1#!/bin/sh 2"""": # -*-python-*- 3# https://sourceware.org/bugzilla/show_bug.cgi?id=26034 4export "BUP_ARGV_0"="$0" 5arg_i=1 6for arg in "$@"; do 7 export "BUP_ARGV_${arg_i}"="$arg" 8 shift 9 arg_i=$((arg_i + 1)) 10done 11# Here to end of preamble replaced during install 12bup_python="$(dirname "$0")/../../config/bin/python" || exit $? 13exec "$bup_python" "$0" 14""" 15# end of bup preamble 16 17from __future__ import absolute_import, print_function 18from binascii import hexlify 19from errno import EACCES 20from io import BytesIO 21import math, os, stat, sys, time 22 23sys.path[:0] = [os.path.dirname(os.path.realpath(__file__)) + '/..'] 24 25from bup import compat, hashsplit, git, options, index, client, metadata 26from bup import hlinkdb 27from bup.compat import argv_bytes, environ 28from bup.hashsplit import GIT_MODE_TREE, GIT_MODE_FILE, GIT_MODE_SYMLINK 29from bup.helpers import (add_error, grafted_path_components, handle_ctrl_c, 30 hostname, istty2, log, parse_date_or_fatal, parse_num, 31 path_components, progress, qprogress, resolve_parent, 32 saved_errors, stripped_path_components, 33 valid_save_name) 34from bup.io import byte_stream, path_msg 35from bup.pwdgrp import userfullname, username 36 37 38optspec = """ 39bup save [-tc] [-n name] <filenames...> 40-- 41r,remote= hostname:/path/to/repo of remote repository 42t,tree output a tree id 43c,commit output a commit id 44n,name= name of backup set to update (if any) 45d,date= date for the commit (seconds since the epoch) 46v,verbose increase log output (can be used more than once) 47q,quiet don't show progress meter 48smaller= only back up files smaller than n bytes 49bwlimit= maximum bytes/sec to transmit to server 50f,indexfile= the name of the index file (normally BUP_DIR/bupindex) 51strip strips the path to every filename given 52strip-path= path-prefix to be stripped when saving 53graft= a graft point *old_path*=*new_path* (can be used more than once) 54#,compress= set compression level to # (0-9, 9 is highest) [1] 55""" 56o = options.Options(optspec) 57opt, flags, extra = o.parse(compat.argv[1:]) 58 59if opt.indexfile: 60 opt.indexfile = argv_bytes(opt.indexfile) 61if opt.name: 62 opt.name = argv_bytes(opt.name) 63if opt.remote: 64 opt.remote = argv_bytes(opt.remote) 65if opt.strip_path: 66 opt.strip_path = argv_bytes(opt.strip_path) 67 68git.check_repo_or_die() 69if not (opt.tree or opt.commit or opt.name): 70 o.fatal("use one or more of -t, -c, -n") 71if not extra: 72 o.fatal("no filenames given") 73 74extra = [argv_bytes(x) for x in extra] 75 76opt.progress = (istty2 and not opt.quiet) 77opt.smaller = parse_num(opt.smaller or 0) 78if opt.bwlimit: 79 client.bwlimit = parse_num(opt.bwlimit) 80 81if opt.date: 82 date = parse_date_or_fatal(opt.date, o.fatal) 83else: 84 date = time.time() 85 86if opt.strip and opt.strip_path: 87 o.fatal("--strip is incompatible with --strip-path") 88 89graft_points = [] 90if opt.graft: 91 if opt.strip: 92 o.fatal("--strip is incompatible with --graft") 93 94 if opt.strip_path: 95 o.fatal("--strip-path is incompatible with --graft") 96 97 for (option, parameter) in flags: 98 if option == "--graft": 99 parameter = argv_bytes(parameter) 100 splitted_parameter = parameter.split(b'=') 101 if len(splitted_parameter) != 2: 102 o.fatal("a graft point must be of the form old_path=new_path") 103 old_path, new_path = splitted_parameter 104 if not (old_path and new_path): 105 o.fatal("a graft point cannot be