lfs: special case the null:// usercache instead of treating it as a url

The previous code worked on Windows, but not on Unix, and a pending patch's test
failed.  The url being used was something like "/tmp/.../client1/null://",
courtesy of ui.configpath().  Looking at the doc comment, this seems like it's
maybe not the right function to call (why should a relative cache path be
expanded relative to the repo root or config file?), but largefiles has been
using it since 8b8dd13295db (Oct 2011).  It was introduced in 1b591f9b7fd2 (Jan
2011) without comment or callers.  A grep over the whole history shows that only
largefiles used it until lfs and infinitepush came along recently.

It looks like if the `if not os.path.isabs(v) or "://" not in v` in configpath()
is changed to an 'and', both Linux and Windows are happy.  I'm guessing that
"://" is to pick off URLs, so that seems reasonable.  But I'm not sure why it
isn't explicitly "file://", and I thought that "file://foo" is relative anyway.
(At least, there are doctests for file:///tmp in util.url.)  There is no mention
of this setting in the help, but it is referenced on the wiki page for
largefiles.  (There's no mention that this is intended to be a URL, and the
example uses an absolute path.)

I don't want this blocking the rest of the lfs server discovery stuff.  It was
also wrong to allow a file:// URL here, but not in largefiles.

hgext/lfs/__init__.py

@@ -1,396 +1,399
 # lfs - hash-preserving large file support using Git-LFS protocol
 #
 # Copyright 2017 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """lfs - large file support (EXPERIMENTAL)
 
 This extension allows large files to be tracked outside of the normal
 repository storage and stored on a centralized server, similar to the
 ``largefiles`` extension.  The ``git-lfs`` protocol is used when
 communicating with the server, so existing git infrastructure can be
 harnessed.  Even though the files are stored outside of the repository,
 they are still integrity checked in the same manner as normal files.
 
 The files stored outside of the repository are downloaded on demand,
 which reduces the time to clone, and possibly the local disk usage.
 This changes fundamental workflows in a DVCS, so careful thought
 should be given before deploying it.  :hg:`convert` can be used to
 convert LFS repositories to normal repositories that no longer
 require this extension, and do so without changing the commit hashes.
 This allows the extension to be disabled if the centralized workflow
 becomes burdensome.  However, the pre and post convert clones will
 not be able to communicate with each other unless the extension is
 enabled on both.
 
 To start a new repository, or to add LFS files to an existing one, just
 create an ``.hglfs`` file as described below in the root directory of
 the repository.  Typically, this file should be put under version
 control, so that the settings will propagate to other repositories with
 push and pull.  During any commit, Mercurial will consult this file to
 determine if an added or modified file should be stored externally.  The
 type of storage depends on the characteristics of the file at each
 commit.  A file that is near a size threshold may switch back and forth
 between LFS and normal storage, as needed.
 
 Alternately, both normal repositories and largefile controlled
 repositories can be converted to LFS by using :hg:`convert` and the
 ``lfs.track`` config option described below.  The ``.hglfs`` file
 should then be created and added, to control subsequent LFS selection.
 The hashes are also unchanged in this case.  The LFS and non-LFS
 repositories can be distinguished because the LFS repository will
 abort any command if this extension is disabled.
 
 Committed LFS files are held locally, until the repository is pushed.
 Prior to pushing the normal repository data, the LFS files that are
 tracked by the outgoing commits are automatically uploaded to the
 configured central server.  No LFS files are transferred on
 :hg:`pull` or :hg:`clone`.  Instead, the files are downloaded on
 demand as they need to be read, if a cached copy cannot be found
 locally.  Both committing and downloading an LFS file will link the
 file to a usercache, to speed up future access.  See the `usercache`
 config setting described below.
 
 .hglfs::
 
     The extension reads its configuration from a versioned ``.hglfs``
     configuration file found in the root of the working directory. The
     ``.hglfs`` file uses the same syntax as all other Mercurial
     configuration files. It uses a single section, ``[track]``.
 
     The ``[track]`` section specifies which files are stored as LFS (or
     not). Each line is keyed by a file pattern, with a predicate value.
     The first file pattern match is used, so put more specific patterns
     first.  The available predicates are ``all()``, ``none()``, and
     ``size()``. See "hg help filesets.size" for the latter.
 
     Example versioned ``.hglfs`` file::
 
       [track]
       # No Makefile or python file, anywhere, will be LFS
       **Makefile = none()
       **.py = none()
 
       **.zip = all()
       **.exe = size(">1MB")
 
       # Catchall for everything not matched above
       ** = size(">10MB")
 
 Configs::
 
     [lfs]
     # Remote endpoint. Multiple protocols are supported:
     # - http(s)://user:[e-mail]/path
     #   git-lfs endpoint
     # - file:///tmp/path
     #   local filesystem, usually for testing
     # if unset, lfs will assume the repository at ``paths.default`` also handles
     # blob storage for http(s) URLs.  Otherwise, lfs will prompt to set this
     # when it must use this value.
     # (default: unset)
     url = https://example.com/repo.git/info/lfs
 
     # Which files to track in LFS.  Path tests are "**.extname" for file
     # extensions, and "path:under/some/directory" for path prefix.  Both
     # are relative to the repository root.
     # File size can be tested with the "size()" fileset, and tests can be
     # joined with fileset operators.  (See "hg help filesets.operators".)
     #
     # Some examples:
     # - all()                       # everything
     # - none()                      # nothing
     # - size(">20MB")               # larger than 20MB
     # - !**.txt                     # anything not a *.txt file
     # - **.zip | **.tar.gz | **.7z  # some types of compressed files
     # - path:bin                    # files under "bin" in the project root
     # - (**.php & size(">2MB")) | (**.js & size(">5MB")) | **.tar.gz
     #     | (path:bin & !path:/bin/README) | size(">1GB")
     # (default: none())
     #
     # This is ignored if there is a tracked '.hglfs' file, and this setting
     # will eventually be deprecated and removed.
     track = size(">10M")
 
