copies: don't filter out copy targets created on other side of merge commit

If file X is copied to Y on one side of merge and the other side
creates Y (no copy), we would not mark that as copy. In the
changeset-centric pathcopies() version, that was done by checking if
the copy target existed on the other branch. Even though merge commits
are pretty uncommon, it still turned out to be too expensive to load
the manifest of the parents of merge commits. In a repo of
mozilla-unified converted to storing copies in changesets, about 2m30s
of `hg debugpathcopies FIREFOX_BETA_59_END FIREFOX_BETA_60_BASE` is
spent on this check of merge commits.

I tried to think of a way of storing more information in the
changesets in order to cheaply detect these cases, but I couldn't
think of a solution. So this patch simply removes those checks.

For reference, these extra copies are reported from the aforementioned
command after this patch:

  browser/base/content/sanitize.js -> browser/modules/Sanitizer.jsm
  testing/mozbase/mozprocess/tests/process_normal_finish_python.ini -> testing/mozbase/mozprocess/tests/process_normal_finish.ini
  testing/mozbase/mozprocess/tests/process_waittimeout_python.ini -> testing/mozbase/mozprocess/tests/process_waittimeout.ini
  testing/mozbase/mozprocess/tests/process_waittimeout_10s_python.ini -> testing/mozbase/mozprocess/tests/process_waittimeout_10s.ini

Since these copies were created on one side of some merge, it still
seems reasonable to include them, so I'm not even sure it's worse than
filelog pathcopies(), just different.

Differential Revision: https://phab.mercurial-scm.org/D6420

mercurial/copies.py

@@ -1,809 +1,804 b''
 # copies.py - copy detection for Mercurial
 #
 # Copyright 2008 Matt Mackall <[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.
 
 from __future__ import absolute_import
 
 import collections
 import heapq
 import os
 
 from .i18n import _
 
 from . import (
     match as matchmod,
     node,
     pathutil,
     util,
 )
 from .utils import (
     stringutil,
 )
 
 def _findlimit(repo, ctxa, ctxb):
     """
     Find the last revision that needs to be checked to ensure that a full
     transitive closure for file copies can be properly calculated.
     Generally, this means finding the earliest revision number that's an
     ancestor of a or b but not both, except when a or b is a direct descendent
     of the other, in which case we can return the minimum revnum of a and b.
     """
 
     # basic idea:
     # - mark a and b with different sides
     # - if a parent's children are all on the same side, the parent is
     #   on that side, otherwise it is on no side
     # - walk the graph in topological order with the help of a heap;
     #   - add unseen parents to side map
     #   - clear side of any parent that has children on different sides
     #   - track number of interesting revs that might still be on a side
     #   - track the lowest interesting rev seen
     #   - quit when interesting revs is zero
 
     cl = repo.changelog
     wdirparents = None
     a = ctxa.rev()
     b = ctxb.rev()
     if a is None:
         wdirparents = (ctxa.p1(), ctxa.p2())
         a = node.wdirrev
     if b is None:
         assert not wdirparents
         wdirparents = (ctxb.p1(), ctxb.p2())
         b = node.wdirrev
 
     side = {a: -1, b: 1}
     visit = [-a, -b]
     heapq.heapify(visit)
     interesting = len(visit)
     limit = node.wdirrev
 
     while interesting:
         r = -heapq.heappop(visit)
         if r == node.wdirrev:
             parents = [pctx.rev() for pctx in wdirparents]
         else:
             parents = cl.parentrevs(r)
         if parents[1] == node.nullrev:
             parents = parents[:1]
         for p in parents:
             if p not in side:
                 # first time we see p; add it to visit
                 side[p] = side[r]
                 if side[p]:
                     interesting += 1
                 heapq.heappush(visit, -p)
             elif side[p] and side[p] != side[r]:
                 # p was interesting but now we know better
                 side[p] = 0
                 interesting -= 1
         if side[r]:
             limit = r # lowest rev visited
             interesting -= 1
 
     # Consider the following flow (see test-commit-amend.t under issue4405):
     # 1/ File 'a0' committed
     # 2/ File renamed from 'a0' to 'a1' in a new commit (call it 'a1')
     # 3/ Move back to first commit
     # 4/ Create a new commit via revert to contents of 'a1' (call it 'a1-amend')
     # 5/ Rename file from 'a1' to 'a2' and commit --amend 'a1-msg'
     #
     # During the amend in step five, we will be in this state:
     #
     # @  3 temporary amend commit for a1-amend
     # |
     # o  2 a1-amend
     # |
     # | o  1 a1
     # |/
     # o  0 a0
     #
     # When _findlimit is called, a and b are revs 3 and 0, so limit will be 2,
     # yet the filelog has the copy information in rev 1 and we will not look
     # back far enough unless we also look at the a and b as candidates.
     # This only occurs when a is a descendent of b or visa-versa.
     return min(limit, a, b)
 
 def _chainandfilter(src, dst, a, b):
     """chain two sets of copies 'a' and 'b' and filter result"""
 
     # When chaining copies in 'a' (from 'src' via some other commit 'mid') with
     # copies in 'b' (from 'mid' to 'dst'), we can get the different cases in the
     # following table (not including trivial cases). For example, case 2 is
     # where a file existed in 'src' and remained under that name in 'mid' and
     # then was renamed between 'mid' and 'dst'.
     #
     # case src mid dst result
     #   1   x   y   -    -
     #   2   x   y   y   x->y
     #   3   x   y   x    -
     #   4   x   y   z   x->z
     #   5   -   x   y    -
     #   6   x   x   y   x->y
     #
     # _chain() takes care of chaining the copies in 'a' and 'b', but it
     # cannot tell the difference between cases 1 and 2, between 3 and 4, or
     # between 5 and 6, so it includes all cases in its result.
     # Cases 1, 3, and 5 are then removed by _filter().
 
