hg: always create new localrepository instance

cachedlocalrepo.copy() didn't actually create new localrepository
instances. This meant that the new thread isolation code in hgweb wasn't
actually using separate localrepository instances, even though it was
properly using separate cachedlocalrepo instances.

Because the behavior of the API changed, the single caller in hgweb had
to be refactored to always call _webifyrepo() or it may not have used
the proper filter.

I confirmed via print() debugging that id(repo) is in fact different on
each thread. This was not the case before.

For reasons I can't yet explain, this does not fix issue4756. I suspect
there is shared cache somewhere that isn't thread safe.

mercurial/hg.py

@@ -1,894 +1,899
 # hg.py - repository classes for mercurial
 #
 # 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.
 
 from __future__ import absolute_import
 
 import errno
 import os
 import shutil
 
 from .i18n import _
 from .node import nullid
 
 from . import (
     bookmarks,
     bundlerepo,
     cmdutil,
     discovery,
     error,
     exchange,
     extensions,
     httppeer,
     localrepo,
     lock,
     merge as mergemod,
     node,
     phases,
     repoview,
     scmutil,
     sshpeer,
     statichttprepo,
     ui as uimod,
     unionrepo,
     url,
     util,
     verify as verifymod,
 )
 
 release = lock.release
 
 def _local(path):
     path = util.expandpath(util.urllocalpath(path))
     return (os.path.isfile(path) and bundlerepo or localrepo)
 
 def addbranchrevs(lrepo, other, branches, revs):
     peer = other.peer() # a courtesy to callers using a localrepo for other
     hashbranch, branches = branches
     if not hashbranch and not branches:
         x = revs or None
         if util.safehasattr(revs, 'first'):
             y =  revs.first()
         elif revs:
             y = revs[0]
         else:
             y = None
         return x, y
     if revs:
         revs = list(revs)
     else:
         revs = []
 
     if not peer.capable('branchmap'):
         if branches:
             raise util.Abort(_("remote branch lookup not supported"))
         revs.append(hashbranch)
         return revs, revs[0]
     branchmap = peer.branchmap()
 
     def primary(branch):
         if branch == '.':
             if not lrepo:
                 raise util.Abort(_("dirstate branch not accessible"))
             branch = lrepo.dirstate.branch()
         if branch in branchmap:
             revs.extend(node.hex(r) for r in reversed(branchmap[branch]))
             return True
         else:
             return False
 
     for branch in branches:
         if not primary(branch):
             raise error.RepoLookupError(_("unknown branch '%s'") % branch)
     if hashbranch:
         if not primary(hashbranch):
             revs.append(hashbranch)
     return revs, revs[0]
 
 def parseurl(path, branches=None):
     '''parse url#branch, returning (url, (branch, branches))'''
 
     u = util.url(path)
     branch = None
     if u.fragment:
         branch = u.fragment
         u.fragment = None
     return str(u), (branch, branches or [])
 
 schemes = {
     'bundle': bundlerepo,
     'union': unionrepo,
     'file': _local,
     'http': httppeer,
     'https': httppeer,
     'ssh': sshpeer,
     'static-http': statichttprepo,
 }
 
 def _peerlookup(path):
     u = util.url(path)
     scheme = u.scheme or 'file'
     thing = schemes.get(scheme) or schemes['file']
     try:
         return thing(path)
     except TypeError:
         # we can't test callable(thing) because 'thing' can be an unloaded
         # module that implements __call__
         if not util.safehasattr(thing, 'instance'):
             raise
         return thing
 
 def islocal(repo):
     '''return true if repo (or path pointing to repo) is local'''
     if isinstance(repo, str):
         try:
             return _peerlookup(repo).islocal(repo)
         except AttributeError:
             return False
     return repo.local()
 
 def openpath(ui, path):
     '''open path with open if local, url.open if remote'''
     pathurl = util.url(path, parsequery=False, parsefragment=False)
     if pathurl.islocal():
         return util.posixfile(pathurl.localpath(), 'rb')
     else:
         return url.open(ui, path)
 
 # a list of (ui, repo) functions called for wire peer initialization
 wirepeersetupfuncs = []
 
 def _peerorrepo(ui, path, create=False):
     """return a repository object for the specified path"""
     obj = _peerlookup(path).instance(ui, path, create)
     ui = getattr(obj, "ui", ui)
     for name, module in extensions.extensions(ui):
         hook = getattr(module, 'reposetup', None)
         if hook:
             hook(ui, obj)
     if not obj.local():
         for f in wirepeersetupfuncs:
             f(ui, obj)
     return obj
 
 def repository(ui, path='', create=False):
     """return a repository object for the specified path"""
     peer = _peerorrepo(ui, path, create)
     repo = peer.local()
     if not repo:
         raise util.Abort(_("repository '%s' is not local") %
                          (path or peer.url()))
     return repo.filtered('visible')
 
 def peer(uiorrepo, opts, path, create=False):
     '''return a repository peer for the specified path'''
     rui = remoteui(uiorrepo, opts)
     return _peerorrepo(rui, path, create).peer()
 
 def defaultdest(source):
     '''return default destination of clone if none is given
 
