comparison hgext/largefiles/overrides.py @ 41062:0a7f582f6f1f

largefiles: port wrapped functions to exthelper

Things get interesting in the commit.  I hadn't seen issue6033 on Windows, and
yet it is now reproducible 100% of the time on Windows 10 with this commit.  I
didn't test Linux.  (For comparison, after seeing this issue, I tested on the
parent with --loop, and it failed 5 times out of over 1300 tests.)

The strange thing is that largefiles has nothing to do with that test (it's not
even mentioned there).  It isn't autoloading run amuck- it occurs even if
largefiles is explicitly disabled, and also if the entry in afterhgrcload() is
commented out.  It's also not the import of lfutil- I disabled that by copying
the function into lfs and removing the import, and the problem still occurs.

Experimenting further, it seems that the problem is isolated to 3 entries:
exchange.pushoperation, hg.clone, and cmdutil.revert.  If those decorators are
all commented out, the test passes when run in a loop for awhile.  (Obviously,
some largefiles tests will fail.)  But if any one is commented back in, the test
fails immediately.

I left one method related to wrapping the wire protocol, because it seemed more
natural with the TODO.  Also, exthelper doesn't support wrapping functions from
another extension, only commands in another extension.  I didn't try to figure
out why rebase is both command wrapped and function wrapped.

41061:98681293c890

import copy
import os

from mercurial.i18n import _

from mercurial import (
    archival,
    cmdutil,
    error,
    exthelper,
    hg,
    logcmdutil,
    match as matchmod,
    pathutil,
    pycompat,
    registrar,
    scmutil,
    smartset,
    util,
)

from . import (
    lfcommands,

    return result

# For overriding mercurial.hgweb.webcommands so that largefiles will
# appear at their right place in the manifests.
def decodepath(orig, path):
    return lfutil.splitstandin(path) or path

# -- Wrappers: modify existing commands --------------------------------

def overrideadd(orig, ui, repo, *pats, **opts):
    if opts.get(r'normal') and opts.get(r'large'):
        raise error.Abort(_('--normal cannot be used with --large'))
    return orig(ui, repo, *pats, **opts)

def cmdutiladd(orig, ui, repo, matcher, prefix, explicitonly, **opts):
    # The --normal flag short circuits this override
    if opts.get(r'normal'):
        return orig(ui, repo, matcher, prefix, explicitonly, **opts)

    bad = orig(ui, repo, normalmatcher, prefix, explicitonly, **opts)

    bad.extend(f for f in lbad)
    return bad

def cmdutilremove(orig, ui, repo, matcher, prefix, after, force, subrepos,
                  dryrun):
    normalmatcher = composenormalfilematcher(matcher, repo[None].manifest())
    result = orig(ui, repo, normalmatcher, prefix, after, force, subrepos,
                  dryrun)
    return removelargefiles(ui, repo, False, matcher, dryrun, after=after,
                            force=force) or result

def overridestatusfn(orig, repo, rev2, **opts):
    try:
        repo._repo.lfstatus = True
        return orig(repo, rev2, **opts)
    finally:

        repo.lfstatus = True
        return orig(ui, repo, *pats, **opts)
    finally:
        repo.lfstatus = False

def overridedirty(orig, repo, ignoreupdate=False, missing=False):
    try:
        repo._repo.lfstatus = True
        return orig(repo, ignoreupdate=ignoreupdate, missing=missing)
    finally:

# in a file 'foo' if we already have a largefile with the same name.
#
# The overridden function filters the unknown files by removing any
# largefiles. This makes the merge proceed and we can then handle this
# case further in the overridden calculateupdates function below.
def overridecheckunknownfile(origfn, repo, wctx, mctx, f, f2=None):
    if lfutil.standin(repo.dirstate.normalize(f)) in wctx:
        return False
    return origfn(repo, wctx, mctx, f, f2)

# presumably changed on purpose.
#
# Finally, the merge.applyupdates function will then take care of
# writing the files into the working copy and lfcommands.updatelfiles
# will update the largefiles.
def overridecalculateupdates(origfn, repo, p1, p2, pas, branchmerge, force,
                             acceptremote, *args, **kwargs):
    overwrite = force and not branchmerge
    actions, diverge, renamedelete = origfn(
        repo, p1, p2, pas, branchmerge, force, acceptremote, *args, **kwargs)

                actions[lfile] = ('g', largs, 'replaces standin')
                actions[standin] = ('r', None, 'replaced by non-standin')

    return actions, diverge, renamedelete

def mergerecordupdates(orig, repo, actions, branchmerge):
    if 'lfmr' in actions:
        lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
        for lfile, args, msg in actions['lfmr']:
            # this should be executed before 'orig', to execute 'remove'

    return orig(repo, actions, branchmerge)

