1 """Virtual File System interface to bup repository content.
3 This module provides a path-based interface to the content of a bup
6 The VFS is structured like this:
9 /SAVE-NAME/SAVE-DATE/...
12 Each path is represented by an item that has least an item.meta which
13 may be either a Metadata object, or an integer mode. Functions like
14 item_mode() and item_size() will return the mode and size in either
15 case. Any item.meta Metadata instances must not be modified directly.
16 Make a copy to modify via item.meta.copy() if needed, or call
19 The want_meta argument is advisory for calls that accept it, and it
20 may not be honored. Callers must be able to handle an item.meta value
21 that is either an instance of Metadata or an integer mode, perhaps
22 via item_mode() or augment_item_meta().
24 Setting want_meta=False is rarely desirable since it can limit the VFS
25 to only the metadata that git itself can represent, and so for
26 example, fifos and sockets will appear to be regular files
27 (e.g. S_ISREG(item_mode(item)) will be true). But the option is still
28 provided because it may be more efficient when just the path names or
29 the more limited metadata is sufficient.
31 Any given metadata object's size may be None, in which case the size
32 can be computed via item_size() or augment_item_meta(...,
35 When traversing a directory using functions like contents(), the meta
36 value for any directories other than '.' will be a default directory
37 mode, not a Metadata object. This is because the actual metadata for
38 a directory is stored inside the directory (see
39 fill_in_metadata_if_dir() or ensure_item_has_metadata()).
41 Commit items represent commits (e.g. /.tag/some-commit or
42 /foo/latest), and for most purposes, they appear as the underlying
43 tree. S_ISDIR(item_mode(item)) will return true for both tree Items
44 and Commits and the commit's oid is the tree hash; the commit hash is
49 from __future__ import absolute_import, print_function
50 from collections import namedtuple
51 from errno import EINVAL, ELOOP, ENOENT, ENOTDIR
52 from itertools import chain, dropwhile, groupby, tee
53 from random import randrange
54 from stat import S_IFDIR, S_IFLNK, S_IFREG, S_ISDIR, S_ISLNK, S_ISREG
55 from time import localtime, strftime
56 import exceptions, re, sys
58 from bup import git, metadata, vint
59 from bup.compat import range
60 from bup.git import BUP_CHUNKED, cp, get_commit_items, parse_commit, tree_decode
61 from bup.helpers import debug2, last
62 from bup.metadata import Metadata
63 from bup.vint import read_bvec, write_bvec
64 from bup.vint import read_vint, write_vint
65 from bup.vint import read_vuint, write_vuint
67 # We currently assume that it's always appropriate to just forward IOErrors
70 class IOError(exceptions.IOError):
71 def __init__(self, errno, message, terminus=None):
72 exceptions.IOError.__init__(self, errno, message)
73 self.terminus = terminus
75 def write_ioerror(port, ex):
76 assert isinstance(ex, IOError)
78 (1 if ex.errno is not None else 0)
79 | (2 if ex.message is not None else 0)
80 | (4 if ex.terminus is not None else 0))
81 if ex.errno is not None:
82 write_vint(port, ex.errno)
83 if ex.message is not None:
84 write_bvec(port, ex.message.encode('utf-8'))
85 if ex.terminus is not None:
86 write_resolution(port, ex.terminus)
88 def read_ioerror(port):
89 mask = read_vuint(port)
90 no = read_vint(port) if 1 & mask else None
91 msg = read_bvec(port).decode('utf-8') if 2 & mask else None
92 term = read_resolution(port) if 4 & mask else None
93 return IOError(errno=no, message=msg, terminus=term)
96 default_file_mode = S_IFREG | 0o644
97 default_dir_mode = S_IFDIR | 0o755
98 default_symlink_mode = S_IFLNK | 0o755
100 def _default_mode_for_gitmode(gitmode):
102 return default_file_mode
104 return default_dir_mode
106 return default_symlink_mode
107 raise Exception('unexpected git mode ' + oct(gitmode))
109 def _normal_or_chunked_file_size(repo, oid):
110 """Return the size of the normal or chunked file indicated by oid."""
111 # FIXME: --batch-format CatPipe?