     # how many times to retry before giving up on transferring an object
     retry = 5
 
     # the local directory to store lfs files for sharing across local clones.
     # If not set, the cache is located in an OS specific cache location.
     usercache = /path/to/global/cache
 """
 
 from __future__ import absolute_import
 
 from mercurial.i18n import _
 
 from mercurial import (
     bundle2,
     changegroup,
     cmdutil,
     config,
     context,
     error,
     exchange,
     extensions,
     filelog,
     fileset,
     hg,
     localrepo,
     minifileset,
     node,
     pycompat,
     registrar,
     revlog,
     scmutil,
     templateutil,
     upgrade,
     util,
     vfs as vfsmod,
     wireproto,
     wireprotoserver,
 )
 
 from . import (
     blobstore,
     wireprotolfsserver,
     wrapper,
 )
 
 # 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'
 
 configtable = {}
 configitem = registrar.configitem(configtable)
 
 configitem('experimental', 'lfs.serve',
     default=True,
 )
 configitem('experimental', 'lfs.user-agent',
     default=None,
 )
+configitem('experimental', 'lfs.disableusercache',
+    default=False,
+)
 configitem('experimental', 'lfs.worker-enable',
     default=False,
 )
 
 configitem('lfs', 'url',
     default=None,
 )
 configitem('lfs', 'usercache',
     default=None,
 )
 # Deprecated
 configitem('lfs', 'threshold',
     default=None,
 )
 configitem('lfs', 'track',
     default='none()',
 )
 configitem('lfs', 'retry',
     default=5,
 )
 
 cmdtable = {}
 command = registrar.command(cmdtable)
 
 templatekeyword = registrar.templatekeyword()
 filesetpredicate = registrar.filesetpredicate()
 
 def featuresetup(ui, supported):
     # don't die on seeing a repo with the lfs requirement
     supported |= {'lfs'}
 
 def uisetup(ui):
     localrepo.featuresetupfuncs.add(featuresetup)
 
 def reposetup(ui, repo):
     # Nothing to do with a remote repo
     if not repo.local():
         return
 
     repo.svfs.lfslocalblobstore = blobstore.local(repo)
     repo.svfs.lfsremoteblobstore = blobstore.remote(repo)
 
     class lfsrepo(repo.__class__):
         @localrepo.unfilteredmethod
         def commitctx(self, ctx, error=False):
             repo.svfs.options['lfstrack'] = _trackedmatcher(self)
             return super(lfsrepo, self).commitctx(ctx, error)
 
     repo.__class__ = lfsrepo
 
     if 'lfs' not in repo.requirements:
         def checkrequireslfs(ui, repo, **kwargs):
             if 'lfs' not in repo.requirements:
                 last = kwargs.get(r'node_last')
                 _bin = node.bin
                 if last:
                     s = repo.set('%n:%n', _bin(kwargs[r'node']), _bin(last))
                 else:
                     s = repo.set('%n', _bin(kwargs[r'node']))
             match = repo.narrowmatch()
             for ctx in s:
                 # TODO: is there a way to just walk the files in the commit?
                 if any(ctx[f].islfs() for f in ctx.files()
                        if f in ctx and match(f)):
                     repo.requirements.add('lfs')
                     repo._writerequirements()
                     repo.prepushoutgoinghooks.add('lfs', wrapper.prepush)
                     break
 
         ui.setconfig('hooks', 'commit.lfs', checkrequireslfs, 'lfs')
         ui.setconfig('hooks', 'pretxnchangegroup.lfs', checkrequireslfs, 'lfs')
     else:
         repo.prepushoutgoinghooks.add('lfs', wrapper.prepush)
 
 def _trackedmatcher(repo):
     """Return a function (path, size) -> bool indicating whether or not to
     track a given file with lfs."""
     if not repo.wvfs.exists('.hglfs'):
         # No '.hglfs' in wdir.  Fallback to config for now.
         trackspec = repo.ui.config('lfs', 'track')
 
         # deprecated config: lfs.threshold
         threshold = repo.ui.configbytes('lfs', 'threshold')
         if threshold:
             fileset.parse(trackspec)  # make sure syntax errors are confined
             trackspec = "(%s) | size('>%d')" % (trackspec, threshold)
 
         return minifileset.compile(trackspec)
 
     data = repo.wvfs.tryread('.hglfs')
     if not data:
         return lambda p, s: False
 
     # Parse errors here will abort with a message that points to the .hglfs file
     # and line number.
     cfg = config.config()
     cfg.parse('.hglfs', data)
 
     try:
         rules = [(minifileset.compile(pattern), minifileset.compile(rule))
                  for pattern, rule in cfg.items('track')]
     except error.ParseError as e:
         # The original exception gives no indicator that the error is in the
         # .hglfs file, so add that.
 
