Abort(_('invalid date: %r') % date)
     # validate explicit (probably user-specified) date and
     # time zone offset. values must fit in signed 32 bits for
     # current 32-bit linux runtimes. timezones go from UTC-12
     # to UTC+14
     if when < -0x80000000 or when > 0x7fffffff:
         raise Abort(_('date exceeds 32 bits: %d') % when)
     if offset < -50400 or offset > 43200:
         raise Abort(_('impossible time zone offset: %d') % offset)
     return when, offset
 
 def matchdate(date):
     """Return a function that matches a given date match specifier
 
     Formats include:
 
     '{date}' match a given date to the accuracy provided
 
     '<{date}' on or before a given date
 
     '>{date}' on or after a given date
 
     >>> p1 = parsedate("10:29:59")
     >>> p2 = parsedate("10:30:00")
     >>> p3 = parsedate("10:30:59")
     >>> p4 = parsedate("10:31:00")
     >>> p5 = parsedate("Sep 15 10:30:00 1999")
     >>> f = matchdate("10:30")
     >>> f(p1[0])
     False
     >>> f(p2[0])
     True
     >>> f(p3[0])
     True
     >>> f(p4[0])
     False
     >>> f(p5[0])
     False
     """
 
     def lower(date):
         d = {'mb': "1", 'd': "1"}
         return parsedate(date, extendeddateformats, d)[0]
 
     def upper(date):
         d = {'mb': "12", 'HI': "23", 'M': "59", 'S': "59"}
         for days in ("31", "30", "29"):
             try:
                 d["d"] = days
                 return parsedate(date, extendeddateformats, d)[0]
             except Abort:
                 pass
         d["d"] = "28"
         return parsedate(date, extendeddateformats, d)[0]
 
     date = date.strip()
 
     if not date:
         raise Abort(_("dates cannot consist entirely of whitespace"))
     elif date[0] == "<":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '<DATE'"))
         when = upper(date[1:])
         return lambda x: x <= when
     elif date[0] == ">":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '>DATE'"))
         when = lower(date[1:])
         return lambda x: x >= when
     elif date[0] == "-":
         try:
             days = int(date[1:])
         except ValueError:
             raise Abort(_("invalid day spec: %s") % date[1:])
         if days < 0:
             raise Abort(_("%s must be nonnegative (see 'hg help dates')")
                 % date[1:])
         when = makedate()[0] - days * 3600 * 24
         return lambda x: x >= when
     elif " to " in date:
         a, b = date.split(" to ")
         start, stop = lower(a), upper(b)
         return lambda x: x >= start and x <= stop
     else:
         start, stop = lower(date), upper(date)
         return lambda x: x >= start and x <= stop
 
 def stringmatcher(pattern, casesensitive=True):
     """
     accepts a string, possibly starting with 're:' or 'literal:' prefix.
     returns the matcher name, pattern, and matcher function.
     missing or unknown prefixes are treated as literal matches.
 
     helper for tests:
     >>> def test(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
     >>> def itest(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern, casesensitive=False)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
 
     exact matching (no prefix):
     >>> test('abcdefg', 'abc', 'def', 'abcdefg')
     ('literal', 'abcdefg', [False, False, True])
 
     regex matching ('re:' prefix)
     >>> test('re:a.+b', 'nomatch', 'fooadef', 'fooadefbar')
     ('re', 'a.+b', [False, False, True])
 
     force exact matches ('literal:' prefix)
     >>> test('literal:re:foobar', 'foobar', 're:foobar')
     ('literal', 're:foobar', [False, True])
 
     unknown prefixes are ignored and treated as literals
     >>> test('foo:bar', 'foo', 'bar', 'foo:bar')
     ('literal', 'foo:bar', [False, False, True])
 
     case insensitive regex matches
     >>> itest('re:A.+b', 'nomatch', 'fooadef', 'fooadefBar')
     ('re', 'A.+b', [False, False, True])
 
     case insensitive literal matches
     >>> itest('ABCDEFG', 'abc', 'def', 'abcdefg')
     ('literal', 'ABCDEFG', [False, False, True])
     """
     if pattern.startswith('re:'):
         pattern = pattern[3:]
         try:
             flags = 0
             if not casesensitive:
                 flags = remod.I
             regex = remod.compile(pattern, flags)
         except remod.error as e:
             raise error.ParseError(_('invalid regular expression: %s')
                                    % e)
         return 're', pattern, regex.search
     elif pattern.startswith('literal:'):
         pattern = pattern[8:]
 
     match = pattern.__eq__
 
     if not casesensitive:
         ipat = encoding.lower(pattern)
         match = lambda s: ipat == encoding.lower(s)
     return 'literal', pattern, match
 
 def shortuser(user):
     """Return a short representation of a user name or email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     f = user.find(' ')
     if f >= 0:
         user = user[:f]
     f = user.find('.')
     if f >= 0:
         user = user[:f]
     return user
 
 def emailuser(user):
     """Return the user portion of an email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     return user
 
 def email(author):
     '''get email of author.'''
     r = author.find('>')
     if r == -1:
         r = None
     return author[author.find('<') + 1:r]
 
 def ellipsis(text, maxlength=400):
     """Trim string to at most maxlength (default: 400) columns in display."""
     return encoding.trim(text, maxlength, ellipsis='...')
 