     >>> defaultdest('foo')
     'foo'
     >>> defaultdest('/foo/bar')
     'bar'
     >>> defaultdest('/')
     ''
     >>> defaultdest('')
     ''
     >>> defaultdest('http://example.org/')
     ''
     >>> defaultdest('http://example.org/foo/')
     'foo'
     '''
     path = util.url(source).path
     if not path:
         return ''
     return os.path.basename(os.path.normpath(path))
 
 def share(ui, source, dest=None, update=True, bookmarks=True):
     '''create a shared repository'''
 
     if not islocal(source):
         raise util.Abort(_('can only share local repositories'))
 
     if not dest:
         dest = defaultdest(source)
     else:
         dest = ui.expandpath(dest)
 
     if isinstance(source, str):
         origsource = ui.expandpath(source)
         source, branches = parseurl(origsource)
         srcrepo = repository(ui, source)
         rev, checkout = addbranchrevs(srcrepo, srcrepo, branches, None)
     else:
         srcrepo = source.local()
         origsource = source = srcrepo.url()
         checkout = None
 
     sharedpath = srcrepo.sharedpath # if our source is already sharing
 
     destwvfs = scmutil.vfs(dest, realpath=True)
     destvfs = scmutil.vfs(os.path.join(destwvfs.base, '.hg'), realpath=True)
 
     if destvfs.lexists():
         raise util.Abort(_('destination already exists'))
 
     if not destwvfs.isdir():
         destwvfs.mkdir()
     destvfs.makedir()
 
     requirements = ''
     try:
         requirements = srcrepo.vfs.read('requires')
     except IOError as inst:
         if inst.errno != errno.ENOENT:
             raise
 
     requirements += 'shared\n'
     destvfs.write('requires', requirements)
     destvfs.write('sharedpath', sharedpath)
 
     r = repository(ui, destwvfs.base)
 
     default = srcrepo.ui.config('paths', 'default')
     if default:
         fp = r.vfs("hgrc", "w", text=True)
         fp.write("[paths]\n")
         fp.write("default = %s\n" % default)
         fp.close()
 
     if update:
         r.ui.status(_("updating working directory\n"))
         if update is not True:
             checkout = update
         for test in (checkout, 'default', 'tip'):
             if test is None:
                 continue
             try:
                 uprev = r.lookup(test)
                 break
             except error.RepoLookupError:
                 continue
         _update(r, uprev)
 
     if bookmarks:
         fp = r.vfs('shared', 'w')
         fp.write('bookmarks\n')
         fp.close()
 
 def copystore(ui, srcrepo, destpath):
     '''copy files from store of srcrepo in destpath
 
     returns destlock
     '''
     destlock = None
     try:
         hardlink = None
         num = 0
         closetopic = [None]
         def prog(topic, pos):
             if pos is None:
                 closetopic[0] = topic
             else:
                 ui.progress(topic, pos + num)
         srcpublishing = srcrepo.publishing()
         srcvfs = scmutil.vfs(srcrepo.sharedpath)
         dstvfs = scmutil.vfs(destpath)
         for f in srcrepo.store.copylist():
             if srcpublishing and f.endswith('phaseroots'):
                 continue
             dstbase = os.path.dirname(f)
             if dstbase and not dstvfs.exists(dstbase):
                 dstvfs.mkdir(dstbase)
             if srcvfs.exists(f):
                 if f.endswith('data'):
                     # 'dstbase' may be empty (e.g. revlog format 0)
                     lockfile = os.path.join(dstbase, "lock")
                     # lock to avoid premature writing to the target
                     destlock = lock.lock(dstvfs, lockfile)
                 hardlink, n = util.copyfiles(srcvfs.join(f), dstvfs.join(f),
                                              hardlink, progress=prog)
                 num += n
         if hardlink:
             ui.debug("linked %d files\n" % num)
             if closetopic[0]:
                 ui.progress(closetopic[0], None)
         else:
             ui.debug("copied %d files\n" % num)
             if closetopic[0]:
                 ui.progress(closetopic[0], None)
         return destlock
     except: # re-raises
         release(destlock)
         raise
 
 def clonewithshare(ui, peeropts, sharepath, source, srcpeer, dest, pull=False,
                    rev=None, update=True, stream=False):
     """Perform a clone using a shared repo.
 
     The store for the repository will be located at <sharepath>/.hg. The
     specified revisions will be cloned or pulled from "source". A shared repo
     will be created at "dest" and a working copy will be created if "update" is
     True.
     """
     revs = None
     if rev:
         if not srcpeer.capable('lookup'):
             raise util.Abort(_("src repository does not support "
                                "revision lookup and so doesn't "
                                "support clone by revision"))
         revs = [srcpeer.lookup(r) for r in rev]
 
     basename = os.path.basename(sharepath)
 
     if os.path.exists(sharepath):
         ui.status(_('(sharing from existing pooled repository %s)\n') %
                   basename)
     else:
         ui.status(_('(sharing from new pooled repository %s)\n') % basename)
         # Always use pull mode because hardlinks in share mode don't work well.
         # Never update because working copies aren't necessary in share mode.
         clone(ui, peeropts, source, dest=sharepath, pull=True,
               rev=rev, update=False, stream=stream)
 
     sharerepo = repository(ui, path=sharepath)
     share(ui, sharerepo, dest=dest, update=update, bookmarks=False)
 
     # We need to perform a pull against the dest repo to fetch bookmarks
     # and other non-store data that isn't shared by default. In the case of
     # non-existing shared repo, this means we pull from the remote twice. This
     # is a bit weird. But at the time it was implemented, there wasn't an easy
     # way to pull just non-changegroup data.
     destrepo = repository(ui, path=dest)
     exchange.pull(destrepo, srcpeer, heads=revs)
 
     return srcpeer, peer(ui, peeropts, dest)
 
 def clone(ui, peeropts, source, dest=None, pull=False, rev=None,
           update=True, stream=False, branch=None, shareopts=None):
     """Make a copy of an existing repository.
 
     Create a copy of an existing repository in a new directory.  The
     source and destination are URLs, as passed to the repository
     function.  Returns a pair of repository peers, the source and
     newly created destination.
 