         # TODO: See if the line number of the file can be made available.
         raise error.Abort(_('parse error in .hglfs: %s') % e)
 
     def _match(path, size):
         for pat, rule in rules:
             if pat(path, size):
                 return rule(path, size)
 
         return False
 
     return _match
 
 def wrapfilelog(filelog):
     wrapfunction = extensions.wrapfunction
 
     wrapfunction(filelog, 'addrevision', wrapper.filelogaddrevision)
     wrapfunction(filelog, 'renamed', wrapper.filelogrenamed)
     wrapfunction(filelog, 'size', wrapper.filelogsize)
 
 def extsetup(ui):
     wrapfilelog(filelog.filelog)
 
     wrapfunction = extensions.wrapfunction
 
     wrapfunction(cmdutil, '_updatecatformatter', wrapper._updatecatformatter)
     wrapfunction(scmutil, 'wrapconvertsink', wrapper.convertsink)
 
     wrapfunction(upgrade, '_finishdatamigration',
                  wrapper.upgradefinishdatamigration)
 
     wrapfunction(upgrade, 'preservedrequirements',
                  wrapper.upgraderequirements)
 
     wrapfunction(upgrade, 'supporteddestrequirements',
                  wrapper.upgraderequirements)
 
     wrapfunction(changegroup,
                  'allsupportedversions',
                  wrapper.allsupportedversions)
 
     wrapfunction(exchange, 'push', wrapper.push)
     wrapfunction(wireproto, '_capabilities', wrapper._capabilities)
     wrapfunction(wireprotoserver, 'handlewsgirequest',
                  wireprotolfsserver.handlewsgirequest)
 
     wrapfunction(context.basefilectx, 'cmp', wrapper.filectxcmp)
     wrapfunction(context.basefilectx, 'isbinary', wrapper.filectxisbinary)
     context.basefilectx.islfs = wrapper.filectxislfs
 
     revlog.addflagprocessor(
         revlog.REVIDX_EXTSTORED,
         (
             wrapper.readfromstore,
             wrapper.writetostore,
             wrapper.bypasscheckhash,
         ),
     )
 
     wrapfunction(hg, 'clone', wrapper.hgclone)
     wrapfunction(hg, 'postshare', wrapper.hgpostshare)
 
     scmutil.fileprefetchhooks.add('lfs', wrapper._prefetchfiles)
 
     # Make bundle choose changegroup3 instead of changegroup2. This affects
     # "hg bundle" command. Note: it does not cover all bundle formats like
     # "packed1". Using "packed1" with lfs will likely cause trouble.
     exchange._bundlespeccontentopts["v2"]["cg.version"] = "03"
 
     # bundlerepo uses "vfsmod.readonlyvfs(othervfs)", we need to make sure lfs
     # options and blob stores are passed from othervfs to the new readonlyvfs.
     wrapfunction(vfsmod.readonlyvfs, '__init__', wrapper.vfsinit)
 
     # when writing a bundle via "hg bundle" command, upload related LFS blobs
     wrapfunction(bundle2, 'writenewbundle', wrapper.writenewbundle)
 
 @filesetpredicate('lfs()', callstatus=True)
 def lfsfileset(mctx, x):
     """File that uses LFS storage."""
     # i18n: "lfs" is a keyword
     fileset.getargs(x, 0, 0, _("lfs takes no arguments"))
     return [f for f in mctx.subset
             if wrapper.pointerfromctx(mctx.ctx, f, removed=True) is not None]
 
 @templatekeyword('lfs_files', requires={'ctx'})
 def lfsfiles(context, mapping):
     """List of strings. All files modified, added, or removed by this
     changeset."""
     ctx = context.resource(mapping, 'ctx')
 
     pointers = wrapper.pointersfromctx(ctx, removed=True) # {path: pointer}
     files = sorted(pointers.keys())
 
     def pointer(v):
         # In the file spec, version is first and the other keys are sorted.
         sortkeyfunc = lambda x: (x[0] != 'version', x)
         items = sorted(pointers[v].iteritems(), key=sortkeyfunc)
         return util.sortdict(items)
 
     makemap = lambda v: {
         'file': v,
         'lfsoid': pointers[v].oid() if pointers[v] else None,
         'lfspointer': templateutil.hybriddict(pointer(v)),
     }
 
     # TODO: make the separator ', '?
     f = templateutil._showcompatlist(context, mapping, 'lfs_file', files)
     return templateutil.hybrid(f, files, makemap, pycompat.identity)
 
 @command('debuglfsupload',
          [('r', 'rev', [], _('upload large files introduced by REV'))])
 def debuglfsupload(ui, repo, **opts):
     """upload lfs blobs added by the working copy parent or given revisions"""
     revs = opts.get(r'rev', [])
     pointers = wrapper.extractpointers(repo, scmutil.revrange(repo, revs))
     wrapper.uploadblobs(repo, pointers)

hgext/lfs/blobstore.py

@@ -1,564 +1,562
 # blobstore.py - local and remote (speaking Git-LFS protocol) blob storages
 #
 # Copyright 2017 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import errno
 import hashlib
 import json
 import os
 import re
 import socket
 
 from mercurial.i18n import _
 
 from mercurial import (
     error,
     pathutil,
     pycompat,
     url as urlmod,
     util,
     vfs as vfsmod,
     worker,
 )
 
 from ..largefiles import lfutil
 
 # 64 bytes for SHA256
 _lfsre = re.compile(br'\A[a-f0-9]{64}\Z')
 
 class lfsvfs(vfsmod.vfs):
     def join(self, path):
         """split the path at first two characters, like: XX/XXXXX..."""
         if not _lfsre.match(path):
             raise error.ProgrammingError('unexpected lfs path: %s' % path)
         return super(lfsvfs, self).join(path[0:2], path[2:])
 
     def walk(self, path=None, onerror=None):
         """Yield (dirpath, [], oids) tuple for blobs under path
 