 def unitcountfn(*unittable):
     '''return a function that renders a readable count of some quantity'''
 
     def go(count):
         for multiplier, divisor, format in unittable:
             if count >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 bytecount = unitcountfn(
     (100, 1 << 30, _('%.0f GB')),
     (10, 1 << 30, _('%.1f GB')),
     (1, 1 << 30, _('%.2f GB')),
     (100, 1 << 20, _('%.0f MB')),
     (10, 1 << 20, _('%.1f MB')),
     (1, 1 << 20, _('%.2f MB')),
     (100, 1 << 10, _('%.0f KB')),
     (10, 1 << 10, _('%.1f KB')),
     (1, 1 << 10, _('%.2f KB')),
     (1, 1, _('%.0f bytes')),
     )
 
 def uirepr(s):
     # Avoid double backslash in Windows path repr()
     return repr(s).replace('\\\\', '\\')
 
 # delay import of textwrap
 def MBTextWrapper(**kwargs):
     class tw(textwrap.TextWrapper):
         """
         Extend TextWrapper for width-awareness.
 
         Neither number of 'bytes' in any encoding nor 'characters' is
         appropriate to calculate terminal columns for specified string.
 
         Original TextWrapper implementation uses built-in 'len()' directly,
         so overriding is needed to use width information of each characters.
 
         In addition, characters classified into 'ambiguous' width are
         treated as wide in East Asian area, but as narrow in other.
 
         This requires use decision to determine width of such characters.
         """
         def _cutdown(self, ucstr, space_left):
             l = 0
             colwidth = encoding.ucolwidth
             for i in xrange(len(ucstr)):
                 l += colwidth(ucstr[i])
                 if space_left < l:
                     return (ucstr[:i], ucstr[i:])
             return ucstr, ''
 
         # overriding of base class
         def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
             space_left = max(width - cur_len, 1)
 
             if self.break_long_words:
                 cut, res = self._cutdown(reversed_chunks[-1], space_left)
                 cur_line.append(cut)
                 reversed_chunks[-1] = res
             elif not cur_line:
                 cur_line.append(reversed_chunks.pop())
 
         # this overriding code is imported from TextWrapper of Python 2.6
         # to calculate columns of string by 'encoding.ucolwidth()'
         def _wrap_chunks(self, chunks):
             colwidth = encoding.ucolwidth
 
             lines = []
             if self.width <= 0:
                 raise ValueError("invalid width %r (must be > 0)" % self.width)
 
             # Arrange in reverse order so items can be efficiently popped
             # from a stack of chucks.
             chunks.reverse()
 
             while chunks:
 
                 # Start the list of chunks that will make up the current line.
                 # cur_len is just the length of all the chunks in cur_line.
                 cur_line = []
                 cur_len = 0
 
                 # Figure out which static string will prefix this line.
                 if lines:
                     indent = self.subsequent_indent
                 else:
                     indent = self.initial_indent
 
                 # Maximum width for this line.
                 width = self.width - len(indent)
 
                 # First chunk on line is whitespace -- drop it, unless this
                 # is the very beginning of the text (i.e. no lines started yet).
                 if self.drop_whitespace and chunks[-1].strip() == '' and lines:
                     del chunks[-1]
 
                 while chunks:
                     l = colwidth(chunks[-1])
 
                     # Can at least squeeze this chunk onto the current line.
                     if cur_len + l <= width:
                         cur_line.append(chunks.pop())
                         cur_len += l
 
                     # Nope, this line is full.
                     else:
                         break
 
                 # The current line is full, and the next chunk is too big to
                 # fit on *any* line (not just this one).
                 if chunks and colwidth(chunks[-1]) > width:
                     self._handle_long_word(chunks, cur_line, cur_len, width)
 
                 # If the last chunk on this line is all whitespace, drop it.
                 if (self.drop_whitespace and
                     cur_line and cur_line[-1].strip() == ''):
                     del cur_line[-1]
 
                 # Convert current line back to a string and store it in list
                 # of all lines (return value).
                 if cur_line:
                     lines.append(indent + ''.join(cur_line))
 
             return lines
 
     global MBTextWrapper
     MBTextWrapper = tw
     return tw(**kwargs)
 
 def wrap(line, width, initindent='', hangindent=''):
     maxindent = max(len(hangindent), len(initindent))
     if width <= maxindent:
         # adjust for weird terminal size
         width = max(78, maxindent + 1)
     line = line.decode(encoding.encoding, encoding.encodingmode)
     initindent = initindent.decode(encoding.encoding, encoding.encodingmode)
     hangindent = hangindent.decode(encoding.encoding, encoding.encodingmode)
     wrapper = MBTextWrapper(width=width,
                             initial_indent=initindent,
                             subsequent_indent=hangindent)
     return wrapper.fill(line).encode(encoding.encoding)
 
 if (pyplatform.python_implementation() == '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*" are ignored for now, as Python < 2.7.4 is a minority.
     #
     # 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.
     if sys.version_info >= (2, 7, 4):
         # fp.readline deals with EINTR correctly, use it as a workaround.
         def _safeiterfile(fp):
             return iter(fp.readline, '')
     else:
         # fp.read* are broken too, manually deal with EINTR in a stupid way.
         # note: this may block longer than necessary because of bufsize.
         def _safeiterfile(fp, bufsize=4096):
             fd = fp.fileno()
             line = ''
             while True:
                 try:
                     buf = os.read(fd, bufsize)
                 except OSError as ex:
                     # os.read only raises EINTR before any data is read
                     if ex.errno == errno.EINTR:
                         continue
                     else:
                         raise
                 line += buf
                 if '\n' in buf:
                     splitted = line.splitlines(True)
                     line = ''
                     for l in splitted:
                         if l[-1] == '\n':
                             yield l
                         else:
                             line = l
                 if not buf:
                     break
             if line:
                 yield line
 