     The location of the source is added to the new repository's
     .hg/hgrc file, as the default to be used for future pulls and
     pushes.
 
     If an exception is raised, the partly cloned/updated destination
     repository will be deleted.
 
     Arguments:
 
     source: repository object or URL
 
     dest: URL of destination repository to create (defaults to base
     name of source repository)
 
     pull: always pull from source repository, even in local case or if the
     server prefers streaming
 
     stream: stream raw data uncompressed from repository (fast over
     LAN, slow over WAN)
 
     rev: revision to clone up to (implies pull=True)
 
     update: update working directory after clone completes, if
     destination is local repository (True means update to default rev,
     anything else is treated as a revision)
 
     branch: branches to clone
 
     shareopts: dict of options to control auto sharing behavior. The "pool" key
     activates auto sharing mode and defines the directory for stores. The
     "mode" key determines how to construct the directory name of the shared
     repository. "identity" means the name is derived from the node of the first
     changeset in the repository. "remote" means the name is derived from the
     remote's path/URL. Defaults to "identity."
     """
 
     if isinstance(source, str):
         origsource = ui.expandpath(source)
         source, branch = parseurl(origsource, branch)
         srcpeer = peer(ui, peeropts, source)
     else:
         srcpeer = source.peer() # in case we were called with a localrepo
         branch = (None, branch or [])
         origsource = source = srcpeer.url()
     rev, checkout = addbranchrevs(srcpeer, srcpeer, branch, rev)
 
     if dest is None:
         dest = defaultdest(source)
         if dest:
             ui.status(_("destination directory: %s\n") % dest)
     else:
         dest = ui.expandpath(dest)
 
     dest = util.urllocalpath(dest)
     source = util.urllocalpath(source)
 
     if not dest:
         raise util.Abort(_("empty destination path is not valid"))
 
     destvfs = scmutil.vfs(dest, expandpath=True)
     if destvfs.lexists():
         if not destvfs.isdir():
             raise util.Abort(_("destination '%s' already exists") % dest)
         elif destvfs.listdir():
             raise util.Abort(_("destination '%s' is not empty") % dest)
 
     shareopts = shareopts or {}
     sharepool = shareopts.get('pool')
     sharenamemode = shareopts.get('mode')
     if sharepool and islocal(dest):
         sharepath = None
         if sharenamemode == 'identity':
             # Resolve the name from the initial changeset in the remote
             # repository. This returns nullid when the remote is empty. It
             # raises RepoLookupError if revision 0 is filtered or otherwise
             # not available. If we fail to resolve, sharing is not enabled.
             try:
                 rootnode = srcpeer.lookup('0')
                 if rootnode != node.nullid:
                     sharepath = os.path.join(sharepool, node.hex(rootnode))
                 else:
                     ui.status(_('(not using pooled storage: '
                                 'remote appears to be empty)\n'))
             except error.RepoLookupError:
                 ui.status(_('(not using pooled storage: '
                             'unable to resolve identity of remote)\n'))
         elif sharenamemode == 'remote':
             sharepath = os.path.join(sharepool, util.sha1(source).hexdigest())
         else:
             raise util.Abort('unknown share naming mode: %s' % sharenamemode)
 
         if sharepath:
             return clonewithshare(ui, peeropts, sharepath, source, srcpeer,
                                   dest, pull=pull, rev=rev, update=update,
                                   stream=stream)
 
     srclock = destlock = cleandir = None
     srcrepo = srcpeer.local()
     try:
         abspath = origsource
         if islocal(origsource):
             abspath = os.path.abspath(util.urllocalpath(origsource))
 
         if islocal(dest):
             cleandir = dest
 
         copy = False
         if (srcrepo and srcrepo.cancopy() and islocal(dest)
             and not phases.hassecret(srcrepo)):
             copy = not pull and not rev
 
         if copy:
             try:
                 # we use a lock here because if we race with commit, we
                 # can end up with extra data in the cloned revlogs that's
                 # not pointed to by changesets, thus causing verify to
                 # fail
                 srclock = srcrepo.lock(wait=False)
             except error.LockError:
                 copy = False
 
         if copy:
             srcrepo.hook('preoutgoing', throw=True, source='clone')
             hgdir = os.path.realpath(os.path.join(dest, ".hg"))
             if not os.path.exists(dest):
                 os.mkdir(dest)
             else:
                 # only clean up directories we create ourselves
                 cleandir = hgdir
             try:
                 destpath = hgdir
                 util.makedir(destpath, notindexed=True)
             except OSError as inst:
                 if inst.errno == errno.EEXIST:
                     cleandir = None
                     raise util.Abort(_("destination '%s' already exists")
                                      % dest)
                 raise
 
             destlock = copystore(ui, srcrepo, destpath)
             # copy bookmarks over
             srcbookmarks = srcrepo.join('bookmarks')
             dstbookmarks = os.path.join(destpath, 'bookmarks')
             if os.path.exists(srcbookmarks):
                 util.copyfile(srcbookmarks, dstbookmarks)
 
             # Recomputing branch cache might be slow on big repos,
             # so just copy it
             def copybranchcache(fname):
                 srcbranchcache = srcrepo.join('cache/%s' % fname)
                 dstbranchcache = os.path.join(dstcachedir, fname)
                 if os.path.exists(srcbranchcache):
                     if not os.path.exists(dstcachedir):
                         os.mkdir(dstcachedir)
                     util.copyfile(srcbranchcache, dstbranchcache)
 
             dstcachedir = os.path.join(destpath, 'cache')
             # In local clones we're copying all nodes, not just served
             # ones. Therefore copy all branch caches over.
             copybranchcache('branch2')
             for cachename in repoview.filtertable:
                 copybranchcache('branch2-%s' % cachename)
 
