manifest: drop the CamelCase name for `manifest.manifestlog`

See 61557734c0ae for the reasoning.

mercurial/manifest.py

@@ -1,2780 +1,2774
 # manifest.py - manifest revision class for mercurial
 #
 # Copyright 2005-2007 Olivia Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import annotations
 
 import heapq
 import itertools
 import struct
 import typing
 import weakref
 
 from typing import (
     ByteString,
     Callable,
     Collection,
     Dict,
     Iterable,
     Iterator,
     List,
     Optional,
     Set,
     Tuple,
     Union,
     cast,
 )
 
 from .i18n import _
 from .node import (
     bin,
     hex,
     nullrev,
 )
 from . import (
     encoding,
     error,
     match as matchmod,
     mdiff,
     pathutil,
     policy,
     pycompat,
     revlog,
     util,
 )
 from .interfaces import (
     repository,
     util as interfaceutil,
 )
 from .revlogutils import (
     constants as revlog_constants,
 )
 
 parsers = policy.importmod('parsers')
 propertycache = util.propertycache
 
 # Allow tests to more easily test the alternate path in manifestdict.fastdelta()
 FASTDELTA_TEXTDIFF_THRESHOLD = 1000
 
 
 def _parse(nodelen, data: bytes):
     # This method does a little bit of excessive-looking
     # precondition checking. This is so that the behavior of this
     # class exactly matches its C counterpart to try and help
     # prevent surprise breakage for anyone that develops against
     # the pure version.
     if data and data[-1:] != b'\n':
         raise ValueError(b'Manifest did not end in a newline.')
     prev = None
     for l in data.splitlines():
         if prev is not None and prev > l:
             raise ValueError(b'Manifest lines not in sorted order.')
         prev = l
         f, n = l.split(b'\0')
         nl = len(n)
         flags = n[-1:]
         if flags in _manifestflags:
             n = n[:-1]
             nl -= 1
         else:
             flags = b''
         if nl != 2 * nodelen:
             raise ValueError(b'Invalid manifest line')
 
         yield f, bin(n), flags
 
 
 def _text(it):
     files = []
     lines = []
     for f, n, fl in it:
         files.append(f)
         # if this is changed to support newlines in filenames,
         # be sure to check the templates/ dir again (especially *-raw.tmpl)
         lines.append(b"%s\0%s%s\n" % (f, hex(n), fl))
 
     _checkforbidden(files)
     return b''.join(lines)
 
 
 class lazymanifestiter:
     def __init__(self, lm: '_LazyManifest') -> None:
         self.pos = 0
         self.lm = lm
 
     def __iter__(self) -> 'lazymanifestiter':
         return self
 
     def next(self) -> bytes:
         try:
             data, pos = self.lm._get(self.pos)
         except IndexError:
             raise StopIteration
         if pos == -1:
             assert isinstance(data, tuple)
             self.pos += 1
             return data[0]
         assert isinstance(data, bytes)
         self.pos += 1
         zeropos = data.find(b'\x00', pos)
         return data[pos:zeropos]
 
     __next__ = next
 
 
 class lazymanifestiterentries:
     def __init__(self, lm: '_LazyManifest') -> None:
         self.lm = lm
         self.pos = 0
 
     def __iter__(self) -> 'lazymanifestiterentries':
         return self
 
     def next(self) -> Tuple[bytes, bytes, bytes]:
         try:
             data, pos = self.lm._get(self.pos)
         except IndexError:
             raise StopIteration
         if pos == -1:
             assert isinstance(data, tuple)
             self.pos += 1
             return data
         assert isinstance(data, bytes)
         zeropos = data.find(b'\x00', pos)
         nlpos = data.find(b'\n', pos)
         if zeropos == -1 or nlpos == -1 or nlpos < zeropos:
             raise error.StorageError(b'Invalid manifest line')
         flags = data[nlpos - 1 : nlpos]
         if flags in _manifestflags:
             hlen = nlpos - zeropos - 2
         else:
             hlen = nlpos - zeropos - 1
             flags = b''
         if hlen != 2 * self.lm._nodelen:
             raise error.StorageError(b'Invalid manifest line')
         hashval = unhexlify(
             data, self.lm.extrainfo[self.pos], zeropos + 1, hlen
         )
         self.pos += 1
         return (data[pos:zeropos], hashval, flags)
 
     __next__ = next
 
 
 def unhexlify(data: bytes, extra: int, pos, length: int):
     s = bin(data[pos : pos + length])
     if extra:
         s += bytes([extra & 0xFF])
     return s
 
 
 def _cmp(a, b):
     return (a > b) - (a < b)
 
 
 _manifestflags = {b'', b'l', b't', b'x'}
 
 
 class _LazyManifest:
     """A pure python manifest backed by a byte string.  It is supplimented with
     internal lists as it is modified, until it is compacted back to a pure byte
     string.
 
     ``data`` is the initial manifest data.
 
     ``positions`` is a list of offsets, one per manifest entry.  Positive
     values are offsets into ``data``, negative values are offsets into the
     ``extradata`` list.  When an entry is removed, its entry is dropped from
     ``positions``.  The values are encoded such that when walking the list and
     indexing into ``data`` or ``extradata`` as appropriate, the entries are
     sorted by filename.
 
     ``extradata`` is a list of (key, hash, flags) for entries that were added or
     modified since the manifest was created or compacted.
     """
 
     def __init__(
         self,
         nodelen: int,
         data: bytes,
         positions=None,
         extrainfo=None,
         extradata=None,
         hasremovals: bool = False,
     ):
         self._nodelen = nodelen
         if positions is None:
             self.positions = self.findlines(data)
             self.extrainfo = [0] * len(self.positions)
             self.data = data
             self.extradata = []
             self.hasremovals = False
         else:
             self.positions = positions[:]
             self.extrainfo = extrainfo[:]
             self.extradata = extradata[:]
             self.data = data
             self.hasremovals = hasremovals
 
     def findlines(self, data: bytes) -> List[int]:
         if not data:
             return []
         pos = data.find(b"\n")
         if pos == -1 or data[-1:] != b'\n':
             raise ValueError(b"Manifest did not end in a newline.")
         positions = [0]
         prev = data[: data.find(b'\x00')]
         while pos < len(data) - 1 and pos != -1:
             positions.append(pos + 1)
             nexts = data[pos + 1 : data.find(b'\x00', pos + 1)]
             if nexts < prev:
                 raise ValueError(b"Manifest lines not in sorted order.")
             prev = nexts
             pos = data.find(b"\n", pos + 1)
         return positions
 
     def _get(
         self, index: int
     ) -> Tuple[Union[bytes, Tuple[bytes, bytes, bytes]], int]:
         # get the position encoded in pos:
         #   positive number is an index in 'data'
         #   negative number is in extrapieces
         pos = self.positions[index]
         if pos >= 0:
             return self.data, pos
         return self.extradata[-pos - 1], -1
 
     def _getkey(self, pos) -> bytes:
         if pos >= 0:
             return self.data[pos : self.data.find(b'\x00', pos + 1)]
         return self.extradata[-pos - 1][0]
 
     def bsearch(self, key: bytes) -> int:
         first = 0
         last = len(self.positions) - 1
 
         while first <= last:
             midpoint = (first + last) // 2
             nextpos = self.positions[midpoint]
             candidate = self._getkey(nextpos)
             r = _cmp(key, candidate)
             if r == 0:
                 return midpoint
             else:
                 if r < 0:
                     last = midpoint - 1
                 else:
                     first = midpoint + 1
         return -1
 
     def bsearch2(self, key: bytes) -> Tuple[int, bool]:
         # same as the above, but will always return the position
         # done for performance reasons
         first = 0
         last = len(self.positions) - 1
 
         while first <= last:
             midpoint = (first + last) // 2
             nextpos = self.positions[midpoint]
             candidate = self._getkey(nextpos)
             r = _cmp(key, candidate)
             if r == 0:
                 return (midpoint, True)
             else:
                 if r < 0:
                     last = midpoint - 1
                 else:
                     first = midpoint + 1
         return (first, False)
 
     def __contains__(self, key: bytes) -> bool:
         return self.bsearch(key) != -1
 
     def __getitem__(self, key: bytes) -> Tuple[bytes, bytes]:
         if not isinstance(key, bytes):
             raise TypeError(b"getitem: manifest keys must be a bytes.")
         needle = self.bsearch(key)
         if needle == -1:
             raise KeyError
         data, pos = self._get(needle)
         if pos == -1:
             assert isinstance(data, tuple)
             return (data[1], data[2])
 
         assert isinstance(data, bytes)
         zeropos = data.find(b'\x00', pos)
         nlpos = data.find(b'\n', zeropos)
         assert 0 <= needle <= len(self.positions)
         assert len(self.extrainfo) == len(self.positions)
         if zeropos == -1 or nlpos == -1 or nlpos < zeropos:
             raise error.StorageError(b'Invalid manifest line')
         hlen = nlpos - zeropos - 1
         flags = data[nlpos - 1 : nlpos]
         if flags in _manifestflags:
             hlen -= 1
         else:
             flags = b''
         if hlen != 2 * self._nodelen:
             raise error.StorageError(b'Invalid manifest line')
         hashval = unhexlify(data, self.extrainfo[needle], zeropos + 1, hlen)
         return (hashval, flags)
 
     def __delitem__(self, key: bytes) -> None:
         needle, found = self.bsearch2(key)
         if not found:
             raise KeyError
         cur = self.positions[needle]
         self.positions = self.positions[:needle] + self.positions[needle + 1 :]
         self.extrainfo = self.extrainfo[:needle] + self.extrainfo[needle + 1 :]
         if cur >= 0:
             # This does NOT unsort the list as far as the search functions are
             # concerned, as they only examine lines mapped by self.positions.
             self.data = self.data[:cur] + b'\x00' + self.data[cur + 1 :]
             self.hasremovals = True
 
     def __setitem__(self, key: bytes, value: Tuple[bytes, bytes]):
         if not isinstance(key, bytes):
             raise TypeError(b"setitem: manifest keys must be a byte string.")
         if not isinstance(value, tuple) or len(value) != 2:
             raise TypeError(
                 b"Manifest values must be a tuple of (node, flags)."
             )
         hashval = value[0]
         if not isinstance(hashval, bytes) or len(hashval) not in (20, 32):
             raise TypeError(b"node must be a 20-byte or 32-byte byte string")
         flags = value[1]
         if not isinstance(flags, bytes) or len(flags) > 1:
             raise TypeError(b"flags must a 0 or 1 byte string, got %r", flags)
         needle, found = self.bsearch2(key)
         if found:
             # put the item
             pos = self.positions[needle]
             if pos < 0:
                 self.extradata[-pos - 1] = (key, hashval, value[1])
             else:
                 # just don't bother
                 self.extradata.append((key, hashval, value[1]))
                 self.positions[needle] = -len(self.extradata)
         else:
             # not found, put it in with extra positions
             self.extradata.append((key, hashval, value[1]))
             self.positions = (
                 self.positions[:needle]
                 + [-len(self.extradata)]
                 + self.positions[needle:]
             )
             self.extrainfo = (
                 self.extrainfo[:needle] + [0] + self.extrainfo[needle:]
             )
 
     def copy(self) -> '_LazyManifest':
         # XXX call _compact like in C?
         return _lazymanifest(
             self._nodelen,
             self.data,
             self.positions,
             self.extrainfo,
             self.extradata,
             self.hasremovals,
         )
 
     def _compact(self) -> None:
         # hopefully not called TOO often
         if len(self.extradata) == 0 and not self.hasremovals:
             return
         l = []
         i = 0
         offset = 0
         self.extrainfo = [0] * len(self.positions)
         while i < len(self.positions):
             if self.positions[i] >= 0:
                 cur = self.positions[i]
                 last_cut = cur
 
                 # Collect all contiguous entries in the buffer at the current
                 # offset, breaking out only for added/modified items held in
                 # extradata, or a deleted line prior to the next position.
                 while True:
                     self.positions[i] = offset
                     i += 1
                     if i == len(self.positions) or self.positions[i] < 0:
                         break
 
                     # A removed file has no positions[] entry, but does have an
                     # overwritten first byte.  Break out and find the end of the
                     # current good entry/entries if there is a removed file
                     # before the next position.
                     if (
                         self.hasremovals
                         and self.data.find(b'\n\x00', cur, self.positions[i])
                         != -1
                     ):
                         break
 
                     offset += self.positions[i] - cur
                     cur = self.positions[i]
                 end_cut = self.data.find(b'\n', cur)
                 if end_cut != -1:
                     end_cut += 1
                 offset += end_cut - cur
                 l.append(self.data[last_cut:end_cut])
             else:
                 while i < len(self.positions) and self.positions[i] < 0:
                     cur = self.positions[i]
                     t = self.extradata[-cur - 1]
                     l.append(self._pack(t))
                     self.positions[i] = offset
                     # Hashes are either 20 bytes (old sha1s) or 32
                     # bytes (new non-sha1).
                     hlen = 20
                     if len(t[1]) > 25:
                         hlen = 32
                     if len(t[1]) > hlen:
                         self.extrainfo[i] = ord(t[1][hlen + 1])
                     offset += len(l[-1])
                     i += 1
         self.data = b''.join(l)
         self.hasremovals = False
         self.extradata = []
 
     def _pack(self, d: Tuple[bytes, bytes, bytes]) -> bytes:
         n = d[1]
         assert len(n) in (20, 32)
         return d[0] + b'\x00' + hex(n) + d[2] + b'\n'
 
     def text(self) -> ByteString:
         self._compact()
         return self.data
 
     def diff(
         self, m2: '_LazyManifest', clean: bool = False
     ) -> Dict[
         bytes,
         Optional[
             Tuple[Tuple[Optional[bytes], bytes], Tuple[Optional[bytes], bytes]]
         ],
     ]:
         '''Finds changes between the current manifest and m2.'''
         # XXX think whether efficiency matters here
         diff = {}
 
         for fn, e1, flags in self.iterentries():
             if fn not in m2:
                 diff[fn] = (e1, flags), (None, b'')
             else:
                 e2 = m2[fn]
                 if (e1, flags) != e2:
                     diff[fn] = (e1, flags), e2
                 elif clean:
                     diff[fn] = None
 
         for fn, e2, flags in m2.iterentries():
             if fn not in self:
                 diff[fn] = (None, b''), (e2, flags)
 
         return diff
 
     def iterentries(self) -> lazymanifestiterentries:
         return lazymanifestiterentries(self)
 
     def iterkeys(self) -> lazymanifestiter:
         return lazymanifestiter(self)
 
     def __iter__(self) -> lazymanifestiter:
         return lazymanifestiter(self)
 
     def __len__(self) -> int:
         return len(self.positions)
 
     def filtercopy(self, filterfn: Callable[[bytes], bool]) -> '_LazyManifest':
         # XXX should be optimized
         c = _lazymanifest(self._nodelen, b'')
         for f, n, fl in self.iterentries():
             if filterfn(f):
                 c[f] = n, fl
         return c
 
 
 try:
     _lazymanifest = parsers.lazymanifest
 except AttributeError:
     _lazymanifest = _LazyManifest
 
 
 class manifestdict:  # (repository.imanifestdict)
     def __init__(self, nodelen: int, data: ByteString = b''):
         self._nodelen = nodelen
         self._lm = _lazymanifest(nodelen, data)
 
     def __getitem__(self, key: bytes) -> bytes:
         return self._lm[key][0]
 
     def find(self, key: bytes) -> Tuple[bytes, bytes]:
         return self._lm[key]
 
     def __len__(self) -> int:
         return len(self._lm)
 
     def __nonzero__(self) -> bool:
         # nonzero is covered by the __len__ function, but implementing it here
         # makes it easier for extensions to override.
         return len(self._lm) != 0
 
     __bool__ = __nonzero__
 
     def set(self, key: bytes, node: bytes, flags: bytes) -> None:
         self._lm[key] = node, flags
 
     def __setitem__(self, key: bytes, node: bytes) -> None:
         self._lm[key] = node, self.flags(key)
 
     def __contains__(self, key: bytes) -> bool:
         if key is None:
             return False
         return key in self._lm
 
     def __delitem__(self, key: bytes) -> None:
         del self._lm[key]
 
     def __iter__(self) -> Iterator[bytes]:
         return self._lm.__iter__()
 
     def iterkeys(self) -> Iterator[bytes]:
         return self._lm.iterkeys()
 
     def keys(self) -> List[bytes]:
         return list(self.iterkeys())
 
     def filesnotin(self, m2, match=None) -> Set[bytes]:
         '''Set of files in this manifest that are not in the other'''
         if match is not None:
             match = matchmod.badmatch(match, lambda path, msg: None)
             sm2 = set(m2.walk(match))
             return {f for f in self.walk(match) if f not in sm2}
         return {f for f in self if f not in m2}
 
     @propertycache
     def _dirs(self) -> pathutil.dirs:
         return pathutil.dirs(self)
 
     def dirs(self) -> pathutil.dirs:
         return self._dirs
 
     def hasdir(self, dir: bytes) -> bool:
         return dir in self._dirs
 
     def _filesfastpath(self, match: matchmod.basematcher) -> bool:
         """Checks whether we can correctly and quickly iterate over matcher
         files instead of over manifest files."""
         files = match.files()
         return len(files) < 100 and (
             match.isexact()
             or (match.prefix() and all(fn in self for fn in files))
         )
 
     def walk(self, match: matchmod.basematcher) -> Iterator[bytes]:
         """Generates matching file names.
 
