This changes the revlog.group and re-implements the localrepo.changeroup

function in terms of it.

revlog.group now takes a list of nodes, and some callback functions
instead of a linkmap.

mercurial/localrepo.py

@@ -1,1420 +1,1562 b''
 # localrepo.py - read/write repository class for mercurial
 #
 # Copyright 2005 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 import struct, os, util
 import filelog, manifest, changelog, dirstate, repo
 from node import *
 from demandload import *
 demandload(globals(), "re lock transaction tempfile stat mdiff errno")
 
 class localrepository:
     def __init__(self, ui, path=None, create=0):
         if not path:
             p = os.getcwd()
             while not os.path.isdir(os.path.join(p, ".hg")):
                 oldp = p
                 p = os.path.dirname(p)
                 if p == oldp: raise repo.RepoError("no repo found")
             path = p
         self.path = os.path.join(path, ".hg")
 
         if not create and not os.path.isdir(self.path):
             raise repo.RepoError("repository %s not found" % self.path)
 
         self.root = os.path.abspath(path)
         self.ui = ui
         self.opener = util.opener(self.path)
         self.wopener = util.opener(self.root)
         self.manifest = manifest.manifest(self.opener)
         self.changelog = changelog.changelog(self.opener)
         self.tagscache = None
         self.nodetagscache = None
         self.encodepats = None
         self.decodepats = None
 
         if create:
             os.mkdir(self.path)
             os.mkdir(self.join("data"))
 
         self.dirstate = dirstate.dirstate(self.opener, ui, self.root)
         try:
             self.ui.readconfig(self.opener("hgrc"))
         except IOError: pass
 
     def hook(self, name, **args):
         s = self.ui.config("hooks", name)
         if s:
             self.ui.note("running hook %s: %s\n" % (name, s))
             old = {}
             for k, v in args.items():
                 k = k.upper()
                 old[k] = os.environ.get(k, None)
                 os.environ[k] = v
 
             # Hooks run in the repository root
             olddir = os.getcwd()
             os.chdir(self.root)
             r = os.system(s)
             os.chdir(olddir)
 
             for k, v in old.items():
                 if v != None:
                     os.environ[k] = v
                 else:
                     del os.environ[k]
 
             if r:
                 self.ui.warn("abort: %s hook failed with status %d!\n" %
                              (name, r))
                 return False
         return True
 
     def tags(self):
         '''return a mapping of tag to node'''
         if not self.tagscache:
             self.tagscache = {}
             def addtag(self, k, n):
                 try:
                     bin_n = bin(n)
                 except TypeError:
                     bin_n = ''
                 self.tagscache[k.strip()] = bin_n
 
             try:
                 # read each head of the tags file, ending with the tip
                 # and add each tag found to the map, with "newer" ones
                 # taking precedence
                 fl = self.file(".hgtags")
                 h = fl.heads()
                 h.reverse()
                 for r in h:
                     for l in fl.read(r).splitlines():
                         if l:
                             n, k = l.split(" ", 1)
                             addtag(self, k, n)
             except KeyError:
                 pass
 
             try:
                 f = self.opener("localtags")
                 for l in f:
                     n, k = l.split(" ", 1)
                     addtag(self, k, n)
             except IOError:
                 pass
 
             self.tagscache['tip'] = self.changelog.tip()
 
         return self.tagscache
 
     def tagslist(self):
         '''return a list of tags ordered by revision'''
         l = []
         for t, n in self.tags().items():
             try:
                 r = self.changelog.rev(n)
             except:
                 r = -2 # sort to the beginning of the list if unknown
             l.append((r,t,n))
         l.sort()
         return [(t,n) for r,t,n in l]
 
     def nodetags(self, node):
         '''return the tags associated with a node'''
         if not self.nodetagscache:
             self.nodetagscache = {}
             for t,n in self.tags().items():
                 self.nodetagscache.setdefault(n,[]).append(t)
         return self.nodetagscache.get(node, [])
 
     def lookup(self, key):
         try:
             return self.tags()[key]
         except KeyError:
             try:
                 return self.changelog.lookup(key)
             except:
                 raise repo.RepoError("unknown revision '%s'" % key)
 
     def dev(self):
         return os.stat(self.path).st_dev
 
     def local(self):
         return True
 
     def join(self, f):
         return os.path.join(self.path, f)
 
     def wjoin(self, f):
         return os.path.join(self.root, f)
 
     def file(self, f):
         if f[0] == '/': f = f[1:]
         return filelog.filelog(self.opener, f)
 
     def getcwd(self):
         return self.dirstate.getcwd()
 
     def wfile(self, f, mode='r'):
         return self.wopener(f, mode)
 
     def wread(self, filename):
         if self.encodepats == None:
             l = []
             for pat, cmd in self.ui.configitems("encode"):
                 mf = util.matcher("", "/", [pat], [], [])[1]
                 l.append((mf, cmd))
             self.encodepats = l
 
         data = self.wopener(filename, 'r').read()
 
         for mf, cmd in self.encodepats:
             if mf(filename):
                 self.ui.debug("filtering %s through %s\n" % (filename, cmd))
                 data = util.filter(data, cmd)
                 break
 
         return data
 
     def wwrite(self, filename, data, fd=None):
         if self.decodepats == None:
             l = []
             for pat, cmd in self.ui.configitems("decode"):
                 mf = util.matcher("", "/", [pat], [], [])[1]
                 l.append((mf, cmd))
             self.decodepats = l
 
         for mf, cmd in self.decodepats:
             if mf(filename):
                 self.ui.debug("filtering %s through %s\n" % (filename, cmd))
                 data = util.filter(data, cmd)
                 break
 
         if fd:
             return fd.write(data)
         return self.wopener(filename, 'w').write(data)
 
     def transaction(self):
         # save dirstate for undo
         try:
             ds = self.opener("dirstate").read()
         except IOError:
             ds = ""
         self.opener("journal.dirstate", "w").write(ds)
 
         def after():
             util.rename(self.join("journal"), self.join("undo"))
             util.rename(self.join("journal.dirstate"),
                         self.join("undo.dirstate"))
 
         return transaction.transaction(self.ui.warn, self.opener,
                                        self.join("journal"), after)
 
     def recover(self):
         lock = self.lock()
         if os.path.exists(self.join("journal")):
             self.ui.status("rolling back interrupted transaction\n")
             return transaction.rollback(self.opener, self.join("journal"))
         else:
             self.ui.warn("no interrupted transaction available\n")
 
     def undo(self):
         lock = self.lock()
         if os.path.exists(self.join("undo")):
             self.ui.status("rolling back last transaction\n")
             transaction.rollback(self.opener, self.join("undo"))
             self.dirstate = None
             util.rename(self.join("undo.dirstate"), self.join("dirstate"))
             self.dirstate = dirstate.dirstate(self.opener, self.ui, self.root)
         else:
             self.ui.warn("no undo information available\n")
 
     def lock(self, wait=1):
         try:
             return lock.lock(self.join("lock"), 0)
         except lock.LockHeld, inst:
             if wait:
                 self.ui.warn("waiting for lock held by %s\n" % inst.args[0])
                 return lock.lock(self.join("lock"), wait)
             raise inst
 
     def rawcommit(self, files, text, user, date, p1=None, p2=None):
         orig_parent = self.dirstate.parents()[0] or nullid
         p1 = p1 or self.dirstate.parents()[0] or nullid
         p2 = p2 or self.dirstate.parents()[1] or nullid
         c1 = self.changelog.read(p1)
         c2 = self.changelog.read(p2)
         m1 = self.manifest.read(c1[0])
         mf1 = self.manifest.readflags(c1[0])
         m2 = self.manifest.read(c2[0])
         changed = []
 
         if orig_parent == p1:
             update_dirstate = 1
         else:
             update_dirstate = 0
 
         tr = self.transaction()
         mm = m1.copy()
         mfm = mf1.copy()
         linkrev = self.changelog.count()
         for f in files:
             try:
                 t = self.wread(f)
                 tm = util.is_exec(self.wjoin(f), mfm.get(f, False))
                 r = self.file(f)
                 mfm[f] = tm
 
                 fp1 = m1.get(f, nullid)
                 fp2 = m2.get(f, nullid)
 
                 # is the same revision on two branches of a merge?
                 if fp2 == fp1:
                     fp2 = nullid
 
                 if fp2 != nullid:
                     # is one parent an ancestor of the other?
                     fpa = r.ancestor(fp1, fp2)
                     if fpa == fp1:
                         fp1, fp2 = fp2, nullid
                     elif fpa == fp2:
                         fp2 = nullid
 