     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:
             return _safeiterfile(fp)
 else:
     # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed.
     def iterfile(fp):
         return fp
 
 def iterlines(iterator):
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line
 
 def expandpath(path):
     return os.path.expanduser(os.path.expandvars(path))
 
 def hgcmd():
     """Return the command used to execute current hg
 
     This is different from hgexecutable() because on Windows we want
     to avoid things opening new shell windows like batch files, so we
     get either the python call or current executable.
     """
     if mainfrozen():
         if getattr(sys, 'frozen', None) == 'macosx_app':
             # Env variable set by py2app
             return [encoding.environ['EXECUTABLEPATH']]
         else:
             return [pycompat.sysexecutable]
     return gethgcmd()
 
 def rundetached(args, condfn):
     """Execute the argument list in a detached process.
 
     condfn is a callable which is called repeatedly and should return
     True once the child process is known to have started successfully.
     At this point, the child process PID is returned. If the child
     process fails to start or finishes before condfn() evaluates to
     True, return -1.
     """
     # Windows case is easier because the child process is either
     # successfully starting and validating the condition or exiting
     # on failure. We just poll on its PID. On Unix, if the child
     # process fails to start, it will be left in a zombie state until
     # the parent wait on it, which we cannot do since we expect a long
     # running process on success. Instead we listen for SIGCHLD telling
     # us our child process terminated.
     terminated = set()
     def handler(signum, frame):
         terminated.add(os.wait())
     prevhandler = None
     SIGCHLD = getattr(signal, 'SIGCHLD', None)
     if SIGCHLD is not None:
         prevhandler = signal.signal(SIGCHLD, handler)
     try:
         pid = spawndetached(args)
         while not condfn():
             if ((pid in terminated or not testpid(pid))
                 and not condfn()):
                 return -1
             time.sleep(0.1)
         return pid
     finally:
         if prevhandler is not None:
             signal.signal(signal.SIGCHLD, prevhandler)
 
 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 = '|'.join(mapping.keys())
     if escape_prefix:
         patterns += '|' + prefix
         if len(prefix) > 1:
             prefix_char = prefix[1:]
         else:
             prefix_char = prefix
         mapping[prefix_char] = prefix_char
     r = remod.compile(r'%s(%s)' % (prefix, patterns))
     return r.sub(lambda x: fn(mapping[x.group()[1:]]), s)
 
 def getport(port):
     """Return the port for a given network service.
 
     If port is an integer, it's returned as is. If it's a string, it's
     looked up using socket.getservbyname(). If there's no matching
     service, error.Abort is raised.
     """
     try:
         return int(port)
     except ValueError:
         pass
 
     try:
         return socket.getservbyname(port)
     except socket.error:
         raise Abort(_("no port number associated with service '%s'") % port)
 
 _booleans = {'1': True, 'yes': True, 'true': True, 'on': True, 'always': True,
              '0': False, 'no': False, 'false': False, 'off': False,
              'never': False}
 
 def parsebool(s):
     """Parse s into a boolean.
 
     If s is not a valid boolean, returns None.
     """
     return _booleans.get(s.lower(), None)
 
 _hextochr = dict((a + b, chr(int(a + b, 16)))
                  for a in string.hexdigits for b in string.hexdigits)
 
 class url(object):
     r"""Reliable URL parser.
 
     This parses URLs and provides attributes for the following
     components:
 
     <scheme>://<user>:<passwd>@<host>:<port>/<path>?<query>#<fragment>
 
     Missing components are set to None. The only exception is
     fragment, which is set to '' if present but empty.
 
     If parsefragment is False, fragment is included in query. If
     parsequery is False, query is included in path. If both are
     False, both fragment and query are included in path.
 
     See http://www.ietf.org/rfc/rfc2396.txt for more information.
 
     Note that for backward compatibility reasons, bundle URLs do not
     take host names. That means 'bundle://../' has a path of '../'.
 
     Examples:
 
     >>> url('http://www.ietf.org/rfc/rfc2396.txt')
     <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'>
     >>> url('ssh://[::1]:2200//home/joe/repo')
     <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'>
     >>> url('file:///home/joe/repo')
     <url scheme: 'file', path: '/home/joe/repo'>
     >>> url('file:///c:/temp/foo/')
     <url scheme: 'file', path: 'c:/temp/foo/'>
     >>> url('bundle:foo')
     <url scheme: 'bundle', path: 'foo'>
     >>> url('bundle://../foo')
     <url scheme: 'bundle', path: '../foo'>
     >>> url(r'c:\foo\bar')
     <url path: 'c:\\foo\\bar'>
     >>> url(r'\\blah\blah\blah')
     <url path: '\\\\blah\\blah\\blah'>
     >>> url(r'\\blah\blah\blah#baz')
     <url path: '\\\\blah\\blah\\blah', fragment: 'baz'>
     >>> url(r'file:///C:\users\me')
     <url scheme: 'file', path: 'C:\\users\\me'>
 
