def _iter_items(cls, repo, common_path = None): if common_path is None: common_path = cls._common_path_default rela_paths = set() # walk loose refs # Currently we do not follow links for root, dirs, files in os.walk(join_path_native(repo.git_dir, common_path)): if 'refs/' not in root: # skip non-refs subfolders refs_id = [ i for i,d in enumerate(dirs) if d == 'refs' ] if refs_id: dirs[0:] = ['refs'] # END prune non-refs folders for f in files: abs_path = to_native_path_linux(join_path(root, f)) rela_paths.add(abs_path.replace(to_native_path_linux(repo.git_dir) + '/', "")) # END for each file in root directory # END for each directory to walk # read packed refs for sha, rela_path in cls._iter_packed_refs(repo): if rela_path.startswith(common_path): rela_paths.add(rela_path) # END relative path matches common path # END packed refs reading # return paths in sorted order for path in sorted(rela_paths): try: yield cls.from_path(repo, path) except ValueError: continue
def store_path(filepath): """Store file at filepath in the database and return the base index entry""" st = os.lstat(filepath) # handles non-symlinks as well stream = None if S_ISLNK(st.st_mode): stream = StringIO(os.readlink(filepath)) else: stream = open(filepath, 'rb') # END handle stream fprogress(filepath, False, filepath) istream = self.repo.odb.store(IStream(Blob.type, st.st_size, stream)) fprogress(filepath, True, filepath) return BaseIndexEntry((stat_mode_to_index_mode(st.st_mode), istream.binsha, 0, to_native_path_linux(filepath)))
def checkout(self, paths=None, force=False, fprogress=lambda *args: None, **kwargs):
"""Checkout the given paths or all files from the version known to the index into
the working tree.
:note: Be sure you have written pending changes using the ``write`` method
in case you have altered the enties dictionary directly
:param paths:
If None, all paths in the index will be checked out. Otherwise an iterable
of relative or absolute paths or a single path pointing to files or directories
in the index is expected.
:param force:
If True, existing files will be overwritten even if they contain local modifications.
If False, these will trigger a CheckoutError.
:param fprogress:
see Index.add_ for signature and explanation.
The provided progress information will contain None as path and item if no
explicit paths are given. Otherwise progress information will be send
prior and after a file has been checked out
:param kwargs:
Additional arguments to be pasesd to pygit-checkout-index
:return:
iterable yielding paths to files which have been checked out and are
guaranteed to match the version stored in the index
:raise CheckoutError:
If at least one file failed to be checked out. This is a summary,
hence it will checkout as many files as it can anyway.
If one of files or directories do not exist in the index
( as opposed to the original pygit command who ignores them ).
Raise GitPyCommandError if error lines could not be parsed - this truly is
an exceptional state
.. note:: The checkout is limited to checking out the files in the
index. Files which are not in the index anymore and exist in
the working tree will not be deleted. This behaviour is fundamentally
different to *head.checkout*, i.e. if you want pygit-checkout like behaviour,
use head.checkout instead of index.checkout.
"""
args = ["--index"]
if force:
args.append("--force")
def handle_stderr(proc, iter_checked_out_files):
stderr = proc.stderr.read()
if not stderr:
return
# line contents:
# pygit-checkout-index: this already exists
failed_files = list()
failed_reasons = list()
unknown_lines = list()
endings = (' already exists', ' is not in the cache', ' does not exist at stage', ' is unmerged')
for line in stderr.splitlines():
if not line.startswith("git checkout-index: ") and not line.startswith("git-checkout-index: "):
is_a_dir = " is a directory"
unlink_issue = "unable to unlink old '"
if line.endswith(is_a_dir):
failed_files.append(line[:-len(is_a_dir)])
failed_reasons.append(is_a_dir)
elif line.startswith(unlink_issue):
failed_files.append(line[len(unlink_issue):line.rfind("'")])
failed_reasons.append(unlink_issue)
else:
unknown_lines.append(line)
continue
# END special lines parsing
for e in endings:
if line.endswith(e):
failed_files.append(line[20:-len(e)])
failed_reasons.append(e)
break
# END if ending matches
# END for each possible ending
# END for each line
if unknown_lines:
raise GitPyCommandError(("git-checkout-index", ), 128, stderr)
if failed_files:
valid_files = list(set(iter_checked_out_files) - set(failed_files))
raise CheckoutError("Some files could not be checked out from the index due to local modifications", failed_files, valid_files, failed_reasons)
# END stderr handler
if paths is None:
args.append("--all")
kwargs['as_process'] = 1
fprogress(None, False, None)