             # we need to re-init the repo after manually copying the data
             # into it
             destpeer = peer(srcrepo, peeropts, dest)
             srcrepo.hook('outgoing', source='clone',
                           node=node.hex(node.nullid))
         else:
             try:
                 destpeer = peer(srcrepo or ui, peeropts, dest, create=True)
                                 # only pass ui when no srcrepo
             except OSError as inst:
                 if inst.errno == errno.EEXIST:
                     cleandir = None
                     raise util.Abort(_("destination '%s' already exists")
                                      % dest)
                 raise
 
             revs = None
             if rev:
                 if not srcpeer.capable('lookup'):
                     raise util.Abort(_("src repository does not support "
                                        "revision lookup and so doesn't "
                                        "support clone by revision"))
                 revs = [srcpeer.lookup(r) for r in rev]
                 checkout = revs[0]
             if destpeer.local():
                 if not stream:
                     if pull:
                         stream = False
                     else:
                         stream = None
                 destpeer.local().clone(srcpeer, heads=revs, stream=stream)
             elif srcrepo:
                 exchange.push(srcrepo, destpeer, revs=revs,
                               bookmarks=srcrepo._bookmarks.keys())
             else:
                 raise util.Abort(_("clone from remote to remote not supported"))
 
         cleandir = None
 
         destrepo = destpeer.local()
         if destrepo:
             template = uimod.samplehgrcs['cloned']
             fp = destrepo.vfs("hgrc", "w", text=True)
             u = util.url(abspath)
             u.passwd = None
             defaulturl = str(u)
             fp.write(template % defaulturl)
             fp.close()
 
             destrepo.ui.setconfig('paths', 'default', defaulturl, 'clone')
 
             if update:
                 if update is not True:
                     checkout = srcpeer.lookup(update)
                 uprev = None
                 status = None
                 if checkout is not None:
                     try:
                         uprev = destrepo.lookup(checkout)
                     except error.RepoLookupError:
                         try:
                             uprev = destrepo.lookup(update)
                         except error.RepoLookupError:
                             pass
                 if uprev is None:
                     try:
                         uprev = destrepo._bookmarks['@']
                         update = '@'
                         bn = destrepo[uprev].branch()
                         if bn == 'default':
                             status = _("updating to bookmark @\n")
                         else:
                             status = (_("updating to bookmark @ on branch %s\n")
                                        % bn)
                     except KeyError:
                         try:
                             uprev = destrepo.branchtip('default')
                         except error.RepoLookupError:
                             uprev = destrepo.lookup('tip')
                 if not status:
                     bn = destrepo[uprev].branch()
                     status = _("updating to branch %s\n") % bn
                 destrepo.ui.status(status)
                 _update(destrepo, uprev)
                 if update in destrepo._bookmarks:
                     bookmarks.activate(destrepo, update)
     finally:
         release(srclock, destlock)
         if cleandir is not None:
             shutil.rmtree(cleandir, True)
         if srcpeer is not None:
             srcpeer.close()
     return srcpeer, destpeer
 
 def _showstats(repo, stats):
     repo.ui.status(_("%d files updated, %d files merged, "
                      "%d files removed, %d files unresolved\n") % stats)
 
 def updaterepo(repo, node, overwrite):
     """Update the working directory to node.
 
     When overwrite is set, changes are clobbered, merged else
 
     returns stats (see pydoc mercurial.merge.applyupdates)"""
     return mergemod.update(repo, node, False, overwrite, None,
                            labels=['working copy', 'destination'])
 
 def update(repo, node):
     """update the working directory to node, merging linear changes"""
     stats = updaterepo(repo, node, False)
     _showstats(repo, stats)
     if stats[3]:
         repo.ui.status(_("use 'hg resolve' to retry unresolved file merges\n"))
     return stats[3] > 0
 
 # naming conflict in clone()
 _update = update
 
 def clean(repo, node, show_stats=True):
     """forcibly switch the working directory to node, clobbering changes"""
     stats = updaterepo(repo, node, True)
     util.unlinkpath(repo.join('graftstate'), ignoremissing=True)
     if show_stats:
         _showstats(repo, stats)
     return stats[3] > 0
 
 def merge(repo, node, force=None, remind=True):
     """Branch merge with node, resolving changes. Return true if any
     unresolved conflicts."""
     stats = mergemod.update(repo, node, True, force, False)
     _showstats(repo, stats)
     if stats[3]:
         repo.ui.status(_("use 'hg resolve' to retry unresolved file merges "
                          "or 'hg update -C .' to abandon\n"))
     elif remind:
         repo.ui.status(_("(branch merge, don't forget to commit)\n"))
     return stats[3] > 0
 
 def _incoming(displaychlist, subreporecurse, ui, repo, source,
         opts, buffered=False):
     """
     Helper for incoming / gincoming.
     displaychlist gets called with
         (remoterepo, incomingchangesetlist, displayer) parameters,
     and is supposed to contain only code that can't be unified.
     """
     source, branches = parseurl(ui.expandpath(source), opts.get('branch'))
     other = peer(repo, opts, source)
     ui.status(_('comparing with %s\n') % util.hidepassword(source))
     revs, checkout = addbranchrevs(repo, other, branches, opts.get('rev'))
 
     if revs:
         revs = [other.lookup(rev) for rev in revs]
     other, chlist, cleanupfn = bundlerepo.getremotechanges(ui, repo, other,
                                 revs, opts["bundle"], opts["force"])
     try:
         if not chlist:
             ui.status(_("no changes found\n"))
             return subreporecurse()
 