empty") 106 graft_points.append((resolve_parent(old_path), 107 resolve_parent(new_path))) 108 109is_reverse = environ.get(b'BUP_SERVER_REVERSE') 110if is_reverse and opt.remote: 111 o.fatal("don't use -r in reverse mode; it's automatic") 112 113name = opt.name 114if name and not valid_save_name(name): 115 o.fatal("'%s' is not a valid branch name" % path_msg(name)) 116refname = name and b'refs/heads/%s' % name or None 117if opt.remote or is_reverse: 118 try: 119 cli = client.Client(opt.remote) 120 except client.ClientError as e: 121 log('error: %s' % e) 122 sys.exit(1) 123 oldref = refname and cli.read_ref(refname) or None 124 w = cli.new_packwriter(compression_level=opt.compress) 125else: 126 cli = None 127 oldref = refname and git.read_ref(refname) or None 128 w = git.PackWriter(compression_level=opt.compress) 129 130handle_ctrl_c() 131 132 133# Metadata is stored in a file named .bupm in each directory. The 134# first metadata entry will be the metadata for the current directory. 135# The remaining entries will be for each of the other directory 136# elements, in the order they're listed in the index. 137# 138# Since the git tree elements are sorted according to 139# git.shalist_item_sort_key, the metalist items are accumulated as 140# (sort_key, metadata) tuples, and then sorted when the .bupm file is 141# created. The sort_key should have been computed using the element's 142# mangled name and git mode (after hashsplitting), but the code isn't 143# actually doing that but rather uses the element's real name and mode. 144# This makes things a bit more difficult when reading it back, see 145# vfs.ordered_tree_entries(). 146 147# Maintain a stack of information representing the current location in 148# the archive being constructed. The current path is recorded in 149# parts, which will be something like ['', 'home', 'someuser'], and 150# the accumulated content and metadata for of the dirs in parts is 151# stored in parallel stacks in shalists and metalists. 152 153parts = [] # Current archive position (stack of dir names). 154shalists = [] # Hashes for each dir in paths. 155metalists = [] # Metadata for each dir in paths. 156 157 158def _push(part, metadata): 159 # Enter a new archive directory -- make it the current directory. 160 parts.append(part) 161 shalists.append([]) 162 metalists.append([(b'', metadata)]) # This dir's metadata (no name). 163 164 165def _pop(force_tree, dir_metadata=None): 166 # Leave the current archive directory and add its tree to its parent. 167 assert(len(parts) >= 1) 168 part = parts.pop() 169 shalist = shalists.pop() 170 metalist = metalists.pop() 171 # FIXME: only test if collision is possible (i.e. given --strip, etc.)? 172 if force_tree: 173 tree = force_tree 174 else: 175 names_seen = set() 176 clean_list = [] 177 metaidx = 1 # entry at 0 is for the dir 178 for x in shalist: 179 name = x[1] 180 if name in names_seen: 181 parent_path = b'/'.join(parts) + b'/' 182 add_error('error: ignoring duplicate path %s in %s' 183 % (path_msg(name), path_msg(parent_path))) 184 if not stat.S_ISDIR(x[0]): 185 del metalist[metaidx] 186 else: 187 names_seen.add(name) 188 clean_list.append(x) 189 if not stat.S_ISDIR(x[0]): 190 metaidx += 1 191 192 if dir_metadata: # Override the original metadata pushed for this dir. 193 metalist = [(b'', dir_metadata)] + metalist[1:] 194 sorted_metalist = sorted(metalist, key = lambda x : x[0]) 195 metadata = b''.join([m[1].encode() for m