mercurial/testing/storage.py
author Pierre-Yves David <pierre-yves.david@octobus.net>
Mon, 03 May 2021 12:35:25 +0200
changeset 47252 2219853a1503
parent 47055 d55b71393907
child 48966 6000f5b25c9b
permissions -rw-r--r--
revlogv2: track pending write in the docket and expose it to hooks The docket is now able to write pending data. We could have used a distinct intermediate files, however keeping everything in the same file will make it simpler to keep track of the various involved files if necessary. However it might prove more complicated for streaming clone. This will be dealt with later. Note that we lifted the stderr redirection in the test since we no longer suffer from "unkown working directory parent" message. Differential Revision: https://phab.mercurial-scm.org/D10631

# storage.py - Testing of storage primitives.
#
# Copyright 2018 Gregory Szorc <gregory.szorc@gmail.com>
#
# 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 absolute_import

import unittest

from ..node import (
    hex,
    nullrev,
)
from ..pycompat import getattr
from .. import (
    error,
    mdiff,
)
from ..interfaces import repository
from ..utils import storageutil


class basetestcase(unittest.TestCase):
    if not getattr(unittest.TestCase, 'assertRaisesRegex', False):
        assertRaisesRegex = (  # camelcase-required
            unittest.TestCase.assertRaisesRegexp
        )


class ifileindextests(basetestcase):
    """Generic tests for the ifileindex interface.

    All file storage backends for index data should conform to the tests in this
    class.

    Use ``makeifileindextests()`` to create an instance of this type.
    """

    def testempty(self):
        f = self._makefilefn()
        self.assertEqual(len(f), 0, b'new file store has 0 length by default')
        self.assertEqual(list(f), [], b'iter yields nothing by default')

        gen = iter(f)
        with self.assertRaises(StopIteration):
            next(gen)

        self.assertFalse(f.hasnode(None))
        self.assertFalse(f.hasnode(0))
        self.assertFalse(f.hasnode(nullrev))
        self.assertFalse(f.hasnode(f.nullid))
        self.assertFalse(f.hasnode(b'0'))
        self.assertFalse(f.hasnode(b'a' * 20))

        # revs() should evaluate to an empty list.
        self.assertEqual(list(f.revs()), [])

        revs = iter(f.revs())
        with self.assertRaises(StopIteration):
            next(revs)

        self.assertEqual(list(f.revs(start=20)), [])

        # parents() and parentrevs() work with f.nullid/nullrev.
        self.assertEqual(f.parents(f.nullid), (f.nullid, f.nullid))
        self.assertEqual(f.parentrevs(nullrev), (nullrev, nullrev))

        with self.assertRaises(error.LookupError):
            f.parents(b'\x01' * 20)

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(IndexError):
                f.parentrevs(i)

        # f.nullid/nullrev lookup always works.
        self.assertEqual(f.rev(f.nullid), nullrev)
        self.assertEqual(f.node(nullrev), f.nullid)

        with self.assertRaises(error.LookupError):
            f.rev(b'\x01' * 20)

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(IndexError):
                f.node(i)

        self.assertEqual(f.lookup(f.nullid), f.nullid)
        self.assertEqual(f.lookup(nullrev), f.nullid)
        self.assertEqual(f.lookup(hex(f.nullid)), f.nullid)
        self.assertEqual(f.lookup(b'%d' % nullrev), f.nullid)

        with self.assertRaises(error.LookupError):
            f.lookup(b'badvalue')

        with self.assertRaises(error.LookupError):
            f.lookup(hex(f.nullid)[0:12])

        with self.assertRaises(error.LookupError):
            f.lookup(b'-2')

        with self.assertRaises(error.LookupError):
            f.lookup(b'0')

        with self.assertRaises(error.LookupError):
            f.lookup(b'1')

        with self.assertRaises(error.LookupError):
            f.lookup(b'11111111111111111111111111111111111111')

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(LookupError):
                f.lookup(i)

        self.assertEqual(f.linkrev(nullrev), nullrev)

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(IndexError):
                f.linkrev(i)

        self.assertFalse(f.iscensored(nullrev))

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(IndexError):
                f.iscensored(i)

        self.assertEqual(list(f.commonancestorsheads(f.nullid, f.nullid)), [])

        with self.assertRaises(ValueError):
            self.assertEqual(list(f.descendants([])), [])

        self.assertEqual(list(f.descendants([nullrev])), [])

        self.assertEqual(f.heads(), [f.nullid])
        self.assertEqual(f.heads(f.nullid), [f.nullid])
        self.assertEqual(f.heads(None, [f.nullid]), [f.nullid])
        self.assertEqual(f.heads(f.nullid, [f.nullid]), [f.nullid])

        self.assertEqual(f.children(f.nullid), [])

        with self.assertRaises(error.LookupError):
            f.children(b'\x01' * 20)

    def testsinglerevision(self):
        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node = f.add(b'initial', None, tr, 0, f.nullid, f.nullid)

        self.assertEqual(len(f), 1)
        self.assertEqual(list(f), [0])

        gen = iter(f)
        self.assertEqual(next(gen), 0)

        with self.assertRaises(StopIteration):
            next(gen)

        self.assertTrue(f.hasnode(node))
        self.assertFalse(f.hasnode(hex(node)))
        self.assertFalse(f.hasnode(nullrev))
        self.assertFalse(f.hasnode(f.nullid))
        self.assertFalse(f.hasnode(node[0:12]))
        self.assertFalse(f.hasnode(hex(node)[0:20]))

