diff hgext/lfs/wrapper.py @ 35098:66c5a8cf2868

lfs: import the Facebook git-lfs client extension

The purpose of this is the same as the built-in largefiles extension- to handle
huge files outside of the normal storage system, generally to keep the amount of
data cloned to a lower amount.  There are several benefits of implementing the
git-lfs protocol, instead of using the largefiles extension:

  - Bitbucket and Github support (and probably wider support in 3rd party
    hosting sites in general). [1][2]

  - The number of hg internals monkey patched are several orders of magnitude
    lower, so it will be easier to reason about and maintain.  Future commands
    will likely just work, without requiring various wrappers.

  - The "standin" files are only written to the filelog, not the disk.  That
    should avoid weird edge cases where the largefile and standin files get out
    of sync. [3]  It also avoids the occasional printing of the "hidden" standin
    file in various messages.

  - Filesets like size() will work, even if the file isn't present.  (It always
    says 41 bytes for largefiles, whether present or not.)

The only place that I see where largefiles comes out on top is that it works
with `hg serve` for simple sharing, without external infrastructure.  Getting
lfs-test-server working was a hassle, and took awhile to figure out.  Maybe we
can do something to make it work in the future.

Long term, I expect that this will be highly preferred over largefiles.  But if
we are to recommend this to largefile users, there are some UI issues to
bikeshed.  Until they are resolved, I've marked this experimental, and am not
putting a pointer to this in the largefiles help.  The (non exhaustive) list of
issues I've seen so far are:

  - It isn't sufficient to just enable the largefiles extension- you have to
    explicitly add a file with --large before it will pay attention to the
    configured sizes and patterns on future adds.  The justification being that
    once you use it, you're stuck with it.  I've seen people confused by this,
    and haven't liked it myself.  But it's also saved me a few times.  Should we
    do something like have a specific enabling config setting that must be set
    in the local repo config, so that enabling this extension in the user or
    system hgrc doesn't silently start storing lfs files?

  - The largefiles extension adds a repo requirement when the first largefile is
    committed, so that the extension must always be enabled in the future.  This
    extension is not doing that, and since I only enabled it locally to avoid
    infecting other repos, I got a cryptic error about missing flag processors
    when I cloned.  Is there no repo requirement due to shallow/narrow clone
    considerations (or other future advanced things)?

  - In the (small amount of) reading I've done about the git implementation, it
    seems that the files and sizes are stored in a tracked .gitattributes file.
    I think a tracked file for this would be extremely useful for consistency
    across developers, but this kind of touches on the tracked hgrc file
    proposal a few months back.

  - The git client can specify file patterns, not just sizes.

  - The largefiles extension has a cache directory in the local repo, but also a
    system wide one.  We should probably implement a system wide cache too, so
    that multiple clones don't have to refetch the files from the server.

  - Jun mentioned other missing features, like SSH authentication, gc, etc.

The code corresponds to c0492b73c7ef in hg-experimental. [4]  The only tweaks
are to load the extension in the tests with 'lfs=' instead of
'lfs=$TESTDIR/../hgext3rd/lfs', change the import in the *.py test to hgext
(from hgext3rd), add the 'testedwith' declaration, and mark it experimental for
now.  The infinite-push, p4fastimport, and remotefilelog tests were left behind.

The devel-warnings for unregistered config options are not corrected yet, nor
are the import check warnings.

[1] https://www.mercurial-scm.org/pipermail/mercurial/2017-November/050699.html
[2] https://bitbucket.org/site/master/issues/3843/largefiles-support-bb-3903
[3] https://bz.mercurial-scm.org/show_bug.cgi?id=5738
[4] https://bitbucket.org/facebook/hg-experimental