         Oids only exist in the root of this vfs, so dirpath is always ''.
         """
         root = os.path.normpath(self.base)
         # when dirpath == root, dirpath[prefixlen:] becomes empty
         # because len(dirpath) < prefixlen.
         prefixlen = len(pathutil.normasprefix(root))
         oids = []
 
         for dirpath, dirs, files in os.walk(self.reljoin(self.base, path or ''),
                                             onerror=onerror):
             dirpath = dirpath[prefixlen:]
 
             # Silently skip unexpected files and directories
             if len(dirpath) == 2:
                 oids.extend([dirpath + f for f in files
                              if _lfsre.match(dirpath + f)])
 
         yield ('', [], oids)
 
 class nullvfs(lfsvfs):
     def __init__(self):
         pass
 
     def exists(self, oid):
         return False
 
     def read(self, oid):
         # store.read() calls into here if the blob doesn't exist in its
         # self.vfs.  Raise the same error as a normal vfs when asked to read a
         # file that doesn't exist.  The only difference is the full file path
         # isn't available in the error.
         raise IOError(errno.ENOENT, '%s: No such file or directory' % oid)
 
     def walk(self, path=None, onerror=None):
         return ('', [], [])
 
     def write(self, oid, data):
         pass
 
 class filewithprogress(object):
     """a file-like object that supports __len__ and read.
 
     Useful to provide progress information for how many bytes are read.
     """
 
     def __init__(self, fp, callback):
         self._fp = fp
         self._callback = callback # func(readsize)
         fp.seek(0, os.SEEK_END)
         self._len = fp.tell()
         fp.seek(0)
 
     def __len__(self):
         return self._len
 
     def read(self, size):
         if self._fp is None:
             return b''
         data = self._fp.read(size)
         if data:
             if self._callback:
                 self._callback(len(data))
         else:
             self._fp.close()
             self._fp = None
         return data
 
 class local(object):
     """Local blobstore for large file contents.
 
     This blobstore is used both as a cache and as a staging area for large blobs
     to be uploaded to the remote blobstore.
     """
 
     def __init__(self, repo):
         fullpath = repo.svfs.join('lfs/objects')
         self.vfs = lfsvfs(fullpath)
-        usercache = util.url(lfutil._usercachedir(repo.ui, 'lfs'))
-        if usercache.scheme in (None, 'file'):
-            self.cachevfs = lfsvfs(usercache.localpath())
-        elif usercache.scheme == 'null':
+
+        if repo.ui.configbool('experimental', 'lfs.disableusercache'):
             self.cachevfs = nullvfs()
         else:
-            raise error.Abort(_('unknown lfs cache scheme: %s')
-                              % usercache.scheme)
+            usercache = lfutil._usercachedir(repo.ui, 'lfs')
+            self.cachevfs = lfsvfs(usercache)
         self.ui = repo.ui
 
     def open(self, oid):
         """Open a read-only file descriptor to the named blob, in either the
         usercache or the local store."""
         # The usercache is the most likely place to hold the file.  Commit will
         # write to both it and the local store, as will anything that downloads
         # the blobs.  However, things like clone without an update won't
         # populate the local store.  For an init + push of a local clone,
         # the usercache is the only place it _could_ be.  If not present, the
         # missing file msg here will indicate the local repo, not the usercache.
         if self.cachevfs.exists(oid):
             return self.cachevfs(oid, 'rb')
 
         return self.vfs(oid, 'rb')
 
     def download(self, oid, src):
         """Read the blob from the remote source in chunks, verify the content,
         and write to this local blobstore."""
         sha256 = hashlib.sha256()
 
         with self.vfs(oid, 'wb', atomictemp=True) as fp:
             for chunk in util.filechunkiter(src, size=1048576):
                 fp.write(chunk)
                 sha256.update(chunk)
 
             realoid = sha256.hexdigest()
             if realoid != oid:
                 raise error.Abort(_('corrupt remote lfs object: %s') % oid)
 
         self._linktousercache(oid)
 
     def write(self, oid, data):
         """Write blob to local blobstore.
 
         This should only be called from the filelog during a commit or similar.
         As such, there is no need to verify the data.  Imports from a remote
         store must use ``download()`` instead."""
         with self.vfs(oid, 'wb', atomictemp=True) as fp:
             fp.write(data)
 
         self._linktousercache(oid)
 
     def _linktousercache(self, oid):
         # XXX: should we verify the content of the cache, and hardlink back to
         # the local store on success, but truncate, write and link on failure?
         if (not self.cachevfs.exists(oid)
             and not isinstance(self.cachevfs, nullvfs)):
             self.ui.note(_('lfs: adding %s to the usercache\n') % oid)
             lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid))
 
     def read(self, oid, verify=True):
         """Read blob from local blobstore."""
         if not self.vfs.exists(oid):
             blob = self._read(self.cachevfs, oid, verify)
 
             # Even if revlog will verify the content, it needs to be verified
             # now before making the hardlink to avoid propagating corrupt blobs.
             # Don't abort if corruption is detected, because `hg verify` will
             # give more useful info about the corruption- simply don't add the
             # hardlink.
             if verify or hashlib.sha256(blob).hexdigest() == oid:
                 self.ui.note(_('lfs: found %s in the usercache\n') % oid)
                 lfutil.link(self.cachevfs.join(oid), self.vfs.join(oid))
         else:
             self.ui.note(_('lfs: found %s in the local lfs store\n') % oid)
             blob = self._read(self.vfs, oid, verify)
         return blob
 
     def _read(self, vfs, oid, verify):
         """Read blob (after verifying) from the given store"""
         blob = vfs.read(oid)
         if verify:
             _verify(oid, blob)
         return blob
 