        self.assertEqual(list(f.revs()), [0])
        self.assertEqual(list(f.revs(start=1)), [])
        self.assertEqual(list(f.revs(start=0)), [0])
        self.assertEqual(list(f.revs(stop=0)), [0])
        self.assertEqual(list(f.revs(stop=1)), [0])
        self.assertEqual(list(f.revs(1, 1)), [])
        # TODO buggy
        self.assertEqual(list(f.revs(1, 0)), [1, 0])
        self.assertEqual(list(f.revs(2, 0)), [2, 1, 0])

        self.assertEqual(f.parents(node), (f.nullid, f.nullid))
        self.assertEqual(f.parentrevs(0), (nullrev, nullrev))

        with self.assertRaises(error.LookupError):
            f.parents(b'\x01' * 20)

        with self.assertRaises(IndexError):
            f.parentrevs(1)

        self.assertEqual(f.rev(node), 0)

        with self.assertRaises(error.LookupError):
            f.rev(b'\x01' * 20)

        self.assertEqual(f.node(0), node)

        with self.assertRaises(IndexError):
            f.node(1)

        self.assertEqual(f.lookup(node), node)
        self.assertEqual(f.lookup(0), node)
        self.assertEqual(f.lookup(-1), f.nullid)
        self.assertEqual(f.lookup(b'0'), node)
        self.assertEqual(f.lookup(hex(node)), node)

        with self.assertRaises(error.LookupError):
            f.lookup(hex(node)[0:12])

        with self.assertRaises(error.LookupError):
            f.lookup(-2)

        with self.assertRaises(error.LookupError):
            f.lookup(b'-2')

        with self.assertRaises(error.LookupError):
            f.lookup(1)

        with self.assertRaises(error.LookupError):
            f.lookup(b'1')

        self.assertEqual(f.linkrev(0), 0)

        with self.assertRaises(IndexError):
            f.linkrev(1)

        self.assertFalse(f.iscensored(0))

        with self.assertRaises(IndexError):
            f.iscensored(1)

        self.assertEqual(list(f.descendants([0])), [])

        self.assertEqual(f.heads(), [node])
        self.assertEqual(f.heads(node), [node])
        self.assertEqual(f.heads(stop=[node]), [node])

        with self.assertRaises(error.LookupError):
            f.heads(stop=[b'\x01' * 20])

        self.assertEqual(f.children(node), [])

    def testmultiplerevisions(self):
        fulltext0 = b'x' * 1024
        fulltext1 = fulltext0 + b'y'
        fulltext2 = b'y' + fulltext0 + b'z'

        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(fulltext1, None, tr, 1, node0, f.nullid)
            node2 = f.add(fulltext2, None, tr, 3, node1, f.nullid)

        self.assertEqual(len(f), 3)
        self.assertEqual(list(f), [0, 1, 2])

        gen = iter(f)
        self.assertEqual(next(gen), 0)
        self.assertEqual(next(gen), 1)
        self.assertEqual(next(gen), 2)

        with self.assertRaises(StopIteration):
            next(gen)

        self.assertEqual(list(f.revs()), [0, 1, 2])
        self.assertEqual(list(f.revs(0)), [0, 1, 2])
        self.assertEqual(list(f.revs(1)), [1, 2])
        self.assertEqual(list(f.revs(2)), [2])
        self.assertEqual(list(f.revs(3)), [])
        self.assertEqual(list(f.revs(stop=1)), [0, 1])
        self.assertEqual(list(f.revs(stop=2)), [0, 1, 2])
        self.assertEqual(list(f.revs(stop=3)), [0, 1, 2])
        self.assertEqual(list(f.revs(2, 0)), [2, 1, 0])
        self.assertEqual(list(f.revs(2, 1)), [2, 1])
        # TODO this is wrong
        self.assertEqual(list(f.revs(3, 2)), [3, 2])

        self.assertEqual(f.parents(node0), (f.nullid, f.nullid))
        self.assertEqual(f.parents(node1), (node0, f.nullid))
        self.assertEqual(f.parents(node2), (node1, f.nullid))

        self.assertEqual(f.parentrevs(0), (nullrev, nullrev))
        self.assertEqual(f.parentrevs(1), (0, nullrev))
        self.assertEqual(f.parentrevs(2), (1, nullrev))

        self.assertEqual(f.rev(node0), 0)
        self.assertEqual(f.rev(node1), 1)
        self.assertEqual(f.rev(node2), 2)

        with self.assertRaises(error.LookupError):
            f.rev(b'\x01' * 20)

        self.assertEqual(f.node(0), node0)
        self.assertEqual(f.node(1), node1)
        self.assertEqual(f.node(2), node2)

        with self.assertRaises(IndexError):
            f.node(3)

        self.assertEqual(f.lookup(node0), node0)
        self.assertEqual(f.lookup(0), node0)
        self.assertEqual(f.lookup(b'0'), node0)
        self.assertEqual(f.lookup(hex(node0)), node0)

        self.assertEqual(f.lookup(node1), node1)
        self.assertEqual(f.lookup(1), node1)
        self.assertEqual(f.lookup(b'1'), node1)
        self.assertEqual(f.lookup(hex(node1)), node1)

        self.assertEqual(f.linkrev(0), 0)
        self.assertEqual(f.linkrev(1), 1)
        self.assertEqual(f.linkrev(2), 3)

