revlog.addrevision(): move computation of nodeid in addrevision()
The check "if node in nodemap" is already done earlier in addgroup().
# revset.py - revision set queries for mercurial
#
# Copyright 2010 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
import re
import parser, util, error, discovery
import match as _match
from i18n import _
elements = {
"(": (20, ("group", 1, ")"), ("func", 1, ")")),
"-": (19, ("negate", 19), ("minus", 19)),
"::": (17, ("dagrangepre", 17), ("dagrange", 17),
("dagrangepost", 17)),
"..": (17, ("dagrangepre", 17), ("dagrange", 17),
("dagrangepost", 17)),
":": (15, ("rangepre", 15), ("range", 15), ("rangepost", 15)),
"not": (10, ("not", 10)),
"!": (10, ("not", 10)),
"and": (5, None, ("and", 5)),
"&": (5, None, ("and", 5)),
"or": (4, None, ("or", 4)),
"|": (4, None, ("or", 4)),
"+": (4, None, ("or", 4)),
",": (2, None, ("list", 2)),
")": (0, None, None),
"symbol": (0, ("symbol",), None),
"string": (0, ("string",), None),
"end": (0, None, None),
}
keywords = set(['and', 'or', 'not'])
def tokenize(program):
pos, l = 0, len(program)
while pos < l:
c = program[pos]
if c.isspace(): # skip inter-token whitespace
pass
elif c == ':' and program[pos:pos + 2] == '::': # look ahead carefully
yield ('::', None, pos)
pos += 1 # skip ahead
elif c == '.' and program[pos:pos + 2] == '..': # look ahead carefully
yield ('..', None, pos)
pos += 1 # skip ahead
elif c in "():,-|&+!": # handle simple operators
yield (c, None, pos)
elif c in '"\'': # handle quoted strings
pos += 1
s = pos
while pos < l: # find closing quote
d = program[pos]
if d == '\\': # skip over escaped characters
pos += 2
continue
if d == c:
yield ('string', program[s:pos].decode('string-escape'), s)
break
pos += 1
else:
raise error.ParseError(_("unterminated string"), s)
elif c.isalnum() or c in '._' or ord(c) > 127: # gather up a symbol/keyword
s = pos
pos += 1
while pos < l: # find end of symbol
d = program[pos]
if not (d.isalnum() or d in "._" or ord(d) > 127):
break
if d == '.' and program[pos - 1] == '.': # special case for ..
pos -= 1
break
pos += 1
sym = program[s:pos]
if sym in keywords: # operator keywords
yield (sym, None, s)
else:
yield ('symbol', sym, s)
pos -= 1
else:
raise error.ParseError(_("syntax error"), pos)
pos += 1
yield ('end', None, pos)
# helpers
def getstring(x, err):
if x and (x[0] == 'string' or x[0] == 'symbol'):
return x[1]
raise error.ParseError(err)
def getlist(x):
if not x:
return []
if x[0] == 'list':
return getlist(x[1]) + [x[2]]
return [x]
def getargs(x, min, max, err):
l = getlist(x)
if len(l) < min or len(l) > max:
raise error.ParseError(err)
return l
def getset(repo, subset, x):
if not x:
raise error.ParseError(_("missing argument"))
return methods[x[0]](repo, subset, *x[1:])
# operator methods
def stringset(repo, subset, x):
x = repo[x].rev()
if x == -1 and len(subset) == len(repo):
return [-1]
if x in subset:
return [x]
return []
def symbolset(repo, subset, x):
if x in symbols:
raise error.ParseError(_("can't use %s here") % x)
return stringset(repo, subset, x)
def rangeset(repo, subset, x, y):
m = getset(repo, subset, x)
if not m:
m = getset(repo, range(len(repo)), x)
n = getset(repo, subset, y)
if not n:
n = getset(repo, range(len(repo)), y)
if not m or not n:
return []
m, n = m[0], n[-1]
if m < n:
r = range(m, n + 1)
else:
r = range(m, n - 1, -1)
s = set(subset)
return [x for x in r if x in s]
def andset(repo, subset, x, y):
return getset(repo, getset(repo, subset, x), y)