     def verify(self, oid):
         """Indicate whether or not the hash of the underlying file matches its
         name."""
         sha256 = hashlib.sha256()
 
         with self.open(oid) as fp:
             for chunk in util.filechunkiter(fp, size=1048576):
                 sha256.update(chunk)
 
         return oid == sha256.hexdigest()
 
     def has(self, oid):
         """Returns True if the local blobstore contains the requested blob,
         False otherwise."""
         return self.cachevfs.exists(oid) or self.vfs.exists(oid)
 
 class _gitlfsremote(object):
 
     def __init__(self, repo, url):
         ui = repo.ui
         self.ui = ui
         baseurl, authinfo = url.authinfo()
         self.baseurl = baseurl.rstrip('/')
         useragent = repo.ui.config('experimental', 'lfs.user-agent')
         if not useragent:
             useragent = 'git-lfs/2.3.4 (Mercurial %s)' % util.version()
         self.urlopener = urlmod.opener(ui, authinfo, useragent)
         self.retry = ui.configint('lfs', 'retry')
 
     def writebatch(self, pointers, fromstore):
         """Batch upload from local to remote blobstore."""
         self._batch(_deduplicate(pointers), fromstore, 'upload')
 
     def readbatch(self, pointers, tostore):
         """Batch download from remote to local blostore."""
         self._batch(_deduplicate(pointers), tostore, 'download')
 
     def _batchrequest(self, pointers, action):
         """Get metadata about objects pointed by pointers for given action
 
         Return decoded JSON object like {'objects': [{'oid': '', 'size': 1}]}
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/batch.md
         """
         objects = [{'oid': p.oid(), 'size': p.size()} for p in pointers]
         requestdata = json.dumps({
             'objects': objects,
             'operation': action,
         })
         batchreq = util.urlreq.request('%s/objects/batch' % self.baseurl,
                                        data=requestdata)
         batchreq.add_header('Accept', 'application/vnd.git-lfs+json')
         batchreq.add_header('Content-Type', 'application/vnd.git-lfs+json')
         try:
             rsp = self.urlopener.open(batchreq)
             rawjson = rsp.read()
         except util.urlerr.httperror as ex:
             raise LfsRemoteError(_('LFS HTTP error: %s (action=%s)')
                                  % (ex, action))
         try:
             response = json.loads(rawjson)
         except ValueError:
             raise LfsRemoteError(_('LFS server returns invalid JSON: %s')
                                  % rawjson)
 
         if self.ui.debugflag:
             self.ui.debug('Status: %d\n' % rsp.status)
             # lfs-test-server and hg serve return headers in different order
             self.ui.debug('%s\n'
                           % '\n'.join(sorted(str(rsp.info()).splitlines())))
 
             if 'objects' in response:
                 response['objects'] = sorted(response['objects'],
                                              key=lambda p: p['oid'])
             self.ui.debug('%s\n'
                           % json.dumps(response, indent=2,
                                        separators=('', ': '), sort_keys=True))
 
         return response
 
     def _checkforservererror(self, pointers, responses, action):
         """Scans errors from objects
 
         Raises LfsRemoteError if any objects have an error"""
         for response in responses:
             # The server should return 404 when objects cannot be found. Some
             # server implementation (ex. lfs-test-server)  does not set "error"
             # but just removes "download" from "actions". Treat that case
             # as the same as 404 error.
             if 'error' not in response:
                 if (action == 'download'
                     and action not in response.get('actions', [])):
                     code = 404
                 else:
                     continue
             else:
                 # An error dict without a code doesn't make much sense, so
                 # treat as a server error.
                 code = response.get('error').get('code', 500)
 
             ptrmap = {p.oid(): p for p in pointers}
             p = ptrmap.get(response['oid'], None)
             if p:
                 filename = getattr(p, 'filename', 'unknown')
                 errors = {
                     404: 'The object does not exist',
                     410: 'The object was removed by the owner',
                     422: 'Validation error',
                     500: 'Internal server error',
                 }
                 msg = errors.get(code, 'status code %d' % code)
                 raise LfsRemoteError(_('LFS server error for "%s": %s')
                                      % (filename, msg))
             else:
                 raise LfsRemoteError(
                     _('LFS server error. Unsolicited response for oid %s')
                     % response['oid'])
 
     def _extractobjects(self, response, pointers, action):
         """extract objects from response of the batch API
 
         response: parsed JSON object returned by batch API
         return response['objects'] filtered by action
         raise if any object has an error
         """
         # Scan errors from objects - fail early
         objects = response.get('objects', [])
         self._checkforservererror(pointers, objects, action)
 
         # Filter objects with given action. Practically, this skips uploading
         # objects which exist in the server.
         filteredobjects = [o for o in objects if action in o.get('actions', [])]
 
         return filteredobjects
 
     def _basictransfer(self, obj, action, localstore):
         """Download or upload a single object using basic transfer protocol
 
         obj: dict, an object description returned by batch API
         action: string, one of ['upload', 'download']
         localstore: blobstore.local
 
