py3: make sure regexes are bytes

# skip-blame because we are adding just b'' prefixes

Differential Revision: https://phab.mercurial-scm.org/D2475

hgext/convert/cvsps.py

@@ -1,952 +1,952 b''
 # Mercurial built-in replacement for cvsps.
 #
 # Copyright 2008, Frank Kingswood <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 from __future__ import absolute_import
 
 import os
 import re
 
 from mercurial.i18n import _
 from mercurial import (
     encoding,
     error,
     hook,
     pycompat,
     util,
 )
 
 pickle = util.pickle
 
 class logentry(object):
     '''Class logentry has the following attributes:
         .author    - author name as CVS knows it
         .branch    - name of branch this revision is on
         .branches  - revision tuple of branches starting at this revision
         .comment   - commit message
         .commitid  - CVS commitid or None
         .date      - the commit date as a (time, tz) tuple
         .dead      - true if file revision is dead
         .file      - Name of file
         .lines     - a tuple (+lines, -lines) or None
         .parent    - Previous revision of this entry
         .rcs       - name of file as returned from CVS
         .revision  - revision number as tuple
         .tags      - list of tags on the file
         .synthetic - is this a synthetic "file ... added on ..." revision?
         .mergepoint - the branch that has been merged from (if present in
                       rlog output) or None
         .branchpoints - the branches that start at the current entry or empty
     '''
     def __init__(self, **entries):
         self.synthetic = False
         self.__dict__.update(entries)
 
     def __repr__(self):
         items = ("%s=%r"%(k, self.__dict__[k]) for k in sorted(self.__dict__))
         return "%s(%s)"%(type(self).__name__, ", ".join(items))
 
 class logerror(Exception):
     pass
 
 def getrepopath(cvspath):
     """Return the repository path from a CVS path.
 
     >>> getrepopath(b'/foo/bar')
     '/foo/bar'
     >>> getrepopath(b'c:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:10/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:10c:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:c:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:[e-mail]:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b':pserver:[e-mail]:c:/foo/bar')
     '/foo/bar'
     >>> getrepopath(b'user@server/path/to/repository')
     '/path/to/repository'
     """
     # According to CVS manual, CVS paths are expressed like:
     # [:method:][[user][:password]@]hostname[:[port]]/path/to/repository
     #
     # CVSpath is splitted into parts and then position of the first occurrence
     # of the '/' char after the '@' is located. The solution is the rest of the
     # string after that '/' sign including it
 
     parts = cvspath.split(':')
     atposition = parts[-1].find('@')
     start = 0
 
     if atposition != -1:
         start = atposition
 
     repopath = parts[-1][parts[-1].find('/', start):]
     return repopath
 
 def createlog(ui, directory=None, root="", rlog=True, cache=None):
     '''Collect the CVS rlog'''
 
     # Because we store many duplicate commit log messages, reusing strings
     # saves a lot of memory and pickle storage space.
     _scache = {}
     def scache(s):
         "return a shared version of a string"
         return _scache.setdefault(s, s)
 
     ui.status(_('collecting CVS rlog\n'))
 
     log = []      # list of logentry objects containing the CVS state
 
     # patterns to match in CVS (r)log output, by state of use
     re_00 = re.compile('RCS file: (.+)$')
     re_01 = re.compile('cvs \\[r?log aborted\\]: (.+)$')
     re_02 = re.compile('cvs (r?log|server): (.+)\n$')
     re_03 = re.compile("(Cannot access.+CVSROOT)|"
                        "(can't create temporary directory.+)$")
     re_10 = re.compile('Working file: (.+)$')
     re_20 = re.compile('symbolic names:')
     re_30 = re.compile('\t(.+): ([\\d.]+)$')
     re_31 = re.compile('----------------------------$')
     re_32 = re.compile('======================================='
                        '======================================$')
     re_50 = re.compile('revision ([\\d.]+)(\s+locked by:\s+.+;)?$')
     re_60 = re.compile(r'date:\s+(.+);\s+author:\s+(.+);\s+state:\s+(.+?);'
                        r'(\s+lines:\s+(\+\d+)?\s+(-\d+)?;)?'
                        r'(\s+commitid:\s+([^;]+);)?'
                        r'(.*mergepoint:\s+([^;]+);)?')
     re_70 = re.compile('branches: (.+);$')
 
     file_added_re = re.compile(r'file [^/]+ was (initially )?added on branch')
 
     prefix = ''   # leading path to strip of what we get from CVS
 
     if directory is None:
         # Current working directory
 
         # Get the real directory in the repository
         try:
             prefix = open(os.path.join('CVS','Repository'), 'rb').read().strip()
             directory = prefix
             if prefix == ".":
                 prefix = ""
         except IOError:
             raise logerror(_('not a CVS sandbox'))
 
         if prefix and not prefix.endswith(pycompat.ossep):
             prefix += pycompat.ossep
 
         # Use the Root file in the sandbox, if it exists
         try:
             root = open(os.path.join('CVS','Root'), 'rb').read().strip()
         except IOError:
             pass
 
     if not root:
         root = encoding.environ.get('CVSROOT', '')
 
     # read log cache if one exists
     oldlog = []
     date = None
 
     if cache:
         cachedir = os.path.expanduser('~/.hg.cvsps')
         if not os.path.exists(cachedir):
             os.mkdir(cachedir)
 
         # The cvsps cache pickle needs a uniquified name, based on the
         # repository location. The address may have all sort of nasties
         # in it, slashes, colons and such. So here we take just the
         # alphanumeric characters, concatenated in a way that does not
         # mix up the various components, so that
         #    :pserver:user@server:/path
         # and
         #    /pserver/user/server/path
         # are mapped to different cache file names.
         cachefile = root.split(":") + [directory, "cache"]
-        cachefile = ['-'.join(re.findall(r'\w+', s)) for s in cachefile if s]
+        cachefile = ['-'.join(re.findall(br'\w+', s)) for s in cachefile if s]
         cachefile = os.path.join(cachedir,
                                  '.'.join([s for s in cachefile if s]))
 
     if cache == 'update':
         try:
             ui.note(_('reading cvs log cache %s\n') % cachefile)
             oldlog = pickle.load(open(cachefile, 'rb'))
             for e in oldlog:
                 if not (util.safehasattr(e, 'branchpoints') and
                         util.safehasattr(e, 'commitid') and
                         util.safehasattr(e, 'mergepoint')):
                     ui.status(_('ignoring old cache\n'))
                     oldlog = []
                     break
 
             ui.note(_('cache has %d log entries\n') % len(oldlog))
         except Exception as e:
             ui.note(_('error reading cache: %r\n') % e)
 
         if oldlog:
             date = oldlog[-1].date    # last commit date as a (time,tz) tuple
             date = util.datestr(date, '%Y/%m/%d %H:%M:%S %1%2')
 
     # build the CVS commandline
     cmd = ['cvs', '-q']
     if root:
         cmd.append('-d%s' % root)
         p = util.normpath(getrepopath(root))
         if not p.endswith('/'):
             p += '/'
         if prefix:
             # looks like normpath replaces "" by "."
             prefix = p + util.normpath(prefix)
         else:
             prefix = p
     cmd.append(['log', 'rlog'][rlog])
     if date:
         # no space between option and date string
         cmd.append('-d>%s' % date)
     cmd.append(directory)
 
     # state machine begins here
     tags = {}     # dictionary of revisions on current file with their tags
     branchmap = {} # mapping between branch names and revision numbers
     rcsmap = {}
     state = 0
     store = False # set when a new record can be appended
 
     cmd = [util.shellquote(arg) for arg in cmd]
     ui.note(_("running %s\n") % (' '.join(cmd)))
     ui.debug("prefix=%r directory=%r root=%r\n" % (prefix, directory, root))
 
     pfp = util.popen(' '.join(cmd))
     peek = pfp.readline()
     while True:
         line = peek
         if line == '':
             break
         peek = pfp.readline()
         if line.endswith('\n'):
             line = line[:-1]
         #ui.debug('state=%d line=%r\n' % (state, line))
 
         if state == 0:
             # initial state, consume input until we see 'RCS file'
             match = re_00.match(line)
             if match:
                 rcs = match.group(1)
                 tags = {}
                 if rlog:
                     filename = util.normpath(rcs[:-2])
                     if filename.startswith(prefix):
                         filename = filename[len(prefix):]
                     if filename.startswith('/'):
                         filename = filename[1:]
                     if filename.startswith('Attic/'):
                         filename = filename[6:]
                     else:
                         filename = filename.replace('/Attic/', '/')
                     state = 2
                     continue
                 state = 1
                 continue
             match = re_01.match(line)
             if match:
                 raise logerror(match.group(1))
             match = re_02.match(line)
             if match:
                 raise logerror(match.group(2))
             if re_03.match(line):
                 raise logerror(line)
 
         elif state == 1:
             # expect 'Working file' (only when using log instead of rlog)
             match = re_10.match(line)
             assert match, _('RCS file must be followed by working file')
             filename = util.normpath(match.group(1))
             state = 2
 
         elif state == 2:
             # expect 'symbolic names'
             if re_20.match(line):
                 branchmap = {}
                 state = 3
 
         elif state == 3:
             # read the symbolic names and store as tags
             match = re_30.match(line)
             if match:
                 rev = [int(x) for x in match.group(2).split('.')]
 
                 # Convert magic branch number to an odd-numbered one
                 revn = len(rev)
                 if revn > 3 and (revn % 2) == 0 and rev[-2] == 0:
                     rev = rev[:-2] + rev[-1:]
                 rev = tuple(rev)
 
                 if rev not in tags:
                     tags[rev] = []
                 tags[rev].append(match.group(1))
                 branchmap[match.group(1)] = match.group(2)
 
             elif re_31.match(line):
                 state = 5
             elif re_32.match(line):
                 state = 0
 
         elif state == 4:
             # expecting '------' separator before first revision
             if re_31.match(line):
                 state = 5
             else:
                 assert not re_32.match(line), _('must have at least '
                                                 'some revisions')
 
         elif state == 5:
             # expecting revision number and possibly (ignored) lock indication
             # we create the logentry here from values stored in states 0 to 4,
             # as this state is re-entered for subsequent revisions of a file.
             match = re_50.match(line)
             assert match, _('expected revision number')
             e = logentry(rcs=scache(rcs),
                          file=scache(filename),
                          revision=tuple([int(x) for x in
                                          match.group(1).split('.')]),
                          branches=[],
                          parent=None,
                          commitid=None,
                          mergepoint=None,
                          branchpoints=set())
 
             state = 6
 
         elif state == 6:
             # expecting date, author, state, lines changed
             match = re_60.match(line)
             assert match, _('revision must be followed by date line')
             d = match.group(1)
             if d[2] == '/':
                 # Y2K
                 d = '19' + d
 
             if len(d.split()) != 3:
                 # cvs log dates always in GMT
                 d = d + ' UTC'
             e.date = util.parsedate(d, ['%y/%m/%d %H:%M:%S',
                                         '%Y/%m/%d %H:%M:%S',
                                         '%Y-%m-%d %H:%M:%S'])
             e.author = scache(match.group(2))
             e.dead = match.group(3).lower() == 'dead'
 
             if match.group(5):
                 if match.group(6):
                     e.lines = (int(match.group(5)), int(match.group(6)))
                 else:
                     e.lines = (int(match.group(5)), 0)
             elif match.group(6):
                 e.lines = (0, int(match.group(6)))
             else:
                 e.lines = None
 
             if match.group(7): # cvs 1.12 commitid
                 e.commitid = match.group(8)
 
             if match.group(9): # cvsnt mergepoint
                 myrev = match.group(10).split('.')
                 if len(myrev) == 2: # head
                     e.mergepoint = 'HEAD'
                 else:
                     myrev = '.'.join(myrev[:-2] + ['0', myrev[-2]])
                     branches = [b for b in branchmap if branchmap[b] == myrev]
                     assert len(branches) == 1, ('unknown branch: %s'
                                                 % e.mergepoint)
                     e.mergepoint = branches[0]
 
             e.comment = []
             state = 7
 
         elif state == 7:
             # read the revision numbers of branches that start at this revision
             # or store the commit log message otherwise
             m = re_70.match(line)
             if m:
                 e.branches = [tuple([int(y) for y in x.strip().split('.')])
                                 for x in m.group(1).split(';')]
                 state = 8
             elif re_31.match(line) and re_50.match(peek):
                 state = 5
                 store = True
             elif re_32.match(line):
                 state = 0
                 store = True
             else:
                 e.comment.append(line)
 
         elif state == 8:
             # store commit log message
             if re_31.match(line):
                 cpeek = peek
                 if cpeek.endswith('\n'):
                     cpeek = cpeek[:-1]
                 if re_50.match(cpeek):
                     state = 5
                     store = True
                 else:
                     e.comment.append(line)
             elif re_32.match(line):
                 state = 0
                 store = True
             else:
                 e.comment.append(line)
 