proc = self.repo.git.checkout_index(*args, **kwargs)
proc.wait()
fprogress(None, True, None)
rval_iter = ( e.path for e in self.entries.itervalues() )
handle_stderr(proc, rval_iter)
return rval_iter
else:
if isinstance(paths, basestring):
paths = [paths]
# make sure we have our entries loaded before we start checkout_index
# which will hold a lock on it. We try to get the lock as well during
# our entries initialization
self.entries
args.append("--stdin")
kwargs['as_process'] = True
kwargs['istream'] = subprocess.PIPE
proc = self.repo.git.checkout_index(args, **kwargs)
make_exc = lambda : GitPyCommandError(("git-checkout-index",)+tuple(args), 128, proc.stderr.read())
checked_out_files = list()
for path in paths:
co_path = to_native_path_linux(self._to_relative_path(path))
# if the item is not in the index, it could be a directory
path_is_directory = False
try:
self.entries[(co_path, 0)]
except KeyError:
dir = co_path
if not dir.endswith('/'):
dir += '/'
for entry in self.entries.itervalues():
if entry.path.startswith(dir):
p = entry.path
self._write_path_to_stdin(proc, p, p, make_exc,
fprogress, read_from_stdout=False)
checked_out_files.append(p)
path_is_directory = True
# END if entry is in directory
# END for each entry
# END path exception handlnig
if not path_is_directory:
self._write_path_to_stdin(proc, co_path, path, make_exc,
fprogress, read_from_stdout=False)
checked_out_files.append(co_path)
# END path is a file
# END for each path
self._flush_stdin_and_wait(proc, ignore_stdout=True)
handle_stderr(proc, checked_out_files)
return checked_out_files
# END paths handling
assert "Should not reach this point"
def add(self, items, force=True, fprogress=lambda *args: None, path_rewriter=None,
write=True):
"""Add files from the working tree, specific blobs or BaseIndexEntries
to the index.
:param items:
Multiple types of items are supported, types can be mixed within one call.
Different types imply a different handling. File paths may generally be
relative or absolute.
- path string
strings denote a relative or absolute path into the repository pointing to
an existing file, i.e. CHANGES, lib/myfile.ext, '/home/gitrepo/lib/myfile.ext'.
Paths provided like this must exist. When added, they will be written
into the object database.
PathStrings may contain globs, such as 'lib/__init__*' or can be directories
like 'lib', the latter ones will add all the files within the dirctory and
subdirectories.
This equals a straight pygit-add.
They are added at stage 0
- Blob or Submodule object
Blobs are added as they are assuming a valid mode is set.
The file they refer to may or may not exist in the file system, but
must be a path relative to our repository.
If their sha is null ( 40*0 ), their path must exist in the file system
relative to the pygit repository as an object will be created from
the data at the path.
The handling now very much equals the way string paths are processed, except that
the mode you have set will be kept. This allows you to create symlinks
by settings the mode respectively and writing the target of the symlink
directly into the file. This equals a default Linux-Symlink which
is not dereferenced automatically, except that it can be created on
filesystems not supporting it as well.
Please note that globs or directories are not allowed in Blob objects.
They are added at stage 0
- BaseIndexEntry or type
Handling equals the one of Blob objects, but the stage may be
explicitly set. Please note that Index Entries require binary sha's.
:param force:
**CURRENTLY INEFFECTIVE**
If True, otherwise ignored or excluded files will be
added anyway.
As opposed to the pygit-add command, we enable this flag by default
as the API user usually wants the item to be added even though
they might be excluded.
:param fprogress:
Function with signature f(path, done=False, item=item) called for each
path to be added, one time once it is about to be added where done==False
and once after it was added where done=True.
item is set to the actual item we handle, either a Path or a BaseIndexEntry
Please note that the processed path is not guaranteed to be present
in the index already as the index is currently being processed.
:param path_rewriter:
Function with signature (string) func(BaseIndexEntry) function returning a path
for each passed entry which is the path to be actually recorded for the
object created from entry.path. This allows you to write an index which
is not identical to the layout of the actual files on your hard-disk.
If not None and ``items`` contain plain paths, these paths will be
converted to Entries beforehand and passed to the path_rewriter.
Please note that entry.path is relative to the pygit repository.
:param write:
If True, the index will be written once it was altered. Otherwise
the changes only exist in memory and are not available to pygit commands.