     Authentication credentials:
 
     >>> url('ssh://joe:xyz@x/repo')
     <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'>
     >>> url('ssh://joe@x/repo')
     <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'>
 
     Query strings and fragments:
 
     >>> url('http://host/a?b#c')
     <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
     >>> url('http://host/a?b#c', parsequery=False, parsefragment=False)
     <url scheme: 'http', host: 'host', path: 'a?b#c'>
 
     Empty path:
 
     >>> url('')
     <url path: ''>
     >>> url('#a')
     <url path: '', fragment: 'a'>
     >>> url('http://host/')
     <url scheme: 'http', host: 'host', path: ''>
     >>> url('http://host/#a')
     <url scheme: 'http', host: 'host', path: '', fragment: 'a'>
 
     Only scheme:
 
     >>> url('http:')
     <url scheme: 'http'>
     """
 
     _safechars = "!~*'()+"
     _safepchars = "/!~*'()+:\\"
     _matchscheme = remod.compile('^[a-zA-Z0-9+.\\-]+:').match
 
     def __init__(self, path, parsequery=True, parsefragment=True):
         # We slowly chomp away at path until we have only the path left
         self.scheme = self.user = self.passwd = self.host = None
         self.port = self.path = self.query = self.fragment = None
         self._localpath = True
         self._hostport = ''
         self._origpath = path
 
         if parsefragment and '#' in path:
             path, self.fragment = path.split('#', 1)
 
         # special case for Windows drive letters and UNC paths
         if hasdriveletter(path) or path.startswith('\\\\'):
             self.path = path
             return
 
         # For compatibility reasons, we can't handle bundle paths as
         # normal URLS
         if path.startswith('bundle:'):
             self.scheme = 'bundle'
             path = path[7:]
             if path.startswith('//'):
                 path = path[2:]
             self.path = path
             return
 
         if self._matchscheme(path):
             parts = path.split(':', 1)
             if parts[0]:
                 self.scheme, path = parts
                 self._localpath = False
 
         if not path:
             path = None
             if self._localpath:
                 self.path = ''
                 return
         else:
             if self._localpath:
                 self.path = path
                 return
 
             if parsequery and '?' in path:
                 path, self.query = path.split('?', 1)
                 if not path:
                     path = None
                 if not self.query:
                     self.query = None
 
             # // is required to specify a host/authority
             if path and path.startswith('//'):
                 parts = path[2:].split('/', 1)
                 if len(parts) > 1:
                     self.host, path = parts
                 else:
                     self.host = parts[0]
                     path = None
                 if not self.host:
                     self.host = None
                     # path of file:///d is /d
                     # path of file:///d:/ is d:/, not /d:/
                     if path and not hasdriveletter(path):
                         path = '/' + path
 
             if self.host and '@' in self.host:
                 self.user, self.host = self.host.rsplit('@', 1)
                 if ':' in self.user:
                     self.user, self.passwd = self.user.split(':', 1)
                 if not self.host:
                     self.host = None
 
             # Don't split on colons in IPv6 addresses without ports
             if (self.host and ':' in self.host and
                 not (self.host.startswith('[') and self.host.endswith(']'))):
                 self._hostport = self.host
                 self.host, self.port = self.host.rsplit(':', 1)
                 if not self.host:
                     self.host = None
 
             if (self.host and self.scheme == 'file' and
                 self.host not in ('localhost', '127.0.0.1', '[::1]')):
                 raise Abort(_('file:// URLs can only refer to localhost'))
 
         self.path = path
 
         # leave the query string escaped
         for a in ('user', 'passwd', 'host', 'port',
                   'path', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 setattr(self, a, pycompat.urlunquote(v))
 
     def __repr__(self):
         attrs = []
         for a in ('scheme', 'user', 'passwd', 'host', 'port', 'path',
                   'query', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 attrs.append('%s: %r' % (a, v))
         return '<url %s>' % ', '.join(attrs)
 
     def __str__(self):
         r"""Join the URL's components back into a URL string.
 
         Examples:
 
         >>> str(url('http://user:pw@host:80/c:/bob?fo:oo#ba:ar'))
         'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'
         >>> str(url('http://user:pw@host:80/?foo=bar&baz=42'))
         'http://user:pw@host:80/?foo=bar&baz=42'
         >>> str(url('http://user:pw@host:80/?foo=bar%3dbaz'))
         'http://user:pw@host:80/?foo=bar%3dbaz'
         >>> str(url('ssh://user:pw@[::1]:2200//home/joe#'))
         'ssh://user:pw@[::1]:2200//home/joe#'
         >>> str(url('http://localhost:80//'))
         'http://localhost:80//'
         >>> str(url('http://localhost:80/'))
         'http://localhost:80/'
         >>> str(url('http://localhost:80'))
         'http://localhost:80/'
         >>> str(url('bundle:foo'))
         'bundle:foo'
         >>> str(url('bundle://../foo'))
         'bundle:../foo'
         >>> str(url('path'))
         'path'
         >>> str(url('file:///tmp/foo/bar'))
         'file:///tmp/foo/bar'
         >>> str(url('file:///c:/tmp/foo/bar'))
         'file:///c:/tmp/foo/bar'
         >>> print url(r'bundle:foo\bar')
         bundle:foo\bar
         >>> print url(r'file:///D:\data\hg')
         file:///D:\data\hg
         """
         if self._localpath:
             s = self.path
             if self.scheme == 'bundle':
                 s = 'bundle:' + s
             if self.fragment:
                 s += '#' + self.fragment
             return s
 