         Equivalent to manifest.matches(match).iterkeys(), but without creating
         an entirely new manifest.
 
         It also reports nonexistent files by marking them bad with match.bad().
         """
         if match.always():
             for f in iter(self):
                 yield f
             return
 
         fset = set(match.files())
 
         # avoid the entire walk if we're only looking for specific files
         if self._filesfastpath(match):
             for fn in sorted(fset):
                 if fn in self:
                     yield fn
             return
 
         for fn in self:
             if fn in fset:
                 # specified pattern is the exact name
                 fset.remove(fn)
             if match(fn):
                 yield fn
 
         # for dirstate.walk, files=[''] means "walk the whole tree".
         # follow that here, too
         fset.discard(b'')
 
         for fn in sorted(fset):
             if not self.hasdir(fn):
                 match.bad(fn, None)
 
     def _matches(self, match: matchmod.basematcher) -> 'manifestdict':
         '''generate a new manifest filtered by the match argument'''
         if match.always():
             return self.copy()
 
         if self._filesfastpath(match):
             m = manifestdict(self._nodelen)
             lm = self._lm
             for fn in match.files():
                 if fn in lm:
                     m._lm[fn] = lm[fn]
             return m
 
         m = manifestdict(self._nodelen)
         m._lm = self._lm.filtercopy(match)
         return m
 
     def diff(
         self,
         m2: 'manifestdict',
         match: Optional[matchmod.basematcher] = None,
         clean: bool = False,
     ) -> Dict[
         bytes,
         Optional[
             Tuple[Tuple[Optional[bytes], bytes], Tuple[Optional[bytes], bytes]]
         ],
     ]:
         """Finds changes between the current manifest and m2.
 
         Args:
           m2: the manifest to which this manifest should be compared.
           clean: if true, include files unchanged between these manifests
                  with a None value in the returned dictionary.
 
         The result is returned as a dict with filename as key and
         values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the
         nodeid in the current/other manifest and fl1/fl2 is the flag
         in the current/other manifest. Where the file does not exist,
         the nodeid will be None and the flags will be the empty
         string.
         """
         if match:
             m1 = self._matches(match)
             m2 = m2._matches(match)
             return m1.diff(m2, clean=clean)
         return self._lm.diff(m2._lm, clean)
 
     def setflag(self, key: bytes, flag: bytes) -> None:
         if flag not in _manifestflags:
             raise TypeError(b"Invalid manifest flag set.")
         self._lm[key] = self[key], flag
 
     def get(self, key: bytes, default=None) -> Optional[bytes]:
         try:
             return self._lm[key][0]
         except KeyError:
             return default
 
     def flags(self, key: bytes) -> bytes:
         try:
             return self._lm[key][1]
         except KeyError:
             return b''
 
     def copy(self) -> 'manifestdict':
         c = manifestdict(self._nodelen)
         c._lm = self._lm.copy()
         return c
 
     def items(self) -> Iterator[Tuple[bytes, bytes]]:
         return (x[:2] for x in self._lm.iterentries())
 
     def iteritems(self) -> Iterator[Tuple[bytes, bytes]]:
         return (x[:2] for x in self._lm.iterentries())
 
     def iterentries(self) -> Iterator[Tuple[bytes, bytes, bytes]]:
         return self._lm.iterentries()
 
     def text(self) -> ByteString:
         # most likely uses native version
         return self._lm.text()
 
     def fastdelta(
         self, base: ByteString, changes: Iterable[Tuple[bytes, bool]]
     ) -> Tuple[ByteString, ByteString]:
         """Given a base manifest text as a bytearray and a list of changes
         relative to that text, compute a delta that can be used by revlog.
         """
         delta = []
         dstart = None
         dend = None
         dline = [b""]
         start = 0
         # zero copy representation of base as a buffer
         addbuf = util.buffer(base)
 
         changes = list(changes)
         if len(changes) < FASTDELTA_TEXTDIFF_THRESHOLD:
             # start with a readonly loop that finds the offset of
             # each line and creates the deltas
             for f, todelete in changes:
                 # bs will either be the index of the item or the insert point
                 start, end = _msearch(addbuf, f, start)
                 if not todelete:
                     h, fl = self._lm[f]
                     l = b"%s\0%s%s\n" % (f, hex(h), fl)
                 else:
                     if start == end:
                         # item we want to delete was not found, error out
                         raise AssertionError(
                             _(b"failed to remove %s from manifest") % f
                         )
                     l = b""
                 if dstart is not None and dstart <= start and dend >= start:
                     if dend < end:
                         dend = end
                     if l:
                         dline.append(l)
                 else:
                     if dstart is not None:
                         delta.append((dstart, dend, b"".join(dline)))
                     dstart = start
                     dend = end
                     dline = [l]
 
             if dstart is not None:
                 delta.append((dstart, dend, b"".join(dline)))
             # apply the delta to the base, and get a delta for addrevision
             deltatext, arraytext = _addlistdelta(base, delta)
         else:
             # For large changes, it's much cheaper to just build the text and
             # diff it.
             arraytext = bytearray(self.text())
             deltatext = mdiff.textdiff(
                 util.buffer(base), util.buffer(arraytext)
             )
 
         return arraytext, deltatext
 
 
 def _msearch(
     m: ByteString, s: bytes, lo: int = 0, hi: Optional[int] = None
 ) -> Tuple[int, int]:
     """return a tuple (start, end) that says where to find s within m.
 
     If the string is found m[start:end] are the line containing
     that string.  If start == end the string was not found and
     they indicate the proper sorted insertion point.
     """
 
     def advance(i: int, c: bytes):
         while i < lenm and m[i : i + 1] != c:
             i += 1
         return i
 
     if not s:
         return (lo, lo)
     lenm = len(m)
     if not hi:
         hi = lenm
     while lo < hi:
         mid = (lo + hi) // 2
         start = mid
         while start > 0 and m[start - 1 : start] != b'\n':
             start -= 1
         end = advance(start, b'\0')
         if bytes(m[start:end]) < s:
             # we know that after the null there are 40 bytes of sha1
             # this translates to the bisect lo = mid + 1
             lo = advance(end + 40, b'\n') + 1
         else:
             # this translates to the bisect hi = mid
             hi = start
     end = advance(lo, b'\0')
     found = m[lo:end]
     if s == found:
         # we know that after the null there are 40 bytes of sha1
         end = advance(end + 40, b'\n')
         return (lo, end + 1)
     else:
         return (lo, lo)
 
 
 def _checkforbidden(l: Iterable[bytes]) -> None:
     """Check filenames for illegal characters."""
     for f in l:
         if b'\n' in f or b'\r' in f:
             raise error.StorageError(
                 _(b"'\\n' and '\\r' disallowed in filenames: %r")
                 % pycompat.bytestr(f)
             )
 
 
 # apply the changes collected during the bisect loop to our addlist
 # return a delta suitable for addrevision
 def _addlistdelta(
     addlist: ByteString,
     x: Iterable[Tuple[int, int, bytes]],
 ) -> Tuple[bytes, ByteString]:
     # for large addlist arrays, building a new array is cheaper
     # than repeatedly modifying the existing one
     currentposition = 0
     newaddlist = bytearray()
 
     for start, end, content in x:
         newaddlist += addlist[currentposition:start]
         if content:
             newaddlist += bytearray(content)
 
         currentposition = end
 
     newaddlist += addlist[currentposition:]
 
     deltatext = b"".join(
         struct.pack(b">lll", start, end, len(content)) + content
         for start, end, content in x
     )
     return deltatext, newaddlist
 
 
 def _splittopdir(f: bytes) -> Tuple[bytes, bytes]:
     if b'/' in f:
         dir, subpath = f.split(b'/', 1)
         return dir + b'/', subpath
     else:
         return b'', f
 
 
 _noop = lambda s: None
 
 
 class treemanifest:  # (repository.imanifestdict)
     _dir: bytes
     _dirs: Dict[bytes, 'treemanifest']
     _dirty: bool
     _files: Dict[bytes, bytes]
     _flags: Dict[bytes, bytes]
 
     def __init__(self, nodeconstants, dir: bytes = b'', text: bytes = b''):
         self._dir = dir
         self.nodeconstants = nodeconstants
         self._node = self.nodeconstants.nullid
         self._nodelen = self.nodeconstants.nodelen
         self._loadfunc = _noop
         self._copyfunc = _noop
         self._dirty = False
         self._dirs = {}
         self._lazydirs: Dict[
             bytes,
             Tuple[bytes, Callable[[bytes, bytes], 'treemanifest'], bool],
         ] = {}
         # Using _lazymanifest here is a little slower than plain old dicts
         self._files = {}
         self._flags = {}
         if text:
 
             def readsubtree(subdir, subm):
                 raise AssertionError(
                     b'treemanifest constructor only accepts flat manifests'
                 )
 
             self.parse(text, readsubtree)
             self._dirty = True  # Mark flat manifest dirty after parsing
 
     def _subpath(self, path: bytes) -> bytes:
         return self._dir + path
 
     def _loadalllazy(self) -> None:
         selfdirs = self._dirs
         subpath = self._subpath
         for d, (node, readsubtree, docopy) in self._lazydirs.items():
             if docopy:
                 selfdirs[d] = readsubtree(subpath(d), node).copy()
             else:
                 selfdirs[d] = readsubtree(subpath(d), node)
         self._lazydirs.clear()
 
     def _loadlazy(self, d: bytes) -> None:
         v = self._lazydirs.get(d)
         if v is not None:
             node, readsubtree, docopy = v
             if docopy:
                 self._dirs[d] = readsubtree(self._subpath(d), node).copy()
             else:
                 self._dirs[d] = readsubtree(self._subpath(d), node)
             del self._lazydirs[d]
 
     def _loadchildrensetlazy(
         self, visit: Union[Set[bytes], bytes]
     ) -> Optional[Set[bytes]]:
         if not visit:
             return None
         if visit == b'all' or visit == b'this':
             self._loadalllazy()
             return None
 
         visit = cast(Set[bytes], visit)
 
         loadlazy = self._loadlazy
         for k in visit:
             loadlazy(k + b'/')
         return visit
 
     def _loaddifflazy(self, t1: 'treemanifest', t2: 'treemanifest'):
         """load items in t1 and t2 if they're needed for diffing.
 
         The criteria currently is:
         - if it's not present in _lazydirs in either t1 or t2, load it in the
           other (it may already be loaded or it may not exist, doesn't matter)
         - if it's present in _lazydirs in both, compare the nodeid; if it
           differs, load it in both
         """
         toloadlazy = []
         for d, v1 in t1._lazydirs.items():
             v2 = t2._lazydirs.get(d)
             if v2 is None or v2[0] != v1[0]:
                 toloadlazy.append(d)
         for d, v1 in t2._lazydirs.items():
             if d not in t1._lazydirs:
                 toloadlazy.append(d)
 
         for d in toloadlazy:
             t1._loadlazy(d)
             t2._loadlazy(d)
 
     def __len__(self) -> int:
         self._load()
         size = len(self._files)
         self._loadalllazy()
         for m in self._dirs.values():
             size += m.__len__()
         return size
 
     def __nonzero__(self) -> bool:
         # Faster than "__len__() != 0" since it avoids loading sub-manifests
         return not self._isempty()
 
     __bool__ = __nonzero__
 
     def _isempty(self) -> bool:
         self._load()  # for consistency; already loaded by all callers
         # See if we can skip loading everything.
         if self._files or (
             self._dirs and any(not m._isempty() for m in self._dirs.values())
         ):
             return False
         self._loadalllazy()
         return not self._dirs or all(m._isempty() for m in self._dirs.values())
 
     @encoding.strmethod
     def __repr__(self) -> bytes:
         return (
             b'<treemanifest dir=%s, node=%s, loaded=%r, dirty=%r at 0x%x>'
             % (
                 self._dir,
                 hex(self._node),
                 bool(self._loadfunc is _noop),
                 self._dirty,
                 id(self),
             )
         )
 
     def dir(self) -> bytes:
         """The directory that this tree manifest represents, including a
         trailing '/'. Empty string for the repo root directory."""
         return self._dir
 
     def node(self) -> bytes:
         """This node of this instance. nullid for unsaved instances. Should
         be updated when the instance is read or written from a revlog.
         """
         assert not self._dirty
         return self._node
 
     def setnode(self, node: bytes) -> None:
         self._node = node
         self._dirty = False
 
     def iterentries(
         self,
     ) -> Iterator[Tuple[bytes, Union[bytes, 'treemanifest'], bytes]]:
         self._load()
         self._loadalllazy()
         for p, n in sorted(
             itertools.chain(self._dirs.items(), self._files.items())
         ):
             if p in self._files:
                 yield self._subpath(p), n, self._flags.get(p, b'')
             else:
                 for x in n.iterentries():
                     yield x
 
     def items(self) -> Iterator[Tuple[bytes, Union[bytes, 'treemanifest']]]:
         self._load()
         self._loadalllazy()
         for p, n in sorted(
             itertools.chain(self._dirs.items(), self._files.items())
         ):
             if p in self._files:
                 yield self._subpath(p), n
             else:
                 for f, sn in n.items():
                     yield f, sn
 
     iteritems = items
 
     def iterkeys(self) -> Iterator[bytes]:
         self._load()
         self._loadalllazy()
         for p in sorted(itertools.chain(self._dirs, self._files)):
             if p in self._files:
                 yield self._subpath(p)
             else:
                 for f in self._dirs[p]:
                     yield f
 
     def keys(self) -> List[bytes]:
         return list(self.iterkeys())
 
     def __iter__(self) -> Iterator[bytes]:
         return self.iterkeys()
 
     def __contains__(self, f: bytes) -> bool:
         if f is None:
             return False
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             if dir not in self._dirs:
                 return False
 
             return self._dirs[dir].__contains__(subpath)
         else:
             return f in self._files
 
     def get(self, f: bytes, default: Optional[bytes] = None) -> Optional[bytes]:
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             if dir not in self._dirs:
                 return default
             return self._dirs[dir].get(subpath, default)
         else:
             return self._files.get(f, default)
 
     def __getitem__(self, f: bytes) -> bytes:
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             return self._dirs[dir].__getitem__(subpath)
         else:
             return self._files[f]
 
     def flags(self, f: bytes) -> bytes:
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             if dir not in self._dirs:
                 return b''
             return self._dirs[dir].flags(subpath)
         else:
             if f in self._lazydirs or f in self._dirs:
                 return b''
             return self._flags.get(f, b'')
 
     def find(self, f: bytes) -> Tuple[bytes, bytes]:
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             return self._dirs[dir].find(subpath)
         else:
             return self._files[f], self._flags.get(f, b'')
 
     def __delitem__(self, f: bytes) -> None:
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
 