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/hgext/lfs/wrapper.py	Tue Nov 14 00:06:23 2017 -0500
@@ -0,0 +1,247 @@
+# wrapper.py - methods wrapping core mercurial logic
+#
+# 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 hashlib
+
+from mercurial import (
+    error,
+    filelog,
+    revlog,
+    util,
+)
+from mercurial.i18n import _
+from mercurial.node import bin, nullid, short
+
+from . import (
+    blobstore,
+    pointer,
+)
+
+def supportedoutgoingversions(orig, repo):
+    versions = orig(repo)
+    versions.discard('01')
+    versions.discard('02')
+    versions.add('03')
+    return versions
+
+def allsupportedversions(orig, ui):
+    versions = orig(ui)
+    versions.add('03')
+    return versions
+
+def bypasscheckhash(self, text):
+    return False
+
+def readfromstore(self, text):
+    """Read filelog content from local blobstore transform for flagprocessor.
+
+    Default tranform for flagprocessor, returning contents from blobstore.
+    Returns a 2-typle (text, validatehash) where validatehash is True as the
+    contents of the blobstore should be checked using checkhash.
+    """
+    p = pointer.deserialize(text)
+    oid = p.oid()
+    store = self.opener.lfslocalblobstore
+    if not store.has(oid):
+        p.filename = getattr(self, 'indexfile', None)
+        self.opener.lfsremoteblobstore.readbatch([p], store)
+    text = store.read(oid)
+
+    # pack hg filelog metadata
+    hgmeta = {}
+    for k in p.keys():
+        if k.startswith('x-hg-'):
+            name = k[len('x-hg-'):]
+            hgmeta[name] = p[k]
+    if hgmeta or text.startswith('\1\n'):
+        text = filelog.packmeta(hgmeta, text)
+
+    return (text, True)
+
+def writetostore(self, text):
+    # hg filelog metadata (includes rename, etc)
+    hgmeta, offset = filelog.parsemeta(text)
+    if offset and offset > 0:
+        # lfs blob does not contain hg filelog metadata
+        text = text[offset:]
+
+    # git-lfs only supports sha256
+    oid = hashlib.sha256(text).hexdigest()
+    self.opener.lfslocalblobstore.write(oid, text)
+
+    # replace contents with metadata
+    longoid = 'sha256:%s' % oid
+    metadata = pointer.gitlfspointer(oid=longoid, size=str(len(text)))
+
+    # by default, we expect the content to be binary. however, LFS could also
+    # be used for non-binary content. add a special entry for non-binary data.
+    # this will be used by filectx.isbinary().
+    if not util.binary(text):
+        # not hg filelog metadata (affecting commit hash), no "x-hg-" prefix
+        metadata['x-is-binary'] = '0'
+
+    # translate hg filelog metadata to lfs metadata with "x-hg-" prefix
+    if hgmeta is not None:
+        for k, v in hgmeta.iteritems():
+            metadata['x-hg-%s' % k] = v
+
+    rawtext = metadata.serialize()
+    return (rawtext, False)
+
+def _islfs(rlog, node=None, rev=None):
+    if rev is None:
+        if node is None:
+            # both None - likely working copy content where node is not ready
+            return False
+        rev = rlog.rev(node)
+    else:
+        node = rlog.node(rev)
+    if node == nullid:
+        return False
+    flags = rlog.flags(rev)
+    return bool(flags & revlog.REVIDX_EXTSTORED)
+
+def filelogaddrevision(orig, self, text, transaction, link, p1, p2,
+                       cachedelta=None, node=None,
+                       flags=revlog.REVIDX_DEFAULT_FLAGS, **kwds):
+    threshold = self.opener.options['lfsthreshold']
+    textlen = len(text)
+    # exclude hg rename meta from file size
+    meta, offset = filelog.parsemeta(text)
+    if offset:
+        textlen -= offset
+
+    if threshold and textlen > threshold:
+        flags |= revlog.REVIDX_EXTSTORED
+
+    return orig(self, text, transaction, link, p1, p2, cachedelta=cachedelta,
+                node=node, flags=flags, **kwds)
+
+def filelogrenamed(orig, self, node):
+    if _islfs(self, node):
+        rawtext = self.revision(node, raw=True)
+        if not rawtext:
+            return False
+        metadata = pointer.deserialize(rawtext)
+        if 'x-hg-copy' in metadata and 'x-hg-copyrev' in metadata:
+            return metadata['x-hg-copy'], bin(metadata['x-hg-copyrev'])