112 it = repo.cat(oid.encode('hex'))
113 _, obj_t, size = next(it)
115 while obj_t == 'tree':
116 mode, name, last_oid = last(tree_decode(''.join(it)))
118 it = repo.cat(last_oid.encode('hex'))
119 _, obj_t, size = next(it)
120 return ofs + sum(len(b) for b in it)
122 def _skip_chunks_before_offset(tree, offset):
123 prev_ent = next(tree, None)
128 ent_ofs = int(ent[1], 16)
130 return chain([prev_ent, ent], tree)
131 if ent_ofs == offset:
132 return chain([ent], tree)
136 def _tree_chunks(repo, tree, startofs):
137 "Tree should be a sequence of (name, mode, hash) as per tree_decode()."
138 assert(startofs >= 0)
139 # name is the chunk's hex offset in the original file
140 for mode, name, oid in _skip_chunks_before_offset(tree, startofs):
142 skipmore = startofs - ofs
145 it = repo.cat(oid.encode('hex'))
146 _, obj_t, size = next(it)
149 assert obj_t == 'tree'
150 for b in _tree_chunks(repo, tree_decode(data), skipmore):
153 assert obj_t == 'blob'
154 yield data[skipmore:]
157 def __init__(self, repo, oid, startofs):
158 it = repo.cat(oid.encode('hex'))
159 _, obj_t, size = next(it)
160 isdir = obj_t == 'tree'
163 self.it = _tree_chunks(repo, tree_decode(data), startofs)
167 self.blob = data[startofs:]
170 def next(self, size):
172 while len(out) < size:
173 if self.it and not self.blob:
175 self.blob = self.it.next()
176 except StopIteration:
179 want = size - len(out)
180 out += self.blob[:want]
181 self.blob = self.blob[want:]
184 debug2('next(%d) returned %d\n' % (size, len(out)))
188 class _FileReader(object):
189 def __init__(self, repo, oid, known_size=None):
190 assert len(oid) == 20
195 self._size = known_size
197 def _compute_size(self):
199 self._size = _normal_or_chunked_file_size(self._repo, self.oid)
203 if ofs < 0 or ofs > self._compute_size():
204 raise IOError(EINVAL, 'Invalid seek offset: %d' % ofs)
210 def read(self, count=-1):
211 size = self._compute_size()
215 count = size - self.ofs
216 if not self.reader or self.reader.ofs != self.ofs:
217 self.reader = _ChunkReader(self._repo, self.oid, self.ofs)
219 buf = self.reader.next(count)
222 raise # our offsets will be all screwed up otherwise
231 def __exit__(self, type, value, traceback):
235 _multiple_slashes_rx = re.compile(r'//+')
237 def _decompose_path(path):
238 """Return a boolean indicating whether the path is absolute, and a
239 reversed list of path elements, omitting any occurrences of "."
240 and ignoring any leading or trailing slash. If the path is
241 effectively '/' or '.', return an empty list.
244 path = re.sub(_multiple_slashes_rx, '/', path)
246 return True, True, []
247 is_absolute = must_be_dir = False
248 if path.startswith('/'):
251 for suffix in ('/', '/.'):
252 if path.endswith(suffix):
254 path = path[:-len(suffix)]
255 parts = [x for x in path.split('/') if x != '.']
258 must_be_dir = True # e.g. path was effectively '.' or '/', etc.
259 return is_absolute, must_be_dir, parts
262 Item = namedtuple('Item', ('meta', 'oid'))
263 Chunky = namedtuple('Chunky', ('meta', 'oid'))
264 FakeLink = namedtuple('FakeLink', ('meta', 'target'))
265 Root = namedtuple('Root', ('meta'))
266 Tags = namedtuple('Tags', ('meta'))
267 RevList = namedtuple('RevList', ('meta', 'oid'))
268 Commit = namedtuple('Commit', ('meta', 'oid', 'coid'))
270 item_types = frozenset((Item, Chunky, Root, Tags, RevList, Commit))
271 real_tree_types = frozenset((Item, Commit))
273 def write_item(port, item):
275 name = bytes(kind.__name__)
277 has_meta = 1 if isinstance(meta, Metadata) else 0
278 if kind in (Item, Chunky, RevList):
279 assert len(item.oid) == 20
281 vint.send(port, 'sVs', name, has_meta, item.oid)
282 Metadata.write(meta, port, include_path=False)
284 vint.send(port, 'sVsV', name, has_meta, item.oid, item.meta)
285 elif kind in (Root, Tags):
287 vint.send(port, 'sV', name, has_meta)
288 Metadata.write(meta, port, include_path=False)
290 vint.send(port, 'sVV', name, has_meta, item.meta)
292 assert len(item.oid) == 20