                     # is the file unmodified from the parent?
                     if t == r.read(fp1):
                         # record the proper existing parent in manifest
                         # no need to add a revision
                         mm[f] = fp1
                         continue
 
                 mm[f] = r.add(t, {}, tr, linkrev, fp1, fp2)
                 changed.append(f)
                 if update_dirstate:
                     self.dirstate.update([f], "n")
             except IOError:
                 try:
                     del mm[f]
                     del mfm[f]
                     if update_dirstate:
                         self.dirstate.forget([f])
                 except:
                     # deleted from p2?
                     pass
 
         mnode = self.manifest.add(mm, mfm, tr, linkrev, c1[0], c2[0])
         user = user or self.ui.username()
         n = self.changelog.add(mnode, changed, text, tr, p1, p2, user, date)
         tr.close()
         if update_dirstate:
             self.dirstate.setparents(n, nullid)
 
     def commit(self, files = None, text = "", user = None, date = None,
                match = util.always, force=False):
         commit = []
         remove = []
         changed = []
 
         if files:
             for f in files:
                 s = self.dirstate.state(f)
                 if s in 'nmai':
                     commit.append(f)
                 elif s == 'r':
                     remove.append(f)
                 else:
                     self.ui.warn("%s not tracked!\n" % f)
         else:
             (c, a, d, u) = self.changes(match=match)
             commit = c + a
             remove = d
 
         p1, p2 = self.dirstate.parents()
         c1 = self.changelog.read(p1)
         c2 = self.changelog.read(p2)
         m1 = self.manifest.read(c1[0])
         mf1 = self.manifest.readflags(c1[0])
         m2 = self.manifest.read(c2[0])
 
         if not commit and not remove and not force and p2 == nullid:
             self.ui.status("nothing changed\n")
             return None
 
         if not self.hook("precommit"):
             return None
 
         lock = self.lock()
         tr = self.transaction()
 
         # check in files
         new = {}
         linkrev = self.changelog.count()
         commit.sort()
         for f in commit:
             self.ui.note(f + "\n")
             try:
                 mf1[f] = util.is_exec(self.wjoin(f), mf1.get(f, False))
                 t = self.wread(f)
             except IOError:
                 self.ui.warn("trouble committing %s!\n" % f)
                 raise
 
             r = self.file(f)
 
             meta = {}
             cp = self.dirstate.copied(f)
             if cp:
                 meta["copy"] = cp
                 meta["copyrev"] = hex(m1.get(cp, m2.get(cp, nullid)))
                 self.ui.debug(" %s: copy %s:%s\n" % (f, cp, meta["copyrev"]))
                 fp1, fp2 = nullid, nullid
             else:
                 fp1 = m1.get(f, nullid)
                 fp2 = m2.get(f, nullid)
 
             # is the same revision on two branches of a merge?
             if fp2 == fp1:
                 fp2 = nullid
 
             if fp2 != nullid:
                 # is one parent an ancestor of the other?
                 fpa = r.ancestor(fp1, fp2)
                 if fpa == fp1:
                     fp1, fp2 = fp2, nullid
                 elif fpa == fp2:
                     fp2 = nullid
 
                 # is the file unmodified from the parent?
                 if not meta and t == r.read(fp1):
                     # record the proper existing parent in manifest
                     # no need to add a revision
                     new[f] = fp1
                     continue
 
             new[f] = r.add(t, meta, tr, linkrev, fp1, fp2)
             # remember what we've added so that we can later calculate
             # the files to pull from a set of changesets
             changed.append(f)
 
         # update manifest
         m1.update(new)
         for f in remove:
             if f in m1:
                 del m1[f]
         mn = self.manifest.add(m1, mf1, tr, linkrev, c1[0], c2[0],
                                (new, remove))
 
         # add changeset
         new = new.keys()
         new.sort()
 
         if not text:
             edittext = ""
             if p2 != nullid:
                 edittext += "HG: branch merge\n"
             edittext += "\n" + "HG: manifest hash %s\n" % hex(mn)
             edittext += "".join(["HG: changed %s\n" % f for f in changed])
             edittext += "".join(["HG: removed %s\n" % f for f in remove])
             if not changed and not remove:
                 edittext += "HG: no files changed\n"
             edittext = self.ui.edit(edittext)
             if not edittext.rstrip():
                 return None
             text = edittext
 
         user = user or self.ui.username()
         n = self.changelog.add(mn, changed, text, tr, p1, p2, user, date)
         tr.close()
 
         self.dirstate.setparents(n)
         self.dirstate.update(new, "n")
         self.dirstate.forget(remove)
 
         if not self.hook("commit", node=hex(n)):
             return None
         return n
 
     def walk(self, node=None, files=[], match=util.always):
         if node:
             for fn in self.manifest.read(self.changelog.read(node)[0]):
                 if match(fn): yield 'm', fn
         else:
             for src, fn in self.dirstate.walk(files, match):
                 yield src, fn
 
     def changes(self, node1 = None, node2 = None, files = [],
                 match = util.always):
         mf2, u = None, []
 
         def fcmp(fn, mf):
             t1 = self.wread(fn)
             t2 = self.file(fn).read(mf.get(fn, nullid))
             return cmp(t1, t2)
 
         def mfmatches(node):
             mf = dict(self.manifest.read(node))
             for fn in mf.keys():
                 if not match(fn):
                     del mf[fn]
             return mf
 
         # are we comparing the working directory?
         if not node2:
             l, c, a, d, u = self.dirstate.changes(files, match)
 
             # are we comparing working dir against its parent?
             if not node1:
                 if l:
                     # do a full compare of any files that might have changed
                     change = self.changelog.read(self.dirstate.parents()[0])
                     mf2 = mfmatches(change[0])
                     for f in l:
                         if fcmp(f, mf2):
                             c.append(f)
 
                 for l in c, a, d, u:
                     l.sort()
 
                 return (c, a, d, u)
 
         # are we comparing working dir against non-tip?
         # generate a pseudo-manifest for the working dir
         if not node2:
             if not mf2:
                 change = self.changelog.read(self.dirstate.parents()[0])
                 mf2 = mfmatches(change[0])
             for f in a + c + l:
                 mf2[f] = ""
             for f in d:
                 if f in mf2: del mf2[f]
         else:
             change = self.changelog.read(node2)
             mf2 = mfmatches(change[0])
 
         # flush lists from dirstate before comparing manifests
         c, a = [], []
 
         change = self.changelog.read(node1)
         mf1 = mfmatches(change[0])
 
         for fn in mf2:
             if mf1.has_key(fn):
                 if mf1[fn] != mf2[fn]:
                     if mf2[fn] != "" or fcmp(fn, mf1):
                         c.append(fn)
                 del mf1[fn]
             else:
                 a.append(fn)
 
         d = mf1.keys()
 
         for l in c, a, d, u:
             l.sort()
 
         return (c, a, d, u)
 
     def add(self, list):
         for f in list:
             p = self.wjoin(f)
             if not os.path.exists(p):
                 self.ui.warn("%s does not exist!\n" % f)
             elif not os.path.isfile(p):
                 self.ui.warn("%s not added: only files supported currently\n" % f)
             elif self.dirstate.state(f) in 'an':
                 self.ui.warn("%s already tracked!\n" % f)
             else:
                 self.dirstate.update([f], "a")
 
     def forget(self, list):
         for f in list:
             if self.dirstate.state(f) not in 'ai':
                 self.ui.warn("%s not added!\n" % f)
             else:
                 self.dirstate.forget([f])
 
     def remove(self, list):
         for f in list:
             p = self.wjoin(f)
             if os.path.exists(p):
                 self.ui.warn("%s still exists!\n" % f)
             elif self.dirstate.state(f) == 'a':
                 self.ui.warn("%s never committed!\n" % f)
                 self.dirstate.forget([f])
             elif f not in self.dirstate:
                 self.ui.warn("%s not tracked!\n" % f)
             else:
                 self.dirstate.update([f], "r")
 
     def copy(self, source, dest):
         p = self.wjoin(dest)
         if not os.path.exists(p):
             self.ui.warn("%s does not exist!\n" % dest)
         elif not os.path.isfile(p):
             self.ui.warn("copy failed: %s is not a file\n" % dest)
         else:
             if self.dirstate.state(dest) == '?':
                 self.dirstate.update([dest], "a")
             self.dirstate.copy(source, dest)
 
     def heads(self):
         return self.changelog.heads()
 
     # branchlookup returns a dict giving a list of branches for
     # each head.  A branch is defined as the tag of a node or
     # the branch of the node's parents.  If a node has multiple
     # branch tags, tags are eliminated if they are visible from other
     # branch tags.
     #
     # So, for this graph:  a->b->c->d->e
     #                       \         /
     #                        aa -----/
     # a has tag 2.6.12
     # d has tag 2.6.13
     # e would have branch tags for 2.6.12 and 2.6.13.  Because the node
     # for 2.6.12 can be reached from the node 2.6.13, that is eliminated
     # from the list.
     #
     # It is possible that more than one head will have the same branch tag.
     # callers need to check the result for multiple heads under the same
     # branch tag if that is a problem for them (ie checkout of a specific
     # branch).
     #
     # passing in a specific branch will limit the depth of the search
     # through the parents.  It won't limit the branches returned in the
     # result though.
     def branchlookup(self, heads=None, branch=None):
         if not heads:
             heads = self.heads()
         headt = [ h for h in heads ]
         chlog = self.changelog
         branches = {}
         merges = []
         seenmerge = {}
 