         # When a file is added on a branch B1, CVS creates a synthetic
         # dead trunk revision 1.1 so that the branch has a root.
         # Likewise, if you merge such a file to a later branch B2 (one
         # that already existed when the file was added on B1), CVS
         # creates a synthetic dead revision 1.1.x.1 on B2.  Don't drop
         # these revisions now, but mark them synthetic so
         # createchangeset() can take care of them.
         if (store and
               e.dead and
               e.revision[-1] == 1 and      # 1.1 or 1.1.x.1
               len(e.comment) == 1 and
               file_added_re.match(e.comment[0])):
             ui.debug('found synthetic revision in %s: %r\n'
                      % (e.rcs, e.comment[0]))
             e.synthetic = True
 
         if store:
             # clean up the results and save in the log.
             store = False
             e.tags = sorted([scache(x) for x in tags.get(e.revision, [])])
             e.comment = scache('\n'.join(e.comment))
 
             revn = len(e.revision)
             if revn > 3 and (revn % 2) == 0:
                 e.branch = tags.get(e.revision[:-1], [None])[0]
             else:
                 e.branch = None
 
             # find the branches starting from this revision
             branchpoints = set()
             for branch, revision in branchmap.iteritems():
                 revparts = tuple([int(i) for i in revision.split('.')])
                 if len(revparts) < 2: # bad tags
                     continue
                 if revparts[-2] == 0 and revparts[-1] % 2 == 0:
                     # normal branch
                     if revparts[:-2] == e.revision:
                         branchpoints.add(branch)
                 elif revparts == (1, 1, 1): # vendor branch
                     if revparts in e.branches:
                         branchpoints.add(branch)
             e.branchpoints = branchpoints
 
             log.append(e)
 
             rcsmap[e.rcs.replace('/Attic/', '/')] = e.rcs
 
             if len(log) % 100 == 0:
                 ui.status(util.ellipsis('%d %s' % (len(log), e.file), 80)+'\n')
 
     log.sort(key=lambda x: (x.rcs, x.revision))
 
     # find parent revisions of individual files
     versions = {}
     for e in sorted(oldlog, key=lambda x: (x.rcs, x.revision)):
         rcs = e.rcs.replace('/Attic/', '/')
         if rcs in rcsmap:
             e.rcs = rcsmap[rcs]
         branch = e.revision[:-1]
         versions[(e.rcs, branch)] = e.revision
 
     for e in log:
         branch = e.revision[:-1]
         p = versions.get((e.rcs, branch), None)
         if p is None:
             p = e.revision[:-2]
         e.parent = p
         versions[(e.rcs, branch)] = e.revision
 
     # update the log cache
     if cache:
         if log:
             # join up the old and new logs
             log.sort(key=lambda x: x.date)
 
             if oldlog and oldlog[-1].date >= log[0].date:
                 raise logerror(_('log cache overlaps with new log entries,'
                                  ' re-run without cache.'))
 
             log = oldlog + log
 
             # write the new cachefile
             ui.note(_('writing cvs log cache %s\n') % cachefile)
             pickle.dump(log, open(cachefile, 'wb'))
         else:
             log = oldlog
 
     ui.status(_('%d log entries\n') % len(log))
 
     encodings = ui.configlist('convert', 'cvsps.logencoding')
     if encodings:
         def revstr(r):
             # this is needed, because logentry.revision is a tuple of "int"
             # (e.g. (1, 2) for "1.2")
             return '.'.join(pycompat.maplist(pycompat.bytestr, r))
 
         for entry in log:
             comment = entry.comment
             for e in encodings:
                 try:
                     entry.comment = comment.decode(e).encode('utf-8')
                     if ui.debugflag:
                         ui.debug("transcoding by %s: %s of %s\n" %
                                  (e, revstr(entry.revision), entry.file))
                     break
                 except UnicodeDecodeError:
                     pass # try next encoding
                 except LookupError as inst: # unknown encoding, maybe
                     raise error.Abort(inst,
                                       hint=_('check convert.cvsps.logencoding'
                                              ' configuration'))
             else:
                 raise error.Abort(_("no encoding can transcode"
                                     " CVS log message for %s of %s")
                                   % (revstr(entry.revision), entry.file),
                                   hint=_('check convert.cvsps.logencoding'
                                          ' configuration'))
 
     hook.hook(ui, None, "cvslog", True, log=log)
 
     return log
 
 
 class changeset(object):
     '''Class changeset has the following attributes:
         .id        - integer identifying this changeset (list index)
         .author    - author name as CVS knows it
         .branch    - name of branch this changeset is on, or None
         .comment   - commit message
         .commitid  - CVS commitid or None
         .date      - the commit date as a (time,tz) tuple
         .entries   - list of logentry objects in this changeset
         .parents   - list of one or two parent changesets
         .tags      - list of tags on this changeset
         .synthetic - from synthetic revision "file ... added on branch ..."
         .mergepoint- the branch that has been merged from or None
         .branchpoints- the branches that start at the current entry or empty
     '''
     def __init__(self, **entries):
         self.id = None
         self.synthetic = False
         self.__dict__.update(entries)
 
     def __repr__(self):
         items = ("%s=%r"%(k, self.__dict__[k]) for k in sorted(self.__dict__))
         return "%s(%s)"%(type(self).__name__, ", ".join(items))
 
 def createchangeset(ui, log, fuzz=60, mergefrom=None, mergeto=None):
     '''Convert log into changesets.'''
 
     ui.status(_('creating changesets\n'))
 
     # try to order commitids by date
     mindate = {}
     for e in log:
         if e.commitid:
             mindate[e.commitid] = min(e.date, mindate.get(e.commitid))
 
     # Merge changesets
     log.sort(key=lambda x: (mindate.get(x.commitid), x.commitid, x.comment,
                             x.author, x.branch, x.date, x.branchpoints))
 
     changesets = []
     files = set()
     c = None
     for i, e in enumerate(log):
 
         # Check if log entry belongs to the current changeset or not.
 
         # Since CVS is file-centric, two different file revisions with
         # different branchpoints should be treated as belonging to two
         # different changesets (and the ordering is important and not
         # honoured by cvsps at this point).
         #
         # Consider the following case:
         # foo 1.1 branchpoints: [MYBRANCH]
         # bar 1.1 branchpoints: [MYBRANCH, MYBRANCH2]
         #
         # Here foo is part only of MYBRANCH, but not MYBRANCH2, e.g. a
         # later version of foo may be in MYBRANCH2, so foo should be the
         # first changeset and bar the next and MYBRANCH and MYBRANCH2
         # should both start off of the bar changeset. No provisions are
         # made to ensure that this is, in fact, what happens.
         if not (c and e.branchpoints == c.branchpoints and
                 (# cvs commitids
                  (e.commitid is not None and e.commitid == c.commitid) or
                  (# no commitids, use fuzzy commit detection
                   (e.commitid is None or c.commitid is None) and
                    e.comment == c.comment and
                    e.author == c.author and
                    e.branch == c.branch and
                    ((c.date[0] + c.date[1]) <=
                     (e.date[0] + e.date[1]) <=
                     (c.date[0] + c.date[1]) + fuzz) and
                    e.file not in files))):
             c = changeset(comment=e.comment, author=e.author,
                           branch=e.branch, date=e.date,
                           entries=[], mergepoint=e.mergepoint,
                           branchpoints=e.branchpoints, commitid=e.commitid)
             changesets.append(c)
 
             files = set()
             if len(changesets) % 100 == 0:
                 t = '%d %s' % (len(changesets), repr(e.comment)[1:-1])
                 ui.status(util.ellipsis(t, 80) + '\n')
 
         c.entries.append(e)
         files.add(e.file)
         c.date = e.date       # changeset date is date of latest commit in it
 
     # Mark synthetic changesets
 
     for c in changesets:
         # Synthetic revisions always get their own changeset, because
         # the log message includes the filename.  E.g. if you add file3
         # and file4 on a branch, you get four log entries and three
         # changesets:
         #   "File file3 was added on branch ..." (synthetic, 1 entry)
         #   "File file4 was added on branch ..." (synthetic, 1 entry)
         #   "Add file3 and file4 to fix ..."     (real, 2 entries)
         # Hence the check for 1 entry here.
         c.synthetic = len(c.entries) == 1 and c.entries[0].synthetic
 
     # Sort files in each changeset
 
     def entitycompare(l, r):
         'Mimic cvsps sorting order'
         l = l.file.split('/')
         r = r.file.split('/')
         nl = len(l)
         nr = len(r)
         n = min(nl, nr)
         for i in range(n):
             if i + 1 == nl and nl < nr:
                 return -1
             elif i + 1 == nr and nl > nr:
                 return +1
             elif l[i] < r[i]:
                 return -1
             elif l[i] > r[i]:
                 return +1
         return 0
 
     for c in changesets:
         c.entries.sort(entitycompare)
 
     # Sort changesets by date
 
     odd = set()
     def cscmp(l, r):
         d = sum(l.date) - sum(r.date)
         if d:
             return d
 
         # detect vendor branches and initial commits on a branch
         le = {}
         for e in l.entries:
             le[e.rcs] = e.revision
         re = {}
         for e in r.entries:
             re[e.rcs] = e.revision
 
         d = 0
         for e in l.entries:
             if re.get(e.rcs, None) == e.parent:
                 assert not d
                 d = 1
                 break
 
         for e in r.entries:
             if le.get(e.rcs, None) == e.parent:
                 if d:
                     odd.add((l, r))
                 d = -1
                 break
         # By this point, the changesets are sufficiently compared that
         # we don't really care about ordering. However, this leaves
         # some race conditions in the tests, so we compare on the
         # number of files modified, the files contained in each
         # changeset, and the branchpoints in the change to ensure test
         # output remains stable.
 
