2 import sys, stat, time, math
3 from bup import hashsplit, git, options, index, client, metadata, hlinkdb
4 from bup.helpers import *
5 from bup.hashsplit import GIT_MODE_TREE, GIT_MODE_FILE, GIT_MODE_SYMLINK
9 bup save [-tc] [-n name] <filenames...>
11 r,remote= hostname:/path/to/repo of remote repository
12 t,tree output a tree id
13 c,commit output a commit id
14 n,name= name of backup set to update (if any)
15 d,date= date for the commit (seconds since the epoch)
16 v,verbose increase log output (can be used more than once)
17 q,quiet don't show progress meter
18 smaller= only back up files smaller than n bytes
19 bwlimit= maximum bytes/sec to transmit to server
20 f,indexfile= the name of the index file (normally BUP_DIR/bupindex)
21 strip strips the path to every filename given
22 strip-path= path-prefix to be stripped when saving
23 graft= a graft point *old_path*=*new_path* (can be used more than once)
24 #,compress= set compression level to # (0-9, 9 is highest) [1]
26 o = options.Options(optspec)
27 (opt, flags, extra) = o.parse(sys.argv[1:])
29 git.check_repo_or_die()
30 if not (opt.tree or opt.commit or opt.name):
31 o.fatal("use one or more of -t, -c, -n")
33 o.fatal("no filenames given")
35 opt.progress = (istty2 and not opt.quiet)
36 opt.smaller = parse_num(opt.smaller or 0)
38 client.bwlimit = parse_num(opt.bwlimit)
41 date = parse_date_or_fatal(opt.date, o.fatal)
45 if opt.strip and opt.strip_path:
46 o.fatal("--strip is incompatible with --strip-path")
51 o.fatal("--strip is incompatible with --graft")
54 o.fatal("--strip-path is incompatible with --graft")
56 for (option, parameter) in flags:
57 if option == "--graft":
58 splitted_parameter = parameter.split('=')
59 if len(splitted_parameter) != 2:
60 o.fatal("a graft point must be of the form old_path=new_path")
61 old_path, new_path = splitted_parameter
62 if not (old_path and new_path):
63 o.fatal("a graft point cannot be empty")
64 graft_points.append((realpath(old_path), realpath(new_path)))
66 is_reverse = os.environ.get('BUP_SERVER_REVERSE')
67 if is_reverse and opt.remote:
68 o.fatal("don't use -r in reverse mode; it's automatic")
70 if opt.name and opt.name.startswith('.'):
71 o.fatal("'%s' is not a valid branch name" % opt.name)
72 refname = opt.name and 'refs/heads/%s' % opt.name or None
73 if opt.remote or is_reverse:
74 cli = client.Client(opt.remote)
75 oldref = refname and cli.read_ref(refname) or None
76 w = cli.new_packwriter()
79 oldref = refname and git.read_ref(refname) or None
80 w = git.PackWriter(compression_level=opt.compress)
92 # Metadata is stored in a file named .bupm in each directory. The
93 # first metadata entry will be the metadata for the current directory.
94 # The remaining entries will be for each of the other directory
95 # elements, in the order they're listed in the index.
97 # Since the git tree elements are sorted according to
98 # git.shalist_item_sort_key, the metalist items are accumulated as
99 # (sort_key, metadata) tuples, and then sorted when the .bupm file is
100 # created. The sort_key must be computed using the element's real
101 # name and mode rather than the git mode and (possibly mangled) name.
103 # Maintain a stack of information representing the current location in
104 # the archive being constructed. The current path is recorded in
105 # parts, which will be something like ['', 'home', 'someuser'], and
106 # the accumulated content and metadata for of the dirs in parts is
107 # stored in parallel stacks in shalists and metalists.
109 parts = [] # Current archive position (stack of dir names).
110 shalists = [] # Hashes for each dir in paths.
111 metalists = [] # Metadata for each dir in paths.
114 def _push(part, metadata):
115 # Enter a new archive directory -- make it the current directory.
118 metalists.append([('', metadata)]) # This dir's metadata (no name).
121 def _pop(force_tree, dir_metadata=None):
122 # Leave the current archive directory and add its tree to its parent.
123 assert(len(parts) >= 1)
125 shalist = shalists.pop()
126 metalist = metalists.pop()
128 if dir_metadata: # Override the original metadata pushed for this dir.
