3 bup_python="$(dirname "$0")/bup-python" || exit $?
4 exec "$bup_python" "$0" ${1+"$@"}
8 from __future__ import absolute_import, print_function
9 from binascii import hexlify
10 from errno import EACCES
11 from io import BytesIO
12 import os, sys, stat, time, math
14 from bup import hashsplit, git, options, index, client, metadata, hlinkdb
15 from bup.compat import argv_bytes, environ
16 from bup.hashsplit import GIT_MODE_TREE, GIT_MODE_FILE, GIT_MODE_SYMLINK
17 from bup.helpers import (add_error, grafted_path_components, handle_ctrl_c,
18 hostname, istty2, log, parse_date_or_fatal, parse_num,
19 path_components, progress, qprogress, resolve_parent,
20 saved_errors, stripped_path_components,
22 from bup.io import byte_stream, path_msg
23 from bup.pwdgrp import userfullname, username
27 bup save [-tc] [-n name] <filenames...>
29 r,remote= hostname:/path/to/repo of remote repository
30 t,tree output a tree id
31 c,commit output a commit id
32 n,name= name of backup set to update (if any)
33 d,date= date for the commit (seconds since the epoch)
34 v,verbose increase log output (can be used more than once)
35 q,quiet don't show progress meter
36 smaller= only back up files smaller than n bytes
37 bwlimit= maximum bytes/sec to transmit to server
38 f,indexfile= the name of the index file (normally BUP_DIR/bupindex)
39 strip strips the path to every filename given
40 strip-path= path-prefix to be stripped when saving
41 graft= a graft point *old_path*=*new_path* (can be used more than once)
42 #,compress= set compression level to # (0-9, 9 is highest) [1]
44 o = options.Options(optspec)
45 (opt, flags, extra) = o.parse(sys.argv[1:])
48 opt.indexfile = argv_bytes(opt.indexfile)
50 opt.name = argv_bytes(opt.name)
52 opt.remote = argv_bytes(opt.remote)
54 opt.strip_path = argv_bytes(opt.strip_path)
56 git.check_repo_or_die()
57 if not (opt.tree or opt.commit or opt.name):
58 o.fatal("use one or more of -t, -c, -n")
60 o.fatal("no filenames given")
62 extra = [argv_bytes(x) for x in extra]
64 opt.progress = (istty2 and not opt.quiet)
65 opt.smaller = parse_num(opt.smaller or 0)
67 client.bwlimit = parse_num(opt.bwlimit)
70 date = parse_date_or_fatal(opt.date, o.fatal)
74 if opt.strip and opt.strip_path:
75 o.fatal("--strip is incompatible with --strip-path")
80 o.fatal("--strip is incompatible with --graft")
83 o.fatal("--strip-path is incompatible with --graft")
85 for (option, parameter) in flags:
86 if option == "--graft":
87 parameter = argv_bytes(parameter)
88 splitted_parameter = parameter.split(b'=')
89 if len(splitted_parameter) != 2:
90 o.fatal("a graft point must be of the form old_path=new_path")
91 old_path, new_path = splitted_parameter
92 if not (old_path and new_path):
93 o.fatal("a graft point cannot be empty")
94 graft_points.append((resolve_parent(old_path),
95 resolve_parent(new_path)))
97 is_reverse = environ.get(b'BUP_SERVER_REVERSE')
98 if is_reverse and opt.remote:
99 o.fatal("don't use -r in reverse mode; it's automatic")
102 if name and not valid_save_name(name):
103 o.fatal("'%s' is not a valid branch name" % path_msg(name))
104 refname = name and b'refs/heads/%s' % name or None
105 if opt.remote or is_reverse:
107 cli = client.Client(opt.remote)
108 except client.ClientError as e:
111 oldref = refname and cli.read_ref(refname) or None
112 w = cli.new_packwriter(compression_level=opt.compress)
115 oldref = refname and git.read_ref(refname) or None
116 w = git.PackWriter(compression_level=opt.compress)
121 # Metadata is stored in a file named .bupm in each directory. The
122 # first metadata entry will be the metadata for the current directory.