             self._dirs[dir].__delitem__(subpath)
             # If the directory is now empty, remove it
             if self._dirs[dir]._isempty():
                 del self._dirs[dir]
         else:
             del self._files[f]
             if f in self._flags:
                 del self._flags[f]
         self._dirty = True
 
     def set(self, f: bytes, node: bytes, flags: bytes) -> None:
         """Set both the node and the flags for path f."""
         assert node is not None
         if flags not in _manifestflags:
             raise TypeError(b"Invalid manifest flag set.")
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
             if dir not in self._dirs:
                 self._dirs[dir] = treemanifest(
                     self.nodeconstants, self._subpath(dir)
                 )
             self._dirs[dir].set(subpath, node, flags)
         else:
             assert len(node) in (20, 32)
             self._files[f] = node
             self._flags[f] = flags
         self._dirty = True
 
     def __setitem__(self, f: bytes, n: bytes) -> None:
         assert n is not None
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
             if dir not in self._dirs:
                 self._dirs[dir] = treemanifest(
                     self.nodeconstants, self._subpath(dir)
                 )
             self._dirs[dir].__setitem__(subpath, n)
         else:
             # manifest nodes are either 20 bytes or 32 bytes,
             # depending on the hash in use. Assert this as historically
             # sometimes extra bytes were added.
             assert len(n) in (20, 32)
             self._files[f] = n
         self._dirty = True
 
     def _load(self) -> None:
         if self._loadfunc is not _noop:
             lf, self._loadfunc = self._loadfunc, _noop
             lf(self)
         elif self._copyfunc is not _noop:
             cf, self._copyfunc = self._copyfunc, _noop
             cf(self)
 
     def setflag(self, f: bytes, flags: bytes) -> None:
         """Set the flags (symlink, executable) for path f."""
         if flags not in _manifestflags:
             raise TypeError(b"Invalid manifest flag set.")
         self._load()
         dir, subpath = _splittopdir(f)
         if dir:
             self._loadlazy(dir)
             if dir not in self._dirs:
                 self._dirs[dir] = treemanifest(
                     self.nodeconstants, self._subpath(dir)
                 )
             self._dirs[dir].setflag(subpath, flags)
         else:
             self._flags[f] = flags
         self._dirty = True
 
     def copy(self) -> 'treemanifest':
         copy = treemanifest(self.nodeconstants, self._dir)
         copy._node = self._node
         copy._dirty = self._dirty
         if self._copyfunc is _noop:
 
             def _copyfunc(s):
                 self._load()
                 s._lazydirs = {
                     d: (n, r, True) for d, (n, r, c) in self._lazydirs.items()
                 }
                 sdirs = s._dirs
                 for d, v in self._dirs.items():
                     sdirs[d] = v.copy()
                 s._files = dict.copy(self._files)
                 s._flags = dict.copy(self._flags)
 
             if self._loadfunc is _noop:
                 _copyfunc(copy)
             else:
                 copy._copyfunc = _copyfunc
         else:
             copy._copyfunc = self._copyfunc
         return copy
 
     def filesnotin(
         self, m2: 'treemanifest', match: Optional[matchmod.basematcher] = None
     ) -> Set[bytes]:
         '''Set of files in this manifest that are not in the other'''
         if match and not match.always():
             m1 = self._matches(match)
             m2 = m2._matches(match)
             return m1.filesnotin(m2)
 
         files = set()
 
         def _filesnotin(t1, t2):
             if t1._node == t2._node and not t1._dirty and not t2._dirty:
                 return
             t1._load()
             t2._load()
             self._loaddifflazy(t1, t2)
             for d, m1 in t1._dirs.items():
                 if d in t2._dirs:
                     m2 = t2._dirs[d]
                     _filesnotin(m1, m2)
                 else:
                     files.update(m1.iterkeys())
 
             for fn in t1._files:
                 if fn not in t2._files:
                     files.add(t1._subpath(fn))
 
         _filesnotin(self, m2)
         return files
 
     @propertycache
     def _alldirs(self) -> pathutil.dirs:
         return pathutil.dirs(self)
 
     def dirs(self) -> pathutil.dirs:
         return self._alldirs
 
     def hasdir(self, dir: bytes) -> bool:
         self._load()
         topdir, subdir = _splittopdir(dir)
         if topdir:
             self._loadlazy(topdir)
             if topdir in self._dirs:
                 return self._dirs[topdir].hasdir(subdir)
             return False
         dirslash = dir + b'/'
         return dirslash in self._dirs or dirslash in self._lazydirs
 
     def walk(self, match: matchmod.basematcher) -> Iterator[bytes]:
         """Generates matching file names.
 
         It also reports nonexistent files by marking them bad with match.bad().
         """
         if match.always():
             for f in iter(self):
                 yield f
             return
 
         fset = set(match.files())
 
         for fn in self._walk(match):
             if fn in fset:
                 # specified pattern is the exact name
                 fset.remove(fn)
             yield fn
 
         # for dirstate.walk, files=[''] means "walk the whole tree".
         # follow that here, too
         fset.discard(b'')
 
         for fn in sorted(fset):
             if not self.hasdir(fn):
                 match.bad(fn, None)
 
     def _walk(self, match: matchmod.basematcher) -> Iterator[bytes]:
         '''Recursively generates matching file names for walk().'''
         visit = match.visitchildrenset(self._dir[:-1])
         if not visit:
             return
 
         # yield this dir's files and walk its submanifests
         self._load()
         visit = self._loadchildrensetlazy(visit)
         for p in sorted(list(self._dirs) + list(self._files)):
             if p in self._files:
                 fullp = self._subpath(p)
                 if match(fullp):
                     yield fullp
             else:
                 if not visit or p[:-1] in visit:
                     for f in self._dirs[p]._walk(match):
                         yield f
 
     def _matches(self, match: matchmod.basematcher) -> 'treemanifest':
         """recursively generate a new manifest filtered by the match argument."""
         if match.always():
             return self.copy()
         return self._matches_inner(match)
 
     def _matches_inner(self, match: matchmod.basematcher) -> 'treemanifest':
         if match.always():
             return self.copy()
 
         visit = match.visitchildrenset(self._dir[:-1])
         if visit == b'all':
             return self.copy()
         ret = treemanifest(self.nodeconstants, self._dir)
         if not visit:
             return ret
 
         self._load()
         for fn in self._files:
             # While visitchildrenset *usually* lists only subdirs, this is
             # actually up to the matcher and may have some files in the set().
             # If visit == 'this', we should obviously look at the files in this
             # directory; if visit is a set, and fn is in it, we should inspect
             # fn (but no need to inspect things not in the set).
             if visit != b'this' and fn not in visit:
                 continue
             fullp = self._subpath(fn)
             # visitchildrenset isn't perfect, we still need to call the regular
             # matcher code to further filter results.
             if not match(fullp):
                 continue
             ret._files[fn] = self._files[fn]
             if fn in self._flags:
                 ret._flags[fn] = self._flags[fn]
 
         visit = self._loadchildrensetlazy(visit)
         for dir, subm in self._dirs.items():
             if visit and dir[:-1] not in visit:
                 continue
             m = subm._matches_inner(match)
             if not m._isempty():
                 ret._dirs[dir] = m
 
         if not ret._isempty():
             ret._dirty = True
         return ret
 
     def fastdelta(
         self, base: ByteString, changes: Iterable[Tuple[bytes, bool]]
     ) -> ByteString:
         raise FastdeltaUnavailable()
 
     def diff(
         self,
         m2: 'treemanifest',
         match: Optional[matchmod.basematcher] = None,
         clean: bool = False,
     ) -> Dict[
         bytes,
         Optional[
             Tuple[Tuple[Optional[bytes], bytes], Tuple[Optional[bytes], bytes]]
         ],
     ]:
         """Finds changes between the current manifest and m2.
 
         Args:
           m2: the manifest to which this manifest should be compared.
           clean: if true, include files unchanged between these manifests
                  with a None value in the returned dictionary.
 
         The result is returned as a dict with filename as key and
         values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the
         nodeid in the current/other manifest and fl1/fl2 is the flag
         in the current/other manifest. Where the file does not exist,
         the nodeid will be None and the flags will be the empty
         string.
         """
         if match and not match.always():
             m1 = self._matches(match)
             m2 = m2._matches(match)
             return m1.diff(m2, clean=clean)
         result = {}
         emptytree = treemanifest(self.nodeconstants)
 
         def _iterativediff(t1, t2, stack):
             """compares two tree manifests and append new tree-manifests which
             needs to be compared to stack"""
             if t1._node == t2._node and not t1._dirty and not t2._dirty:
                 return
             t1._load()
             t2._load()
             self._loaddifflazy(t1, t2)
 
             for d, m1 in t1._dirs.items():
                 m2 = t2._dirs.get(d, emptytree)
                 stack.append((m1, m2))
 
             for d, m2 in t2._dirs.items():
                 if d not in t1._dirs:
                     stack.append((emptytree, m2))
 
             for fn, n1 in t1._files.items():
                 fl1 = t1._flags.get(fn, b'')
                 n2 = t2._files.get(fn, None)
                 fl2 = t2._flags.get(fn, b'')
                 if n1 != n2 or fl1 != fl2:
                     result[t1._subpath(fn)] = ((n1, fl1), (n2, fl2))
                 elif clean:
                     result[t1._subpath(fn)] = None
 
             for fn, n2 in t2._files.items():
                 if fn not in t1._files:
                     fl2 = t2._flags.get(fn, b'')
                     result[t2._subpath(fn)] = ((None, b''), (n2, fl2))
 
         stackls = []
         _iterativediff(self, m2, stackls)
         while stackls:
             t1, t2 = stackls.pop()
             # stackls is populated in the function call
             _iterativediff(t1, t2, stackls)
         return result
 
     def unmodifiedsince(self, m2: 'treemanifest') -> bool:
         return not self._dirty and not m2._dirty and self._node == m2._node
 
     def parse(
         self,
         text: bytes,
         readsubtree: Callable[[bytes, bytes], 'treemanifest'],
     ) -> None:
         selflazy = self._lazydirs
         for f, n, fl in _parse(self._nodelen, text):
             if fl == b't':
                 f = f + b'/'
                 # False below means "doesn't need to be copied" and can use the
                 # cached value from readsubtree directly.
                 selflazy[f] = (n, readsubtree, False)
             elif b'/' in f:
                 # This is a flat manifest, so use __setitem__ and setflag rather
                 # than assigning directly to _files and _flags, so we can
                 # assign a path in a subdirectory, and to mark dirty (compared
                 # to nullid).
                 self[f] = n
                 if fl:
                     self.setflag(f, fl)
             else:
                 # Assigning to _files and _flags avoids marking as dirty,
                 # and should be a little faster.
                 self._files[f] = n
                 if fl:
                     self._flags[f] = fl
 
     def text(self) -> ByteString:
         """Get the full data of this manifest as a bytestring."""
         self._load()
         return _text(self.iterentries())
 
     def dirtext(self) -> ByteString:
         """Get the full data of this directory as a bytestring. Make sure that
         any submanifests have been written first, so their nodeids are correct.
         """
         self._load()
         flags = self.flags
         lazydirs = [(d[:-1], v[0], b't') for d, v in self._lazydirs.items()]
         dirs = [(d[:-1], self._dirs[d]._node, b't') for d in self._dirs]
         files = [(f, self._files[f], flags(f)) for f in self._files]
         return _text(sorted(dirs + files + lazydirs))
 
     def read(
         self,
         gettext: Callable[[], ByteString],
         readsubtree: Callable[[bytes, bytes], 'treemanifest'],
     ) -> None:
         def _load_for_read(s):
             s.parse(gettext(), readsubtree)
             s._dirty = False
 
         self._loadfunc = _load_for_read
 
     def writesubtrees(
         self,
         m1: 'treemanifest',
         m2: 'treemanifest',
         writesubtree: Callable[
             [
                 Callable[['treemanifest'], None],
                 bytes,
                 bytes,
                 matchmod.basematcher,
             ],
             None,
         ],
         match: matchmod.basematcher,
     ) -> None:
         self._load()  # for consistency; should never have any effect here
         m1._load()
         m2._load()
         emptytree = treemanifest(self.nodeconstants)
 
         def getnode(m, d):
             ld = m._lazydirs.get(d)
             if ld:
                 return ld[0]
             tree = m._dirs.get(d, emptytree)
             assert tree is not None  # helps pytype
             return tree._node
 
         # let's skip investigating things that `match` says we do not need.
         visit = match.visitchildrenset(self._dir[:-1])
         visit = self._loadchildrensetlazy(visit)
         if visit == b'this' or visit == b'all':
             visit = None
         for d, subm in self._dirs.items():
             if visit and d[:-1] not in visit:
                 continue
             subp1 = getnode(m1, d)
             subp2 = getnode(m2, d)
             if subp1 == self.nodeconstants.nullid:
                 subp1, subp2 = subp2, subp1
             writesubtree(subm, subp1, subp2, match)
 
     def walksubtrees(
         self, matcher: Optional[matchmod.basematcher] = None
     ) -> Iterator['treemanifest']:
         """Returns an iterator of the subtrees of this manifest, including this
         manifest itself.
 
         If `matcher` is provided, it only returns subtrees that match.
         """
         if matcher and not matcher.visitdir(self._dir[:-1]):
             return
         if not matcher or matcher(self._dir[:-1]):
             yield self
 
         self._load()
         # OPT: use visitchildrenset to avoid loading everything.
         self._loadalllazy()
         for d, subm in self._dirs.items():
             for subtree in subm.walksubtrees(matcher=matcher):
                 yield subtree
 
 
 class manifestfulltextcache(util.lrucachedict):
     """File-backed LRU cache for the manifest cache
 
     File consists of entries, up to EOF:
 
     - 20 bytes node, 4 bytes length, <length> manifest data
 
     These are written in reverse cache order (oldest to newest).
 
     """
 
     _file = b'manifestfulltextcache'
 
     def __init__(self, max):
         super(manifestfulltextcache, self).__init__(max)
         self._dirty = False
         self._read = False
         self._opener = None
 
     def read(self):
         if self._read or self._opener is None:
             return
 
         try:
             with self._opener(self._file) as fp:
                 set = super(manifestfulltextcache, self).__setitem__
                 # ignore trailing data, this is a cache, corruption is skipped
                 while True:
                     # TODO do we need to do work here for sha1 portability?
                     node = fp.read(20)
                     if len(node) < 20:
                         break
                     try:
                         size = struct.unpack(b'>L', fp.read(4))[0]
                     except struct.error:
                         break
                     value = bytearray(fp.read(size))
                     if len(value) != size:
                         break
                     set(node, value)
         except IOError:
             # the file is allowed to be missing
             pass
 
         self._read = True
         self._dirty = False
 
     def write(self):
         if not self._dirty or self._opener is None:
             return
         # rotate backwards to the first used node
         try:
             with self._opener(
                 self._file, b'w', atomictemp=True, checkambig=True
             ) as fp:
                 node = self._head.prev
                 while True:
                     if node.key in self._cache:
                         fp.write(node.key)
                         fp.write(struct.pack(b'>L', len(node.value)))
                         fp.write(node.value)
                     if node is self._head:
                         break
                     node = node.prev
         except IOError:
             # We could not write the cache (eg: permission error)
             # the content can be missing.
             #
             # We could try harder and see if we could recreate a wcache
             # directory were we coudl write too.
             #
             # XXX the error pass silently, having some way to issue an error
             # log `ui.log` would be nice.
             pass
 
     def __len__(self):
         if not self._read:
             self.read()
         return super(manifestfulltextcache, self).__len__()
 
     def __contains__(self, k):
         if not self._read:
             self.read()
         return super(manifestfulltextcache, self).__contains__(k)
 
     def __iter__(self):
         if not self._read:
             self.read()
         return super(manifestfulltextcache, self).__iter__()
 
     def __getitem__(self, k):
         if not self._read:
             self.read()
         # the cache lru order can change on read
         setdirty = self._cache.get(k) is not self._head
         value = super(manifestfulltextcache, self).__getitem__(k)
         if setdirty:
             self._dirty = True
         return value
 
     def __setitem__(self, k, v):
         if not self._read:
             self.read()
         super(manifestfulltextcache, self).__setitem__(k, v)
         self._dirty = True
 
     def __delitem__(self, k):
         if not self._read:
             self.read()
         super(manifestfulltextcache, self).__delitem__(k)
         self._dirty = True
 
     def get(self, k, default=None):
         if not self._read:
             self.read()
         return super(manifestfulltextcache, self).get(k, default=default)
 
     def clear(self, clear_persisted_data=False):
         super(manifestfulltextcache, self).clear()
         if clear_persisted_data:
             self._dirty = True
             self.write()
         self._read = False
 
 
 # and upper bound of what we expect from compression
 # (real live value seems to be "3")
 MAXCOMPRESSION = 3
 
 
 class FastdeltaUnavailable(Exception):
     """Exception raised when fastdelta isn't usable on a manifest."""
 
 
 class manifestrevlog:  # (repository.imanifeststorage)
     """A revlog that stores manifest texts. This is responsible for caching the
     full-text manifest contents.
     """
 
     def __init__(
         self,
         nodeconstants,
         opener,
         tree=b'',
         dirlogcache=None,
         treemanifest=False,
     ):
         """Constructs a new manifest revlog
 
         `indexfile` - used by extensions to have two manifests at once, like
         when transitioning between flatmanifeset and treemanifests.
 