         # recommended replacement for cmp from
         # https://docs.python.org/3.0/whatsnew/3.0.html
         c = lambda x, y: (x > y) - (x < y)
         # Sort bigger changes first.
         if not d:
             d = c(len(l.entries), len(r.entries))
         # Try sorting by filename in the change.
         if not d:
             d = c([e.file for e in l.entries], [e.file for e in r.entries])
         # Try and put changes without a branch point before ones with
         # a branch point.
         if not d:
             d = c(len(l.branchpoints), len(r.branchpoints))
         return d
 
     changesets.sort(cscmp)
 
     # Collect tags
 
     globaltags = {}
     for c in changesets:
         for e in c.entries:
             for tag in e.tags:
                 # remember which is the latest changeset to have this tag
                 globaltags[tag] = c
 
     for c in changesets:
         tags = set()
         for e in c.entries:
             tags.update(e.tags)
         # remember tags only if this is the latest changeset to have it
         c.tags = sorted(tag for tag in tags if globaltags[tag] is c)
 
     # Find parent changesets, handle {{mergetobranch BRANCHNAME}}
     # by inserting dummy changesets with two parents, and handle
     # {{mergefrombranch BRANCHNAME}} by setting two parents.
 
     if mergeto is None:
         mergeto = r'{{mergetobranch ([-\w]+)}}'
     if mergeto:
         mergeto = re.compile(mergeto)
 
     if mergefrom is None:
         mergefrom = r'{{mergefrombranch ([-\w]+)}}'
     if mergefrom:
         mergefrom = re.compile(mergefrom)
 
     versions = {}    # changeset index where we saw any particular file version
     branches = {}    # changeset index where we saw a branch
     n = len(changesets)
     i = 0
     while i < n:
         c = changesets[i]
 
         for f in c.entries:
             versions[(f.rcs, f.revision)] = i
 
         p = None
         if c.branch in branches:
             p = branches[c.branch]
         else:
             # first changeset on a new branch
             # the parent is a changeset with the branch in its
             # branchpoints such that it is the latest possible
             # commit without any intervening, unrelated commits.
 
             for candidate in xrange(i):
                 if c.branch not in changesets[candidate].branchpoints:
                     if p is not None:
                         break
                     continue
                 p = candidate
 
         c.parents = []
         if p is not None:
             p = changesets[p]
 
             # Ensure no changeset has a synthetic changeset as a parent.
             while p.synthetic:
                 assert len(p.parents) <= 1, \
                        _('synthetic changeset cannot have multiple parents')
                 if p.parents:
                     p = p.parents[0]
                 else:
                     p = None
                     break
 
             if p is not None:
                 c.parents.append(p)
 
         if c.mergepoint:
             if c.mergepoint == 'HEAD':
                 c.mergepoint = None
             c.parents.append(changesets[branches[c.mergepoint]])
 
         if mergefrom:
             m = mergefrom.search(c.comment)
             if m:
                 m = m.group(1)
                 if m == 'HEAD':
                     m = None
                 try:
                     candidate = changesets[branches[m]]
                 except KeyError:
                     ui.warn(_("warning: CVS commit message references "
                               "non-existent branch %r:\n%s\n")
                             % (m, c.comment))
                 if m in branches and c.branch != m and not candidate.synthetic:
                     c.parents.append(candidate)
 
         if mergeto:
             m = mergeto.search(c.comment)
             if m:
                 if m.groups():
                     m = m.group(1)
                     if m == 'HEAD':
                         m = None
                 else:
                     m = None   # if no group found then merge to HEAD
                 if m in branches and c.branch != m:
                     # insert empty changeset for merge
                     cc = changeset(
                         author=c.author, branch=m, date=c.date,
                         comment='convert-repo: CVS merge from branch %s'
                         % c.branch,
                         entries=[], tags=[],
                         parents=[changesets[branches[m]], c])
                     changesets.insert(i + 1, cc)
                     branches[m] = i + 1
 
                     # adjust our loop counters now we have inserted a new entry
                     n += 1
                     i += 2
                     continue
 
         branches[c.branch] = i
         i += 1
 
     # Drop synthetic changesets (safe now that we have ensured no other
     # changesets can have them as parents).
     i = 0
     while i < len(changesets):
         if changesets[i].synthetic:
             del changesets[i]
         else:
             i += 1
 
     # Number changesets
 
     for i, c in enumerate(changesets):
         c.id = i + 1
 
     if odd:
         for l, r in odd:
             if l.id is not None and r.id is not None:
                 ui.warn(_('changeset %d is both before and after %d\n')
                         % (l.id, r.id))
 
     ui.status(_('%d changeset entries\n') % len(changesets))
 
     hook.hook(ui, None, "cvschangesets", True, changesets=changesets)
 
     return changesets
 
 
 def debugcvsps(ui, *args, **opts):
     '''Read CVS rlog for current directory or named path in
     repository, and convert the log to changesets based on matching
     commit log entries and dates.
     '''
     opts = pycompat.byteskwargs(opts)
     if opts["new_cache"]:
         cache = "write"
     elif opts["update_cache"]:
         cache = "update"
     else:
         cache = None
 
     revisions = opts["revisions"]
 
     try:
         if args:
             log = []
             for d in args:
                 log += createlog(ui, d, root=opts["root"], cache=cache)
         else:
             log = createlog(ui, root=opts["root"], cache=cache)
     except logerror as e:
         ui.write("%r\n"%e)
         return
 
     changesets = createchangeset(ui, log, opts["fuzz"])
     del log
 
     # Print changesets (optionally filtered)
 
     off = len(revisions)
     branches = {}    # latest version number in each branch
     ancestors = {}   # parent branch
     for cs in changesets:
 
         if opts["ancestors"]:
             if cs.branch not in branches and cs.parents and cs.parents[0].id:
                 ancestors[cs.branch] = (changesets[cs.parents[0].id - 1].branch,
                                         cs.parents[0].id)
             branches[cs.branch] = cs.id
 
         # limit by branches
         if opts["branches"] and (cs.branch or 'HEAD') not in opts["branches"]:
             continue
 
         if not off:
             # Note: trailing spaces on several lines here are needed to have
             #       bug-for-bug compatibility with cvsps.
             ui.write('---------------------\n')
             ui.write(('PatchSet %d \n' % cs.id))
             ui.write(('Date: %s\n' % util.datestr(cs.date,
                                                  '%Y/%m/%d %H:%M:%S %1%2')))
             ui.write(('Author: %s\n' % cs.author))
             ui.write(('Branch: %s\n' % (cs.branch or 'HEAD')))
             ui.write(('Tag%s: %s \n' % (['', 's'][len(cs.tags) > 1],
                                   ','.join(cs.tags) or '(none)')))
             if cs.branchpoints:
                 ui.write(('Branchpoints: %s \n') %
                          ', '.join(sorted(cs.branchpoints)))
             if opts["parents"] and cs.parents:
                 if len(cs.parents) > 1:
                     ui.write(('Parents: %s\n' %
                              (','.join([str(p.id) for p in cs.parents]))))
                 else:
                     ui.write(('Parent: %d\n' % cs.parents[0].id))
 
             if opts["ancestors"]:
                 b = cs.branch
                 r = []
                 while b:
                     b, c = ancestors[b]
                     r.append('%s:%d:%d' % (b or "HEAD", c, branches[b]))
                 if r:
                     ui.write(('Ancestors: %s\n' % (','.join(r))))
 
             ui.write(('Log:\n'))
             ui.write('%s\n\n' % cs.comment)
             ui.write(('Members: \n'))
             for f in cs.entries:
                 fn = f.file
                 if fn.startswith(opts["prefix"]):
                     fn = fn[len(opts["prefix"]):]
                 ui.write('\t%s:%s->%s%s \n' % (
                         fn, '.'.join([str(x) for x in f.parent]) or 'INITIAL',
                         '.'.join([str(x) for x in f.revision]),
                         ['', '(DEAD)'][f.dead]))
             ui.write('\n')
 
         # have we seen the start tag?
         if revisions and off:
             if revisions[0] == str(cs.id) or \
                 revisions[0] in cs.tags:
                 off = False
 
         # see if we reached the end tag
         if len(revisions) > 1 and not off:
             if revisions[1] == str(cs.id) or \
                 revisions[1] in cs.tags:
                 break

hgext/convert/subversion.py

@@ -1,1356 +1,1356 b''
 # Subversion 1.4/1.5 Python API backend
 #
 # Copyright(C) 2007 Daniel Holth et al
 from __future__ import absolute_import
 
 import os
 import re
 import tempfile
 import xml.dom.minidom
 
 from mercurial.i18n import _
 from mercurial import (
     encoding,
     error,
     pycompat,
     util,
     vfs as vfsmod,
 )
 
 from . import common
 
 pickle = util.pickle
 stringio = util.stringio
 propertycache = util.propertycache
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 commandline = common.commandline
 commit = common.commit
 converter_sink = common.converter_sink
 converter_source = common.converter_source
 decodeargs = common.decodeargs
 encodeargs = common.encodeargs
 makedatetimestamp = common.makedatetimestamp
 mapfile = common.mapfile
 MissingTool = common.MissingTool
 NoRepo = common.NoRepo
 
 # Subversion stuff. Works best with very recent Python SVN bindings
 # e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing
 # these bindings.
 
 try:
     import svn
     import svn.client
     import svn.core
     import svn.ra
     import svn.delta
     from . import transport
     import warnings
     warnings.filterwarnings('ignore',
             module='svn.core',
             category=DeprecationWarning)
     svn.core.SubversionException # trigger import to catch error
 
 except ImportError:
     svn = None
 
 class SvnPathNotFound(Exception):
     pass
 
 def revsplit(rev):
     """Parse a revision string and return (uuid, path, revnum).
     >>> revsplit(b'svn:a2147622-4a9f-4db4-a8d3-13562ff547b2'
     ...          b'/proj%20B/mytrunk/mytrunk@1')
     ('a2147622-4a9f-4db4-a8d3-13562ff547b2', '/proj%20B/mytrunk/mytrunk', 1)
     >>> revsplit(b'svn:8af66a51-67f5-4354-b62c-98d67cc7be1d@1')
     ('', '', 1)
     >>> revsplit(b'@7')
     ('', '', 7)
     >>> revsplit(b'7')
     ('', '', 0)
     >>> revsplit(b'bad')
     ('', '', 0)
     """
     parts = rev.rsplit('@', 1)
     revnum = 0
     if len(parts) > 1:
         revnum = int(parts[1])
     parts = parts[0].split('/', 1)
     uuid = ''
     mod = ''
     if len(parts) > 1 and parts[0].startswith('svn:'):
         uuid = parts[0][4:]
         mod = '/' + parts[1]
     return uuid, mod, revnum
 
 def quote(s):
     # As of svn 1.7, many svn calls expect "canonical" paths. In
     # theory, we should call svn.core.*canonicalize() on all paths
     # before passing them to the API.  Instead, we assume the base url
     # is canonical and copy the behaviour of svn URL encoding function
     # so we can extend it safely with new components. The "safe"
     # characters were taken from the "svn_uri__char_validity" table in
     # libsvn_subr/path.c.
     return urlreq.quote(s, "!$&'()*+,-./:=@_~")
 
 def geturl(path):
     try:
         return svn.client.url_from_path(svn.core.svn_path_canonicalize(path))
     except svn.core.SubversionException:
         # svn.client.url_from_path() fails with local repositories
         pass
     if os.path.isdir(path):
         path = os.path.normpath(os.path.abspath(path))
         if pycompat.iswindows:
             path = '/' + util.normpath(path)
         # Module URL is later compared with the repository URL returned
         # by svn API, which is UTF-8.
         path = encoding.tolocal(path)
         path = 'file://%s' % quote(path)
     return svn.core.svn_path_canonicalize(path)
 
 def optrev(number):
     optrev = svn.core.svn_opt_revision_t()
     optrev.kind = svn.core.svn_opt_revision_number
     optrev.value.number = number
     return optrev
 
 class changedpath(object):
     def __init__(self, p):
         self.copyfrom_path = p.copyfrom_path
         self.copyfrom_rev = p.copyfrom_rev
         self.action = p.action
 
