view tests/test-parseindex2.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
author Pierre-Yves David <pierre-yves.david@octobus.net>
date Tue, 11 Feb 2020 11:18:52 +0100
parents ac627ed8a911
children e7a4c018b563
line wrap: on
line source

"""This unit test primarily tests parsers.parse_index2().

It also checks certain aspects of the parsers module as a whole.
"""

from __future__ import absolute_import, print_function

import struct
import subprocess
import sys
import unittest

from mercurial.node import (
    nullid,
    nullrev,
)
from mercurial import (
    node as nodemod,
    policy,
    pycompat,
)

parsers = policy.importmod('parsers')

# original python implementation
def gettype(q):
    return int(q & 0xFFFF)


def offset_type(offset, type):
    return int(int(offset) << 16 | type)


indexformatng = ">Qiiiiii20s12x"


def py_parseindex(data, inline):
    s = 64
    cache = None
    index = []
    nodemap = {nullid: nullrev}
    n = off = 0

    l = len(data) - s
    append = index.append
    if inline:
        cache = (0, data)
        while off <= l:
            e = struct.unpack(indexformatng, data[off : off + s])
            nodemap[e[7]] = n
            append(e)
            n += 1
            if e[1] < 0:
                break
            off += e[1] + s
    else:
        while off <= l:
            e = struct.unpack(indexformatng, data[off : off + s])
            nodemap[e[7]] = n
            append(e)
            n += 1
            off += s

    e = list(index[0])
    type = gettype(e[0])
    e[0] = offset_type(0, type)
    index[0] = tuple(e)

    return index, cache


data_inlined = (
    b'\x00\x01\x00\x01\x00\x00\x00\x00\x00\x00\x01\x8c'
    b'\x00\x00\x04\x07\x00\x00\x00\x00\x00\x00\x15\x15\xff\xff\xff'
    b'\xff\xff\xff\xff\xff\xebG\x97\xb7\x1fB\x04\xcf\x13V\x81\tw\x1b'
    b'w\xdduR\xda\xc6\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
    b'x\x9c\x9d\x93?O\xc30\x10\xc5\xf7|\x8a\xdb\x9a\xa8m\x06\xd8*\x95'
    b'\x81B\xa1\xa2\xa2R\xcb\x86Pd\x9a\x0b5$vd_\x04\xfd\xf6\x9c\xff@'
    b'\x11!\x0b\xd9\xec\xf7\xbbw\xe7gG6\xad6\x04\xdaN\xc0\x92\xa0$)'
    b'\xb1\x82\xa2\xd1%\x16\xa4\x8b7\xa9\xca\xd4-\xb2Y\x02\xfc\xc9'
    b'\xcaS\xf9\xaeX\xed\xb6\xd77Q\x02\x83\xd4\x19\xf5--Y\xea\xe1W'
    b'\xab\xed\x10\xceR\x0f_\xdf\xdf\r\xe1,\xf5\xf0\xcb\xf5 \xceR\x0f'
    b'_\xdc\x0e\x0e\xc3R\x0f_\xae\x96\x9b!\x9e\xa5\x1e\xbf\xdb,\x06'
    b'\xc7q\x9a/\x88\x82\xc3B\xea\xb5\xb4TJ\x93\xb6\x82\x0e\xe16\xe6'
    b'KQ\xdb\xaf\xecG\xa3\xd1 \x01\xd3\x0b_^\xe8\xaa\xa0\xae\xad\xd1'
    b'&\xbef\x1bz\x08\xb0|\xc9Xz\x06\xf6Z\x91\x90J\xaa\x17\x90\xaa'
    b'\xd2\xa6\x11$5C\xcf\xba#\xa0\x03\x02*2\x92-\xfc\xb1\x94\xdf\xe2'
    b'\xae\xb8\'m\x8ey0^\x85\xd3\x82\xb4\xf0`:\x9c\x00\x8a\xfd\x01'
    b'\xb0\xc6\x86\x8b\xdd\xae\x80\xf3\xa9\x9fd\x16\n\x00R%\x1a\x06'
    b'\xe9\xd8b\x98\x1d\xf4\xf3+\x9bf\x01\xd8p\x1b\xf3.\xed\x9f^g\xc3'
    b'^\xd9W81T\xdb\xd5\x04sx|\xf2\xeb\xd6`%?x\xed"\x831\xbf\xf3\xdc'
    b'b\xeb%gaY\xe1\xad\x9f\xb9f\'1w\xa9\xa5a\x83s\x82J\xb98\xbc4\x8b'
    b'\x83\x00\x9f$z\xb8#\xa5\xb1\xdf\x98\xd9\xec\x1b\x89O\xe3Ts\x9a4'
    b'\x17m\x8b\xfc\x8f\xa5\x95\x9a\xfc\xfa\xed,\xe5|\xa1\xfe\x15\xb9'
    b'\xbc\xb2\x93\x1f\xf2\x95\xff\xdf,\x1a\xc5\xe7\x17*\x93Oz:>\x0e'
)