        with self.assertRaises(IndexError):
            f.linkrev(3)

        self.assertFalse(f.iscensored(0))
        self.assertFalse(f.iscensored(1))
        self.assertFalse(f.iscensored(2))

        with self.assertRaises(IndexError):
            f.iscensored(3)

        self.assertEqual(f.commonancestorsheads(node1, f.nullid), [])
        self.assertEqual(f.commonancestorsheads(node1, node0), [node0])
        self.assertEqual(f.commonancestorsheads(node1, node1), [node1])
        self.assertEqual(f.commonancestorsheads(node0, node1), [node0])
        self.assertEqual(f.commonancestorsheads(node1, node2), [node1])
        self.assertEqual(f.commonancestorsheads(node2, node1), [node1])

        self.assertEqual(list(f.descendants([0])), [1, 2])
        self.assertEqual(list(f.descendants([1])), [2])
        self.assertEqual(list(f.descendants([0, 1])), [1, 2])

        self.assertEqual(f.heads(), [node2])
        self.assertEqual(f.heads(node0), [node2])
        self.assertEqual(f.heads(node1), [node2])
        self.assertEqual(f.heads(node2), [node2])

        # TODO this behavior seems wonky. Is it correct? If so, the
        # docstring for heads() should be updated to reflect desired
        # behavior.
        self.assertEqual(f.heads(stop=[node1]), [node1, node2])
        self.assertEqual(f.heads(stop=[node0]), [node0, node2])
        self.assertEqual(f.heads(stop=[node1, node2]), [node1, node2])

        with self.assertRaises(error.LookupError):
            f.heads(stop=[b'\x01' * 20])

        self.assertEqual(f.children(node0), [node1])
        self.assertEqual(f.children(node1), [node2])
        self.assertEqual(f.children(node2), [])

    def testmultipleheads(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            node0 = f.add(b'0', None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(b'1', None, tr, 1, node0, f.nullid)
            node2 = f.add(b'2', None, tr, 2, node1, f.nullid)
            node3 = f.add(b'3', None, tr, 3, node0, f.nullid)
            node4 = f.add(b'4', None, tr, 4, node3, f.nullid)
            node5 = f.add(b'5', None, tr, 5, node0, f.nullid)

        self.assertEqual(len(f), 6)

        self.assertEqual(list(f.descendants([0])), [1, 2, 3, 4, 5])
        self.assertEqual(list(f.descendants([1])), [2])
        self.assertEqual(list(f.descendants([2])), [])
        self.assertEqual(list(f.descendants([3])), [4])
        self.assertEqual(list(f.descendants([0, 1])), [1, 2, 3, 4, 5])
        self.assertEqual(list(f.descendants([1, 3])), [2, 4])

        self.assertEqual(f.heads(), [node2, node4, node5])
        self.assertEqual(f.heads(node0), [node2, node4, node5])
        self.assertEqual(f.heads(node1), [node2])
        self.assertEqual(f.heads(node2), [node2])
        self.assertEqual(f.heads(node3), [node4])
        self.assertEqual(f.heads(node4), [node4])
        self.assertEqual(f.heads(node5), [node5])

        # TODO this seems wrong.
        self.assertEqual(f.heads(stop=[node0]), [node0, node2, node4, node5])
        self.assertEqual(f.heads(stop=[node1]), [node1, node2, node4, node5])

        self.assertEqual(f.children(node0), [node1, node3, node5])
        self.assertEqual(f.children(node1), [node2])
        self.assertEqual(f.children(node2), [])
        self.assertEqual(f.children(node3), [node4])
        self.assertEqual(f.children(node4), [])
        self.assertEqual(f.children(node5), [])


class ifiledatatests(basetestcase):
    """Generic tests for the ifiledata interface.

    All file storage backends for data should conform to the tests in this
    class.

    Use ``makeifiledatatests()`` to create an instance of this type.
    """

    def testempty(self):
        f = self._makefilefn()

        self.assertEqual(f.storageinfo(), {})
        self.assertEqual(
            f.storageinfo(revisionscount=True, trackedsize=True),
            {b'revisionscount': 0, b'trackedsize': 0},
        )

        self.assertEqual(f.size(nullrev), 0)

        for i in range(-5, 5):
            if i == nullrev:
                continue

            with self.assertRaises(IndexError):
                f.size(i)

        self.assertEqual(f.revision(f.nullid), b'')
        self.assertEqual(f.rawdata(f.nullid), b'')

        with self.assertRaises(error.LookupError):
            f.revision(b'\x01' * 20)

        self.assertEqual(f.read(f.nullid), b'')

        with self.assertRaises(error.LookupError):
            f.read(b'\x01' * 20)

        self.assertFalse(f.renamed(f.nullid))

        with self.assertRaises(error.LookupError):
            f.read(b'\x01' * 20)

        self.assertTrue(f.cmp(f.nullid, b''))
        self.assertTrue(f.cmp(f.nullid, b'foo'))

        with self.assertRaises(error.LookupError):
            f.cmp(b'\x01' * 20, b'irrelevant')

        # Emitting empty list is an empty generator.
        gen = f.emitrevisions([])
        with self.assertRaises(StopIteration):
            next(gen)

        # Emitting null node yields nothing.
        gen = f.emitrevisions([f.nullid])
        with self.assertRaises(StopIteration):
            next(gen)

        # Requesting unknown node fails.
        with self.assertRaises(error.LookupError):
            list(f.emitrevisions([b'\x01' * 20]))

    def testsinglerevision(self):
        fulltext = b'initial'