         # traverse the tree once for each head, recording in the branches
         # dict which tags are visible from this head.  The branches
         # dict also records which tags are visible from each tag
         # while we traverse.
         while headt or merges:
             if merges:
                 n, found = merges.pop()
                 visit = [n]
             else:
                 h = headt.pop()
                 visit = [h]
                 found = [h]
                 seen = {}
             while visit:
                 n = visit.pop()
                 if n in seen:
                     continue
                 pp = chlog.parents(n)
                 tags = self.nodetags(n)
                 if tags:
                     for x in tags:
                         if x == 'tip':
                             continue
                         for f in found:
                             branches.setdefault(f, {})[n] = 1
                         branches.setdefault(n, {})[n] = 1
                         break
                     if n not in found:
                         found.append(n)
                     if branch in tags:
                         continue
                 seen[n] = 1
                 if pp[1] != nullid and n not in seenmerge:
                     merges.append((pp[1], [x for x in found]))
                     seenmerge[n] = 1
                 if pp[0] != nullid:
                     visit.append(pp[0])
         # traverse the branches dict, eliminating branch tags from each
         # head that are visible from another branch tag for that head.
         out = {}
         viscache = {}
         for h in heads:
             def visible(node):
                 if node in viscache:
                     return viscache[node]
                 ret = {}
                 visit = [node]
                 while visit:
                     x = visit.pop()
                     if x in viscache:
                         ret.update(viscache[x])
                     elif x not in ret:
                         ret[x] = 1
                         if x in branches:
                             visit[len(visit):] = branches[x].keys()
                 viscache[node] = ret
                 return ret
             if h not in branches:
                 continue
             # O(n^2), but somewhat limited.  This only searches the
             # tags visible from a specific head, not all the tags in the
             # whole repo.
             for b in branches[h]:
                 vis = False
                 for bb in branches[h].keys():
                     if b != bb:
                         if b in visible(bb):
                             vis = True
                             break
                 if not vis:
                     l = out.setdefault(h, [])
                     l[len(l):] = self.nodetags(b)
         return out
 
     def branches(self, nodes):
         if not nodes: nodes = [self.changelog.tip()]
         b = []
         for n in nodes:
             t = n
             while n:
                 p = self.changelog.parents(n)
                 if p[1] != nullid or p[0] == nullid:
                     b.append((t, n, p[0], p[1]))
                     break
                 n = p[0]
         return b
 
     def between(self, pairs):
         r = []
 
         for top, bottom in pairs:
             n, l, i = top, [], 0
             f = 1
 
             while n != bottom:
                 p = self.changelog.parents(n)[0]
                 if i == f:
                     l.append(n)
                     f = f * 2
                 n = p
                 i += 1
 
             r.append(l)
 
         return r
 
     def findincoming(self, remote, base=None, heads=None):
         m = self.changelog.nodemap
         search = []
         fetch = {}
         seen = {}
         seenbranch = {}
         if base == None:
             base = {}
 
         # assume we're closer to the tip than the root
         # and start by examining the heads
         self.ui.status("searching for changes\n")
 
         if not heads:
             heads = remote.heads()
 
         unknown = []
         for h in heads:
             if h not in m:
                 unknown.append(h)
             else:
                 base[h] = 1
 
         if not unknown:
             return None
 
         rep = {}
         reqcnt = 0
 
         # search through remote branches
         # a 'branch' here is a linear segment of history, with four parts:
         # head, root, first parent, second parent
         # (a branch always has two parents (or none) by definition)
         unknown = remote.branches(unknown)
         while unknown:
             r = []
             while unknown:
                 n = unknown.pop(0)
                 if n[0] in seen:
                     continue
 
                 self.ui.debug("examining %s:%s\n" % (short(n[0]), short(n[1])))
                 if n[0] == nullid:
                     break
                 if n in seenbranch:
                     self.ui.debug("branch already found\n")
                     continue
                 if n[1] and n[1] in m: # do we know the base?
                     self.ui.debug("found incomplete branch %s:%s\n"
                                   % (short(n[0]), short(n[1])))
                     search.append(n) # schedule branch range for scanning
                     seenbranch[n] = 1
                 else:
                     if n[1] not in seen and n[1] not in fetch:
                         if n[2] in m and n[3] in m:
                             self.ui.debug("found new changeset %s\n" %
                                           short(n[1]))
                             fetch[n[1]] = 1 # earliest unknown
                             base[n[2]] = 1 # latest known
                             continue
 
                     for a in n[2:4]:
                         if a not in rep:
                             r.append(a)
                             rep[a] = 1
 
                 seen[n[0]] = 1
 
             if r:
                 reqcnt += 1
                 self.ui.debug("request %d: %s\n" %
                             (reqcnt, " ".join(map(short, r))))
                 for p in range(0, len(r), 10):
                     for b in remote.branches(r[p:p+10]):
                         self.ui.debug("received %s:%s\n" %
                                       (short(b[0]), short(b[1])))
                         if b[0] in m:
                             self.ui.debug("found base node %s\n" % short(b[0]))
                             base[b[0]] = 1
                         elif b[0] not in seen:
                             unknown.append(b)
 
         # do binary search on the branches we found
         while search:
             n = search.pop(0)
             reqcnt += 1
             l = remote.between([(n[0], n[1])])[0]
             l.append(n[1])
             p = n[0]
             f = 1
             for i in l:
                 self.ui.debug("narrowing %d:%d %s\n" % (f, len(l), short(i)))
                 if i in m:
                     if f <= 2:
                         self.ui.debug("found new branch changeset %s\n" %
                                           short(p))
                         fetch[p] = 1
                         base[i] = 1
                     else:
                         self.ui.debug("narrowed branch search to %s:%s\n"
                                       % (short(p), short(i)))
                         search.append((p, i))
                     break
                 p, f = i, f * 2
 
         # sanity check our fetch list
         for f in fetch.keys():
             if f in m:
                 raise repo.RepoError("already have changeset " + short(f[:4]))
 
         if base.keys() == [nullid]:
             self.ui.warn("warning: pulling from an unrelated repository!\n")
 
         self.ui.note("found new changesets starting at " +
                      " ".join([short(f) for f in fetch]) + "\n")
 
         self.ui.debug("%d total queries\n" % reqcnt)
 
         return fetch.keys()
 
     def findoutgoing(self, remote, base=None, heads=None):
         if base == None:
             base = {}
             self.findincoming(remote, base, heads)
 
         self.ui.debug("common changesets up to "
                       + " ".join(map(short, base.keys())) + "\n")
 
         remain = dict.fromkeys(self.changelog.nodemap)
 
         # prune everything remote has from the tree
         del remain[nullid]
         remove = base.keys()
         while remove:
             n = remove.pop(0)
             if n in remain:
                 del remain[n]
                 for p in self.changelog.parents(n):
                     remove.append(p)
 
         # find every node whose parents have been pruned
         subset = []
         for n in remain:
             p1, p2 = self.changelog.parents(n)
             if p1 not in remain and p2 not in remain:
                 subset.append(n)
 
         # this is the set of all roots we have to push
         return subset
 
     def pull(self, remote):
         lock = self.lock()
 
         # if we have an empty repo, fetch everything
         if self.changelog.tip() == nullid:
             self.ui.status("requesting all changes\n")
             fetch = [nullid]
         else:
             fetch = self.findincoming(remote)
 
         if not fetch:
             self.ui.status("no changes found\n")
             return 1
 
         cg = remote.changegroup(fetch)
         return self.addchangegroup(cg)
 
     def push(self, remote, force=False):
         lock = remote.lock()
 
         base = {}
         heads = remote.heads()
         inc = self.findincoming(remote, base, heads)
         if not force and inc:
             self.ui.warn("abort: unsynced remote changes!\n")
             self.ui.status("(did you forget to sync? use push -f to force)\n")
             return 1
 
         update = self.findoutgoing(remote, base)
         if not update:
             self.ui.status("no changes found\n")
             return 1
         elif not force:
             if len(heads) < len(self.changelog.heads()):
                 self.ui.warn("abort: push creates new remote branches!\n")
                 self.ui.status("(did you forget to merge?" +
                                " use push -f to force)\n")
                 return 1
 
         cg = self.changegroup(update)
         return remote.addchangegroup(cg)
 