129 metalist = [('', dir_metadata)] + metalist[1:]
130 sorted_metalist = sorted(metalist, key = lambda x : x[0])
131 metadata = ''.join([m[1].encode() for m in sorted_metalist])
132 shalist.append((0100644, '.bupm', w.new_blob(metadata)))
133 tree = force_tree or w.new_tree(shalist)
135 shalists[-1].append((GIT_MODE_TREE,
136 git.mangle_name(part,
137 GIT_MODE_TREE, GIT_MODE_TREE),
143 def progress_report(n):
144 global count, subcount, lastremain
146 cc = count + subcount
147 pct = total and (cc*100.0/total) or 0
149 elapsed = now - tstart
150 kps = elapsed and int(cc/1024./elapsed)
151 kps_frac = 10 ** int(math.log(kps+1, 10) - 1)
152 kps = int(kps/kps_frac)*kps_frac
154 remain = elapsed*1.0/cc * (total-cc)
157 if (lastremain and (remain > lastremain)
158 and ((remain - lastremain)/lastremain < 0.05)):
162 hours = int(remain/60/60)
163 mins = int(remain/60 - hours*60)
164 secs = int(remain - hours*60*60 - mins*60)
169 kpsstr = '%dk/s' % kps
171 remainstr = '%dh%dm' % (hours, mins)
173 remainstr = '%dm%d' % (mins, secs)
175 remainstr = '%ds' % secs
176 qprogress('Saving: %.2f%% (%d/%dk, %d/%d files) %s %s\r'
177 % (pct, cc/1024, total/1024, fcount, ftotal,
181 indexfile = opt.indexfile or git.repo('bupindex')
182 r = index.Reader(indexfile)
183 hlink_db = hlinkdb.HLinkDB(indexfile + '.hlink')
185 def already_saved(ent):
186 return ent.is_valid() and w.exists(ent.sha) and ent.sha
188 def wantrecurse_pre(ent):
189 return not already_saved(ent)
191 def wantrecurse_during(ent):
192 return not already_saved(ent) or ent.sha_missing()
194 def find_hardlink_target(hlink_db, ent):
195 if hlink_db and not stat.S_ISDIR(ent.mode) and ent.nlink > 1:
196 link_paths = hlink_db.node_paths(ent.dev, ent.ino)
202 for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_pre):
203 if not (ftotal % 10024):
204 qprogress('Reading index: %d\r' % ftotal)
205 exists = ent.exists()
206 hashvalid = already_saved(ent)
207 ent.set_sha_missing(not hashvalid)
208 if not opt.smaller or ent.size < opt.smaller:
209 if exists and not hashvalid:
212 progress('Reading index: %d, done.\n' % ftotal)
213 hashsplit.progress_callback = progress_report
215 # Root collisions occur when strip or graft options map more than one
216 # path to the same directory (paths which originally had separate
217 # parents). When that situation is detected, use empty metadata for
218 # the parent. Otherwise, use the metadata for the common parent.
219 # Collision example: "bup save ... --strip /foo /foo/bar /bar".
221 # FIXME: Add collision tests, or handle collisions some other way.
223 # FIXME: Detect/handle strip/graft name collisions (other than root),
224 # i.e. if '/foo/bar' and '/bar' both map to '/'.
227 root_collision = None
229 count = subcount = fcount = 0
232 for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_during):
233 (dir, file) = os.path.split(ent.name)
234 exists = (ent.flags & index.IX_EXISTS)
235 hashvalid = already_saved(ent)
236 wasmissing = ent.sha_missing()
242 if ent.sha == index.EMPTY_SHA:
249 log('%s %-70s\n' % (status, ent.name))
250 elif not stat.S_ISDIR(ent.mode) and lastdir != dir:
251 if not lastdir.startswith(dir):
252 log('%s %-70s\n' % (status, os.path.join(dir, '')))
261 if opt.smaller and ent.size >= opt.smaller:
262 if exists and not hashvalid:
263 add_error('skipping large file "%s"' % ent.name)
264 lastskip_name = ent.name
267 assert(dir.startswith('/'))
269 dirp = stripped_path_components(dir, extra)
271 dirp = stripped_path_components(dir, [opt.strip_path])
273 dirp = grafted_path_components(graft_points, dir)
275 dirp = path_components(dir)
277 # At this point, dirp contains a representation of the archive
278 # path that looks like [(archive_dir_name, real_fs_path), ...].
279 # So given "bup save ... --strip /foo/bar /foo/bar/baz", dirp
280 # might look like this at some point:
281 # [('', '/foo/bar'), ('baz', '/foo/bar/baz'), ...].