:return:
List(BaseIndexEntries) representing the entries just actually added.
:raise OSError:
if a supplied Path did not exist. Please note that BaseIndexEntry
Objects that do not have a null sha will be added even if their paths
do not exist.
"""
# sort the entries into strings and Entries, Blobs are converted to entries
# automatically
# paths can be pygit-added, for everything else we use git-update-index
entries_added = list()
paths, entries = self._preprocess_add_items(items)
if paths and path_rewriter:
for path in paths:
abspath = os.path.abspath(path)
gitrelative_path = abspath[len(self.repo.working_tree_dir)+1:]
blob = Blob(self.repo, Blob.NULL_BIN_SHA,
stat_mode_to_index_mode(os.stat(abspath).st_mode),
to_native_path_linux(gitrelative_path))
entries.append(BaseIndexEntry.from_blob(blob))
# END for each path
del(paths[:])
# END rewrite paths
def store_path(filepath):
"""Store file at filepath in the database and return the base index entry"""
st = os.lstat(filepath) # handles non-symlinks as well
stream = None
if S_ISLNK(st.st_mode):
stream = StringIO(os.readlink(filepath))
else:
stream = open(filepath, 'rb')
# END handle stream
fprogress(filepath, False, filepath)
istream = self.repo.odb.store(IStream(Blob.type, st.st_size, stream))
fprogress(filepath, True, filepath)
return BaseIndexEntry((stat_mode_to_index_mode(st.st_mode),
istream.binsha, 0, to_native_path_linux(filepath)))
# END utility method
# HANDLE PATHS
if paths:
assert len(entries_added) == 0
added_files = list()
for filepath in self._iter_expand_paths(paths):
entries_added.append(store_path(filepath))
# END for each filepath
# END path handling
# HANDLE ENTRIES
if entries:
null_mode_entries = [ e for e in entries if e.mode == 0 ]
if null_mode_entries:
raise ValueError("At least one Entry has a null-mode - please use index.remove to remove files for clarity")
# END null mode should be remove
# HANLDE ENTRY OBJECT CREATION
# create objects if required, otherwise go with the existing shas
null_entries_indices = [ i for i,e in enumerate(entries) if e.binsha == Object.NULL_BIN_SHA ]
if null_entries_indices:
for ei in null_entries_indices:
null_entry = entries[ei]
new_entry = store_path(null_entry.path)
# update null entry
entries[ei] = BaseIndexEntry((null_entry.mode, new_entry.binsha, null_entry.stage, null_entry.path))
# END for each entry index
# END null_entry handling
# REWRITE PATHS
# If we have to rewrite the entries, do so now, after we have generated
# all object sha's
if path_rewriter:
for i,e in enumerate(entries):
entries[i] = BaseIndexEntry((e.mode, e.binsha, e.stage, path_rewriter(e)))
# END for each entry
# END handle path rewriting
# just go through the remaining entries and provide progress info
for i, entry in enumerate(entries):
progress_sent = i in null_entries_indices
if not progress_sent:
fprogress(entry.path, False, entry)
fprogress(entry.path, True, entry)
# END handle progress
# END for each enty
entries_added.extend(entries)
# END if there are base entries
# FINALIZE
# add the new entries to this instance
for entry in entries_added:
self.entries[(entry.path, 0)] = IndexEntry.from_base(entry)
if write:
self.write()
# END handle write
return entries_added
def test_root_module(self, rwrepo):
# Can query everything without problems
rm = RootModule(self.rorepo)
assert rm.module() is self.rorepo
# try attributes
rm.binsha
rm.mode
rm.path
assert rm.name == rm.k_root_name
assert rm.parent_commit == self.rorepo.head.commit
rm.url
rm.branch
assert len(rm.list_items(rm.module())) == 1
rm.config_reader()
rm.config_writer()
# deep traversal gitdb / async
rsmsp = [sm.path for sm in rm.traverse()]
assert len(rsmsp) == 2 # gitdb and async, async being a child of gitdb
# cannot set the parent commit as root module's path didn't exist
self.failUnlessRaises(ValueError, rm.set_parent_commit, 'HEAD')
# TEST UPDATE
#############
# setup commit which remove existing, add new and modify existing submodules
rm = RootModule(rwrepo)
assert len(rm.children()) == 1
# modify path without modifying the index entry
# ( which is what the move method would do properly )
#==================================================
sm = rm.children()[0]
pp = "path/prefix"
fp = join_path_native(pp, sm.path)
prep = sm.path
assert not sm.module_exists() # was never updated after rwrepo's clone
# assure we clone from a local source
sm.config_writer().set_value('url', to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, sm.path)))
# dry-run does nothing
sm.update(recursive=False, dry_run=True, progress=prog)
assert not sm.module_exists()
sm.update(recursive=False)
assert sm.module_exists()
sm.config_writer().set_value('path', fp) # change path to something with prefix AFTER url change
# update fails as list_items in such a situations cannot work, as it cannot
# find the entry at the changed path
self.failUnlessRaises(InvalidGitPyRepositoryError, rm.update, recursive=False)