         displayer = cmdutil.show_changeset(ui, other, opts, buffered)
         displaychlist(other, chlist, displayer)
         displayer.close()
     finally:
         cleanupfn()
     subreporecurse()
     return 0 # exit code is zero since we found incoming changes
 
 def incoming(ui, repo, source, opts):
     def subreporecurse():
         ret = 1
         if opts.get('subrepos'):
             ctx = repo[None]
             for subpath in sorted(ctx.substate):
                 sub = ctx.sub(subpath)
                 ret = min(ret, sub.incoming(ui, source, opts))
         return ret
 
     def display(other, chlist, displayer):
         limit = cmdutil.loglimit(opts)
         if opts.get('newest_first'):
             chlist.reverse()
         count = 0
         for n in chlist:
             if limit is not None and count >= limit:
                 break
             parents = [p for p in other.changelog.parents(n) if p != nullid]
             if opts.get('no_merges') and len(parents) == 2:
                 continue
             count += 1
             displayer.show(other[n])
     return _incoming(display, subreporecurse, ui, repo, source, opts)
 
 def _outgoing(ui, repo, dest, opts):
     dest = ui.expandpath(dest or 'default-push', dest or 'default')
     dest, branches = parseurl(dest, opts.get('branch'))
     ui.status(_('comparing with %s\n') % util.hidepassword(dest))
     revs, checkout = addbranchrevs(repo, repo, branches, opts.get('rev'))
     if revs:
         revs = [repo.lookup(rev) for rev in scmutil.revrange(repo, revs)]
 
     other = peer(repo, opts, dest)
     outgoing = discovery.findcommonoutgoing(repo.unfiltered(), other, revs,
                                             force=opts.get('force'))
     o = outgoing.missing
     if not o:
         scmutil.nochangesfound(repo.ui, repo, outgoing.excluded)
     return o, other
 
 def outgoing(ui, repo, dest, opts):
     def recurse():
         ret = 1
         if opts.get('subrepos'):
             ctx = repo[None]
             for subpath in sorted(ctx.substate):
                 sub = ctx.sub(subpath)
                 ret = min(ret, sub.outgoing(ui, dest, opts))
         return ret
 
     limit = cmdutil.loglimit(opts)
     o, other = _outgoing(ui, repo, dest, opts)
     if not o:
         cmdutil.outgoinghooks(ui, repo, other, opts, o)
         return recurse()
 
     if opts.get('newest_first'):
         o.reverse()
     displayer = cmdutil.show_changeset(ui, repo, opts)
     count = 0
     for n in o:
         if limit is not None and count >= limit:
             break
         parents = [p for p in repo.changelog.parents(n) if p != nullid]
         if opts.get('no_merges') and len(parents) == 2:
             continue
         count += 1
         displayer.show(repo[n])
     displayer.close()
     cmdutil.outgoinghooks(ui, repo, other, opts, o)
     recurse()
     return 0 # exit code is zero since we found outgoing changes
 
 def revert(repo, node, choose):
     """revert changes to revision in node without updating dirstate"""
     return mergemod.update(repo, node, False, True, choose)[3] > 0
 
 def verify(repo):
     """verify the consistency of a repository"""
     ret = verifymod.verify(repo)
 
     # Broken subrepo references in hidden csets don't seem worth worrying about,
     # since they can't be pushed/pulled, and --hidden can be used if they are a
     # concern.
 
     # pathto() is needed for -R case
     revs = repo.revs("filelog(%s)",
                      util.pathto(repo.root, repo.getcwd(), '.hgsubstate'))
 
     if revs:
         repo.ui.status(_('checking subrepo links\n'))
         for rev in revs:
             ctx = repo[rev]
             try:
                 for subpath in ctx.substate:
                     ret = ctx.sub(subpath).verify() or ret
             except Exception:
                 repo.ui.warn(_('.hgsubstate is corrupt in revision %s\n') %
                              node.short(ctx.node()))
 
     return ret
 
 def remoteui(src, opts):
     'build a remote ui from ui or repo and opts'
     if util.safehasattr(src, 'baseui'): # looks like a repository
         dst = src.baseui.copy() # drop repo-specific config
         src = src.ui # copy target options from repo
     else: # assume it's a global ui object
         dst = src.copy() # keep all global options
 
     # copy ssh-specific options
     for o in 'ssh', 'remotecmd':
         v = opts.get(o) or src.config('ui', o)
         if v:
             dst.setconfig("ui", o, v, 'copied')
 
     # copy bundle-specific options
     r = src.config('bundle', 'mainreporoot')
     if r:
         dst.setconfig('bundle', 'mainreporoot', r, 'copied')
 
     # copy selected local settings to the remote ui
     for sect in ('auth', 'hostfingerprints', 'http_proxy'):
         for key, val in src.configitems(sect):
             dst.setconfig(sect, key, val, 'copied')
     v = src.config('web', 'cacerts')
     if v == '!':
         dst.setconfig('web', 'cacerts', v, 'copied')
     elif v:
         dst.setconfig('web', 'cacerts', util.expandpath(v), 'copied')
 
     return dst
 
 # Files of interest
 # Used to check if the repository has changed looking at mtime and size of
 # theses files.
 foi = [('spath', '00changelog.i'),
        ('spath', 'phaseroots'), # ! phase can change content at the same size
        ('spath', 'obsstore'),
        ('path', 'bookmarks'), # ! bookmark can change content at the same size
       ]
 
 class cachedlocalrepo(object):
     """Holds a localrepository that can be cached and reused."""
 
     def __init__(self, repo):
         """Create a new cached repo from an existing repo.
 
