hgext/automv.py

match: convert O(n) to O(log n) in exactmatcher.visitchildrenset

When using narrow, during rebase this is called (at least) once per directory in
the set of files in the commit being rebased. Every time it's called, we did the
set arithmetic (now extracted and cached), which was probably pretty cheap but
not necessary to repeat each time, looped over every item in the matcher and
kept things that started with the directory we were querying.

With very large narrowspecs, and a commit that touched a file in a large number
of directories, this was slow. In a pathological repo, the rebase of a single
commit (that touched over 17k files, I believe in approximately as many
directories) with a narrowspec that had >32k entries took 8,246s of profiled
time, with 5,007s of that spent in visitchildrenset (transitively). With this
change, the time spent in visitchildrenset is less than 34s (which is where my
profile cut off). Most of the remaining time was network access due to our
custom remotefilelog-based setup not properly prefetching.

Differential Revision: https://phab.mercurial-scm.org/D10294

# automv.py
#
# Copyright 2013-2016 Facebook, Inc.
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
"""check for unrecorded moves at commit time (EXPERIMENTAL)

This extension checks at commit/amend time if any of the committed files
comes from an unrecorded mv.

The threshold at which a file is considered a move can be set with the
``automv.similarity`` config option. This option takes a percentage between 0
(disabled) and 100 (files must be identical), the default is 95.

"""

# Using 95 as a default similarity is based on an analysis of the mercurial
# repositories of the cpython, mozilla-central & mercurial repositories, as
# well as 2 very large facebook repositories. At 95 50% of all potential
# missed moves would be caught, as well as correspond with 87% of all
# explicitly marked moves.  Together, 80% of moved files are 95% similar or
# more.
#
# See http://markmail.org/thread/5pxnljesvufvom57 for context.

from __future__ import absolute_import

from mercurial.i18n import _
from mercurial import (
    commands,
    copies,
    error,
    extensions,
    pycompat,
    registrar,
    scmutil,
    similar,
)

configtable = {}
configitem = registrar.configitem(configtable)

configitem(
    b'automv',
    b'similarity',
    default=95,
)


def extsetup(ui):
    entry = extensions.wrapcommand(commands.table, b'commit', mvcheck)
    entry[1].append(
        (b'', b'no-automv', None, _(b'disable automatic file move detection'))
    )


def mvcheck(orig, ui, repo, *pats, **opts):
    """Hook to check for moves at commit time"""
    opts = pycompat.byteskwargs(opts)
    renames = None
    disabled = opts.pop(b'no_automv', False)
    if not disabled:
        threshold = ui.configint(b'automv', b'similarity')
        if not 0 <= threshold <= 100:
            raise error.Abort(_(b'automv.similarity must be between 0 and 100'))
        if threshold > 0:
            match = scmutil.match(repo[None], pats, opts)
            added, removed = _interestingfiles(repo, match)
            uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
            renames = _findrenames(
                repo, uipathfn, added, removed, threshold / 100.0
            )

    with repo.wlock():
        if renames is not None:
            scmutil._markchanges(repo, (), (), renames)
        return orig(ui, repo, *pats, **pycompat.strkwargs(opts))


def _interestingfiles(repo, matcher):
    """Find what files were added or removed in this commit.

    Returns a tuple of two lists: (added, removed). Only files not *already*
    marked as moved are included in the added list.

    """
    stat = repo.status(match=matcher)
    added = stat.added
    removed = stat.removed

    copy = copies.pathcopies(repo[b'.'], repo[None], matcher)
    # remove the copy files for which we already have copy info
    added = [f for f in added if f not in copy]

    return added, removed


def _findrenames(repo, uipathfn, added, removed, similarity):
    """Find what files in added are really moved files.

    Any file named in removed that is at least similarity% similar to a file
    in added is seen as a rename.

    """
    renames = {}
    if similarity > 0:
        for src, dst, score in similar.findrenames(
            repo, added, removed, similarity
        ):
            if repo.ui.verbose:
                repo.ui.status(
                    _(b'detected move of %s as %s (%d%% similar)\n')
                    % (uipathfn(src), uipathfn(dst), score * 100)
                )
            renames[dst] = src
    if renames:
        repo.ui.status(_(b'detected move of %d files\n') % len(renames))
    return renames