# Override filemerge to prompt the user about how they wish to merge
# largefiles. This will handle identical edits without prompting the user.
def overridefilemerge(origfn, premerge, repo, wctx, mynode, orig, fcd, fco, fca,
                      labels=None):
    if not lfutil.isstandin(orig) or fcd.isabsent() or fco.isabsent():
        return origfn(premerge, repo, wctx, mynode, orig, fcd, fco, fca,
                      labels=labels)

               (lfutil.splitstandin(orig), ahash, dhash, ohash),
             0) == 1)):
        repo.wwrite(fcd.path(), fco.data(), fco.flags())
    return True, 0, False

def copiespathcopies(orig, ctx1, ctx2, match=None):
    copies = orig(ctx1, ctx2, match=match)
    updated = {}

    for k, v in copies.iteritems():

# Copy first changes the matchers to match standins instead of
# largefiles.  Then it overrides util.copyfile in that function it
# checks if the destination largefile already exists. It also keeps a
# list of copied files so that the largefiles can be copied and the
# dirstate updated.
def overridecopy(orig, ui, repo, pats, opts, rename=False):
    # doesn't remove largefile on rename
    if len(pats) < 2:
        # this isn't legal, let the original function deal with it
        return orig(ui, repo, pats, opts, rename)

#
# Standins are only updated (to match the hash of largefiles) before
# commits. Update the standins then run the original revert, changing
# the matcher to hit standins instead of largefiles. Based on the
# resulting standins update the largefiles.
def overriderevert(orig, ui, repo, ctx, parents, *pats, **opts):
    # Because we put the standins in a bad state (by updating them)
    # and then return them to a correct state we need to lock to
    # prevent others from changing them in their incorrect state.
    with repo.wlock():

    if lfrevs:
        opargs = kwargs.setdefault(r'opargs', {})
        opargs['lfrevs'] = scmutil.revrange(repo, lfrevs)
    return orig(ui, repo, *args, **kwargs)

def exchangepushoperation(orig, *args, **kwargs):
    """Override pushoperation constructor and store lfrevs parameter"""
    lfrevs = kwargs.pop(r'lfrevs', None)
    pushop = orig(*args, **kwargs)
    pushop.lfrevs = lfrevs

            '--all-largefiles is incompatible with non-local destination %s') %
            d)

    return orig(ui, source, dest, **opts)

def hgclone(orig, ui, opts, *args, **kwargs):
    result = orig(ui, opts, *args, **kwargs)

    if result is not None:
        sourcerepo, destrepo = result

    try:
        return orig(ui, repo.unfiltered(), dest, **opts)
    finally:
        repo.unfiltered().lfstatus = False

def hgwebarchive(orig, web):
    web.repo.lfstatus = True

    try:
        return orig(web)
    finally:
        web.repo.lfstatus = False

def overridearchive(orig, repo, dest, node, kind, decode=True, match=None,
            prefix='', mtime=None, subrepos=None):
    # For some reason setting repo.lfstatus in hgwebarchive only changes the
    # unfiltered repo's attr, so check that as well.
    if not repo.lfstatus and not repo.unfiltered().lfstatus:

            sub._repo.lfstatus = True
            sub.archive(archiver, prefix, submatch)

    archiver.done()

def hgsubrepoarchive(orig, repo, archiver, prefix, match=None, decode=True):
    lfenabled = util.safehasattr(repo._repo, '_largefilesenabled')
    if not lfenabled or not repo._repo.lfstatus:
        return orig(repo, archiver, prefix, match, decode)