         `treemanifest` - used to indicate this is a tree manifest revlog. Opener
         options can also be used to make this a tree manifest revlog. The opener
         option takes precedence, so if it is set to True, we ignore whatever
         value is passed in to the constructor.
         """
         self.nodeconstants = nodeconstants
         # During normal operations, we expect to deal with not more than four
         # revs at a time (such as during commit --amend). When rebasing large
         # stacks of commits, the number can go up, hence the config knob below.
         cachesize = 4
         optiontreemanifest = False
         persistentnodemap = False
         opts = getattr(opener, 'options', None)
         if opts is not None:
             cachesize = opts.get(b'manifestcachesize', cachesize)
             optiontreemanifest = opts.get(b'treemanifest', False)
             persistentnodemap = opts.get(b'persistent-nodemap', False)
 
         self._treeondisk = optiontreemanifest or treemanifest
 
         self._fulltextcache = manifestfulltextcache(cachesize)
 
         if tree:
             assert self._treeondisk, (tree, b'opts is %r' % opts)
 
         radix = b'00manifest'
         if tree:
             radix = b"meta/" + tree + radix
 
         self.tree = tree
 
         # The dirlogcache is kept on the root manifest log
         if tree:
             self._dirlogcache = dirlogcache
         else:
             self._dirlogcache = {b'': self}
 
         self._revlog = revlog.revlog(
             opener,
             target=(revlog_constants.KIND_MANIFESTLOG, self.tree),
             radix=radix,
             # only root indexfile is cached
             checkambig=not bool(tree),
             mmaplargeindex=True,
             upperboundcomp=MAXCOMPRESSION,
             persistentnodemap=persistentnodemap,
         )
 
         self.index = self._revlog.index
 
     def get_revlog(self):
         """return an actual revlog instance if any
 
         This exist because a lot of code leverage the fact the underlying
         storage is a revlog for optimization, so giving simple way to access
         the revlog instance helps such code.
         """
         return self._revlog
 
     def _setupmanifestcachehooks(self, repo):
         """Persist the manifestfulltextcache on lock release"""
         if not hasattr(repo, '_wlockref'):
             return
 
         self._fulltextcache._opener = repo.wcachevfs
         if repo._currentlock(repo._wlockref) is None:
             return
 
         reporef = weakref.ref(repo)
         manifestrevlogref = weakref.ref(self)
 
         def persistmanifestcache(success):
             # Repo is in an unknown state, do not persist.
             if not success:
                 return
 
             repo = reporef()
             self = manifestrevlogref()
             if repo is None or self is None:
                 return
             if repo.manifestlog.getstorage(b'') is not self:
                 # there's a different manifest in play now, abort
                 return
             self._fulltextcache.write()
 
         repo._afterlock(persistmanifestcache)
 
     @property
     def fulltextcache(self):
         return self._fulltextcache
 
     def clearcaches(self, clear_persisted_data: bool = False) -> None:
         self._revlog.clearcaches()
         self._fulltextcache.clear(clear_persisted_data=clear_persisted_data)
         self._dirlogcache = {self.tree: self}
 
     def dirlog(self, d):
         if d:
             assert self._treeondisk
         if d not in self._dirlogcache:
             mfrevlog = manifestrevlog(
                 self.nodeconstants,
                 self.opener,
                 d,
                 self._dirlogcache,
                 treemanifest=self._treeondisk,
             )
             self._dirlogcache[d] = mfrevlog
         return self._dirlogcache[d]
 
     def add(
         self,
         m,
         transaction,
         link,
         p1,
         p2,
         added: Iterable[bytes],
         removed: Iterable[bytes],
         readtree=None,
         match=None,
     ):
         """add some manifest entry in to the manifest log
 
         input:
 
           m:           the manifest dict we want to store
           transaction: the open transaction
           p1:          manifest-node of p1
           p2:          manifest-node of p2
           added:       file added/changed compared to parent
           removed:     file removed compared to parent
 
         tree manifest input:
 
           readtree:    a function to read a subtree
           match:       a filematcher for the subpart of the tree manifest
         """
         try:
             if p1 not in self.fulltextcache:
                 raise FastdeltaUnavailable()
             # If our first parent is in the manifest cache, we can
             # compute a delta here using properties we know about the
             # manifest up-front, which may save time later for the
             # revlog layer.
 
             _checkforbidden(added)
             # combine the changed lists into one sorted iterator
             work = heapq.merge(
                 [(x, False) for x in sorted(added)],
                 [(x, True) for x in sorted(removed)],
             )
 
             arraytext, deltatext = m.fastdelta(self.fulltextcache[p1], work)
             cachedelta = self._revlog.rev(p1), deltatext
             text = util.buffer(arraytext)
             rev = self._revlog.addrevision(
                 text, transaction, link, p1, p2, cachedelta
             )
             n = self._revlog.node(rev)
         except FastdeltaUnavailable:
             # The first parent manifest isn't already loaded or the
             # manifest implementation doesn't support fastdelta, so
             # we'll just encode a fulltext of the manifest and pass
             # that through to the revlog layer, and let it handle the
             # delta process.
             if self._treeondisk:
                 assert readtree, b"readtree must be set for treemanifest writes"
                 assert match, b"match must be specified for treemanifest writes"
                 m1 = readtree(self.tree, p1)
                 m2 = readtree(self.tree, p2)
                 n = self._addtree(
                     m, transaction, link, m1, m2, readtree, match=match
                 )
                 arraytext = None
             else:
                 text = m.text()
                 rev = self._revlog.addrevision(text, transaction, link, p1, p2)
                 n = self._revlog.node(rev)
                 arraytext = bytearray(text)
 
         if arraytext is not None:
             self.fulltextcache[n] = arraytext
 
         return n
 
     def _addtree(self, m, transaction, link, m1, m2, readtree, match):
         # If the manifest is unchanged compared to one parent,
         # don't write a new revision
         if self.tree != b'' and (
             m.unmodifiedsince(m1) or m.unmodifiedsince(m2)
         ):
             return m.node()
 
         def writesubtree(subm, subp1, subp2, match):
             sublog = self.dirlog(subm.dir())
             sublog.add(
                 subm,
                 transaction,
                 link,
                 subp1,
                 subp2,
                 None,
                 None,
                 readtree=readtree,
                 match=match,
             )
 
         m.writesubtrees(m1, m2, writesubtree, match)
         text = m.dirtext()
         n = None
         if self.tree != b'':
             # Double-check whether contents are unchanged to one parent
             if text == m1.dirtext():
                 n = m1.node()
             elif text == m2.dirtext():
                 n = m2.node()
 
         if not n:
             rev = self._revlog.addrevision(
                 text, transaction, link, m1.node(), m2.node()
             )
             n = self._revlog.node(rev)
 
         # Save nodeid so parent manifest can calculate its nodeid
         m.setnode(n)
         return n
 
     def __len__(self):
         return len(self._revlog)
 
     def __iter__(self):
         return self._revlog.__iter__()
 
     def rev(self, node):
         return self._revlog.rev(node)
 
     def node(self, rev):
         return self._revlog.node(rev)
 
     def lookup(self, value):
         return self._revlog.lookup(value)
 
     def parentrevs(self, rev):
         return self._revlog.parentrevs(rev)
 
     def parents(self, node):
         return self._revlog.parents(node)
 
     def linkrev(self, rev):
         return self._revlog.linkrev(rev)
 
     def checksize(self):
         return self._revlog.checksize()
 
     def revision(self, node):
         return self._revlog.revision(node)
 
     def rawdata(self, node):
         return self._revlog.rawdata(node)
 
     def revdiff(self, rev1, rev2):
         return self._revlog.revdiff(rev1, rev2)
 
     def cmp(self, node, text):
         return self._revlog.cmp(node, text)
 
     def deltaparent(self, rev):
         return self._revlog.deltaparent(rev)
 
     def emitrevisions(
         self,
         nodes,
         nodesorder=None,
         revisiondata=False,
         assumehaveparentrevisions=False,
         deltamode=repository.CG_DELTAMODE_STD,
         sidedata_helpers=None,
         debug_info=None,
     ):
         return self._revlog.emitrevisions(
             nodes,
             nodesorder=nodesorder,
             revisiondata=revisiondata,
             assumehaveparentrevisions=assumehaveparentrevisions,
             deltamode=deltamode,
             sidedata_helpers=sidedata_helpers,
             debug_info=debug_info,
         )
 
     def addgroup(
         self,
         deltas,
         linkmapper,
         transaction,
         alwayscache=False,
         addrevisioncb=None,
         duplicaterevisioncb=None,
         debug_info=None,
         delta_base_reuse_policy=None,
     ):
         return self._revlog.addgroup(
             deltas,
             linkmapper,
             transaction,
             alwayscache=alwayscache,
             addrevisioncb=addrevisioncb,
             duplicaterevisioncb=duplicaterevisioncb,
             debug_info=debug_info,
             delta_base_reuse_policy=delta_base_reuse_policy,
         )
 
     def rawsize(self, rev):
         return self._revlog.rawsize(rev)
 
     def getstrippoint(self, minlink):
         return self._revlog.getstrippoint(minlink)
 
     def strip(self, minlink, transaction):
         return self._revlog.strip(minlink, transaction)
 
     def files(self):
         return self._revlog.files()
 
     def clone(self, tr, destrevlog, **kwargs):
         if not isinstance(destrevlog, manifestrevlog):
             raise error.ProgrammingError(b'expected manifestrevlog to clone()')
 
         return self._revlog.clone(tr, destrevlog._revlog, **kwargs)
 
     def storageinfo(
         self,
         exclusivefiles=False,
         sharedfiles=False,
         revisionscount=False,
         trackedsize=False,
         storedsize=False,
     ):
         return self._revlog.storageinfo(
             exclusivefiles=exclusivefiles,
             sharedfiles=sharedfiles,
             revisionscount=revisionscount,
             trackedsize=trackedsize,
             storedsize=storedsize,
         )
 
     @property
     def opener(self):
         return self._revlog.opener
 
     @opener.setter
     def opener(self, value):
         self._revlog.opener = value
 
 
 AnyManifestCtx = Union['ManifestCtx', 'TreeManifestCtx']
 # TODO: drop this in favor of repository.imanifestdict
 AnyManifestDict = Union[manifestdict, treemanifest]
 
 
-class ManifestLog:
+class manifestlog:  # (repository.imanifestlog)
     """A collection class representing the collection of manifest snapshots
     referenced by commits in the repository.
 
     In this situation, 'manifest' refers to the abstract concept of a snapshot
     of the list of files in the given commit. Consumers of the output of this
     class do not care about the implementation details of the actual manifests
     they receive (i.e. tree or flat or lazily loaded, etc)."""
 
     def __init__(self, opener, repo, rootstore, narrowmatch):
         self.nodeconstants = repo.nodeconstants
         usetreemanifest = False
         cachesize = 4
 
         opts = getattr(opener, 'options', None)
         if opts is not None:
             usetreemanifest = opts.get(b'treemanifest', usetreemanifest)
             cachesize = opts.get(b'manifestcachesize', cachesize)
 
         self._treemanifests = usetreemanifest
 
         self._rootstore = rootstore
         self._rootstore._setupmanifestcachehooks(repo)
         self._narrowmatch = narrowmatch
 
         # A cache of the manifestctx or treemanifestctx for each directory
         self._dirmancache = {}
         self._dirmancache[b''] = util.lrucachedict(cachesize)
 
         self._cachesize = cachesize
 
     def __getitem__(self, node):
         """Retrieves the manifest instance for the given node. Throws a
         LookupError if not found.
         """
         return self.get(b'', node)
 
     @property
     def narrowed(self):
         return not (self._narrowmatch is None or self._narrowmatch.always())
 
     def get(
         self, tree: bytes, node: bytes, verify: bool = True
     ) -> AnyManifestCtx:
         """Retrieves the manifest instance for the given node. Throws a
         LookupError if not found.
 
         `verify` - if True an exception will be thrown if the node is not in
                    the revlog
         """
         if node in self._dirmancache.get(tree, ()):
             return self._dirmancache[tree][node]
 
         if not self._narrowmatch.always():
             if not self._narrowmatch.visitdir(tree[:-1]):
                 return excludeddirmanifestctx(self.nodeconstants, tree, node)
         if tree:
             if self._rootstore._treeondisk:
                 if verify:
                     # Side-effect is LookupError is raised if node doesn't
                     # exist.
                     self.getstorage(tree).rev(node)
 
                 m = treemanifestctx(self, tree, node)
             else:
                 raise error.Abort(
                     _(
                         b"cannot ask for manifest directory '%s' in a flat "
                         b"manifest"
                     )
                     % tree
                 )
         else:
             if verify:
                 # Side-effect is LookupError is raised if node doesn't exist.
                 self._rootstore.rev(node)
 
             if self._treemanifests:
                 m = treemanifestctx(self, b'', node)
             else:
                 m = manifestctx(self, node)
 
         if node != self.nodeconstants.nullid:
             mancache = self._dirmancache.get(tree)
             if not mancache:
                 mancache = util.lrucachedict(self._cachesize)
                 self._dirmancache[tree] = mancache
             mancache[node] = m
         return m
 
     def getstorage(self, tree):
         return self._rootstore.dirlog(tree)
 
     def clearcaches(self, clear_persisted_data: bool = False) -> None:
         self._dirmancache.clear()
         self._rootstore.clearcaches(clear_persisted_data=clear_persisted_data)
 
     def rev(self, node) -> int:
         return self._rootstore.rev(node)
 
     def update_caches(self, transaction) -> None:
         return self._rootstore._revlog.update_caches(transaction=transaction)
 
 
-manifestlog = interfaceutil.implementer(repository.imanifestlog)(ManifestLog)
-
-if typing.TYPE_CHECKING:
-    manifestlog = ManifestLog
-
-
 class MemManifestCtx:
     _manifestdict: manifestdict
 
     def __init__(self, manifestlog):
         self._manifestlog = manifestlog
         self._manifestdict = manifestdict(manifestlog.nodeconstants.nodelen)
 
     def _storage(self) -> manifestrevlog:
         return self._manifestlog.getstorage(b'')
 
     def copy(self) -> 'MemManifestCtx':
         memmf = memmanifestctx(self._manifestlog)
         memmf._manifestdict = self.read().copy()
         return memmf
 
     def read(self) -> 'manifestdict':
         return self._manifestdict
 
     def write(self, transaction, link, p1, p2, added, removed, match=None):
         return self._storage().add(
             self._manifestdict,
             transaction,
             link,
             p1,
             p2,
             added,
             removed,
             match=match,
         )
 
 
 memmanifestctx = interfaceutil.implementer(
     repository.imanifestrevisionwritable
 )(MemManifestCtx)
 
 if typing.TYPE_CHECKING:
     memmanifestctx = MemManifestCtx
 
 
 class ManifestCtx:
     """A class representing a single revision of a manifest, including its
     contents, its parent revs, and its linkrev.
     """
 
     _data: Optional[manifestdict]
 
     def __init__(self, manifestlog, node):
         self._manifestlog = manifestlog
         self._data = None
 
         self._node = node
 
         # TODO: We eventually want p1, p2, and linkrev exposed on this class,
         # but let's add it later when something needs it and we can load it
         # lazily.
         # self.p1, self.p2 = store.parents(node)
         # rev = store.rev(node)
         # self.linkrev = store.linkrev(rev)
 
     def _storage(self) -> 'manifestrevlog':
         return self._manifestlog.getstorage(b'')
 
     def node(self) -> bytes:
         return self._node
 
     def copy(self) -> MemManifestCtx:
         memmf = memmanifestctx(self._manifestlog)
         memmf._manifestdict = self.read().copy()
         return memmf
 
     @propertycache
     def parents(self) -> Tuple[bytes, bytes]:
         return self._storage().parents(self._node)
 
     def read(self) -> 'manifestdict':
         if self._data is None:
             nc = self._manifestlog.nodeconstants
             if self._node == nc.nullid:
                 self._data = manifestdict(nc.nodelen)
             else:
                 store = self._storage()
                 if self._node in store.fulltextcache:
                     text = pycompat.bytestr(store.fulltextcache[self._node])
                 else:
                     text = store.revision(self._node)
                     arraytext = bytearray(text)
                     store.fulltextcache[self._node] = arraytext
                 self._data = manifestdict(nc.nodelen, text)
         return self._data
 
     def readfast(self, shallow: bool = False) -> 'manifestdict':
         """Calls either readdelta or read, based on which would be less work.
         readdelta is called if the delta is against the p1, and therefore can be
         read quickly.
 