 def get_log_child(fp, url, paths, start, end, limit=0,
                   discover_changed_paths=True, strict_node_history=False):
     protocol = -1
     def receiver(orig_paths, revnum, author, date, message, pool):
         paths = {}
         if orig_paths is not None:
             for k, v in orig_paths.iteritems():
                 paths[k] = changedpath(v)
         pickle.dump((paths, revnum, author, date, message),
                     fp, protocol)
 
     try:
         # Use an ra of our own so that our parent can consume
         # our results without confusing the server.
         t = transport.SvnRaTransport(url=url)
         svn.ra.get_log(t.ra, paths, start, end, limit,
                        discover_changed_paths,
                        strict_node_history,
                        receiver)
     except IOError:
         # Caller may interrupt the iteration
         pickle.dump(None, fp, protocol)
     except Exception as inst:
         pickle.dump(str(inst), fp, protocol)
     else:
         pickle.dump(None, fp, protocol)
     fp.close()
     # With large history, cleanup process goes crazy and suddenly
     # consumes *huge* amount of memory. The output file being closed,
     # there is no need for clean termination.
     os._exit(0)
 
 def debugsvnlog(ui, **opts):
     """Fetch SVN log in a subprocess and channel them back to parent to
     avoid memory collection issues.
     """
     if svn is None:
         raise error.Abort(_('debugsvnlog could not load Subversion python '
                            'bindings'))
 
     args = decodeargs(ui.fin.read())
     get_log_child(ui.fout, *args)
 
 class logstream(object):
     """Interruptible revision log iterator."""
     def __init__(self, stdout):
         self._stdout = stdout
 
     def __iter__(self):
         while True:
             try:
                 entry = pickle.load(self._stdout)
             except EOFError:
                 raise error.Abort(_('Mercurial failed to run itself, check'
                                    ' hg executable is in PATH'))
             try:
                 orig_paths, revnum, author, date, message = entry
             except (TypeError, ValueError):
                 if entry is None:
                     break
                 raise error.Abort(_("log stream exception '%s'") % entry)
             yield entry
 
     def close(self):
         if self._stdout:
             self._stdout.close()
             self._stdout = None
 
 class directlogstream(list):
     """Direct revision log iterator.
     This can be used for debugging and development but it will probably leak
     memory and is not suitable for real conversions."""
     def __init__(self, url, paths, start, end, limit=0,
                   discover_changed_paths=True, strict_node_history=False):
 
         def receiver(orig_paths, revnum, author, date, message, pool):
             paths = {}
             if orig_paths is not None:
                 for k, v in orig_paths.iteritems():
                     paths[k] = changedpath(v)
             self.append((paths, revnum, author, date, message))
 
         # Use an ra of our own so that our parent can consume
         # our results without confusing the server.
         t = transport.SvnRaTransport(url=url)
         svn.ra.get_log(t.ra, paths, start, end, limit,
                        discover_changed_paths,
                        strict_node_history,
                        receiver)
 
     def close(self):
         pass
 
 # Check to see if the given path is a local Subversion repo. Verify this by
 # looking for several svn-specific files and directories in the given
 # directory.
 def filecheck(ui, path, proto):
     for x in ('locks', 'hooks', 'format', 'db'):
         if not os.path.exists(os.path.join(path, x)):
             return False
     return True
 
 # Check to see if a given path is the root of an svn repo over http. We verify
 # this by requesting a version-controlled URL we know can't exist and looking
 # for the svn-specific "not found" XML.
 def httpcheck(ui, path, proto):
     try:
         opener = urlreq.buildopener()
         rsp = opener.open('%s://%s/!svn/ver/0/.svn' % (proto, path), 'rb')
         data = rsp.read()
     except urlerr.httperror as inst:
         if inst.code != 404:
             # Except for 404 we cannot know for sure this is not an svn repo
             ui.warn(_('svn: cannot probe remote repository, assume it could '
                       'be a subversion repository. Use --source-type if you '
                       'know better.\n'))
             return True
         data = inst.fp.read()
     except Exception:
         # Could be urlerr.urlerror if the URL is invalid or anything else.
         return False
     return '<m:human-readable errcode="160013">' in data
 
 protomap = {'http': httpcheck,
             'https': httpcheck,
             'file': filecheck,
             }
 def issvnurl(ui, url):
     try:
         proto, path = url.split('://', 1)
         if proto == 'file':
             if (pycompat.iswindows and path[:1] == '/'
                   and path[1:2].isalpha() and path[2:6].lower() == '%3a/'):
                 path = path[:2] + ':/' + path[6:]
             path = urlreq.url2pathname(path)
     except ValueError:
         proto = 'file'
         path = os.path.abspath(url)
     if proto == 'file':
         path = util.pconvert(path)
     check = protomap.get(proto, lambda *args: False)
     while '/' in path:
         if check(ui, path, proto):
             return True
         path = path.rsplit('/', 1)[0]
     return False
 
 # SVN conversion code stolen from bzr-svn and tailor
 #
 # Subversion looks like a versioned filesystem, branches structures
 # are defined by conventions and not enforced by the tool. First,
 # we define the potential branches (modules) as "trunk" and "branches"
 # children directories. Revisions are then identified by their
 # module and revision number (and a repository identifier).
 #
 # The revision graph is really a tree (or a forest). By default, a
 # revision parent is the previous revision in the same module. If the
 # module directory is copied/moved from another module then the
 # revision is the module root and its parent the source revision in
 # the parent module. A revision has at most one parent.
 #
 class svn_source(converter_source):
     def __init__(self, ui, repotype, url, revs=None):
         super(svn_source, self).__init__(ui, repotype, url, revs=revs)
 
         if not (url.startswith('svn://') or url.startswith('svn+ssh://') or
                 (os.path.exists(url) and
                  os.path.exists(os.path.join(url, '.svn'))) or
                 issvnurl(ui, url)):
             raise NoRepo(_("%s does not look like a Subversion repository")
                          % url)
         if svn is None:
             raise MissingTool(_('could not load Subversion python bindings'))
 
         try:
             version = svn.core.SVN_VER_MAJOR, svn.core.SVN_VER_MINOR
             if version < (1, 4):
                 raise MissingTool(_('Subversion python bindings %d.%d found, '
                                     '1.4 or later required') % version)
         except AttributeError:
             raise MissingTool(_('Subversion python bindings are too old, 1.4 '
                                 'or later required'))
 
         self.lastrevs = {}
 
         latest = None
         try:
             # Support file://path@rev syntax. Useful e.g. to convert
             # deleted branches.
             at = url.rfind('@')
             if at >= 0:
                 latest = int(url[at + 1:])
                 url = url[:at]
         except ValueError:
             pass
         self.url = geturl(url)
         self.encoding = 'UTF-8' # Subversion is always nominal UTF-8
         try:
             self.transport = transport.SvnRaTransport(url=self.url)
             self.ra = self.transport.ra
             self.ctx = self.transport.client
             self.baseurl = svn.ra.get_repos_root(self.ra)
             # Module is either empty or a repository path starting with
             # a slash and not ending with a slash.
             self.module = urlreq.unquote(self.url[len(self.baseurl):])
             self.prevmodule = None
             self.rootmodule = self.module
             self.commits = {}
             self.paths = {}
             self.uuid = svn.ra.get_uuid(self.ra)
         except svn.core.SubversionException:
             ui.traceback()
             svnversion = '%d.%d.%d' % (svn.core.SVN_VER_MAJOR,
                                        svn.core.SVN_VER_MINOR,
                                        svn.core.SVN_VER_MICRO)
             raise NoRepo(_("%s does not look like a Subversion repository "
                            "to libsvn version %s")
                          % (self.url, svnversion))
 
         if revs:
             if len(revs) > 1:
                 raise error.Abort(_('subversion source does not support '
                                    'specifying multiple revisions'))
             try:
                 latest = int(revs[0])
             except ValueError:
                 raise error.Abort(_('svn: revision %s is not an integer') %
                                  revs[0])
 
         trunkcfg = self.ui.config('convert', 'svn.trunk')
         if trunkcfg is None:
             trunkcfg = 'trunk'
         self.trunkname = trunkcfg.strip('/')
         self.startrev = self.ui.config('convert', 'svn.startrev')
         try:
             self.startrev = int(self.startrev)
             if self.startrev < 0:
                 self.startrev = 0
         except ValueError:
             raise error.Abort(_('svn: start revision %s is not an integer')
                              % self.startrev)
 
         try:
             self.head = self.latest(self.module, latest)
         except SvnPathNotFound:
             self.head = None
         if not self.head:
             raise error.Abort(_('no revision found in module %s')
                              % self.module)
         self.last_changed = self.revnum(self.head)
 
         self._changescache = (None, None)
 
         if os.path.exists(os.path.join(url, '.svn/entries')):
             self.wc = url
         else:
             self.wc = None
         self.convertfp = None
 
     def setrevmap(self, revmap):
         lastrevs = {}
         for revid in revmap:
             uuid, module, revnum = revsplit(revid)
             lastrevnum = lastrevs.setdefault(module, revnum)
             if revnum > lastrevnum:
                 lastrevs[module] = revnum
         self.lastrevs = lastrevs
 
     def exists(self, path, optrev):
         try:
             svn.client.ls(self.url.rstrip('/') + '/' + quote(path),
                                  optrev, False, self.ctx)
             return True
         except svn.core.SubversionException:
             return False
 
     def getheads(self):
 
         def isdir(path, revnum):
             kind = self._checkpath(path, revnum)
             return kind == svn.core.svn_node_dir
 
         def getcfgpath(name, rev):
             cfgpath = self.ui.config('convert', 'svn.' + name)
             if cfgpath is not None and cfgpath.strip() == '':
                 return None
             path = (cfgpath or name).strip('/')
             if not self.exists(path, rev):
                 if self.module.endswith(path) and name == 'trunk':
                     # we are converting from inside this directory
                     return None
                 if cfgpath:
                     raise error.Abort(_('expected %s to be at %r, but not found'
                                        ) % (name, path))
                 return None
             self.ui.note(_('found %s at %r\n') % (name, path))
             return path
 
         rev = optrev(self.last_changed)
         oldmodule = ''
         trunk = getcfgpath('trunk', rev)
         self.tags = getcfgpath('tags', rev)
         branches = getcfgpath('branches', rev)
 
         # If the project has a trunk or branches, we will extract heads
         # from them. We keep the project root otherwise.
         if trunk:
             oldmodule = self.module or ''
             self.module += '/' + trunk
             self.head = self.latest(self.module, self.last_changed)
             if not self.head:
                 raise error.Abort(_('no revision found in module %s')
                                  % self.module)
 
         # First head in the list is the module's head
         self.heads = [self.head]
         if self.tags is not None:
             self.tags = '%s/%s' % (oldmodule , (self.tags or 'tags'))
 