# If a largefile is modified, the change is not reflected in its
# standin until a commit. cmdutil.bailifchanged() raises an exception
# if the repo has uncommitted changes. Wrap it to also check if
# largefiles were changed. This is used by bisect, backout and fetch.
def overridebailifchanged(orig, repo, *args, **kwargs):
    orig(repo, *args, **kwargs)
    repo.lfstatus = True
    s = repo.status()
    repo.lfstatus = False
    if s.modified or s.added or s.removed or s.deleted:
        raise error.Abort(_('uncommitted changes'))

def postcommitstatus(orig, repo, *args, **kwargs):
    repo.lfstatus = True
    try:
        return orig(repo, *args, **kwargs)
    finally:
        repo.lfstatus = False

def cmdutilforget(orig, ui, repo, match, prefix, explicitonly, dryrun,
                  interactive):
    normalmatcher = composenormalfilematcher(match, repo[None].manifest())
    bad, forgot = orig(ui, repo, normalmatcher, prefix, explicitonly, dryrun,
                       interactive)

        repo.lfstatus = True
        orig(ui, repo, *pats, **opts)
    finally:
        repo.lfstatus = False

def scmutiladdremove(orig, repo, matcher, prefix, opts=None):
    if opts is None:
        opts = {}
    if not lfutil.islfilesrepo(repo):
        return orig(repo, matcher, prefix, opts)

                    for chunk in util.filechunkiter(fpin):
                        fp.write(chunk)
        err = 0
    return err

def mergeupdate(orig, repo, node, branchmerge, force,
                *args, **kwargs):
    matcher = kwargs.get(r'matcher', None)
    # note if this is a partial update
    partial = matcher and not matcher.always()

        lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                normallookup=partial)

        return result

def scmutilmarktouched(orig, repo, files, *args, **kwargs):
    result = orig(repo, files, *args, **kwargs)

    filelist = []
    for f in files:

        lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                printmessage=False, normallookup=True)

    return result

def upgraderequirements(orig, repo):
    reqs = orig(repo)
    if 'largefiles' in repo.requirements:
        reqs.add('largefiles')
    return reqs

_lfscheme = 'largefile://'
def openlargefile(orig, ui, url_, data=None):
    if url_.startswith(_lfscheme):
        if data:
            msg = "cannot use data on a 'largefile://' url"
            raise error.ProgrammingError(msg)

41062:0a7f582f6f1f

import copy
import os

from mercurial.i18n import _

from mercurial.hgweb import (
    webcommands,
)

from mercurial import (
    archival,
    cmdutil,
    copies as copiesmod,
    error,
    exchange,
    exthelper,
    filemerge,
    hg,
    logcmdutil,
    match as matchmod,
    merge,
    pathutil,
    pycompat,
    registrar,
    scmutil,
    smartset,
    subrepo,
    upgrade,
    url as urlmod,
    util,
)

from . import (
    lfcommands,

    return result

# For overriding mercurial.hgweb.webcommands so that largefiles will
# appear at their right place in the manifests.
@eh.wrapfunction(webcommands, 'decodepath')
def decodepath(orig, path):
    return lfutil.splitstandin(path) or path

# -- Wrappers: modify existing commands --------------------------------

def overrideadd(orig, ui, repo, *pats, **opts):
    if opts.get(r'normal') and opts.get(r'large'):
        raise error.Abort(_('--normal cannot be used with --large'))
    return orig(ui, repo, *pats, **opts)

@eh.wrapfunction(cmdutil, 'add')
def cmdutiladd(orig, ui, repo, matcher, prefix, explicitonly, **opts):
    # The --normal flag short circuits this override
    if opts.get(r'normal'):
        return orig(ui, repo, matcher, prefix, explicitonly, **opts)

    bad = orig(ui, repo, normalmatcher, prefix, explicitonly, **opts)

    bad.extend(f for f in lbad)
    return bad

@eh.wrapfunction(cmdutil, 'remove')
def cmdutilremove(orig, ui, repo, matcher, prefix, after, force, subrepos,
                  dryrun):
    normalmatcher = composenormalfilematcher(matcher, repo[None].manifest())
    result = orig(ui, repo, normalmatcher, prefix, after, force, subrepos,
                  dryrun)
    return removelargefiles(ui, repo, False, matcher, dryrun, after=after,
                            force=force) or result

@eh.wrapfunction(subrepo.hgsubrepo, 'status')
def overridestatusfn(orig, repo, rev2, **opts):
    try:
        repo._repo.lfstatus = True
        return orig(repo, rev2, **opts)
    finally:

        repo.lfstatus = True
        return orig(ui, repo, *pats, **opts)
    finally:
        repo.lfstatus = False