-    def changegroup(self, basenodes):
-        genread = util.chunkbuffer
+    def changegroupsubset(self, bases, heads):
+        cl = self.changelog
+        # msng = missing
+        msng_cl_lst, bases, heads = cl.nodesbetween(basenodes, headnodes)
+        junk = None
+        knownheads = {}
+        for n in basenodes:
+            for p in cl.parents(n):
+                if p != nullid:
+                    knownheads[p] = 1
+        knownheads = knownheads.keys()
+        has_cl_set, junk, junk = cl.nodesbetween(None, knownheads)
+        has_cl_set = dict.fromkeys(hasnodeset)
+
+        mnfst = self.manifest
+        msng_mnfst_set = {}
+        msng_filenode_set = {}
+
+        def identity(x):
+            return x
+
+        def cmp_by_rev_func(revlog):
+            def cmpfunc(a, b):
+                return cmp(revlog.rev(a), revlog.rev(b))
+            return cmpfunc
+
+        def prune_parents(revlog, hasset, msngset):
+            haslst = hasset.keys()
+            haslst.sort(cmp_by_rev_func(revlog))
+            for node in haslst:
+                parentlst = [p for p in revlog.parents(node) if p != nullid]
+                while parentlst:
+                    n = parentlst.pop()
+                    if n not in hasset:
+                        hasset[n] = 1
+                        p = [p for p in revlog.parents(n) if p != nullid]
+                        parentlst.extend(p)
+            for n in hasset:
+                msngset.pop(n, None)
+
+        def manifest_and_file_collector(changedfileset):
+            def collect_manifests_and_files(clnode):
+                c = cl.read(clnode)
+                for f in c[3]:
+                    # This is to make sure we only have one instance of each
+                    # filename string for each filename.
+                    changedfileset.set_default(f, f)
+                msng_mnfst_set.set_default(c[0], clnode)
+            return collect_manifests_and_files
+
+        def prune_manifests():
+            has_mnfst_set = {}
+            for n in msng_mnfst_set:
+                linknode = cl.node(mnfst.linkrev(n))
+                if linknode in has_cl_set:
+                    has_mnfst_set[n] = 1
+            prune_parents(mnfst, has_mnfst_set, msng_mnfst_set)
+
+        def lookup_manifest_link(mnfstnode):
+            return msng_mnfst_set[mnfstnode]
+
+        def filenode_collector(changedfiles):
+            def collect_msng_filenodes(mnfstnode):
+                m = mnfst.read(mnfstnode)
+                for f in changedfiles:
+                    fnode = m.get(f, None)
+                    if fnode is not None:
+                        clnode = msng_mnfst_set[mnfstnode]
+                        ndset = msng_filenode_set.setdefault(f, {})
+                        ndset.set_default(fnode, clnode)
+
+        def prune_filenodes(f, filerevlog):
+            msngset = msng_filenode_set[f]
+            hasset = {}
+            for n in msngset:
+                clnode = cl.node(filerevlog.linkrev(n))
+                if clnode in has_cl_set:
+                    hasset[n] = 1
+            prune_parents(filerevlog, hasset, msngset)
+
+        def lookup_filenode_link_func(fname):
+            msngset = msng_filenode_set[fname]
+            def lookup_filenode_link(fnode):
+                return msngset[fnode]
+            return lookup_filenode_link
 
         def gengroup():
-            nodes = self.changelog.nodesbetween(basenodes)[0]
+            changedfiles = {}
+            group = cl.group(msng_cl_lst, identity,
+                             manifest_and_file_collector(changedfiles))
+            for chnk in group:
+                yield chnk
+            prune_manifests()
+            msng_mnfst_lst = msng_mnfst_set.keys()
+            msng_mnfst_lst.sort(cmp_by_rev_func(mnfst))
+            changedfiles = changedfiles.keys()
+            changedfiles.sort()
+            group = mnfst.group(mnfst, lookup_manifest_link,
+                                filenode_collector(changedfiles))
+            for chnk in group:
+                yield chnk
+            msng_mnfst_lst = None
+            msng_mnfst_set.clear()
+            for fname in changedfiles:
+                filerevlog = self.file(fname)
+                prune_filenodes(fname, filerevlog)
+                msng_filenode_lst = msng_filenode_set[fname].keys()
+                if len(msng_filenode_lst) > 0:
+                    yield struct.pack(">l", len(f) + 4) + f
+                    msng_filenode_lst.sort(cmp_by_rev_func(filerevlog))
+                    group = filerevlog.group(msng_filenode_lst,
+                                             lookup_filenode_link)
+                    for chnk in group:
+                        yield chnk
+                del msng_filenode_set[fname]
+            yield struct.pack(">l", 0)
 
-            # construct the link map
-            linkmap = {}
-            for n in nodes:
-                linkmap[self.changelog.rev(n)] = n
+        return util.chunkbuffer(gengroup())
+
+    def changegroup(self, basenodes):
+        cl = self.changelog
+        nodes = cl.nodesbetween(basenodes, None)[0]
+        revset = dict.fromkeys([cl.rev(n) for n in nodes])
+
+        def identity(x):
+            return x
 
+        def gennodelst(revlog):
+            for r in xrange(0, revlog.count()):
+                n = revlog.node(r)
+                if revlog.linkrev(n) in revset:
+                    yield n
+
+        def changed_file_collector(changedfileset):
+            def collect_changed_files(clnode):
+                c = cl.read(clnode)
+                for fname in c[3]:
+                    changedfileset[fname] = 1
+            return collect_changed_files
+
+        def lookuprevlink_func(revlog):
+            def lookuprevlink(n):
+                return cl.node(revlog.linkrev(n))
+            return lookuprevlink
+
+        def gengroup():
             # construct a list of all changed files
-            changed = {}
-            for n in nodes:
-                c = self.changelog.read(n)
-                for f in c[3]:
-                    changed[f] = 1
-            changed = changed.keys()
-            changed.sort()
+            changedfiles = {}
+
+            for chnk in cl.group(nodes, identity,
+                                 changed_file_collector(changedfiles)):
+                yield chnk
+            changedfiles = changedfiles.keys()
+            changedfiles.sort()
 
-            # the changegroup is changesets + manifests + all file revs
-            revs = [ self.changelog.rev(n) for n in nodes ]
+            mnfst = self.manifest
+            def nodegen(revlog, reviter):
+                for r in reviter:
+                    yield revlog.node(r)
+            nodeiter = gennodelst(mnfst)
+            for chnk in mnfst.group(nodeiter, lookuprevlink_func(mnfst)):
+                yield chnk
 
-            for y in self.changelog.group(linkmap): yield y
-            for y in self.manifest.group(linkmap): yield y
-            for f in changed:
-                yield struct.pack(">l", len(f) + 4) + f
-                g = self.file(f).group(linkmap)
-                for y in g:
-                    yield y
+            for fname in changedfiles:
+                filerevlog = self.file(fname)
+                nodeiter = gennodelst(filerevlog)
+                nodeiter = list(nodeiter)
+                if nodeiter:
+                    yield struct.pack(">l", len(fname) + 4) + fname
+                    lookup = lookuprevlink_func(filerevlog)
+                    for chnk in filerevlog.group(nodeiter, lookup):
+                        yield chnk
 
             yield struct.pack(">l", 0)
 
-        return genread(gengroup())
+        return util.chunkbuffer(gengroup())
 
     def addchangegroup(self, source):
 
         def getchunk():
             d = source.read(4)
             if not d: return ""
             l = struct.unpack(">l", d)[0]
             if l <= 4: return ""
             d = source.read(l - 4)
             if len(d) < l - 4:
                 raise repo.RepoError("premature EOF reading chunk" +
                                      " (got %d bytes, expected %d)"
                                      % (len(d), l - 4))
             return d
 
         def getgroup():
             while 1:
                 c = getchunk()
                 if not c: break
                 yield c
 
         def csmap(x):
             self.ui.debug("add changeset %s\n" % short(x))
             return self.changelog.count()
 
         def revmap(x):
             return self.changelog.rev(x)
 
         if not source: return
         changesets = files = revisions = 0
 
         tr = self.transaction()
 
         oldheads = len(self.changelog.heads())
 
         # pull off the changeset group
         self.ui.status("adding changesets\n")
         co = self.changelog.tip()
         cn = self.changelog.addgroup(getgroup(), csmap, tr, 1) # unique
         cnr, cor = map(self.changelog.rev, (cn, co))
         if cn == nullid:
             cnr = cor
         changesets = cnr - cor
 
         # pull off the manifest group
         self.ui.status("adding manifests\n")
         mm = self.manifest.tip()
         mo = self.manifest.addgroup(getgroup(), revmap, tr)
 
         # process the files
         self.ui.status("adding file changes\n")
         while 1:
             f = getchunk()
             if not f: break
             self.ui.debug("adding %s revisions\n" % f)
             fl = self.file(f)
             o = fl.count()
             n = fl.addgroup(getgroup(), revmap, tr)
             revisions += fl.count() - o
             files += 1
 
         newheads = len(self.changelog.heads())
         heads = ""
         if oldheads and newheads > oldheads:
             heads = " (+%d heads)" % (newheads - oldheads)
 
         self.ui.status(("added %d changesets" +
                         " with %d changes to %d files%s\n")
                        % (changesets, revisions, files, heads))
 
         tr.close()
 
         if changesets > 0:
             if not self.hook("changegroup",
                               node=hex(self.changelog.node(cor+1))):
                 self.ui.warn("abort: changegroup hook returned failure!\n")
                 return 1
 
             for i in range(cor + 1, cnr + 1):
                 self.hook("commit", node=hex(self.changelog.node(i)))
 
         return
 
     def update(self, node, allow=False, force=False, choose=None,
                moddirstate=True):
         pl = self.dirstate.parents()
         if not force and pl[1] != nullid:
             self.ui.warn("aborting: outstanding uncommitted merges\n")
             return 1
 
         p1, p2 = pl[0], node
         pa = self.changelog.ancestor(p1, p2)
         m1n = self.changelog.read(p1)[0]
         m2n = self.changelog.read(p2)[0]
         man = self.manifest.ancestor(m1n, m2n)
         m1 = self.manifest.read(m1n)
         mf1 = self.manifest.readflags(m1n)
         m2 = self.manifest.read(m2n)
         mf2 = self.manifest.readflags(m2n)
         ma = self.manifest.read(man)
         mfa = self.manifest.readflags(man)
 