        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node = f.add(fulltext, None, tr, 0, f.nullid, f.nullid)

        self.assertEqual(f.storageinfo(), {})
        self.assertEqual(
            f.storageinfo(revisionscount=True, trackedsize=True),
            {b'revisionscount': 1, b'trackedsize': len(fulltext)},
        )

        self.assertEqual(f.size(0), len(fulltext))

        with self.assertRaises(IndexError):
            f.size(1)

        self.assertEqual(f.revision(node), fulltext)
        self.assertEqual(f.rawdata(node), fulltext)

        self.assertEqual(f.read(node), fulltext)

        self.assertFalse(f.renamed(node))

        self.assertFalse(f.cmp(node, fulltext))
        self.assertTrue(f.cmp(node, fulltext + b'extra'))

        # Emitting a single revision works.
        gen = f.emitrevisions([node])
        rev = next(gen)

        self.assertEqual(rev.node, node)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertIsNone(rev.delta)

        with self.assertRaises(StopIteration):
            next(gen)

        # Requesting revision data works.
        gen = f.emitrevisions([node], revisiondata=True)
        rev = next(gen)

        self.assertEqual(rev.node, node)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertEqual(rev.revision, fulltext)
        self.assertIsNone(rev.delta)

        with self.assertRaises(StopIteration):
            next(gen)

        # Emitting an unknown node after a known revision results in error.
        with self.assertRaises(error.LookupError):
            list(f.emitrevisions([node, b'\x01' * 20]))

    def testmultiplerevisions(self):
        fulltext0 = b'x' * 1024
        fulltext1 = fulltext0 + b'y'
        fulltext2 = b'y' + fulltext0 + b'z'

        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(fulltext1, None, tr, 1, node0, f.nullid)
            node2 = f.add(fulltext2, None, tr, 3, node1, f.nullid)

        self.assertEqual(f.storageinfo(), {})
        self.assertEqual(
            f.storageinfo(revisionscount=True, trackedsize=True),
            {
                b'revisionscount': 3,
                b'trackedsize': len(fulltext0)
                + len(fulltext1)
                + len(fulltext2),
            },
        )

        self.assertEqual(f.size(0), len(fulltext0))
        self.assertEqual(f.size(1), len(fulltext1))
        self.assertEqual(f.size(2), len(fulltext2))

        with self.assertRaises(IndexError):
            f.size(3)

        self.assertEqual(f.revision(node0), fulltext0)
        self.assertEqual(f.rawdata(node0), fulltext0)
        self.assertEqual(f.revision(node1), fulltext1)
        self.assertEqual(f.rawdata(node1), fulltext1)
        self.assertEqual(f.revision(node2), fulltext2)
        self.assertEqual(f.rawdata(node2), fulltext2)

        with self.assertRaises(error.LookupError):
            f.revision(b'\x01' * 20)

        self.assertEqual(f.read(node0), fulltext0)
        self.assertEqual(f.read(node1), fulltext1)
        self.assertEqual(f.read(node2), fulltext2)

        with self.assertRaises(error.LookupError):
            f.read(b'\x01' * 20)

        self.assertFalse(f.renamed(node0))
        self.assertFalse(f.renamed(node1))
        self.assertFalse(f.renamed(node2))

        with self.assertRaises(error.LookupError):
            f.renamed(b'\x01' * 20)

        self.assertFalse(f.cmp(node0, fulltext0))
        self.assertFalse(f.cmp(node1, fulltext1))
        self.assertFalse(f.cmp(node2, fulltext2))

        self.assertTrue(f.cmp(node1, fulltext0))
        self.assertTrue(f.cmp(node2, fulltext1))

        with self.assertRaises(error.LookupError):
            f.cmp(b'\x01' * 20, b'irrelevant')

        # Nodes should be emitted in order.
        gen = f.emitrevisions([node0, node1, node2], revisiondata=True)

        rev = next(gen)

        self.assertEqual(rev.node, node0)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertEqual(rev.revision, fulltext0)
        self.assertIsNone(rev.delta)

        rev = next(gen)

        self.assertEqual(rev.node, node1)
        self.assertEqual(rev.p1node, node0)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, node0)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
        )

        rev = next(gen)

        self.assertEqual(rev.node, node2)
        self.assertEqual(rev.p1node, node1)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, node1)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
        )

        with self.assertRaises(StopIteration):
            next(gen)

        # Request not in DAG order is reordered to be in DAG order.
        gen = f.emitrevisions([node2, node1, node0], revisiondata=True)

        rev = next(gen)

        self.assertEqual(rev.node, node0)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertEqual(rev.revision, fulltext0)
        self.assertIsNone(rev.delta)

        rev = next(gen)

        self.assertEqual(rev.node, node1)
        self.assertEqual(rev.p1node, node0)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, node0)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
        )

        rev = next(gen)

        self.assertEqual(rev.node, node2)
        self.assertEqual(rev.p1node, node1)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertIsNone(rev.linknode)
        self.assertEqual(rev.basenode, node1)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
        )

        with self.assertRaises(StopIteration):
            next(gen)

        # Unrecognized nodesorder value raises ProgrammingError.
        with self.assertRaises(error.ProgrammingError):
            list(f.emitrevisions([], nodesorder=b'bad'))