         If `shallow` is True, nothing changes since this is a flat manifest.
         """
         util.nouideprecwarn(
             b'"readfast" is deprecated use "read_any_fast_delta" or "read_delta_parents"',
             b"6.9",
             stacklevel=2,
         )
         store = self._storage()
         r = store.rev(self._node)
         deltaparent = store.deltaparent(r)
         if deltaparent != nullrev and deltaparent in store.parentrevs(r):
             return self.readdelta()
         return self.read()
 
     def readdelta(self, shallow: bool = False) -> 'manifestdict':
         """Returns a manifest containing just the entries that are present
         in this manifest, but not in its p1 manifest. This is efficient to read
         if the revlog delta is already p1.
 
         Changing the value of `shallow` has no effect on flat manifests.
         """
         util.nouideprecwarn(
             b'"readfast" is deprecated use "read_any_fast_delta" or "read_delta_new_entries"',
             b"6.9",
             stacklevel=2,
         )
         store = self._storage()
         r = store.rev(self._node)
         d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r))
         return manifestdict(store.nodeconstants.nodelen, d)
 
     def read_any_fast_delta(
         self,
         valid_bases: Optional[Collection[int]] = None,
         *,
         shallow: bool = False,
     ) -> Tuple[Optional[int], manifestdict]:
         """see `imanifestrevisionstored` documentation"""
         store = self._storage()
         r = store.rev(self._node)
         deltaparent = store.deltaparent(r)
         if valid_bases is None:
             # make sure the next check is True
             valid_bases = (deltaparent,)
         if deltaparent != nullrev and deltaparent in valid_bases:
             d = mdiff.patchtext(store.revdiff(deltaparent, r))
             return (
                 deltaparent,
                 manifestdict(store.nodeconstants.nodelen, d),
             )
         return (None, self.read())
 
     def read_delta_parents(
         self,
         *,
         shallow: bool = False,
         exact: bool = True,
     ) -> manifestdict:
         """see `interface.imanifestrevisionbase` documentations"""
         store = self._storage()
         r = store.rev(self._node)
         deltaparent = store.deltaparent(r)
         parents = [p for p in store.parentrevs(r) if p is not nullrev]
         if not exact and deltaparent in parents:
             d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r))
             return manifestdict(store.nodeconstants.nodelen, d)
         elif not exact or len(parents) == 0:
             return self.read()
         elif len(parents) == 1:
             p = parents[0]
             d = mdiff.patchtext(store.revdiff(p, r))
             return manifestdict(store.nodeconstants.nodelen, d)
         else:
             p1, p2 = parents
             d1 = mdiff.patchtext(store.revdiff(p1, r))
             d2 = mdiff.patchtext(store.revdiff(p2, r))
             d1 = manifestdict(store.nodeconstants.nodelen, d1)
             d2 = manifestdict(store.nodeconstants.nodelen, d2)
             md = manifestdict(store.nodeconstants.nodelen)
             for f, new_node, new_flag in d1.iterentries():
                 if f not in d2:
                     continue
                 if new_node is not None:
                     md.set(f, new_node, new_flag)
             return md
 
     def read_delta_new_entries(self, *, shallow=False) -> manifestdict:
         """see `interface.imanifestrevisionbase` documentations"""
         # If we are using narrow, returning a delta against an arbitrary
         # changeset might return file outside the narrowspec. This can create
         # issue when running validation server side with strict security as
         # push from low priviledge usage might be seen as adding new revision
         # for files they cannot touch. So we are strict if narrow is involved.
         if self._manifestlog.narrowed:
             return self.read_delta_parents(shallow=shallow, exact=True)
         store = self._storage()
         r = store.rev(self._node)
         d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r))
         return manifestdict(store.nodeconstants.nodelen, d)
 
     def find(self, key: bytes) -> Tuple[bytes, bytes]:
         return self.read().find(key)
 
 
 manifestctx = interfaceutil.implementer(repository.imanifestrevisionstored)(
     ManifestCtx
 )
 
 if typing.TYPE_CHECKING:
     manifestctx = ManifestCtx
 
 
 class MemTreeManifestCtx:
     _treemanifest: treemanifest
 
     def __init__(self, manifestlog, dir=b''):
         self._manifestlog = manifestlog
         self._dir = dir
         self._treemanifest = treemanifest(manifestlog.nodeconstants)
 
     def _storage(self) -> manifestrevlog:
         return self._manifestlog.getstorage(b'')
 
     def copy(self) -> 'MemTreeManifestCtx':
         memmf = memtreemanifestctx(self._manifestlog, dir=self._dir)
         memmf._treemanifest = self._treemanifest.copy()
         return memmf
 
     def read(self) -> 'treemanifest':
         return self._treemanifest
 
     def write(self, transaction, link, p1, p2, added, removed, match=None):
         def readtree(dir, node):
             return self._manifestlog.get(dir, node).read()
 
         return self._storage().add(
             self._treemanifest,
             transaction,
             link,
             p1,
             p2,
             added,
             removed,
             readtree=readtree,
             match=match,
         )
 
 
 memtreemanifestctx = interfaceutil.implementer(
     repository.imanifestrevisionwritable
 )(MemTreeManifestCtx)
 
 if typing.TYPE_CHECKING:
     memtreemanifestctx = MemTreeManifestCtx
 
 
 class TreeManifestCtx:
     _data: Optional[treemanifest]
 
     def __init__(self, manifestlog, dir, node):
         self._manifestlog = manifestlog
         self._dir = dir
         self._data = None
 
         self._node = node
 
         # TODO: Load p1/p2/linkrev lazily. They need to be lazily loaded so that
         # we can instantiate treemanifestctx objects for directories we don't
         # have on disk.
         # self.p1, self.p2 = store.parents(node)
         # rev = store.rev(node)
         # self.linkrev = store.linkrev(rev)
 
     def _storage(self) -> manifestrevlog:
         narrowmatch = self._manifestlog._narrowmatch
         if not narrowmatch.always():
             if not narrowmatch.visitdir(self._dir[:-1]):
                 return excludedmanifestrevlog(
                     self._manifestlog.nodeconstants, self._dir
                 )
         return self._manifestlog.getstorage(self._dir)
 
     def read(self) -> 'treemanifest':
         if self._data is None:
             store = self._storage()
             if self._node == self._manifestlog.nodeconstants.nullid:
                 self._data = treemanifest(self._manifestlog.nodeconstants)
             # TODO accessing non-public API
             elif store._treeondisk:
                 m = treemanifest(self._manifestlog.nodeconstants, dir=self._dir)
 
                 def gettext():
                     return store.revision(self._node)
 
                 def readsubtree(dir, subm):
                     # Set verify to False since we need to be able to create
                     # subtrees for trees that don't exist on disk.
                     return self._manifestlog.get(dir, subm, verify=False).read()
 
                 m.read(gettext, readsubtree)
                 m.setnode(self._node)
                 self._data = m
             else:
                 if self._node in store.fulltextcache:
                     text = pycompat.bytestr(store.fulltextcache[self._node])
                 else:
                     text = store.revision(self._node)
                     arraytext = bytearray(text)
                     store.fulltextcache[self._node] = arraytext
                 self._data = treemanifest(
                     self._manifestlog.nodeconstants, dir=self._dir, text=text
                 )
 
         return self._data
 
     def node(self) -> bytes:
         return self._node
 
     def copy(self) -> 'MemTreeManifestCtx':
         memmf = memtreemanifestctx(self._manifestlog, dir=self._dir)
         memmf._treemanifest = self.read().copy()
         return memmf
 
     @propertycache
     def parents(self) -> Tuple[bytes, bytes]:
         return self._storage().parents(self._node)
 
     def readdelta(self, shallow: bool = False) -> AnyManifestDict:
         """see `imanifestrevisionstored` documentation"""
         util.nouideprecwarn(
             b'"readdelta" is deprecated use "read_any_fast_delta" or "read_delta_new_entries"',
             b"6.9",
             stacklevel=2,
         )
         store = self._storage()
         if shallow:
             r = store.rev(self._node)
             d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r))
             return manifestdict(store.nodeconstants.nodelen, d)
         else:
             # Need to perform a slow delta
             r0 = store.deltaparent(store.rev(self._node))
             m0 = self._manifestlog.get(self._dir, store.node(r0)).read()
             m1 = self.read()
             md = treemanifest(self._manifestlog.nodeconstants, dir=self._dir)
             for f, ((n0, fl0), (n1, fl1)) in m0.diff(m1).items():
                 if n1:
                     md[f] = n1
                     if fl1:
                         md.setflag(f, fl1)
             return md
 
     def read_any_fast_delta(
         self,
         valid_bases: Optional[Collection[int]] = None,
         *,
         shallow: bool = False,
     ) -> Tuple[Optional[int], AnyManifestDict]:
         """see `imanifestrevisionstored` documentation"""
         store = self._storage()
         r = store.rev(self._node)
         deltaparent = store.deltaparent(r)
 
         if valid_bases is None:
             # make sure the next check is True
             valid_bases = (deltaparent,)
         can_use_delta = deltaparent != nullrev and deltaparent in valid_bases
 
         if shallow:
             if can_use_delta:
                 return (deltaparent, self._read_storage_delta_shallow())
             else:
                 d = store.revision(self._node)
                 return (None, manifestdict(store.nodeconstants.nodelen, d))
         else:
             # note: This use "slow_delta" here is cargo culted from the previous
             # implementation. I am not sure it make sense since the goal here is to
             # be fast, so why are we computing a delta? On the other hand, tree
             # manifest delta as fairly "cheap" and allow for skipping whole part of
             # the tree that a full read would access. So it might be a good idea.
             #
             # If we realize we don't need delta here, we should simply use:
             #
             #     return (None, self.read())
             if can_use_delta:
                 return (None, self._read_storage_slow_delta(base=deltaparent))
             else:
                 parents = [
                     p
                     for p in store.parentrevs(r)
                     if p is not nullrev and p in valid_bases
                 ]
                 if parents:
                     best_base = max(parents)
                 else:
                     best_base = max(valid_bases)
                 return (None, self._read_storage_slow_delta(base=best_base))
 
     def _read_storage_delta_shallow(self) -> manifestdict:
         store = self._storage()
         r = store.rev(self._node)
         d = mdiff.patchtext(store.revdiff(store.deltaparent(r), r))
         return manifestdict(store.nodeconstants.nodelen, d)
 
     def _read_storage_slow_delta(self, base) -> 'treemanifest':
         store = self._storage()
         if base is None:
             base = store.deltaparent(store.rev(self._node))
         m0 = self._manifestlog.get(self._dir, store.node(base)).read()
         m1 = self.read()
         md = treemanifest(self._manifestlog.nodeconstants, dir=self._dir)
         for f, ((n0, fl0), (n1, fl1)) in m0.diff(m1).items():
             if n1:
                 md[f] = n1
                 if fl1:
                     md.setflag(f, fl1)
         return md
 
     def read_delta_parents(
         self,
         *,
         shallow: bool = False,
         exact: bool = True,
     ) -> AnyManifestDict:
         """see `interface.imanifestrevisionbase` documentations"""
         store = self._storage()
         r = store.rev(self._node)
         parents = [p for p in store.parentrevs(r) if p is not nullrev]
         if not exact:
             return self.read_any_fast_delta(parents, shallow=shallow)[1]
         elif len(parents) == 0:
             if shallow:
                 d = store.revision(self._node)
                 return manifestdict(store.nodeconstants.nodelen, d)
             else:
                 return self.read()
         elif len(parents) == 1:
             p = parents[0]
             if shallow:
                 d = mdiff.patchtext(store.revdiff(p, r))
                 return manifestdict(store.nodeconstants.nodelen, d)
             else:
                 return self._read_storage_slow_delta(base=p)
         else:
             p1, p2 = parents
             if shallow:
                 d1 = mdiff.patchtext(store.revdiff(p1, r))
                 d2 = mdiff.patchtext(store.revdiff(p2, r))
                 d1 = manifestdict(store.nodeconstants.nodelen, d1)
                 d2 = manifestdict(store.nodeconstants.nodelen, d2)
                 md = manifestdict(store.nodeconstants.nodelen)
                 for f, new_node, new_flag in d1.iterentries():
                     if f not in d2:
                         continue
                     if new_node is not None:
                         md.set(f, new_node, new_flag)
                 return md
             else:
                 m1 = self._manifestlog.get(self._dir, store.node(p1)).read()
                 m2 = self._manifestlog.get(self._dir, store.node(p2)).read()
                 mc = self.read()
                 d1 = m1.diff(mc)
                 d2 = m2.diff(mc)
                 md = treemanifest(
                     self._manifestlog.nodeconstants,
                     dir=self._dir,
                 )
                 for f, new_node, new_flag in d1.iterentries():
                     if f not in d2:
                         continue
                     if new_node is not None:
                         md.set(f, new_node, new_flag)
                 return md
 
     def read_delta_new_entries(
         self, *, shallow: bool = False
     ) -> AnyManifestDict:
         """see `interface.imanifestrevisionbase` documentations"""
         # If we are using narrow, returning a delta against an arbitrary
         # changeset might return file outside the narrowspec. This can create
         # issue when running validation server side with strict security as
         # push from low priviledge usage might be seen as adding new revision
         # for files they cannot touch. So we are strict if narrow is involved.
         if self._manifestlog.narrowed:
             return self.read_delta_parents(shallow=shallow, exact=True)
         # delegate to existing another existing method for simplicity
         store = self._storage()
         r = store.rev(self._node)
         bases = (store.deltaparent(r),)
         return self.read_any_fast_delta(bases, shallow=shallow)[1]
 
     def readfast(self, shallow=False) -> AnyManifestDict:
         """Calls either readdelta or read, based on which would be less work.
         readdelta is called if the delta is against the p1, and therefore can be
         read quickly.
 
         If `shallow` is True, it only returns the entries from this manifest,
         and not any submanifests.
         """
         util.nouideprecwarn(
             b'"readdelta" is deprecated use "read_any_fast_delta" or "read_delta_parents"',
             b"6.9",
             stacklevel=2,
         )
         store = self._storage()
         r = store.rev(self._node)
         deltaparent = store.deltaparent(r)
         if deltaparent != nullrev and deltaparent in store.parentrevs(r):
             return self.readdelta(shallow=shallow)
 
         if shallow:
             return manifestdict(
                 store.nodeconstants.nodelen, store.revision(self._node)
             )
         else:
             return self.read()
 
     def find(self, key: bytes) -> Tuple[bytes, bytes]:
         return self.read().find(key)
 
 
 treemanifestctx = interfaceutil.implementer(repository.imanifestrevisionstored)(
     TreeManifestCtx
 )
 
 if typing.TYPE_CHECKING:
     treemanifestctx = TreeManifestCtx
 
 
 class excludeddir(treemanifest):
     """Stand-in for a directory that is excluded from the repository.
 
