kill trailing whitespace

hgext/convert/subversion.py

@@ -1,1136 +1,1136
 # Subversion 1.4/1.5 Python API backend
 #
 # Copyright(C) 2007 Daniel Holth et al
 
 import os
 import re
 import sys
 import cPickle as pickle
 import tempfile
 import urllib
 
 from mercurial import strutil, util, encoding
 from mercurial.i18n import _
 
 # 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.
 
 from cStringIO import StringIO
 
 from common import NoRepo, MissingTool, commit, encodeargs, decodeargs
 from common import commandline, converter_source, converter_sink, mapfile
 
 try:
     from svn.core import SubversionException, Pool
     import svn
     import svn.client
     import svn.core
     import svn.ra
     import svn.delta
     import transport
     import warnings
     warnings.filterwarnings('ignore',
             module='svn.core',
             category=DeprecationWarning)
 
 except ImportError:
     pass
 
 class SvnPathNotFound(Exception):
     pass
 
 def geturl(path):
     try:
         return svn.client.url_from_path(svn.core.svn_path_canonicalize(path))
     except SubversionException:
         pass
     if os.path.isdir(path):
         path = os.path.normpath(os.path.abspath(path))
         if os.name == 'nt':
             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)
         return 'file://%s' % urllib.quote(path)
     return 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):
         if orig_paths is not None:
             for k, v in orig_paths.iteritems():
                 orig_paths[k] = changedpath(v)
         pickle.dump((orig_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 SubversionException, (inst, num):
         pickle.dump(num, fp, protocol)
     except IOError:
         # Caller may interrupt the iteration
         pickle.dump(None, 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.
     """
     util.set_binary(sys.stdin)
     util.set_binary(sys.stdout)
     args = decodeargs(sys.stdin.read())
     get_log_child(sys.stdout, *args)
 
 class logstream(object):
     """Interruptible revision log iterator."""
     def __init__(self, stdout):
         self._stdout = stdout
 
     def __iter__(self):
         while True:
             entry = pickle.load(self._stdout)
             try:
                 orig_paths, revnum, author, date, message = entry
             except:
                 if entry is None:
                     break
                 raise SubversionException("child raised exception", entry)
             yield entry
 
     def close(self):
         if self._stdout:
             self._stdout.close()
             self._stdout = None
 
 
 # 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(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(path, proto):
     return ('<m:human-readable errcode="160013">' in
             urllib.urlopen('%s://%s/!svn/ver/0/.svn' % (proto, path)).read())
 
 protomap = {'http': httpcheck,
             'https': httpcheck,
             'file': filecheck,
             }
 def issvnurl(url):
     try:
         proto, path = url.split('://', 1)
         path = urllib.url2pathname(path)
     except ValueError:
         proto = 'file'
         path = os.path.abspath(url)
     path = path.replace(os.sep, '/')
     check = protomap.get(proto, lambda p, p2: False)
     while '/' in path:
         if check(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, url, rev=None):
         super(svn_source, self).__init__(ui, url, rev=rev)
 
         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(url)):
             raise NoRepo("%s does not look like a Subversion repo" % url)
 
         try:
             SubversionException
         except NameError:
             raise MissingTool(_('Subversion python bindings could not be loaded'))
 
         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 = urllib.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 SubversionException:
             ui.traceback()
             raise NoRepo("%s does not look like a Subversion repo" % self.url)
 
         if rev:
             try:
                 latest = int(rev)
             except ValueError:
                 raise util.Abort(_('svn: revision %s is not an integer') % rev)
 
         self.startrev = self.ui.config('convert', 'svn.startrev', default=0)
         try:
             self.startrev = int(self.startrev)
             if self.startrev < 0:
                 self.startrev = 0
         except ValueError:
             raise util.Abort(_('svn: start revision %s is not an integer')
                              % self.startrev)
 
         self.head = self.latest(self.module, latest)
         if not self.head:
             raise util.Abort(_('no revision found in module %s')
                              % self.module)
         self.last_changed = self.revnum(self.head)
 