123 # The remaining entries will be for each of the other directory
124 # elements, in the order they're listed in the index.
126 # Since the git tree elements are sorted according to
127 # git.shalist_item_sort_key, the metalist items are accumulated as
128 # (sort_key, metadata) tuples, and then sorted when the .bupm file is
129 # created. The sort_key must be computed using the element's real
130 # name and mode rather than the git mode and (possibly mangled) name.
132 # Maintain a stack of information representing the current location in
133 # the archive being constructed. The current path is recorded in
134 # parts, which will be something like ['', 'home', 'someuser'], and
135 # the accumulated content and metadata for of the dirs in parts is
136 # stored in parallel stacks in shalists and metalists.
138 parts = [] # Current archive position (stack of dir names).
139 shalists = [] # Hashes for each dir in paths.
140 metalists = [] # Metadata for each dir in paths.
143 def _push(part, metadata):
144 # Enter a new archive directory -- make it the current directory.
147 metalists.append([(b'', metadata)]) # This dir's metadata (no name).
150 def _pop(force_tree, dir_metadata=None):
151 # Leave the current archive directory and add its tree to its parent.
152 assert(len(parts) >= 1)
154 shalist = shalists.pop()
155 metalist = metalists.pop()
156 # FIXME: only test if collision is possible (i.e. given --strip, etc.)?
164 if name in names_seen:
165 parent_path = b'/'.join(parts) + b'/'
166 add_error('error: ignoring duplicate path %s in %s'
167 % (path_msg(name), path_msg(parent_path)))
173 if dir_metadata: # Override the original metadata pushed for this dir.
174 metalist = [(b'', dir_metadata)] + metalist[1:]
175 sorted_metalist = sorted(metalist, key = lambda x : x[0])
176 metadata = b''.join([m[1].encode() for m in sorted_metalist])
177 metadata_f = BytesIO(metadata)
178 mode, id = hashsplit.split_to_blob_or_tree(w.new_blob, w.new_tree,
180 keep_boundaries=False)
181 clean_list.append((mode, b'.bupm', id))
182 tree = w.new_tree(clean_list)
184 shalists[-1].append((GIT_MODE_TREE,
185 git.mangle_name(part,
186 GIT_MODE_TREE, GIT_MODE_TREE),
192 def progress_report(n):
193 global count, subcount, lastremain
195 cc = count + subcount
196 pct = total and (cc*100.0/total) or 0
198 elapsed = now - tstart
199 kps = elapsed and int(cc/1024./elapsed)
200 kps_frac = 10 ** int(math.log(kps+1, 10) - 1)
201 kps = int(kps/kps_frac)*kps_frac
203 remain = elapsed*1.0/cc * (total-cc)
206 if (lastremain and (remain > lastremain)
207 and ((remain - lastremain)/lastremain < 0.05)):
211 hours = int(remain/60/60)
212 mins = int(remain/60 - hours*60)
213 secs = int(remain - hours*60*60 - mins*60)
218 kpsstr = '%dk/s' % kps
220 remainstr = '%dh%dm' % (hours, mins)
222 remainstr = '%dm%d' % (mins, secs)
224 remainstr = '%ds' % secs
225 qprogress('Saving: %.2f%% (%d/%dk, %d/%d files) %s %s\r'
226 % (pct, cc/1024, total/1024, fcount, ftotal,
230 indexfile = opt.indexfile or git.repo(b'bupindex')
231 r = index.Reader(indexfile)
233 msr = index.MetaStoreReader(indexfile + b'.meta')
234 except IOError as ex:
235 if ex.errno != EACCES:
237 log('error: cannot access %r; have you run bup index?'