293 assert len(item.coid) == 20
295 vint.send(port, 'sVss', name, has_meta, item.oid, item.coid)
296 Metadata.write(meta, port, include_path=False)
298 vint.send(port, 'sVssV', name, has_meta, item.oid, item.coid,
300 elif kind == FakeLink:
302 vint.send(port, 'sVs', name, has_meta, item.target)
303 Metadata.write(meta, port, include_path=False)
305 vint.send(port, 'sVsV', name, has_meta, item.target, item.meta)
310 def read_m(port, has_meta):
312 m = Metadata.read(port)
314 return read_vuint(port)
315 kind, has_meta = vint.recv(port, 'sV')
317 oid, meta = read_bvec(port), read_m(port, has_meta)
318 return Item(oid=oid, meta=meta)
319 if kind == b'Chunky':
320 oid, meta = read_bvec(port), read_m(port, has_meta)
321 return Chunky(oid=oid, meta=meta)
322 if kind == b'RevList':
323 oid, meta = read_bvec(port), read_m(port, has_meta)
324 return RevList(oid=oid, meta=meta)
326 return Root(meta=read_m(port, has_meta))
328 return Tags(meta=read_m(port, has_meta))
329 if kind == b'Commit':
330 oid, coid = vint.recv(port, 'ss')
331 meta = read_m(port, has_meta)
332 return Commit(oid=oid, coid=coid, meta=meta)
333 if kind == b'FakeLink':
334 target, meta = read_bvec(port), read_m(port, has_meta)
335 return FakeLink(target=target, meta=meta)
338 def write_resolution(port, resolution):
339 write_vuint(port, len(resolution))
340 for name, item in resolution:
341 write_bvec(port, name)
344 write_item(port, item)
348 def read_resolution(port):
352 name = read_bvec(port)
353 have_item = ord(port.read(1))
354 assert have_item in (0, 1)
355 item = read_item(port) if have_item else None
356 result.append((name, item))
360 _root = Root(meta=default_dir_mode)
361 _tags = Tags(meta=default_dir_mode)
366 ### A general purpose shared cache with (currently) cheap random
367 ### eviction. At the moment there is no weighting so a single commit
368 ### item is just as likely to be evicted as an entire "rev-list". See
369 ### is_valid_cache_key for a description of the expected content.
373 _cache_max_items = 30000
376 global _cache, _cache_keys
380 def is_valid_cache_key(x):
381 """Return logically true if x looks like it could be a valid cache key
382 (with respect to structure). Current valid cache entries:
383 res:... -> resolution
385 rvl:OID -> {'.', commit, '2012...', next_commit, ...}
387 # Suspect we may eventually add "(container_oid, name) -> ...", and others.
391 if tag in ('itm:', 'rvl:') and len(x) == 24:
398 if not is_valid_cache_key(key):
399 raise Exception('invalid cache key: ' + repr(key))
400 return _cache.get(key)
402 def cache_notice(key, value):
403 global _cache, _cache_keys, _cache_max_items
404 if not is_valid_cache_key(key):
405 raise Exception('invalid cache key: ' + repr(key))
408 if len(_cache) < _cache_max_items:
409 _cache_keys.append(key)
412 victim_i = randrange(0, len(_cache_keys))
413 victim = _cache_keys[victim_i]
415 _cache_keys[victim_i] = key
418 def cache_get_commit_item(oid, need_meta=True):
419 """Return the requested tree item if it can be found in the cache.
420 When need_meta is true don't return a cached item that only has a
422 # tree might be stored independently, or as '.' with its entries.
423 commit_key = b'itm:' + oid
424 item = cache_get(commit_key)
428 if isinstance(item.meta, Metadata):
430 entries = cache_get(b'rvl:' + oid)
434 def cache_get_revlist_item(oid, need_meta=True):
435 commit = cache_get_commit_item(oid, need_meta=need_meta)
437 return RevList(oid=oid, meta=commit.meta)
440 """Return a completely independent copy of item, such that
441 modifications will not affect the original.
444 meta = getattr(item, 'meta', None)
445 if isinstance(meta, Metadata):
446 return(item._replace(meta=meta.copy()))
450 """Return the integer mode (stat st_mode) for item."""
452 if isinstance(m, Metadata):
456 def _read_dir_meta(bupm):
457 # This is because save writes unmodified Metadata() entries for
458 # fake parents -- test-save-strip-graft.sh demonstrates.