# move it properly - doesn't work as it its path currently points to an indexentry
# which doesn't exist ( move it to some path, it doesn't matter here )
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.move, pp)
# reset the path(cache) to where it was, now it works
sm.path = prep
sm.move(fp, module=False) # leave it at the old location
assert not sm.module_exists()
cpathchange = rwrepo.index.commit("changed sm path") # finally we can commit
# update puts the module into place
rm.update(recursive=False, progress=prog)
sm.set_parent_commit(cpathchange)
assert sm.module_exists()
# add submodule
#================
nsmn = "newsubmodule"
nsmp = "submrepo"
async_url = to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, rsmsp[0], rsmsp[1]))
nsm = Submodule.add(rwrepo, nsmn, nsmp, url=async_url)
csmadded = rwrepo.index.commit("Added submodule").hexsha # make sure we don't keep the repo reference
nsm.set_parent_commit(csmadded)
assert nsm.module_exists()
# in our case, the module should not exist, which happens if we update a parent
# repo and a new submodule comes into life
nsm.remove(configuration=False, module=True)
assert not nsm.module_exists() and nsm.exists()
# dry-run does nothing
rm.update(recursive=False, dry_run=True, progress=prog)
# otherwise it will work
rm.update(recursive=False, progress=prog)
assert nsm.module_exists()
# remove submodule - the previous one
#====================================
sm.set_parent_commit(csmadded)
smp = sm.abspath
assert not sm.remove(module=False).exists()
assert os.path.isdir(smp) # module still exists
csmremoved = rwrepo.index.commit("Removed submodule")
# an update will remove the module
# not in dry_run
rm.update(recursive=False, dry_run=True)
assert os.path.isdir(smp)
rm.update(recursive=False)
assert not os.path.isdir(smp)
# change url
#=============
# to the first repository, this way we have a fast checkout, and a completely different
# repository at the different url
nsm.set_parent_commit(csmremoved)
nsmurl = to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, rsmsp[0]))
nsm.config_writer().set_value('url', nsmurl)
csmpathchange = rwrepo.index.commit("changed url")
nsm.set_parent_commit(csmpathchange)
prev_commit = nsm.module().head.commit
# dry-run does nothing
rm.update(recursive=False, dry_run=True, progress=prog)
assert nsm.module().remotes.origin.url != nsmurl
rm.update(recursive=False, progress=prog)
assert nsm.module().remotes.origin.url == nsmurl
# head changed, as the remote url and its commit changed
assert prev_commit != nsm.module().head.commit
# add the submodule's changed commit to the index, which is what the
# user would do
# beforehand, update our instance's binsha with the new one
nsm.binsha = nsm.module().head.commit.binsha
rwrepo.index.add([nsm])
# change branch
#=================
# we only have one branch, so we switch to a virtual one, and back
# to the current one to trigger the difference
cur_branch = nsm.branch
nsmm = nsm.module()
prev_commit = nsmm.head.commit
for branch in ("some_virtual_branch", cur_branch.name):
nsm.config_writer().set_value(Submodule.k_head_option, pygit.Head.to_full_path(branch))
csmbranchchange = rwrepo.index.commit("changed branch to %s" % branch)
nsm.set_parent_commit(csmbranchchange)
# END for each branch to change
# Lets remove our tracking branch to simulate some changes
nsmmh = nsmm.head
assert nsmmh.ref.tracking_branch() is None # never set it up until now
assert not nsmmh.is_detached
#dry run does nothing
rm.update(recursive=False, dry_run=True, progress=prog)
assert nsmmh.ref.tracking_branch() is None
# the real thing does
rm.update(recursive=False, progress=prog)
assert nsmmh.ref.tracking_branch() is not None
assert not nsmmh.is_detached
# recursive update
# =================
# finally we recursively update a module, just to run the code at least once
# remove the module so that it has more work
assert len(nsm.children()) == 1
assert nsm.exists() and nsm.module_exists() and len(nsm.children()) == 1
# assure we pull locally only
nsmc = nsm.children()[0]
nsmc.config_writer().set_value('url', async_url)
rm.update(recursive=True, progress=prog, dry_run=True) # just to run the code
rm.update(recursive=True, progress=prog)
assert len(nsm.children()) == 1 and nsmc.module_exists()
def _do_base_tests(self, rwrepo):
"""Perform all tests in the given repository, it may be bare or nonbare"""
# manual instantiation
smm = Submodule(rwrepo, "\0"*20)
# name needs to be set in advance
self.failUnlessRaises(AttributeError, getattr, smm, 'name')
# iterate - 1 submodule
sms = Submodule.list_items(rwrepo, self.k_subm_current)
assert len(sms) == 1
sm = sms[0]
# at a different time, there is None
assert len(Submodule.list_items(rwrepo, self.k_no_subm_tag)) == 0
assert sm.path == 'git/ext/gitdb'
assert sm.path != sm.name # in our case, we have ids there, which don't equal the path
assert sm.url == 'git://github.com/gitpython-developers/gitdb.git'
assert sm.branch_path == 'refs/heads/master' # the default ...