         # Check if branches bring a few more heads to the list
         if branches:
             rpath = self.url.strip('/')
             branchnames = svn.client.ls(rpath + '/' + quote(branches),
                                         rev, False, self.ctx)
             for branch in sorted(branchnames):
                 module = '%s/%s/%s' % (oldmodule, branches, branch)
                 if not isdir(module, self.last_changed):
                     continue
                 brevid = self.latest(module, self.last_changed)
                 if not brevid:
                     self.ui.note(_('ignoring empty branch %s\n') % branch)
                     continue
                 self.ui.note(_('found branch %s at %d\n') %
                              (branch, self.revnum(brevid)))
                 self.heads.append(brevid)
 
         if self.startrev and self.heads:
             if len(self.heads) > 1:
                 raise error.Abort(_('svn: start revision is not supported '
                                    'with more than one branch'))
             revnum = self.revnum(self.heads[0])
             if revnum < self.startrev:
                 raise error.Abort(
                     _('svn: no revision found after start revision %d')
                                  % self.startrev)
 
         return self.heads
 
     def _getchanges(self, rev, full):
         (paths, parents) = self.paths[rev]
         copies = {}
         if parents:
             files, self.removed, copies = self.expandpaths(rev, paths, parents)
         if full or not parents:
             # Perform a full checkout on roots
             uuid, module, revnum = revsplit(rev)
             entries = svn.client.ls(self.baseurl + quote(module),
                                     optrev(revnum), True, self.ctx)
             files = [n for n, e in entries.iteritems()
                      if e.kind == svn.core.svn_node_file]
             self.removed = set()
 
         files.sort()
         files = zip(files, [rev] * len(files))
         return (files, copies)
 
     def getchanges(self, rev, full):
         # reuse cache from getchangedfiles
         if self._changescache[0] == rev and not full:
             (files, copies) = self._changescache[1]
         else:
             (files, copies) = self._getchanges(rev, full)
             # caller caches the result, so free it here to release memory
             del self.paths[rev]
         return (files, copies, set())
 
     def getchangedfiles(self, rev, i):
         # called from filemap - cache computed values for reuse in getchanges
         (files, copies) = self._getchanges(rev, False)
         self._changescache = (rev, (files, copies))
         return [f[0] for f in files]
 
     def getcommit(self, rev):
         if rev not in self.commits:
             uuid, module, revnum = revsplit(rev)
             self.module = module
             self.reparent(module)
             # We assume that:
             # - requests for revisions after "stop" come from the
             # revision graph backward traversal. Cache all of them
             # down to stop, they will be used eventually.
             # - requests for revisions before "stop" come to get
             # isolated branches parents. Just fetch what is needed.
             stop = self.lastrevs.get(module, 0)
             if revnum < stop:
                 stop = revnum + 1
             self._fetch_revisions(revnum, stop)
             if rev not in self.commits:
                 raise error.Abort(_('svn: revision %s not found') % revnum)
         revcommit = self.commits[rev]
         # caller caches the result, so free it here to release memory
         del self.commits[rev]
         return revcommit
 
     def checkrevformat(self, revstr, mapname='splicemap'):
         """ fails if revision format does not match the correct format"""
         if not re.match(r'svn:[0-9a-f]{8,8}-[0-9a-f]{4,4}-'
                               r'[0-9a-f]{4,4}-[0-9a-f]{4,4}-[0-9a-f]'
                               r'{12,12}(.*)\@[0-9]+$',revstr):
             raise error.Abort(_('%s entry %s is not a valid revision'
                                ' identifier') % (mapname, revstr))
 
     def numcommits(self):
         return int(self.head.rsplit('@', 1)[1]) - self.startrev
 
     def gettags(self):
         tags = {}
         if self.tags is None:
             return tags
 
         # svn tags are just a convention, project branches left in a
         # 'tags' directory. There is no other relationship than
         # ancestry, which is expensive to discover and makes them hard
         # to update incrementally.  Worse, past revisions may be
         # referenced by tags far away in the future, requiring a deep
         # history traversal on every calculation.  Current code
         # performs a single backward traversal, tracking moves within
         # the tags directory (tag renaming) and recording a new tag
         # everytime a project is copied from outside the tags
         # directory. It also lists deleted tags, this behaviour may
         # change in the future.
         pendings = []
         tagspath = self.tags
         start = svn.ra.get_latest_revnum(self.ra)
         stream = self._getlog([self.tags], start, self.startrev)
         try:
             for entry in stream:
                 origpaths, revnum, author, date, message = entry
                 if not origpaths:
                     origpaths = []
                 copies = [(e.copyfrom_path, e.copyfrom_rev, p) for p, e
                           in origpaths.iteritems() if e.copyfrom_path]
                 # Apply moves/copies from more specific to general
                 copies.sort(reverse=True)
 
                 srctagspath = tagspath
                 if copies and copies[-1][2] == tagspath:
                     # Track tags directory moves
                     srctagspath = copies.pop()[0]
 
                 for source, sourcerev, dest in copies:
                     if not dest.startswith(tagspath + '/'):
                         continue
                     for tag in pendings:
                         if tag[0].startswith(dest):
                             tagpath = source + tag[0][len(dest):]
                             tag[:2] = [tagpath, sourcerev]
                             break
                     else:
                         pendings.append([source, sourcerev, dest])
 
                 # Filter out tags with children coming from different
                 # parts of the repository like:
                 # /tags/tag.1 (from /trunk:10)
                 # /tags/tag.1/foo (from /branches/foo:12)
                 # Here/tags/tag.1 discarded as well as its children.
                 # It happens with tools like cvs2svn. Such tags cannot
                 # be represented in mercurial.
                 addeds = dict((p, e.copyfrom_path) for p, e
                               in origpaths.iteritems()
                               if e.action == 'A' and e.copyfrom_path)
                 badroots = set()
                 for destroot in addeds:
                     for source, sourcerev, dest in pendings:
                         if (not dest.startswith(destroot + '/')
                             or source.startswith(addeds[destroot] + '/')):
                             continue
                         badroots.add(destroot)
                         break
 
                 for badroot in badroots:
                     pendings = [p for p in pendings if p[2] != badroot
                                 and not p[2].startswith(badroot + '/')]
 
                 # Tell tag renamings from tag creations
                 renamings = []
                 for source, sourcerev, dest in pendings:
                     tagname = dest.split('/')[-1]
                     if source.startswith(srctagspath):
                         renamings.append([source, sourcerev, tagname])
                         continue
                     if tagname in tags:
                         # Keep the latest tag value
                         continue
                     # From revision may be fake, get one with changes
                     try:
                         tagid = self.latest(source, sourcerev)
                         if tagid and tagname not in tags:
                             tags[tagname] = tagid
                     except SvnPathNotFound:
                         # It happens when we are following directories
                         # we assumed were copied with their parents
                         # but were really created in the tag
                         # directory.
                         pass
                 pendings = renamings
                 tagspath = srctagspath
         finally:
             stream.close()
         return tags
 
     def converted(self, rev, destrev):
         if not self.wc:
             return
         if self.convertfp is None:
             self.convertfp = open(os.path.join(self.wc, '.svn', 'hg-shamap'),
                                   'ab')
         self.convertfp.write(util.tonativeeol('%s %d\n'
                                               % (destrev, self.revnum(rev))))
         self.convertfp.flush()
 
     def revid(self, revnum, module=None):
         return 'svn:%s%s@%s' % (self.uuid, module or self.module, revnum)
 
     def revnum(self, rev):
         return int(rev.split('@')[-1])
 
     def latest(self, path, stop=None):
         """Find the latest revid affecting path, up to stop revision
         number. If stop is None, default to repository latest
         revision. It may return a revision in a different module,
         since a branch may be moved without a change being
         reported. Return None if computed module does not belong to
         rootmodule subtree.
         """
         def findchanges(path, start, stop=None):
             stream = self._getlog([path], start, stop or 1)
             try:
                 for entry in stream:
                     paths, revnum, author, date, message = entry
                     if stop is None and paths:
                         # We do not know the latest changed revision,
                         # keep the first one with changed paths.
                         break
                     if revnum <= stop:
                         break
 
                     for p in paths:
                         if (not path.startswith(p) or
                             not paths[p].copyfrom_path):
                             continue
                         newpath = paths[p].copyfrom_path + path[len(p):]
                         self.ui.debug("branch renamed from %s to %s at %d\n" %
                                       (path, newpath, revnum))
                         path = newpath
                         break
                 if not paths:
                     revnum = None
                 return revnum, path
             finally:
                 stream.close()
 
         if not path.startswith(self.rootmodule):
             # Requests on foreign branches may be forbidden at server level
             self.ui.debug('ignoring foreign branch %r\n' % path)
             return None
 
         if stop is None:
             stop = svn.ra.get_latest_revnum(self.ra)
         try:
             prevmodule = self.reparent('')
             dirent = svn.ra.stat(self.ra, path.strip('/'), stop)
             self.reparent(prevmodule)
         except svn.core.SubversionException:
             dirent = None
         if not dirent:
             raise SvnPathNotFound(_('%s not found up to revision %d')
                                   % (path, stop))
 
         # stat() gives us the previous revision on this line of
         # development, but it might be in *another module*. Fetch the
         # log and detect renames down to the latest revision.
         revnum, realpath = findchanges(path, stop, dirent.created_rev)
         if revnum is None:
             # Tools like svnsync can create empty revision, when
             # synchronizing only a subtree for instance. These empty
             # revisions created_rev still have their original values
             # despite all changes having disappeared and can be
             # returned by ra.stat(), at least when stating the root
             # module. In that case, do not trust created_rev and scan
             # the whole history.
             revnum, realpath = findchanges(path, stop)
             if revnum is None:
                 self.ui.debug('ignoring empty branch %r\n' % realpath)
                 return None
 
         if not realpath.startswith(self.rootmodule):
             self.ui.debug('ignoring foreign branch %r\n' % realpath)
             return None
         return self.revid(revnum, realpath)
 
     def reparent(self, module):
         """Reparent the svn transport and return the previous parent."""
         if self.prevmodule == module:
             return module
         svnurl = self.baseurl + quote(module)
         prevmodule = self.prevmodule
         if prevmodule is None:
             prevmodule = ''
         self.ui.debug("reparent to %s\n" % svnurl)
         svn.ra.reparent(self.ra, svnurl)
         self.prevmodule = module
         return prevmodule
 
     def expandpaths(self, rev, paths, parents):
         changed, removed = set(), set()
         copies = {}
 
         new_module, revnum = revsplit(rev)[1:]
         if new_module != self.module:
             self.module = new_module
             self.reparent(self.module)
 
         for i, (path, ent) in enumerate(paths):
             self.ui.progress(_('scanning paths'), i, item=path,
                              total=len(paths), unit=_('paths'))
             entrypath = self.getrelpath(path)
 
             kind = self._checkpath(entrypath, revnum)
             if kind == svn.core.svn_node_file:
                 changed.add(self.recode(entrypath))
                 if not ent.copyfrom_path or not parents:
                     continue
                 # Copy sources not in parent revisions cannot be
                 # represented, ignore their origin for now
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 if ent.copyfrom_rev < prevnum:
                     continue
                 copyfrom_path = self.getrelpath(ent.copyfrom_path, pmodule)
                 if not copyfrom_path:
                     continue
                 self.ui.debug("copied to %s from %s@%s\n" %
                               (entrypath, copyfrom_path, ent.copyfrom_rev))
                 copies[self.recode(entrypath)] = self.recode(copyfrom_path)
             elif kind == 0: # gone, but had better be a deleted *file*
                 self.ui.debug("gone from %s\n" % ent.copyfrom_rev)
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 parentpath = pmodule + "/" + entrypath
                 fromkind = self._checkpath(entrypath, prevnum, pmodule)
 
                 if fromkind == svn.core.svn_node_file:
                     removed.add(self.recode(entrypath))
                 elif fromkind == svn.core.svn_node_dir:
                     oroot = parentpath.strip('/')
                     nroot = path.strip('/')
                     children = self._iterfiles(oroot, prevnum)
                     for childpath in children:
                         childpath = childpath.replace(oroot, nroot)
                         childpath = self.getrelpath("/" + childpath, pmodule)
                         if childpath:
                             removed.add(self.recode(childpath))
                 else:
                     self.ui.debug('unknown path in revision %d: %s\n' % \
                                   (revnum, path))
             elif kind == svn.core.svn_node_dir:
                 if ent.action == 'M':
                     # If the directory just had a prop change,
                     # then we shouldn't need to look for its children.
                     continue
                 if ent.action == 'R' and parents:
                     # If a directory is replacing a file, mark the previous
                     # file as deleted
                     pmodule, prevnum = revsplit(parents[0])[1:]
                     pkind = self._checkpath(entrypath, prevnum, pmodule)
                     if pkind == svn.core.svn_node_file:
                         removed.add(self.recode(entrypath))
                     elif pkind == svn.core.svn_node_dir:
                         # We do not know what files were kept or removed,
                         # mark them all as changed.
                         for childpath in self._iterfiles(pmodule, prevnum):
                             childpath = self.getrelpath("/" + childpath)
                             if childpath:
                                 changed.add(self.recode(childpath))
 
                 for childpath in self._iterfiles(path, revnum):
                     childpath = self.getrelpath("/" + childpath)
                     if childpath:
                         changed.add(self.recode(childpath))
 
