3 bup_python="$(dirname "$0")/bup-python" || exit $?
4 exec "$bup_python" "$0" ${1+"$@"}
8 from __future__ import absolute_import
9 from stat import S_ISDIR
10 import copy, errno, os, sys, stat, re
12 from bup import options, git, metadata, vfs
13 from bup._helpers import write_sparsely
14 from bup.compat import argv_bytes, fsencode, wrap_main
15 from bup.helpers import (add_error, chunkyreader, die_if_errors, handle_ctrl_c,
16 log, mkdirp, parse_rx_excludes, progress, qprogress,
17 saved_errors, should_rx_exclude_path, unlink)
18 from bup.io import byte_stream
19 from bup.repo import LocalRepo, RemoteRepo
23 bup restore [-r host:path] [-C outdir] </branch/revision/path/to/dir ...>
25 r,remote= remote repository path
26 C,outdir= change to given outdir before extracting files
27 numeric-ids restore numeric IDs (user, group, etc.) rather than names
28 exclude-rx= skip paths matching the unanchored regex (may be repeated)
29 exclude-rx-from= skip --exclude-rx patterns in file (may be repeated)
30 sparse create sparse files
31 v,verbose increase log output (can be used more than once)
32 map-user= given OLD=NEW, restore OLD user as NEW user
33 map-group= given OLD=NEW, restore OLD group as NEW group
34 map-uid= given OLD=NEW, restore OLD uid as NEW uid
35 map-gid= given OLD=NEW, restore OLD gid as NEW gid
36 q,quiet don't show progress meter
41 # stdout should be flushed after each line, even when not connected to a tty
42 stdoutfd = sys.stdout.fileno()
44 sys.stdout = os.fdopen(stdoutfd, 'w', 1)
45 out = byte_stream(sys.stdout)
47 def valid_restore_path(path):
48 path = os.path.normpath(path)
49 if path.startswith(b'/'):
54 def parse_owner_mappings(type, options, fatal):
55 """Traverse the options and parse all --map-TYPEs, or call Option.fatal()."""
56 opt_name = '--map-' + type
57 if type in ('uid', 'gid'):
58 value_rx = re.compile(br'^(-?[0-9]+)=(-?[0-9]+)$')
60 value_rx = re.compile(br'^([^=]+)=([^=]*)$')
63 (option, parameter) = flag
64 if option != opt_name:
66 parameter = argv_bytes(parameter)
67 match = value_rx.match(parameter)
69 raise fatal("couldn't parse %r as %s mapping" % (parameter, type))
70 old_id, new_id = match.groups()
71 if type in ('uid', 'gid'):
74 owner_map[old_id] = new_id
77 def apply_metadata(meta, name, restore_numeric_ids, owner_map):
78 m = copy.deepcopy(meta)
79 m.user = owner_map['user'].get(m.user, m.user)
80 m.group = owner_map['group'].get(m.group, m.group)
81 m.uid = owner_map['uid'].get(m.uid, m.uid)
82 m.gid = owner_map['gid'].get(m.gid, m.gid)
83 m.apply_to_path(name, restore_numeric_ids = restore_numeric_ids)
85 def hardlink_compatible(prev_path, prev_item, new_item, top):
86 prev_candidate = top + prev_path
87 if not os.path.exists(prev_candidate):
89 prev_meta, new_meta = prev_item.meta, new_item.meta
90 if new_item.oid != prev_item.oid \
91 or new_meta.mtime != prev_meta.mtime \
92 or new_meta.ctime != prev_meta.ctime \
93 or new_meta.mode != prev_meta.mode:
95 # FIXME: should we be checking the path on disk, or the recorded metadata?
96 # The exists() above might seem to suggest the former.
97 if not new_meta.same_file(prev_meta):
101 def hardlink_if_possible(fullname, item, top, hardlinks):
102 """Find a suitable hardlink target, link to it, and return true,
103 otherwise return false."""
104 # The cwd will be dirname(fullname), and fullname will be
105 # absolute, i.e. /foo/bar, and the caller is expected to handle
106 # restoring the metadata if hardlinking isn't possible.
108 # FIXME: we can probably replace the target_vfs_path with the
111 # hardlinks tracks a list of (restore_path, vfs_path, meta)
112 # triples for each path we've written for a given hardlink_target.
113 # This allows us to handle the case where we restore a set of
114 # hardlinks out of order (with respect to the original save
115 # call(s)) -- i.e. when we don't restore the hardlink_target path
116 # first. This data also allows us to attempt to handle other
117 # situations like hardlink sets that change on disk during a save,
118 # or between index and save.
120 target = item.meta.hardlink_target
122 assert(fullname.startswith(b'/'))
123 target_versions = hardlinks.get(target)
125 # Check every path in the set that we've written so far for a match.