        # nodesorder=storage is recognized. But we can't test it thoroughly
        # because behavior is storage-dependent.
        res = list(
            f.emitrevisions([node2, node1, node0], nodesorder=b'storage')
        )
        self.assertEqual(len(res), 3)
        self.assertEqual({o.node for o in res}, {node0, node1, node2})

        # nodesorder=nodes forces the order.
        gen = f.emitrevisions(
            [node2, node0], nodesorder=b'nodes', revisiondata=True
        )

        rev = next(gen)
        self.assertEqual(rev.node, node2)
        self.assertEqual(rev.p1node, node1)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertEqual(rev.revision, fulltext2)
        self.assertIsNone(rev.delta)

        rev = next(gen)
        self.assertEqual(rev.node, node0)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        # Delta behavior is storage dependent, so we can't easily test it.

        with self.assertRaises(StopIteration):
            next(gen)

        # assumehaveparentrevisions=False (the default) won't send a delta for
        # the first revision.
        gen = f.emitrevisions({node2, node1}, revisiondata=True)

        rev = next(gen)
        self.assertEqual(rev.node, node1)
        self.assertEqual(rev.p1node, node0)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertEqual(rev.revision, fulltext1)
        self.assertIsNone(rev.delta)

        rev = next(gen)
        self.assertEqual(rev.node, node2)
        self.assertEqual(rev.p1node, node1)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, node1)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
        )

        with self.assertRaises(StopIteration):
            next(gen)

        # assumehaveparentrevisions=True allows delta against initial revision.
        gen = f.emitrevisions(
            [node2, node1], revisiondata=True, assumehaveparentrevisions=True
        )

        rev = next(gen)
        self.assertEqual(rev.node, node1)
        self.assertEqual(rev.p1node, node0)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, node0)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
        )

        # forceprevious=True forces a delta against the previous revision.
        # Special case for initial revision.
        gen = f.emitrevisions(
            [node0], revisiondata=True, deltamode=repository.CG_DELTAMODE_PREV
        )

        rev = next(gen)
        self.assertEqual(rev.node, node0)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0,
        )

        with self.assertRaises(StopIteration):
            next(gen)

        gen = f.emitrevisions(
            [node0, node2],
            revisiondata=True,
            deltamode=repository.CG_DELTAMODE_PREV,
        )

        rev = next(gen)
        self.assertEqual(rev.node, node0)
        self.assertEqual(rev.p1node, f.nullid)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, f.nullid)
        self.assertIsNone(rev.baserevisionsize)
        self.assertIsNone(rev.revision)
        self.assertEqual(
            rev.delta,
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0,
        )

        rev = next(gen)
        self.assertEqual(rev.node, node2)
        self.assertEqual(rev.p1node, node1)
        self.assertEqual(rev.p2node, f.nullid)
        self.assertEqual(rev.basenode, node0)

        with self.assertRaises(StopIteration):
            next(gen)

    def testrenamed(self):
        fulltext0 = b'foo'
        fulltext1 = b'bar'
        fulltext2 = b'baz'

        meta1 = {
            b'copy': b'source0',
            b'copyrev': b'a' * 40,
        }

        meta2 = {
            b'copy': b'source1',
            b'copyrev': b'b' * 40,
        }

        stored1 = b''.join(
            [
                b'\x01\ncopy: source0\n',
                b'copyrev: aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\n\x01\n',
                fulltext1,
            ]
        )

        stored2 = b''.join(
            [
                b'\x01\ncopy: source1\n',
                b'copyrev: bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\n\x01\n',
                fulltext2,
            ]
        )

        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(fulltext1, meta1, tr, 1, node0, f.nullid)
            node2 = f.add(fulltext2, meta2, tr, 2, f.nullid, f.nullid)

        # Metadata header isn't recognized when parent isn't f.nullid.
        self.assertEqual(f.size(1), len(stored1))
        self.assertEqual(f.size(2), len(fulltext2))

        self.assertEqual(f.revision(node1), stored1)
        self.assertEqual(f.rawdata(node1), stored1)
        self.assertEqual(f.revision(node2), stored2)
        self.assertEqual(f.rawdata(node2), stored2)

        self.assertEqual(f.read(node1), fulltext1)
        self.assertEqual(f.read(node2), fulltext2)

        # Returns False when first parent is set.
        self.assertFalse(f.renamed(node1))
        self.assertEqual(f.renamed(node2), (b'source1', b'\xbb' * 20))

        self.assertTrue(f.cmp(node1, fulltext1))
        self.assertTrue(f.cmp(node1, stored1))
        self.assertFalse(f.cmp(node2, fulltext2))
        self.assertTrue(f.cmp(node2, stored2))

    def testmetadataprefix(self):
        # Content with metadata prefix has extra prefix inserted in storage.
        fulltext0 = b'\x01\nfoo'
        stored0 = b'\x01\n\x01\n\x01\nfoo'

        fulltext1 = b'\x01\nbar'
        meta1 = {
            b'copy': b'source0',
            b'copyrev': b'b' * 40,
        }
        stored1 = b''.join(
            [
                b'\x01\ncopy: source0\n',
                b'copyrev: %s\n' % (b'b' * 40),
                b'\x01\n\x01\nbar',
            ]
        )

        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, {}, tr, 0, f.nullid, f.nullid)
            node1 = f.add(fulltext1, meta1, tr, 1, f.nullid, f.nullid)