         self._changescache = 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.iterkeys():
             uuid, module, revnum = self.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('/') + '/' + urllib.quote(path),
                                  optrev, False, self.ctx)
             return True
         except 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 cfgpath:
                     raise util.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 util.Abort(_('no revision found in module %s') 
+                raise util.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 + '/' + urllib.quote(branches),
                                         rev, False, self.ctx)
             for branch in branchnames.keys():
                 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 util.Abort(_('svn: start revision is not supported '
                                    'with more than one branch'))
             revnum = self.revnum(self.heads[0])
             if revnum < self.startrev:
                 raise util.Abort(_('svn: no revision found after start revision %d')
                                  % self.startrev)
 
         return self.heads
 
     def getfile(self, file, rev):
         data, mode = self._getfile(file, rev)
         self.modecache[(file, rev)] = mode
         return data
 
     def getmode(self, file, rev):
         return self.modecache[(file, rev)]
 
     def getchanges(self, rev):
         if self._changescache and self._changescache[0] == rev:
             return self._changescache[1]
         self._changescache = None
         self.modecache = {}
         (paths, parents) = self.paths[rev]
         if parents:
             files, copies = self.expandpaths(rev, paths, parents)
         else:
             # Perform a full checkout on roots
             uuid, module, revnum = self.revsplit(rev)
             entries = svn.client.ls(self.baseurl + urllib.quote(module),
                                     optrev(revnum), True, self.ctx)
             files = [n for n,e in entries.iteritems()
                      if e.kind == svn.core.svn_node_file]
             copies = {}
 
         files.sort()
         files = zip(files, [rev] * len(files))
 
         # caller caches the result, so free it here to release memory
         del self.paths[rev]
         return (files, copies)
 
     def getchangedfiles(self, rev, i):
         changes = self.getchanges(rev)
         self._changescache = (rev, changes)
         return [f[0] for f in changes[0]]
 
     def getcommit(self, rev):
         if rev not in self.commits:
             uuid, module, revnum = self.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)
         commit = self.commits[rev]
         # caller caches the result, so free it here to release memory
         del self.commits[rev]
         return commit
 
     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)
         try:
             for entry in self._getlog([self.tags], start, self.startrev):
                 origpaths, revnum, author, date, message = entry
                 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() 
+                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
                 remainings = []
                 for source, sourcerev, dest in pendings:
                     tagname = dest.split('/')[-1]
                     if source.startswith(srctagspath):
                         remainings.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 = remainings
                 tagspath = srctagspath
 
         except SubversionException:
             self.ui.note(_('no tags found at revision %d\n') % start)
         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'),
                                   'a')
         self.convertfp.write('%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 revsplit(self, rev):
         url, revnum = rev.rsplit('@', 1)
         revnum = int(revnum)
         parts = url.split('/', 1)
         uuid = parts.pop(0)[4:]
         mod = ''
         if parts:
             mod = '/' + parts[0]
         return uuid, mod, revnum
 
     def latest(self, path, stop=0):
         """Find the latest revid affecting path, up to stop. 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.
         """
         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 not stop:
             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 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.
         stream = self._getlog([path], stop, dirent.created_rev)
         try:
             for entry in stream:
                 paths, revnum, author, date, message = entry
                 if revnum <= dirent.created_rev:
                     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
         finally:
             stream.close()
 
         if not path.startswith(self.rootmodule):
             self.ui.debug(_('ignoring foreign branch %r\n') % path)
             return None
         return self.revid(dirent.created_rev, path)
 
     def reparent(self, module):
         """Reparent the svn transport and return the previous parent."""
         if self.prevmodule == module:
             return module
         svnurl = self.baseurl + urllib.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):
         entries = []
         # Map of entrypath, revision for finding source of deleted
         # revisions.
         copyfrom = {}
         copies = {}
 
         new_module, revnum = self.revsplit(rev)[1:]
         if new_module != self.module:
             self.module = new_module
             self.reparent(self.module)
 
         for path, ent in paths:
             entrypath = self.getrelpath(path)
 
             kind = self._checkpath(entrypath, revnum)
             if kind == svn.core.svn_node_file:
                 entries.append(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 = self.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 = self.revsplit(parents[0])[1:]
                 parentpath = pmodule + "/" + entrypath
                 self.ui.debug(_("entry %s\n") % parentpath)
 