         (c, a, d, u) = self.changes()
 
         # is this a jump, or a merge?  i.e. is there a linear path
         # from p1 to p2?
         linear_path = (pa == p1 or pa == p2)
 
         # resolve the manifest to determine which files
         # we care about merging
         self.ui.note("resolving manifests\n")
         self.ui.debug(" force %s allow %s moddirstate %s linear %s\n" %
                       (force, allow, moddirstate, linear_path))
         self.ui.debug(" ancestor %s local %s remote %s\n" %
                       (short(man), short(m1n), short(m2n)))
 
         merge = {}
         get = {}
         remove = []
 
         # construct a working dir manifest
         mw = m1.copy()
         mfw = mf1.copy()
         umap = dict.fromkeys(u)
 
         for f in a + c + u:
             mw[f] = ""
             mfw[f] = util.is_exec(self.wjoin(f), mfw.get(f, False))
 
         for f in d:
             if f in mw: del mw[f]
 
             # If we're jumping between revisions (as opposed to merging),
             # and if neither the working directory nor the target rev has
             # the file, then we need to remove it from the dirstate, to
             # prevent the dirstate from listing the file when it is no
             # longer in the manifest.
             if moddirstate and linear_path and f not in m2:
                 self.dirstate.forget((f,))
 
         # Compare manifests
         for f, n in mw.iteritems():
             if choose and not choose(f): continue
             if f in m2:
                 s = 0
 
                 # is the wfile new since m1, and match m2?
                 if f not in m1:
                     t1 = self.wread(f)
                     t2 = self.file(f).read(m2[f])
                     if cmp(t1, t2) == 0:
                         n = m2[f]
                     del t1, t2
 
                 # are files different?
                 if n != m2[f]:
                     a = ma.get(f, nullid)
                     # are both different from the ancestor?
                     if n != a and m2[f] != a:
                         self.ui.debug(" %s versions differ, resolve\n" % f)
                         # merge executable bits
                         # "if we changed or they changed, change in merge"
                         a, b, c = mfa.get(f, 0), mfw[f], mf2[f]
                         mode = ((a^b) | (a^c)) ^ a
                         merge[f] = (m1.get(f, nullid), m2[f], mode)
                         s = 1
                     # are we clobbering?
                     # is remote's version newer?
                     # or are we going back in time?
                     elif force or m2[f] != a or (p2 == pa and mw[f] == m1[f]):
                         self.ui.debug(" remote %s is newer, get\n" % f)
                         get[f] = m2[f]
                         s = 1
                 elif f in umap:
                     # this unknown file is the same as the checkout
                     get[f] = m2[f]
 
                 if not s and mfw[f] != mf2[f]:
                     if force:
                         self.ui.debug(" updating permissions for %s\n" % f)
                         util.set_exec(self.wjoin(f), mf2[f])
                     else:
                         a, b, c = mfa.get(f, 0), mfw[f], mf2[f]
                         mode = ((a^b) | (a^c)) ^ a
                         if mode != b:
                             self.ui.debug(" updating permissions for %s\n" % f)
                             util.set_exec(self.wjoin(f), mode)
                 del m2[f]
             elif f in ma:
                 if n != ma[f]:
                     r = "d"
                     if not force and (linear_path or allow):
                         r = self.ui.prompt(
                             (" local changed %s which remote deleted\n" % f) +
                             "(k)eep or (d)elete?", "[kd]", "k")
                     if r == "d":
                         remove.append(f)
                 else:
                     self.ui.debug("other deleted %s\n" % f)
                     remove.append(f) # other deleted it
             else:
                 # file is created on branch or in working directory
                 if force and f not in umap:
                     self.ui.debug("remote deleted %s, clobbering\n" % f)
                     remove.append(f)
                 elif n == m1.get(f, nullid): # same as parent
                     if p2 == pa: # going backwards?
                         self.ui.debug("remote deleted %s\n" % f)
                         remove.append(f)
                     else:
                         self.ui.debug("local modified %s, keeping\n" % f)
                 else:
                     self.ui.debug("working dir created %s, keeping\n" % f)
 
         for f, n in m2.iteritems():
             if choose and not choose(f): continue
             if f[0] == "/": continue
             if f in ma and n != ma[f]:
                 r = "k"
                 if not force and (linear_path or allow):
                     r = self.ui.prompt(
                         ("remote changed %s which local deleted\n" % f) +
                         "(k)eep or (d)elete?", "[kd]", "k")
                 if r == "k": get[f] = n
             elif f not in ma:
                 self.ui.debug("remote created %s\n" % f)
                 get[f] = n
             else:
                 if force or p2 == pa: # going backwards?
                     self.ui.debug("local deleted %s, recreating\n" % f)
                     get[f] = n
                 else:
                     self.ui.debug("local deleted %s\n" % f)
 
         del mw, m1, m2, ma
 
         if force:
             for f in merge:
                 get[f] = merge[f][1]
             merge = {}
 
         if linear_path or force:
             # we don't need to do any magic, just jump to the new rev
             branch_merge = False
             p1, p2 = p2, nullid
         else:
             if not allow:
                 self.ui.status("this update spans a branch" +
                                " affecting the following files:\n")
                 fl = merge.keys() + get.keys()
                 fl.sort()
                 for f in fl:
                     cf = ""
                     if f in merge: cf = " (resolve)"
                     self.ui.status(" %s%s\n" % (f, cf))
                 self.ui.warn("aborting update spanning branches!\n")
                 self.ui.status("(use update -m to merge across branches" +
                                " or -C to lose changes)\n")
                 return 1
             branch_merge = True
 
         if moddirstate:
             self.dirstate.setparents(p1, p2)
 
         # get the files we don't need to change
         files = get.keys()
         files.sort()
         for f in files:
             if f[0] == "/": continue
             self.ui.note("getting %s\n" % f)
             t = self.file(f).read(get[f])
             try:
                 self.wwrite(f, t)
             except IOError, e:
                 if e.errno != errno.ENOENT:
                     raise
                 os.makedirs(os.path.dirname(self.wjoin(f)))
                 self.wwrite(f, t)
             util.set_exec(self.wjoin(f), mf2[f])
             if moddirstate:
                 if branch_merge:
                     self.dirstate.update([f], 'n', st_mtime=-1)
                 else:
                     self.dirstate.update([f], 'n')
 
         # merge the tricky bits
         files = merge.keys()
         files.sort()
         for f in files:
             self.ui.status("merging %s\n" % f)
             my, other, flag = merge[f]
             self.merge3(f, my, other)
             util.set_exec(self.wjoin(f), flag)
             if moddirstate:
                 if branch_merge:
                     # We've done a branch merge, mark this file as merged
                     # so that we properly record the merger later
                     self.dirstate.update([f], 'm')
                 else:
                     # We've update-merged a locally modified file, so
                     # we set the dirstate to emulate a normal checkout
                     # of that file some time in the past. Thus our
                     # merge will appear as a normal local file
                     # modification.
                     f_len = len(self.file(f).read(other))
                     self.dirstate.update([f], 'n', st_size=f_len, st_mtime=-1)
 
         remove.sort()
         for f in remove:
             self.ui.note("removing %s\n" % f)
             try:
                 os.unlink(self.wjoin(f))
             except OSError, inst:
                 self.ui.warn("update failed to remove %s: %s!\n" % (f, inst))
             # try removing directories that might now be empty
             try: os.removedirs(os.path.dirname(self.wjoin(f)))
             except: pass
         if moddirstate:
             if branch_merge:
                 self.dirstate.update(remove, 'r')
             else:
                 self.dirstate.forget(remove)
 
     def merge3(self, fn, my, other):
         """perform a 3-way merge in the working directory"""
 
         def temp(prefix, node):
             pre = "%s~%s." % (os.path.basename(fn), prefix)
             (fd, name) = tempfile.mkstemp("", pre)
             f = os.fdopen(fd, "wb")
             self.wwrite(fn, fl.read(node), f)
             f.close()
             return name
 
         fl = self.file(fn)
         base = fl.ancestor(my, other)
         a = self.wjoin(fn)
         b = temp("base", base)
         c = temp("other", other)
 
         self.ui.note("resolving %s\n" % fn)
         self.ui.debug("file %s: my %s other %s ancestor %s\n" %
                               (fn, short(my), short(other), short(base)))
 