        # TODO this is buggy.
        self.assertEqual(f.size(0), len(fulltext0) + 4)

        self.assertEqual(f.size(1), len(fulltext1))

        self.assertEqual(f.revision(node0), stored0)
        self.assertEqual(f.rawdata(node0), stored0)

        self.assertEqual(f.revision(node1), stored1)
        self.assertEqual(f.rawdata(node1), stored1)

        self.assertEqual(f.read(node0), fulltext0)
        self.assertEqual(f.read(node1), fulltext1)

        self.assertFalse(f.cmp(node0, fulltext0))
        self.assertTrue(f.cmp(node0, stored0))

        self.assertFalse(f.cmp(node1, fulltext1))
        self.assertTrue(f.cmp(node1, stored0))

    def testbadnoderead(self):
        f = self._makefilefn()

        fulltext0 = b'foo\n' * 30
        fulltext1 = fulltext0 + b'bar\n'

        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = b'\xaa' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, rawtext=fulltext1
            )

        self.assertEqual(len(f), 2)
        self.assertEqual(f.parents(node1), (node0, f.nullid))

        # revision() raises since it performs hash verification.
        with self.assertRaises(error.StorageError):
            f.revision(node1)

        # rawdata() still verifies because there are no special storage
        # settings.
        with self.assertRaises(error.StorageError):
            f.rawdata(node1)

        # read() behaves like revision().
        with self.assertRaises(error.StorageError):
            f.read(node1)

        # We can't test renamed() here because some backends may not require
        # reading/validating the fulltext to return rename metadata.

    def testbadnoderevisionraw(self):
        # Like above except we test rawdata() first to isolate
        # revision caching behavior.
        f = self._makefilefn()

        fulltext0 = b'foo\n' * 30
        fulltext1 = fulltext0 + b'bar\n'

        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = b'\xaa' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, rawtext=fulltext1
            )

        with self.assertRaises(error.StorageError):
            f.rawdata(node1)

        with self.assertRaises(error.StorageError):
            f.rawdata(node1)

    def testbadnoderevision(self):
        # Like above except we test read() first to isolate revision caching
        # behavior.
        f = self._makefilefn()

        fulltext0 = b'foo\n' * 30
        fulltext1 = fulltext0 + b'bar\n'

        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = b'\xaa' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, rawtext=fulltext1
            )

        with self.assertRaises(error.StorageError):
            f.read(node1)

        with self.assertRaises(error.StorageError):
            f.read(node1)

    def testbadnodedelta(self):
        f = self._makefilefn()

        fulltext0 = b'foo\n' * 31
        fulltext1 = fulltext0 + b'bar\n'
        fulltext2 = fulltext1 + b'baz\n'

        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)
            node1 = b'\xaa' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, rawtext=fulltext1
            )

        with self.assertRaises(error.StorageError):
            f.read(node1)

        node2 = storageutil.hashrevisionsha1(fulltext2, node1, f.nullid)

        with self._maketransactionfn() as tr:
            delta = mdiff.textdiff(fulltext1, fulltext2)
            self._addrawrevisionfn(
                f, tr, node2, node1, f.nullid, 2, delta=(1, delta)
            )

        self.assertEqual(len(f), 3)

        # Assuming a delta is stored, we shouldn't need to validate node1 in
        # order to retrieve node2.
        self.assertEqual(f.read(node2), fulltext2)

    def testcensored(self):
        f = self._makefilefn()

        stored1 = storageutil.packmeta(
            {
                b'censored': b'tombstone',
            },
            b'',
        )

        with self._maketransactionfn() as tr:
            node0 = f.add(b'foo', None, tr, 0, f.nullid, f.nullid)

            # The node value doesn't matter since we can't verify it.
            node1 = b'\xbb' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, stored1, censored=True
            )

        self.assertTrue(f.iscensored(1))

        with self.assertRaises(error.CensoredNodeError):
            f.revision(1)

        with self.assertRaises(error.CensoredNodeError):
            f.rawdata(1)

        with self.assertRaises(error.CensoredNodeError):
            f.read(1)

    def testcensoredrawrevision(self):
        # Like above, except we do the rawdata() request first to
        # isolate revision caching behavior.

        f = self._makefilefn()

        stored1 = storageutil.packmeta(
            {
                b'censored': b'tombstone',
            },
            b'',
        )

        with self._maketransactionfn() as tr:
            node0 = f.add(b'foo', None, tr, 0, f.nullid, f.nullid)

            # The node value doesn't matter since we can't verify it.
            node1 = b'\xbb' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, stored1, censored=True
            )

        with self.assertRaises(error.CensoredNodeError):
            f.rawdata(1)


class ifilemutationtests(basetestcase):
    """Generic tests for the ifilemutation interface.

    All file storage backends that support writing should conform to this
    interface.