238 % path_msg(indexfile))
240 hlink_db = hlinkdb.HLinkDB(indexfile + b'.hlink')
242 def already_saved(ent):
243 return ent.is_valid() and w.exists(ent.sha) and ent.sha
245 def wantrecurse_pre(ent):
246 return not already_saved(ent)
248 def wantrecurse_during(ent):
249 return not already_saved(ent) or ent.sha_missing()
251 def find_hardlink_target(hlink_db, ent):
252 if hlink_db and not stat.S_ISDIR(ent.mode) and ent.nlink > 1:
253 link_paths = hlink_db.node_paths(ent.dev, ent.ino)
259 for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_pre):
260 if not (ftotal % 10024):
261 qprogress('Reading index: %d\r' % ftotal)
262 exists = ent.exists()
263 hashvalid = already_saved(ent)
264 ent.set_sha_missing(not hashvalid)
265 if not opt.smaller or ent.size < opt.smaller:
266 if exists and not hashvalid:
269 progress('Reading index: %d, done.\n' % ftotal)
270 hashsplit.progress_callback = progress_report
272 # Root collisions occur when strip or graft options map more than one
273 # path to the same directory (paths which originally had separate
274 # parents). When that situation is detected, use empty metadata for
275 # the parent. Otherwise, use the metadata for the common parent.
276 # Collision example: "bup save ... --strip /foo /foo/bar /bar".
278 # FIXME: Add collision tests, or handle collisions some other way.
280 # FIXME: Detect/handle strip/graft name collisions (other than root),
281 # i.e. if '/foo/bar' and '/bar' both map to '/'.
284 root_collision = None
286 count = subcount = fcount = 0
289 for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_during):
290 (dir, file) = os.path.split(ent.name)
291 exists = (ent.flags & index.IX_EXISTS)
292 hashvalid = already_saved(ent)
293 wasmissing = ent.sha_missing()
299 if ent.sha == index.EMPTY_SHA:
306 log('%s %-70s\n' % (status, path_msg(ent.name)))
307 elif not stat.S_ISDIR(ent.mode) and lastdir != dir:
308 if not lastdir.startswith(dir):
309 log('%s %-70s\n' % (status, path_msg(os.path.join(dir, b''))))
318 if opt.smaller and ent.size >= opt.smaller:
319 if exists and not hashvalid:
321 log('skipping large file "%s"\n' % path_msg(ent.name))
322 lastskip_name = ent.name
325 assert(dir.startswith(b'/'))
327 dirp = stripped_path_components(dir, extra)
329 dirp = stripped_path_components(dir, [opt.strip_path])
331 dirp = grafted_path_components(graft_points, dir)
333 dirp = path_components(dir)
335 # At this point, dirp contains a representation of the archive
336 # path that looks like [(archive_dir_name, real_fs_path), ...].
337 # So given "bup save ... --strip /foo/bar /foo/bar/baz", dirp
338 # might look like this at some point:
339 # [('', '/foo/bar'), ('baz', '/foo/bar/baz'), ...].
341 # This dual representation supports stripping/grafting, where the
342 # archive path may not have a direct correspondence with the
343 # filesystem. The root directory is represented by an initial
344 # component named '', and any component that doesn't have a
345 # corresponding filesystem directory (due to grafting, for
346 # example) will have a real_fs_path of None, i.e. [('', None),
349 if first_root == None:
351 elif first_root != dirp[0]:
352 root_collision = True
354 # If switching to a new sub-tree, finish the current sub-tree.
355 while parts > [x[0] for x in dirp]:
356 _pop(force_tree = None)
358 # If switching to a new sub-tree, start a new sub-tree.
359 for path_component in dirp[len(parts):]:
360 dir_name, fs_path = path_component
361 # Not indexed, so just grab the FS metadata or use empty metadata.
363 meta = metadata.from_path(fs_path, normalized=True) \
364 if fs_path else metadata.Metadata()
365 except (OSError, IOError) as e:
367 lastskip_name = dir_name
368 meta = metadata.Metadata()
369 _push(dir_name, meta)
373 continue # We're at the top level -- keep the current root dir
374 # Since there's no filename, this is a subdir -- finish it.