data_non_inlined = (
    b'\x00\x00\x00\x01\x00\x00\x00\x00\x00\x01D\x19'
    b'\x00\x07e\x12\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff'
    b'\xff\xff\xff\xff\xd1\xf4\xbb\xb0\xbe\xfc\x13\xbd\x8c\xd3\x9d'
    b'\x0f\xcd\xd9;\x8c\x07\x8cJ/\x00\x00\x00\x00\x00\x00\x00\x00\x00'
    b'\x00\x00\x00\x00\x00\x00\x01D\x19\x00\x00\x00\x00\x00\xdf\x00'
    b'\x00\x01q\x00\x00\x00\x01\x00\x00\x00\x01\x00\x00\x00\x00\xff'
    b'\xff\xff\xff\xc1\x12\xb9\x04\x96\xa4Z1t\x91\xdfsJ\x90\xf0\x9bh'
    b'\x07l&\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
    b'\x00\x01D\xf8\x00\x00\x00\x00\x01\x1b\x00\x00\x01\xb8\x00\x00'
    b'\x00\x01\x00\x00\x00\x02\x00\x00\x00\x01\xff\xff\xff\xff\x02\n'
    b'\x0e\xc6&\xa1\x92\xae6\x0b\x02i\xfe-\xe5\xbao\x05\xd1\xe7\x00'
    b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01F'
    b'\x13\x00\x00\x00\x00\x01\xec\x00\x00\x03\x06\x00\x00\x00\x01'
    b'\x00\x00\x00\x03\x00\x00\x00\x02\xff\xff\xff\xff\x12\xcb\xeby1'
    b'\xb6\r\x98B\xcb\x07\xbd`\x8f\x92\xd9\xc4\x84\xbdK\x00\x00\x00'
    b'\x00\x00\x00\x00\x00\x00\x00\x00\x00'
)


def parse_index2(data, inline):
    index, chunkcache = parsers.parse_index2(data, inline)
    return list(index), chunkcache


def importparsers(hexversion):
    """Import mercurial.parsers with the given sys.hexversion."""
    # The file parsers.c inspects sys.hexversion to determine the version
    # of the currently-running Python interpreter, so we monkey-patch
    # sys.hexversion to simulate using different versions.
    code = (
        "import sys; sys.hexversion=%s; "
        "import mercurial.cext.parsers" % hexversion
    )
    cmd = "python -c \"%s\"" % code
    # We need to do these tests inside a subprocess because parser.c's
    # version-checking code happens inside the module init function, and
    # when using reload() to reimport an extension module, "The init function
    # of extension modules is not called a second time"
    # (from http://docs.python.org/2/library/functions.html?#reload).
    p = subprocess.Popen(
        cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT
    )
    return p.communicate()  # returns stdout, stderr


def hexfailmsg(testnumber, hexversion, stdout, expected):
    try:
        hexstring = hex(hexversion)
    except TypeError:
        hexstring = None
    return (
        "FAILED: version test #%s with Python %s and patched "
        "sys.hexversion %r (%r):\n Expected %s but got:\n-->'%s'\n"
        % (
            testnumber,
            sys.version_info,
            hexversion,
            hexstring,
            expected,
            stdout,
        )
    )