    Use ``makeifilemutationtests()`` to create an instance of this type.
    """

    def testaddnoop(self):
        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            node0 = f.add(b'foo', None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(b'foo', None, tr, 0, f.nullid, f.nullid)
            # Varying by linkrev shouldn't impact hash.
            node2 = f.add(b'foo', None, tr, 1, f.nullid, f.nullid)

        self.assertEqual(node1, node0)
        self.assertEqual(node2, node0)
        self.assertEqual(len(f), 1)

    def testaddrevisionbadnode(self):
        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            # Adding a revision with bad node value fails.
            with self.assertRaises(error.StorageError):
                f.addrevision(
                    b'foo', tr, 0, f.nullid, f.nullid, node=b'\x01' * 20
                )

    def testaddrevisionunknownflag(self):
        f = self._makefilefn()
        with self._maketransactionfn() as tr:
            for i in range(15, 0, -1):
                if (1 << i) & ~repository.REVISION_FLAGS_KNOWN:
                    flags = 1 << i
                    break

            with self.assertRaises(error.StorageError):
                f.addrevision(b'foo', tr, 0, f.nullid, f.nullid, flags=flags)

    def testaddgroupsimple(self):
        f = self._makefilefn()

        callbackargs = []

        def cb(*args, **kwargs):
            callbackargs.append((args, kwargs))

        def linkmapper(node):
            return 0

        with self._maketransactionfn() as tr:
            nodes = []

            def onchangeset(cl, rev):
                node = cl.node(rev)
                nodes.append(node)
                cb(cl, node)

            def ondupchangeset(cl, rev):
                nodes.append(cl.node(rev))

            f.addgroup(
                [],
                None,
                tr,
                addrevisioncb=onchangeset,
                duplicaterevisioncb=ondupchangeset,
            )

        self.assertEqual(nodes, [])
        self.assertEqual(callbackargs, [])
        self.assertEqual(len(f), 0)

        fulltext0 = b'foo'
        delta0 = mdiff.trivialdiffheader(len(fulltext0)) + fulltext0

        with self._maketransactionfn() as tr:
            node0 = f.add(fulltext0, None, tr, 0, f.nullid, f.nullid)

        f = self._makefilefn()

        deltas = [
            (node0, f.nullid, f.nullid, f.nullid, f.nullid, delta0, 0, {}),
        ]

        with self._maketransactionfn() as tr:
            nodes = []

            def onchangeset(cl, rev):
                node = cl.node(rev)
                nodes.append(node)
                cb(cl, node)

            def ondupchangeset(cl, rev):
                nodes.append(cl.node(rev))

            f.addgroup(
                deltas,
                linkmapper,
                tr,
                addrevisioncb=onchangeset,
                duplicaterevisioncb=ondupchangeset,
            )

        self.assertEqual(
            nodes,
            [
                b'\x49\xd8\xcb\xb1\x5c\xe2\x57\x92\x04\x47'
                b'\x00\x6b\x46\x97\x8b\x7a\xf9\x80\xa9\x79'
            ],
        )

        self.assertEqual(len(callbackargs), 1)
        self.assertEqual(callbackargs[0][0][1], nodes[0])

        self.assertEqual(list(f.revs()), [0])
        self.assertEqual(f.rev(nodes[0]), 0)
        self.assertEqual(f.node(0), nodes[0])

    def testaddgroupmultiple(self):
        f = self._makefilefn()

        fulltexts = [
            b'foo',
            b'bar',
            b'x' * 1024,
        ]

        nodes = []
        with self._maketransactionfn() as tr:
            for fulltext in fulltexts:
                nodes.append(f.add(fulltext, None, tr, 0, f.nullid, f.nullid))

        f = self._makefilefn()
        deltas = []
        for i, fulltext in enumerate(fulltexts):
            delta = mdiff.trivialdiffheader(len(fulltext)) + fulltext

            deltas.append(
                (nodes[i], f.nullid, f.nullid, f.nullid, f.nullid, delta, 0, {})
            )

        with self._maketransactionfn() as tr:
            newnodes = []

            def onchangeset(cl, rev):
                newnodes.append(cl.node(rev))

            f.addgroup(
                deltas,
                lambda x: 0,
                tr,
                addrevisioncb=onchangeset,
                duplicaterevisioncb=onchangeset,
            )
            self.assertEqual(newnodes, nodes)

        self.assertEqual(len(f), len(deltas))
        self.assertEqual(list(f.revs()), [0, 1, 2])
        self.assertEqual(f.rev(nodes[0]), 0)
        self.assertEqual(f.rev(nodes[1]), 1)
        self.assertEqual(f.rev(nodes[2]), 2)
        self.assertEqual(f.node(0), nodes[0])
        self.assertEqual(f.node(1), nodes[1])
        self.assertEqual(f.node(2), nodes[2])

    def testdeltaagainstcensored(self):
        # Attempt to apply a delta made against a censored revision.
        f = self._makefilefn()

        stored1 = storageutil.packmeta(
            {
                b'censored': b'tombstone',
            },
            b'',
        )

        with self._maketransactionfn() as tr:
            node0 = f.add(b'foo\n' * 30, None, tr, 0, f.nullid, f.nullid)

            # The node value doesn't matter since we can't verify it.
            node1 = b'\xbb' * 20

            self._addrawrevisionfn(
                f, tr, node1, node0, f.nullid, 1, stored1, censored=True
            )

        delta = mdiff.textdiff(b'bar\n' * 30, (b'bar\n' * 30) + b'baz\n')
        deltas = [
            (b'\xcc' * 20, node1, f.nullid, b'\x01' * 20, node1, delta, 0, {})
        ]

        with self._maketransactionfn() as tr:
            with self.assertRaises(error.CensoredBaseError):
                f.addgroup(deltas, lambda x: 0, tr)

    def testcensorrevisionbasic(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            node0 = f.add(b'foo\n' * 30, None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(b'foo\n' * 31, None, tr, 1, node0, f.nullid)
            node2 = f.add(b'foo\n' * 32, None, tr, 2, node1, f.nullid)

        with self._maketransactionfn() as tr:
            f.censorrevision(tr, node1)

