##// END OF EJS Templates
revlog: introduce an explicit tracking of what the revlog is about...
revlog: introduce an explicit tracking of what the revlog is about Since the dawn of time, people have been forced to rely to lossy introspection of the index filename to determine what the purpose and role of the revlog they encounter is. This is hacky, error prone, inflexible, abstraction-leaky, <insert-your-own-complaints-here>. In f63299ee7e4d Raphaël introduced a new attribute to track this information: `revlog_kind`. However it is initialized in an odd place and various instances end up not having it set. In addition is only tracking some of the information we end up having to introspect in various pieces of code. So we add a new attribute that holds more data and is more strictly enforced. This work is done in collaboration with Raphaël. The `revlog_kind` one will be removed/adapted in the next changeset. We expect to be able to clean up various existing piece of code and to simplify coming work around the newer revlog format. Differential Revision: https://phab.mercurial-scm.org/D10352

File last commit:

r47506:63a3941d default
r47838:4c041c71 default
Show More
subsetmaker.py
170 lines | 4.8 KiB | text/x-python | PythonLexer
"""revset to select sample of repository
Hopefully this is useful to create interesting discovery cases.
"""
import collections
import random
from mercurial.i18n import _
from mercurial import (
registrar,
revset,
revsetlang,
smartset,
)
revsetpredicate = registrar.revsetpredicate()
@revsetpredicate(b'subsetspec("<spec>")')
def subsetmarkerspec(repo, subset, x):
"""use a shorthand spec as used by search-discovery-case
Supported format are:
- "scratch-count-seed": not scratch(all(), count, "seed")
- "randomantichain-seed": ::randomantichain(all(), "seed")
- "rev-REV": "::REV"
"""
args = revsetlang.getargs(
x, 0, 1, _(b'subsetspec("spec") required an argument')
)
spec = revsetlang.getstring(args[0], _(b"spec should be a string"))
case = spec.split(b'-')
t = case[0]
if t == b'scratch':
spec_revset = b'not scratch(all(), %s, "%s")' % (case[1], case[2])
elif t == b'randomantichain':
spec_revset = b'::randomantichain(all(), "%s")' % case[1]
elif t == b'rev':
spec_revset = b'::%d' % case[1]
else:
assert False, spec
selected = repo.revs(spec_revset)
return selected & subset
@revsetpredicate(b'scratch(REVS, <count>, [seed])')
def scratch(repo, subset, x):
"""randomly remove <count> revision from the repository top
This subset is created by recursively picking changeset starting from the
heads. It can be summarized using the following algorithm::
selected = set()
for i in range(<count>):
unselected = repo.revs("not <selected>")
candidates = repo.revs("heads(<unselected>)")
pick = random.choice(candidates)
selected.add(pick)
"""
m = _(b"scratch expects revisions, count argument and an optional seed")
args = revsetlang.getargs(x, 2, 3, m)
if len(args) == 2:
x, n = args
rand = random
elif len(args) == 3:
x, n, seed = args
seed = revsetlang.getinteger(seed, _(b"seed should be a number"))
rand = random.Random(seed)
else:
assert False
n = revsetlang.getinteger(n, _(b"scratch expects a number"))
selected = set()
heads = set()
children_count = collections.defaultdict(lambda: 0)
parents = repo.changelog._uncheckedparentrevs
baseset = revset.getset(repo, smartset.fullreposet(repo), x)
baseset.sort()
for r in baseset:
heads.add(r)
p1, p2 = parents(r)
if p1 >= 0:
heads.discard(p1)
children_count[p1] += 1
if p2 >= 0:
heads.discard(p2)
children_count[p2] += 1
for h in heads:
assert children_count[h] == 0
selected = set()
for x in range(n):
if not heads:
break
pick = rand.choice(list(heads))
heads.remove(pick)
assert pick not in selected
selected.add(pick)
p1, p2 = parents(pick)
if p1 in children_count:
assert p1 in children_count
children_count[p1] -= 1
assert children_count[p1] >= 0
if children_count[p1] == 0:
assert p1 not in selected, (r, p1)
heads.add(p1)
if p2 in children_count:
assert p2 in children_count
children_count[p2] -= 1
assert children_count[p2] >= 0
if children_count[p2] == 0:
assert p2 not in selected, (r, p2)
heads.add(p2)
return smartset.baseset(selected) & subset
@revsetpredicate(b'randomantichain(REVS, [seed])')
def antichain(repo, subset, x):
"""Pick a random anti-chain in the repository
A antichain is a set of changeset where there isn't any element that is
either a descendant or ancestors of any other element in the set. In other
word, all the elements are independant. It can be summarized with the
following algorithm::
selected = set()
unselected = repo.revs('all()')
while unselected:
pick = random.choice(unselected)
selected.add(pick)
unselected -= repo.revs('::<pick> + <pick>::')
"""
args = revsetlang.getargs(
x, 1, 2, _(b"randomantichain expects revisions and an optional seed")
)
if len(args) == 1:
(x,) = args
rand = random
elif len(args) == 2:
x, seed = args
seed = revsetlang.getinteger(seed, _(b"seed should be a number"))
rand = random.Random(seed)
else:
assert False
selected = set()
baseset = revset.getset(repo, smartset.fullreposet(repo), x)
undecided = baseset
while undecided:
pick = rand.choice(list(undecided))
selected.add(pick)
undecided = repo.revs(
'%ld and not (::%ld or %ld::head())', baseset, selected, selected
)
return smartset.baseset(selected) & subset