view tests/test-manifest.py @ 44363:f7459da77f23

nodemap: introduce an option to use mmap to read the nodemap mapping

The performance and memory benefit is much greater if we don't have to copy all
the data in memory for each information. So we introduce an option (on by
default) to read the data using mmap.

This changeset is the last one definition the API for index support nodemap
data. (they have to be able to use the mmaping).

Below are some benchmark comparing the best we currently have in 5.3 with the
final step of this series (using the persistent nodemap implementation in
Rust). The benchmark run `hg perfindex` with various revset and the following
variants:

Before:

 * do not use the persistent nodemap
 * use the CPython implementation of the index for nodemap
 * use mmapping of the changelog index

After:

 * use the MixedIndex Rust code, with the NodeTree object for nodemap access
   (still in review)
 * use the persistent nodemap data from disk
 * access the persistent nodemap data through mmap
 * use mmapping of the changelog index

The persistent nodemap greatly speed up  most operation on very large
repositories. Some of the previously very fast lookup end up a bit slower because
the persistent nodemap has to be setup. However the absolute slowdown is very
small and won't matters in the big picture.

Here are some numbers (in seconds) for the reference copy of mozilla-try:

Revset            Before   After     abs-change speedup

-10000:           0.004622 0.005532   0.000910  ×   0.83
-10:              0.000050 0.000132   0.000082  ×   0.37
tip               0.000052 0.000085   0.000033  ×   0.61
0 + (-10000:)     0.028222 0.005337  -0.022885  ×   5.29
0                 0.023521 0.000084  -0.023437  × 280.01
(-10000:) + 0     0.235539 0.005308  -0.230231  ×  44.37
(-10:) + :9       0.232883 0.000180  -0.232703  ×1293.79
(-10000:) + (:99) 0.238735 0.005358  -0.233377  ×  44.55
:99 + (-10000:)   0.317942 0.005593  -0.312349  ×  56.84
:9 + (-10:)       0.313372 0.000179  -0.313193  ×1750.68
:9                0.316450 0.000143  -0.316307  ×2212.93


On smaller repositories, the cost of nodemap related operation is not as big, so
the win is much more modest. Yet it helps shaving a handful of millisecond here
and there.

Here are some numbers (in seconds) for the reference copy of mercurial:

Revset            Before   After     abs-change speedup
-10:              0.000065 0.000097   0.000032  × 0.67
tip               0.000063 0.000078   0.000015  × 0.80
0                 0.000561 0.000079  -0.000482  × 7.10
-10000:           0.004609 0.003648  -0.000961  × 1.26
0 + (-10000:)     0.005023 0.003715  -0.001307  × 1.35
(-10:) + :9       0.002187 0.000108  -0.002079  ×20.25
(-10000:) + 0     0.006252 0.003716  -0.002536  × 1.68
(-10000:) + (:99) 0.006367 0.003707  -0.002660  × 1.71
:9 + (-10:)       0.003846 0.000110  -0.003736  ×34.96
:9                0.003854 0.000099  -0.003755  ×38.92
:99 + (-10000:)   0.007644 0.003778  -0.003866  × 2.02

Differential Revision: https://phab.mercurial-scm.org/D7894

from __future__ import absolute_import

import binascii
import itertools
import silenttestrunner
import unittest
import zlib

from mercurial import (
    manifest as manifestmod,
    match as matchmod,
    util,
)

EMTPY_MANIFEST = b''

HASH_1 = b'1' * 40
BIN_HASH_1 = binascii.unhexlify(HASH_1)
HASH_2 = b'f' * 40
BIN_HASH_2 = binascii.unhexlify(HASH_2)
HASH_3 = b'1234567890abcdef0987654321deadbeef0fcafe'
BIN_HASH_3 = binascii.unhexlify(HASH_3)
A_SHORT_MANIFEST = (
    b'bar/baz/qux.py\0%(hash2)s%(flag2)s\n' b'foo\0%(hash1)s%(flag1)s\n'
) % {b'hash1': HASH_1, b'flag1': b'', b'hash2': HASH_2, b'flag2': b'l',}