        self.assertEqual(len(f), 3)
        self.assertEqual(list(f.revs()), [0, 1, 2])

        self.assertEqual(f.read(node0), b'foo\n' * 30)
        self.assertEqual(f.read(node2), b'foo\n' * 32)

        with self.assertRaises(error.CensoredNodeError):
            f.read(node1)

    def testgetstrippointnoparents(self):
        # N revisions where none have parents.
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            for rev in range(10):
                f.add(b'%d' % rev, None, tr, rev, f.nullid, f.nullid)

        for rev in range(10):
            self.assertEqual(f.getstrippoint(rev), (rev, set()))

    def testgetstrippointlinear(self):
        # N revisions in a linear chain.
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            p1 = f.nullid

            for rev in range(10):
                f.add(b'%d' % rev, None, tr, rev, p1, f.nullid)

        for rev in range(10):
            self.assertEqual(f.getstrippoint(rev), (rev, set()))

    def testgetstrippointmultipleheads(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            node0 = f.add(b'0', None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(b'1', None, tr, 1, node0, f.nullid)
            f.add(b'2', None, tr, 2, node1, f.nullid)
            f.add(b'3', None, tr, 3, node0, f.nullid)
            f.add(b'4', None, tr, 4, node0, f.nullid)

        for rev in range(5):
            self.assertEqual(f.getstrippoint(rev), (rev, set()))

    def testgetstrippointearlierlinkrevs(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            node0 = f.add(b'0', None, tr, 0, f.nullid, f.nullid)
            f.add(b'1', None, tr, 10, node0, f.nullid)
            f.add(b'2', None, tr, 5, node0, f.nullid)

        self.assertEqual(f.getstrippoint(0), (0, set()))
        self.assertEqual(f.getstrippoint(1), (1, set()))
        self.assertEqual(f.getstrippoint(2), (1, set()))
        self.assertEqual(f.getstrippoint(3), (1, set()))
        self.assertEqual(f.getstrippoint(4), (1, set()))
        self.assertEqual(f.getstrippoint(5), (1, set()))
        self.assertEqual(f.getstrippoint(6), (1, {2}))
        self.assertEqual(f.getstrippoint(7), (1, {2}))
        self.assertEqual(f.getstrippoint(8), (1, {2}))
        self.assertEqual(f.getstrippoint(9), (1, {2}))
        self.assertEqual(f.getstrippoint(10), (1, {2}))
        self.assertEqual(f.getstrippoint(11), (3, set()))

    def teststripempty(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            f.strip(0, tr)

        self.assertEqual(len(f), 0)

    def teststripall(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            p1 = f.nullid
            for rev in range(10):
                p1 = f.add(b'%d' % rev, None, tr, rev, p1, f.nullid)

        self.assertEqual(len(f), 10)

        with self._maketransactionfn() as tr:
            f.strip(0, tr)

        self.assertEqual(len(f), 0)

    def teststrippartial(self):
        f = self._makefilefn()

        with self._maketransactionfn() as tr:
            f.add(b'0', None, tr, 0, f.nullid, f.nullid)
            node1 = f.add(b'1', None, tr, 5, f.nullid, f.nullid)
            node2 = f.add(b'2', None, tr, 10, f.nullid, f.nullid)

        self.assertEqual(len(f), 3)

        with self._maketransactionfn() as tr:
            f.strip(11, tr)

        self.assertEqual(len(f), 3)

        with self._maketransactionfn() as tr:
            f.strip(10, tr)

        self.assertEqual(len(f), 2)

        with self.assertRaises(error.LookupError):
            f.rev(node2)

        with self._maketransactionfn() as tr:
            f.strip(6, tr)

        self.assertEqual(len(f), 2)

        with self._maketransactionfn() as tr:
            f.strip(3, tr)

        self.assertEqual(len(f), 1)

        with self.assertRaises(error.LookupError):
            f.rev(node1)


def makeifileindextests(makefilefn, maketransactionfn, addrawrevisionfn):
    """Create a unittest.TestCase class suitable for testing file storage.

    ``makefilefn`` is a callable which receives the test case as an
    argument and returns an object implementing the ``ifilestorage`` interface.

    ``maketransactionfn`` is a callable which receives the test case as an
    argument and returns a transaction object.

    ``addrawrevisionfn`` is a callable which receives arguments describing a
    low-level revision to add. This callable allows the insertion of
    potentially bad data into the store in order to facilitate testing.

    Returns a type that is a ``unittest.TestCase`` that can be used for
    testing the object implementing the file storage interface. Simply
    assign the returned value to a module-level attribute and a test loader
    should find and run it automatically.
    """
    d = {
        '_makefilefn': makefilefn,
        '_maketransactionfn': maketransactionfn,
        '_addrawrevisionfn': addrawrevisionfn,
    }
    return type('ifileindextests', (ifileindextests,), d)


def makeifiledatatests(makefilefn, maketransactionfn, addrawrevisionfn):
    d = {
        '_makefilefn': makefilefn,
        '_maketransactionfn': maketransactionfn,
        '_addrawrevisionfn': addrawrevisionfn,
    }
    return type('ifiledatatests', (ifiledatatests,), d)


def makeifilemutationtests(makefilefn, maketransactionfn, addrawrevisionfn):
    d = {
        '_makefilefn': makefilefn,
        '_maketransactionfn': maketransactionfn,
        '_addrawrevisionfn': addrawrevisionfn,
    }
    return type('ifilemutationtests', (ifilemutationtests,), d)