         We assume the passed in repo was recently created. If the
         repo has changed between when it was created and when it was
         turned into a cache, it may not refresh properly.
         """
         assert isinstance(repo, localrepo.localrepository)
         self._repo = repo
         self._state, self.mtime = self._repostate()
 
     def fetch(self):
         """Refresh (if necessary) and return a repository.
 
         If the cached instance is out of date, it will be recreated
         automatically and returned.
 
         Returns a tuple of the repo and a boolean indicating whether a new
         repo instance was created.
         """
         # We compare the mtimes and sizes of some well-known files to
         # determine if the repo changed. This is not precise, as mtimes
         # are susceptible to clock skew and imprecise filesystems and
         # file content can change while maintaining the same size.
 
         state, mtime = self._repostate()
         if state == self._state:
             return self._repo, False
 
         self._repo = repository(self._repo.baseui, self._repo.url())
         self._state = state
         self.mtime = mtime
 
         return self._repo, True
 
     def _repostate(self):
         state = []
         maxmtime = -1
         for attr, fname in foi:
             prefix = getattr(self._repo, attr)
             p = os.path.join(prefix, fname)
             try:
                 st = os.stat(p)
             except OSError:
                 st = os.stat(prefix)
             state.append((st.st_mtime, st.st_size))
             maxmtime = max(maxmtime, st.st_mtime)
 
         return tuple(state), maxmtime
 
     def copy(self):
-        """Obtain a copy of this class instance."""
-        c = cachedlocalrepo(self._repo)
+        """Obtain a copy of this class instance.
+
+        A new localrepository instance is obtained. The new instance should be
+        completely independent of the original.
+        """
+        repo = repository(self._repo.baseui, self._repo.origroot)
+        c = cachedlocalrepo(repo)
         c._state = self._state
         c.mtime = self.mtime
         return c

mercurial/hgweb/hgweb_mod.py

@@ -1,437 +1,437
 # hgweb/hgweb_mod.py - Web interface for a repository.
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # 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.
 
 import contextlib
 import os
 from mercurial import ui, hg, hook, error, encoding, templater, util, repoview
 from mercurial.templatefilters import websub
 from common import ErrorResponse, permhooks, caching
 from common import HTTP_OK, HTTP_NOT_MODIFIED, HTTP_BAD_REQUEST
 from common import HTTP_NOT_FOUND, HTTP_SERVER_ERROR
 from request import wsgirequest
 import webcommands, protocol, webutil
 
 perms = {
     'changegroup': 'pull',
     'changegroupsubset': 'pull',
     'getbundle': 'pull',
     'stream_out': 'pull',
     'listkeys': 'pull',
     'unbundle': 'push',
     'pushkey': 'push',
 }
 
 def makebreadcrumb(url, prefix=''):
     '''Return a 'URL breadcrumb' list
 
     A 'URL breadcrumb' is a list of URL-name pairs,
     corresponding to each of the path items on a URL.
     This can be used to create path navigation entries.
     '''
     if url.endswith('/'):
         url = url[:-1]
     if prefix:
         url = '/' + prefix + url
     relpath = url
     if relpath.startswith('/'):
         relpath = relpath[1:]
 
     breadcrumb = []
     urlel = url
     pathitems = [''] + relpath.split('/')
     for pathel in reversed(pathitems):
         if not pathel or not urlel:
             break
         breadcrumb.append({'url': urlel, 'name': pathel})
         urlel = os.path.dirname(urlel)
     return reversed(breadcrumb)
 
 class requestcontext(object):
     """Holds state/context for an individual request.
 
     Servers can be multi-threaded. Holding state on the WSGI application
     is prone to race conditions. Instances of this class exist to hold
     mutable and race-free state for requests.
     """
     def __init__(self, app, repo):
         self.repo = repo
         self.reponame = app.reponame
 
         self.archives = ('zip', 'gz', 'bz2')
 
         self.maxchanges = self.configint('web', 'maxchanges', 10)
         self.stripecount = self.configint('web', 'stripes', 1)
         self.maxshortchanges = self.configint('web', 'maxshortchanges', 60)
         self.maxfiles = self.configint('web', 'maxfiles', 10)
         self.allowpull = self.configbool('web', 'allowpull', True)
 
         # we use untrusted=False to prevent a repo owner from using
         # web.templates in .hg/hgrc to get access to any file readable
         # by the user running the CGI script
         self.templatepath = self.config('web', 'templates', untrusted=False)
 
         # This object is more expensive to build than simple config values.
         # It is shared across requests. The app will replace the object
         # if it is updated. Since this is a reference and nothing should
         # modify the underlying object, it should be constant for the lifetime
         # of the request.
         self.websubtable = app.websubtable
 
     # Trust the settings from the .hg/hgrc files by default.
     def config(self, section, name, default=None, untrusted=True):
         return self.repo.ui.config(section, name, default,
                                    untrusted=untrusted)
 
     def configbool(self, section, name, default=False, untrusted=True):
         return self.repo.ui.configbool(section, name, default,
                                        untrusted=untrusted)
 
     def configint(self, section, name, default=None, untrusted=True):
         return self.repo.ui.configint(section, name, default,
                                       untrusted=untrusted)
 
     def configlist(self, section, name, default=None, untrusted=True):
         return self.repo.ui.configlist(section, name, default,
                                        untrusted=untrusted)
 
     archivespecs = {
         'bz2': ('application/x-bzip2', 'tbz2', '.tar.bz2', None),
         'gz': ('application/x-gzip', 'tgz', '.tar.gz', None),
         'zip': ('application/zip', 'zip', '.zip', None),
     }
 
     def archivelist(self, nodeid):
         allowed = self.configlist('web', 'allow_archive')
         for typ, spec in self.archivespecs.iteritems():
             if typ in allowed or self.configbool('web', 'allow%s' % typ):
                 yield {'type': typ, 'extension': spec[2], 'node': nodeid}
 