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/\
         basic-transfers.md
         """
         oid = pycompat.bytestr(obj['oid'])
 
         href = pycompat.bytestr(obj['actions'][action].get('href'))
         headers = obj['actions'][action].get('header', {}).items()
 
         request = util.urlreq.request(href)
         if action == 'upload':
             # If uploading blobs, read data from local blobstore.
             if not localstore.verify(oid):
                 raise error.Abort(_('detected corrupt lfs object: %s') % oid,
                                   hint=_('run hg verify'))
             request.data = filewithprogress(localstore.open(oid), None)
             request.get_method = lambda: 'PUT'
             request.add_header('Content-Type', 'application/octet-stream')
 
         for k, v in headers:
             request.add_header(k, v)
 
         response = b''
         try:
             req = self.urlopener.open(request)
 
             if self.ui.debugflag:
                 self.ui.debug('Status: %d\n' % req.status)
                 # lfs-test-server and hg serve return headers in different order
                 self.ui.debug('%s\n'
                               % '\n'.join(sorted(str(req.info()).splitlines())))
 
             if action == 'download':
                 # If downloading blobs, store downloaded data to local blobstore
                 localstore.download(oid, req)
             else:
                 while True:
                     data = req.read(1048576)
                     if not data:
                         break
                     response += data
                 if response:
                     self.ui.debug('lfs %s response: %s' % (action, response))
         except util.urlerr.httperror as ex:
             if self.ui.debugflag:
                 self.ui.debug('%s: %s\n' % (oid, ex.read()))
             raise LfsRemoteError(_('HTTP error: %s (oid=%s, action=%s)')
                                  % (ex, oid, action))
 
     def _batch(self, pointers, localstore, action):
         if action not in ['upload', 'download']:
             raise error.ProgrammingError('invalid Git-LFS action: %s' % action)
 
         response = self._batchrequest(pointers, action)
         objects = self._extractobjects(response, pointers, action)
         total = sum(x.get('size', 0) for x in objects)
         sizes = {}
         for obj in objects:
             sizes[obj.get('oid')] = obj.get('size', 0)
         topic = {'upload': _('lfs uploading'),
                  'download': _('lfs downloading')}[action]
         if len(objects) > 1:
             self.ui.note(_('lfs: need to transfer %d objects (%s)\n')
                          % (len(objects), util.bytecount(total)))
         self.ui.progress(topic, 0, total=total)
         def transfer(chunk):
             for obj in chunk:
                 objsize = obj.get('size', 0)
                 if self.ui.verbose:
                     if action == 'download':
                         msg = _('lfs: downloading %s (%s)\n')
                     elif action == 'upload':
                         msg = _('lfs: uploading %s (%s)\n')
                     self.ui.note(msg % (obj.get('oid'),
                                  util.bytecount(objsize)))
                 retry = self.retry
                 while True:
                     try:
                         self._basictransfer(obj, action, localstore)
                         yield 1, obj.get('oid')
                         break
                     except socket.error as ex:
                         if retry > 0:
                             self.ui.note(
                                 _('lfs: failed: %r (remaining retry %d)\n')
                                 % (ex, retry))
                             retry -= 1
                             continue
                         raise
 
         # Until https multiplexing gets sorted out
         if self.ui.configbool('experimental', 'lfs.worker-enable'):
             oids = worker.worker(self.ui, 0.1, transfer, (),
                                  sorted(objects, key=lambda o: o.get('oid')))
         else:
             oids = transfer(sorted(objects, key=lambda o: o.get('oid')))
 
         processed = 0
         blobs = 0
         for _one, oid in oids:
             processed += sizes[oid]
             blobs += 1
             self.ui.progress(topic, processed, total=total)
             self.ui.note(_('lfs: processed: %s\n') % oid)
         self.ui.progress(topic, pos=None, total=total)
 
         if blobs > 0:
             if action == 'upload':
                 self.ui.status(_('lfs: uploaded %d files (%s)\n')
                                % (blobs, util.bytecount(processed)))
             # TODO: coalesce the download requests, and comment this in
             #elif action == 'download':
             #    self.ui.status(_('lfs: downloaded %d files (%s)\n')
             #                   % (blobs, util.bytecount(processed)))
 
     def __del__(self):
         # copied from mercurial/httppeer.py
         urlopener = getattr(self, 'urlopener', None)
         if urlopener:
             for h in urlopener.handlers:
                 h.close()
                 getattr(h, "close_all", lambda : None)()
 
 class _dummyremote(object):
     """Dummy store storing blobs to temp directory."""
 
     def __init__(self, repo, url):
         fullpath = repo.vfs.join('lfs', url.path)
         self.vfs = lfsvfs(fullpath)
 
     def writebatch(self, pointers, fromstore):
         for p in _deduplicate(pointers):
             content = fromstore.read(p.oid(), verify=True)
             with self.vfs(p.oid(), 'wb', atomictemp=True) as fp:
                 fp.write(content)
 
     def readbatch(self, pointers, tostore):
         for p in _deduplicate(pointers):
             with self.vfs(p.oid(), 'rb') as fp:
                 tostore.download(p.oid(), fp)
 
 class _nullremote(object):
     """Null store storing blobs to /dev/null."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore):
         pass
 
     def readbatch(self, pointers, tostore):
         pass
 
 class _promptremote(object):
     """Prompt user to set lfs.url when accessed."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore, ui=None):
         self._prompt()
 
     def readbatch(self, pointers, tostore, ui=None):
         self._prompt()
 
     def _prompt(self):
         raise error.Abort(_('lfs.url needs to be configured'))
 
 _storemap = {
     'https': _gitlfsremote,
     'http': _gitlfsremote,
     'file': _dummyremote,
     'null': _nullremote,
     None: _promptremote,
 }
 
 def _deduplicate(pointers):
     """Remove any duplicate oids that exist in the list"""
     reduced = util.sortdict()
     for p in pointers:
         reduced[p.oid()] = p
     return reduced.values()
 
 def _verify(oid, content):
     realoid = hashlib.sha256(content).hexdigest()
     if realoid != oid:
         raise error.Abort(_('detected corrupt lfs object: %s') % oid,
                           hint=_('run hg verify'))
 
 def remote(repo):
     """remotestore factory. return a store in _storemap depending on config
 
