match: have visitdir() consider includes and excludes

match.visitdir() used to only look at the match's primary pattern roots to
decide if a treemanifest traverser should descend into a particular directory.
This change logically makes visitdir also consider the match's include and
exclude pattern roots (if applicable) to make this decision.

This is especially important for situations like using narrowhg with multiple
treemanifest revlogs.

mercurial/match.py

@@ -1,587 +1,610
 # match.py - filename matching
 #
 #  Copyright 2008, 2009 Matt Mackall <[e-mail]> and others
 #
 # 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 util, pathutil
 from i18n import _
 
 propertycache = util.propertycache
 
 def _rematcher(regex):
     '''compile the regexp with the best available regexp engine and return a
     matcher function'''
     m = util.re.compile(regex)
     try:
         # slightly faster, provided by facebook's re2 bindings
         return m.test_match
     except AttributeError:
         return m.match
 
 def _expandsets(kindpats, ctx, listsubrepos):
     '''Returns the kindpats list with the 'set' patterns expanded.'''
     fset = set()
     other = []
 
     for kind, pat, source in kindpats:
         if kind == 'set':
             if not ctx:
                 raise util.Abort("fileset expression with no context")
             s = ctx.getfileset(pat)
             fset.update(s)
 
             if listsubrepos:
                 for subpath in ctx.substate:
                     s = ctx.sub(subpath).getfileset(pat)
                     fset.update(subpath + '/' + f for f in s)
 
             continue
         other.append((kind, pat, source))
     return fset, other
 
 def _kindpatsalwaysmatch(kindpats):
     """"Checks whether the kindspats match everything, as e.g.
     'relpath:.' does.
     """
     for kind, pat, source in kindpats:
         if pat != '' or kind not in ['relpath', 'glob']:
             return False
     return True
 
 class match(object):
     def __init__(self, root, cwd, patterns, include=[], exclude=[],
                  default='glob', exact=False, auditor=None, ctx=None,
                  listsubrepos=False, warn=None):
         """build an object to match a set of file patterns
 
         arguments:
         root - the canonical root of the tree you're matching against
         cwd - the current working directory, if relevant
         patterns - patterns to find
         include - patterns to include (unless they are excluded)
         exclude - patterns to exclude (even if they are included)
         default - if a pattern in patterns has no explicit type, assume this one
         exact - patterns are actually filenames (include/exclude still apply)
         warn - optional function used for printing warnings
 
         a pattern is one of:
         'glob:<glob>' - a glob relative to cwd
         're:<regexp>' - a regular expression
         'path:<path>' - a path relative to repository root
         'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs)
         'relpath:<path>' - a path relative to cwd
         'relre:<regexp>' - a regexp that needn't match the start of a name
         'set:<fileset>' - a fileset expression
         'include:<path>' - a file of patterns to read and include
         '<something>' - a pattern of the specified default type
         """
 
         self._root = root
         self._cwd = cwd
         self._files = [] # exact files and roots of patterns
         self._anypats = bool(include or exclude)
         self._always = False
         self._pathrestricted = bool(include or exclude or patterns)
         self._warn = warn
+        self._includeroots = set()
+        self._includedirs = set(['.'])
+        self._excluderoots = set()
 
         matchfns = []
         if include:
             kindpats = self._normalize(include, 'glob', root, cwd, auditor)
             self.includepat, im = _buildmatch(ctx, kindpats, '(?:/|$)',
                                               listsubrepos)
+            self._includeroots.update(_roots(kindpats))
+            self._includeroots.discard('.')
+            self._includedirs.update(util.dirs(self._includeroots))
             matchfns.append(im)
         if exclude:
             kindpats = self._normalize(exclude, 'glob', root, cwd, auditor)
             self.excludepat, em = _buildmatch(ctx, kindpats, '(?:/|$)',
                                               listsubrepos)
+            self._excluderoots.update(_roots(kindpats))
+            self._excluderoots.discard('.')
             matchfns.append(lambda f: not em(f))
         if exact:
             if isinstance(patterns, list):
                 self._files = patterns
             else:
                 self._files = list(patterns)
             matchfns.append(self.exact)
         elif patterns:
             kindpats = self._normalize(patterns, default, root, cwd, auditor)
             if not _kindpatsalwaysmatch(kindpats):
                 self._files = _roots(kindpats)
                 self._anypats = self._anypats or _anypats(kindpats)
                 self.patternspat, pm = _buildmatch(ctx, kindpats, '$',
                                                    listsubrepos)
                 matchfns.append(pm)
 