     t = _chain(a, b)
     _filter(src, dst, t)
     return t
 
 def _filter(src, dst, t):
     """filters out invalid copies after chaining"""
     for k, v in list(t.items()):
         # remove copies from files that didn't exist
         if v not in src:
             del t[k]
         # remove criss-crossed copies
         elif k in src and v in dst:
             del t[k]
         # remove copies to files that were then removed
         elif k not in dst:
             del t[k]
 
 def _chain(a, b):
     """chain two sets of copies 'a' and 'b'"""
     t = a.copy()
     for k, v in b.iteritems():
         if v in t:
             t[k] = t[v]
         else:
             t[k] = v
     return t
 
 def _tracefile(fctx, am, limit):
     """return file context that is the ancestor of fctx present in ancestor
     manifest am, stopping after the first ancestor lower than limit"""
 
     for f in fctx.ancestors():
         if am.get(f.path(), None) == f.filenode():
             return f
         if not f.isintroducedafter(limit):
             return None
 
 def _dirstatecopies(repo, match=None):
     ds = repo.dirstate
     c = ds.copies().copy()
     for k in list(c):
         if ds[k] not in 'anm' or (match and not match(k)):
             del c[k]
     return c
 
 def _computeforwardmissing(a, b, match=None):
     """Computes which files are in b but not a.
     This is its own function so extensions can easily wrap this call to see what
     files _forwardcopies is about to process.
     """
     ma = a.manifest()
     mb = b.manifest()
     return mb.filesnotin(ma, match=match)
 
 def usechangesetcentricalgo(repo):
     """Checks if we should use changeset-centric copy algorithms"""
     return (repo.ui.config('experimental', 'copies.read-from') in
             ('changeset-only', 'compatibility'))
 
 def _committedforwardcopies(a, b, match):
     """Like _forwardcopies(), but b.rev() cannot be None (working copy)"""
     # files might have to be traced back to the fctx parent of the last
     # one-side-only changeset, but not further back than that
     repo = a._repo
 
     if usechangesetcentricalgo(repo):
         return _changesetforwardcopies(a, b, match)
 
     debug = repo.ui.debugflag and repo.ui.configbool('devel', 'debug.copies')
     dbg = repo.ui.debug
     if debug:
         dbg('debug.copies:    looking into rename from %s to %s\n'
             % (a, b))
     limit = _findlimit(repo, a, b)
     if debug:
         dbg('debug.copies:      search limit: %d\n' % limit)
     am = a.manifest()
 
     # find where new files came from
     # we currently don't try to find where old files went, too expensive
     # this means we can miss a case like 'hg rm b; hg cp a b'
     cm = {}
 
     # Computing the forward missing is quite expensive on large manifests, since
     # it compares the entire manifests. We can optimize it in the common use
     # case of computing what copies are in a commit versus its parent (like
     # during a rebase or histedit). Note, we exclude merge commits from this
     # optimization, since the ctx.files() for a merge commit is not correct for
     # this comparison.
     forwardmissingmatch = match
     if b.p1() == a and b.p2().node() == node.nullid:
         filesmatcher = matchmod.exact(b.files())
         forwardmissingmatch = matchmod.intersectmatchers(match, filesmatcher)
     missing = _computeforwardmissing(a, b, match=forwardmissingmatch)
 
     ancestrycontext = a._repo.changelog.ancestors([b.rev()], inclusive=True)
 
     if debug:
         dbg('debug.copies:      missing files to search: %d\n' % len(missing))
 
     for f in sorted(missing):
         if debug:
             dbg('debug.copies:        tracing file: %s\n' % f)
         fctx = b[f]
         fctx._ancestrycontext = ancestrycontext
 
         if debug:
             start = util.timer()
         ofctx = _tracefile(fctx, am, limit)
         if ofctx:
             if debug:
                 dbg('debug.copies:          rename of: %s\n' % ofctx._path)
             cm[f] = ofctx.path()
         if debug:
             dbg('debug.copies:          time: %f seconds\n'
                 % (util.timer() - start))
     return cm
 
 def _changesetforwardcopies(a, b, match):
     if a.rev() == node.nullrev:
         return {}
 
     repo = a.repo()
     children = {}
     cl = repo.changelog
     missingrevs = cl.findmissingrevs(common=[a.rev()], heads=[b.rev()])
     for r in missingrevs:
         for p in cl.parentrevs(r):
             if p == node.nullrev:
                 continue
             if p not in children:
                 children[p] = [r]
             else:
                 children[p].append(r)
 
     roots = set(children) - set(missingrevs)
     # 'work' contains 3-tuples of a (revision number, parent number, copies).
     # The parent number is only used for knowing which parent the copies dict
     # came from.
     work = [(r, 1, {}) for r in roots]
     heapq.heapify(work)
     while work:
         r, i1, copies1 = heapq.heappop(work)
         if work and work[0][0] == r:
             # We are tracing copies from both parents
             r, i2, copies2 = heapq.heappop(work)
             copies = {}
-            ctx = repo[r]
-            p1man, p2man = ctx.p1().manifest(), ctx.p2().manifest()
             allcopies = set(copies1) | set(copies2)
             # TODO: perhaps this filtering should be done as long as ctx
             # is merge, whether or not we're tracing from both parent.
             for dst in allcopies:
                 if not match(dst):
                     continue
-                if dst not in copies2:
-                    # Copied on p1 side: mark as copy from p1 side if it didn't
-                    # already exist on p2 side
-                    if dst not in p2man:
-                        copies[dst] = copies1[dst]
-                elif dst not in copies1:
-                    # Copied on p2 side: mark as copy from p2 side if it didn't
-                    # already exist on p1 side
-                    if dst not in p1man:
-                        copies[dst] = copies2[dst]
+                # Unlike when copies are stored in the filelog, we consider
+                # it a copy even if the destination already existed on the
+                # other branch. It's simply too expensive to check if the
+                # file existed in the manifest.
+                if dst in copies1:
+                    # If it was copied on the p1 side, mark it as copied from
+                    # that side, even if it was also copied on the p2 side.
+                    copies[dst] = copies1[dst]
                 else:
-                    # Copied on both sides: mark as copy from p1 side
-                    copies[dst] = copies1[dst]
+                    copies[dst] = copies2[dst]
         else:
             copies = copies1
         if r == b.rev():
             _filter(a, b, copies)
             return copies
         for c in children[r]:
             childctx = repo[c]
             if r == childctx.p1().rev():
                 parent = 1
                 childcopies = childctx.p1copies()
             else:
                 assert r == childctx.p2().rev()
                 parent = 2
                 childcopies = childctx.p2copies()
             if not match.always():
                 childcopies = {dst: src for dst, src in childcopies.items()
                                if match(dst)}
             childcopies = _chain(copies, childcopies)
             for f in childctx.filesremoved():
                 if f in childcopies:
                     del childcopies[f]
             heapq.heappush(work, (c, parent, childcopies))
     assert False
 