         cmd = (os.environ.get("HGMERGE") or self.ui.config("ui", "merge")
                or "hgmerge")
         r = os.system("%s %s %s %s" % (cmd, a, b, c))
         if r:
             self.ui.warn("merging %s failed!\n" % fn)
 
         os.unlink(b)
         os.unlink(c)
 
     def verify(self):
         filelinkrevs = {}
         filenodes = {}
         changesets = revisions = files = 0
         errors = [0]
         neededmanifests = {}
 
         def err(msg):
             self.ui.warn(msg + "\n")
             errors[0] += 1
 
         seen = {}
         self.ui.status("checking changesets\n")
         for i in range(self.changelog.count()):
             changesets += 1
             n = self.changelog.node(i)
             l = self.changelog.linkrev(n)
             if l != i:
                 err("incorrect link (%d) for changeset revision %d" % (l, i))
             if n in seen:
                 err("duplicate changeset at revision %d" % i)
             seen[n] = 1
 
             for p in self.changelog.parents(n):
                 if p not in self.changelog.nodemap:
                     err("changeset %s has unknown parent %s" %
                                  (short(n), short(p)))
             try:
                 changes = self.changelog.read(n)
             except Exception, inst:
                 err("unpacking changeset %s: %s" % (short(n), inst))
 
             neededmanifests[changes[0]] = n
 
             for f in changes[3]:
                 filelinkrevs.setdefault(f, []).append(i)
 
         seen = {}
         self.ui.status("checking manifests\n")
         for i in range(self.manifest.count()):
             n = self.manifest.node(i)
             l = self.manifest.linkrev(n)
 
             if l < 0 or l >= self.changelog.count():
                 err("bad manifest link (%d) at revision %d" % (l, i))
 
             if n in neededmanifests:
                 del neededmanifests[n]
 
             if n in seen:
                 err("duplicate manifest at revision %d" % i)
 
             seen[n] = 1
 
             for p in self.manifest.parents(n):
                 if p not in self.manifest.nodemap:
                     err("manifest %s has unknown parent %s" %
                         (short(n), short(p)))
 
             try:
                 delta = mdiff.patchtext(self.manifest.delta(n))
             except KeyboardInterrupt:
                 self.ui.warn("interrupted")
                 raise
             except Exception, inst:
                 err("unpacking manifest %s: %s" % (short(n), inst))
 
             ff = [ l.split('\0') for l in delta.splitlines() ]
             for f, fn in ff:
                 filenodes.setdefault(f, {})[bin(fn[:40])] = 1
 
         self.ui.status("crosschecking files in changesets and manifests\n")
 
         for m,c in neededmanifests.items():
             err("Changeset %s refers to unknown manifest %s" %
                 (short(m), short(c)))
         del neededmanifests
 
         for f in filenodes:
             if f not in filelinkrevs:
                 err("file %s in manifest but not in changesets" % f)
 
         for f in filelinkrevs:
             if f not in filenodes:
                 err("file %s in changeset but not in manifest" % f)
 
         self.ui.status("checking files\n")
         ff = filenodes.keys()
         ff.sort()
         for f in ff:
             if f == "/dev/null": continue
             files += 1
             fl = self.file(f)
             nodes = { nullid: 1 }
             seen = {}
             for i in range(fl.count()):
                 revisions += 1
                 n = fl.node(i)
 
                 if n in seen:
                     err("%s: duplicate revision %d" % (f, i))
                 if n not in filenodes[f]:
                     err("%s: %d:%s not in manifests" % (f, i, short(n)))
                 else:
                     del filenodes[f][n]
 
                 flr = fl.linkrev(n)
                 if flr not in filelinkrevs[f]:
                     err("%s:%s points to unexpected changeset %d"
                             % (f, short(n), flr))
                 else:
                     filelinkrevs[f].remove(flr)
 
                 # verify contents
                 try:
                     t = fl.read(n)
                 except Exception, inst:
                     err("unpacking file %s %s: %s" % (f, short(n), inst))
 
                 # verify parents
                 (p1, p2) = fl.parents(n)
                 if p1 not in nodes:
                     err("file %s:%s unknown parent 1 %s" %
                         (f, short(n), short(p1)))
                 if p2 not in nodes:
                     err("file %s:%s unknown parent 2 %s" %
                             (f, short(n), short(p1)))
                 nodes[n] = 1
 
             # cross-check
             for node in filenodes[f]:
                 err("node %s in manifests not in %s" % (hex(node), f))
 
         self.ui.status("%d files, %d changesets, %d total revisions\n" %
                        (files, changesets, revisions))
 
         if errors[0]:
             self.ui.warn("%d integrity errors encountered!\n" % errors[0])
             return 1

mercurial/revlog.py

@@ -1,817 +1,814 b''
 """
 revlog.py - storage back-end for mercurial
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches
 
 Copyright 2005 Matt Mackall <[e-mail]>
 
 This software may be used and distributed according to the terms
 of the GNU General Public License, incorporated herein by reference.
 """
 
 from node import *
 from demandload import demandload
 demandload(globals(), "binascii errno heapq mdiff sha struct zlib")
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     l = [p1, p2]
     l.sort()
     s = sha.new(l[0])
     s.update(l[1])
     s.update(text)
     return s.digest()
 
 def compress(text):
     """ generate a possibly-compressed representation of text """
     if not text: return text
     if len(text) < 44:
         if text[0] == '\0': return text
         return 'u' + text
     bin = zlib.compress(text)
     if len(bin) > len(text):
         if text[0] == '\0': return text
         return 'u' + text
     return bin
 
 def decompress(bin):
     """ decompress the given input """
     if not bin: return bin
     t = bin[0]
     if t == '\0': return bin
     if t == 'x': return zlib.decompress(bin)
     if t == 'u': return bin[1:]
     raise RevlogError("unknown compression type %s" % t)
 
 indexformat = ">4l20s20s20s"
 
 class lazyparser:
     """
     this class avoids the need to parse the entirety of large indices
 
     By default we parse and load 1000 entries at a time.
 
     If no position is specified, we load the whole index, and replace
     the lazy objects in revlog with the underlying objects for
     efficiency in cases where we look at most of the nodes.
     """
     def __init__(self, data, revlog):
         self.data = data
         self.s = struct.calcsize(indexformat)
         self.l = len(data)/self.s
         self.index = [None] * self.l
         self.map = {nullid: -1}
         self.all = 0
         self.revlog = revlog
 
     def load(self, pos=None):
         if self.all: return
         if pos is not None:
             block = pos / 1000
             i = block * 1000
             end = min(self.l, i + 1000)
         else:
             self.all = 1
             i = 0
             end = self.l
             self.revlog.index = self.index
             self.revlog.nodemap = self.map
 
         while i < end:
             d = self.data[i * self.s: (i + 1) * self.s]
             e = struct.unpack(indexformat, d)
             self.index[i] = e
             self.map[e[6]] = i
             i += 1
 
 class lazyindex:
     """a lazy version of the index array"""
     def __init__(self, parser):
         self.p = parser
     def __len__(self):
         return len(self.p.index)
     def load(self, pos):
         self.p.load(pos)
         return self.p.index[pos]
     def __getitem__(self, pos):
         return self.p.index[pos] or self.load(pos)
     def append(self, e):
         self.p.index.append(e)
 
 class lazymap:
     """a lazy version of the node map"""
     def __init__(self, parser):
         self.p = parser
     def load(self, key):
         if self.p.all: return
         n = self.p.data.find(key)
         if n < 0:
             raise KeyError(key)
         pos = n / self.p.s
         self.p.load(pos)
     def __contains__(self, key):
         self.p.load()
         return key in self.p.map
     def __iter__(self):
         yield nullid
         for i in xrange(self.p.l):
             try:
                 yield self.p.index[i][6]
             except:
                 self.p.load(i)
                 yield self.p.index[i][6]
     def __getitem__(self, key):
         try:
             return self.p.map[key]
         except KeyError:
             try:
                 self.load(key)
                 return self.p.map[key]
             except KeyError:
                 raise KeyError("node " + hex(key))
     def __setitem__(self, key, val):
         self.p.map[key] = val
 
 class RevlogError(Exception): pass
 
 class revlog:
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, includings its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
     """
     def __init__(self, opener, indexfile, datafile):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = datafile
         self.opener = opener
         self.cache = None
 
         try:
             i = self.opener(self.indexfile).read()
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             i = ""
 
         if len(i) > 10000:
             # big index, let's parse it on demand
             parser = lazyparser(i, self)
             self.index = lazyindex(parser)
             self.nodemap = lazymap(parser)
         else:
             s = struct.calcsize(indexformat)
             l = len(i) / s
             self.index = [None] * l
             m = [None] * l
 
             n = 0
             for f in xrange(0, len(i), s):
                 # offset, size, base, linkrev, p1, p2, nodeid
                 e = struct.unpack(indexformat, i[f:f + s])
                 m[n] = (e[6], n)
                 self.index[n] = e
                 n += 1
 