     If ``lfs.url`` is specified, use that remote endpoint.  Otherwise, try to
     infer the endpoint, based on the remote repository using the same path
     adjustments as git.  As an extension, 'http' is supported as well so that
     ``hg serve`` works out of the box.
 
     https://github.com/git-lfs/git-lfs/blob/master/docs/api/server-discovery.md
     """
     url = util.url(repo.ui.config('lfs', 'url') or '')
     if url.scheme is None:
         # TODO: investigate 'paths.remote:lfsurl' style path customization,
         # and fall back to inferring from 'paths.remote' if unspecified.
         defaulturl = util.url(repo.ui.config('paths', 'default') or b'')
 
         # TODO: support local paths as well.
         # TODO: consider the ssh -> https transformation that git applies
         if defaulturl.scheme in (b'http', b'https'):
             defaulturl.path = defaulturl.path or b'' + b'.git/info/lfs'
 
             url = util.url(bytes(defaulturl))
             repo.ui.note(_('lfs: assuming remote store: %s\n') % url)
 
     scheme = url.scheme
     if scheme not in _storemap:
         raise error.Abort(_('lfs: unknown url scheme: %s') % scheme)
     return _storemap[scheme](repo, url)
 
 class LfsRemoteError(error.RevlogError):
     pass

tests/test-lfs-serve.t

@@ -1,283 +1,284
 #testcases lfsremote-on lfsremote-off
 #require serve no-reposimplestore
 
 This test splits `hg serve` with and without using the extension into separate
 tests cases.  The tests are broken down as follows, where "LFS"/"No-LFS"
 indicates whether or not there are commits that use an LFS file, and "D"/"E"
 indicates whether or not the extension is loaded.  The "X" cases are not tested
 individually, because the lfs requirement causes the process to bail early if
 the extension is disabled.
 
 .                        Server
 .
 .                    No-LFS        LFS
 .            +----------------------------+
 .            |   ||  D  |  E  |  D  |  E  |
 .            |---++=======================|
 .  C         | D || N/A | #1  |  X  | #4  |
 .  l    No   +---++-----------------------|
 .  i    LFS  | E || #2  | #2  |  X  | #5  |
 .  e         +---++-----------------------|
 .  n         | D ||  X  |  X  |  X  |  X  |
 .  t    LFS  |---++-----------------------|
 .            | E || #3  | #3  |  X  | #6  |
 .            |---++-----------------------+
 
   $ hg init server
   $ SERVER_REQUIRES="$TESTTMP/server/.hg/requires"
 
 Skip the experimental.changegroup3=True config.  Failure to agree on this comes
 first, and causes a "ValueError: no common changegroup version" or "abort:
 HTTP Error 500: Internal Server Error", if the extension is only loaded on one
 side.  If that *is* enabled, the subsequent failure is "abort: missing processor
 for flag '0x2000'!" if the extension is only loaded on one side (possibly also
 masked by the Internal Server Error message).
   $ cat >> $HGRCPATH <<EOF
+  > [experimental]
+  > lfs.disableusercache = True
   > [lfs]
-  > usercache = null://
   > threshold=10
   > [web]
   > allow_push=*
   > push_ssl=False
   > EOF
 
 #if lfsremote-on
   $ hg --config extensions.lfs= -R server \
   >    serve -p $HGPORT -d --pid-file=hg.pid --errorlog=$TESTTMP/errors.log
 #else
   $ hg --config extensions.lfs=! -R server \
   >    serve -p $HGPORT -d --pid-file=hg.pid --errorlog=$TESTTMP/errors.log
 #endif
 
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg clone -q http://localhost:$HGPORT client
   $ grep 'lfs' client/.hg/requires $SERVER_REQUIRES
   [1]
 
 --------------------------------------------------------------------------------
 Case #1: client with non-lfs content and the extension disabled; server with
 non-lfs content, and the extension enabled.
 
   $ cd client
   $ echo 'non-lfs' > nonlfs.txt
   $ hg ci -Aqm 'non-lfs'
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   [1]
 
 #if lfsremote-on
 
   $ hg push -q
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client1_clone
   $ grep 'lfs' $TESTTMP/client1_clone/.hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg init $TESTTMP/client1_pull
   $ hg -R $TESTTMP/client1_pull pull -q http://localhost:$HGPORT
   $ grep 'lfs' $TESTTMP/client1_pull/.hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg identify http://localhost:$HGPORT
   d437e1d24fbd
 
 #endif
 
 --------------------------------------------------------------------------------
 Case #2: client with non-lfs content and the extension enabled; server with
 non-lfs content, and the extension state controlled by #testcases.
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > lfs =
   > EOF
   $ echo 'non-lfs' > nonlfs2.txt
   $ hg ci -Aqm 'non-lfs file with lfs client'
 
 Since no lfs content has been added yet, the push is allowed, even when the
 extension is not enabled remotely.
 
   $ hg push -q
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client2_clone
   $ grep 'lfs' $TESTTMP/client2_clone/.hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg init $TESTTMP/client2_pull
   $ hg -R $TESTTMP/client2_pull pull -q http://localhost:$HGPORT
   $ grep 'lfs' $TESTTMP/client2_pull/.hg/requires $SERVER_REQUIRES
   [1]
 
   $ hg identify http://localhost:$HGPORT
   1477875038c6
 
 --------------------------------------------------------------------------------
 Case #3: client with lfs content and the extension enabled; server with
 non-lfs content, and the extension state controlled by #testcases.  The server
 should have an 'lfs' requirement after it picks up its first commit with a blob.
 