         if not matchfns:
             m = util.always
             self._always = True
         elif len(matchfns) == 1:
             m = matchfns[0]
         else:
             def m(f):
                 for matchfn in matchfns:
                     if not matchfn(f):
                         return False
                 return True
 
         self.matchfn = m
         self._fileroots = set(self._files)
 
     def __call__(self, fn):
         return self.matchfn(fn)
     def __iter__(self):
         for f in self._files:
             yield f
 
     # Callbacks related to how the matcher is used by dirstate.walk.
     # Subscribers to these events must monkeypatch the matcher object.
     def bad(self, f, msg):
         '''Callback from dirstate.walk for each explicit file that can't be
         found/accessed, with an error message.'''
         pass
 
     # If an explicitdir is set, it will be called when an explicitly listed
     # directory is visited.
     explicitdir = None
 
     # If an traversedir is set, it will be called when a directory discovered
     # by recursive traversal is visited.
     traversedir = None
 
     def abs(self, f):
         '''Convert a repo path back to path that is relative to the root of the
         matcher.'''
         return f
 
     def rel(self, f):
         '''Convert repo path back to path that is relative to cwd of matcher.'''
         return util.pathto(self._root, self._cwd, f)
 
     def uipath(self, f):
         '''Convert repo path to a display path.  If patterns or -I/-X were used
         to create this matcher, the display path will be relative to cwd.
         Otherwise it is relative to the root of the repo.'''
         return (self._pathrestricted and self.rel(f)) or self.abs(f)
 
     def files(self):
         '''Explicitly listed files or patterns or roots:
         if no patterns or .always(): empty list,
         if exact: list exact files,
         if not .anypats(): list all files and dirs,
         else: optimal roots'''
         return self._files
 
     @propertycache
     def _dirs(self):
         return set(util.dirs(self._fileroots)) | set(['.'])
 
     def visitdir(self, dir):
+        '''Decides whether a directory should be visited based on whether it
+        has potential matches in it or one of its subdirectories. This is
+        based on the match's primary, included, and excluded patterns.
+
+        This function's behavior is undefined if it has returned False for
+        one of the dir's parent directories.
+        '''
+        if dir in self._excluderoots:
+            return False
+        parentdirs = None
+        if (self._includeroots and dir not in self._includeroots and
+                dir not in self._includedirs):
+            parentdirs = util.finddirs(dir)
+            if not any(parent in self._includeroots for parent in parentdirs):
+                return False
         return (not self._fileroots or '.' in self._fileroots or
                 dir in self._fileroots or dir in self._dirs or
                 any(parentdir in self._fileroots
-                    for parentdir in util.finddirs(dir)))
+                    for parentdir in parentdirs or util.finddirs(dir)))
 
     def exact(self, f):
         '''Returns True if f is in .files().'''
         return f in self._fileroots
 
     def anypats(self):
         '''Matcher uses patterns or include/exclude.'''
         return self._anypats
 
     def always(self):
         '''Matcher will match everything and .files() will be empty
         - optimization might be possible and necessary.'''
         return self._always
 
     def ispartial(self):
         '''True if the matcher won't always match.
 
         Although it's just the inverse of _always in this implementation,
         an extenion such as narrowhg might make it return something
         slightly different.'''
         return not self._always
 