             self.nodemap = dict(m)
             self.nodemap[nullid] = -1
 
     def tip(self): return self.node(len(self.index) - 1)
     def count(self): return len(self.index)
     def node(self, rev): return (rev < 0) and nullid or self.index[rev][6]
     def rev(self, node):
         try:
             return self.nodemap[node]
         except KeyError:
             raise RevlogError('%s: no node %s' % (self.indexfile, hex(node)))
     def linkrev(self, node): return self.index[self.rev(node)][3]
     def parents(self, node):
         if node == nullid: return (nullid, nullid)
         return self.index[self.rev(node)][4:6]
 
     def start(self, rev): return self.index[rev][0]
     def length(self, rev): return self.index[rev][1]
     def end(self, rev): return self.start(rev) + self.length(rev)
     def base(self, rev): return self.index[rev][2]
 
     def reachable(self, rev, stop=None):
         reachable = {}
         visit = [rev]
         reachable[rev] = 1
         if stop:
             stopn = self.rev(stop)
         else:
             stopn = 0
         while visit:
             n = visit.pop(0)
             if n == stop:
                 continue
             if n == nullid:
                 continue
             for p in self.parents(n):
                 if self.rev(p) < stopn:
                     continue
                 if p not in reachable:
                     reachable[p] = 1
                     visit.append(p)
         return reachable
 
     def nodesbetween(self, roots=None, heads=None):
         """Return a tuple containing three elements. Elements 1 and 2 contain
         a final list bases and heads after all the unreachable ones have been
         pruned.  Element 0 contains a topologically sorted list of all
 
         nodes that satisfy these constraints:
         1. All nodes must be descended from a node in roots (the nodes on
            roots are considered descended from themselves).
         2. All nodes must also be ancestors of a node in heads (the nodes in
            heads are considered to be their own ancestors).
 
         If roots is unspecified, nullid is assumed as the only root.
         If heads is unspecified, it is taken to be the output of the
         heads method (i.e. a list of all nodes in the repository that
         have no children)."""
         if roots is not None:
             roots = list(roots)
             lowestrev = min([self.rev(n) for n in roots])
         else:
             roots = [nullid] # Everybody's a descendent of nullid
             lowestrev = -1
         if (lowestrev == -1) and (heads is None):
             # We want _all_ the nodes!
             return ([self.node(r) for r in xrange(0, self.count())],
                     [nullid], list(self.heads()))
         if heads is None:
             # All nodes are ancestors, so the latest ancestor is the last
             # node.
             highestrev = self.count() - 1
             # Set ancestors to None to signal that every node is an ancestor.
             ancestors = None
             # Set heads to an empty dictionary for later discovery of heads
             heads = {}
         else:
             ancestors = {}
             # Start at the top and keep marking parents until we're done.
             nodestotag = list(heads)
             # Turn heads into a dictionary so we can remove 'fake' heads.
             # Also, later we will be using it to filter out the heads we can't
             # find from roots.
             heads = dict.fromkeys(heads, 0)
             # Remember where the top was so we can use it as a limit later.
             highestrev = max([self.rev(n) for n in nodestotag])
             while nodestotag:
                 # grab a node to tag
                 n = nodestotag.pop()
                 # Never tag nullid
                 if n == nullid:
                     continue
                 # A node's revision number represents its place in a
                 # topologically sorted list of nodes.
                 r = self.rev(n)
                 if r >= lowestrev:
                     if n not in ancestors:
                         # If we are possibly a descendent of one of the roots
                         # and we haven't already been marked as an ancestor
                         ancestors[n] = 1 # Mark as ancestor
                         # Add non-nullid parents to list of nodes to tag.
                         nodestotag.extend([p for p in self.parents(n) if
                                            p != nullid])
                     elif n in heads: # We've seen it before, is it a fake head?
                         # So it is, real heads should not be the ancestors of
                         # any other heads.
                         heads.pop(n)
             # Now that we have our set of ancestors, we want to remove any
             # roots that are not ancestors.
 
             # If one of the roots was nullid, everything is included anyway.
             if lowestrev > -1:
                 # But, since we weren't, let's recompute the lowest rev to not
                 # include roots that aren't ancestors.
 
                 # Filter out roots that aren't ancestors of heads
                 roots = [n for n in roots if n in ancestors]
                 # Recompute the lowest revision
                 if roots:
                     lowestrev = min([self.rev(n) for n in roots])
                 else:
                     # No more roots?  Return empty list
                     return ([], [], [])
             else:
                 # We are descending from nullid, and don't need to care about
                 # any other roots.
                 lowestrev = -1
                 roots = [nullid]
         # Transform our roots list into a 'set' (i.e. a dictionary where the
         # values don't matter.
         descendents = dict.fromkeys(roots, 1)
         # Also, keep the original roots so we can filter out roots that aren't
         # 'real' roots (i.e. are descended from other roots).
         roots = descendents.copy()
         # Our topologically sorted list of output nodes.
         orderedout = []
         # Don't start at nullid since we don't want nullid in our output list,
         # and if nullid shows up in descedents, empty parents will look like
         # they're descendents.
         for r in xrange(max(lowestrev, 0), highestrev + 1):
             n = self.node(r)
             isdescendent = False
             if lowestrev == -1:  # Everybody is a descendent of nullid
                 isdescendent = True
             elif n in descendents:
                 # n is already a descendent
                 isdescendent = True
                 # This check only needs to be done here because all the roots
                 # will start being marked is descendents before the loop.
                 if n in roots:
                     # If n was a root, check if it's a 'real' root.
                     p = tuple(self.parents(n))
                     # If any of its parents are descendents, it's not a root.
                     if (p[0] in descendents) or (p[1] in descendents):
                         roots.pop(n)
             else:
                 p = tuple(self.parents(n))
                 # A node is a descendent if either of its parents are
                 # descendents.  (We seeded the dependents list with the roots
                 # up there, remember?)
                 if (p[0] in descendents) or (p[1] in descendents):
                     descendents[n] = 1
                     isdescendent = True
             if isdescendent and ((ancestors is None) or (n in ancestors)):
                 # Only include nodes that are both descendents and ancestors.
                 orderedout.append(n)
                 if (ancestors is not None) and (n in heads):
                     # We're trying to figure out which heads are reachable
                     # from roots.
                     # Mark this head as having been reached
                     heads[n] = 1
                 elif ancestors is None:
                     # Otherwise, we're trying to discover the heads.
                     # Assume this is a head because if it isn't, the next step
                     # will eventually remove it.
                     heads[n] = 1
                     # But, obviously its parents aren't.
                     for p in self.parents(n):
                         heads.pop(p, None)
         heads = [n for n in heads.iterkeys() if heads[n] != 0]
         roots = roots.keys()
         assert orderedout
         assert roots
         assert heads
         return (orderedout, roots, heads)
 
     def heads(self, stop=None):
         """return the list of all nodes that have no children"""
         p = {}
         h = []
         stoprev = 0
         if stop and stop in self.nodemap:
             stoprev = self.rev(stop)
 
         for r in range(self.count() - 1, -1, -1):
             n = self.node(r)
             if n not in p:
                 h.append(n)
             if n == stop:
                 break
             if r < stoprev:
                 break
             for pn in self.parents(n):
                 p[pn] = 1
         return h
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in range(p + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn == node:
                     c.append(n)
                     continue
                 elif pn == nullid:
                     continue
         return c
 
     def lookup(self, id):
         """locate a node based on revision number or subset of hex nodeid"""
         try:
             rev = int(id)
             if str(rev) != id: raise ValueError
             if rev < 0: rev = self.count() + rev
             if rev < 0 or rev >= self.count(): raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             c = []
             for n in self.nodemap:
                 if hex(n).startswith(id):
                     c.append(n)
             if len(c) > 1: raise KeyError("Ambiguous identifier")
             if len(c) < 1: raise KeyError("No match found")
             return c[0]
 
         return None
 
     def diff(self, a, b):
         """return a delta between two revisions"""
         return mdiff.textdiff(a, b)
 
     def patches(self, t, pl):
         """apply a list of patches to a string"""
         return mdiff.patches(t, pl)
 
     def delta(self, node):
         """return or calculate a delta between a node and its predecessor"""
         r = self.rev(node)
         b = self.base(r)
         if r == b:
             return self.diff(self.revision(self.node(r - 1)),
                              self.revision(node))
         else:
             f = self.opener(self.datafile)
             f.seek(self.start(r))
             data = f.read(self.length(r))
         return decompress(data)
 
     def revision(self, node):
         """return an uncompressed revision of a given"""
         if node == nullid: return ""
         if self.cache and self.cache[0] == node: return self.cache[2]
 
         # look up what we need to read
         text = None
         rev = self.rev(node)
         start, length, base, link, p1, p2, node = self.index[rev]
         end = start + length
         if base != rev: start = self.start(base)
 