@eh.wrapfunction(subrepo.hgsubrepo, 'dirty')
def overridedirty(orig, repo, ignoreupdate=False, missing=False):
    try:
        repo._repo.lfstatus = True
        return orig(repo, ignoreupdate=ignoreupdate, missing=missing)
    finally:

# in a file 'foo' if we already have a largefile with the same name.
#
# The overridden function filters the unknown files by removing any
# largefiles. This makes the merge proceed and we can then handle this
# case further in the overridden calculateupdates function below.
@eh.wrapfunction(merge, '_checkunknownfile')
def overridecheckunknownfile(origfn, repo, wctx, mctx, f, f2=None):
    if lfutil.standin(repo.dirstate.normalize(f)) in wctx:
        return False
    return origfn(repo, wctx, mctx, f, f2)

# presumably changed on purpose.
#
# Finally, the merge.applyupdates function will then take care of
# writing the files into the working copy and lfcommands.updatelfiles
# will update the largefiles.
@eh.wrapfunction(merge, 'calculateupdates')
def overridecalculateupdates(origfn, repo, p1, p2, pas, branchmerge, force,
                             acceptremote, *args, **kwargs):
    overwrite = force and not branchmerge
    actions, diverge, renamedelete = origfn(
        repo, p1, p2, pas, branchmerge, force, acceptremote, *args, **kwargs)

                actions[lfile] = ('g', largs, 'replaces standin')
                actions[standin] = ('r', None, 'replaced by non-standin')

    return actions, diverge, renamedelete

@eh.wrapfunction(merge, 'recordupdates')
def mergerecordupdates(orig, repo, actions, branchmerge):
    if 'lfmr' in actions:
        lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
        for lfile, args, msg in actions['lfmr']:
            # this should be executed before 'orig', to execute 'remove'

    return orig(repo, actions, branchmerge)

# Override filemerge to prompt the user about how they wish to merge
# largefiles. This will handle identical edits without prompting the user.
@eh.wrapfunction(filemerge, '_filemerge')
def overridefilemerge(origfn, premerge, repo, wctx, mynode, orig, fcd, fco, fca,
                      labels=None):
    if not lfutil.isstandin(orig) or fcd.isabsent() or fco.isabsent():
        return origfn(premerge, repo, wctx, mynode, orig, fcd, fco, fca,
                      labels=labels)

               (lfutil.splitstandin(orig), ahash, dhash, ohash),
             0) == 1)):
        repo.wwrite(fcd.path(), fco.data(), fco.flags())
    return True, 0, False

@eh.wrapfunction(copiesmod, 'pathcopies')
def copiespathcopies(orig, ctx1, ctx2, match=None):
    copies = orig(ctx1, ctx2, match=match)
    updated = {}

    for k, v in copies.iteritems():

# Copy first changes the matchers to match standins instead of
# largefiles.  Then it overrides util.copyfile in that function it
# checks if the destination largefile already exists. It also keeps a
# list of copied files so that the largefiles can be copied and the
# dirstate updated.
@eh.wrapfunction(cmdutil, 'copy')
def overridecopy(orig, ui, repo, pats, opts, rename=False):
    # doesn't remove largefile on rename
    if len(pats) < 2:
        # this isn't legal, let the original function deal with it
        return orig(ui, repo, pats, opts, rename)

#
# Standins are only updated (to match the hash of largefiles) before
# commits. Update the standins then run the original revert, changing
# the matcher to hit standins instead of largefiles. Based on the
# resulting standins update the largefiles.
@eh.wrapfunction(cmdutil, 'revert')
def overriderevert(orig, ui, repo, ctx, parents, *pats, **opts):
    # Because we put the standins in a bad state (by updating them)
    # and then return them to a correct state we need to lock to
    # prevent others from changing them in their incorrect state.
    with repo.wlock():

    if lfrevs:
        opargs = kwargs.setdefault(r'opargs', {})
        opargs['lfrevs'] = scmutil.revrange(repo, lfrevs)
    return orig(ui, repo, *args, **kwargs)

@eh.wrapfunction(exchange, 'pushoperation')
def exchangepushoperation(orig, *args, **kwargs):
    """Override pushoperation constructor and store lfrevs parameter"""
    lfrevs = kwargs.pop(r'lfrevs', None)
    pushop = orig(*args, **kwargs)
    pushop.lfrevs = lfrevs

            '--all-largefiles is incompatible with non-local destination %s') %
            d)

    return orig(ui, source, dest, **opts)