+        else:
+            return False
+    return orig(self, node)
+
+def filelogsize(orig, self, rev):
+    if _islfs(self, rev=rev):
+        # fast path: use lfs metadata to answer size
+        rawtext = self.revision(rev, raw=True)
+        metadata = pointer.deserialize(rawtext)
+        return int(metadata['size'])
+    return orig(self, rev)
+
+def filectxcmp(orig, self, fctx):
+    """returns True if text is different than fctx"""
+    # some fctx (ex. hg-git) is not based on basefilectx and do not have islfs
+    if self.islfs() and getattr(fctx, 'islfs', lambda: False)():
+        # fast path: check LFS oid
+        p1 = pointer.deserialize(self.rawdata())
+        p2 = pointer.deserialize(fctx.rawdata())
+        return p1.oid() != p2.oid()
+    return orig(self, fctx)
+
+def filectxisbinary(orig, self):
+    if self.islfs():
+        # fast path: use lfs metadata to answer isbinary
+        metadata = pointer.deserialize(self.rawdata())
+        # if lfs metadata says nothing, assume it's binary by default
+        return bool(int(metadata.get('x-is-binary', 1)))
+    return orig(self)
+
+def filectxislfs(self):
+    return _islfs(self.filelog(), self.filenode())
+
+def vfsinit(orig, self, othervfs):
+    orig(self, othervfs)
+    # copy lfs related options
+    for k, v in othervfs.options.items():
+        if k.startswith('lfs'):
+            self.options[k] = v
+    # also copy lfs blobstores. note: this can run before reposetup, so lfs
+    # blobstore attributes are not always ready at this time.
+    for name in ['lfslocalblobstore', 'lfsremoteblobstore']:
+        if util.safehasattr(othervfs, name):
+            setattr(self, name, getattr(othervfs, name))
+
+def _canskipupload(repo):
+    # if remotestore is a null store, upload is a no-op and can be skipped
+    return isinstance(repo.svfs.lfsremoteblobstore, blobstore._nullremote)
+
+def candownload(repo):
+    # if remotestore is a null store, downloads will lead to nothing
+    return not isinstance(repo.svfs.lfsremoteblobstore, blobstore._nullremote)
+
+def uploadblobsfromrevs(repo, revs):
+    '''upload lfs blobs introduced by revs
+
+    Note: also used by other extensions e. g. infinitepush. avoid renaming.
+    '''
+    if _canskipupload(repo):
+        return
+    pointers = extractpointers(repo, revs)
+    uploadblobs(repo, pointers)
+
+def prepush(pushop):
+    """Prepush hook.
+
+    Read through the revisions to push, looking for filelog entries that can be
+    deserialized into metadata so that we can block the push on their upload to
+    the remote blobstore.
+    """
+    return uploadblobsfromrevs(pushop.repo, pushop.outgoing.missing)
+
+def writenewbundle(orig, ui, repo, source, filename, bundletype, outgoing,
+                   *args, **kwargs):
+    """upload LFS blobs added by outgoing revisions on 'hg bundle'"""
+    uploadblobsfromrevs(repo, outgoing.missing)
+    return orig(ui, repo, source, filename, bundletype, outgoing, *args,
+                **kwargs)
+
+def extractpointers(repo, revs):
+    """return a list of lfs pointers added by given revs"""
+    ui = repo.ui
+    if ui.debugflag:
+        ui.write(_('lfs: computing set of blobs to upload\n'))
+    pointers = {}
+    for r in revs:
+        ctx = repo[r]
+        for p in pointersfromctx(ctx).values():
+            pointers[p.oid()] = p
+    return pointers.values()
+
+def pointersfromctx(ctx):
+    """return a dict {path: pointer} for given single changectx"""
+    result = {}
+    for f in ctx.files():
+        if f not in ctx:
+            continue
+        fctx = ctx[f]
+        if not _islfs(fctx.filelog(), fctx.filenode()):
+            continue
+        try:
+            result[f] = pointer.deserialize(fctx.rawdata())
+        except pointer.InvalidPointer as ex:
+            raise error.Abort(_('lfs: corrupted pointer (%s@%s): %s\n')
+                              % (f, short(ctx.node()), ex))
+    return result
+
+def uploadblobs(repo, pointers):
+    """upload given pointers from local blobstore"""
+    if not pointers:
+        return
+
+    remoteblob = repo.svfs.lfsremoteblobstore
+    remoteblob.writebatch(pointers, repo.svfs.lfslocalblobstore)