                 # Handle directory copies
                 if not ent.copyfrom_path or not parents:
                     continue
                 # Copy sources not in parent revisions cannot be
                 # represented, ignore their origin for now
                 pmodule, prevnum = revsplit(parents[0])[1:]
                 if ent.copyfrom_rev < prevnum:
                     continue
                 copyfrompath = self.getrelpath(ent.copyfrom_path, pmodule)
                 if not copyfrompath:
                     continue
                 self.ui.debug("mark %s came from %s:%d\n"
                               % (path, copyfrompath, ent.copyfrom_rev))
                 children = self._iterfiles(ent.copyfrom_path, ent.copyfrom_rev)
                 for childpath in children:
                     childpath = self.getrelpath("/" + childpath, pmodule)
                     if not childpath:
                         continue
                     copytopath = path + childpath[len(copyfrompath):]
                     copytopath = self.getrelpath(copytopath)
                     copies[self.recode(copytopath)] = self.recode(childpath)
 
         self.ui.progress(_('scanning paths'), None)
         changed.update(removed)
         return (list(changed), removed, copies)
 
     def _fetch_revisions(self, from_revnum, to_revnum):
         if from_revnum < to_revnum:
             from_revnum, to_revnum = to_revnum, from_revnum
 
         self.child_cset = None
 
         def parselogentry(orig_paths, revnum, author, date, message):
             """Return the parsed commit object or None, and True if
             the revision is a branch root.
             """
             self.ui.debug("parsing revision %d (%d changes)\n" %
                           (revnum, len(orig_paths)))
 
             branched = False
             rev = self.revid(revnum)
             # branch log might return entries for a parent we already have
 
             if rev in self.commits or revnum < to_revnum:
                 return None, branched
 
             parents = []
             # check whether this revision is the start of a branch or part
             # of a branch renaming
             orig_paths = sorted(orig_paths.iteritems())
             root_paths = [(p, e) for p, e in orig_paths
                           if self.module.startswith(p)]
             if root_paths:
                 path, ent = root_paths[-1]
                 if ent.copyfrom_path:
                     branched = True
                     newpath = ent.copyfrom_path + self.module[len(path):]
                     # ent.copyfrom_rev may not be the actual last revision
                     previd = self.latest(newpath, ent.copyfrom_rev)
                     if previd is not None:
                         prevmodule, prevnum = revsplit(previd)[1:]
                         if prevnum >= self.startrev:
                             parents = [previd]
                             self.ui.note(
                                 _('found parent of branch %s at %d: %s\n') %
                                 (self.module, prevnum, prevmodule))
                 else:
                     self.ui.debug("no copyfrom path, don't know what to do.\n")
 
             paths = []
             # filter out unrelated paths
             for path, ent in orig_paths:
                 if self.getrelpath(path) is None:
                     continue
                 paths.append((path, ent))
 
             # Example SVN datetime. Includes microseconds.
             # ISO-8601 conformant
             # '2007-01-04T17:35:00.902377Z'
             date = util.parsedate(date[:19] + " UTC", ["%Y-%m-%dT%H:%M:%S"])
             if self.ui.configbool('convert', 'localtimezone'):
                 date = makedatetimestamp(date[0])
 
             if message:
                 log = self.recode(message)
             else:
                 log = ''
 
             if author:
                 author = self.recode(author)
             else:
                 author = ''
 
             try:
                 branch = self.module.split("/")[-1]
                 if branch == self.trunkname:
                     branch = None
             except IndexError:
                 branch = None
 
             cset = commit(author=author,
                           date=util.datestr(date, '%Y-%m-%d %H:%M:%S %1%2'),
                           desc=log,
                           parents=parents,
                           branch=branch,
                           rev=rev)
 
             self.commits[rev] = cset
             # The parents list is *shared* among self.paths and the
             # commit object. Both will be updated below.
             self.paths[rev] = (paths, cset.parents)
             if self.child_cset and not self.child_cset.parents:
                 self.child_cset.parents[:] = [rev]
             self.child_cset = cset
             return cset, branched
 
         self.ui.note(_('fetching revision log for "%s" from %d to %d\n') %
                      (self.module, from_revnum, to_revnum))
 
         try:
             firstcset = None
             lastonbranch = False
             stream = self._getlog([self.module], from_revnum, to_revnum)
             try:
                 for entry in stream:
                     paths, revnum, author, date, message = entry
                     if revnum < self.startrev:
                         lastonbranch = True
                         break
                     if not paths:
                         self.ui.debug('revision %d has no entries\n' % revnum)
                         # If we ever leave the loop on an empty
                         # revision, do not try to get a parent branch
                         lastonbranch = lastonbranch or revnum == 0
                         continue
                     cset, lastonbranch = parselogentry(paths, revnum, author,
                                                        date, message)
                     if cset:
                         firstcset = cset
                     if lastonbranch:
                         break
             finally:
                 stream.close()
 
             if not lastonbranch and firstcset and not firstcset.parents:
                 # The first revision of the sequence (the last fetched one)
                 # has invalid parents if not a branch root. Find the parent
                 # revision now, if any.
                 try:
                     firstrevnum = self.revnum(firstcset.rev)
                     if firstrevnum > 1:
                         latest = self.latest(self.module, firstrevnum - 1)
                         if latest:
                             firstcset.parents.append(latest)
                 except SvnPathNotFound:
                     pass
         except svn.core.SubversionException as xxx_todo_changeme:
             (inst, num) = xxx_todo_changeme.args
             if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION:
                 raise error.Abort(_('svn: branch has no revision %s')
                                  % to_revnum)
             raise
 
     def getfile(self, file, rev):
         # TODO: ra.get_file transmits the whole file instead of diffs.
         if file in self.removed:
             return None, None
         mode = ''
         try:
             new_module, revnum = revsplit(rev)[1:]
             if self.module != new_module:
                 self.module = new_module
                 self.reparent(self.module)
             io = stringio()
             info = svn.ra.get_file(self.ra, file, revnum, io)
             data = io.getvalue()
             # ra.get_file() seems to keep a reference on the input buffer
             # preventing collection. Release it explicitly.
             io.close()
             if isinstance(info, list):
                 info = info[-1]
             mode = ("svn:executable" in info) and 'x' or ''
             mode = ("svn:special" in info) and 'l' or mode
         except svn.core.SubversionException as e:
             notfound = (svn.core.SVN_ERR_FS_NOT_FOUND,
                 svn.core.SVN_ERR_RA_DAV_PATH_NOT_FOUND)
             if e.apr_err in notfound: # File not found
                 return None, None
             raise
         if mode == 'l':
             link_prefix = "link "
             if data.startswith(link_prefix):
                 data = data[len(link_prefix):]
         return data, mode
 
     def _iterfiles(self, path, revnum):
         """Enumerate all files in path at revnum, recursively."""
         path = path.strip('/')
         pool = svn.core.Pool()
         rpath = '/'.join([self.baseurl, quote(path)]).strip('/')
         entries = svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool)
         if path:
             path += '/'
         return ((path + p) for p, e in entries.iteritems()
                 if e.kind == svn.core.svn_node_file)
 
     def getrelpath(self, path, module=None):
         if module is None:
             module = self.module
         # Given the repository url of this wc, say
         #   "http://server/plone/CMFPlone/branches/Plone-2_0-branch"
         # extract the "entry" portion (a relative path) from what
         # svn log --xml says, i.e.
         #   "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py"
         # that is to say "tests/PloneTestCase.py"
         if path.startswith(module):
             relative = path.rstrip('/')[len(module):]
             if relative.startswith('/'):
                 return relative[1:]
             elif relative == '':
                 return relative
 
         # The path is outside our tracked tree...
         self.ui.debug('%r is not under %r, ignoring\n' % (path, module))
         return None
 
     def _checkpath(self, path, revnum, module=None):
         if module is not None:
             prevmodule = self.reparent('')
             path = module + '/' + path
         try:
             # ra.check_path does not like leading slashes very much, it leads
             # to PROPFIND subversion errors
             return svn.ra.check_path(self.ra, path.strip('/'), revnum)
         finally:
             if module is not None:
                 self.reparent(prevmodule)
 
     def _getlog(self, paths, start, end, limit=0, discover_changed_paths=True,
                 strict_node_history=False):
         # Normalize path names, svn >= 1.5 only wants paths relative to
         # supplied URL
         relpaths = []
         for p in paths:
             if not p.startswith('/'):
                 p = self.module + '/' + p
             relpaths.append(p.strip('/'))
         args = [self.baseurl, relpaths, start, end, limit,
                 discover_changed_paths, strict_node_history]
         # developer config: convert.svn.debugsvnlog
         if not self.ui.configbool('convert', 'svn.debugsvnlog'):
             return directlogstream(*args)
         arg = encodeargs(args)
         hgexe = util.hgexecutable()
         cmd = '%s debugsvnlog' % util.shellquote(hgexe)
         stdin, stdout = util.popen2(util.quotecommand(cmd))
         stdin.write(arg)
         try:
             stdin.close()
         except IOError:
             raise error.Abort(_('Mercurial failed to run itself, check'
                                ' hg executable is in PATH'))
         return logstream(stdout)
 
 pre_revprop_change = '''#!/bin/sh
 
 REPOS="$1"
 REV="$2"
 USER="$3"
 PROPNAME="$4"
 ACTION="$5"
 
 if [ "$ACTION" = "M" -a "$PROPNAME" = "svn:log" ]; then exit 0; fi
 if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-branch" ]; then exit 0; fi
 if [ "$ACTION" = "A" -a "$PROPNAME" = "hg:convert-rev" ]; then exit 0; fi
 
 echo "Changing prohibited revision property" >&2
 exit 1
 '''
 
 class svn_sink(converter_sink, commandline):
     commit_re = re.compile(r'Committed revision (\d+).', re.M)
     uuid_re = re.compile(r'Repository UUID:\s*(\S+)', re.M)
 
     def prerun(self):
         if self.wc:
             os.chdir(self.wc)
 
     def postrun(self):
         if self.wc:
             os.chdir(self.cwd)
 
     def join(self, name):
         return os.path.join(self.wc, '.svn', name)
 
     def revmapfile(self):
         return self.join('hg-shamap')
 
     def authorfile(self):
         return self.join('hg-authormap')
 
     def __init__(self, ui, repotype, path):
 
         converter_sink.__init__(self, ui, repotype, path)
         commandline.__init__(self, ui, 'svn')
         self.delete = []
         self.setexec = []
         self.delexec = []
         self.copies = []
         self.wc = None
         self.cwd = pycompat.getcwd()
 
         created = False
         if os.path.isfile(os.path.join(path, '.svn', 'entries')):
             self.wc = os.path.realpath(path)
             self.run0('update')
         else:
-            if not re.search(r'^(file|http|https|svn|svn\+ssh)\://', path):
+            if not re.search(br'^(file|http|https|svn|svn\+ssh)\://', path):
                 path = os.path.realpath(path)
                 if os.path.isdir(os.path.dirname(path)):
                     if not os.path.exists(os.path.join(path, 'db', 'fs-type')):
                         ui.status(_('initializing svn repository %r\n') %
                                   os.path.basename(path))
                         commandline(ui, 'svnadmin').run0('create', path)
                         created = path
                     path = util.normpath(path)
                     if not path.startswith('/'):
                         path = '/' + path
                     path = 'file://' + path
 
             wcpath = os.path.join(pycompat.getcwd(), os.path.basename(path) +
                                 '-wc')
             ui.status(_('initializing svn working copy %r\n')
                       % os.path.basename(wcpath))
             self.run0('checkout', path, wcpath)
 
             self.wc = wcpath
         self.opener = vfsmod.vfs(self.wc)
         self.wopener = vfsmod.vfs(self.wc)
         self.childmap = mapfile(ui, self.join('hg-childmap'))
         if util.checkexec(self.wc):
             self.is_exec = util.isexec
         else:
             self.is_exec = None
 
         if created:
             hook = os.path.join(created, 'hooks', 'pre-revprop-change')
             fp = open(hook, 'wb')
             fp.write(pre_revprop_change)
             fp.close()
             util.setflags(hook, False, True)
 
         output = self.run0('info')
         self.uuid = self.uuid_re.search(output).group(1).strip()
 
     def wjoin(self, *names):
         return os.path.join(self.wc, *names)
 