459 m = Metadata.read(bupm)
461 return default_dir_mode
462 assert m.mode is not None
467 def tree_data_and_bupm(repo, oid):
468 """Return (tree_bytes, bupm_oid) where bupm_oid will be None if the
469 tree has no metadata (i.e. older bup save, or non-bup tree).
472 assert len(oid) == 20
473 it = repo.cat(oid.encode('hex'))
474 _, item_t, size = next(it)
476 if item_t == 'commit':
477 commit = parse_commit(data)
478 it = repo.cat(commit.tree)
479 _, item_t, size = next(it)
481 assert item_t == 'tree'
482 elif item_t != 'tree':
483 raise Exception('%r is not a tree or commit' % oid.encode('hex'))
484 for _, mangled_name, sub_oid in tree_decode(data):
485 if mangled_name == '.bupm':
487 if mangled_name > '.bupm':
491 def _find_treeish_oid_metadata(repo, oid):
492 """Return the metadata for the tree or commit oid, or None if the tree
493 has no metadata (i.e. older bup save, or non-bup tree).
496 tree_data, bupm_oid = tree_data_and_bupm(repo, oid)
498 with _FileReader(repo, bupm_oid) as meta_stream:
499 return _read_dir_meta(meta_stream)
502 def _readlink(repo, oid):
503 return ''.join(repo.join(oid.encode('hex')))
505 def readlink(repo, item):
506 """Return the link target of item, which must be a symlink. Reads the
507 target from the repository if necessary."""
509 assert S_ISLNK(item_mode(item))
510 if isinstance(item, FakeLink):
512 if isinstance(item.meta, Metadata):
513 target = item.meta.symlink_target
516 return _readlink(repo, item.oid)
518 def _compute_item_size(repo, item):
519 mode = item_mode(item)
521 size = _normal_or_chunked_file_size(repo, item.oid)
524 if isinstance(item, FakeLink):
525 return len(item.target)
526 return len(_readlink(repo, item.oid))
529 def item_size(repo, item):
530 """Return the size of item, computing it if necessary."""
532 if isinstance(m, Metadata) and m.size is not None:
534 return _compute_item_size(repo, item)
536 def tree_data_reader(repo, oid):
537 """Return an open reader for all of the data contained within oid. If
538 oid refers to a tree, recursively concatenate all of its contents."""
539 return _FileReader(repo, oid)
541 def fopen(repo, item):
542 """Return an open reader for the given file item."""
543 assert S_ISREG(item_mode(item))
544 return tree_data_reader(repo, item.oid)
546 def _commit_item_from_data(oid, data):
547 info = parse_commit(data)
548 return Commit(meta=default_dir_mode,
549 oid=info.tree.decode('hex'),
552 def _commit_item_from_oid(repo, oid, require_meta):
553 commit = cache_get_commit_item(oid, need_meta=require_meta)
554 if commit and ((not require_meta) or isinstance(commit.meta, Metadata)):
556 it = repo.cat(oid.encode('hex'))
557 _, typ, size = next(it)
558 assert typ == 'commit'
559 commit = _commit_item_from_data(oid, ''.join(it))
561 meta = _find_treeish_oid_metadata(repo, commit.oid)
563 commit = commit._replace(meta=meta)
564 commit_key = b'itm:' + oid
565 cache_notice(commit_key, commit)
568 def _revlist_item_from_oid(repo, oid, require_meta):
569 commit = _commit_item_from_oid(repo, oid, require_meta)
570 return RevList(oid=oid, meta=commit.meta)
572 def root_items(repo, names=None, want_meta=True):
573 """Yield (name, item) for the items in '/' in the VFS. Return
574 everything if names is logically false, otherwise return only
575 items with a name in the collection.
578 # FIXME: what about non-leaf refs like 'refs/heads/foo/bar/baz?
584 # FIXME: maybe eventually support repo.clone() or something
585 # and pass in two repos, so we can drop the tuple() and stream
586 # in parallel (i.e. meta vs refs).
587 for name, oid in tuple(repo.refs([], limit_to_heads=True)):
588 assert(name.startswith('refs/heads/'))
589 yield name[11:], _revlist_item_from_oid(repo, oid, want_meta)
597 if ref in ('.', '.tag'):
599 it = repo.cat('refs/heads/' + ref)
600 oidx, typ, size = next(it)
604 assert typ == 'commit'
605 commit = parse_commit(''.join(it))
606 yield ref, _revlist_item_from_oid(repo, oidx.decode('hex'), want_meta)
608 def ordered_tree_entries(tree_data, bupm=None):
609 """Yields (name, mangled_name, kind, gitmode, oid) for each item in
610 tree, sorted by name.