A_DEEPER_MANIFEST = (
    b'a/b/c/bar.py\0%(hash3)s%(flag1)s\n'
    b'a/b/c/bar.txt\0%(hash1)s%(flag1)s\n'
    b'a/b/c/foo.py\0%(hash3)s%(flag1)s\n'
    b'a/b/c/foo.txt\0%(hash2)s%(flag2)s\n'
    b'a/b/d/baz.py\0%(hash3)s%(flag1)s\n'
    b'a/b/d/qux.py\0%(hash1)s%(flag2)s\n'
    b'a/b/d/ten.txt\0%(hash3)s%(flag2)s\n'
    b'a/b/dog.py\0%(hash3)s%(flag1)s\n'
    b'a/b/fish.py\0%(hash2)s%(flag1)s\n'
    b'a/c/london.py\0%(hash3)s%(flag2)s\n'
    b'a/c/paper.txt\0%(hash2)s%(flag2)s\n'
    b'a/c/paris.py\0%(hash2)s%(flag1)s\n'
    b'a/d/apple.py\0%(hash3)s%(flag1)s\n'
    b'a/d/pizza.py\0%(hash3)s%(flag2)s\n'
    b'a/green.py\0%(hash1)s%(flag2)s\n'
    b'a/purple.py\0%(hash2)s%(flag1)s\n'
    b'app.py\0%(hash3)s%(flag1)s\n'
    b'readme.txt\0%(hash2)s%(flag1)s\n'
) % {
    b'hash1': HASH_1,
    b'flag1': b'',
    b'hash2': HASH_2,
    b'flag2': b'l',
    b'hash3': HASH_3,
}

HUGE_MANIFEST_ENTRIES = 200001

izip = getattr(itertools, 'izip', zip)
if 'xrange' not in globals():
    xrange = range

A_HUGE_MANIFEST = b''.join(
    sorted(
        b'file%d\0%s%s\n' % (i, h, f)
        for i, h, f in izip(
            xrange(200001),
            itertools.cycle((HASH_1, HASH_2)),
            itertools.cycle((b'', b'x', b'l')),
        )
    )
)


class basemanifesttests(object):
    def parsemanifest(self, text):
        raise NotImplementedError('parsemanifest not implemented by test case')

    def testEmptyManifest(self):
        m = self.parsemanifest(EMTPY_MANIFEST)
        self.assertEqual(0, len(m))
        self.assertEqual([], list(m))

    def testManifest(self):
        m = self.parsemanifest(A_SHORT_MANIFEST)
        self.assertEqual([b'bar/baz/qux.py', b'foo'], list(m))
        self.assertEqual(BIN_HASH_2, m[b'bar/baz/qux.py'])
        self.assertEqual(b'l', m.flags(b'bar/baz/qux.py'))
        self.assertEqual(BIN_HASH_1, m[b'foo'])
        self.assertEqual(b'', m.flags(b'foo'))
        with self.assertRaises(KeyError):
            m[b'wat']

    def testSetItem(self):
        want = BIN_HASH_1

        m = self.parsemanifest(EMTPY_MANIFEST)
        m[b'a'] = want
        self.assertIn(b'a', m)
        self.assertEqual(want, m[b'a'])
        self.assertEqual(b'a\0' + HASH_1 + b'\n', m.text())

        m = self.parsemanifest(A_SHORT_MANIFEST)
        m[b'a'] = want
        self.assertEqual(want, m[b'a'])
        self.assertEqual(b'a\0' + HASH_1 + b'\n' + A_SHORT_MANIFEST, m.text())

    def testSetFlag(self):
        want = b'x'

        m = self.parsemanifest(EMTPY_MANIFEST)
        # first add a file; a file-less flag makes no sense
        m[b'a'] = BIN_HASH_1
        m.setflag(b'a', want)
        self.assertEqual(want, m.flags(b'a'))
        self.assertEqual(b'a\0' + HASH_1 + want + b'\n', m.text())

        m = self.parsemanifest(A_SHORT_MANIFEST)
        # first add a file; a file-less flag makes no sense
        m[b'a'] = BIN_HASH_1
        m.setflag(b'a', want)
        self.assertEqual(want, m.flags(b'a'))
        self.assertEqual(
            b'a\0' + HASH_1 + want + b'\n' + A_SHORT_MANIFEST, m.text()
        )

    def testCopy(self):
        m = self.parsemanifest(A_SHORT_MANIFEST)
        m[b'a'] = BIN_HASH_1
        m2 = m.copy()
        del m
        del m2  # make sure we don't double free() anything