assert sm.branch_name == 'master'
assert sm.parent_commit == rwrepo.head.commit
# size is always 0
assert sm.size == 0
# the module is not checked-out yet
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.module)
# which is why we can't get the branch either - it points into the module() repository
self.failUnlessRaises(InvalidGitPyRepositoryError, getattr, sm, 'branch')
# branch_path works, as its just a string
assert isinstance(sm.branch_path, basestring)
# some commits earlier we still have a submodule, but its at a different commit
smold = Submodule.iter_items(rwrepo, self.k_subm_changed).next()
assert smold.binsha != sm.binsha
assert smold != sm # the name changed
# force it to reread its information
del(smold._url)
smold.url == sm.url
# test config_reader/writer methods
sm.config_reader()
new_smclone_path = None # keep custom paths for later
new_csmclone_path = None #
if rwrepo.bare:
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.config_writer)
else:
writer = sm.config_writer()
# for faster checkout, set the url to the local path
new_smclone_path = to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, sm.path))
writer.set_value('url', new_smclone_path)
del(writer)
assert sm.config_reader().get_value('url') == new_smclone_path
assert sm.url == new_smclone_path
# END handle bare repo
smold.config_reader()
# cannot get a writer on historical submodules
if not rwrepo.bare:
self.failUnlessRaises(ValueError, smold.config_writer)
# END handle bare repo
# make the old into a new - this doesn't work as the name changed
prev_parent_commit = smold.parent_commit
self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_subm_current)
# the sha is properly updated
smold.set_parent_commit(self.k_subm_changed+"~1")
assert smold.binsha != sm.binsha
# raises if the sm didn't exist in new parent - it keeps its
# parent_commit unchanged
self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_no_subm_tag)
# TEST TODO: if a path in the pygitmodules file, but not in the index, it raises
# TEST UPDATE
##############
# module retrieval is not always possible
if rwrepo.bare:
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.module)
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.remove)
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.add, rwrepo, 'here', 'there')
else:
# its not checked out in our case
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.module)
assert not sm.module_exists()
# currently there is only one submodule
assert len(list(rwrepo.iter_submodules())) == 1
assert sm.binsha != "\0"*20
# TEST ADD
###########
# preliminary tests
# adding existing returns exactly the existing
sma = Submodule.add(rwrepo, sm.name, sm.path)
assert sma.path == sm.path
# no url and no module at path fails
self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", "pathtorepo", url=None)
# CONTINUE UPDATE
#################
# lets update it - its a recursive one too
newdir = os.path.join(sm.abspath, 'dir')
os.makedirs(newdir)
# update fails if the path already exists non-empty
self.failUnlessRaises(OSError, sm.update)
os.rmdir(newdir)
# dry-run does nothing
sm.update(dry_run=True, progress=prog)
assert not sm.module_exists()
assert sm.update() is sm
sm_repopath = sm.path # cache for later
assert sm.module_exists()
assert isinstance(sm.module(), pygit.Repo)
assert sm.module().working_tree_dir == sm.abspath
# INTERLEAVE ADD TEST
#####################
# url must match the one in the existing repository ( if submodule name suggests a new one )
# or we raise
self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", sm.path, "git://someurl/repo.git")