613 # Sadly, the .bupm entries currently aren't in git tree order,
614 # but in unmangled name order. They _do_ account for the fact
615 # that git sorts trees (including chunked trees) as if their
616 # names ended with "/" (so "fo" sorts after "fo." iff fo is a
617 # directory), but we apply this on the unmangled names in save
618 # rather than on the mangled names.
619 # This makes streaming impossible when we need the metadata.
620 def result_from_tree_entry(tree_entry):
621 gitmode, mangled_name, oid = tree_entry
622 name, kind = git.demangle_name(mangled_name, gitmode)
623 return name, mangled_name, kind, gitmode, oid
625 tree_ents = (result_from_tree_entry(x) for x in tree_decode(tree_data))
627 tree_ents = sorted(tree_ents, key=lambda x: x[0])
628 for ent in tree_ents:
631 def tree_items(oid, tree_data, names=frozenset(), bupm=None):
633 def tree_item(ent_oid, kind, gitmode):
634 if kind == BUP_CHUNKED:
635 meta = Metadata.read(bupm) if bupm else default_file_mode
636 return Chunky(oid=ent_oid, meta=meta)
639 # No metadata here (accessable via '.' inside ent_oid).
640 return Item(meta=default_dir_mode, oid=ent_oid)
642 meta = Metadata.read(bupm) if bupm else None
643 # handle the case of metadata being empty/missing in bupm
644 # (or there not being bupm at all)
646 meta = _default_mode_for_gitmode(gitmode)
647 return Item(oid=ent_oid, meta=meta)
649 assert len(oid) == 20
651 dot_meta = _read_dir_meta(bupm) if bupm else default_dir_mode
652 yield '.', Item(oid=oid, meta=dot_meta)
653 tree_entries = ordered_tree_entries(tree_data, bupm)
654 for name, mangled_name, kind, gitmode, ent_oid in tree_entries:
655 if mangled_name == '.bupm':
658 yield name, tree_item(ent_oid, kind, gitmode)
661 # Assumes the tree is properly formed, i.e. there are no
662 # duplicates, and entries will be in git tree order.
663 if type(names) not in (frozenset, set):
664 names = frozenset(names)
665 remaining = len(names)
667 # Account for the bupm sort order issue (cf. ordered_tree_entries above)
668 last_name = max(names) if bupm else max(names) + '/'
671 dot_meta = _read_dir_meta(bupm) if bupm else default_dir_mode
672 yield '.', Item(oid=oid, meta=dot_meta)
677 tree_entries = ordered_tree_entries(tree_data, bupm)
678 for name, mangled_name, kind, gitmode, ent_oid in tree_entries:
679 if mangled_name == '.bupm':
682 if name not in names:
684 break # given bupm sort order, we're finished
685 if (kind == BUP_CHUNKED or not S_ISDIR(gitmode)) and bupm:
688 yield name, tree_item(ent_oid, kind, gitmode)
693 def tree_items_with_meta(repo, oid, tree_data, names):
694 # For now, the .bupm order doesn't quite match git's, and we don't
695 # load the tree data incrementally anyway, so we just work in RAM
697 assert len(oid) == 20
699 for _, mangled_name, sub_oid in tree_decode(tree_data):
700 if mangled_name == '.bupm':
701 bupm = _FileReader(repo, sub_oid)
703 if mangled_name > '.bupm':
705 for item in tree_items(oid, tree_data, names, bupm):
708 _save_name_rx = re.compile(r'^\d\d\d\d-\d\d-\d\d-\d{6}(-\d+)?$')
710 def _reverse_suffix_duplicates(strs):
711 """Yields the elements of strs, with any runs of duplicate values
712 suffixed with -N suffixes, where the zero padded integer N
713 decreases to 0 by 1 (e.g. 10, 09, ..., 00).