     With narrowing active on a repository that uses treemanifests,
     some of the directory revlogs will be excluded from the resulting
     clone. This is a huge storage win for clients, but means we need
     some sort of pseudo-manifest to surface to internals so we can
     detect a merge conflict outside the narrowspec. That's what this
     class is: it stands in for a directory whose node is known, but
     whose contents are unknown.
     """
 
     _files: Dict[bytes, bytes]
     _flags: Dict[bytes, bytes]
 
     def __init__(self, nodeconstants, dir, node):
         super(excludeddir, self).__init__(nodeconstants, dir)
         self._node = node
         # Add an empty file, which will be included by iterators and such,
         # appearing as the directory itself (i.e. something like "dir/")
         self._files[b''] = node
         self._flags[b''] = b't'
 
     # Manifests outside the narrowspec should never be modified, so avoid
     # copying. This makes a noticeable difference when there are very many
     # directories outside the narrowspec. Also, it makes sense for the copy to
     # be of the same type as the original, which would not happen with the
     # super type's copy().
     def copy(self):
         return self
 
 
 class excludeddirmanifestctx(treemanifestctx):
     """context wrapper for excludeddir - see that docstring for rationale"""
 
     def __init__(self, nodeconstants, dir, node):
         self.nodeconstants = nodeconstants
         self._dir = dir
         self._node = node
 
     def read(self):
         return excludeddir(self.nodeconstants, self._dir, self._node)
 
     def readfast(self, shallow=False):
         # special version of readfast since we don't have underlying storage
         return self.read()
 
     def write(self, *args):
         raise error.ProgrammingError(
             b'attempt to write manifest from excluded dir %s' % self._dir
         )
 
 
 class excludedmanifestrevlog(manifestrevlog):
     """Stand-in for excluded treemanifest revlogs.
 
     When narrowing is active on a treemanifest repository, we'll have
     references to directories we can't see due to the revlog being
     skipped. This class exists to conform to the manifestrevlog
     interface for those directories and proactively prevent writes to
     outside the narrowspec.
     """
 
     def __init__(self, nodeconstants, dir):
         self.nodeconstants = nodeconstants
         self._dir = dir
 
     def __len__(self):
         raise error.ProgrammingError(
             b'attempt to get length of excluded dir %s' % self._dir
         )
 
     def rev(self, node):
         raise error.ProgrammingError(
             b'attempt to get rev from excluded dir %s' % self._dir
         )
 
     def linkrev(self, node):
         raise error.ProgrammingError(
             b'attempt to get linkrev from excluded dir %s' % self._dir
         )
 
     def node(self, rev):
         raise error.ProgrammingError(
             b'attempt to get node from excluded dir %s' % self._dir
         )
 
     def add(self, *args, **kwargs):
         # We should never write entries in dirlogs outside the narrow clone.
         # However, the method still gets called from writesubtree() in
         # _addtree(), so we need to handle it. We should possibly make that
         # avoid calling add() with a clean manifest (_dirty is always False
         # in excludeddir instances).
         pass

mercurial/statichttprepo.py

@@ -1,281 +1,281
 # statichttprepo.py - simple http repository class for mercurial
 #
 # This provides read-only repo access to repositories exported via static http
 #
 # Copyright 2005-2007 Olivia Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import annotations
 
 import errno
 
 from .i18n import _
 from .node import sha1nodeconstants
 from . import (
     branchmap,
     changelog,
     error,
     localrepo,
     manifest,
     namespaces,
     pathutil,
     pycompat,
     requirements as requirementsmod,
     url,
     util,
     vfs as vfsmod,
 )
 from .utils import (
     urlutil,
 )
 
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 
 class httprangereader:
     def __init__(self, url, opener):
         # we assume opener has HTTPRangeHandler
         self.url = url
         self.pos = 0
         self.opener = opener
         self.name = url
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_value, traceback):
         self.close()
 
     def seek(self, pos):
         self.pos = pos
 
     def read(self, n: int = -1):
         req = urlreq.request(pycompat.strurl(self.url))
         end = ''
 
         if n == 0:
             return b''
         elif n > 0:
             end = "%d" % (self.pos + n - 1)
         if self.pos or end:
             req.add_header('Range', 'bytes=%d-%s' % (self.pos, end))
 
         try:
             f = self.opener.open(req)
             data = f.read()
             code = f.code
         except urlerr.httperror as inst:
             num = inst.code == 404 and errno.ENOENT or None
             # Explicitly convert the exception to str as Py3 will try
             # convert it to local encoding and with as the HTTPResponse
             # instance doesn't support encode.
             raise IOError(num, str(inst))
         except urlerr.urlerror as inst:
             raise IOError(None, inst.reason)
 
         if code == 200:
             # HTTPRangeHandler does nothing if remote does not support
             # Range headers and returns the full entity. Let's slice it.
             if n > 0 and (self.pos + n) < len(data):
                 data = data[self.pos : self.pos + n]
             elif self.pos < len(data):
                 data = data[self.pos :]
             else:
                 data = b''
         elif 0 < n < len(data):
             data = data[:n]
         self.pos += len(data)
         return data
 
     def readlines(self):
         return self.read().splitlines(True)
 
     def __iter__(self):
         return iter(self.readlines())
 
     def close(self):
         pass
 
 
 # _RangeError and _HTTPRangeHandler were originally in byterange.py,
 # which was itself extracted from urlgrabber. See the last version of
 # byterange.py from history if you need more information.
 class _RangeError(IOError):
     """Error raised when an unsatisfiable range is requested."""
 
 
 class _HTTPRangeHandler(urlreq.basehandler):
     """Handler that enables HTTP Range headers.
 
     This was extremely simple. The Range header is a HTTP feature to
     begin with so all this class does is tell urllib2 that the
     "206 Partial Content" response from the HTTP server is what we
     expected.
     """
 
     def http_error_206(self, req, fp, code, msg, hdrs):
         # 206 Partial Content Response
         r = urlreq.addinfourl(fp, hdrs, req.get_full_url())
         r.code = code
         r.msg = msg
         return r
 
     def http_error_416(self, req, fp, code, msg, hdrs):
         # HTTP's Range Not Satisfiable error
         raise _RangeError('Requested Range Not Satisfiable')
 
 
 def build_opener(ui, authinfo):
     # urllib cannot handle URLs with embedded user or passwd
     urlopener = url.opener(ui, authinfo)
     urlopener.add_handler(_HTTPRangeHandler())
 
     class statichttpvfs(vfsmod.abstractvfs):
         def __init__(self, base):
             self.base = base
             self.options = {}
 
         def __call__(self, path, mode=b'r', *args, **kw):
             if mode not in (b'r', b'rb'):
                 raise IOError('Permission denied')
             f = b"/".join((self.base, urlreq.quote(path)))
             return httprangereader(f, urlopener)
 
         def _auditpath(self, path: bytes, mode: bytes) -> None:
             raise NotImplementedError
 
         def join(self, path, *insidef):
             if path:
                 return pathutil.join(self.base, path, *insidef)
             else:
                 return self.base
 
     return statichttpvfs
 
 
 class statichttppeer(localrepo.localpeer):
     def local(self):
         return None
 
     def canpush(self):
         return False
 
 
 class statichttprepository(
     localrepo.localrepository, localrepo.revlogfilestorage
 ):
     supported = localrepo.localrepository._basesupported
 
-    manifestlog: manifest.ManifestLog
+    manifestlog: manifest.manifestlog
 
     def __init__(self, ui, path):
         self._url = path
         self.ui = ui
 
         self.root = path
         u = urlutil.url(path.rstrip(b'/') + b"/.hg")
         self.path, authinfo = u.authinfo()
 
         vfsclass = build_opener(ui, authinfo)
         self.vfs = vfsclass(self.path)
         self.cachevfs = vfsclass(self.vfs.join(b'cache'))
         self._phasedefaults = []
 
         self.names = namespaces.namespaces()
         self.filtername = None
         self._extrafilterid = None
         self._wanted_sidedata = set()
         self.features = set()
 
         try:
             requirements = set(self.vfs.read(b'requires').splitlines())
         except FileNotFoundError:
             requirements = set()
 
             # check if it is a non-empty old-style repository
             try:
                 with self.vfs(b"00changelog.i") as fp:
                     fp.read(1)
             except FileNotFoundError:
                 # we do not care about empty old-style repositories here
                 msg = _(b"'%s' does not appear to be an hg repository") % path
                 raise error.RepoError(msg)
         if requirementsmod.SHARESAFE_REQUIREMENT in requirements:
             storevfs = vfsclass(self.vfs.join(b'store'))
             requirements |= set(storevfs.read(b'requires').splitlines())
 
         supportedrequirements = localrepo.gathersupportedrequirements(ui)
         localrepo.ensurerequirementsrecognized(
             requirements, supportedrequirements
         )
         localrepo.ensurerequirementscompatible(ui, requirements)
         self.nodeconstants = sha1nodeconstants
         self.nullid = self.nodeconstants.nullid
 
         # setup store
         self.store = localrepo.makestore(requirements, self.path, vfsclass)
         self.spath = self.store.path
         self.svfs = self.store.opener
         self.sjoin = self.store.join
         self._filecache = {}
         self.requirements = requirements
 
         rootmanifest = manifest.manifestrevlog(self.nodeconstants, self.svfs)
         self.manifestlog = manifest.manifestlog(
             self.svfs, self, rootmanifest, self.narrowmatch()
         )
         self.changelog = changelog.changelog(self.svfs)
         self._tags = None
         self.nodetagscache = None
         self._branchcaches = branchmap.BranchMapCache()
         self._revbranchcache = None
         self.encodepats = None
         self.decodepats = None
         self._transref = None
         self._dirstate = None
 
     def _restrictcapabilities(self, caps):
         caps = super(statichttprepository, self)._restrictcapabilities(caps)
         return caps.difference([b"pushkey"])
 
     def url(self):
         return self._url
 
     def local(self):
         return False
 
     def peer(self, path=None, remotehidden=False):
         return statichttppeer(self, path=path, remotehidden=remotehidden)
 
     def wlock(self, wait=True):
         raise error.LockUnavailable(
             0,
             pycompat.sysstr(_(b'lock not available')),
             b'lock',
             _(b'cannot lock static-http repository'),
         )
 
     def lock(self, wait=True):
         raise error.LockUnavailable(
             0,
             pycompat.sysstr(_(b'lock not available')),
             b'lock',
             _(b'cannot lock static-http repository'),
         )
 
     def _writecaches(self):
         pass  # statichttprepository are read only
 
 
 def make_peer(
     ui, path, create, intents=None, createopts=None, remotehidden=False
 ):
     if create:
         raise error.Abort(_(b'cannot create new static-http repository'))
     url = path.loc[7:]
     return statichttprepository(ui, url).peer(
         path=path, remotehidden=remotehidden
     )

mercurial/store.py

@@ -1,1258 +1,1258
 # store.py - repository store handling for Mercurial)
 #
 # Copyright 2008 Olivia Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import annotations
 
 import collections
 import functools
 import os
 import re
 import stat
 import typing
 
 from typing import (
     Generator,
     List,
     Optional,
 )
 
 from .i18n import _
 from .thirdparty import attr
 
 # Force pytype to use the non-vendored package
 if typing.TYPE_CHECKING:
     # noinspection PyPackageRequirements
     import attr
 
 from .node import hex
 from .revlogutils.constants import (
     INDEX_HEADER,
     KIND_CHANGELOG,
     KIND_FILELOG,
     KIND_MANIFESTLOG,
 )
 from . import (
     changelog,
     error,
     filelog,
     manifest,
     policy,
     pycompat,
     revlog as revlogmod,
     util,
     vfs as vfsmod,
 )
 from .utils import hashutil
 
 parsers = policy.importmod('parsers')
 # how much bytes should be read from fncache in one read
 # It is done to prevent loading large fncache files into memory
 fncache_chunksize = 10**6
 
 
 def _match_tracked_entry(entry: "BaseStoreEntry", matcher):
     """parses a fncache entry and returns whether the entry is tracking a path
     matched by matcher or not.
 
     If matcher is None, returns True"""
 
     if matcher is None:
         return True
 
     # TODO: make this safe for other entry types.  Currently, the various
     #  store.data_entry generators only yield  RevlogStoreEntry, so the
     #  attributes do exist on `entry`.
     # pytype: disable=attribute-error
     if entry.is_filelog:
         return matcher(entry.target_id)
     elif entry.is_manifestlog:
         return matcher.visitdir(entry.target_id.rstrip(b'/'))
     # pytype: enable=attribute-error
     raise error.ProgrammingError(b"cannot process entry %r" % entry)
 
 
 # This avoids a collision between a file named foo and a dir named
 # foo.i or foo.d
 def _encodedir(path):
     """
     >>> _encodedir(b'data/foo.i')
     'data/foo.i'
     >>> _encodedir(b'data/foo.i/bla.i')
     'data/foo.i.hg/bla.i'
     >>> _encodedir(b'data/foo.i.hg/bla.i')
     'data/foo.i.hg.hg/bla.i'
     >>> _encodedir(b'data/foo.i\\ndata/foo.i/bla.i\\ndata/foo.i.hg/bla.i\\n')
     'data/foo.i\\ndata/foo.i.hg/bla.i\\ndata/foo.i.hg.hg/bla.i\\n'
     """
     return (
         path.replace(b".hg/", b".hg.hg/")
         .replace(b".i/", b".i.hg/")
         .replace(b".d/", b".d.hg/")
     )
 
 
 encodedir = getattr(parsers, 'encodedir', _encodedir)
 
 
 def decodedir(path):
     """
     >>> decodedir(b'data/foo.i')
     'data/foo.i'
     >>> decodedir(b'data/foo.i.hg/bla.i')
     'data/foo.i/bla.i'
     >>> decodedir(b'data/foo.i.hg.hg/bla.i')
     'data/foo.i.hg/bla.i'
     """
     if b".hg/" not in path:
         return path
     return (
         path.replace(b".d.hg/", b".d/")
         .replace(b".i.hg/", b".i/")
         .replace(b".hg.hg/", b".hg/")
     )
 
 
 def _reserved():
     """characters that are problematic for filesystems
 
     * ascii escapes (0..31)
     * ascii hi (126..255)
     * windows specials
 
     these characters will be escaped by encodefunctions
     """
     winreserved = [ord(x) for x in u'\\:*?"<>|']
     for x in range(32):
         yield x
     for x in range(126, 256):
         yield x
     for x in winreserved:
         yield x
 
 
 def _buildencodefun():
     """
     >>> enc, dec = _buildencodefun()
 
     >>> enc(b'nothing/special.txt')
     'nothing/special.txt'
     >>> dec(b'nothing/special.txt')
     'nothing/special.txt'
 
     >>> enc(b'HELLO')
     '_h_e_l_l_o'
     >>> dec(b'_h_e_l_l_o')
     'HELLO'
 
     >>> enc(b'hello:world?')
     'hello~3aworld~3f'
     >>> dec(b'hello~3aworld~3f')
     'hello:world?'
 