         s = self.scheme + ':'
         if self.user or self.passwd or self.host:
             s += '//'
         elif self.scheme and (not self.path or self.path.startswith('/')
                               or hasdriveletter(self.path)):
             s += '//'
             if hasdriveletter(self.path):
                 s += '/'
         if self.user:
             s += urlreq.quote(self.user, safe=self._safechars)
         if self.passwd:
             s += ':' + urlreq.quote(self.passwd, safe=self._safechars)
         if self.user or self.passwd:
             s += '@'
         if self.host:
             if not (self.host.startswith('[') and self.host.endswith(']')):
                 s += urlreq.quote(self.host)
             else:
                 s += self.host
         if self.port:
             s += ':' + urlreq.quote(self.port)
         if self.host:
             s += '/'
         if self.path:
             # TODO: similar to the query string, we should not unescape the
             # path when we store it, the path might contain '%2f' = '/',
             # which we should *not* escape.
             s += urlreq.quote(self.path, safe=self._safepchars)
         if self.query:
             # we store the query in escaped form.
             s += '?' + self.query
         if self.fragment is not None:
             s += '#' + urlreq.quote(self.fragment, safe=self._safepchars)
         return s
 
     def authinfo(self):
         user, passwd = self.user, self.passwd
         try:
             self.user, self.passwd = None, None
             s = str(self)
         finally:
             self.user, self.passwd = user, passwd
         if not self.user:
             return (s, None)
         # authinfo[1] is passed to urllib2 password manager, and its
         # URIs must not contain credentials. The host is passed in the
         # URIs list because Python < 2.4.3 uses only that to search for
         # a password.
         return (s, (None, (s, self.host),
                     self.user, self.passwd or ''))
 
     def isabs(self):
         if self.scheme and self.scheme != 'file':
             return True # remote URL
         if hasdriveletter(self.path):
             return True # absolute for our purposes - can't be joined()
         if self.path.startswith(r'\\'):
             return True # Windows UNC path
         if self.path.startswith('/'):
             return True # POSIX-style
         return False
 
     def localpath(self):
         if self.scheme == 'file' or self.scheme == 'bundle':
             path = self.path or '/'
             # For Windows, we need to promote hosts containing drive
             # letters to paths with drive letters.
             if hasdriveletter(self._hostport):
                 path = self._hostport + '/' + self.path
             elif (self.host is not None and self.path
                   and not hasdriveletter(path)):
                 path = '/' + path
             return path
         return self._origpath
 
     def islocal(self):
         '''whether localpath will return something that posixfile can open'''
         return (not self.scheme or self.scheme == 'file'
                 or self.scheme == 'bundle')
 
 def hasscheme(path):
     return bool(url(path).scheme)
 
 def hasdriveletter(path):
     return path and path[1:2] == ':' and path[0:1].isalpha()
 
 def urllocalpath(path):
     return url(path, parsequery=False, parsefragment=False).localpath()
 
 def hidepassword(u):
     '''hide user credential in a url string'''
     u = url(u)
     if u.passwd:
         u.passwd = '***'
     return str(u)
 
 def removeauth(u):
     '''remove all authentication information from a url string'''
     u = url(u)
     u.user = u.passwd = None
     return str(u)
 
 timecount = unitcountfn(
     (1, 1e3, _('%.0f s')),
     (100, 1, _('%.1f s')),
     (10, 1, _('%.2f s')),
     (1, 1, _('%.3f s')),
     (100, 0.001, _('%.1f ms')),
     (10, 0.001, _('%.2f ms')),
     (1, 0.001, _('%.3f ms')),
     (100, 0.000001, _('%.1f us')),
     (10, 0.000001, _('%.2f us')),
     (1, 0.000001, _('%.3f us')),
     (100, 0.000000001, _('%.1f ns')),
     (10, 0.000000001, _('%.2f ns')),
     (1, 0.000000001, _('%.3f ns')),
     )
 
 _timenesting = [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):
         start = time.time()
         indent = 2
         _timenesting[0] += indent
         try:
             return func(*args, **kwargs)
         finally:
             elapsed = time.time() - start
             _timenesting[0] -= indent
             stderr.write('%s%s: %s\n' %
                          (' ' * _timenesting[0], func.__name__,
                           timecount(elapsed)))
     return wrapper
 
 _sizeunits = (('m', 2**20), ('k', 2**10), ('g', 2**30),
               ('kb', 2**10), ('mb', 2**20), ('gb', 2**30), ('b', 1))
 
 def sizetoint(s):
     '''Convert a space specifier to a byte count.
 