in sorted_metalist]) 196 metadata_f = BytesIO(metadata) 197 mode, id = hashsplit.split_to_blob_or_tree(w.new_blob, w.new_tree, 198 [metadata_f], 199 keep_boundaries=False) 200 clean_list.append((mode, b'.bupm', id)) 201 202 tree = w.new_tree(clean_list) 203 if shalists: 204 shalists[-1].append((GIT_MODE_TREE, 205 git.mangle_name(part, 206 GIT_MODE_TREE, GIT_MODE_TREE), 207 tree)) 208 return tree 209 210 211lastremain = None 212def progress_report(n): 213 global count, subcount, lastremain 214 subcount += n 215 cc = count + subcount 216 pct = total and (cc*100.0/total) or 0 217 now = time.time() 218 elapsed = now - tstart 219 kps = elapsed and int(cc/1024./elapsed) 220 kps_frac = 10 ** int(math.log(kps+1, 10) - 1) 221 kps = int(kps/kps_frac)*kps_frac 222 if cc: 223 remain = elapsed*1.0/cc * (total-cc) 224 else: 225 remain = 0.0 226 if (lastremain and (remain > lastremain) 227 and ((remain - lastremain)/lastremain < 0.05)): 228 remain = lastremain 229 else: 230 lastremain = remain 231 hours = int(remain/60/60) 232 mins = int(remain/60 - hours*60) 233 secs = int(remain - hours*60*60 - mins*60) 234 if elapsed < 30: 235 remainstr = '' 236 kpsstr = '' 237 else: 238 kpsstr = '%dk/s' % kps 239 if hours: 240 remainstr = '%dh%dm' % (hours, mins) 241 elif mins: 242 remainstr = '%dm%d' % (mins, secs) 243 else: 244 remainstr = '%ds' % secs 245 qprogress('Saving: %.2f%% (%d/%dk, %d/%d files) %s %s\r' 246 % (pct, cc/1024, total/1024, fcount, ftotal, 247 remainstr, kpsstr)) 248 249 250indexfile = opt.indexfile or git.repo(b'bupindex') 251r = index.Reader(indexfile) 252try: 253 msr = index.MetaStoreReader(indexfile + b'.meta') 254except IOError as ex: 255 if ex.errno != EACCES: 256 raise 257 log('error: cannot access %r; have you run bup index?' 258 % path_msg(indexfile)) 259 sys.exit(1) 260hlink_db = hlinkdb.HLinkDB(indexfile + b'.hlink') 261 262def already_saved(ent): 263 return ent.is_valid() and w.exists(ent.sha) and ent.sha 264 265def wantrecurse_pre(ent): 266 return not already_saved(ent) 267 268def wantrecurse_during(ent): 269 return not already_saved(ent) or ent.sha_missing() 270 271def find_hardlink_target(hlink_db, ent): 272 if hlink_db and not stat.S_ISDIR(ent.mode) and ent.nlink > 1: 273 link_paths = hlink_db.node_paths(ent.dev, ent.ino) 274 if link_paths: 275 return link_paths[0] 276 277total = ftotal = 0 278if opt.progress: 279 for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_pre): 280 if not (ftotal % 10024): 281 qprogress('Reading index: %d\r' % ftotal) 282 exists = ent.exists() 283 hashvalid = already_saved(ent) 284 ent.set_sha_missing(not hashvalid) 285 if not opt.smaller or ent.size < opt.smaller: 286 if exists and not hashvalid: 287 total += ent.size 288 ftotal += 1 289 progress('Reading index: %d, done.\n' % ftotal) 290 hashsplit.progress_callback = progress_report 291 292# Root collisions occur when strip or graft options map more than one 293# path to the same directory (paths which originally had separate 294# parents). When that situation is detected, use empty metadata for 295# the parent. Otherwise, use the metadata for the common parent. 296# Collision example: "bup save ... --strip /foo /foo/bar /bar". 297 298# FIXME: Add collision tests, or handle collisions some other way. 299 300# FIXME: Detect/handle strip/graft name collisions (other than root), 301# i.e. if '/foo/bar' and '/bar' both map to '/'. 