     def templater(self, req):
         # determine scheme, port and server name
         # this is needed to create absolute urls
 
         proto = req.env.get('wsgi.url_scheme')
         if proto == 'https':
             proto = 'https'
             default_port = '443'
         else:
             proto = 'http'
             default_port = '80'
 
         port = req.env['SERVER_PORT']
         port = port != default_port and (':' + port) or ''
         urlbase = '%s://%s%s' % (proto, req.env['SERVER_NAME'], port)
         logourl = self.config('web', 'logourl', 'http://mercurial.selenic.com/')
         logoimg = self.config('web', 'logoimg', 'hglogo.png')
         staticurl = self.config('web', 'staticurl') or req.url + 'static/'
         if not staticurl.endswith('/'):
             staticurl += '/'
 
         # some functions for the templater
 
         def motd(**map):
             yield self.config('web', 'motd', '')
 
         # figure out which style to use
 
         vars = {}
         styles = (
             req.form.get('style', [None])[0],
             self.config('web', 'style'),
             'paper',
         )
         style, mapfile = templater.stylemap(styles, self.templatepath)
         if style == styles[0]:
             vars['style'] = style
 
         start = req.url[-1] == '?' and '&' or '?'
         sessionvars = webutil.sessionvars(vars, start)
 
         if not self.reponame:
             self.reponame = (self.config('web', 'name')
                              or req.env.get('REPO_NAME')
                              or req.url.strip('/') or self.repo.root)
 
         def websubfilter(text):
             return websub(text, self.websubtable)
 
         # create the templater
 
         tmpl = templater.templater(mapfile,
                                    filters={'websub': websubfilter},
                                    defaults={'url': req.url,
                                              'logourl': logourl,
                                              'logoimg': logoimg,
                                              'staticurl': staticurl,
                                              'urlbase': urlbase,
                                              'repo': self.reponame,
                                              'encoding': encoding.encoding,
                                              'motd': motd,
                                              'sessionvars': sessionvars,
                                              'pathdef': makebreadcrumb(req.url),
                                              'style': style,
                                             })
         return tmpl
 
 
 class hgweb(object):
     """HTTP server for individual repositories.
 
     Instances of this class serve HTTP responses for a particular
     repository.
 
     Instances are typically used as WSGI applications.
 
     Some servers are multi-threaded. On these servers, there may
     be multiple active threads inside __call__.
     """
     def __init__(self, repo, name=None, baseui=None):
         if isinstance(repo, str):
             if baseui:
                 u = baseui.copy()
             else:
                 u = ui.ui()
             r = hg.repository(u, repo)
         else:
             # we trust caller to give us a private copy
             r = repo
 
         r.ui.setconfig('ui', 'report_untrusted', 'off', 'hgweb')
         r.baseui.setconfig('ui', 'report_untrusted', 'off', 'hgweb')
         r.ui.setconfig('ui', 'nontty', 'true', 'hgweb')
         r.baseui.setconfig('ui', 'nontty', 'true', 'hgweb')
         # displaying bundling progress bar while serving feel wrong and may
         # break some wsgi implementation.
         r.ui.setconfig('progress', 'disable', 'true', 'hgweb')
         r.baseui.setconfig('progress', 'disable', 'true', 'hgweb')
         self._repos = [hg.cachedlocalrepo(self._webifyrepo(r))]
         self._lastrepo = self._repos[0]
         hook.redirect(True)
         self.reponame = name
 
     def _webifyrepo(self, repo):
         repo = getwebview(repo)
         self.websubtable = webutil.getwebsubs(repo)
         return repo
 
     @contextlib.contextmanager
     def _obtainrepo(self):
         """Obtain a repo unique to the caller.
 
         Internally we maintain a stack of cachedlocalrepo instances
         to be handed out. If one is available, we pop it and return it,
         ensuring it is up to date in the process. If one is not available,
         we clone the most recently used repo instance and return it.
 
         It is currently possible for the stack to grow without bounds
         if the server allows infinite threads. However, servers should
         have a thread limit, thus establishing our limit.
         """
         if self._repos:
             cached = self._repos.pop()
             r, created = cached.fetch()
-            if created:
-                r = self._webifyrepo(r)
         else:
             cached = self._lastrepo.copy()
             r, created = cached.fetch()
+        if created:
+            r = self._webifyrepo(r)
 
         self._lastrepo = cached
         self.mtime = cached.mtime
         try:
             yield r
         finally:
             self._repos.append(cached)
 
     def run(self):
         """Start a server from CGI environment.
 
         Modern servers should be using WSGI and should avoid this
         method, if possible.
         """
         if not os.environ.get('GATEWAY_INTERFACE', '').startswith("CGI/1."):
             raise RuntimeError("This function is only intended to be "
                                "called while running as a CGI script.")
         import mercurial.hgweb.wsgicgi as wsgicgi
         wsgicgi.launch(self)
 
     def __call__(self, env, respond):
         """Run the WSGI application.
 
         This may be called by multiple threads.
         """
         req = wsgirequest(env, respond)
         return self.run_wsgi(req)
 
     def run_wsgi(self, req):
         """Internal method to run the WSGI application.
 
