revlogv2: use a unique filename for index

Having a unique index will allow for ambiguity less rewriting of revlog content,
something useful to clarify handling of some operation like censoring or
stripping.

Differential Revision: https://phab.mercurial-scm.org/D10771

mercurial/revlogutils/docket.py

@@ -1,235 +1,249 b''
 # docket - code related to revlog "docket"
 #
 # Copyright 2021 Pierre-Yves David <[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.
 
 ### Revlog docket file
 #
 # The revlog is stored on disk using multiple files:
 #
 # * a small docket file, containing metadata and a pointer,
 #
 # * an index file, containing fixed width information about revisions,
 #
 # * a data file, containing variable width data for these revisions,
 
 from __future__ import absolute_import
 
 import errno
 import os
 import random
 import struct
 
 from .. import (
     encoding,
     error,
     node,
     pycompat,
     util,
 )
 
 from . import (
     constants,
 )
 
 
 def make_uid(id_size=8):
     """return a new unique identifier.
 
     The identifier is random and composed of ascii characters."""
     # size we "hex" the result we need half the number of bits to have a final
     # uuid of size ID_SIZE
     return node.hex(os.urandom(id_size // 2))
 
 
 # some special test logic to avoid anoying random output in the test
 stable_docket_file = encoding.environ.get(b'HGTEST_UUIDFILE')
 
 if stable_docket_file:
 
     def make_uid(id_size=8):
         try:
             with open(stable_docket_file, mode='rb') as f:
                 seed = f.read().strip()
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
             seed = b'04'  # chosen by a fair dice roll. garanteed to be random
         if pycompat.ispy3:
             iter_seed = iter(seed)
         else:
             iter_seed = (ord(c) for c in seed)
         # some basic circular sum hashing on 64 bits
         int_seed = 0
         low_mask = int('1' * 35, 2)
         for i in iter_seed:
             high_part = int_seed >> 35
             low_part = (int_seed & low_mask) << 28
             int_seed = high_part + low_part + i
         r = random.Random()
         if pycompat.ispy3:
             r.seed(int_seed, version=1)
         else:
             r.seed(int_seed)
         # once we drop python 3.8 support we can simply use r.randbytes
         raw = r.getrandbits(id_size * 4)
         assert id_size == 8
         p = struct.pack('>L', raw)
         new = node.hex(p)
         with open(stable_docket_file, 'wb') as f:
             f.write(new)
         return new
 
 
 # Docket format
 #
 # * 4 bytes: revlog version
 #          |   This is mandatory as docket must be compatible with the previous
 #          |   revlog index header.
+# * 1 bytes: size of index uuid
 # * 8 bytes: size of index-data
 # * 8 bytes: pending size of index-data
 # * 8 bytes: size of data
 # * 8 bytes: pending size of data
 # * 1 bytes: default compression header
-S_HEADER = struct.Struct(constants.INDEX_HEADER.format + 'LLLLc')
+S_HEADER = struct.Struct(constants.INDEX_HEADER.format + 'BLLLLc')
 
 
 class RevlogDocket(object):
     """metadata associated with revlog"""
 
     def __init__(
         self,
         revlog,
         use_pending=False,
         version_header=None,
+        index_uuid=None,
         index_end=0,
         pending_index_end=0,
         data_end=0,
         pending_data_end=0,
         default_compression_header=None,
     ):
         self._version_header = version_header
         self._read_only = bool(use_pending)
         self._dirty = False
         self._radix = revlog.radix
         self._path = revlog._docket_file
         self._opener = revlog.opener
+        self._index_uuid = index_uuid
         # thes asserts should be True as long as we have a single index filename
         assert index_end <= pending_index_end
         assert data_end <= pending_data_end
         self._initial_index_end = index_end
         self._pending_index_end = pending_index_end
         self._initial_data_end = data_end
         self._pending_data_end = pending_data_end
         if use_pending:
             self._index_end = self._pending_index_end
             self._data_end = self._pending_data_end
         else:
             self._index_end = self._initial_index_end
             self._data_end = self._initial_data_end
         self.default_compression_header = default_compression_header
 
     def index_filepath(self):
         """file path to the current index file associated to this docket"""
         # very simplistic version at first
-        return b"%s.idx" % self._radix
+        if self._index_uuid is None:
+            self._index_uuid = make_uid()
+        return b"%s-%s.idx" % (self._radix, self._index_uuid)
 
     @property
     def index_end(self):
         return self._index_end
 
     @index_end.setter
     def index_end(self, new_size):
         if new_size != self._index_end:
             self._index_end = new_size
             self._dirty = True
 
     @property
     def data_end(self):
         return self._data_end
 
     @data_end.setter
     def data_end(self, new_size):
         if new_size != self._data_end:
             self._data_end = new_size
             self._dirty = True
 
     def write(self, transaction, pending=False, stripping=False):
         """write the modification of disk if any
 