302 303first_root = None 304root_collision = None 305tstart = time.time() 306count = subcount = fcount = 0 307lastskip_name = None 308lastdir = b'' 309for (transname,ent) in r.filter(extra, wantrecurse=wantrecurse_during): 310 (dir, file) = os.path.split(ent.name) 311 exists = (ent.flags & index.IX_EXISTS) 312 hashvalid = already_saved(ent) 313 wasmissing = ent.sha_missing() 314 oldsize = ent.size 315 if opt.verbose: 316 if not exists: 317 status = 'D' 318 elif not hashvalid: 319 if ent.sha == index.EMPTY_SHA: 320 status = 'A' 321 else: 322 status = 'M' 323 else: 324 status = ' ' 325 if opt.verbose >= 2: 326 log('%s %-70s\n' % (status, path_msg(ent.name))) 327 elif not stat.S_ISDIR(ent.mode) and lastdir != dir: 328 if not lastdir.startswith(dir): 329 log('%s %-70s\n' % (status, path_msg(os.path.join(dir, b'')))) 330 lastdir = dir 331 332 if opt.progress: 333 progress_report(0) 334 fcount += 1 335 336 if not exists: 337 continue 338 if opt.smaller and ent.size >= opt.smaller: 339 if exists and not hashvalid: 340 if opt.verbose: 341 log('skipping large file "%s"\n' % path_msg(ent.name)) 342 lastskip_name = ent.name 343 continue 344 345 assert(dir.startswith(b'/')) 346 if opt.strip: 347 dirp = stripped_path_components(dir, extra) 348 elif opt.strip_path: 349 dirp = stripped_path_components(dir, [opt.strip_path]) 350 elif graft_points: 351 dirp = grafted_path_components(graft_points, dir) 352 else: 353 dirp = path_components(dir) 354 355 # At this point, dirp contains a representation of the archive 356 # path that looks like [(archive_dir_name, real_fs_path), ...]. 357 # So given "bup save ... --strip /foo/bar /foo/bar/baz", dirp 358 # might look like this at some point: 359 # [('', '/foo/bar'), ('baz', '/foo/bar/baz'), ...]. 360 361 # This dual representation supports stripping/grafting, where the 362 # archive path may not have a direct correspondence with the 363 # filesystem. The root directory is represented by an initial 364 # component named '', and any component that doesn't have a 365 # corresponding filesystem directory (due to grafting, for 366 # example) will have a real_fs_path of None, i.e. [('', None), 367 # ...]. 368 369 if first_root == None: 370 first_root = dirp[0] 371 elif first_root != dirp[0]: 372 root_collision = True 373 374 # If switching to a new sub-tree, finish the current sub-tree. 375 while parts > [x[0] for x in dirp]: 376 _pop(force_tree = None) 377 378 # If switching to a new sub-tree, start a new sub-tree. 379 for path_component in dirp[len(parts):]: 380 dir_name, fs_path = path_component 381 # Not indexed, so just grab the FS metadata or use empty metadata. 382 try: 383 meta = metadata.from_path(fs_path, normalized=True) \ 384 if fs_path else metadata.Metadata() 385 except (OSError, IOError) as e: 386 add_error(e) 387 lastskip_name = dir_name 388 meta = metadata.Metadata() 389 _push(dir_name, meta) 390 391 if not file: 392 if len(parts) == 1: 393 continue # We're at the top level -- keep the current root dir 394 # Since there's no filename, this is a subdir -- finish it. 395 oldtree = already_saved(ent) # may be None 396 newtree = _pop(force_tree = oldtree) 397 if not oldtree: 398 if lastskip_name and lastskip_name.startswith(ent.name): 399 ent.invalidate() 400 else: 401 ent.validate(GIT_MODE_TREE, newtree) 402 ent.repack() 403 if exists and wasmissing: 404 count += oldsize 405 continue 406 407 # it's not a directory 408 if hashvalid: 409 id = ent.sha 410 git_name = git.mangle_name(file, ent.mode, ent.gitmode) 411 git_info = (ent.gitmode, git_name, id) 412 shalists[-1].append(git_info) 413 sort_key = git.shalist_item_sort_key((ent.mode, file, id)) 414 meta = msr.metadata_at(ent.meta_ofs) 415 meta.hardlink_target = find_hardlink_target(hlink_db, ent) 416 # Restore the times that were cleared to 0 in the metastore. 