 def _forwardcopies(a, b, match=None):
     """find {dst@b: src@a} copy mapping where a is an ancestor of b"""
 
     match = a.repo().narrowmatch(match)
     # check for working copy
     if b.rev() is None:
         if a == b.p1():
             # short-circuit to avoid issues with merge states
             return _dirstatecopies(b._repo, match)
 
         cm = _committedforwardcopies(a, b.p1(), match)
         # combine copies from dirstate if necessary
         return _chainandfilter(a, b, cm, _dirstatecopies(b._repo, match))
     return _committedforwardcopies(a, b, match)
 
 def _backwardrenames(a, b, match):
     if a._repo.ui.config('experimental', 'copytrace') == 'off':
         return {}
 
     # Even though we're not taking copies into account, 1:n rename situations
     # can still exist (e.g. hg cp a b; hg mv a c). In those cases we
     # arbitrarily pick one of the renames.
     # We don't want to pass in "match" here, since that would filter
     # the destination by it. Since we're reversing the copies, we want
     # to filter the source instead.
     f = _forwardcopies(b, a)
     r = {}
     for k, v in sorted(f.iteritems()):
         if match and not match(v):
             continue
         # remove copies
         if v in a:
             continue
         r[v] = k
     return r
 
 def pathcopies(x, y, match=None):
     """find {dst@y: src@x} copy mapping for directed compare"""
     repo = x._repo
     debug = repo.ui.debugflag and repo.ui.configbool('devel', 'debug.copies')
     if debug:
         repo.ui.debug('debug.copies: searching copies from %s to %s\n'
                       % (x, y))
     if x == y or not x or not y:
         return {}
     a = y.ancestor(x)
     if a == x:
         if debug:
             repo.ui.debug('debug.copies: search mode: forward\n')
         return _forwardcopies(x, y, match=match)
     if a == y:
         if debug:
             repo.ui.debug('debug.copies: search mode: backward\n')
         return _backwardrenames(x, y, match=match)
     if debug:
         repo.ui.debug('debug.copies: search mode: combined\n')
     return _chainandfilter(x, y, _backwardrenames(x, a, match=match),
                            _forwardcopies(a, y, match=match))
 
 def mergecopies(repo, c1, c2, base):
     """
     Finds moves and copies between context c1 and c2 that are relevant for
     merging. 'base' will be used as the merge base.
 
     Copytracing is used in commands like rebase, merge, unshelve, etc to merge
     files that were moved/ copied in one merge parent and modified in another.
     For example:
 
     o          ---> 4 another commit
     |
     |   o      ---> 3 commit that modifies a.txt
     |  /
     o /        ---> 2 commit that moves a.txt to b.txt
     |/
     o          ---> 1 merge base
 
     If we try to rebase revision 3 on revision 4, since there is no a.txt in
     revision 4, and if user have copytrace disabled, we prints the following
     message:
 
     ```other changed <file> which local deleted```
 
     Returns five dicts: "copy", "movewithdir", "diverge", "renamedelete" and
     "dirmove".
 
     "copy" is a mapping from destination name -> source name,
     where source is in c1 and destination is in c2 or vice-versa.
 
     "movewithdir" is a mapping from source name -> destination name,
     where the file at source present in one context but not the other
     needs to be moved to destination by the merge process, because the
     other context moved the directory it is in.
 
     "diverge" is a mapping of source name -> list of destination names
     for divergent renames.
 
     "renamedelete" is a mapping of source name -> list of destination
     names for files deleted in c1 that were renamed in c2 or vice-versa.
 
     "dirmove" is a mapping of detected source dir -> destination dir renames.
     This is needed for handling changes to new files previously grafted into
     renamed directories.
 
     This function calls different copytracing algorithms based on config.
     """
     # avoid silly behavior for update from empty dir
     if not c1 or not c2 or c1 == c2:
         return {}, {}, {}, {}, {}
 
     narrowmatch = c1.repo().narrowmatch()
 
     # avoid silly behavior for parent -> working dir
     if c2.node() is None and c1.node() == repo.dirstate.p1():
         return _dirstatecopies(repo, narrowmatch), {}, {}, {}, {}
 
     copytracing = repo.ui.config('experimental', 'copytrace')
     if stringutil.parsebool(copytracing) is False:
         # stringutil.parsebool() returns None when it is unable to parse the
         # value, so we should rely on making sure copytracing is on such cases
         return {}, {}, {}, {}, {}
 
     if usechangesetcentricalgo(repo):
         # The heuristics don't make sense when we need changeset-centric algos
         return _fullcopytracing(repo, c1, c2, base)
 
     # Copy trace disabling is explicitly below the node == p1 logic above
     # because the logic above is required for a simple copy to be kept across a
     # rebase.
     if copytracing == 'heuristics':
         # Do full copytracing if only non-public revisions are involved as
         # that will be fast enough and will also cover the copies which could
         # be missed by heuristics
         if _isfullcopytraceable(repo, c1, base):
             return _fullcopytracing(repo, c1, c2, base)
         return _heuristicscopytracing(repo, c1, c2, base)
     else:
         return _fullcopytracing(repo, c1, c2, base)
 
 def _isfullcopytraceable(repo, c1, base):
     """ Checks that if base, source and destination are all no-public branches,
     if yes let's use the full copytrace algorithm for increased capabilities
     since it will be fast enough.
 