def orset(repo, subset, x, y):
s = set(getset(repo, subset, x))
s |= set(getset(repo, [r for r in subset if r not in s], y))
return [r for r in subset if r in s]
def notset(repo, subset, x):
s = set(getset(repo, subset, x))
return [r for r in subset if r not in s]
def listset(repo, subset, a, b):
raise error.ParseError(_("can't use a list in this context"))
def func(repo, subset, a, b):
if a[0] == 'symbol' and a[1] in symbols:
return symbols[a[1]](repo, subset, b)
raise error.ParseError(_("not a function: %s") % a[1])
# functions
def p1(repo, subset, x):
ps = set()
cl = repo.changelog
for r in getset(repo, subset, x):
ps.add(cl.parentrevs(r)[0])
return [r for r in subset if r in ps]
def p2(repo, subset, x):
ps = set()
cl = repo.changelog
for r in getset(repo, subset, x):
ps.add(cl.parentrevs(r)[1])
return [r for r in subset if r in ps]
def parents(repo, subset, x):
ps = set()
cl = repo.changelog
for r in getset(repo, subset, x):
ps.update(cl.parentrevs(r))
return [r for r in subset if r in ps]
def maxrev(repo, subset, x):
s = getset(repo, subset, x)
if s:
m = max(s)
if m in subset:
return [m]
return []
def minrev(repo, subset, x):
s = getset(repo, subset, x)
if s:
m = min(s)
if m in subset:
return [m]
return []
def limit(repo, subset, x):
l = getargs(x, 2, 2, _("limit wants two arguments"))
try:
lim = int(getstring(l[1], _("limit wants a number")))
except ValueError:
raise error.ParseError(_("limit expects a number"))
return getset(repo, subset, l[0])[:lim]
def children(repo, subset, x):
cs = set()
cl = repo.changelog
s = set(getset(repo, subset, x))
for r in xrange(0, len(repo)):
for p in cl.parentrevs(r):
if p in s:
cs.add(r)
return [r for r in subset if r in cs]
def branch(repo, subset, x):
s = getset(repo, range(len(repo)), x)
b = set()
for r in s:
b.add(repo[r].branch())
s = set(s)
return [r for r in subset if r in s or repo[r].branch() in b]
def ancestor(repo, subset, x):
l = getargs(x, 2, 2, _("ancestor wants two arguments"))
r = range(len(repo))
a = getset(repo, r, l[0])
b = getset(repo, r, l[1])
if len(a) != 1 or len(b) != 1:
raise error.ParseError(_("ancestor arguments must be single revisions"))
an = [repo[a[0]].ancestor(repo[b[0]]).rev()]
return [r for r in an if r in subset]
def ancestors(repo, subset, x):
args = getset(repo, range(len(repo)), x)
if not args:
return []
s = set(repo.changelog.ancestors(*args)) | set(args)
return [r for r in subset if r in s]
def descendants(repo, subset, x):
args = getset(repo, range(len(repo)), x)
if not args:
return []
s = set(repo.changelog.descendants(*args)) | set(args)
return [r for r in subset if r in s]
def follow(repo, subset, x):
getargs(x, 0, 0, _("follow takes no arguments"))
p = repo['.'].rev()
s = set(repo.changelog.ancestors(p)) | set([p])
return [r for r in subset if r in s]
def date(repo, subset, x):
ds = getstring(x, _("date wants a string"))
dm = util.matchdate(ds)
return [r for r in subset if dm(repo[r].date()[0])]
def keyword(repo, subset, x):
kw = getstring(x, _("keyword wants a string")).lower()
l = []
for r in subset:
c = repo[r]
t = " ".join(c.files() + [c.user(), c.description()])
if kw in t.lower():
l.append(r)
return l
def grep(repo, subset, x):
gr = re.compile(getstring(x, _("grep wants a string")))
l = []
for r in subset:
c = repo[r]
for e in c.files() + [c.user(), c.description()]:
if gr.search(e):
l.append(r)
continue
return l
def author(repo, subset, x):