     def isexact(self):
         return self.matchfn == self.exact
 
     def _normalize(self, patterns, default, root, cwd, auditor):
         '''Convert 'kind:pat' from the patterns list to tuples with kind and
         normalized and rooted patterns and with listfiles expanded.'''
         kindpats = []
         for kind, pat in [_patsplit(p, default) for p in patterns]:
             if kind in ('glob', 'relpath'):
                 pat = pathutil.canonpath(root, cwd, pat, auditor)
             elif kind in ('relglob', 'path'):
                 pat = util.normpath(pat)
             elif kind in ('listfile', 'listfile0'):
                 try:
                     files = util.readfile(pat)
                     if kind == 'listfile0':
                         files = files.split('\0')
                     else:
                         files = files.splitlines()
                     files = [f for f in files if f]
                 except EnvironmentError:
                     raise util.Abort(_("unable to read file list (%s)") % pat)
                 for k, p, source in self._normalize(files, default, root, cwd,
                                                     auditor):
                     kindpats.append((k, p, pat))
                 continue
             elif kind == 'include':
                 try:
                     includepats = readpatternfile(pat, self._warn)
                     for k, p, source in self._normalize(includepats, default,
                                                         root, cwd, auditor):
                         kindpats.append((k, p, source or pat))
                 except util.Abort, inst:
                     raise util.Abort('%s: %s' % (pat, inst[0]))
                 except IOError, inst:
                     if self._warn:
                         self._warn(_("skipping unreadable pattern file "
                                      "'%s': %s\n") % (pat, inst.strerror))
                 continue
             # else: re or relre - which cannot be normalized
             kindpats.append((kind, pat, ''))
         return kindpats
 
 def exact(root, cwd, files):
     return match(root, cwd, files, exact=True)
 
 def always(root, cwd):
     return match(root, cwd, [])
 
 class narrowmatcher(match):
     """Adapt a matcher to work on a subdirectory only.
 
     The paths are remapped to remove/insert the path as needed:
 
     >>> m1 = match('root', '', ['a.txt', 'sub/b.txt'])
     >>> m2 = narrowmatcher('sub', m1)
     >>> bool(m2('a.txt'))
     False
     >>> bool(m2('b.txt'))
     True
     >>> bool(m2.matchfn('a.txt'))
     False
     >>> bool(m2.matchfn('b.txt'))
     True
     >>> m2.files()
     ['b.txt']
     >>> m2.exact('b.txt')
     True
     >>> util.pconvert(m2.rel('b.txt'))
     'sub/b.txt'
     >>> def bad(f, msg):
     ...     print "%s: %s" % (f, msg)
     >>> m1.bad = bad
     >>> m2.bad('x.txt', 'No such file')
     sub/x.txt: No such file
     >>> m2.abs('c.txt')
     'sub/c.txt'
     """
 
     def __init__(self, path, matcher):
         self._root = matcher._root
         self._cwd = matcher._cwd
         self._path = path
         self._matcher = matcher
         self._always = matcher._always
         self._pathrestricted = matcher._pathrestricted
 
         self._files = [f[len(path) + 1:] for f in matcher._files
                        if f.startswith(path + "/")]
 
         # If the parent repo had a path to this subrepo and no patterns are
         # specified, this submatcher always matches.
         if not self._always and not matcher._anypats:
             self._always = any(f == path for f in matcher._files)
 
         self._anypats = matcher._anypats
         self.matchfn = lambda fn: matcher.matchfn(self._path + "/" + fn)
         self._fileroots = set(self._files)
 
     def abs(self, f):
         return self._matcher.abs(self._path + "/" + f)
 
     def bad(self, f, msg):
         self._matcher.bad(self._path + "/" + f, msg)
 
     def rel(self, f):
         return self._matcher.rel(self._path + "/" + f)
 
 class icasefsmatcher(match):
     """A matcher for wdir on case insensitive filesystems, which normalizes the
     given patterns to the case in the filesystem.
     """
 
     def __init__(self, root, cwd, patterns, include, exclude, default, auditor,
                  ctx, listsubrepos=False):
         init = super(icasefsmatcher, self).__init__
         self._dsnormalize = ctx.repo().dirstate.normalize
 
         init(root, cwd, patterns, include, exclude, default, auditor=auditor,
              ctx=ctx, listsubrepos=listsubrepos)
 