417 (meta.atime, meta.mtime, meta.ctime) = (ent.atime, ent.mtime, ent.ctime) 418 metalists[-1].append((sort_key, meta)) 419 else: 420 id = None 421 if stat.S_ISREG(ent.mode): 422 try: 423 with hashsplit.open_noatime(ent.name) as f: 424 (mode, id) = hashsplit.split_to_blob_or_tree( 425 w.new_blob, w.new_tree, [f], 426 keep_boundaries=False) 427 except (IOError, OSError) as e: 428 add_error('%s: %s' % (ent.name, e)) 429 lastskip_name = ent.name 430 elif stat.S_ISDIR(ent.mode): 431 assert(0) # handled above 432 elif stat.S_ISLNK(ent.mode): 433 try: 434 rl = os.readlink(ent.name) 435 except (OSError, IOError) as e: 436 add_error(e) 437 lastskip_name = ent.name 438 else: 439 (mode, id) = (GIT_MODE_SYMLINK, w.new_blob(rl)) 440 else: 441 # Everything else should be fully described by its 442 # metadata, so just record an empty blob, so the paths 443 # in the tree and .bupm will match up. 444 (mode, id) = (GIT_MODE_FILE, w.new_blob(b'')) 445 446 if id: 447 ent.validate(mode, id) 448 ent.repack() 449 git_name = git.mangle_name(file, ent.mode, ent.gitmode) 450 git_info = (mode, git_name, id) 451 shalists[-1].append(git_info) 452 sort_key = git.shalist_item_sort_key((ent.mode, file, id)) 453 hlink = find_hardlink_target(hlink_db, ent) 454 try: 455 meta = metadata.from_path(ent.name, hardlink_target=hlink, 456 normalized=True) 457 except (OSError, IOError) as e: 458 add_error(e) 459 lastskip_name = ent.name 460 meta = metadata.Metadata() 461 metalists[-1].append((sort_key, meta)) 462 463 if exists and wasmissing: 464 count += oldsize 465 subcount = 0 466 467 468if opt.progress: 469 pct = total and count*100.0/total or 100 470 progress('Saving: %.2f%% (%d/%dk, %d/%d files), done. \n' 471 % (pct, count/1024, total/1024, fcount, ftotal)) 472 473while len(parts) > 1: # _pop() all the parts above the root 474 _pop(force_tree = None) 475assert(len(shalists) == 1) 476assert(len(metalists) == 1) 477 478# Finish the root directory. 479tree = _pop(force_tree = None, 480 # When there's a collision, use empty metadata for the root. 481 dir_metadata = metadata.Metadata() if root_collision else None) 482 483sys.stdout.flush() 484out = byte_stream(sys.stdout) 485 486if opt.tree: 487 out.write(hexlify(tree)) 488 out.write(b'\n') 489if opt.commit or name: 490 if compat.py_maj > 2: 491 # Strip b prefix from python 3 bytes reprs to preserve previous format 492 msgcmd = b'[%s]' % b', '.join([repr(argv_bytes(x))[1:].encode('ascii') 493 for x in compat.argv]) 494 else: 495 msgcmd = repr(compat.argv) 496 msg = b'bup save\n\nGenerated by command:\n%s\n' % msgcmd 497 userline = (b'%s <%s@%s>' % (userfullname(), username(), hostname())) 498 commit = w.new_commit(tree, oldref, userline, date, None, 499 userline, date, None, msg) 500 if opt.commit: 501 out.write(hexlify(commit)) 502 out.write(b'\n') 503 504msr.close() 505w.close() # must close before we can update the ref 506 507if opt.name: 508 if cli: 509 cli.update_ref(refname, commit, oldref) 510 else: 511 git.update_ref(refname, commit, oldref) 512 513if cli: 514 cli.close() 515 516if saved_errors: 517 log('WARNING: %d errors encountered while saving.\n' % len(saved_errors)) 518 sys.exit(1) 519