716 for name, duplicates in groupby(strs):
717 ndup = len(tuple(duplicates))
721 ndig = len(str(ndup - 1))
722 fmt = '%s-' + '%0' + str(ndig) + 'd'
723 for i in range(ndup - 1, -1, -1):
724 yield fmt % (name, i)
727 items = f.readline().split(None)
728 assert len(items) == 2
729 tree, auth_sec = items
730 return tree.decode('hex'), int(auth_sec)
732 def _name_for_rev(rev):
733 commit_oidx, (tree_oid, utc) = rev
734 return strftime('%Y-%m-%d-%H%M%S', localtime(utc))
736 def _item_for_rev(rev):
737 commit_oidx, (tree_oid, utc) = rev
738 coid = commit_oidx.decode('hex')
739 item = cache_get_commit_item(coid, need_meta=False)
742 item = Commit(meta=default_dir_mode, oid=tree_oid, coid=coid)
743 commit_key = b'itm:' + coid
744 cache_notice(commit_key, item)
747 def cache_commit(repo, oid):
748 """Build, cache, and return a "name -> commit_item" dict of the entire
752 # For now, always cache with full metadata
754 entries['.'] = _revlist_item_from_oid(repo, oid, True)
755 revs = repo.rev_list((oid.encode('hex'),), format='%T %at',
757 rev_items, rev_names = tee(revs)
758 revs = None # Don't disturb the tees
759 rev_names = _reverse_suffix_duplicates(_name_for_rev(x) for x in rev_names)
760 rev_items = (_item_for_rev(x) for x in rev_items)
762 for item in rev_items:
763 name = next(rev_names)
764 tip = tip or (name, item)
766 entries['latest'] = FakeLink(meta=default_symlink_mode, target=tip[0])
767 revlist_key = b'rvl:' + tip[1].coid
768 cache_notice(revlist_key, entries)
771 def revlist_items(repo, oid, names):
772 assert len(oid) == 20
774 # Special case '.' instead of caching the whole history since it's
775 # the only way to get the metadata for the commit.
776 if names and all(x == '.' for x in names):
777 yield '.', _revlist_item_from_oid(repo, oid, True)
780 # For now, don't worry about the possibility of the contents being
781 # "too big" for the cache.
782 revlist_key = b'rvl:' + oid
783 entries = cache_get(revlist_key)
785 entries = cache_commit(repo, oid)
788 for name in sorted(entries.keys()):
789 yield name, entries[name]
792 names = frozenset(name for name in names
793 if _save_name_rx.match(name) or name in ('.', 'latest'))
796 yield '.', entries['.']
797 for name in (n for n in names if n != '.'):
798 commit = entries.get(name)
802 def tags_items(repo, names):
806 assert len(oid) == 20
807 oidx = oid.encode('hex')
809 _, typ, size = next(it)
811 return cache_get_commit_item(oid, need_meta=False) \
812 or _commit_item_from_data(oid, ''.join(it))
815 return Item(meta=default_file_mode, oid=oid)
817 return Item(meta=default_dir_mode, oid=oid)
818 raise Exception('unexpected tag type ' + typ + ' for tag ' + name)
822 # We have to pull these all into ram because tag_item calls cat()
823 for name, oid in tuple(repo.refs(names, limit_to_tags=True)):
824 assert(name.startswith('refs/tags/'))
826 yield name, tag_item(oid)
829 # Assumes no duplicate refs
830 if type(names) not in (frozenset, set):
831 names = frozenset(names)
832 remaining = len(names)
833 last_name = max(names)
840 for name, oid in repo.refs(names, limit_to_tags=True):
841 assert(name.startswith('refs/tags/'))
845 if name not in names:
847 yield name, tag_item(oid)
852 def contents(repo, item, names=None, want_meta=True):
853 """Yields information about the items contained in item. Yields
854 (name, item) for each name in names, if the name exists, in an
855 unspecified order. If there are no names, then yields (name,
856 item) for all items, including, a first item named '.'
857 representing the container itself.
859 The meta value for any directories other than '.' will be a
860 default directory mode, not a Metadata object. This is because
861 the actual metadata for a directory is stored inside the directory
862 (see fill_in_metadata_if_dir() or ensure_item_has_metadata()).
864 Note that want_meta is advisory. For any given item, item.meta
865 might be a Metadata instance or a mode, and if the former,
866 meta.size might be None. Missing sizes can be computed via via
867 item_size() or augment_item_meta(..., include_size=True).
869 Do not modify any item.meta Metadata instances directly. If
870 needed, make a copy via item.meta.copy() and modify that instead.
873 # Q: are we comfortable promising '.' first when no names?