     `experimental.copytrace.sourcecommitlimit` can be used to set a limit for
     number of changesets from c1 to base such that if number of changesets are
     more than the limit, full copytracing algorithm won't be used.
     """
     if c1.rev() is None:
         c1 = c1.p1()
     if c1.mutable() and base.mutable():
         sourcecommitlimit = repo.ui.configint('experimental',
                                               'copytrace.sourcecommitlimit')
         commits = len(repo.revs('%d::%d', base.rev(), c1.rev()))
         return commits < sourcecommitlimit
     return False
 
 def _checksinglesidecopies(src, dsts1, m1, m2, mb, c2, base,
                            copy, renamedelete):
     if src not in m2:
         # deleted on side 2
         if src not in m1:
             # renamed on side 1, deleted on side 2
             renamedelete[src] = dsts1
     elif m2[src] != mb[src]:
         if not _related(c2[src], base[src]):
             return
         # modified on side 2
         for dst in dsts1:
             if dst not in m2:
                 # dst not added on side 2 (handle as regular
                 # "both created" case in manifestmerge otherwise)
                 copy[dst] = src
 
 def _fullcopytracing(repo, c1, c2, base):
     """ The full copytracing algorithm which finds all the new files that were
     added from merge base up to the top commit and for each file it checks if
     this file was copied from another file.
 
     This is pretty slow when a lot of changesets are involved but will track all
     the copies.
     """
     m1 = c1.manifest()
     m2 = c2.manifest()
     mb = base.manifest()
 
     copies1 = pathcopies(base, c1)
     copies2 = pathcopies(base, c2)
 
     inversecopies1 = {}
     inversecopies2 = {}
     for dst, src in copies1.items():
         inversecopies1.setdefault(src, []).append(dst)
     for dst, src in copies2.items():
         inversecopies2.setdefault(src, []).append(dst)
 
     copy = {}
     diverge = {}
     renamedelete = {}
     allsources = set(inversecopies1) | set(inversecopies2)
     for src in allsources:
         dsts1 = inversecopies1.get(src)
         dsts2 = inversecopies2.get(src)
         if dsts1 and dsts2:
             # copied/renamed on both sides
             if src not in m1 and src not in m2:
                 # renamed on both sides
                 dsts1 = set(dsts1)
                 dsts2 = set(dsts2)
                 # If there's some overlap in the rename destinations, we
                 # consider it not divergent. For example, if side 1 copies 'a'
                 # to 'b' and 'c' and deletes 'a', and side 2 copies 'a' to 'c'
                 # and 'd' and deletes 'a'.
                 if dsts1 & dsts2:
                     for dst in (dsts1 & dsts2):
                         copy[dst] = src
                 else:
                     diverge[src] = sorted(dsts1 | dsts2)
             elif src in m1 and src in m2:
                 # copied on both sides
                 dsts1 = set(dsts1)
                 dsts2 = set(dsts2)
                 for dst in (dsts1 & dsts2):
                     copy[dst] = src
             # TODO: Handle cases where it was renamed on one side and copied
             # on the other side
         elif dsts1:
             # copied/renamed only on side 1
             _checksinglesidecopies(src, dsts1, m1, m2, mb, c2, base,
                                    copy, renamedelete)
         elif dsts2:
             # copied/renamed only on side 2
             _checksinglesidecopies(src, dsts2, m2, m1, mb, c1, base,
                                    copy, renamedelete)
 
     renamedeleteset = set()
     divergeset = set()
     for dsts in diverge.values():
         divergeset.update(dsts)
     for dsts in renamedelete.values():
         renamedeleteset.update(dsts)
 
     # find interesting file sets from manifests
     addedinm1 = m1.filesnotin(mb, repo.narrowmatch())
     addedinm2 = m2.filesnotin(mb, repo.narrowmatch())
     u1 = sorted(addedinm1 - addedinm2)
     u2 = sorted(addedinm2 - addedinm1)
 
     header = "  unmatched files in %s"
     if u1:
         repo.ui.debug("%s:\n   %s\n" % (header % 'local', "\n   ".join(u1)))
     if u2:
         repo.ui.debug("%s:\n   %s\n" % (header % 'other', "\n   ".join(u2)))
 
     fullcopy = copies1.copy()
     fullcopy.update(copies2)
     if not fullcopy:
         return copy, {}, diverge, renamedelete, {}
 
     if repo.ui.debugflag:
         repo.ui.debug("  all copies found (* = to merge, ! = divergent, "
                       "% = renamed and deleted):\n")
         for f in sorted(fullcopy):
             note = ""
             if f in copy:
                 note += "*"
             if f in divergeset:
                 note += "!"
             if f in renamedeleteset:
                 note += "%"
             repo.ui.debug("   src: '%s' -> dst: '%s' %s\n" % (fullcopy[f], f,
                                                               note))
     del divergeset
 
     repo.ui.debug("  checking for directory renames\n")
 
     # generate a directory move map
     d1, d2 = c1.dirs(), c2.dirs()
     invalid = set()
     dirmove = {}
 
     # examine each file copy for a potential directory move, which is
     # when all the files in a directory are moved to a new directory
     for dst, src in fullcopy.iteritems():
         dsrc, ddst = pathutil.dirname(src), pathutil.dirname(dst)
         if dsrc in invalid:
             # already seen to be uninteresting
             continue
         elif dsrc in d1 and ddst in d1:
             # directory wasn't entirely moved locally
             invalid.add(dsrc)
         elif dsrc in d2 and ddst in d2:
             # directory wasn't entirely moved remotely
             invalid.add(dsrc)
         elif dsrc in dirmove and dirmove[dsrc] != ddst:
             # files from the same directory moved to two different places
             invalid.add(dsrc)
         else:
             # looks good so far
             dirmove[dsrc] = ddst
 
     for i in invalid:
         if i in dirmove:
             del dirmove[i]
     del d1, d2, invalid
 
     if not dirmove:
         return copy, {}, diverge, renamedelete, {}
 
     dirmove = {k + "/": v + "/" for k, v in dirmove.iteritems()}
 
     for d in dirmove:
         repo.ui.debug("   discovered dir src: '%s' -> dst: '%s'\n" %
                       (d, dirmove[d]))
 
     movewithdir = {}
     # check unaccounted nonoverlapping files against directory moves
     for f in u1 + u2:
         if f not in fullcopy:
             for d in dirmove:
                 if f.startswith(d):
                     # new file added in a directory that was moved, move it
                     df = dirmove[d] + f[len(d):]
                     if df not in copy:
                         movewithdir[f] = df
                         repo.ui.debug(("   pending file src: '%s' -> "
                                        "dst: '%s'\n") % (f, df))
                     break
 
     return copy, movewithdir, diverge, renamedelete, dirmove
 
 def _heuristicscopytracing(repo, c1, c2, base):
     """ Fast copytracing using filename heuristics
 
     Assumes that moves or renames are of following two types:
 
     1) Inside a directory only (same directory name but different filenames)
     2) Move from one directory to another
                     (same filenames but different directory names)
 
     Works only when there are no merge commits in the "source branch".
     Source branch is commits from base up to c2 not including base.
 