         This make the new content visible to all process"""
         if not self._dirty:
             return False
         else:
             if self._read_only:
                 msg = b'writing read-only docket: %s'
                 msg %= self._path
                 raise error.ProgrammingError(msg)
             if not stripping:
                 # XXX we could, leverage the docket while stripping. However it
                 # is not powerfull enough at the time of this comment
                 transaction.addbackup(self._path, location=b'store')
             with self._opener(self._path, mode=b'w', atomictemp=True) as f:
                 f.write(self._serialize(pending=pending))
             # if pending we still need to the write final data eventually
             self._dirty = pending
             return True
 
     def _serialize(self, pending=False):
         if pending:
             official_index_end = self._initial_index_end
             official_data_end = self._initial_data_end
         else:
             official_index_end = self._index_end
             official_data_end = self._data_end
 
         # this assert should be True as long as we have a single index filename
         assert official_data_end <= self._data_end
         data = (
             self._version_header,
+            len(self._index_uuid),
             official_index_end,
             self._index_end,
             official_data_end,
             self._data_end,
             self.default_compression_header,
         )
-        return S_HEADER.pack(*data)
+        s = []
+        s.append(S_HEADER.pack(*data))
+        s.append(self._index_uuid)
+        return b''.join(s)
 
 
 def default_docket(revlog, version_header):
     """given a revlog version a new docket object for the given revlog"""
     rl_version = version_header & 0xFFFF
     if rl_version not in (constants.REVLOGV2, constants.CHANGELOGV2):
         return None
     comp = util.compengines[revlog._compengine].revlogheader()
     docket = RevlogDocket(
         revlog,
         version_header=version_header,
         default_compression_header=comp,
     )
     docket._dirty = True
     return docket
 
 
 def parse_docket(revlog, data, use_pending=False):
     """given some docket data return a docket object for the given revlog"""
     header = S_HEADER.unpack(data[: S_HEADER.size])
+    offset = S_HEADER.size
     version_header = header[0]
-    index_size = header[1]
-    pending_index_size = header[2]
-    data_size = header[3]
-    pending_data_size = header[4]
-    default_compression_header = header[5]
+    index_uuid_size = header[1]
+    index_uuid = data[offset : offset + index_uuid_size]
+    offset += index_uuid_size
+    index_size = header[2]
+    pending_index_size = header[3]
+    data_size = header[4]
+    pending_data_size = header[5]
+    default_compression_header = header[6]
     docket = RevlogDocket(
         revlog,
         use_pending=use_pending,
         version_header=version_header,
+        index_uuid=index_uuid,
         index_end=index_size,
         pending_index_end=pending_index_size,
         data_end=data_size,
         pending_data_end=pending_data_size,
         default_compression_header=default_compression_header,
     )
     return docket

tests/test-revlog-v2.t

@@ -1,69 +1,85 b''
 #require reporevlogstore
 
 A repo with unknown revlogv2 requirement string cannot be opened
 
   $ hg init invalidreq
   $ cd invalidreq
   $ echo exp-revlogv2.unknown >> .hg/requires
   $ hg log
   abort: repository requires features unknown to this Mercurial: exp-revlogv2.unknown
   (see https://mercurial-scm.org/wiki/MissingRequirement for more information)
   [255]
   $ cd ..
 
 Can create and open repo with revlog v2 requirement
 
   $ cat >> $HGRCPATH << EOF
   > [experimental]
   > revlogv2 = enable-unstable-format-and-corrupt-my-data
   > EOF
 
   $ hg init empty-repo
   $ cd empty-repo
   $ cat .hg/requires
   dotencode
   exp-dirstate-v2 (dirstate-v2 !)
   exp-revlogv2.2
   fncache
   generaldelta
   persistent-nodemap (rust !)
   revlog-compression-zstd (zstd !)
   sparserevlog
   store
 
   $ hg log
 
 Unknown flags to revlog are rejected
 
   >>> with open('.hg/store/00changelog.i', 'wb') as fh:
   ...     fh.write(b'\xff\x00\xde\xad') and None
 
   $ hg log
   abort: unknown flags (0xff00) in version 57005 revlog 00changelog
   [50]
 
   $ cd ..
 
 Writing a simple revlog v2 works
 
   $ hg init simple
   $ cd simple
   $ touch foo
   $ hg -q commit -A -m initial
 
   $ hg log
   changeset:   0:96ee1d7354c4
   tag:         tip
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     initial
   
 Header written as expected
 
   $ f --hexdump --bytes 4 .hg/store/00changelog.i
   .hg/store/00changelog.i:
   0000: 00 00 de ad                                     |....|
 
   $ f --hexdump --bytes 4 .hg/store/data/foo.i
   .hg/store/data/foo.i:
   0000: 00 00 de ad                                     |....|
+
+The expected files are generated
+--------------------------------
+
+We should have have:
+- a docket
+- a index file with a unique name
+- a data file
+
+  $ ls .hg/store/00changelog* .hg/store/00manifest*
+  .hg/store/00changelog-b870a51b.idx
+  .hg/store/00changelog.d
+  .hg/store/00changelog.i
+  .hg/store/00manifest-88698448.idx
+  .hg/store/00manifest.d
+  .hg/store/00manifest.i