n = getstring(x, _("author wants a string")).lower()
return [r for r in subset if n in repo[r].user().lower()]
def hasfile(repo, subset, x):
pat = getstring(x, _("file wants a pattern"))
m = _match.match(repo.root, repo.getcwd(), [pat])
s = []
for r in subset:
for f in repo[r].files():
if m(f):
s.append(r)
continue
return s
def contains(repo, subset, x):
pat = getstring(x, _("contains wants a pattern"))
m = _match.match(repo.root, repo.getcwd(), [pat])
s = []
if m.files() == [pat]:
for r in subset:
if pat in repo[r]:
s.append(r)
continue
else:
for r in subset:
for f in repo[r].manifest():
if m(f):
s.append(r)
continue
return s
def checkstatus(repo, subset, pat, field):
m = _match.match(repo.root, repo.getcwd(), [pat])
s = []
fast = (m.files() == [pat])
for r in subset:
c = repo[r]
if fast:
if pat not in c.files():
continue
else:
for f in c.files():
if m(f):
break
else:
continue
files = repo.status(c.p1().node(), c.node())[field]
if fast:
if pat in files:
s.append(r)
continue
else:
for f in files:
if m(f):
s.append(r)
continue
return s
def modifies(repo, subset, x):
pat = getstring(x, _("modifies wants a pattern"))
return checkstatus(repo, subset, pat, 0)
def adds(repo, subset, x):
pat = getstring(x, _("adds wants a pattern"))
return checkstatus(repo, subset, pat, 1)
def removes(repo, subset, x):
pat = getstring(x, _("removes wants a pattern"))
return checkstatus(repo, subset, pat, 2)
def merge(repo, subset, x):
getargs(x, 0, 0, _("merge takes no arguments"))
cl = repo.changelog
return [r for r in subset if cl.parentrevs(r)[1] != -1]
def closed(repo, subset, x):
getargs(x, 0, 0, _("closed takes no arguments"))
return [r for r in subset if repo[r].extra().get('close')]
def head(repo, subset, x):
getargs(x, 0, 0, _("head takes no arguments"))
hs = set()
for b, ls in repo.branchmap().iteritems():
hs.update(repo[h].rev() for h in ls)
return [r for r in subset if r in hs]
def reverse(repo, subset, x):
l = getset(repo, subset, x)
l.reverse()
return l
def present(repo, subset, x):
try:
return getset(repo, subset, x)
except error.RepoLookupError:
return []
def sort(repo, subset, x):
l = getargs(x, 1, 2, _("sort wants one or two arguments"))
keys = "rev"
if len(l) == 2:
keys = getstring(l[1], _("sort spec must be a string"))
s = l[0]
keys = keys.split()
l = []
def invert(s):
return "".join(chr(255 - ord(c)) for c in s)
for r in getset(repo, subset, s):
c = repo[r]
e = []
for k in keys:
if k == 'rev':
e.append(r)
elif k == '-rev':
e.append(-r)
elif k == 'branch':
e.append(c.branch())
elif k == '-branch':
e.append(invert(c.branch()))
elif k == 'desc':
e.append(c.description())
elif k == '-desc':
e.append(invert(c.description()))
elif k in 'user author':
e.append(c.user())
elif k in '-user -author':
e.append(invert(c.user()))
elif k == 'date':
e.append(c.date()[0])
elif k == '-date':
e.append(-c.date()[0])
else:
raise error.ParseError(_("unknown sort key %r") % k)
e.append(r)
l.append(e)
l.sort()
return [e[-1] for e in l]
def getall(repo, subset, x):
getargs(x, 0, 0, _("all takes no arguments"))
return subset
def heads(repo, subset, x):
s = getset(repo, subset, x)
ps = set(parents(repo, subset, x))
return [r for r in s if r not in ps]
def roots(repo, subset, x):
s = getset(repo, subset, x)
cs = set(children(repo, subset, x))
return [r for r in s if r not in cs]
def outgoing(repo, subset, x):
import hg # avoid start-up nasties
l = getargs(x, 0, 1, _("outgoing wants a repository path"))