     @propertycache
     def manifest(self):
         # As of svn 1.7, the "add" command fails when receiving
         # already tracked entries, so we have to track and filter them
         # ourselves.
         m = set()
         output = self.run0('ls', recursive=True, xml=True)
         doc = xml.dom.minidom.parseString(output)
         for e in doc.getElementsByTagName('entry'):
             for n in e.childNodes:
                 if n.nodeType != n.ELEMENT_NODE or n.tagName != 'name':
                     continue
                 name = ''.join(c.data for c in n.childNodes
                                if c.nodeType == c.TEXT_NODE)
                 # Entries are compared with names coming from
                 # mercurial, so bytes with undefined encoding. Our
                 # best bet is to assume they are in local
                 # encoding. They will be passed to command line calls
                 # later anyway, so they better be.
                 m.add(encoding.unitolocal(name))
                 break
         return m
 
     def putfile(self, filename, flags, data):
         if 'l' in flags:
             self.wopener.symlink(data, filename)
         else:
             try:
                 if os.path.islink(self.wjoin(filename)):
                     os.unlink(filename)
             except OSError:
                 pass
             self.wopener.write(filename, data)
 
             if self.is_exec:
                 if self.is_exec(self.wjoin(filename)):
                     if 'x' not in flags:
                         self.delexec.append(filename)
                 else:
                     if 'x' in flags:
                         self.setexec.append(filename)
                 util.setflags(self.wjoin(filename), False, 'x' in flags)
 
     def _copyfile(self, source, dest):
         # SVN's copy command pukes if the destination file exists, but
         # our copyfile method expects to record a copy that has
         # already occurred.  Cross the semantic gap.
         wdest = self.wjoin(dest)
         exists = os.path.lexists(wdest)
         if exists:
             fd, tempname = tempfile.mkstemp(
                 prefix='hg-copy-', dir=os.path.dirname(wdest))
             os.close(fd)
             os.unlink(tempname)
             os.rename(wdest, tempname)
         try:
             self.run0('copy', source, dest)
         finally:
             self.manifest.add(dest)
             if exists:
                 try:
                     os.unlink(wdest)
                 except OSError:
                     pass
                 os.rename(tempname, wdest)
 
     def dirs_of(self, files):
         dirs = set()
         for f in files:
             if os.path.isdir(self.wjoin(f)):
                 dirs.add(f)
             i = len(f)
             for i in iter(lambda: f.rfind('/', 0, i), -1):
                 dirs.add(f[:i])
         return dirs
 
     def add_dirs(self, files):
         add_dirs = [d for d in sorted(self.dirs_of(files))
                     if d not in self.manifest]
         if add_dirs:
             self.manifest.update(add_dirs)
             self.xargs(add_dirs, 'add', non_recursive=True, quiet=True)
         return add_dirs
 
     def add_files(self, files):
         files = [f for f in files if f not in self.manifest]
         if files:
             self.manifest.update(files)
             self.xargs(files, 'add', quiet=True)
         return files
 
     def addchild(self, parent, child):
         self.childmap[parent] = child
 
     def revid(self, rev):
         return u"svn:%s@%s" % (self.uuid, rev)
 
     def putcommit(self, files, copies, parents, commit, source, revmap, full,
                   cleanp2):
         for parent in parents:
             try:
                 return self.revid(self.childmap[parent])
             except KeyError:
                 pass
 
         # Apply changes to working copy
         for f, v in files:
             data, mode = source.getfile(f, v)
             if data is None:
                 self.delete.append(f)
             else:
                 self.putfile(f, mode, data)
                 if f in copies:
                     self.copies.append([copies[f], f])
         if full:
             self.delete.extend(sorted(self.manifest.difference(files)))
         files = [f[0] for f in files]
 
         entries = set(self.delete)
         files = frozenset(files)
         entries.update(self.add_dirs(files.difference(entries)))
         if self.copies:
             for s, d in self.copies:
                 self._copyfile(s, d)
             self.copies = []
         if self.delete:
             self.xargs(self.delete, 'delete')
             for f in self.delete:
                 self.manifest.remove(f)
             self.delete = []
         entries.update(self.add_files(files.difference(entries)))
         if self.delexec:
             self.xargs(self.delexec, 'propdel', 'svn:executable')
             self.delexec = []
         if self.setexec:
             self.xargs(self.setexec, 'propset', 'svn:executable', '*')
             self.setexec = []
 
         fd, messagefile = tempfile.mkstemp(prefix='hg-convert-')
         fp = os.fdopen(fd, pycompat.sysstr('wb'))
         fp.write(util.tonativeeol(commit.desc))
         fp.close()
         try:
             output = self.run0('commit',
                                username=util.shortuser(commit.author),
                                file=messagefile,
                                encoding='utf-8')
             try:
                 rev = self.commit_re.search(output).group(1)
             except AttributeError:
                 if parents and not files:
                     return parents[0]
                 self.ui.warn(_('unexpected svn output:\n'))
                 self.ui.warn(output)
                 raise error.Abort(_('unable to cope with svn output'))
             if commit.rev:
                 self.run('propset', 'hg:convert-rev', commit.rev,
                          revprop=True, revision=rev)
             if commit.branch and commit.branch != 'default':
                 self.run('propset', 'hg:convert-branch', commit.branch,
                          revprop=True, revision=rev)
             for parent in parents:
                 self.addchild(parent, rev)
             return self.revid(rev)
         finally:
             os.unlink(messagefile)
 
     def puttags(self, tags):
         self.ui.warn(_('writing Subversion tags is not yet implemented\n'))
         return None, None
 
     def hascommitfrommap(self, rev):
         # We trust that revisions referenced in a map still is present
         # TODO: implement something better if necessary and feasible
         return True
 
     def hascommitforsplicemap(self, rev):
         # This is not correct as one can convert to an existing subversion
         # repository and childmap would not list all revisions. Too bad.
         if rev in self.childmap:
             return True
         raise error.Abort(_('splice map revision %s not found in subversion '
                            'child map (revision lookups are not implemented)')
                          % rev)

hgext/lfs/blobstore.py

@@ -1,481 +1,481 b''
 # blobstore.py - local and remote (speaking Git-LFS protocol) blob storages
 #
 # Copyright 2017 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import hashlib
 import json
 import os
 import re
 import socket
 
 from mercurial.i18n import _
 
 from mercurial import (
     error,
     pathutil,
     url as urlmod,
     util,
     vfs as vfsmod,
     worker,
 )
 
 from ..largefiles import lfutil
 
 # 64 bytes for SHA256
-_lfsre = re.compile(r'\A[a-f0-9]{64}\Z')
+_lfsre = re.compile(br'\A[a-f0-9]{64}\Z')
 
 class lfsvfs(vfsmod.vfs):
     def join(self, path):
         """split the path at first two characters, like: XX/XXXXX..."""
         if not _lfsre.match(path):
             raise error.ProgrammingError('unexpected lfs path: %s' % path)
         return super(lfsvfs, self).join(path[0:2], path[2:])
 
     def walk(self, path=None, onerror=None):
         """Yield (dirpath, [], oids) tuple for blobs under path
 
         Oids only exist in the root of this vfs, so dirpath is always ''.
         """
         root = os.path.normpath(self.base)
         # when dirpath == root, dirpath[prefixlen:] becomes empty
         # because len(dirpath) < prefixlen.
         prefixlen = len(pathutil.normasprefix(root))
         oids = []
 
         for dirpath, dirs, files in os.walk(self.reljoin(self.base, path or ''),
                                             onerror=onerror):
             dirpath = dirpath[prefixlen:]
 
             # Silently skip unexpected files and directories
             if len(dirpath) == 2:
                 oids.extend([dirpath + f for f in files
                              if _lfsre.match(dirpath + f)])
 
         yield ('', [], oids)
 
 class filewithprogress(object):
     """a file-like object that supports __len__ and read.
 
     Useful to provide progress information for how many bytes are read.
     """
 
     def __init__(self, fp, callback):
         self._fp = fp
         self._callback = callback # func(readsize)
         fp.seek(0, os.SEEK_END)
         self._len = fp.tell()
         fp.seek(0)
 
     def __len__(self):
         return self._len
 
     def read(self, size):
         if self._fp is None:
             return b''
         data = self._fp.read(size)
         if data:
             if self._callback:
                 self._callback(len(data))
         else:
             self._fp.close()
             self._fp = None
         return data
 
 class local(object):
     """Local blobstore for large file contents.
 
     This blobstore is used both as a cache and as a staging area for large blobs
     to be uploaded to the remote blobstore.
     """
 
     def __init__(self, repo):
         fullpath = repo.svfs.join('lfs/objects')
         self.vfs = lfsvfs(fullpath)
         usercache = lfutil._usercachedir(repo.ui, 'lfs')
         self.cachevfs = lfsvfs(usercache)
         self.ui = repo.ui
 
     def open(self, oid):
         """Open a read-only file descriptor to the named blob, in either the
         usercache or the local store."""
         # The usercache is the most likely place to hold the file.  Commit will
         # write to both it and the local store, as will anything that downloads
         # the blobs.  However, things like clone without an update won't
         # populate the local store.  For an init + push of a local clone,
         # the usercache is the only place it _could_ be.  If not present, the
         # missing file msg here will indicate the local repo, not the usercache.
         if self.cachevfs.exists(oid):
             return self.cachevfs(oid, 'rb')
 
         return self.vfs(oid, 'rb')
 
     def download(self, oid, src):
         """Read the blob from the remote source in chunks, verify the content,
         and write to this local blobstore."""
         sha256 = hashlib.sha256()
 
         with self.vfs(oid, 'wb', atomictemp=True) as fp:
             for chunk in util.filechunkiter(src, size=1048576):
                 fp.write(chunk)
                 sha256.update(chunk)
 
             realoid = sha256.hexdigest()
             if realoid != oid:
                 raise error.Abort(_('corrupt remote lfs object: %s') % oid)
 
         # XXX: should we verify the content of the cache, and hardlink back to
         # the local store on success, but truncate, write and link on failure?
         if not self.cachevfs.exists(oid):
             self.ui.note(_('lfs: adding %s to the usercache\n') % oid)
             lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid))
 
     def write(self, oid, data):
         """Write blob to local blobstore.
 