126 for prev_path, prev_item in target_versions:
127 if hardlink_compatible(prev_path, prev_item, item, top):
129 os.link(top + prev_path, top + fullname)
132 if e.errno != errno.EXDEV:
136 hardlinks[target] = target_versions
137 target_versions.append((fullname, item))
140 def write_file_content(repo, dest_path, vfs_file):
141 with vfs.fopen(repo, vfs_file) as inf:
142 with open(dest_path, 'wb') as outf:
143 for b in chunkyreader(inf):
146 def write_file_content_sparsely(repo, dest_path, vfs_file):
147 with vfs.fopen(repo, vfs_file) as inf:
148 outfd = os.open(dest_path, os.O_WRONLY | os.O_CREAT | os.O_TRUNC, 0o600)
151 for b in chunkyreader(inf):
152 trailing_zeros = write_sparsely(outfd, b, 512, trailing_zeros)
153 pos = os.lseek(outfd, trailing_zeros, os.SEEK_END)
154 os.ftruncate(outfd, pos)
158 def restore(repo, parent_path, name, item, top, sparse, numeric_ids, owner_map,
159 exclude_rxs, verbosity, hardlinks):
160 global total_restored
161 mode = vfs.item_mode(item)
162 treeish = S_ISDIR(mode)
163 fullname = parent_path + b'/' + name
164 # Match behavior of index --exclude-rx with respect to paths.
165 if should_rx_exclude_path(fullname + (b'/' if treeish else b''),
170 # Do this now so we'll have meta.symlink_target for verbose output
171 item = vfs.augment_item_meta(repo, item, include_size=True)
173 assert(meta.mode == mode)
175 if stat.S_ISDIR(mode):
177 out.write(b'%s/\n' % fullname)
178 elif stat.S_ISLNK(mode):
179 assert(meta.symlink_target)
181 out.write(b'%s@ -> %s\n' % (fullname, meta.symlink_target))
184 out.write(fullname + '\n')
186 orig_cwd = os.getcwd()
189 # Assumes contents() returns '.' with the full metadata first
190 sub_items = vfs.contents(repo, item, want_meta=True)
191 dot, item = next(sub_items, None)
193 item = vfs.augment_item_meta(repo, item, include_size=True)
195 meta.create_path(name)
199 qprogress('Restoring: %d\r' % total_restored)
200 for sub_name, sub_item in sub_items:
201 restore(repo, fullname, sub_name, sub_item, top, sparse,
202 numeric_ids, owner_map, exclude_rxs, verbosity,
205 apply_metadata(meta, name, numeric_ids, owner_map)
207 created_hardlink = False
208 if meta.hardlink_target:
209 created_hardlink = hardlink_if_possible(fullname, item, top,
211 if not created_hardlink:
212 meta.create_path(name)
213 if stat.S_ISREG(meta.mode):
215 write_file_content_sparsely(repo, name, item)
217 write_file_content(repo, name, item)
220 qprogress('Restoring: %d\r' % total_restored)
221 if not created_hardlink:
222 apply_metadata(meta, name, numeric_ids, owner_map)
227 o = options.Options(optspec)
228 opt, flags, extra = o.parse(sys.argv[1:])
229 verbosity = (opt.verbose or 0) if not opt.quiet else -1
231 opt.remote = argv_bytes(opt.remote)
233 opt.outdir = argv_bytes(opt.outdir)
235 git.check_repo_or_die()
238 o.fatal('must specify at least one filename to restore')
240 exclude_rxs = parse_rx_excludes(flags, o.fatal)
243 for map_type in ('user', 'group', 'uid', 'gid'):
244 owner_map[map_type] = parse_owner_mappings(map_type, flags, o.fatal)
250 repo = RemoteRepo(opt.remote) if opt.remote else LocalRepo()
251 top = fsencode(os.getcwd())
253 for path in [argv_bytes(x) for x in extra]:
254 if not valid_restore_path(path):
255 add_error("path %r doesn't include a branch and revision" % path)
258 resolved = vfs.resolve(repo, path, want_meta=True, follow=False)
259 except vfs.IOError as e:
262 if len(resolved) == 3 and resolved[2][0] == b'latest':
263 # Follow latest symlink to the actual save
265 resolved = vfs.resolve(repo, b'latest', parent=resolved[:-1],
267 except vfs.IOError as e:
270 # Rename it back to 'latest'
271 resolved = tuple(elt if i != 2 else (b'latest',) + elt[1:]
272 for i, elt in enumerate(resolved))
273 path_parent, path_name = os.path.split(path)
274 leaf_name, leaf_item = resolved[-1]
276 add_error('error: cannot access %r in %r'
277 % ('/'.join(name for name, item in resolved),
280 if not path_name or path_name == b'.':
281 # Source is /foo/what/ever/ or /foo/what/ever/. -- extract
282 # what/ever/* to the current directory, and if name == '.'
283 # (i.e. /foo/what/ever/.), then also restore what/ever's
284 # metadata to the current directory.
285 treeish = vfs.item_mode(leaf_item)
287 add_error('%r cannot be restored as a directory' % path)
289 items = vfs.contents(repo, leaf_item, want_meta=True)
290 dot, leaf_item = next(items, None)
292 for sub_name, sub_item in items:
293 restore(repo, b'', sub_name, sub_item, top,
294 opt.sparse, opt.numeric_ids, owner_map,
295 exclude_rxs, verbosity, hardlinks)
296 if path_name == b'.':
297 leaf_item = vfs.augment_item_meta(repo, leaf_item,
299 apply_metadata(leaf_item.meta, b'.',
300 opt.numeric_ids, owner_map)
302 restore(repo, b'', leaf_name, leaf_item, top,
303 opt.sparse, opt.numeric_ids, owner_map,
304 exclude_rxs, verbosity, hardlinks)
307 progress('Restoring: %d, done.\n' % total_restored)