         # m.exact(file) must be based off of the actual user input, otherwise
         # inexact case matches are treated as exact, and not noted without -v.
         if self._files:
             self._fileroots = set(_roots(self._kp))
 
     def _normalize(self, patterns, default, root, cwd, auditor):
         self._kp = super(icasefsmatcher, self)._normalize(patterns, default,
                                                           root, cwd, auditor)
         kindpats = []
         for kind, pats, source in self._kp:
             if kind not in ('re', 'relre'):  # regex can't be normalized
                 pats = self._dsnormalize(pats)
             kindpats.append((kind, pats, source))
         return kindpats
 
 def patkind(pattern, default=None):
     '''If pattern is 'kind:pat' with a known kind, return kind.'''
     return _patsplit(pattern, default)[0]
 
 def _patsplit(pattern, default):
     """Split a string into the optional pattern kind prefix and the actual
     pattern."""
     if ':' in pattern:
         kind, pat = pattern.split(':', 1)
         if kind in ('re', 'glob', 'path', 'relglob', 'relpath', 'relre',
                     'listfile', 'listfile0', 'set', 'include'):
             return kind, pat
     return default, pattern
 
 def _globre(pat):
     r'''Convert an extended glob string to a regexp string.
 
     >>> print _globre(r'?')
     .
     >>> print _globre(r'*')
     [^/]*
     >>> print _globre(r'**')
     .*
     >>> print _globre(r'**/a')
     (?:.*/)?a
     >>> print _globre(r'a/**/b')
     a\/(?:.*/)?b
     >>> print _globre(r'[a*?!^][^b][!c]')
     [a*?!^][\^b][^c]
     >>> print _globre(r'{a,b}')
     (?:a|b)
     >>> print _globre(r'.\*\?')
     \.\*\?
     '''
     i, n = 0, len(pat)
     res = ''
     group = 0
     escape = util.re.escape
     def peek():
         return i < n and pat[i]
     while i < n:
         c = pat[i]
         i += 1
         if c not in '*?[{},\\':
             res += escape(c)
         elif c == '*':
             if peek() == '*':
                 i += 1
                 if peek() == '/':
                     i += 1
                     res += '(?:.*/)?'
                 else:
                     res += '.*'
             else:
                 res += '[^/]*'
         elif c == '?':
             res += '.'
         elif c == '[':
             j = i
             if j < n and pat[j] in '!]':
                 j += 1
             while j < n and pat[j] != ']':
                 j += 1
             if j >= n:
                 res += '\\['
             else:
                 stuff = pat[i:j].replace('\\','\\\\')
                 i = j + 1
                 if stuff[0] == '!':
                     stuff = '^' + stuff[1:]
                 elif stuff[0] == '^':
                     stuff = '\\' + stuff
                 res = '%s[%s]' % (res, stuff)
         elif c == '{':
             group += 1
             res += '(?:'
         elif c == '}' and group:
             res += ')'
             group -= 1
         elif c == ',' and group:
             res += '|'
         elif c == '\\':
             p = peek()
             if p:
                 i += 1
                 res += escape(p)
             else:
                 res += escape(c)
         else:
             res += escape(c)
     return res
 
 def _regex(kind, pat, globsuffix):
     '''Convert a (normalized) pattern of any kind into a regular expression.
     globsuffix is appended to the regexp of globs.'''
     if not pat:
         return ''
     if kind == 're':
         return pat
     if kind == 'path':
         return '^' + util.re.escape(pat) + '(?:/|$)'
     if kind == 'relglob':
         return '(?:|.*/)' + _globre(pat) + globsuffix
     if kind == 'relpath':
         return util.re.escape(pat) + '(?:/|$)'
     if kind == 'relre':
         if pat.startswith('^'):
             return pat
         return '.*' + pat
     return _globre(pat) + globsuffix
 
 def _buildmatch(ctx, kindpats, globsuffix, listsubrepos):
     '''Return regexp string and a matcher function for kindpats.
     globsuffix is appended to the regexp of globs.'''
     fset, kindpats = _expandsets(kindpats, ctx, listsubrepos)
     if not kindpats:
         return "", fset.__contains__
 
     regex, mf = _buildregexmatch(kindpats, globsuffix)
     if fset:
         return regex, lambda f: f in fset or mf(f)
     return regex, mf
 