def makehex(major, minor, micro):
    return int("%x%02x%02x00" % (major, minor, micro), 16)


class parseindex2tests(unittest.TestCase):
    def assertversionokay(self, testnumber, hexversion):
        stdout, stderr = importparsers(hexversion)
        self.assertFalse(
            stdout, hexfailmsg(testnumber, hexversion, stdout, 'no stdout')
        )

    def assertversionfail(self, testnumber, hexversion):
        stdout, stderr = importparsers(hexversion)
        # We include versionerrortext to distinguish from other ImportErrors.
        errtext = b"ImportError: %s" % pycompat.sysbytes(
            parsers.versionerrortext
        )
        self.assertIn(
            errtext,
            stdout,
            hexfailmsg(
                testnumber,
                hexversion,
                stdout,
                expected="stdout to contain %r" % errtext,
            ),
        )

    def testversiondetection(self):
        """Check the version-detection logic when importing parsers."""
        # Only test the version-detection logic if it is present.
        try:
            parsers.versionerrortext
        except AttributeError:
            return
        info = sys.version_info
        major, minor, micro = info[0], info[1], info[2]
        # Test same major-minor versions.
        self.assertversionokay(1, makehex(major, minor, micro))
        self.assertversionokay(2, makehex(major, minor, micro + 1))
        # Test different major-minor versions.
        self.assertversionfail(3, makehex(major + 1, minor, micro))
        self.assertversionfail(4, makehex(major, minor + 1, micro))
        self.assertversionfail(5, "'foo'")

    def testbadargs(self):
        # Check that parse_index2() raises TypeError on bad arguments.
        with self.assertRaises(TypeError):
            parse_index2(0, True)

    def testparseindexfile(self):
        # Check parsers.parse_index2() on an index file against the
        # original Python implementation of parseindex, both with and
        # without inlined data.

        want = py_parseindex(data_inlined, True)
        got = parse_index2(data_inlined, True)
        self.assertEqual(want, got)  # inline data

        want = py_parseindex(data_non_inlined, False)
        got = parse_index2(data_non_inlined, False)
        self.assertEqual(want, got)  # no inline data

        ix = parsers.parse_index2(data_inlined, True)[0]
        for i, r in enumerate(ix):
            if r[7] == nullid:
                i = -1
            try:
                self.assertEqual(
                    ix[r[7]],
                    i,
                    'Reverse lookup inconsistent for %r' % nodemod.hex(r[7]),
                )
            except TypeError:
                # pure version doesn't support this
                break

    def testminusone(self):
        want = (0, 0, 0, -1, -1, -1, -1, nullid)
        index, junk = parsers.parse_index2(data_inlined, True)
        got = index[-1]
        self.assertEqual(want, got)  # inline data

        index, junk = parsers.parse_index2(data_non_inlined, False)
        got = index[-1]
        self.assertEqual(want, got)  # no inline data

    def testdelitemwithoutnodetree(self):
        index, _junk = parsers.parse_index2(data_non_inlined, False)

        def hexrev(rev):
            if rev == nullrev:
                return b'\xff\xff\xff\xff'
            else:
                return nodemod.bin('%08x' % rev)

        def appendrev(p1, p2=nullrev):
            # node won't matter for this test, let's just make sure
            # they don't collide. Other data don't matter either.
            node = hexrev(p1) + hexrev(p2) + b'.' * 12
            index.append((0, 0, 12, 1, 34, p1, p2, node))

        appendrev(4)
        appendrev(5)
        appendrev(6)
        self.assertEqual(len(index), 7)

        del index[1:-1]

        # assertions that failed before correction
        self.assertEqual(len(index), 1)  # was 4
        headrevs = getattr(index, 'headrevs', None)
        if headrevs is not None:  # not implemented in pure
            self.assertEqual(index.headrevs(), [0])  # gave ValueError


if __name__ == '__main__':
    import silenttestrunner

    silenttestrunner.main(__name__)