# CONTINUE UPDATE
#################
# we should have setup a tracking branch, which is also active
assert sm.module().head.ref.tracking_branch() is not None
# delete the whole directory and re-initialize
shutil.rmtree(sm.abspath)
assert len(sm.children()) == 0
# dry-run does nothing
sm.update(dry_run=True, recursive=False, progress=prog)
assert len(sm.children()) == 0
sm.update(recursive=False)
assert len(list(rwrepo.iter_submodules())) == 2
assert len(sm.children()) == 1 # its not checked out yet
csm = sm.children()[0]
assert not csm.module_exists()
csm_repopath = csm.path
# adjust the path of the submodules module to point to the local destination
new_csmclone_path = to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, sm.path, csm.path))
csm.config_writer().set_value('url', new_csmclone_path)
assert csm.url == new_csmclone_path
# dry-run does nothing
assert not csm.module_exists()
sm.update(recursive=True, dry_run=True, progress=prog)
assert not csm.module_exists()
# update recursively again
sm.update(recursive=True)
assert csm.module_exists()
# tracking branch once again
csm.module().head.ref.tracking_branch() is not None
# this flushed in a sub-submodule
assert len(list(rwrepo.iter_submodules())) == 2
# reset both heads to the previous version, verify that to_latest_revision works
smods = (sm.module(), csm.module())
for repo in smods:
repo.head.reset('HEAD~2', working_tree=1)
# END for each repo to reset
# dry run does nothing
sm.update(recursive=True, dry_run=True, progress=prog)
for repo in smods:
assert repo.head.commit != repo.head.ref.tracking_branch().commit
# END for each repo to check
sm.update(recursive=True, to_latest_revision=True)
for repo in smods:
assert repo.head.commit == repo.head.ref.tracking_branch().commit
# END for each repo to check
del(smods)
# if the head is detached, it still works ( but warns )
smref = sm.module().head.ref
sm.module().head.ref = 'HEAD~1'
# if there is no tracking branch, we get a warning as well
csm_tracking_branch = csm.module().head.ref.tracking_branch()
csm.module().head.ref.set_tracking_branch(None)
sm.update(recursive=True, to_latest_revision=True)
# to_latest_revision changes the child submodule's commit, it needs an
# update now
csm.set_parent_commit(csm.repo.head.commit)
# undo the changes
sm.module().head.ref = smref
csm.module().head.ref.set_tracking_branch(csm_tracking_branch)
# REMOVAL OF REPOSITOTRY
########################
# must delete something
self.failUnlessRaises(ValueError, csm.remove, module=False, configuration=False)
# We have modified the configuration, hence the index is dirty, and the
# deletion will fail
# NOTE: As we did a few updates in the meanwhile, the indices were reset
# Hence we create some changes
csm.set_parent_commit(csm.repo.head.commit)
sm.config_writer().set_value("somekey", "somevalue")
csm.config_writer().set_value("okey", "ovalue")
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.remove)
# if we remove the dirty index, it would work
sm.module().index.reset()
# still, we have the file modified
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.remove, dry_run=True)
sm.module().index.reset(working_tree=True)
# this would work
assert sm.remove(dry_run=True) is sm
assert sm.module_exists()
sm.remove(force=True, dry_run=True)
assert sm.module_exists()
# but ... we have untracked files in the child submodule
fn = join_path_native(csm.module().working_tree_dir, "newfile")
open(fn, 'w').write("hi")
self.failUnlessRaises(InvalidGitPyRepositoryError, sm.remove)
# forcibly delete the child repository
assert csm.remove(force=True) is csm
assert not csm.exists()
assert not csm.module_exists()