 def _buildregexmatch(kindpats, globsuffix):
     """Build a match function from a list of kinds and kindpats,
     return regexp string and a matcher function."""
     try:
         regex = '(?:%s)' % '|'.join([_regex(k, p, globsuffix)
                                      for (k, p, s) in kindpats])
         if len(regex) > 20000:
             raise OverflowError
         return regex, _rematcher(regex)
     except OverflowError:
         # We're using a Python with a tiny regex engine and we
         # made it explode, so we'll divide the pattern list in two
         # until it works
         l = len(kindpats)
         if l < 2:
             raise
         regexa, a = _buildregexmatch(kindpats[:l//2], globsuffix)
         regexb, b = _buildregexmatch(kindpats[l//2:], globsuffix)
         return regex, lambda s: a(s) or b(s)
     except re.error:
         for k, p, s in kindpats:
             try:
                 _rematcher('(?:%s)' % _regex(k, p, globsuffix))
             except re.error:
                 if s:
                     raise util.Abort(_("%s: invalid pattern (%s): %s") %
                                      (s, k, p))
                 else:
                     raise util.Abort(_("invalid pattern (%s): %s") % (k, p))
         raise util.Abort(_("invalid pattern"))
 
 def _roots(kindpats):
     '''return roots and exact explicitly listed files from patterns
 
     >>> _roots([('glob', 'g/*', ''), ('glob', 'g', ''), ('glob', 'g*', '')])
     ['g', 'g', '.']
     >>> _roots([('relpath', 'r', ''), ('path', 'p/p', ''), ('path', '', '')])
     ['r', 'p/p', '.']
     >>> _roots([('relglob', 'rg*', ''), ('re', 're/', ''), ('relre', 'rr', '')])
     ['.', '.', '.']
     '''
     r = []
     for kind, pat, source in kindpats:
         if kind == 'glob': # find the non-glob prefix
             root = []
             for p in pat.split('/'):
                 if '[' in p or '{' in p or '*' in p or '?' in p:
                     break
                 root.append(p)
             r.append('/'.join(root) or '.')
         elif kind in ('relpath', 'path'):
             r.append(pat or '.')
         else: # relglob, re, relre
             r.append('.')
     return r
 
 def _anypats(kindpats):
     for kind, pat, source in kindpats:
         if kind in ('glob', 're', 'relglob', 'relre', 'set'):
             return True
 
 _commentre = None
 
 def readpatternfile(filepath, warn):
     '''parse a pattern file, returning a list of
     patterns. These patterns should be given to compile()
     to be validated and converted into a match function.
 
     trailing white space is dropped.
     the escape character is backslash.
     comments start with #.
     empty lines are skipped.
 
     lines can be of the following formats:
 
     syntax: regexp # defaults following lines to non-rooted regexps
     syntax: glob   # defaults following lines to non-rooted globs
     re:pattern     # non-rooted regular expression
     glob:pattern   # non-rooted glob
     pattern        # pattern of the current default type'''
 
     syntaxes = {'re': 'relre:', 'regexp': 'relre:', 'glob': 'relglob:',
                 'include': 'include'}
     syntax = 'relre:'
     patterns = []
 
     fp = open(filepath)
     for line in fp:
         if "#" in line:
             global _commentre
             if not _commentre:
                 _commentre = re.compile(r'((^|[^\\])(\\\\)*)#.*')
             # remove comments prefixed by an even number of escapes
             line = _commentre.sub(r'\1', line)
             # fixup properly escaped comments that survived the above
             line = line.replace("\\#", "#")
         line = line.rstrip()
         if not line:
             continue
 
         if line.startswith('syntax:'):
             s = line[7:].strip()
             try:
                 syntax = syntaxes[s]
             except KeyError:
                 if warn:
                     warn(_("%s: ignoring invalid syntax '%s'\n") %
                          (filepath, s))
             continue
 
         linesyntax = syntax
         for s, rels in syntaxes.iteritems():
             if line.startswith(rels):
                 linesyntax = rels
                 line = line[len(rels):]
                 break
             elif line.startswith(s+':'):
                 linesyntax = rels
                 line = line[len(s) + 1:]
                 break
         patterns.append(linesyntax + line)
     fp.close()
     return patterns

tests/test-treemanifest.t

@@ -1,282 +1,379
 
 Set up repo
 
   $ hg --config experimental.treemanifest=True init repo
   $ cd repo
 
 Requirements get set on init
 
   $ grep treemanifest .hg/requires
   treemanifest
 
 Without directories, looks like any other repo
 
   $ echo 0 > a
   $ echo 0 > b
   $ hg ci -Aqm initial
   $ hg debugdata -m 0
   a\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   b\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
 