   $ echo 'this is a big lfs file' > lfs.bin
   $ hg ci -Aqm 'lfs'
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   .hg/requires:lfs
 
 #if lfsremote-off
   $ hg push -q
   abort: required features are not supported in the destination: lfs
   (enable the lfs extension on the server)
   [255]
 #else
   $ hg push -q
 #endif
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   .hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs (lfsremote-on !)
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client3_clone
   $ grep 'lfs' $TESTTMP/client3_clone/.hg/requires $SERVER_REQUIRES || true
   $TESTTMP/client3_clone/.hg/requires:lfs (lfsremote-on !)
   $TESTTMP/server/.hg/requires:lfs (lfsremote-on !)
 
   $ hg init $TESTTMP/client3_pull
   $ hg -R $TESTTMP/client3_pull pull -q http://localhost:$HGPORT
   $ grep 'lfs' $TESTTMP/client3_pull/.hg/requires $SERVER_REQUIRES || true
   $TESTTMP/client3_pull/.hg/requires:lfs (lfsremote-on !)
   $TESTTMP/server/.hg/requires:lfs (lfsremote-on !)
 
 The difference here is the push failed above when the extension isn't
 enabled on the server.
   $ hg identify http://localhost:$HGPORT
   8374dc4052cb (lfsremote-on !)
   1477875038c6 (lfsremote-off !)
 
 Don't bother testing the lfsremote-off cases- the server won't be able
 to launch if there's lfs content and the extension is disabled.
 
 #if lfsremote-on
 
 --------------------------------------------------------------------------------
 Case #4: client with non-lfs content and the extension disabled; server with
 lfs content, and the extension enabled.
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > lfs = !
   > EOF
 
   $ hg init $TESTTMP/client4
   $ cd $TESTTMP/client4
   $ cat >> .hg/hgrc <<EOF
   > [paths]
   > default = http://localhost:$HGPORT
   > EOF
   $ echo 'non-lfs' > nonlfs2.txt
   $ hg ci -Aqm 'non-lfs'
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg push -q --force
   warning: repository is unrelated
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   $TESTTMP/server/.hg/requires:lfs
 
 TODO: fail more gracefully.
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client4_clone
   abort: HTTP Error 500: Internal Server Error
   [255]
   $ grep 'lfs' $TESTTMP/client4_clone/.hg/requires $SERVER_REQUIRES
   grep: $TESTTMP/client4_clone/.hg/requires: $ENOENT$
   $TESTTMP/server/.hg/requires:lfs
   [2]
 
 TODO: fail more gracefully.
 
   $ hg init $TESTTMP/client4_pull
   $ hg -R $TESTTMP/client4_pull pull -q http://localhost:$HGPORT
   abort: HTTP Error 500: Internal Server Error
   [255]
   $ grep 'lfs' $TESTTMP/client4_pull/.hg/requires $SERVER_REQUIRES
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg identify http://localhost:$HGPORT
   03b080fa9d93
 
 --------------------------------------------------------------------------------
 Case #5: client with non-lfs content and the extension enabled; server with
 lfs content, and the extension enabled.
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > lfs =
   > EOF
   $ echo 'non-lfs' > nonlfs3.txt
   $ hg ci -Aqm 'non-lfs file with lfs client'
 
   $ hg push -q
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client5_clone
   $ grep 'lfs' $TESTTMP/client5_clone/.hg/requires $SERVER_REQUIRES
   $TESTTMP/client5_clone/.hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg init $TESTTMP/client5_pull
   $ hg -R $TESTTMP/client5_pull pull -q http://localhost:$HGPORT
   $ grep 'lfs' $TESTTMP/client5_pull/.hg/requires $SERVER_REQUIRES
   $TESTTMP/client5_pull/.hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg identify http://localhost:$HGPORT
   c729025cc5e3
 
 --------------------------------------------------------------------------------
 Case #6: client with lfs content and the extension enabled; server with
 lfs content, and the extension enabled.
 
   $ echo 'this is another lfs file' > lfs2.txt
   $ hg ci -Aqm 'lfs file with lfs client'
 
   $ hg push -q
   $ grep 'lfs' .hg/requires $SERVER_REQUIRES
   .hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg clone -q http://localhost:$HGPORT $TESTTMP/client6_clone
   $ grep 'lfs' $TESTTMP/client6_clone/.hg/requires $SERVER_REQUIRES
   $TESTTMP/client6_clone/.hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg init $TESTTMP/client6_pull
   $ hg -R $TESTTMP/client6_pull pull -q http://localhost:$HGPORT
   $ grep 'lfs' $TESTTMP/client6_pull/.hg/requires $SERVER_REQUIRES
   $TESTTMP/client6_pull/.hg/requires:lfs
   $TESTTMP/server/.hg/requires:lfs
 
   $ hg identify http://localhost:$HGPORT
   d3b84d50eacb
 
 --------------------------------------------------------------------------------
 Misc: process dies early if a requirement exists and the extension is disabled
 
   $ hg --config extensions.lfs=! summary
   abort: repository requires features unknown to this Mercurial: lfs!
   (see https://mercurial-scm.org/wiki/MissingRequirement for more information)
   [255]
 
 #endif
 
   $ $PYTHON $TESTDIR/killdaemons.py $DAEMON_PIDS
 
 #if lfsremote-on
   $ cat $TESTTMP/errors.log | grep '^[A-Z]'
   Traceback (most recent call last):
   ValueError: no common changegroup version
   Traceback (most recent call last):
   ValueError: no common changegroup version
 #else
   $ cat $TESTTMP/errors.log
 #endif