     >>> sizetoint('30')
     30
     >>> sizetoint('2.2kb')
     2252
     >>> sizetoint('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(_("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=' %-*s in %s\n', fileline='%s:%s'):
     '''Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last 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 % (fn, ln), func)
         for fn, ln, func, _text in traceback.extract_stack()[:-skip - 1]]
     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='stacktrace', skip=0, f=stderr, otherf=stdout):
     '''Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' last 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('%s at:\n' % msg)
     for line in getstackframes(skip + 1):
         f.write(line)
     f.flush()
 
 class dirs(object):
     '''a multiset of directory names from a dirstate or manifest'''
 
     def __init__(self, map, skip=None):
         self._dirs = {}
         addpath = self.addpath
         if safehasattr(map, 'iteritems') and skip is not None:
             for f, s in map.iteritems():
                 if s[0] != skip:
                     addpath(f)
         else:
             for f in map:
                 addpath(f)
 
     def addpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if base in dirs:
                 dirs[base] += 1
                 return
             dirs[base] = 1
 
     def delpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if dirs[base] > 1:
                 dirs[base] -= 1
                 return
             del dirs[base]
 
     def __iter__(self):
         return self._dirs.iterkeys()
 
     def __contains__(self, d):
         return d in self._dirs
 
 if safehasattr(parsers, 'dirs'):
     dirs = parsers.dirs
 
 def finddirs(path):
     pos = path.rfind('/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind('/', 0, pos)
 
 class ctxmanager(object):
     '''A context manager for use in 'with' blocks to allow multiple
     contexts to be entered at once.  This is both safer and more
     flexible than contextlib.nested.
 
     Once Mercurial supports Python 2.7+, this will become mostly
     unnecessary.
     '''
 
     def __init__(self, *args):
         '''Accepts a list of no-argument functions that return context
         managers.  These will be invoked at __call__ time.'''
         self._pending = args
         self._atexit = []
 
     def __enter__(self):
         return self
 
     def enter(self):
         '''Create and enter context managers in the order in which they were
         passed to the constructor.'''
         values = []
         for func in self._pending:
             obj = func()
             values.append(obj.__enter__())
             self._atexit.append(obj.__exit__)
         del self._pending
         return values
 
     def atexit(self, func, *args, **kwargs):
         '''Add a function to call when this context manager exits.  The
         ordering of multiple atexit calls is unspecified, save that
         they will happen before any __exit__ functions.'''
         def wrapper(exc_type, exc_val, exc_tb):
             func(*args, **kwargs)
         self._atexit.append(wrapper)
         return func
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         '''Context managers are exited in the reverse order from which
         they were created.'''
         received = exc_type is not None
         suppressed = False
         pending = None
         self._atexit.reverse()
         for exitfunc in self._atexit:
             try:
                 if exitfunc(exc_type, exc_val, exc_tb):
                     suppressed = True
                     exc_type = None
                     exc_val = None
                     exc_tb = None
             except BaseException:
                 pending = sys.exc_info()
                 exc_type, exc_val, exc_tb = pending = sys.exc_info()
         del self._atexit
         if pending:
             raise exc_val
         return received and suppressed
 
 # compression code
 
 SERVERROLE = 'server'
 CLIENTROLE = 'client'
 
 compewireprotosupport = collections.namedtuple(u'compenginewireprotosupport',
                                                (u'name', u'serverpriority',
                                                 u'clientpriority'))
 
 class compressormanager(object):
     """Holds registrations of various compression engines.
 
     This class essentially abstracts the differences between compression
     engines to allow new compression formats to be added easily, possibly from
     extensions.
 
     Compressors are registered against the global instance by calling its
     ``register()`` method.
     """
     def __init__(self):
         self._engines = {}
         # Bundle spec human name to engine name.
         self._bundlenames = {}
         # Internal bundle identifier to engine name.
         self._bundletypes = {}
         # Revlog header to engine name.
         self._revlogheaders = {}
         # Wire proto identifier to engine name.
         self._wiretypes = {}
 
     def __getitem__(self, key):
         return self._engines[key]
 
     def __contains__(self, key):
         return key in self._engines
 
     def __iter__(self):
         return iter(self._engines.keys())
 
     def register(self, engine):
         """Register a compression engine with the manager.
 
         The argument must be a ``compressionengine`` instance.
         """
         if not isinstance(engine, compressionengine):
             raise ValueError(_('argument must be a compressionengine'))
 
         name = engine.name()
 
         if name in self._engines:
             raise error.Abort(_('compression engine %s already registered') %
                               name)
 
         bundleinfo = engine.bundletype()
         if bundleinfo:
             bundlename, bundletype = bundleinfo
 
             if bundlename in self._bundlenames:
                 raise error.Abort(_('bundle name %s already registered') %
                                   bundlename)
             if bundletype in self._bundletypes:
                 raise error.Abort(_('bundle type %s already registered by %s') %
                                   (bundletype, self._bundletypes[bundletype]))
 
             # No external facing name declared.
             if bundlename:
                 self._bundlenames[bundlename] = name
 
             self._bundletypes[bundletype] = name
 
         wiresupport = engine.wireprotosupport()
         if wiresupport:
             wiretype = wiresupport.name
             if wiretype in self._wiretypes:
                 raise error.Abort(_('wire protocol compression %s already '
                                     'registered by %s') %
                                   (wiretype, self._wiretypes[wiretype]))
 
             self._wiretypes[wiretype] = name
 
         revlogheader = engine.revlogheader()
         if revlogheader and revlogheader in self._revlogheaders:
             raise error.Abort(_('revlog header %s already registered by %s') %
                               (revlogheader, self._revlogheaders[revlogheader]))
 
         if revlogheader:
             self._revlogheaders[revlogheader] = name
 
         self._engines[name] = engine
 
     @property
     def supportedbundlenames(self):
         return set(self._bundlenames.keys())
 
     @property
     def supportedbundletypes(self):
         return set(self._bundletypes.keys())
 
     def forbundlename(self, bundlename):
         """Obtain a compression engine registered to a bundle name.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundlenames[bundlename]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forbundletype(self, bundletype):
         """Obtain a compression engine registered to a bundle type.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundletypes[bundletype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def supportedwireengines(self, role, onlyavailable=True):
         """Obtain compression engines that support the wire protocol.
 