 Submanifest is stored in separate revlog
 
   $ mkdir dir1
   $ echo 1 > dir1/a
   $ echo 1 > dir1/b
   $ echo 1 > e
   $ hg ci -Aqm 'add dir1'
   $ hg debugdata -m 1
   a\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   b\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   dir1\x008b3ffd73f901e83304c83d33132c8e774ceac44ed (esc)
   e\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
   $ hg debugdata --dir dir1 0
   a\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
   b\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
 
 Can add nested directories
 
   $ mkdir dir1/dir1
   $ echo 2 > dir1/dir1/a
   $ echo 2 > dir1/dir1/b
   $ mkdir dir1/dir2
   $ echo 2 > dir1/dir2/a
   $ echo 2 > dir1/dir2/b
   $ hg ci -Aqm 'add dir1/dir1'
   $ hg files -r .
   a
   b
   dir1/a (glob)
   dir1/b (glob)
   dir1/dir1/a (glob)
   dir1/dir1/b (glob)
   dir1/dir2/a (glob)
   dir1/dir2/b (glob)
   e
 
 Revision is not created for unchanged directory
 
   $ mkdir dir2
   $ echo 3 > dir2/a
   $ hg add dir2
   adding dir2/a (glob)
   $ hg debugindex --dir dir1 > before
   $ hg ci -qm 'add dir2'
   $ hg debugindex --dir dir1 > after
   $ diff before after
   $ rm before after
 
 Removing directory does not create an revlog entry
 
   $ hg rm dir1/dir1
   removing dir1/dir1/a (glob)
   removing dir1/dir1/b (glob)
   $ hg debugindex --dir dir1/dir1 > before
   $ hg ci -qm 'remove dir1/dir1'
   $ hg debugindex --dir dir1/dir1 > after
   $ diff before after
   $ rm before after
 
 Check that hg files (calls treemanifest.walk()) works
 without loading all directory revlogs
 
   $ hg co 'desc("add dir2")'
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ mv .hg/store/meta/dir2 .hg/store/meta/dir2-backup
   $ hg files -r . dir1
   dir1/a (glob)
   dir1/b (glob)
   dir1/dir1/a (glob)
   dir1/dir1/b (glob)
   dir1/dir2/a (glob)
   dir1/dir2/b (glob)
 
 Check that status between revisions works (calls treemanifest.matches())
 without loading all directory revlogs
 
   $ hg status --rev 'desc("add dir1")' --rev . dir1
   A dir1/dir1/a
   A dir1/dir1/b
   A dir1/dir2/a
   A dir1/dir2/b
   $ mv .hg/store/meta/dir2-backup .hg/store/meta/dir2
 
 Merge creates 2-parent revision of directory revlog
 
   $ echo 5 > dir1/a
   $ hg ci -Aqm 'modify dir1/a'
   $ hg co '.^'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 6 > dir1/b
   $ hg ci -Aqm 'modify dir1/b'
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'conflict-free merge involving dir1/'
   $ cat dir1/a
   5
   $ cat dir1/b
   6
   $ hg debugindex --dir dir1
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0      54      0       1 8b3ffd73f901 000000000000 000000000000
        1        54      68      0       2 b66d046c644f 8b3ffd73f901 000000000000
        2       122      12      0       4 b87265673c8a b66d046c644f 000000000000
        3       134      95      0       5 aa5d3adcec72 b66d046c644f 000000000000
        4       229      81      0       6 e29b066b91ad b66d046c644f 000000000000
        5       310     107      5       7 a120ce2b83f5 e29b066b91ad aa5d3adcec72
 
 Merge keeping directory from parent 1 does not create revlog entry. (Note that
 dir1's manifest does change, but only because dir1/a's filelog changes.)
 