283 # This dual representation supports stripping/grafting, where the
284 # archive path may not have a direct correspondence with the
285 # filesystem. The root directory is represented by an initial
286 # component named '', and any component that doesn't have a
287 # corresponding filesystem directory (due to grafting, for
288 # example) will have a real_fs_path of None, i.e. [('', None),
291 if first_root == None:
292 dir_name, fs_path = dirp[0]
294 meta = metadata.from_path(fs_path) if fs_path else metadata.Metadata()
295 _push(dir_name, meta)
296 elif first_root != dirp[0]:
297 root_collision = True
299 # If switching to a new sub-tree, finish the current sub-tree.
300 while parts > [x[0] for x in dirp]:
301 _pop(force_tree = None)
303 # If switching to a new sub-tree, start a new sub-tree.
304 for path_component in dirp[len(parts):]:
305 dir_name, fs_path = path_component
306 meta = metadata.from_path(fs_path) if fs_path else metadata.Metadata()
307 _push(dir_name, meta)
311 continue # We're at the top level -- keep the current root dir
312 # Since there's no filename, this is a subdir -- finish it.
313 oldtree = already_saved(ent) # may be None
314 newtree = _pop(force_tree = oldtree)
316 if lastskip_name and lastskip_name.startswith(ent.name):
319 ent.validate(GIT_MODE_TREE, newtree)
321 if exists and wasmissing:
325 # it's not a directory
329 git_name = git.mangle_name(file, ent.mode, ent.gitmode)
330 git_info = (ent.gitmode, git_name, id)
331 shalists[-1].append(git_info)
332 sort_key = git.shalist_item_sort_key((ent.mode, file, id))
333 hlink = find_hardlink_target(hlink_db, ent)
334 metalists[-1].append((sort_key,
335 metadata.from_path(ent.name,
336 hardlink_target=hlink)))
338 if stat.S_ISREG(ent.mode):
340 f = hashsplit.open_noatime(ent.name)
341 except (IOError, OSError), e:
343 lastskip_name = ent.name
346 (mode, id) = hashsplit.split_to_blob_or_tree(
347 w.new_blob, w.new_tree, [f],
348 keep_boundaries=False)
349 except (IOError, OSError), e:
350 add_error('%s: %s' % (ent.name, e))
351 lastskip_name = ent.name
353 if stat.S_ISDIR(ent.mode):
354 assert(0) # handled above
355 elif stat.S_ISLNK(ent.mode):
357 rl = os.readlink(ent.name)
358 except (OSError, IOError), e:
360 lastskip_name = ent.name
362 (mode, id) = (GIT_MODE_SYMLINK, w.new_blob(rl))
364 # Everything else should be fully described by its
365 # metadata, so just record an empty blob, so the paths
366 # in the tree and .bupm will match up.
367 (mode, id) = (GIT_MODE_FILE, w.new_blob(""))
370 ent.validate(mode, id)
372 git_name = git.mangle_name(file, ent.mode, ent.gitmode)
373 git_info = (mode, git_name, id)
374 shalists[-1].append(git_info)
375 sort_key = git.shalist_item_sort_key((ent.mode, file, id))
376 hlink = find_hardlink_target(hlink_db, ent)
377 metalists[-1].append((sort_key,
378 metadata.from_path(ent.name,
379 hardlink_target=hlink)))
380 if exists and wasmissing:
386 pct = total and count*100.0/total or 100
387 progress('Saving: %.2f%% (%d/%dk, %d/%d files), done. \n'
388 % (pct, count/1024, total/1024, fcount, ftotal))
390 while len(parts) > 1: # _pop() all the parts above the root
391 _pop(force_tree = None)
392 assert(len(shalists) == 1)
393 assert(len(metalists) == 1)
395 # Finish the root directory.
396 tree = _pop(force_tree = None,
397 dir_metadata = metadata.Metadata() if root_collision else None)
400 print tree.encode('hex')
401 if opt.commit or opt.name:
402 msg = 'bup save\n\nGenerated by command:\n%r' % sys.argv
403 commit = w.new_commit(oldref, tree, date, msg)
405 print commit.encode('hex')
407 w.close() # must close before we can update the ref
411 cli.update_ref(refname, commit, oldref)
413 git.update_ref(refname, commit, oldref)
419 log('WARNING: %d errors encountered while saving.\n' % len(saved_errors))