375 oldtree = already_saved(ent) # may be None
376 newtree = _pop(force_tree = oldtree)
378 if lastskip_name and lastskip_name.startswith(ent.name):
381 ent.validate(GIT_MODE_TREE, newtree)
383 if exists and wasmissing:
387 # it's not a directory
390 git_name = git.mangle_name(file, ent.mode, ent.gitmode)
391 git_info = (ent.gitmode, git_name, id)
392 shalists[-1].append(git_info)
393 sort_key = git.shalist_item_sort_key((ent.mode, file, id))
394 meta = msr.metadata_at(ent.meta_ofs)
395 meta.hardlink_target = find_hardlink_target(hlink_db, ent)
396 # Restore the times that were cleared to 0 in the metastore.
397 (meta.atime, meta.mtime, meta.ctime) = (ent.atime, ent.mtime, ent.ctime)
398 metalists[-1].append((sort_key, meta))
401 if stat.S_ISREG(ent.mode):
403 f = hashsplit.open_noatime(ent.name)
404 except (IOError, OSError) as e:
406 lastskip_name = ent.name
409 (mode, id) = hashsplit.split_to_blob_or_tree(
410 w.new_blob, w.new_tree, [f],
411 keep_boundaries=False)
412 except (IOError, OSError) as e:
413 add_error('%s: %s' % (ent.name, e))
414 lastskip_name = ent.name
415 elif stat.S_ISDIR(ent.mode):
416 assert(0) # handled above
417 elif stat.S_ISLNK(ent.mode):
419 rl = os.readlink(ent.name)
420 except (OSError, IOError) as e:
422 lastskip_name = ent.name
424 (mode, id) = (GIT_MODE_SYMLINK, w.new_blob(rl))
426 # Everything else should be fully described by its
427 # metadata, so just record an empty blob, so the paths
428 # in the tree and .bupm will match up.
429 (mode, id) = (GIT_MODE_FILE, w.new_blob(b''))
432 ent.validate(mode, id)
434 git_name = git.mangle_name(file, ent.mode, ent.gitmode)
435 git_info = (mode, git_name, id)
436 shalists[-1].append(git_info)
437 sort_key = git.shalist_item_sort_key((ent.mode, file, id))
438 hlink = find_hardlink_target(hlink_db, ent)
440 meta = metadata.from_path(ent.name, hardlink_target=hlink,
442 except (OSError, IOError) as e:
444 lastskip_name = ent.name
445 meta = metadata.Metadata()
446 metalists[-1].append((sort_key, meta))
448 if exists and wasmissing:
454 pct = total and count*100.0/total or 100
455 progress('Saving: %.2f%% (%d/%dk, %d/%d files), done. \n'
456 % (pct, count/1024, total/1024, fcount, ftotal))
458 while len(parts) > 1: # _pop() all the parts above the root
459 _pop(force_tree = None)
460 assert(len(shalists) == 1)
461 assert(len(metalists) == 1)
463 # Finish the root directory.
464 tree = _pop(force_tree = None,
465 # When there's a collision, use empty metadata for the root.
466 dir_metadata = metadata.Metadata() if root_collision else None)
469 out = byte_stream(sys.stdout)
472 out.write(hexlify(tree))
474 if opt.commit or name:
475 msg = (b'bup save\n\nGenerated by command:\n%r\n'
476 % [argv_bytes(x) for x in sys.argv])
477 userline = (b'%s <%s@%s>' % (userfullname(), username(), hostname()))
478 commit = w.new_commit(tree, oldref, userline, date, None,
479 userline, date, None, msg)
481 out.write(hexlify(commit))
485 w.close() # must close before we can update the ref
489 cli.update_ref(refname, commit, oldref)
491 git.update_ref(refname, commit, oldref)
497 log('WARNING: %d errors encountered while saving.\n' % len(saved_errors))