     >>> enc(b'the\\x07quick\\xADshot')
     'the~07quick~adshot'
     >>> dec(b'the~07quick~adshot')
     'the\\x07quick\\xadshot'
     """
     e = b'_'
     xchr = pycompat.bytechr
     asciistr = list(map(xchr, range(127)))
     capitals = list(range(ord(b"A"), ord(b"Z") + 1))
 
     cmap = {x: x for x in asciistr}
     for x in _reserved():
         cmap[xchr(x)] = b"~%02x" % x
     for x in capitals + [ord(e)]:
         cmap[xchr(x)] = e + xchr(x).lower()
 
     dmap = {}
     for k, v in cmap.items():
         dmap[v] = k
 
     def decode(s):
         i = 0
         while i < len(s):
             for l in range(1, 4):
                 try:
                     yield dmap[s[i : i + l]]
                     i += l
                     break
                 except KeyError:
                     pass
             else:
                 raise KeyError
 
     return (
         lambda s: b''.join([cmap[s[c : c + 1]] for c in range(len(s))]),
         lambda s: b''.join(list(decode(s))),
     )
 
 
 _encodefname, _decodefname = _buildencodefun()
 
 
 def encodefilename(s):
     """
     >>> encodefilename(b'foo.i/bar.d/bla.hg/hi:world?/HELLO')
     'foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o'
     """
     return _encodefname(encodedir(s))
 
 
 def decodefilename(s):
     """
     >>> decodefilename(b'foo.i.hg/bar.d.hg/bla.hg.hg/hi~3aworld~3f/_h_e_l_l_o')
     'foo.i/bar.d/bla.hg/hi:world?/HELLO'
     """
     return decodedir(_decodefname(s))
 
 
 def _buildlowerencodefun():
     """
     >>> f = _buildlowerencodefun()
     >>> f(b'nothing/special.txt')
     'nothing/special.txt'
     >>> f(b'HELLO')
     'hello'
     >>> f(b'hello:world?')
     'hello~3aworld~3f'
     >>> f(b'the\\x07quick\\xADshot')
     'the~07quick~adshot'
     """
     xchr = pycompat.bytechr
     cmap = {xchr(x): xchr(x) for x in range(127)}
     for x in _reserved():
         cmap[xchr(x)] = b"~%02x" % x
     for x in range(ord(b"A"), ord(b"Z") + 1):
         cmap[xchr(x)] = xchr(x).lower()
 
     def lowerencode(s):
         return b"".join([cmap[c] for c in pycompat.iterbytestr(s)])
 
     return lowerencode
 
 
 lowerencode = getattr(parsers, 'lowerencode', None) or _buildlowerencodefun()
 
 # Windows reserved names: con, prn, aux, nul, com1..com9, lpt1..lpt9
 _winres3 = (b'aux', b'con', b'prn', b'nul')  # length 3
 _winres4 = (b'com', b'lpt')  # length 4 (with trailing 1..9)
 
 
 def _auxencode(path, dotencode):
     """
     Encodes filenames containing names reserved by Windows or which end in
     period or space. Does not touch other single reserved characters c.
     Specifically, c in '\\:*?"<>|' or ord(c) <= 31 are *not* encoded here.
     Additionally encodes space or period at the beginning, if dotencode is
     True. Parameter path is assumed to be all lowercase.
     A segment only needs encoding if a reserved name appears as a
     basename (e.g. "aux", "aux.foo"). A directory or file named "foo.aux"
     doesn't need encoding.
 
     >>> s = b'.foo/aux.txt/txt.aux/con/prn/nul/foo.'
     >>> _auxencode(s.split(b'/'), True)
     ['~2efoo', 'au~78.txt', 'txt.aux', 'co~6e', 'pr~6e', 'nu~6c', 'foo~2e']
     >>> s = b'.com1com2/lpt9.lpt4.lpt1/conprn/com0/lpt0/foo.'
     >>> _auxencode(s.split(b'/'), False)
     ['.com1com2', 'lp~749.lpt4.lpt1', 'conprn', 'com0', 'lpt0', 'foo~2e']
     >>> _auxencode([b'foo. '], True)
     ['foo.~20']
     >>> _auxencode([b' .foo'], True)
     ['~20.foo']
     """
     for i, n in enumerate(path):
         if not n:
             continue
         if dotencode and n[0] in b'. ':
             n = b"~%02x" % ord(n[0:1]) + n[1:]
             path[i] = n
         else:
             l = n.find(b'.')
             if l == -1:
                 l = len(n)
             if (l == 3 and n[:3] in _winres3) or (
                 l == 4
                 and n[3:4] <= b'9'
                 and n[3:4] >= b'1'
                 and n[:3] in _winres4
             ):
                 # encode third letter ('aux' -> 'au~78')
                 ec = b"~%02x" % ord(n[2:3])
                 n = n[0:2] + ec + n[3:]
                 path[i] = n
         if n[-1] in b'. ':
             # encode last period or space ('foo...' -> 'foo..~2e')
             path[i] = n[:-1] + b"~%02x" % ord(n[-1:])
     return path
 
 
 _maxstorepathlen = 120
 _dirprefixlen = 8
 _maxshortdirslen = 8 * (_dirprefixlen + 1) - 4
 
 
 def _hashencode(path, dotencode):
     digest = hex(hashutil.sha1(path).digest())
     le = lowerencode(path[5:]).split(b'/')  # skips prefix 'data/' or 'meta/'
     parts = _auxencode(le, dotencode)
     basename = parts[-1]
     _root, ext = os.path.splitext(basename)
     sdirs = []
     sdirslen = 0
     for p in parts[:-1]:
         d = p[:_dirprefixlen]
         if d[-1] in b'. ':
             # Windows can't access dirs ending in period or space
             d = d[:-1] + b'_'
         if sdirslen == 0:
             t = len(d)
         else:
             t = sdirslen + 1 + len(d)
             if t > _maxshortdirslen:
                 break
         sdirs.append(d)
         sdirslen = t
     dirs = b'/'.join(sdirs)
     if len(dirs) > 0:
         dirs += b'/'
     res = b'dh/' + dirs + digest + ext
     spaceleft = _maxstorepathlen - len(res)
     if spaceleft > 0:
         filler = basename[:spaceleft]
         res = b'dh/' + dirs + filler + digest + ext
     return res
 
 
 def _hybridencode(path, dotencode):
     """encodes path with a length limit
 
     Encodes all paths that begin with 'data/', according to the following.
 
     Default encoding (reversible):
 
     Encodes all uppercase letters 'X' as '_x'. All reserved or illegal
     characters are encoded as '~xx', where xx is the two digit hex code
     of the character (see encodefilename).
     Relevant path components consisting of Windows reserved filenames are
     masked by encoding the third character ('aux' -> 'au~78', see _auxencode).
 
     Hashed encoding (not reversible):
 
     If the default-encoded path is longer than _maxstorepathlen, a
     non-reversible hybrid hashing of the path is done instead.
     This encoding uses up to _dirprefixlen characters of all directory
     levels of the lowerencoded path, but not more levels than can fit into
     _maxshortdirslen.
     Then follows the filler followed by the sha digest of the full path.
     The filler is the beginning of the basename of the lowerencoded path
     (the basename is everything after the last path separator). The filler
     is as long as possible, filling in characters from the basename until
     the encoded path has _maxstorepathlen characters (or all chars of the
     basename have been taken).
     The extension (e.g. '.i' or '.d') is preserved.
 
     The string 'data/' at the beginning is replaced with 'dh/', if the hashed
     encoding was used.
     """
     path = encodedir(path)
     ef = _encodefname(path).split(b'/')
     res = b'/'.join(_auxencode(ef, dotencode))
     if len(res) > _maxstorepathlen:
         res = _hashencode(path, dotencode)
     return res
 
 
 def _pathencode(path):
     de = encodedir(path)
     if len(path) > _maxstorepathlen:
         return _hashencode(de, True)
     ef = _encodefname(de).split(b'/')
     res = b'/'.join(_auxencode(ef, True))
     if len(res) > _maxstorepathlen:
         return _hashencode(de, True)
     return res
 
 
 _pathencode = getattr(parsers, 'pathencode', _pathencode)
 
 
 def _plainhybridencode(f):
     return _hybridencode(f, False)
 
 
 def _calcmode(vfs):
     try:
         # files in .hg/ will be created using this mode
         mode = vfs.stat().st_mode
         # avoid some useless chmods
         if (0o777 & ~util.umask) == (0o777 & mode):
             mode = None
     except OSError:
         mode = None
     return mode
 
 
 _data = [
     b'bookmarks',
     b'narrowspec',
     b'data',
     b'meta',
     b'00manifest.d',
     b'00manifest.i',
     b'00changelog.d',
     b'00changelog.i',
     b'phaseroots',
     b'obsstore',
     b'requires',
 ]
 
 REVLOG_FILES_EXT = (
     b'.i',
     b'.idx',
     b'.d',
     b'.dat',
     b'.n',
     b'.nd',
     b'.sda',
 )
 # file extension that also use a `-SOMELONGIDHASH.ext` form
 REVLOG_FILES_LONG_EXT = (
     b'.nd',
     b'.idx',
     b'.dat',
     b'.sda',
 )
 # files that are "volatile" and might change between listing and streaming
 #
 # note: the ".nd" file are nodemap data and won't "change" but they might be
 # deleted.
 REVLOG_FILES_VOLATILE_EXT = (b'.n', b'.nd')
 
 # some exception to the above matching
 #
 # XXX This is currently not in use because of issue6542
 EXCLUDED = re.compile(br'.*undo\.[^/]+\.(nd?|i)$')
 
 
 def is_revlog(f, kind, st):
     if kind != stat.S_IFREG:
         return False
     if f.endswith(REVLOG_FILES_EXT):
         return True
     return False
 
 
 def is_revlog_file(f):
     if f.endswith(REVLOG_FILES_EXT):
         return True
     return False
 
 
 @attr.s(slots=True)
 class StoreFile:
     """a file matching a store entry"""
 
     unencoded_path = attr.ib()
     _file_size = attr.ib(default=None)
     is_volatile = attr.ib(default=False)
 
     def file_size(self, vfs):
         if self._file_size is None:
             if vfs is None:
                 msg = b"calling vfs-less file_size without prior call: %s"
                 msg %= self.unencoded_path
                 raise error.ProgrammingError(msg)
             try:
                 self._file_size = vfs.stat(self.unencoded_path).st_size
             except FileNotFoundError:
                 self._file_size = 0
         return self._file_size
 
     @property
     def has_size(self):
         return self._file_size is not None
 
     def get_stream(self, vfs, volatiles):
         """return data "stream" information for this file
 
         (unencoded_file_path, content_iterator, content_size)
         """
         size = self.file_size(None)
 
         def get_stream():
             path = vfs.join(self.unencoded_path)
             with volatiles.open(path) as fp:
                 yield None  # ready to stream
                 if size <= 65536:
                     yield fp.read(size)
                 else:
                     yield from util.filechunkiter(fp, limit=size)
 
         s = get_stream()
         next(s)
         return (self.unencoded_path, s, size)
 
 
 @attr.s(slots=True, init=False)
 class BaseStoreEntry:
     """An entry in the store
 
     This is returned by `store.walk` and represent some data in the store."""
 
     maybe_volatile = True
 
     def files(self) -> List[StoreFile]:
         raise NotImplementedError
 
     def get_streams(
         self,
         repo=None,
         vfs=None,
         volatiles=None,
         max_changeset=None,
         preserve_file_count=False,
     ):
         """return a list of data stream associated to files for this entry
 
         return [(unencoded_file_path, content_iterator, content_size), …]
         """
         assert vfs is not None
         return [f.get_stream(vfs, volatiles) for f in self.files()]
 
 
 @attr.s(slots=True, init=False)
 class SimpleStoreEntry(BaseStoreEntry):
     """A generic entry in the store"""
 
     is_revlog = False
 
     maybe_volatile = attr.ib()
     _entry_path = attr.ib()
     _is_volatile = attr.ib(default=False)
     _file_size = attr.ib(default=None)
     _files = attr.ib(default=None)
 
     def __init__(
         self,
         entry_path,
         is_volatile=False,
         file_size=None,
     ):
         super().__init__()
         self._entry_path = entry_path
         self._is_volatile = is_volatile
         self._file_size = file_size
         self._files = None
         self.maybe_volatile = is_volatile
 
     def files(self) -> List[StoreFile]:
         if self._files is None:
             self._files = [
                 StoreFile(
                     unencoded_path=self._entry_path,
                     file_size=self._file_size,
                     is_volatile=self._is_volatile,
                 )
             ]
         return self._files
 
 
 @attr.s(slots=True, init=False)
 class RevlogStoreEntry(BaseStoreEntry):
     """A revlog entry in the store"""
 
     is_revlog = True
 
     revlog_type = attr.ib(default=None)
     target_id = attr.ib(default=None)
     maybe_volatile = attr.ib(default=True)
     _path_prefix = attr.ib(default=None)
     _details = attr.ib(default=None)
     _files = attr.ib(default=None)
 
     def __init__(
         self,
         revlog_type,
         path_prefix,
         target_id,
         details,
     ):
         super().__init__()
         self.revlog_type = revlog_type
         self.target_id = target_id
         self._path_prefix = path_prefix
         assert b'.i' in details, (path_prefix, details)
         for ext in details:
             if ext.endswith(REVLOG_FILES_VOLATILE_EXT):
                 self.maybe_volatile = True
                 break
         else:
             self.maybe_volatile = False
         self._details = details
         self._files = None
 
     @property
     def is_changelog(self):
         return self.revlog_type == KIND_CHANGELOG
 
     @property
     def is_manifestlog(self):
         return self.revlog_type == KIND_MANIFESTLOG
 
     @property
     def is_filelog(self):
         return self.revlog_type == KIND_FILELOG
 
     def main_file_path(self):
         """unencoded path of the main revlog file"""
         return self._path_prefix + b'.i'
 
     def files(self) -> List[StoreFile]:
         if self._files is None:
             self._files = []
             for ext in sorted(self._details, key=_ext_key):
                 path = self._path_prefix + ext
                 file_size = self._details[ext]
                 # files that are "volatile" and might change between
                 # listing and streaming
                 #
                 # note: the ".nd" file are nodemap data and won't "change"
                 # but they might be deleted.
                 volatile = ext.endswith(REVLOG_FILES_VOLATILE_EXT)
                 f = StoreFile(path, file_size, volatile)
                 self._files.append(f)
         return self._files
 
     def get_streams(
         self,
         repo=None,
         vfs=None,
         volatiles=None,
         max_changeset=None,
         preserve_file_count=False,
     ):
         pre_sized = all(f.has_size for f in self.files())
         if pre_sized and (
             repo is None
             or max_changeset is None
             # This use revlog-v2, ignore for now
             or any(k.endswith(b'.idx') for k in self._details.keys())
             # This is not inline, no race expected
             or b'.d' in self._details
         ):
             return super().get_streams(
                 repo=repo,
                 vfs=vfs,
                 volatiles=volatiles,
                 max_changeset=max_changeset,
                 preserve_file_count=preserve_file_count,
             )
         elif not preserve_file_count:
             stream = [
                 f.get_stream(vfs, volatiles)
                 for f in self.files()
                 if not f.unencoded_path.endswith((b'.i', b'.d'))
             ]
             rl = self.get_revlog_instance(repo).get_revlog()
             rl_stream = rl.get_streams(max_changeset)
             stream.extend(rl_stream)
             return stream
 
         name_to_size = {}
         for f in self.files():
             name_to_size[f.unencoded_path] = f.file_size(None)
 
         stream = [
             f.get_stream(vfs, volatiles)
             for f in self.files()
             if not f.unencoded_path.endswith(b'.i')
         ]
 
         index_path = self._path_prefix + b'.i'
 
         index_file = None
         try:
             index_file = vfs(index_path)
             header = index_file.read(INDEX_HEADER.size)
             if revlogmod.revlog.is_inline_index(header):
                 size = name_to_size[index_path]
 
                 # no split underneath, just return the stream
                 def get_stream():
                     fp = index_file
                     try:
                         fp.seek(0)
                         yield None
                         if size <= 65536:
                             yield fp.read(size)
                         else:
                             yield from util.filechunkiter(fp, limit=size)
                     finally:
                         fp.close()
 
                 s = get_stream()
                 next(s)
                 index_file = None
                 stream.append((index_path, s, size))
             else:
                 rl = self.get_revlog_instance(repo).get_revlog()
                 rl_stream = rl.get_streams(max_changeset, force_inline=True)
                 for name, s, size in rl_stream:
                     if name_to_size.get(name, 0) != size:
                         msg = _(b"expected %d bytes but %d provided for %s")
                         msg %= name_to_size.get(name, 0), size, name
                         raise error.Abort(msg)
                 stream.extend(rl_stream)
         finally:
             if index_file is not None:
                 index_file.close()
 
         files = self.files()
         assert len(stream) == len(files), (
             stream,
             files,
             self._path_prefix,
             self.target_id,
         )
         return stream
 
     def get_revlog_instance(self, repo):
         """Obtain a revlog instance from this store entry
 