   $ hg co 'desc("add dir2")'
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 8 > dir2/a
   $ hg ci -m 'modify dir2/a'
   created new head
 
   $ hg debugindex --dir dir2 > before
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg revert -r 'desc("modify dir2/a")' .
   reverting dir1/a (glob)
   $ hg ci -m 'merge, keeping parent 1'
   $ hg debugindex --dir dir2 > after
   $ diff before after
   $ rm before after
 
 Merge keeping directory from parent 2 does not create revlog entry. (Note that
 dir2's manifest does change, but only because dir2/a's filelog changes.)
 
   $ hg co 'desc("modify dir2/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg debugindex --dir dir1 > before
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg revert -r 'desc("modify dir1/a")' .
   reverting dir2/a (glob)
   $ hg ci -m 'merge, keeping parent 2'
   created new head
   $ hg debugindex --dir dir1 > after
   $ diff before after
   $ rm before after
 
 Create flat source repo for tests with mixed flat/tree manifests
 
   $ cd ..
   $ hg init repo-flat
   $ cd repo-flat
 
 Create a few commits with flat manifest
 
   $ echo 0 > a
   $ echo 0 > b
   $ echo 0 > e
   $ for d in dir1 dir1/dir1 dir1/dir2 dir2
   > do
   >   mkdir $d
   >   echo 0 > $d/a
   >   echo 0 > $d/b
   > done
   $ hg ci -Aqm initial
 
   $ echo 1 > a
   $ echo 1 > dir1/a
   $ echo 1 > dir1/dir1/a
   $ hg ci -Aqm 'modify on branch 1'
 
   $ hg co 0
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 2 > b
   $ echo 2 > dir1/b
   $ echo 2 > dir1/dir1/b
   $ hg ci -Aqm 'modify on branch 2'
 
   $ hg merge 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'merge of flat manifests to new flat manifest'
 
 Create clone with tree manifests enabled
 
   $ cd ..
   $ hg clone --pull --config experimental.treemanifest=1 repo-flat repo-mixed
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 17 changes to 11 files
   updating to branch default
   11 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cd repo-mixed
   $ test -f .hg/store/meta
   [1]
   $ grep treemanifest .hg/requires
   treemanifest
 
 Commit should store revlog per directory
 
   $ hg co 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 3 > a
   $ echo 3 > dir1/a
   $ echo 3 > dir1/dir1/a
   $ hg ci -m 'first tree'
   created new head
   $ find .hg/store/meta | sort
   .hg/store/meta
   .hg/store/meta/dir1
   .hg/store/meta/dir1/00manifest.i
   .hg/store/meta/dir1/dir1
   .hg/store/meta/dir1/dir1/00manifest.i
   .hg/store/meta/dir1/dir2
   .hg/store/meta/dir1/dir2/00manifest.i
   .hg/store/meta/dir2
   .hg/store/meta/dir2/00manifest.i
 
 Merge of two trees
 
   $ hg co 2
   6 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg merge 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'merge of flat manifests to new tree manifest'
   created new head
   $ hg diff -r 3
 
 Parent of tree root manifest should be flat manifest, and two for merge
 
   $ hg debugindex -m
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0      80      0       0 40536115ed9e 000000000000 000000000000
        1        80      83      0       1 f3376063c255 40536115ed9e 000000000000
        2       163     103      0       2 5d9b9da231a2 40536115ed9e 000000000000
        3       266      83      0       3 d17d663cbd8a 5d9b9da231a2 f3376063c255
        4       349     132      4       4 c05a51345f86 f3376063c255 000000000000
        5       481     110      4       5 82594b1f557d 5d9b9da231a2 f3376063c255
 