    def testCompaction(self):
        unhex = binascii.unhexlify
        h1, h2 = unhex(HASH_1), unhex(HASH_2)
        m = self.parsemanifest(A_SHORT_MANIFEST)
        m[b'alpha'] = h1
        m[b'beta'] = h2
        del m[b'foo']
        want = b'alpha\0%s\nbar/baz/qux.py\0%sl\nbeta\0%s\n' % (
            HASH_1,
            HASH_2,
            HASH_2,
        )
        self.assertEqual(want, m.text())
        self.assertEqual(3, len(m))
        self.assertEqual([b'alpha', b'bar/baz/qux.py', b'beta'], list(m))
        self.assertEqual(h1, m[b'alpha'])
        self.assertEqual(h2, m[b'bar/baz/qux.py'])
        self.assertEqual(h2, m[b'beta'])
        self.assertEqual(b'', m.flags(b'alpha'))
        self.assertEqual(b'l', m.flags(b'bar/baz/qux.py'))
        self.assertEqual(b'', m.flags(b'beta'))
        with self.assertRaises(KeyError):
            m[b'foo']

    def testSetGetNodeSuffix(self):
        clean = self.parsemanifest(A_SHORT_MANIFEST)
        m = self.parsemanifest(A_SHORT_MANIFEST)
        h = m[b'foo']
        f = m.flags(b'foo')
        want = h + b'a'
        # Merge code wants to set 21-byte fake hashes at times
        m[b'foo'] = want
        self.assertEqual(want, m[b'foo'])
        self.assertEqual(
            [(b'bar/baz/qux.py', BIN_HASH_2), (b'foo', BIN_HASH_1 + b'a')],
            list(m.items()),
        )
        # Sometimes it even tries a 22-byte fake hash, but we can
        # return 21 and it'll work out
        m[b'foo'] = want + b'+'
        self.assertEqual(want, m[b'foo'])
        # make sure the suffix survives a copy
        match = matchmod.match(util.localpath(b'/repo'), b'', [b're:foo'])
        m2 = m._matches(match)
        self.assertEqual(want, m2[b'foo'])
        self.assertEqual(1, len(m2))
        m2 = m.copy()
        self.assertEqual(want, m2[b'foo'])
        # suffix with iteration
        self.assertEqual(
            [(b'bar/baz/qux.py', BIN_HASH_2), (b'foo', want)], list(m.items())
        )

        # shows up in diff
        self.assertEqual({b'foo': ((want, f), (h, b''))}, m.diff(clean))
        self.assertEqual({b'foo': ((h, b''), (want, f))}, clean.diff(m))

    def testMatchException(self):
        m = self.parsemanifest(A_SHORT_MANIFEST)
        match = matchmod.match(util.localpath(b'/repo'), b'', [b're:.*'])

        def filt(path):
            if path == b'foo':
                assert False
            return True

        match.matchfn = filt
        with self.assertRaises(AssertionError):
            m._matches(match)

    def testRemoveItem(self):
        m = self.parsemanifest(A_SHORT_MANIFEST)
        del m[b'foo']
        with self.assertRaises(KeyError):
            m[b'foo']
        self.assertEqual(1, len(m))
        self.assertEqual(1, len(list(m)))
        # now restore and make sure everything works right
        m[b'foo'] = b'a' * 20
        self.assertEqual(2, len(m))
        self.assertEqual(2, len(list(m)))

    def testManifestDiff(self):
        MISSING = (None, b'')
        addl = b'z-only-in-left\0' + HASH_1 + b'\n'
        addr = b'z-only-in-right\0' + HASH_2 + b'x\n'
        left = self.parsemanifest(
            A_SHORT_MANIFEST.replace(HASH_1, HASH_3 + b'x') + addl
        )
        right = self.parsemanifest(A_SHORT_MANIFEST + addr)
        want = {
            b'foo': ((BIN_HASH_3, b'x'), (BIN_HASH_1, b'')),
            b'z-only-in-left': ((BIN_HASH_1, b''), MISSING),
            b'z-only-in-right': (MISSING, (BIN_HASH_2, b'x')),
        }
        self.assertEqual(want, left.diff(right))