     If merge is involved it fallbacks to _fullcopytracing().
 
     Can be used by setting the following config:
 
         [experimental]
         copytrace = heuristics
 
     In some cases the copy/move candidates found by heuristics can be very large
     in number and that will make the algorithm slow. The number of possible
     candidates to check can be limited by using the config
     `experimental.copytrace.movecandidateslimit` which defaults to 100.
     """
 
     if c1.rev() is None:
         c1 = c1.p1()
     if c2.rev() is None:
         c2 = c2.p1()
 
     copies = {}
 
     changedfiles = set()
     m1 = c1.manifest()
     if not repo.revs('%d::%d', base.rev(), c2.rev()):
         # If base is not in c2 branch, we switch to fullcopytracing
         repo.ui.debug("switching to full copytracing as base is not "
                       "an ancestor of c2\n")
         return _fullcopytracing(repo, c1, c2, base)
 
     ctx = c2
     while ctx != base:
         if len(ctx.parents()) == 2:
             # To keep things simple let's not handle merges
             repo.ui.debug("switching to full copytracing because of merges\n")
             return _fullcopytracing(repo, c1, c2, base)
         changedfiles.update(ctx.files())
         ctx = ctx.p1()
 
     cp = _forwardcopies(base, c2)
     for dst, src in cp.iteritems():
         if src in m1:
             copies[dst] = src
 
     # file is missing if it isn't present in the destination, but is present in
     # the base and present in the source.
     # Presence in the base is important to exclude added files, presence in the
     # source is important to exclude removed files.
     filt = lambda f: f not in m1 and f in base and f in c2
     missingfiles = [f for f in changedfiles if filt(f)]
 
     if missingfiles:
         basenametofilename = collections.defaultdict(list)
         dirnametofilename = collections.defaultdict(list)
 
         for f in m1.filesnotin(base.manifest()):
             basename = os.path.basename(f)
             dirname = os.path.dirname(f)
             basenametofilename[basename].append(f)
             dirnametofilename[dirname].append(f)
 
         for f in missingfiles:
             basename = os.path.basename(f)
             dirname = os.path.dirname(f)
             samebasename = basenametofilename[basename]
             samedirname = dirnametofilename[dirname]
             movecandidates = samebasename + samedirname
             # f is guaranteed to be present in c2, that's why
             # c2.filectx(f) won't fail
             f2 = c2.filectx(f)
             # we can have a lot of candidates which can slow down the heuristics
             # config value to limit the number of candidates moves to check
             maxcandidates = repo.ui.configint('experimental',
                                               'copytrace.movecandidateslimit')
 
             if len(movecandidates) > maxcandidates:
                 repo.ui.status(_("skipping copytracing for '%s', more "
                                  "candidates than the limit: %d\n")
                                % (f, len(movecandidates)))
                 continue
 
             for candidate in movecandidates:
                 f1 = c1.filectx(candidate)
                 if _related(f1, f2):
                     # if there are a few related copies then we'll merge
                     # changes into all of them. This matches the behaviour
                     # of upstream copytracing
                     copies[candidate] = f
 
     return copies, {}, {}, {}, {}
 
 def _related(f1, f2):
     """return True if f1 and f2 filectx have a common ancestor
 
     Walk back to common ancestor to see if the two files originate
     from the same file. Since workingfilectx's rev() is None it messes
     up the integer comparison logic, hence the pre-step check for
     None (f1 and f2 can only be workingfilectx's initially).
     """
 
     if f1 == f2:
         return True # a match
 
     g1, g2 = f1.ancestors(), f2.ancestors()
     try:
         f1r, f2r = f1.linkrev(), f2.linkrev()
 
         if f1r is None:
             f1 = next(g1)
         if f2r is None:
             f2 = next(g2)
 
         while True:
             f1r, f2r = f1.linkrev(), f2.linkrev()
             if f1r > f2r:
                 f1 = next(g1)
             elif f2r > f1r:
                 f2 = next(g2)
             else: # f1 and f2 point to files in the same linkrev
                 return f1 == f2 # true if they point to the same file
     except StopIteration:
         return False
 
 def duplicatecopies(repo, wctx, rev, fromrev, skiprev=None):
     """reproduce copies from fromrev to rev in the dirstate
 
     If skiprev is specified, it's a revision that should be used to
     filter copy records. Any copies that occur between fromrev and
     skiprev will not be duplicated, even if they appear in the set of
     copies between fromrev and rev.
     """
     exclude = {}
     ctraceconfig = repo.ui.config('experimental', 'copytrace')
     bctrace = stringutil.parsebool(ctraceconfig)
     if (skiprev is not None and
         (ctraceconfig == 'heuristics' or bctrace or bctrace is None)):
         # copytrace='off' skips this line, but not the entire function because
         # the line below is O(size of the repo) during a rebase, while the rest
         # of the function is much faster (and is required for carrying copy
         # metadata across the rebase anyway).
         exclude = pathcopies(repo[fromrev], repo[skiprev])
     for dst, src in pathcopies(repo[fromrev], repo[rev]).iteritems():
         if dst in exclude:
             continue
         if dst in wctx:
             wctx[dst].markcopied(src)

tests/test-copies.t

@@ -1,647 +1,652 b''
 #testcases filelog compatibility changeset
 
   $ cat >> $HGRCPATH << EOF
   > [extensions]
   > rebase=
   > [alias]
   > l = log -G -T '{rev} {desc}\n{files}\n'
   > EOF
 