876 assert S_ISDIR(item_mode(item))
878 if item_t in real_tree_types:
879 it = repo.cat(item.oid.encode('hex'))
880 _, obj_t, size = next(it)
884 # Note: it shouldn't be possible to see an Item with type
885 # 'commit' since a 'commit' should always produce a Commit.
886 raise Exception('unexpected git ' + obj_t)
888 item_gen = tree_items_with_meta(repo, item.oid, data, names)
890 item_gen = tree_items(item.oid, data, names)
891 elif item_t == RevList:
892 item_gen = revlist_items(repo, item.oid, names)
894 item_gen = root_items(repo, names, want_meta)
896 item_gen = tags_items(repo, names)
898 raise Exception('unexpected VFS item ' + str(item))
902 def _resolve_path(repo, path, parent=None, want_meta=True, follow=True):
903 cache_key = b'res:%d%d%d:%s\0%s' \
904 % (bool(want_meta), bool(follow), repo.id(),
905 ('/'.join(x[0] for x in parent) if parent else ''),
907 resolution = cache_get(cache_key)
911 def notice_resolution(r):
912 cache_notice(cache_key, r)
915 def raise_dir_required_but_not_dir(path, parent, past):
916 raise IOError(ENOTDIR,
917 "path %r%s resolves to non-directory %r"
919 ' (relative to %r)' % parent if parent else '',
928 assert type(x[0]) in (bytes, str)
929 assert type(x[1]) in item_types
930 assert parent[0][1] == _root
931 if not S_ISDIR(item_mode(parent[-1][1])):
932 raise IOError(ENOTDIR,
933 'path resolution parent %r is not a directory'
935 is_absolute, must_be_dir, future = _decompose_path(path)
938 if not future: # path was effectively '.' or '/'
940 return notice_resolution((('', _root),))
942 return notice_resolution(tuple(parent))
943 return notice_resolution((('', _root),))
947 past = list(parent) if parent else [('', _root)]
951 if must_be_dir and not S_ISDIR(item_mode(past[-1][1])):
952 raise_dir_required_but_not_dir(path, parent, past)
953 return notice_resolution(tuple(past))
954 segment = future.pop()
957 if len(past) > 1: # .. from / is /
958 assert S_ISDIR(item_mode(past[-1][1]))
961 parent_name, parent_item = past[-1]
962 wanted = (segment,) if not want_meta else ('.', segment)
963 items = tuple(contents(repo, parent_item, names=wanted,
964 want_meta=want_meta))
966 item = items[0][1] if items else None
967 else: # First item will be '.' and have the metadata
968 item = items[1][1] if len(items) == 2 else None
969 dot, dot_item = items[0]
971 past[-1] = parent_name, parent_item
973 past.append((segment, None),)
974 return notice_resolution(tuple(past))
975 mode = item_mode(item)
976 if not S_ISLNK(mode):
977 if not S_ISDIR(mode):
978 past.append((segment, item),)
980 raise IOError(ENOTDIR,
981 'path %r%s ends internally in non-directory here: %r'
983 ' (relative to %r)' % parent if parent else '',
987 raise_dir_required_but_not_dir(path, parent, past)
988 return notice_resolution(tuple(past))
990 if want_meta and type(item) in real_tree_types:
991 dir_meta = _find_treeish_oid_metadata(repo, item.oid)
993 item = item._replace(meta=dir_meta)
994 past.append((segment, item))
996 if not future and not follow:
997 past.append((segment, item),)
1000 raise IOError(ELOOP,
1001 'too many symlinks encountered while resolving %r%s'
1002 % (path, ' relative to %r' % parent if parent else ''),
1003 terminus=tuple(past + [(segment, item)]))
1004 target = readlink(repo, item)
1005 is_absolute, _, target_future = _decompose_path(target)
1007 if not target_future: # path was effectively '/'
1008 return notice_resolution((('', _root),))
1009 past = [('', _root)]
1010 future = target_future
1012 future.extend(target_future)
1015 def resolve(repo, path, parent=None, want_meta=True, follow=True):
1016 """Follow the path in the virtual filesystem and return a tuple
1017 representing the location, if any, denoted by the path. Each
1018 element in the result tuple will be (name, info), where info will
1019 be a VFS item that can be passed to functions like item_mode().
1021 If follow is false, and if the final path element is a symbolic
1022 link, don't follow it, just return it in the result.
1024 If a path segment that does not exist is encountered during
1025 resolution, the result will represent the location of the missing
1026 item, and that item in the result will be None.