        want = {
            b'bar/baz/qux.py': (MISSING, (BIN_HASH_2, b'l')),
            b'foo': (MISSING, (BIN_HASH_3, b'x')),
            b'z-only-in-left': (MISSING, (BIN_HASH_1, b'')),
        }
        self.assertEqual(want, self.parsemanifest(EMTPY_MANIFEST).diff(left))

        want = {
            b'bar/baz/qux.py': ((BIN_HASH_2, b'l'), MISSING),
            b'foo': ((BIN_HASH_3, b'x'), MISSING),
            b'z-only-in-left': ((BIN_HASH_1, b''), MISSING),
        }
        self.assertEqual(want, left.diff(self.parsemanifest(EMTPY_MANIFEST)))
        copy = right.copy()
        del copy[b'z-only-in-right']
        del right[b'foo']
        want = {
            b'foo': (MISSING, (BIN_HASH_1, b'')),
            b'z-only-in-right': ((BIN_HASH_2, b'x'), MISSING),
        }
        self.assertEqual(want, right.diff(copy))

        short = self.parsemanifest(A_SHORT_MANIFEST)
        pruned = short.copy()
        del pruned[b'foo']
        want = {
            b'foo': ((BIN_HASH_1, b''), MISSING),
        }
        self.assertEqual(want, short.diff(pruned))
        want = {
            b'foo': (MISSING, (BIN_HASH_1, b'')),
        }
        self.assertEqual(want, pruned.diff(short))
        want = {
            b'bar/baz/qux.py': None,
            b'foo': (MISSING, (BIN_HASH_1, b'')),
        }
        self.assertEqual(want, pruned.diff(short, clean=True))

    def testReversedLines(self):
        backwards = b''.join(
            l + b'\n' for l in reversed(A_SHORT_MANIFEST.split(b'\n')) if l
        )
        try:
            self.parsemanifest(backwards)
            self.fail('Should have raised ValueError')
        except ValueError as v:
            self.assertIn('Manifest lines not in sorted order.', str(v))

    def testNoTerminalNewline(self):
        try:
            self.parsemanifest(A_SHORT_MANIFEST + b'wat')
            self.fail('Should have raised ValueError')
        except ValueError as v:
            self.assertIn('Manifest did not end in a newline.', str(v))

    def testNoNewLineAtAll(self):
        try:
            self.parsemanifest(b'wat')
            self.fail('Should have raised ValueError')
        except ValueError as v:
            self.assertIn('Manifest did not end in a newline.', str(v))

    def testHugeManifest(self):
        m = self.parsemanifest(A_HUGE_MANIFEST)
        self.assertEqual(HUGE_MANIFEST_ENTRIES, len(m))
        self.assertEqual(len(m), len(list(m)))

    def testMatchesMetadata(self):
        '''Tests matches() for a few specific files to make sure that both
        the set of files as well as their flags and nodeids are correct in
        the resulting manifest.'''
        m = self.parsemanifest(A_HUGE_MANIFEST)

        match = matchmod.exact([b'file1', b'file200', b'file300'])
        m2 = m._matches(match)

        w = (b'file1\0%sx\n' b'file200\0%sl\n' b'file300\0%s\n') % (
            HASH_2,
            HASH_1,
            HASH_1,
        )
        self.assertEqual(w, m2.text())

    def testMatchesNonexistentFile(self):
        '''Tests matches() for a small set of specific files, including one
        nonexistent file to make sure in only matches against existing files.
        '''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.exact(
            [b'a/b/c/bar.txt', b'a/b/d/qux.py', b'readme.txt', b'nonexistent']
        )
        m2 = m._matches(match)

        self.assertEqual(
            [b'a/b/c/bar.txt', b'a/b/d/qux.py', b'readme.txt'], m2.keys()
        )

    def testMatchesNonexistentDirectory(self):
        '''Tests matches() for a relpath match on a directory that doesn't
        actually exist.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.match(
            util.localpath(b'/repo'), b'', [b'a/f'], default=b'relpath'
        )
        m2 = m._matches(match)

        self.assertEqual([], m2.keys())

    def testMatchesExactLarge(self):
        '''Tests matches() for files matching a large list of exact files.
        '''
        m = self.parsemanifest(A_HUGE_MANIFEST)

        flist = m.keys()[80:300]
        match = matchmod.exact(flist)
        m2 = m._matches(match)

        self.assertEqual(flist, m2.keys())