assert len(sm.children()) == 0
# now we have a changed index, as configuration was altered.
# fix this
sm.module().index.reset(working_tree=True)
# now delete only the module of the main submodule
assert sm.module_exists()
sm.remove(configuration=False)
assert sm.exists()
assert not sm.module_exists()
assert sm.config_reader().get_value('url')
# delete the rest
sm.remove()
assert not sm.exists()
assert not sm.module_exists()
assert len(rwrepo.submodules) == 0
# ADD NEW SUBMODULE
###################
# add a simple remote repo - trailing slashes are no problem
smid = "newsub"
osmid = "othersub"
nsm = Submodule.add(rwrepo, smid, sm_repopath, new_smclone_path+"/", None, no_checkout=True)
assert nsm.name == smid
assert nsm.module_exists()
assert nsm.exists()
# its not checked out
assert not os.path.isfile(join_path_native(nsm.module().working_tree_dir, Submodule.k_modules_file))
assert len(rwrepo.submodules) == 1
# add another submodule, but into the root, not as submodule
osm = Submodule.add(rwrepo, osmid, csm_repopath, new_csmclone_path, Submodule.k_head_default)
assert osm != nsm
assert osm.module_exists()
assert osm.exists()
assert os.path.isfile(join_path_native(osm.module().working_tree_dir, 'setup.py'))
assert len(rwrepo.submodules) == 2
# commit the changes, just to finalize the operation
rwrepo.index.commit("my submod commit")
assert len(rwrepo.submodules) == 2
# needs update as the head changed, it thinks its in the history
# of the repo otherwise
nsm.set_parent_commit(rwrepo.head.commit)
osm.set_parent_commit(rwrepo.head.commit)
# MOVE MODULE
#############
# invalid inptu
self.failUnlessRaises(ValueError, nsm.move, 'doesntmatter', module=False, configuration=False)
# renaming to the same path does nothing
assert nsm.move(sm.path) is nsm
# rename a module
nmp = join_path_native("new", "module", "dir") + "/" # new module path
pmp = nsm.path
abspmp = nsm.abspath
assert nsm.move(nmp) is nsm
nmp = nmp[:-1] # cut last /
assert nsm.path == nmp
assert rwrepo.submodules[0].path == nmp
mpath = 'newsubmodule'
absmpath = join_path_native(rwrepo.working_tree_dir, mpath)
open(absmpath, 'w').write('')
self.failUnlessRaises(ValueError, nsm.move, mpath)
os.remove(absmpath)
# now it works, as we just move it back
nsm.move(pmp)
assert nsm.path == pmp
assert rwrepo.submodules[0].path == pmp
# TODO lowprio: test remaining exceptions ... for now its okay, the code looks right
# REMOVE 'EM ALL
################
# if a submodule's repo has no remotes, it can't be added without an explicit url
osmod = osm.module()
osm.remove(module=False)
for remote in osmod.remotes:
remote.remove(osmod, remote.name)
assert not osm.exists()
self.failUnlessRaises(ValueError, Submodule.add, rwrepo, osmid, csm_repopath, url=None)
# END handle bare mode
# Error if there is no submodule file here
self.failUnlessRaises(IOError, Submodule._config_parser, rwrepo, rwrepo.commit(self.k_no_subm_tag), True)
def move(self, module_path, configuration=True, module=True):
"""Move the submodule to a another module path. This involves physically moving
the repository at our current path, changing the configuration, as well as
adjusting our index entry accordingly.
:param module_path: the path to which to move our module, given as
repository-relative path. Intermediate directories will be created
accordingly. If the path already exists, it must be empty.
Trailling (back)slashes are removed automatically
:param configuration: if True, the configuration will be adjusted to let
the submodule point to the given path.
:param module: if True, the repository managed by this submodule
will be moved, not the configuration. This will effectively
leave your repository in an inconsistent state unless the configuration
and index already point to the target location.
:return: self
:raise ValueError: if the module path existed and was not empty, or was a file
:note: Currently the method is not atomic, and it could leave the repository
in an inconsistent state if a sub-step fails for some reason
"""
if module + configuration < 1:
raise ValueError("You must specify to move at least the module or the configuration of the submodule")
#END handle input
module_path = to_native_path_linux(module_path)
if module_path.endswith('/'):
module_path = module_path[:-1]
# END handle trailing slash
# VERIFY DESTINATION
if module_path == self.path:
return self
#END handle no change
dest_path = join_path_native(self.repo.working_tree_dir, module_path)
if os.path.isfile(dest_path):
raise ValueError("Cannot move repository onto a file: %s" % dest_path)
# END handle target files
index = self.repo.index
tekey = index.entry_key(module_path, 0)
# if the target item already exists, fail
if configuration and tekey in index.entries:
raise ValueError("Index entry for target path did alredy exist")
#END handle index key already there
# remove existing destination
if module:
if os.path.exists(dest_path):
if len(os.listdir(dest_path)):
raise ValueError("Destination module directory was not empty")
#END handle non-emptyness
if os.path.islink(dest_path):
os.remove(dest_path)
else:
os.rmdir(dest_path)
#END handle link
else:
# recreate parent directories
# NOTE: renames() does that now
pass
#END handle existance
# END handle module
# move the module into place if possible
cur_path = self.abspath
renamed_module = False
if module and os.path.exists(cur_path):
os.renames(cur_path, dest_path)
renamed_module = True
#END move physical module
# rename the index entry - have to manipulate the index directly as
# pygit-mv cannot be used on submodules ... yeah
try:
if configuration:
try:
ekey = index.entry_key(self.path, 0)
entry = index.entries[ekey]
del(index.entries[ekey])
nentry = pygit.IndexEntry(entry[:3]+(module_path,)+entry[4:])
index.entries[tekey] = nentry
except KeyError:
raise InvalidGitPyRepositoryError("Submodule's entry at %r did not exist" % (self.path))
#END handle submodule doesn't exist
# update configuration
writer = self.config_writer(index=index) # auto-write
writer.set_value('path', module_path)
self.path = module_path
del(writer)
# END handle configuration flag
except Exception:
if renamed_module:
os.renames(dest_path, cur_path)
# END undo module renaming
raise
#END handle undo rename
return self
def add(cls, repo, name, path, url=None, branch=None, no_checkout=False):
"""Add a new submodule to the given repository. This will alter the index
as well as the .gitmodules file, but will not create a new commit.