 #if compatibility
   $ cat >> $HGRCPATH << EOF
   > [experimental]
   > copies.read-from = compatibility
   > EOF
 #endif
 
 #if changeset
   $ cat >> $HGRCPATH << EOF
   > [experimental]
   > copies.read-from = changeset-only
   > copies.write-to = changeset-only
   > EOF
 #endif
 
   $ REPONUM=0
   $ newrepo() {
   >     cd $TESTTMP
   >     REPONUM=`expr $REPONUM + 1`
   >     hg init repo-$REPONUM
   >     cd repo-$REPONUM
   > }
 
 Simple rename case
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg mv x y
   $ hg debugp1copies
   x -> y
   $ hg debugp2copies
   $ hg ci -m 'rename x to y'
   $ hg l
   @  1 rename x to y
   |  x y
   o  0 add x
      x
   $ hg debugp1copies -r 1
   x -> y
   $ hg debugpathcopies 0 1
   x -> y
   $ hg debugpathcopies 1 0
   y -> x
 Test filtering copies by path. We do filtering by destination.
   $ hg debugpathcopies 0 1 x
   $ hg debugpathcopies 1 0 x
   y -> x
   $ hg debugpathcopies 0 1 y
   x -> y
   $ hg debugpathcopies 1 0 y
 
 Copy a file onto another file
   $ newrepo
   $ echo x > x
   $ echo y > y
   $ hg ci -Aqm 'add x and y'
   $ hg cp -f x y
   $ hg debugp1copies
   x -> y
   $ hg debugp2copies
   $ hg ci -m 'copy x onto y'
   $ hg l
   @  1 copy x onto y
   |  y
   o  0 add x and y
      x y
   $ hg debugp1copies -r 1
   x -> y
 Incorrectly doesn't show the rename
   $ hg debugpathcopies 0 1
 
 Copy a file onto another file with same content. If metadata is stored in changeset, this does not
 produce a new filelog entry. The changeset's "files" entry should still list the file.
   $ newrepo
   $ echo x > x
   $ echo x > x2
   $ hg ci -Aqm 'add x and x2 with same content'
   $ hg cp -f x x2
   $ hg ci -m 'copy x onto x2'
   $ hg l
   @  1 copy x onto x2
   |  x2
   o  0 add x and x2 with same content
      x x2
   $ hg debugp1copies -r 1
   x -> x2
 Incorrectly doesn't show the rename
   $ hg debugpathcopies 0 1
 
 Copy a file, then delete destination, then copy again. This does not create a new filelog entry.
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg cp x y
   $ hg ci -m 'copy x to y'
   $ hg rm y
   $ hg ci -m 'remove y'
   $ hg cp -f x y
   $ hg ci -m 'copy x onto y (again)'
   $ hg l
   @  3 copy x onto y (again)
   |  y
   o  2 remove y
   |  y
   o  1 copy x to y
   |  y
   o  0 add x
      x
   $ hg debugp1copies -r 3
   x -> y
   $ hg debugpathcopies 0 3
   x -> y
 
 Rename file in a loop: x->y->z->x
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg mv x y
   $ hg debugp1copies
   x -> y
   $ hg debugp2copies
   $ hg ci -m 'rename x to y'
   $ hg mv y z
   $ hg ci -m 'rename y to z'
   $ hg mv z x
   $ hg ci -m 'rename z to x'
   $ hg l
   @  3 rename z to x
   |  x z
   o  2 rename y to z
   |  y z
   o  1 rename x to y
   |  x y
   o  0 add x
      x
   $ hg debugpathcopies 0 3
 
 Copy x to y, then remove y, then add back y. With copy metadata in the changeset, this could easily
 end up reporting y as copied from x (if we don't unmark it as a copy when it's removed).
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg mv x y
   $ hg ci -m 'rename x to y'
   $ hg rm y
   $ hg ci -qm 'remove y'
   $ echo x > y
   $ hg ci -Aqm 'add back y'
   $ hg l
   @  3 add back y
   |  y
   o  2 remove y
   |  y
   o  1 rename x to y
   |  x y
   o  0 add x
      x
   $ hg debugp1copies -r 3
   $ hg debugpathcopies 0 3
 
 Copy x to z, then remove z, then copy x2 (same content as x) to z. With copy metadata in the
 changeset, the two copies here will have the same filelog entry, so ctx['z'].introrev() might point
 to the first commit that added the file. We should still report the copy as being from x2.
   $ newrepo
   $ echo x > x
   $ echo x > x2
   $ hg ci -Aqm 'add x and x2 with same content'
   $ hg cp x z
   $ hg ci -qm 'copy x to z'
   $ hg rm z
   $ hg ci -m 'remove z'
   $ hg cp x2 z
   $ hg ci -m 'copy x2 to z'
   $ hg l
   @  3 copy x2 to z
   |  z
   o  2 remove z
   |  z
   o  1 copy x to z
   |  z
   o  0 add x and x2 with same content
      x x2
   $ hg debugp1copies -r 3
   x2 -> z
   $ hg debugpathcopies 0 3
   x2 -> z
 
 Create x and y, then rename them both to the same name, but on different sides of a fork
   $ newrepo
   $ echo x > x
   $ echo y > y
   $ hg ci -Aqm 'add x and y'
   $ hg mv x z
   $ hg ci -qm 'rename x to z'
   $ hg co -q 0
   $ hg mv y z
   $ hg ci -qm 'rename y to z'
   $ hg l
   @  2 rename y to z
   |  y z
   | o  1 rename x to z
   |/   x z
   o  0 add x and y
      x y
   $ hg debugpathcopies 1 2
   z -> x
   y -> z
 
 Fork renames x to y on one side and removes x on the other
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg mv x y
   $ hg ci -m 'rename x to y'
   $ hg co -q 0
   $ hg rm x
   $ hg ci -m 'remove x'
   created new head
   $ hg l
   @  2 remove x
   |  x
   | o  1 rename x to y
   |/   x y
   o  0 add x
      x
   $ hg debugpathcopies 1 2
 