dest = l and getstring(l[0], _("outgoing wants a repository path")) or ''
dest = repo.ui.expandpath(dest or 'default-push', dest or 'default')
dest, branches = hg.parseurl(dest)
other = hg.repository(hg.remoteui(repo, {}), dest)
repo.ui.pushbuffer()
o = discovery.findoutgoing(repo, other)
repo.ui.popbuffer()
cl = repo.changelog
o = set([cl.rev(r) for r in repo.changelog.nodesbetween(o, None)[0]])
return [r for r in subset if r in o]
def tagged(repo, subset, x):
getargs(x, 0, 0, _("tagged takes no arguments"))
cl = repo.changelog
s = set([cl.rev(n) for t, n in repo.tagslist() if t != 'tip'])
return [r for r in subset if r in s]
symbols = {
"adds": adds,
"all": getall,
"ancestor": ancestor,
"ancestors": ancestors,
"author": author,
"branch": branch,
"children": children,
"closed": closed,
"contains": contains,
"date": date,
"descendants": descendants,
"file": hasfile,
"follow": follow,
"grep": grep,
"head": head,
"heads": heads,
"keyword": keyword,
"limit": limit,
"max": maxrev,
"min": minrev,
"merge": merge,
"modifies": modifies,
"outgoing": outgoing,
"p1": p1,
"p2": p2,
"parents": parents,
"present": present,
"removes": removes,
"reverse": reverse,
"roots": roots,
"sort": sort,
"tagged": tagged,
"user": author,
}
methods = {
"range": rangeset,
"string": stringset,
"symbol": symbolset,
"and": andset,
"or": orset,
"not": notset,
"list": listset,
"func": func,
}
def optimize(x, small):
if x == None:
return 0, x
smallbonus = 1
if small:
smallbonus = .5
op = x[0]
if op == 'minus':
return optimize(('and', x[1], ('not', x[2])), small)
elif op == 'dagrange':
return optimize(('and', ('func', ('symbol', 'descendants'), x[1]),
('func', ('symbol', 'ancestors'), x[2])), small)
elif op == 'dagrangepre':
return optimize(('func', ('symbol', 'ancestors'), x[1]), small)
elif op == 'dagrangepost':
return optimize(('func', ('symbol', 'descendants'), x[1]), small)
elif op == 'rangepre':
return optimize(('range', ('string', '0'), x[1]), small)
elif op == 'rangepost':
return optimize(('range', x[1], ('string', 'tip')), small)
elif op == 'negate':
return optimize(('string',
'-' + getstring(x[1], _("can't negate that"))), small)
elif op in 'string symbol negate':
return smallbonus, x # single revisions are small
elif op == 'and' or op == 'dagrange':
wa, ta = optimize(x[1], True)
wb, tb = optimize(x[2], True)
w = min(wa, wb)
if wa > wb:
return w, (op, tb, ta)
return w, (op, ta, tb)
elif op == 'or':
wa, ta = optimize(x[1], False)
wb, tb = optimize(x[2], False)
if wb < wa:
wb, wa = wa, wb
return max(wa, wb), (op, ta, tb)
elif op == 'not':
o = optimize(x[1], not small)
return o[0], (op, o[1])
elif op == 'group':
return optimize(x[1], small)
elif op in 'range list':
wa, ta = optimize(x[1], small)
wb, tb = optimize(x[2], small)
return wa + wb, (op, ta, tb)
elif op == 'func':
f = getstring(x[1], _("not a symbol"))
wa, ta = optimize(x[2], small)
if f in "grep date user author keyword branch file":
w = 10 # slow
elif f in "modifies adds removes outgoing":
w = 30 # slower
elif f == "contains":
w = 100 # very slow
elif f == "ancestor":
w = 1 * smallbonus
elif f == "reverse limit":
w = 0
elif f in "sort":
w = 10 # assume most sorts look at changelog
else:
w = 1
return w + wa, (op, x[1], ta)
return 1, x
parse = parser.parser(tokenize, elements).parse
def match(spec):
if not spec:
raise error.ParseError(_("empty query"))
tree = parse(spec)
weight, tree = optimize(tree, True)
def mfunc(repo, subset):
return getset(repo, subset, tree)
return mfunc