1028 Any attempt to traverse a non-directory will raise a VFS ENOTDIR
1031 Any symlinks along the path, including at the end, will be
1032 resolved. A VFS IOError with the errno attribute set to ELOOP
1033 will be raised if too many symlinks are traversed while following
1034 the path. That exception is effectively like a normal
1035 ELOOP IOError exception, but will include a terminus element
1036 describing the location of the failure, which will be a tuple of
1037 (name, info) elements.
1039 The parent, if specified, must be a sequence of (name, item)
1040 tuples, and will provide the starting point for the resolution of
1041 the path. If no parent is specified, resolution will start at
1044 The result may include elements of parent directly, so they must
1045 not be modified later. If this is a concern, pass in "name,
1046 copy_item(item) for name, item in parent" instead.
1048 When want_meta is true, detailed metadata will be included in each
1049 result item if it's avaiable, otherwise item.meta will be an
1050 integer mode. The metadata size may or may not be provided, but
1051 can be computed by item_size() or augment_item_meta(...,
1052 include_size=True). Setting want_meta=False is rarely desirable
1053 since it can limit the VFS to just the metadata git itself can
1054 represent, and so, as an example, fifos and sockets will appear to
1055 be regular files (e.g. S_ISREG(item_mode(item)) will be true) .
1056 But the option is provided because it may be more efficient when
1057 only the path names or the more limited metadata is sufficient.
1059 Do not modify any item.meta Metadata instances directly. If
1060 needed, make a copy via item.meta.copy() and modify that instead.
1063 if repo.is_remote():
1064 # Redirect to the more efficient remote version
1065 return repo.resolve(path, parent=parent, want_meta=want_meta,
1067 result = _resolve_path(repo, path, parent=parent, want_meta=want_meta,
1069 _, leaf_item = result[-1]
1070 if leaf_item and follow:
1071 assert not S_ISLNK(item_mode(leaf_item))
1074 def try_resolve(repo, path, parent=None, want_meta=True):
1075 """If path does not refer to a symlink, does not exist, or refers to a
1076 valid symlink, behave exactly like resolve(..., follow=True). If
1077 path refers to an invalid symlink, behave like resolve(...,
1081 res = resolve(repo, path, parent=parent, want_meta=want_meta, follow=False)
1082 leaf_name, leaf_item = res[-1]
1085 if not S_ISLNK(item_mode(leaf_item)):
1087 follow = resolve(repo, leaf_name, parent=res[:-1], want_meta=want_meta)
1088 follow_name, follow_item = follow[-1]
1093 def augment_item_meta(repo, item, include_size=False):
1094 """Ensure item has a Metadata instance for item.meta. If item.meta is
1095 currently a mode, replace it with a compatible "fake" Metadata
1096 instance. If include_size is true, ensure item.meta.size is
1097 correct, computing it if needed. If item.meta is a Metadata
1098 instance, this call may modify it in place or replace it.
1101 # If we actually had parallelism, we'd need locking...
1104 if isinstance(m, Metadata):
1105 if include_size and m.size is None:
1106 m.size = _compute_item_size(repo, item)
1107 return item._replace(meta=m)
1112 meta.uid = meta.gid = meta.atime = meta.mtime = meta.ctime = 0
1114 if isinstance(item, FakeLink):
1115 target = item.target
1117 target = _readlink(repo, item.oid)
1118 meta.symlink_target = target
1119 meta.size = len(target)
1121 meta.size = _compute_item_size(repo, item)
1122 return item._replace(meta=meta)
1124 def fill_in_metadata_if_dir(repo, item):
1125 """If item is a directory and item.meta is not a Metadata instance,
1126 attempt to find the metadata for the directory. If found, return
1127 a new item augmented to include that metadata. Otherwise, return
1128 item. May be useful for the output of contents().
1131 if S_ISDIR(item_mode(item)) and not isinstance(item.meta, Metadata):
1132 items = tuple(contents(repo, item, ('.',), want_meta=True))
1133 assert len(items) == 1
1134 assert items[0][0] == '.'
1138 def ensure_item_has_metadata(repo, item, include_size=False):
1139 """If item is a directory, attempt to find and add its metadata. If
1140 the item still doesn't have a Metadata instance for item.meta,
1141 give it one via augment_item_meta(). May be useful for the output
1145 return augment_item_meta(repo,
1146 fill_in_metadata_if_dir(repo, item),
1147 include_size=include_size)