 Copies via null revision (there shouldn't be any)
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg cp x y
   $ hg ci -m 'copy x to y'
   $ hg co -q null
   $ echo x > x
   $ hg ci -Aqm 'add x (again)'
   $ hg l
   @  2 add x (again)
      x
   o  1 copy x to y
   |  y
   o  0 add x
      x
   $ hg debugpathcopies 1 2
   $ hg debugpathcopies 2 1
 
 Merge rename from other branch
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg mv x y
   $ hg ci -m 'rename x to y'
   $ hg co -q 0
   $ echo z > z
   $ hg ci -Aqm 'add z'
   $ hg merge -q 1
   $ hg debugp1copies
   $ hg debugp2copies
   $ hg ci -m 'merge rename from p2'
   $ hg l
   @    3 merge rename from p2
   |\   x
   | o  2 add z
   | |  z
   o |  1 rename x to y
   |/   x y
   o  0 add x
      x
 Perhaps we should indicate the rename here, but `hg status` is documented to be weird during
 merges, so...
   $ hg debugp1copies -r 3
   $ hg debugp2copies -r 3
   $ hg debugpathcopies 0 3
   x -> y
   $ hg debugpathcopies 1 2
   y -> x
   $ hg debugpathcopies 1 3
   $ hg debugpathcopies 2 3
   x -> y
 
 Copy file from either side in a merge
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg co -q null
   $ echo y > y
   $ hg ci -Aqm 'add y'
   $ hg merge -q 0
   $ hg cp y z
   $ hg debugp1copies
   y -> z
   $ hg debugp2copies
   $ hg ci -m 'copy file from p1 in merge'
   $ hg co -q 1
   $ hg merge -q 0
   $ hg cp x z
   $ hg debugp1copies
   $ hg debugp2copies
   x -> z
   $ hg ci -qm 'copy file from p2 in merge'
   $ hg l
   @    3 copy file from p2 in merge
   |\   z
   +---o  2 copy file from p1 in merge
   | |/   z
   | o  1 add y
   |    y
   o  0 add x
      x
   $ hg debugp1copies -r 2
   y -> z
   $ hg debugp2copies -r 2
   $ hg debugpathcopies 1 2
   y -> z
   $ hg debugpathcopies 0 2
   $ hg debugp1copies -r 3
   $ hg debugp2copies -r 3
   x -> z
   $ hg debugpathcopies 1 3
   $ hg debugpathcopies 0 3
   x -> z
 
 Copy file that exists on both sides of the merge, same content on both sides
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x on branch 1'
   $ hg co -q null
   $ echo x > x
   $ hg ci -Aqm 'add x on branch 2'
   $ hg merge -q 0
   $ hg cp x z
   $ hg debugp1copies
   x -> z
   $ hg debugp2copies
   $ hg ci -qm 'merge'
   $ hg l
   @    2 merge
   |\   z
   | o  1 add x on branch 2
   |    x
   o  0 add x on branch 1
      x
   $ hg debugp1copies -r 2
   x -> z
   $ hg debugp2copies -r 2
 It's a little weird that it shows up on both sides
   $ hg debugpathcopies 1 2
   x -> z
   $ hg debugpathcopies 0 2
   x -> z (filelog !)
 
 Copy file that exists on both sides of the merge, different content
   $ newrepo
   $ echo branch1 > x
   $ hg ci -Aqm 'add x on branch 1'
   $ hg co -q null
   $ echo branch2 > x
   $ hg ci -Aqm 'add x on branch 2'
   $ hg merge -q 0
   warning: conflicts while merging x! (edit, then use 'hg resolve --mark')
   [1]
   $ echo resolved > x
   $ hg resolve -m x
   (no more unresolved files)
   $ hg cp x z
   $ hg debugp1copies
   x -> z
   $ hg debugp2copies
   $ hg ci -qm 'merge'
   $ hg l
   @    2 merge
   |\   x z
   | o  1 add x on branch 2
   |    x
   o  0 add x on branch 1
      x
   $ hg debugp1copies -r 2
   x -> z (changeset !)
   $ hg debugp2copies -r 2
   x -> z (no-changeset !)
   $ hg debugpathcopies 1 2
   x -> z (changeset !)
   $ hg debugpathcopies 0 2
   x -> z (no-changeset !)
 
 Copy x->y on one side of merge and copy x->z on the other side. Pathcopies from one parent
 of the merge to the merge should include the copy from the other side.
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg cp x y
   $ hg ci -qm 'copy x to y'
   $ hg co -q 0
   $ hg cp x z
   $ hg ci -qm 'copy x to z'
   $ hg merge -q 1
   $ hg ci -m 'merge copy x->y and copy x->z'
   $ hg l
   @    3 merge copy x->y and copy x->z
   |\
   | o  2 copy x to z
   | |  z
   o |  1 copy x to y
   |/   y
   o  0 add x
      x
   $ hg debugp1copies -r 3
   $ hg debugp2copies -r 3
   $ hg debugpathcopies 2 3
   x -> y
   $ hg debugpathcopies 1 3
   x -> z
 
 Copy x to y on one side of merge, create y and rename to z on the other side. Pathcopies from the
 first side should not include the y->z rename since y didn't exist in the merge base.
   $ newrepo
   $ echo x > x
   $ hg ci -Aqm 'add x'
   $ hg cp x y
   $ hg ci -qm 'copy x to y'
   $ hg co -q 0
   $ echo y > y
   $ hg ci -Aqm 'add y'
   $ hg mv y z
   $ hg ci -m 'rename y to z'
   $ hg merge -q 1
   $ hg ci -m 'merge'
   $ hg l
   @    4 merge
   |\
   | o  3 rename y to z
   | |  y z
   | o  2 add y
   | |  y
   o |  1 copy x to y
   |/   y
   o  0 add x
      x
   $ hg debugp1copies -r 3
   y -> z
   $ hg debugp2copies -r 3
   $ hg debugpathcopies 2 3
   y -> z
   $ hg debugpathcopies 1 3
 