    def testMatchesFull(self):
        '''Tests matches() for what should be a full match.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.match(util.localpath(b'/repo'), b'', [b''])
        m2 = m._matches(match)

        self.assertEqual(m.keys(), m2.keys())

    def testMatchesDirectory(self):
        '''Tests matches() on a relpath match on a directory, which should
        match against all files within said directory.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.match(
            util.localpath(b'/repo'), b'', [b'a/b'], default=b'relpath'
        )
        m2 = m._matches(match)

        self.assertEqual(
            [
                b'a/b/c/bar.py',
                b'a/b/c/bar.txt',
                b'a/b/c/foo.py',
                b'a/b/c/foo.txt',
                b'a/b/d/baz.py',
                b'a/b/d/qux.py',
                b'a/b/d/ten.txt',
                b'a/b/dog.py',
                b'a/b/fish.py',
            ],
            m2.keys(),
        )

    def testMatchesExactPath(self):
        '''Tests matches() on an exact match on a directory, which should
        result in an empty manifest because you can't perform an exact match
        against a directory.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.exact([b'a/b'])
        m2 = m._matches(match)

        self.assertEqual([], m2.keys())

    def testMatchesCwd(self):
        '''Tests matches() on a relpath match with the current directory ('.')
        when not in the root directory.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.match(
            util.localpath(b'/repo'), b'a/b', [b'.'], default=b'relpath'
        )
        m2 = m._matches(match)

        self.assertEqual(
            [
                b'a/b/c/bar.py',
                b'a/b/c/bar.txt',
                b'a/b/c/foo.py',
                b'a/b/c/foo.txt',
                b'a/b/d/baz.py',
                b'a/b/d/qux.py',
                b'a/b/d/ten.txt',
                b'a/b/dog.py',
                b'a/b/fish.py',
            ],
            m2.keys(),
        )

    def testMatchesWithPattern(self):
        '''Tests matches() for files matching a pattern that reside
        deeper than the specified directory.'''
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        match = matchmod.match(util.localpath(b'/repo'), b'', [b'a/b/*/*.txt'])
        m2 = m._matches(match)

        self.assertEqual(
            [b'a/b/c/bar.txt', b'a/b/c/foo.txt', b'a/b/d/ten.txt'], m2.keys()
        )


class testmanifestdict(unittest.TestCase, basemanifesttests):
    def parsemanifest(self, text):
        return manifestmod.manifestdict(text)

    def testObviouslyBogusManifest(self):
        # This is a 163k manifest that came from oss-fuzz. It was a
        # timeout there, but when run normally it doesn't seem to
        # present any particular slowness.
        data = zlib.decompress(
            b'x\x9c\xed\xce;\n\x83\x00\x10\x04\xd0\x8deNa\x93~\xf1\x03\xc9q\xf4'
            b'\x14\xeaU\xbdB\xda\xd4\xe6Cj\xc1FA\xde+\x86\xe9f\xa2\xfci\xbb\xfb'
            b'\xa3\xef\xea\xba\xca\x7fk\x86q\x9a\xc6\xc8\xcc&\xb3\xcf\xf8\xb8|#'
            b'\x8a9\x00\xd8\xe6v\xf4\x01N\xe1\n\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
            b'\x00\x00\xc0\x8aey\x1d}\x01\xd8\xe0\xb9\xf3\xde\x1b\xcf\x17'
            b'\xac\xbe'
        )
        with self.assertRaises(ValueError):
            self.parsemanifest(data)


class testtreemanifest(unittest.TestCase, basemanifesttests):
    def parsemanifest(self, text):
        return manifestmod.treemanifest(b'', text)

    def testWalkSubtrees(self):
        m = self.parsemanifest(A_DEEPER_MANIFEST)

        dirs = [s._dir for s in m.walksubtrees()]
        self.assertEqual(
            sorted(
                [b'', b'a/', b'a/c/', b'a/d/', b'a/b/', b'a/b/c/', b'a/b/d/']
            ),
            sorted(dirs),
        )

        match = matchmod.match(util.localpath(b'/repo'), b'', [b'path:a/b/'])
        dirs = [s._dir for s in m.walksubtrees(matcher=match)]
        self.assertEqual(sorted([b'a/b/', b'a/b/c/', b'a/b/d/']), sorted(dirs))


if __name__ == '__main__':
    silenttestrunner.main(__name__)