@eh.wrapfunction(hg, 'clone')
def hgclone(orig, ui, opts, *args, **kwargs):
    result = orig(ui, opts, *args, **kwargs)

    if result is not None:
        sourcerepo, destrepo = result

    try:
        return orig(ui, repo.unfiltered(), dest, **opts)
    finally:
        repo.unfiltered().lfstatus = False

@eh.wrapfunction(webcommands, 'archive')
def hgwebarchive(orig, web):
    web.repo.lfstatus = True

    try:
        return orig(web)
    finally:
        web.repo.lfstatus = False

@eh.wrapfunction(archival, 'archive')
def overridearchive(orig, repo, dest, node, kind, decode=True, match=None,
            prefix='', mtime=None, subrepos=None):
    # For some reason setting repo.lfstatus in hgwebarchive only changes the
    # unfiltered repo's attr, so check that as well.
    if not repo.lfstatus and not repo.unfiltered().lfstatus:

            sub._repo.lfstatus = True
            sub.archive(archiver, prefix, submatch)

    archiver.done()

@eh.wrapfunction(subrepo.hgsubrepo, 'archive')
def hgsubrepoarchive(orig, repo, archiver, prefix, match=None, decode=True):
    lfenabled = util.safehasattr(repo._repo, '_largefilesenabled')
    if not lfenabled or not repo._repo.lfstatus:
        return orig(repo, archiver, prefix, match, decode)

# If a largefile is modified, the change is not reflected in its
# standin until a commit. cmdutil.bailifchanged() raises an exception
# if the repo has uncommitted changes. Wrap it to also check if
# largefiles were changed. This is used by bisect, backout and fetch.
@eh.wrapfunction(cmdutil, 'bailifchanged')
def overridebailifchanged(orig, repo, *args, **kwargs):
    orig(repo, *args, **kwargs)
    repo.lfstatus = True
    s = repo.status()
    repo.lfstatus = False
    if s.modified or s.added or s.removed or s.deleted:
        raise error.Abort(_('uncommitted changes'))

@eh.wrapfunction(cmdutil, 'postcommitstatus')
def postcommitstatus(orig, repo, *args, **kwargs):
    repo.lfstatus = True
    try:
        return orig(repo, *args, **kwargs)
    finally:
        repo.lfstatus = False

@eh.wrapfunction(cmdutil, 'forget')
def cmdutilforget(orig, ui, repo, match, prefix, explicitonly, dryrun,
                  interactive):
    normalmatcher = composenormalfilematcher(match, repo[None].manifest())
    bad, forgot = orig(ui, repo, normalmatcher, prefix, explicitonly, dryrun,
                       interactive)

        repo.lfstatus = True
        orig(ui, repo, *pats, **opts)
    finally:
        repo.lfstatus = False

@eh.wrapfunction(scmutil, 'addremove')
def scmutiladdremove(orig, repo, matcher, prefix, opts=None):
    if opts is None:
        opts = {}
    if not lfutil.islfilesrepo(repo):
        return orig(repo, matcher, prefix, opts)

                    for chunk in util.filechunkiter(fpin):
                        fp.write(chunk)
        err = 0
    return err

@eh.wrapfunction(merge, 'update')
def mergeupdate(orig, repo, node, branchmerge, force,
                *args, **kwargs):
    matcher = kwargs.get(r'matcher', None)
    # note if this is a partial update
    partial = matcher and not matcher.always()

        lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                normallookup=partial)

        return result

@eh.wrapfunction(scmutil, 'marktouched')
def scmutilmarktouched(orig, repo, files, *args, **kwargs):
    result = orig(repo, files, *args, **kwargs)

    filelist = []
    for f in files:

        lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                printmessage=False, normallookup=True)

    return result

@eh.wrapfunction(upgrade, 'preservedrequirements')
@eh.wrapfunction(upgrade, 'supporteddestrequirements')
def upgraderequirements(orig, repo):
    reqs = orig(repo)
    if 'largefiles' in repo.requirements:
        reqs.add('largefiles')
    return reqs

_lfscheme = 'largefile://'

@eh.wrapfunction(urlmod, 'open')
def openlargefile(orig, ui, url_, data=None):
    if url_.startswith(_lfscheme):
        if data:
            msg = "cannot use data on a 'largefile://' url"
            raise error.ProgrammingError(msg)