         # do we have useful data cached?
         if self.cache and self.cache[1] >= base and self.cache[1] < rev:
             base = self.cache[1]
             start = self.start(base + 1)
             text = self.cache[2]
             last = 0
 
         f = self.opener(self.datafile)
         f.seek(start)
         data = f.read(end - start)
 
         if text is None:
             last = self.length(base)
             text = decompress(data[:last])
 
         bins = []
         for r in xrange(base + 1, rev + 1):
             s = self.length(r)
             bins.append(decompress(data[last:last + s]))
             last = last + s
 
         text = mdiff.patches(text, bins)
 
         if node != hash(text, p1, p2):
             raise RevlogError("integrity check failed on %s:%d"
                           % (self.datafile, rev))
 
         self.cache = (node, rev, text)
         return text
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         d - an optional precomputed delta
         """
         if text is None: text = ""
         if p1 is None: p1 = self.tip()
         if p2 is None: p2 = nullid
 
         node = hash(text, p1, p2)
 
         if node in self.nodemap:
             return node
 
         n = self.count()
         t = n - 1
 
         if n:
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
             if not d:
                 prev = self.revision(self.tip())
                 d = self.diff(prev, text)
             data = compress(d)
             dist = end - start + len(data)
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if not n or dist > len(text) * 2:
             data = compress(text)
             base = n
         else:
             base = self.base(t)
 
         offset = 0
         if t >= 0:
             offset = self.end(t)
 
         e = (offset, len(data), base, link, p1, p2, node)
 
         self.index.append(e)
         self.nodemap[node] = n
         entry = struct.pack(indexformat, *e)
 
         transaction.add(self.datafile, e[0])
         self.opener(self.datafile, "a").write(data)
         transaction.add(self.indexfile, n * len(entry))
         self.opener(self.indexfile, "a").write(entry)
 
         self.cache = (node, n, text)
         return node
 
     def ancestor(self, a, b):
         """calculate the least common ancestor of nodes a and b"""
         # calculate the distance of every node from root
         dist = {nullid: 0}
         for i in xrange(self.count()):
             n = self.node(i)
             p1, p2 = self.parents(n)
             dist[n] = max(dist[p1], dist[p2]) + 1
 
         # traverse ancestors in order of decreasing distance from root
         def ancestors(node):
             # we store negative distances because heap returns smallest member
             h = [(-dist[node], node)]
             seen = {}
             earliest = self.count()
             while h:
                 d, n = heapq.heappop(h)
                 if n not in seen:
                     seen[n] = 1
                     r = self.rev(n)
                     yield (-d, n)
                     for p in self.parents(n):
                         heapq.heappush(h, (-dist[p], p))
 
         def generations(node):
             sg, s = None, {}
             for g,n in ancestors(node):
                 if g != sg:
                     if sg:
                         yield sg, s
                     sg, s = g, {n:1}
                 else:
                     s[n] = 1
             yield sg, s
 
         x = generations(a)
         y = generations(b)
         gx = x.next()
         gy = y.next()
 
         # increment each ancestor list until it is closer to root than
         # the other, or they match
         while 1:
             #print "ancestor gen %s %s" % (gx[0], gy[0])
             if gx[0] == gy[0]:
                 # find the intersection
                 i = [ n for n in gx[1] if n in gy[1] ]
                 if i:
                     return i[0]
                 else:
                     #print "next"
                     gy = y.next()
                     gx = x.next()
             elif gx[0] < gy[0]:
                 #print "next y"
                 gy = y.next()
             else:
                 #print "next x"
                 gx = x.next()
 
-    def group(self, linkmap):
+    def group(self, nodelist, lookup, infocollect = None):
         """calculate a delta group
 
         Given a list of changeset revs, return a set of deltas and
         metadata corresponding to nodes. the first delta is
         parent(nodes[0]) -> nodes[0] the receiver is guaranteed to
         have this parent as it has all history before these
         changesets. parent is parent[0]
         """
-        revs = []
-        needed = {}
-
-        # find file nodes/revs that match changeset revs
-        for i in xrange(0, self.count()):
-            if self.index[i][3] in linkmap:
-                revs.append(i)
-                needed[i] = 1
+        revs = [self.rev(n) for n in nodelist]
+        needed = dict.fromkeys(revs, 1)
 
         # if we don't have any revisions touched by these changesets, bail
         if not revs:
             yield struct.pack(">l", 0)
             return
 
         # add the parent of the first rev
         p = self.parents(self.node(revs[0]))[0]
         revs.insert(0, self.rev(p))
 
         # for each delta that isn't contiguous in the log, we need to
         # reconstruct the base, reconstruct the result, and then
         # calculate the delta. We also need to do this where we've
         # stored a full version and not a delta
         for i in xrange(0, len(revs) - 1):
             a, b = revs[i], revs[i + 1]
             if a + 1 != b or self.base(b) == b:
                 for j in xrange(self.base(a), a + 1):
                     needed[j] = 1
                 for j in xrange(self.base(b), b + 1):
                     needed[j] = 1
 
         # calculate spans to retrieve from datafile
         needed = needed.keys()
         needed.sort()
         spans = []
         oo = -1
         ol = 0
         for n in needed:
             if n < 0: continue
             o = self.start(n)
             l = self.length(n)
             if oo + ol == o: # can we merge with the previous?
                 nl = spans[-1][2]
                 nl.append((n, l))
                 ol += l
                 spans[-1] = (oo, ol, nl)
             else:
                 oo = o
                 ol = l
                 spans.append((oo, ol, [(n, l)]))
 
         # read spans in, divide up chunks
         chunks = {}
         for span in spans:
             # we reopen the file for each span to make http happy for now
             f = self.opener(self.datafile)
             f.seek(span[0])
             data = f.read(span[1])
 
             # divide up the span
             pos = 0
             for r, l in span[2]:
                 chunks[r] = decompress(data[pos: pos + l])
                 pos += l
 
         # helper to reconstruct intermediate versions
         def construct(text, base, rev):
             bins = [chunks[r] for r in xrange(base + 1, rev + 1)]
             return mdiff.patches(text, bins)
 
         # build deltas
         deltas = []
         for d in xrange(0, len(revs) - 1):
             a, b = revs[d], revs[d + 1]
             n = self.node(b)
 
+            if infocollect is not None:
+                infocollect(n)
+
             # do we need to construct a new delta?
             if a + 1 != b or self.base(b) == b:
                 if a >= 0:
                     base = self.base(a)
                     ta = chunks[self.base(a)]
                     ta = construct(ta, base, a)
                 else:
                     ta = ""
 
                 base = self.base(b)
                 if a > base:
                     base = a
                     tb = ta
                 else:
                     tb = chunks[self.base(b)]
                 tb = construct(tb, base, b)
                 d = self.diff(ta, tb)
             else:
                 d = chunks[b]
 
             p = self.parents(n)
-            meta = n + p[0] + p[1] + linkmap[self.linkrev(n)]
+            meta = n + p[0] + p[1] + lookup(n)
             l = struct.pack(">l", len(meta) + len(d) + 4)
             yield l
             yield meta
             yield d
 
         yield struct.pack(">l", 0)
 
     def addgroup(self, revs, linkmapper, transaction, unique=0):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
         """
 
         #track the base of the current delta log
         r = self.count()
         t = r - 1
         node = nullid
 
         base = prev = -1
         start = end = measure = 0
         if r:
             start = self.start(self.base(t))
             end = self.end(t)
             measure = self.length(self.base(t))
             base = self.base(t)
             prev = self.tip()
 
         transaction.add(self.datafile, end)
         transaction.add(self.indexfile, r * struct.calcsize(indexformat))
         dfh = self.opener(self.datafile, "a")
         ifh = self.opener(self.indexfile, "a")
 
         # loop through our set of deltas
         chain = None
         for chunk in revs:
             node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
             link = linkmapper(cs)
             if node in self.nodemap:
                 # this can happen if two branches make the same change
                 # if unique:
                 #    raise RevlogError("already have %s" % hex(node[:4]))
                 chain = node
                 continue
             delta = chunk[80:]
 
             if not chain:
                 # retrieve the parent revision of the delta chain
                 chain = p1
                 if not chain in self.nodemap:
                     raise RevlogError("unknown base %s" % short(chain[:4]))
 
             # full versions are inserted when the needed deltas become
             # comparable to the uncompressed text or when the previous
             # version is not the one we have a delta against. We use
             # the size of the previous full rev as a proxy for the
             # current size.
 
             if chain == prev:
                 cdelta = compress(delta)
 
             if chain != prev or (end - start + len(cdelta)) > measure * 2:
                 # flush our writes here so we can read it in revision
                 dfh.flush()
                 ifh.flush()
                 text = self.revision(chain)
                 text = self.patches(text, [delta])
                 chk = self.addrevision(text, transaction, link, p1, p2)
                 if chk != node:
                     raise RevlogError("consistency error adding group")
                 measure = len(text)
             else:
                 e = (end, len(cdelta), self.base(t), link, p1, p2, node)
                 self.index.append(e)
                 self.nodemap[node] = r
                 dfh.write(cdelta)
                 ifh.write(struct.pack(indexformat, *e))
 
             t, r, chain, prev = r, r + 1, node, node
             start = self.start(self.base(t))
             end = self.end(t)
 
         dfh.close()
         ifh.close()
         return node