-Create x and y, then rename x to z on one side of merge, and rename y to z and modify z on the
-other side.
+Create x and y, then rename x to z on one side of merge, and rename y to z and
+modify z on the other side. When storing copies in the changeset, we don't
+filter out copies whose target was created on the other side of the merge.
   $ newrepo
   $ echo x > x
   $ echo y > y
   $ hg ci -Aqm 'add x and y'
   $ hg mv x z
   $ hg ci -qm 'rename x to z'
   $ hg co -q 0
   $ hg mv y z
   $ hg ci -qm 'rename y to z'
   $ echo z >> z
   $ hg ci -m 'modify z'
   $ hg merge -q 1
   warning: conflicts while merging z! (edit, then use 'hg resolve --mark')
   [1]
   $ echo z > z
   $ hg resolve -qm z
   $ hg ci -m 'merge 1 into 3'
 Try merging the other direction too
   $ hg co -q 1
   $ hg merge -q 3
   warning: conflicts while merging z! (edit, then use 'hg resolve --mark')
   [1]
   $ echo z > z
   $ hg resolve -qm z
   $ hg ci -m 'merge 3 into 1'
   created new head
   $ hg l
   @    5 merge 3 into 1
   |\   y z
   +---o  4 merge 1 into 3
   | |/   x z
   | o  3 modify z
   | |  z
   | o  2 rename y to z
   | |  y z
   o |  1 rename x to z
   |/   x z
   o  0 add x and y
      x y
   $ hg debugpathcopies 1 4
+  y -> z (no-filelog !)
   $ hg debugpathcopies 2 4
+  x -> z (no-filelog !)
   $ hg debugpathcopies 0 4
   x -> z (filelog !)
   y -> z (compatibility !)
   $ hg debugpathcopies 1 5
+  y -> z (no-filelog !)
   $ hg debugpathcopies 2 5
+  x -> z (no-filelog !)
   $ hg debugpathcopies 0 5
   x -> z
 
 
 Test for a case in fullcopytracing algorithm where neither of the merging csets
 is a descendant of the merge base. This test reflects that the algorithm
 correctly finds the copies:
 
   $ cat >> $HGRCPATH << EOF
   > [experimental]
   > evolution.createmarkers=True
   > evolution.allowunstable=True
   > EOF
 
   $ newrepo
   $ echo a > a
   $ hg add a
   $ hg ci -m "added a"
   $ echo b > b
   $ hg add b
   $ hg ci -m "added b"
 
   $ hg mv b b1
   $ hg ci -m "rename b to b1"
 
   $ hg up ".^"
   1 files updated, 0 files merged, 1 files removed, 0 files unresolved
   $ echo d > d
   $ hg add d
   $ hg ci -m "added d"
   created new head
 
   $ echo baba >> b
   $ hg ci --amend -m "added d, modified b"
 
   $ hg l --hidden
   @  4 added d, modified b
   |  b d
   | x  3 added d
   |/   d
   | o  2 rename b to b1
   |/   b b1
   o  1 added b
   |  b
   o  0 added a
      a
 
 Grafting revision 4 on top of revision 2, showing that it respect the rename:
 
   $ hg up 2 -q
   $ hg graft -r 4 --base 3 --hidden
   grafting 4:af28412ec03c "added d, modified b" (tip) (no-changeset !)
   grafting 4:6325ca0b7a1c "added d, modified b" (tip) (changeset !)
   merging b1 and b to b1
 
   $ hg l -l1 -p
   @  5 added d, modified b
   |  b1
   ~  diff -r 5a4825cc2926 -r 94a2f1a0e8e2 b1 (no-changeset !)
   ~  diff -r 0a0ed3b3251c -r d544fb655520 b1 (changeset !)
      --- a/b1	Thu Jan 01 00:00:00 1970 +0000
      +++ b/b1	Thu Jan 01 00:00:00 1970 +0000
      @@ -1,1 +1,2 @@
       b
      +baba
   
 Test to make sure that fullcopytracing algorithm doesn't fail when neither of the
 merging csets is a descendant of the base.
 -------------------------------------------------------------------------------------------------
 
   $ newrepo
   $ echo a > a
   $ hg add a
   $ hg ci -m "added a"
   $ echo b > b
   $ hg add b
   $ hg ci -m "added b"
 
   $ echo foobar > willconflict
   $ hg add willconflict
   $ hg ci -m "added willconflict"
   $ echo c > c
   $ hg add c
   $ hg ci -m "added c"
 
   $ hg l
   @  3 added c
   |  c
   o  2 added willconflict
   |  willconflict
   o  1 added b
   |  b
   o  0 added a
      a
 
   $ hg up ".^^"
   0 files updated, 0 files merged, 2 files removed, 0 files unresolved
   $ echo d > d
   $ hg add d
   $ hg ci -m "added d"
   created new head
 
   $ echo barfoo > willconflict
   $ hg add willconflict
   $ hg ci --amend -m "added willconflict and d"
 
   $ hg l
   @  5 added willconflict and d
   |  d willconflict
   | o  3 added c
   | |  c
   | o  2 added willconflict
   |/   willconflict
   o  1 added b
   |  b
   o  0 added a
      a
 
   $ hg rebase -r . -d 2 -t :other
   rebasing 5:5018b1509e94 "added willconflict and d" (tip) (no-changeset !)
   rebasing 5:af8d273bf580 "added willconflict and d" (tip) (changeset !)
 
   $ hg up 3 -q
   $ hg l --hidden
   o  6 added willconflict and d
   |  d willconflict
   | x  5 added willconflict and d
   | |  d willconflict
   | | x  4 added d
   | |/   d
   +---@  3 added c
   | |    c
   o |  2 added willconflict
   |/   willconflict
   o  1 added b
   |  b
   o  0 added a
      a
 
 Now if we trigger a merge between revision 3 and 6 using base revision 4,
 neither of the merging csets will be a descendant of the base revision:
 
   $ hg graft -r 6 --base 4 --hidden -t :other
   grafting 6:99802e4f1e46 "added willconflict and d" (tip) (no-changeset !)
   grafting 6:b19f0df72728 "added willconflict and d" (tip) (changeset !)