         This should only be called from the filelog during a commit or similar.
         As such, there is no need to verify the data.  Imports from a remote
         store must use ``download()`` instead."""
         with self.vfs(oid, 'wb', atomictemp=True) as fp:
             fp.write(data)
 
         # XXX: should we verify the content of the cache, and hardlink back to
         # the local store on success, but truncate, write and link on failure?
         if not self.cachevfs.exists(oid):
             self.ui.note(_('lfs: adding %s to the usercache\n') % oid)
             lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid))
 
     def read(self, oid, verify=True):
         """Read blob from local blobstore."""
         if not self.vfs.exists(oid):
             blob = self._read(self.cachevfs, oid, verify)
 
             # Even if revlog will verify the content, it needs to be verified
             # now before making the hardlink to avoid propagating corrupt blobs.
             # Don't abort if corruption is detected, because `hg verify` will
             # give more useful info about the corruption- simply don't add the
             # hardlink.
             if verify or hashlib.sha256(blob).hexdigest() == oid:
                 self.ui.note(_('lfs: found %s in the usercache\n') % oid)
                 lfutil.link(self.cachevfs.join(oid), self.vfs.join(oid))
         else:
             self.ui.note(_('lfs: found %s in the local lfs store\n') % oid)
             blob = self._read(self.vfs, oid, verify)
         return blob
 
     def _read(self, vfs, oid, verify):
         """Read blob (after verifying) from the given store"""
         blob = vfs.read(oid)
         if verify:
             _verify(oid, blob)
         return blob
 
     def has(self, oid):
         """Returns True if the local blobstore contains the requested blob,
         False otherwise."""
         return self.cachevfs.exists(oid) or self.vfs.exists(oid)
 
 class _gitlfsremote(object):
 
     def __init__(self, repo, url):
         ui = repo.ui
         self.ui = ui
         baseurl, authinfo = url.authinfo()
         self.baseurl = baseurl.rstrip('/')
         useragent = repo.ui.config('experimental', 'lfs.user-agent')
         if not useragent:
             useragent = 'git-lfs/2.3.4 (Mercurial %s)' % util.version()
         self.urlopener = urlmod.opener(ui, authinfo, useragent)
         self.retry = ui.configint('lfs', 'retry')
 
     def writebatch(self, pointers, fromstore):
         """Batch upload from local to remote blobstore."""
         self._batch(_deduplicate(pointers), fromstore, 'upload')
 
     def readbatch(self, pointers, tostore):
         """Batch download from remote to local blostore."""
         self._batch(_deduplicate(pointers), tostore, 'download')
 
     def _batchrequest(self, pointers, action):
         """Get metadata about objects pointed by pointers for given action
 
         Return decoded JSON object like {'objects': [{'oid': '', 'size': 1}]}
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/batch.md
         """
         objects = [{'oid': p.oid(), 'size': p.size()} for p in pointers]
         requestdata = json.dumps({
             'objects': objects,
             'operation': action,
         })
         batchreq = util.urlreq.request('%s/objects/batch' % self.baseurl,
                                        data=requestdata)
         batchreq.add_header('Accept', 'application/vnd.git-lfs+json')
         batchreq.add_header('Content-Type', 'application/vnd.git-lfs+json')
         try:
             rawjson = self.urlopener.open(batchreq).read()
         except util.urlerr.httperror as ex:
             raise LfsRemoteError(_('LFS HTTP error: %s (action=%s)')
                                  % (ex, action))
         try:
             response = json.loads(rawjson)
         except ValueError:
             raise LfsRemoteError(_('LFS server returns invalid JSON: %s')
                                  % rawjson)
         return response
 
     def _checkforservererror(self, pointers, responses, action):
         """Scans errors from objects
 
         Raises LfsRemoteError if any objects have an error"""
         for response in responses:
             # The server should return 404 when objects cannot be found. Some
             # server implementation (ex. lfs-test-server)  does not set "error"
             # but just removes "download" from "actions". Treat that case
             # as the same as 404 error.
             notfound = (response.get('error', {}).get('code') == 404
                         or (action == 'download'
                             and action not in response.get('actions', [])))
             if notfound:
                 ptrmap = {p.oid(): p for p in pointers}
                 p = ptrmap.get(response['oid'], None)
                 if p:
                     filename = getattr(p, 'filename', 'unknown')
                     raise LfsRemoteError(
                         _(('LFS server error. Remote object '
                           'for "%s" not found: %r')) % (filename, response))
                 else:
                     raise LfsRemoteError(
                         _('LFS server error. Unsolicited response for oid %s')
                         % response['oid'])
             if 'error' in response:
                 raise LfsRemoteError(_('LFS server error: %r') % response)
 
     def _extractobjects(self, response, pointers, action):
         """extract objects from response of the batch API
 
         response: parsed JSON object returned by batch API
         return response['objects'] filtered by action
         raise if any object has an error
         """
         # Scan errors from objects - fail early
         objects = response.get('objects', [])
         self._checkforservererror(pointers, objects, action)
 
         # Filter objects with given action. Practically, this skips uploading
         # objects which exist in the server.
         filteredobjects = [o for o in objects if action in o.get('actions', [])]
 
         return filteredobjects
 
     def _basictransfer(self, obj, action, localstore):
         """Download or upload a single object using basic transfer protocol
 
         obj: dict, an object description returned by batch API
         action: string, one of ['upload', 'download']
         localstore: blobstore.local
 
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/\
         basic-transfers.md
         """
         oid = str(obj['oid'])
 
         href = str(obj['actions'][action].get('href'))
         headers = obj['actions'][action].get('header', {}).items()
 
         request = util.urlreq.request(href)
         if action == 'upload':
             # If uploading blobs, read data from local blobstore.
             with localstore.open(oid) as fp:
                 _verifyfile(oid, fp)
             request.data = filewithprogress(localstore.open(oid), None)
             request.get_method = lambda: 'PUT'
 
         for k, v in headers:
             request.add_header(k, v)
 
         response = b''
         try:
             req = self.urlopener.open(request)
             if action == 'download':
                 # If downloading blobs, store downloaded data to local blobstore
                 localstore.download(oid, req)
             else:
                 while True:
                     data = req.read(1048576)
                     if not data:
                         break
                     response += data
                 if response:
                     self.ui.debug('lfs %s response: %s' % (action, response))
         except util.urlerr.httperror as ex:
             if self.ui.debugflag:
                 self.ui.debug('%s: %s\n' % (oid, ex.read()))
             raise LfsRemoteError(_('HTTP error: %s (oid=%s, action=%s)')
                                  % (ex, oid, action))
 
     def _batch(self, pointers, localstore, action):
         if action not in ['upload', 'download']:
             raise error.ProgrammingError('invalid Git-LFS action: %s' % action)
 
         response = self._batchrequest(pointers, action)
         objects = self._extractobjects(response, pointers, action)
         total = sum(x.get('size', 0) for x in objects)
         sizes = {}
         for obj in objects:
             sizes[obj.get('oid')] = obj.get('size', 0)
         topic = {'upload': _('lfs uploading'),
                  'download': _('lfs downloading')}[action]
         if len(objects) > 1:
             self.ui.note(_('lfs: need to transfer %d objects (%s)\n')
                          % (len(objects), util.bytecount(total)))
         self.ui.progress(topic, 0, total=total)
         def transfer(chunk):
             for obj in chunk:
                 objsize = obj.get('size', 0)
                 if self.ui.verbose:
                     if action == 'download':
                         msg = _('lfs: downloading %s (%s)\n')
                     elif action == 'upload':
                         msg = _('lfs: uploading %s (%s)\n')
                     self.ui.note(msg % (obj.get('oid'),
                                  util.bytecount(objsize)))
                 retry = self.retry
                 while True:
                     try:
                         self._basictransfer(obj, action, localstore)
                         yield 1, obj.get('oid')
                         break
                     except socket.error as ex:
                         if retry > 0:
                             self.ui.note(
                                 _('lfs: failed: %r (remaining retry %d)\n')
                                 % (ex, retry))
                             retry -= 1
                             continue
                         raise
 
         # Until https multiplexing gets sorted out
         if self.ui.configbool('experimental', 'lfs.worker-enable'):
             oids = worker.worker(self.ui, 0.1, transfer, (),
                                  sorted(objects, key=lambda o: o.get('oid')))
         else:
             oids = transfer(sorted(objects, key=lambda o: o.get('oid')))
 
         processed = 0
         blobs = 0
         for _one, oid in oids:
             processed += sizes[oid]
             blobs += 1
             self.ui.progress(topic, processed, total=total)
             self.ui.note(_('lfs: processed: %s\n') % oid)
         self.ui.progress(topic, pos=None, total=total)
 
         if blobs > 0:
             if action == 'upload':
                 self.ui.status(_('lfs: uploaded %d files (%s)\n')
                                % (blobs, util.bytecount(processed)))
             # TODO: coalesce the download requests, and comment this in
             #elif action == 'download':
             #    self.ui.status(_('lfs: downloaded %d files (%s)\n')
             #                   % (blobs, util.bytecount(processed)))
 
     def __del__(self):
         # copied from mercurial/httppeer.py
         urlopener = getattr(self, 'urlopener', None)
         if urlopener:
             for h in urlopener.handlers:
                 h.close()
                 getattr(h, "close_all", lambda : None)()
 
 class _dummyremote(object):
     """Dummy store storing blobs to temp directory."""
 
     def __init__(self, repo, url):
         fullpath = repo.vfs.join('lfs', url.path)
         self.vfs = lfsvfs(fullpath)
 
     def writebatch(self, pointers, fromstore):
         for p in _deduplicate(pointers):
             content = fromstore.read(p.oid(), verify=True)
             with self.vfs(p.oid(), 'wb', atomictemp=True) as fp:
                 fp.write(content)
 
     def readbatch(self, pointers, tostore):
         for p in _deduplicate(pointers):
             with self.vfs(p.oid(), 'rb') as fp:
                 tostore.download(p.oid(), fp)
 
 class _nullremote(object):
     """Null store storing blobs to /dev/null."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore):
         pass
 
     def readbatch(self, pointers, tostore):
         pass
 
 class _promptremote(object):
     """Prompt user to set lfs.url when accessed."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore, ui=None):
         self._prompt()
 
     def readbatch(self, pointers, tostore, ui=None):
         self._prompt()
 
     def _prompt(self):
         raise error.Abort(_('lfs.url needs to be configured'))
 
 _storemap = {
     'https': _gitlfsremote,
     'http': _gitlfsremote,
     'file': _dummyremote,
     'null': _nullremote,
     None: _promptremote,
 }
 
 def _deduplicate(pointers):
     """Remove any duplicate oids that exist in the list"""
     reduced = util.sortdict()
     for p in pointers:
         reduced[p.oid()] = p
     return reduced.values()
 
 def _verify(oid, content):
     realoid = hashlib.sha256(content).hexdigest()
     if realoid != oid:
         raise error.Abort(_('detected corrupt lfs object: %s') % oid,
                           hint=_('run hg verify'))
 
 def _verifyfile(oid, fp):
     sha256 = hashlib.sha256()
     while True:
         data = fp.read(1024 * 1024)
         if not data:
             break
         sha256.update(data)
     realoid = sha256.hexdigest()
     if realoid != oid:
         raise error.Abort(_('detected corrupt lfs object: %s') % oid,
                           hint=_('run hg verify'))
 
 def remote(repo):
     """remotestore factory. return a store in _storemap depending on config"""
     url = util.url(repo.ui.config('lfs', 'url') or '')
     scheme = url.scheme
     if scheme not in _storemap:
         raise error.Abort(_('lfs: unknown url scheme: %s') % scheme)
     return _storemap[scheme](repo, url)
 
 class LfsRemoteError(error.RevlogError):
     pass