         Returns a list of engines in prioritized order, most desired first.
 
         If ``onlyavailable`` is set, filter out engines that can't be
         loaded.
         """
         assert role in (SERVERROLE, CLIENTROLE)
 
         attr = 'serverpriority' if role == SERVERROLE else 'clientpriority'
 
         engines = [self._engines[e] for e in self._wiretypes.values()]
         if onlyavailable:
             engines = [e for e in engines if e.available()]
 
         def getkey(e):
             # Sort first by priority, highest first. In case of tie, sort
             # alphabetically. This is arbitrary, but ensures output is
             # stable.
             w = e.wireprotosupport()
             return -1 * getattr(w, attr), w.name
 
         return list(sorted(engines, key=getkey))
 
     def forwiretype(self, wiretype):
         engine = self._engines[self._wiretypes[wiretype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forrevlogheader(self, header):
         """Obtain a compression engine registered to a revlog header.
 
         Will raise KeyError if the revlog header value isn't registered.
         """
         return self._engines[self._revlogheaders[header]]
 
 compengines = compressormanager()
 
 class compressionengine(object):
     """Base class for compression engines.
 
     Compression engines must implement the interface defined by this class.
     """
     def name(self):
         """Returns the name of the compression engine.
 
         This is the key the engine is registered under.
 
         This method must be implemented.
         """
         raise NotImplementedError()
 
     def available(self):
         """Whether the compression engine is available.
 
         The intent of this method is to allow optional compression engines
         that may not be available in all installations (such as engines relying
         on C extensions that may not be present).
         """
         return True
 
     def bundletype(self):
         """Describes bundle identifiers for this engine.
 
         If this compression engine isn't supported for bundles, returns None.
 
         If this engine can be used for bundles, returns a 2-tuple of strings of
         the user-facing "bundle spec" compression name and an internal
         identifier used to denote the compression format within bundles. To
         exclude the name from external usage, set the first element to ``None``.
 
         If bundle compression is supported, the class must also implement
         ``compressstream`` and `decompressorreader``.
         """
         return None
 
     def wireprotosupport(self):
         """Declare support for this compression format on the wire protocol.
 
         If this compression engine isn't supported for compressing wire
         protocol payloads, returns None.
 
         Otherwise, returns ``compenginewireprotosupport`` with the following
         fields:
 
         * String format identifier
         * Integer priority for the server
         * Integer priority for the client
 
         The integer priorities are used to order the advertisement of format
         support by server and client. The highest integer is advertised
         first. Integers with non-positive values aren't advertised.
 
         The priority values are somewhat arbitrary and only used for default
         ordering. The relative order can be changed via config options.
 
         If wire protocol compression is supported, the class must also implement
         ``compressstream`` and ``decompressorreader``.
         """
         return None
 
     def revlogheader(self):
         """Header added to revlog chunks that identifies this engine.
 
         If this engine can be used to compress revlogs, this method should
         return the bytes used to identify chunks compressed with this engine.
         Else, the method should return ``None`` to indicate it does not
         participate in revlog compression.
         """
         return None
 
     def compressstream(self, it, opts=None):
         """Compress an iterator of chunks.
 
         The method receives an iterator (ideally a generator) of chunks of
         bytes to be compressed. It returns an iterator (ideally a generator)
         of bytes of chunks representing the compressed output.
 
         Optionally accepts an argument defining how to perform compression.
         Each engine treats this argument differently.
         """
         raise NotImplementedError()
 
     def decompressorreader(self, fh):
         """Perform decompression on a file object.
 
         Argument is an object with a ``read(size)`` method that returns
         compressed data. Return value is an object with a ``read(size)`` that
         returns uncompressed data.
         """
         raise NotImplementedError()
 
     def revlogcompressor(self, opts=None):
         """Obtain an object that can be used to compress revlog entries.
 
         The object has a ``compress(data)`` method that compresses binary
         data. This method returns compressed binary data or ``None`` if
         the data could not be compressed (too small, not compressible, etc).
         The returned data should have a header uniquely identifying this
         compression format so decompression can be routed to this engine.
         This header should be identified by the ``revlogheader()`` return
         value.
 
         The object has a ``decompress(data)`` method that decompresses
         data. The method will only be called if ``data`` begins with
         ``revlogheader()``. The method should return the raw, uncompressed
         data or raise a ``RevlogError``.
 