If the submodule already exists, no matter if the configuration differs
from the one provided, the existing submodule will be returned.
:param repo: Repository instance which should receive the submodule
:param name: The name/identifier for the submodule
:param path: repository-relative or absolute path at which the submodule
should be located
It will be created as required during the repository initialization.
:param url: pygit-clone compatible URL, see git-clone reference for more information
If None, the repository is assumed to exist, and the url of the first
remote is taken instead. This is useful if you want to make an existing
repository a submodule of anotherone.
:param branch: name of branch at which the submodule should (later) be checked out.
The given branch must exist in the remote repository, and will be checked
out locally as a tracking branch.
It will only be written into the configuration if it not None, which is
when the checked out branch will be the one the remote HEAD pointed to.
The result you get in these situation is somewhat fuzzy, and it is recommended
to specify at least 'master' here.
Examples are 'master' or 'feature/new'
:param no_checkout: if True, and if the repository has to be cloned manually,
no checkout will be performed
:return: The newly created submodule instance
:note: works atomically, such that no change will be done if the repository
update fails for instance"""
if repo.bare:
raise InvalidGitPyRepositoryError("Cannot add submodules to bare repositories")
# END handle bare repos
path = to_native_path_linux(path)
if path.endswith('/'):
path = path[:-1]
# END handle trailing slash
# assure we never put backslashes into the url, as some operating systems
# like it ...
if url != None:
url = to_native_path_linux(url)
#END assure url correctness
# INSTANTIATE INTERMEDIATE SM
sm = cls(repo, cls.NULL_BIN_SHA, cls.k_default_mode, path, name)
if sm.exists():
# reretrieve submodule from tree
try:
return repo.head.commit.tree[path]
except KeyError:
# could only be in index
index = repo.index
entry = index.entries[index.entry_key(path, 0)]
sm.binsha = entry.binsha
return sm
# END handle exceptions
# END handle existing
# fake-repo - we only need the functionality on the branch instance
br = pygit.Head(repo, git.Head.to_full_path(str(branch) or cls.k_head_default))
has_module = sm.module_exists()
branch_is_default = branch is None
if has_module and url is not None:
if url not in [r.url for r in sm.module().remotes]:
raise ValueError("Specified URL '%s' does not match any remote url of the repository at '%s'" % (url, sm.abspath))
# END check url
# END verify urls match
mrepo = None
if url is None:
if not has_module:
raise ValueError("A URL was not given and existing repository did not exsit at %s" % path)
# END check url
mrepo = sm.module()
urls = [r.url for r in mrepo.remotes]
if not urls:
raise ValueError("Didn't find any remote url in repository at %s" % sm.abspath)
# END verify we have url
url = urls[0]
else:
# clone new repo
kwargs = {'n' : no_checkout}
if not branch_is_default:
kwargs['b'] = br.name
# END setup checkout-branch
mrepo = pygit.Repo.clone_from(url, path, **kwargs)
# END verify url
# update configuration and index
index = sm.repo.index
writer = sm.config_writer(index=index, write=False)
writer.set_value('url', url)
writer.set_value('path', path)
sm._url = url
if not branch_is_default:
# store full path
writer.set_value(cls.k_head_option, br.path)
sm._branch_path = br.path
# END handle path
del(writer)
# we deliberatly assume that our head matches our index !
pcommit = repo.head.commit
sm._parent_commit = pcommit
sm.binsha = mrepo.head.commit.binsha
index.add([sm], write=True)
return sm