         An instance of the appropriate class is returned.
         """
         if self.is_changelog:
             return changelog.changelog(repo.svfs)
         elif self.is_manifestlog:
             mandir = self.target_id
             return manifest.manifestrevlog(
                 repo.nodeconstants, repo.svfs, tree=mandir
             )
         else:
             return filelog.filelog(repo.svfs, self.target_id)
 
 
 def _gather_revlog(files_data):
     """group files per revlog prefix
 
     The returns a two level nested dict. The top level key is the revlog prefix
     without extension, the second level is all the file "suffix" that were
     seen for this revlog and arbitrary file data as value.
     """
     revlogs = collections.defaultdict(dict)
     for u, value in files_data:
         name, ext = _split_revlog_ext(u)
         revlogs[name][ext] = value
     return sorted(revlogs.items())
 
 
 def _split_revlog_ext(filename):
     """split the revlog file prefix from the variable extension"""
     if filename.endswith(REVLOG_FILES_LONG_EXT):
         char = b'-'
     else:
         char = b'.'
     idx = filename.rfind(char)
     return filename[:idx], filename[idx:]
 
 
 def _ext_key(ext):
     """a key to order revlog suffix
 
     important to issue .i after other entry."""
     # the only important part of this order is to keep the `.i` last.
     if ext.endswith(b'.n'):
         return (0, ext)
     elif ext.endswith(b'.nd'):
         return (10, ext)
     elif ext.endswith(b'.d'):
         return (20, ext)
     elif ext.endswith(b'.i'):
         return (50, ext)
     else:
         return (40, ext)
 
 
 class basicstore:
     '''base class for local repository stores'''
 
     def __init__(self, path, vfstype):
         vfs = vfstype(path)
         self.path = vfs.base
         self.createmode = _calcmode(vfs)
         vfs.createmode = self.createmode
         self.rawvfs = vfs
         self.vfs = vfsmod.filtervfs(vfs, encodedir)
         self.opener = self.vfs
 
     def join(self, f):
         return self.path + b'/' + encodedir(f)
 
     def _walk(self, relpath, recurse, undecodable=None):
         '''yields (revlog_type, unencoded, size)'''
         path = self.path
         if relpath:
             path += b'/' + relpath
         striplen = len(self.path) + 1
         l = []
         if self.rawvfs.isdir(path):
             visit = [path]
             readdir = self.rawvfs.readdir
             while visit:
                 p = visit.pop()
                 for f, kind, st in readdir(p, stat=True):
                     fp = p + b'/' + f
                     if is_revlog(f, kind, st):
                         n = util.pconvert(fp[striplen:])
                         l.append((decodedir(n), st.st_size))
                     elif kind == stat.S_IFDIR and recurse:
                         visit.append(fp)
 
         l.sort()
         return l
 
     def changelog(self, trypending, concurrencychecker=None):
         return changelog.changelog(
             self.vfs,
             trypending=trypending,
             concurrencychecker=concurrencychecker,
         )
 
-    def manifestlog(self, repo, storenarrowmatch) -> manifest.ManifestLog:
+    def manifestlog(self, repo, storenarrowmatch) -> manifest.manifestlog:
         rootstore = manifest.manifestrevlog(repo.nodeconstants, self.vfs)
         return manifest.manifestlog(self.vfs, repo, rootstore, storenarrowmatch)
 
     def data_entries(
         self, matcher=None, undecodable=None
     ) -> Generator[BaseStoreEntry, None, None]:
         """Like walk, but excluding the changelog and root manifest.
 
         When [undecodable] is None, revlogs names that can't be
         decoded cause an exception. When it is provided, it should
         be a list and the filenames that can't be decoded are added
         to it instead. This is very rarely needed."""
         dirs = [
             (b'data', KIND_FILELOG, False),
             (b'meta', KIND_MANIFESTLOG, True),
         ]
         for base_dir, rl_type, strip_filename in dirs:
             files = self._walk(base_dir, True, undecodable=undecodable)
             for revlog, details in _gather_revlog(files):
                 revlog_target_id = revlog.split(b'/', 1)[1]
                 if strip_filename and b'/' in revlog:
                     revlog_target_id = revlog_target_id.rsplit(b'/', 1)[0]
                     revlog_target_id += b'/'
                 yield RevlogStoreEntry(
                     path_prefix=revlog,
                     revlog_type=rl_type,
                     target_id=revlog_target_id,
                     details=details,
                 )
 
     def top_entries(
         self, phase=False, obsolescence=False
     ) -> Generator[BaseStoreEntry, None, None]:
         if phase and self.vfs.exists(b'phaseroots'):
             yield SimpleStoreEntry(
                 entry_path=b'phaseroots',
                 is_volatile=True,
             )
 
         if obsolescence and self.vfs.exists(b'obsstore'):
             # XXX if we had the file size it could be non-volatile
             yield SimpleStoreEntry(
                 entry_path=b'obsstore',
                 is_volatile=True,
             )
 
         files = reversed(self._walk(b'', False))
 
         changelogs = collections.defaultdict(dict)
         manifestlogs = collections.defaultdict(dict)
 
         for u, s in files:
             if u.startswith(b'00changelog'):
                 name, ext = _split_revlog_ext(u)
                 changelogs[name][ext] = s
             elif u.startswith(b'00manifest'):
                 name, ext = _split_revlog_ext(u)
                 manifestlogs[name][ext] = s
             else:
                 yield SimpleStoreEntry(
                     entry_path=u,
                     is_volatile=False,
                     file_size=s,
                 )
         # yield manifest before changelog
         top_rl = [
             (manifestlogs, KIND_MANIFESTLOG),
             (changelogs, KIND_CHANGELOG),
         ]
         assert len(manifestlogs) <= 1
         assert len(changelogs) <= 1
         for data, revlog_type in top_rl:
             for revlog, details in sorted(data.items()):
                 yield RevlogStoreEntry(
                     path_prefix=revlog,
                     revlog_type=revlog_type,
                     target_id=b'',
                     details=details,
                 )
 
     def walk(
         self, matcher=None, phase=False, obsolescence=False
     ) -> Generator[BaseStoreEntry, None, None]:
         """return files related to data storage (ie: revlogs)
 
         yields instance from BaseStoreEntry subclasses
 
         if a matcher is passed, storage files of only those tracked paths
         are passed with matches the matcher
         """
         # yield data files first
         for x in self.data_entries(matcher):
             yield x
         for x in self.top_entries(phase=phase, obsolescence=obsolescence):
             yield x
 
     def copylist(self):
         return _data
 
     def write(self, tr):
         pass
 
     def invalidatecaches(self):
         pass
 
     def markremoved(self, fn):
         pass
 
     def __contains__(self, path):
         '''Checks if the store contains path'''
         path = b"/".join((b"data", path))
         # file?
         if self.vfs.exists(path + b".i"):
             return True
         # dir?
         if not path.endswith(b"/"):
             path = path + b"/"
         return self.vfs.exists(path)
 
 
 class encodedstore(basicstore):
     def __init__(self, path, vfstype):
         vfs = vfstype(path + b'/store')
         self.path = vfs.base
         self.createmode = _calcmode(vfs)
         vfs.createmode = self.createmode
         self.rawvfs = vfs
         self.vfs = vfsmod.filtervfs(vfs, encodefilename)
         self.opener = self.vfs
 
     def _walk(self, relpath, recurse, undecodable=None):
         old = super()._walk(relpath, recurse)
         new = []
         for f1, value in old:
             try:
                 f2 = decodefilename(f1)
             except KeyError:
                 if undecodable is None:
                     msg = _(b'undecodable revlog name %s') % f1
                     raise error.StorageError(msg)
                 else:
                     undecodable.append(f1)
                     continue
             new.append((f2, value))
         return new
 
     def data_entries(
         self, matcher=None, undecodable=None
     ) -> Generator[BaseStoreEntry, None, None]:
         entries = super(encodedstore, self).data_entries(
             undecodable=undecodable
         )
         for entry in entries:
             if _match_tracked_entry(entry, matcher):
                 yield entry
 
     def join(self, f):
         return self.path + b'/' + encodefilename(f)
 
     def copylist(self):
         return [b'requires', b'00changelog.i'] + [b'store/' + f for f in _data]
 
 
 class fncache:
     # the filename used to be partially encoded
     # hence the encodedir/decodedir dance
     def __init__(self, vfs):
         self.vfs = vfs
         self._ignores = set()
         self.entries = None
         self._dirty = False
         # set of new additions to fncache
         self.addls = set()
 
     def ensureloaded(self, warn=None):
         """read the fncache file if not already read.
 
         If the file on disk is corrupted, raise. If warn is provided,
         warn and keep going instead."""
         if self.entries is None:
             self._load(warn)
 
     def _load(self, warn=None):
         '''fill the entries from the fncache file'''
         self._dirty = False
         try:
             fp = self.vfs(b'fncache', mode=b'rb')
         except IOError:
             # skip nonexistent file
             self.entries = set()
             return
 
         self.entries = set()
         chunk = b''
         for c in iter(functools.partial(fp.read, fncache_chunksize), b''):
             chunk += c
             try:
                 p = chunk.rindex(b'\n')
                 self.entries.update(decodedir(chunk[: p + 1]).splitlines())
                 chunk = chunk[p + 1 :]
             except ValueError:
                 # substring '\n' not found, maybe the entry is bigger than the
                 # chunksize, so let's keep iterating
                 pass
 
         if chunk:
             msg = _(b"fncache does not ends with a newline")
             if warn:
                 warn(msg + b'\n')
             else:
                 raise error.Abort(
                     msg,
                     hint=_(
                         b"use 'hg debugrebuildfncache' to "
                         b"rebuild the fncache"
                     ),
                 )
         self._checkentries(fp, warn)
         fp.close()
 
     def _checkentries(self, fp, warn):
         """make sure there is no empty string in entries"""
         if b'' in self.entries:
             fp.seek(0)
             for n, line in enumerate(fp):
                 if not line.rstrip(b'\n'):
                     t = _(b'invalid entry in fncache, line %d') % (n + 1)
                     if warn:
                         warn(t + b'\n')
                     else:
                         raise error.Abort(t)
 
     def write(self, tr):
         if self._dirty:
             assert self.entries is not None
             self.entries = self.entries | self.addls
             self.addls = set()
             tr.addbackup(b'fncache')
             fp = self.vfs(b'fncache', mode=b'wb', atomictemp=True)
             if self.entries:
                 fp.write(encodedir(b'\n'.join(self.entries) + b'\n'))
             fp.close()
             self._dirty = False
         if self.addls:
             # if we have just new entries, let's append them to the fncache
             tr.addbackup(b'fncache')
             fp = self.vfs(b'fncache', mode=b'ab', atomictemp=True)
             if self.addls:
                 fp.write(encodedir(b'\n'.join(self.addls) + b'\n'))
             fp.close()
             self.entries = None
             self.addls = set()
 
     def addignore(self, fn):
         self._ignores.add(fn)
 
     def add(self, fn):
         if fn in self._ignores:
             return
         if self.entries is None:
             self._load()
         if fn not in self.entries:
             self.addls.add(fn)
 
     def remove(self, fn):
         if self.entries is None:
             self._load()
         if fn in self.addls:
             self.addls.remove(fn)
             return
         try:
             self.entries.remove(fn)
             self._dirty = True
         except KeyError:
             pass
 
     def __contains__(self, fn):
         if fn in self.addls:
             return True
         if self.entries is None:
             self._load()
         return fn in self.entries
 
     def __iter__(self):
         if self.entries is None:
             self._load()
         return iter(self.entries | self.addls)
 
 
 class _fncachevfs(vfsmod.proxyvfs):
     def __init__(self, vfs, fnc, encode):
         vfsmod.proxyvfs.__init__(self, vfs)
         self.fncache = fnc
         self.encode = encode
 
     def __call__(self, path, mode=b'r', *args, **kw):
         encoded = self.encode(path)
         if (
             mode not in (b'r', b'rb')
             and (path.startswith(b'data/') or path.startswith(b'meta/'))
             and is_revlog_file(path)
         ):
             # do not trigger a fncache load when adding a file that already is
             # known to exist.
             notload = self.fncache.entries is None and (
                 # if the file has size zero, it should be considered as missing.
                 # Such zero-size files are the result of truncation when a
                 # transaction is aborted.
                 self.vfs.exists(encoded)
                 and self.vfs.stat(encoded).st_size
             )
             if not notload:
                 self.fncache.add(path)
         return self.vfs(encoded, mode, *args, **kw)
 
     def join(self, path: Optional[bytes], *insidef: bytes) -> bytes:
         insidef = (self.encode(f) for f in insidef)
 
         if path:
             return self.vfs.join(self.encode(path), *insidef)
         else:
             return self.vfs.join(path, *insidef)
 
     def register_file(self, path):
         """generic hook point to lets fncache steer its stew"""
         if path.startswith(b'data/') or path.startswith(b'meta/'):
             self.fncache.add(path)
 
 
 class fncachestore(basicstore):
     def __init__(self, path, vfstype, dotencode):
         if dotencode:
             encode = _pathencode
         else:
             encode = _plainhybridencode
         self.encode = encode
         vfs = vfstype(path + b'/store')
         self.path = vfs.base
         self.pathsep = self.path + b'/'
         self.createmode = _calcmode(vfs)
         vfs.createmode = self.createmode
         self.rawvfs = vfs
         fnc = fncache(vfs)
         self.fncache = fnc
         self.vfs = _fncachevfs(vfs, fnc, encode)
         self.opener = self.vfs
 
     def join(self, f):
         return self.pathsep + self.encode(f)
 
     def getsize(self, path):
         return self.rawvfs.stat(path).st_size
 
     def data_entries(
         self, matcher=None, undecodable=None
     ) -> Generator[BaseStoreEntry, None, None]:
         # Note: all files in fncache should be revlog related, However the
         # fncache might contains such file added by previous version of
         # Mercurial.
         files = ((f, None) for f in self.fncache if is_revlog_file(f))
         by_revlog = _gather_revlog(files)
         for revlog, details in by_revlog:
             if revlog.startswith(b'data/'):
                 rl_type = KIND_FILELOG
                 revlog_target_id = revlog.split(b'/', 1)[1]
             elif revlog.startswith(b'meta/'):
                 rl_type = KIND_MANIFESTLOG
                 # drop the initial directory and the `00manifest` file part
                 tmp = revlog.split(b'/', 1)[1]
                 revlog_target_id = tmp.rsplit(b'/', 1)[0] + b'/'
             else:
                 # unreachable
                 assert False, revlog
             entry = RevlogStoreEntry(
                 path_prefix=revlog,
                 revlog_type=rl_type,
                 target_id=revlog_target_id,
                 details=details,
             )
             if _match_tracked_entry(entry, matcher):
                 yield entry
 
     def copylist(self):
         d = (
             b'bookmarks',
             b'narrowspec',
             b'data',
             b'meta',
             b'dh',
             b'fncache',
             b'phaseroots',
             b'obsstore',
             b'00manifest.d',
             b'00manifest.i',
             b'00changelog.d',
             b'00changelog.i',
             b'requires',
         )
         return [b'requires', b'00changelog.i'] + [b'store/' + f for f in d]
 
     def write(self, tr):
         self.fncache.write(tr)
 
     def invalidatecaches(self):
         self.fncache.entries = None
         self.fncache.addls = set()
 
     def markremoved(self, fn):
         self.fncache.remove(fn)
 
     def _exists(self, f):
         ef = self.encode(f)
         try:
             self.getsize(ef)
             return True
         except FileNotFoundError:
             return False
 
     def __contains__(self, path):
         '''Checks if the store contains path'''
         path = b"/".join((b"data", path))
         # check for files (exact match)
         e = path + b'.i'
         if e in self.fncache and self._exists(e):
             return True
         # now check for directories (prefix match)
         if not path.endswith(b'/'):
             path += b'/'
         for e in self.fncache:
             if e.startswith(path) and self._exists(e):
                 return True
         return False