         The object is reusable but is not thread safe.
         """
         raise NotImplementedError()
 
 class _zlibengine(compressionengine):
     def name(self):
         return 'zlib'
 
     def bundletype(self):
         return 'gzip', 'GZ'
 
     def wireprotosupport(self):
         return compewireprotosupport('zlib', 20, 20)
 
     def revlogheader(self):
         return 'x'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
 
         z = zlib.compressobj(opts.get('level', -1))
         for chunk in it:
             data = z.compress(chunk)
             # Not all calls to compress emit data. It is cheaper to inspect
             # here than to feed empty chunks through generator.
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = zlib.decompressobj()
             for chunk in filechunkiter(fh):
                 while chunk:
                     # Limit output size to limit memory.
                     yield d.decompress(chunk, 2 ** 18)
                     chunk = d.unconsumed_tail
 
         return chunkbuffer(gen())
 
     class zlibrevlogcompressor(object):
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 44:
                 return None
 
             elif insize <= 1000000:
                 compressed = zlib.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
 
             # zlib makes an internal copy of the input buffer, doubling
             # memory usage for large inputs. So do streaming compression
             # on large inputs.
             else:
                 z = zlib.compressobj()
                 parts = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + 2**20
                     parts.append(z.compress(data[pos:pos2]))
                     pos = pos2
                 parts.append(z.flush())
 
                 if sum(map(len, parts)) < insize:
                     return ''.join(parts)
                 return None
 
         def decompress(self, data):
             try:
                 return zlib.decompress(data)
             except zlib.error as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         return self.zlibrevlogcompressor()
 
 compengines.register(_zlibengine())
 
 class _bz2engine(compressionengine):
     def name(self):
         return 'bz2'
 
     def bundletype(self):
         return 'bzip2', 'BZ'
 
     # We declare a protocol name but don't advertise by default because
     # it is slow.
     def wireprotosupport(self):
         return compewireprotosupport('bzip2', 0, 0)
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         z = bz2.BZ2Compressor(opts.get('level', 9))
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = bz2.BZ2Decompressor()
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_bz2engine())
 
 class _truncatedbz2engine(compressionengine):
     def name(self):
         return 'bz2truncated'
 
     def bundletype(self):
         return None, '_truncatedBZ'
 
     # We don't implement compressstream because it is hackily handled elsewhere.
 
     def decompressorreader(self, fh):
         def gen():
             # The input stream doesn't have the 'BZ' header. So add it back.
             d = bz2.BZ2Decompressor()
             d.decompress('BZ')
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_truncatedbz2engine())
 
 class _noopengine(compressionengine):
     def name(self):
         return 'none'
 
     def bundletype(self):
         return 'none', 'UN'
 
     # Clients always support uncompressed payloads. Servers don't because
     # unless you are on a fast network, uncompressed payloads can easily
     # saturate your network pipe.
     def wireprotosupport(self):
         return compewireprotosupport('none', 0, 10)
 
     # We don't implement revlogheader because it is handled specially
     # in the revlog class.
 
     def compressstream(self, it, opts=None):
         return it
 
     def decompressorreader(self, fh):
         return fh
 
     class nooprevlogcompressor(object):
         def compress(self, data):
             return None
 
     def revlogcompressor(self, opts=None):
         return self.nooprevlogcompressor()
 
 compengines.register(_noopengine())
 
 class _zstdengine(compressionengine):
     def name(self):
         return 'zstd'
 
     @propertycache
     def _module(self):
         # Not all installs have the zstd module available. So defer importing
         # until first access.
         try:
             from . import zstd
             # Force delayed import.
             zstd.__version__
             return zstd
         except ImportError:
             return None
 
     def available(self):
         return bool(self._module)
 
     def bundletype(self):
         return 'zstd', 'ZS'
 
     def wireprotosupport(self):
         return compewireprotosupport('zstd', 50, 50)
 
     def revlogheader(self):
         return '\x28'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         # zstd level 3 is almost always significantly faster than zlib
         # while providing no worse compression. It strikes a good balance
         # between speed and compression.
         level = opts.get('level', 3)
 
         zstd = self._module
         z = zstd.ZstdCompressor(level=level).compressobj()
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         zstd = self._module
         dctx = zstd.ZstdDecompressor()
         return chunkbuffer(dctx.read_from(fh))
 
     class zstdrevlogcompressor(object):
         def __init__(self, zstd, level=3):
             # Writing the content size adds a few bytes to the output. However,
             # it allows decompression to be more optimal since we can
             # pre-allocate a buffer to hold the result.
             self._cctx = zstd.ZstdCompressor(level=level,
                                              write_content_size=True)
             self._dctx = zstd.ZstdDecompressor()
             self._compinsize = zstd.COMPRESSION_RECOMMENDED_INPUT_SIZE
             self._decompinsize = zstd.DECOMPRESSION_RECOMMENDED_INPUT_SIZE
 
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 50:
                 return None
 
             elif insize <= 1000000:
                 compressed = self._cctx.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
             else:
                 z = self._cctx.compressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._compinsize
                     chunk = z.compress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 chunks.append(z.flush())
 
                 if sum(map(len, chunks)) < insize:
                     return ''.join(chunks)
                 return None
 
         def decompress(self, data):
             insize = len(data)
 
             try:
                 # This was measured to be faster than other streaming
                 # decompressors.
                 dobj = self._dctx.decompressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._decompinsize
                     chunk = dobj.decompress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 # Frame should be exhausted, so no finish() API.
 
                 return ''.join(chunks)
             except Exception as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         opts = opts or {}
         return self.zstdrevlogcompressor(self._module,
                                          level=opts.get('level', 3))
 
 compengines.register(_zstdengine())
 
 # convenient shortcut
 dst = debugstacktrace