 
 Status across flat/tree boundary should work
 
   $ hg status --rev '.^' --rev .
   M a
   M dir1/a
   M dir1/dir1/a
 
 
 Turning off treemanifest config has no effect
 
   $ hg debugindex .hg/store/meta/dir1/00manifest.i
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0     125      0       4 63c9c0557d24 000000000000 000000000000
        1       125     109      0       5 23d12a1f6e0e 000000000000 000000000000
   $ echo 2 > dir1/a
   $ hg --config experimental.treemanifest=False ci -qm 'modify dir1/a'
   $ hg debugindex .hg/store/meta/dir1/00manifest.i
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0     125      0       4 63c9c0557d24 000000000000 000000000000
        1       125     109      0       5 23d12a1f6e0e 000000000000 000000000000
        2       234      55      0       6 3cb2d87b4250 23d12a1f6e0e 000000000000
+
+Create deeper repo with tree manifests.
+
+  $ cd ..
+  $ hg --config experimental.treemanifest=True init deeprepo
+  $ cd deeprepo
+
+  $ mkdir a
+  $ mkdir b
+  $ mkdir b/bar
+  $ mkdir b/bar/orange
+  $ mkdir b/bar/orange/fly
+  $ mkdir b/foo
+  $ mkdir b/foo/apple
+  $ mkdir b/foo/apple/bees
+
+  $ touch a/one.txt
+  $ touch a/two.txt
+  $ touch b/bar/fruits.txt
+  $ touch b/bar/orange/fly/gnat.py
+  $ touch b/bar/orange/fly/housefly.txt
+  $ touch b/foo/apple/bees/flower.py
+  $ touch c.txt
+  $ touch d.py
+
+  $ hg ci -Aqm 'initial'
+
+We'll see that visitdir works by removing some treemanifest revlogs and running
+the files command with various parameters.
+
+Test files from the root.
+
+  $ hg files -r .
+  a/one.txt
+  a/two.txt
+  b/bar/fruits.txt
+  b/bar/orange/fly/gnat.py
+  b/bar/orange/fly/housefly.txt
+  b/foo/apple/bees/flower.py
+  c.txt
+  d.py
+
+Test files for a subdirectory.
+
+  $ mv .hg/store/meta/a oldmf
+  $ hg files -r . b
+  b/bar/fruits.txt
+  b/bar/orange/fly/gnat.py
+  b/bar/orange/fly/housefly.txt
+  b/foo/apple/bees/flower.py
+  $ mv oldmf .hg/store/meta/a
+
+Test files with just includes and excludes.
+
+  $ mv .hg/store/meta/a oldmf
+  $ mv .hg/store/meta/b/bar/orange/fly oldmf2
+  $ mv .hg/store/meta/b/foo/apple/bees oldmf3
+  $ hg files -r . -I b/bar -X b/bar/orange/fly -I b/foo -X b/foo/apple/bees
+  b/bar/fruits.txt
+  $ mv oldmf .hg/store/meta/a
+  $ mv oldmf2 .hg/store/meta/b/bar/orange/fly
+  $ mv oldmf3 .hg/store/meta/b/foo/apple/bees
+
+Test files for a subdirectory, excluding a directory within it.
+
+  $ mv .hg/store/meta/a oldmf
+  $ mv .hg/store/meta/b/foo oldmf2
+  $ hg files -r . -X b/foo b
+  b/bar/fruits.txt
+  b/bar/orange/fly/gnat.py
+  b/bar/orange/fly/housefly.txt
+  $ mv oldmf .hg/store/meta/a
+  $ mv oldmf2 .hg/store/meta/b/foo
+
+Test files for a sub directory, including only a directory within it, and
+including an unrelated directory.
+
+  $ mv .hg/store/meta/a oldmf
+  $ mv .hg/store/meta/b/foo oldmf2
+  $ hg files -r . -I b/bar/orange -I a b
+  b/bar/orange/fly/gnat.py
+  b/bar/orange/fly/housefly.txt
+  $ mv oldmf .hg/store/meta/a
+  $ mv oldmf2 .hg/store/meta/b/foo
+
+Test files for a pattern, including a directory, and excluding a directory
+within that.
+
+  $ mv .hg/store/meta/a oldmf
+  $ mv .hg/store/meta/b/foo oldmf2
+  $ mv .hg/store/meta/b/bar/orange oldmf3
+  $ hg files -r . glob:**.txt -I b/bar -X b/bar/orange
+  b/bar/fruits.txt
+  $ mv oldmf .hg/store/meta/a
+  $ mv oldmf2 .hg/store/meta/b/foo
+  $ mv oldmf3 .hg/store/meta/b/bar/orange
+