         This is typically only called by Mercurial. External consumers
         should be using instances of this class as the WSGI application.
         """
         with self._obtainrepo() as repo:
             return self._runwsgi(req, repo)
 
     def _runwsgi(self, req, repo):
         rctx = requestcontext(self, repo)
 
         # This state is global across all threads.
         encoding.encoding = rctx.config('web', 'encoding', encoding.encoding)
         rctx.repo.ui.environ = req.env
 
         # work with CGI variables to create coherent structure
         # use SCRIPT_NAME, PATH_INFO and QUERY_STRING as well as our REPO_NAME
 
         req.url = req.env['SCRIPT_NAME']
         if not req.url.endswith('/'):
             req.url += '/'
         if 'REPO_NAME' in req.env:
             req.url += req.env['REPO_NAME'] + '/'
 
         if 'PATH_INFO' in req.env:
             parts = req.env['PATH_INFO'].strip('/').split('/')
             repo_parts = req.env.get('REPO_NAME', '').split('/')
             if parts[:len(repo_parts)] == repo_parts:
                 parts = parts[len(repo_parts):]
             query = '/'.join(parts)
         else:
             query = req.env['QUERY_STRING'].split('&', 1)[0]
             query = query.split(';', 1)[0]
 
         # process this if it's a protocol request
         # protocol bits don't need to create any URLs
         # and the clients always use the old URL structure
 
         cmd = req.form.get('cmd', [''])[0]
         if protocol.iscmd(cmd):
             try:
                 if query:
                     raise ErrorResponse(HTTP_NOT_FOUND)
                 if cmd in perms:
                     self.check_perm(rctx, req, perms[cmd])
                 return protocol.call(rctx.repo, req, cmd)
             except ErrorResponse as inst:
                 # A client that sends unbundle without 100-continue will
                 # break if we respond early.
                 if (cmd == 'unbundle' and
                     (req.env.get('HTTP_EXPECT',
                                  '').lower() != '100-continue') or
                     req.env.get('X-HgHttp2', '')):
                     req.drain()
                 else:
                     req.headers.append(('Connection', 'Close'))
                 req.respond(inst, protocol.HGTYPE,
                             body='0\n%s\n' % inst)
                 return ''
 
         # translate user-visible url structure to internal structure
 
         args = query.split('/', 2)
         if 'cmd' not in req.form and args and args[0]:
 
             cmd = args.pop(0)
             style = cmd.rfind('-')
             if style != -1:
                 req.form['style'] = [cmd[:style]]
                 cmd = cmd[style + 1:]
 
             # avoid accepting e.g. style parameter as command
             if util.safehasattr(webcommands, cmd):
                 req.form['cmd'] = [cmd]
 
             if cmd == 'static':
                 req.form['file'] = ['/'.join(args)]
             else:
                 if args and args[0]:
                     node = args.pop(0).replace('%2F', '/')
                     req.form['node'] = [node]
                 if args:
                     req.form['file'] = args
 
             ua = req.env.get('HTTP_USER_AGENT', '')
             if cmd == 'rev' and 'mercurial' in ua:
                 req.form['style'] = ['raw']
 
             if cmd == 'archive':
                 fn = req.form['node'][0]
                 for type_, spec in rctx.archivespecs.iteritems():
                     ext = spec[2]
                     if fn.endswith(ext):
                         req.form['node'] = [fn[:-len(ext)]]
                         req.form['type'] = [type_]
 
         # process the web interface request
 
         try:
             tmpl = rctx.templater(req)
             ctype = tmpl('mimetype', encoding=encoding.encoding)
             ctype = templater.stringify(ctype)
 
             # check read permissions non-static content
             if cmd != 'static':
                 self.check_perm(rctx, req, None)
 
             if cmd == '':
                 req.form['cmd'] = [tmpl.cache['default']]
                 cmd = req.form['cmd'][0]
 
             if rctx.configbool('web', 'cache', True):
                 caching(self, req) # sets ETag header or raises NOT_MODIFIED
             if cmd not in webcommands.__all__:
                 msg = 'no such method: %s' % cmd
                 raise ErrorResponse(HTTP_BAD_REQUEST, msg)
             elif cmd == 'file' and 'raw' in req.form.get('style', []):
                 rctx.ctype = ctype
                 content = webcommands.rawfile(rctx, req, tmpl)
             else:
                 content = getattr(webcommands, cmd)(rctx, req, tmpl)
                 req.respond(HTTP_OK, ctype)
 
             return content
 
         except (error.LookupError, error.RepoLookupError) as err:
             req.respond(HTTP_NOT_FOUND, ctype)
             msg = str(err)
             if (util.safehasattr(err, 'name') and
                 not isinstance(err,  error.ManifestLookupError)):
                 msg = 'revision not found: %s' % err.name
             return tmpl('error', error=msg)
         except (error.RepoError, error.RevlogError) as inst:
             req.respond(HTTP_SERVER_ERROR, ctype)
             return tmpl('error', error=str(inst))
         except ErrorResponse as inst:
             req.respond(inst, ctype)
             if inst.code == HTTP_NOT_MODIFIED:
                 # Not allowed to return a body on a 304
                 return ['']
             return tmpl('error', error=str(inst))
 
     def check_perm(self, rctx, req, op):
         for permhook in permhooks:
             permhook(rctx, req, op)
 
 def getwebview(repo):
     """The 'web.view' config controls changeset filter to hgweb. Possible
     values are ``served``, ``visible`` and ``all``. Default is ``served``.
     The ``served`` filter only shows changesets that can be pulled from the
     hgweb instance.  The``visible`` filter includes secret changesets but
     still excludes "hidden" one.
 
     See the repoview module for details.
 
     The option has been around undocumented since Mercurial 2.5, but no
     user ever asked about it. So we better keep it undocumented for now."""
     viewconfig = repo.ui.config('web', 'view', 'served',
                                 untrusted=True)
     if viewconfig == 'all':
         return repo.unfiltered()
     elif viewconfig in repoview.filtertable:
         return repo.filtered(viewconfig)
     else:
         return repo.filtered('served')