                 # We can avoid the reparent calls if the module has
                 # not changed but it probably does not worth the pain.
                 prevmodule = self.reparent('')
                 fromkind = svn.ra.check_path(self.ra, parentpath.strip('/'), prevnum)
                 self.reparent(prevmodule)
 
                 if fromkind == svn.core.svn_node_file:
                     entries.append(self.recode(entrypath))
                 elif fromkind == svn.core.svn_node_dir:
                     if ent.action == 'C':
                         children = self._find_children(path, prevnum)
                     else:
                         oroot = parentpath.strip('/')
                         nroot = path.strip('/')
                         children = self._find_children(oroot, prevnum)
                         children = [s.replace(oroot,nroot) for s in children]
 
                     for child in children:
                         childpath = self.getrelpath("/" + child, pmodule)
                         if not childpath:
                             continue
                         if childpath in copies:
                             del copies[childpath]
                         entries.append(childpath)
                 else:
                     self.ui.debug(_('unknown path in revision %d: %s\n') % \
                                   (revnum, path))
             elif kind == svn.core.svn_node_dir:
                 # If the directory just had a prop change,
                 # then we shouldn't need to look for its children.
                 if ent.action == 'M':
                     continue
 
                 children = sorted(self._find_children(path, revnum))
                 for child in children:
                     # Can we move a child directory and its
                     # parent in the same commit? (probably can). Could
                     # cause problems if instead of revnum -1,
                     # we have to look in (copyfrom_path, revnum - 1)
                     entrypath = self.getrelpath("/" + child)
                     if entrypath:
                         # Need to filter out directories here...
                         kind = self._checkpath(entrypath, revnum)
                         if kind != svn.core.svn_node_dir:
                             entries.append(self.recode(entrypath))
 
                 # 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 = self.revsplit(parents[0])[1:]
                 if ent.copyfrom_rev < prevnum:
                     continue
                 copyfrompath = self.getrelpath(ent.copyfrom_path, pmodule)
                 if not copyfrompath:
                     continue
                 copyfrom[path] = ent
                 self.ui.debug(_("mark %s came from %s:%d\n")
                               % (path, copyfrompath, ent.copyfrom_rev))
                 children = self._find_children(ent.copyfrom_path, ent.copyfrom_rev)
                 children.sort()
                 for child in children:
                     entrypath = self.getrelpath("/" + child, pmodule)
                     if not entrypath:
                         continue
                     copytopath = path + entrypath[len(copyfrompath):]
                     copytopath = self.getrelpath(copytopath)
                     copies[self.recode(copytopath)] = self.recode(entrypath)
 
         return (list(set(entries)), 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 = self.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"])
 
             log = message and self.recode(message) or ''
             author = author and self.recode(author) or ''
             try:
                 branch = self.module.split("/")[-1]
                 if branch == 'trunk':
                     branch = ''
             except IndexError:
                 branch = None
 
             cset = commit(author=author,
                           date=util.datestr(date),
                           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)
                         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 SubversionException, (inst, num):
             if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION:
                 raise util.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.
         mode = ''
         try:
             new_module, revnum = self.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_files() seems to keep a reference on the input buffer
             # preventing collection. Release it explicitely.
             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 SubversionException, 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
                 raise IOError()
             raise
         if mode == 'l':
             link_prefix = "link "
             if data.startswith(link_prefix):
                 data = data[len(link_prefix):]
         return data, mode
 
     def _find_children(self, path, revnum):
         path = path.strip('/')
         pool = Pool()
         rpath = '/'.join([self.baseurl, urllib.quote(path)]).strip('/')
         return ['%s/%s' % (path, x) for x in
                 svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool).keys()]
 
     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, ie
         #   "/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):
         # 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)
 
     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]
         arg = encodeargs(args)
         hgexe = util.hgexecutable()
         cmd = '%s debugsvnlog' % util.shellquote(hgexe)
         stdin, stdout = util.popen2(cmd)
         stdin.write(arg)
         stdin.close()
         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)
 
     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, path):
         converter_sink.__init__(self, ui, path)
         commandline.__init__(self, ui, 'svn')
         self.delete = []
         self.setexec = []
         self.delexec = []
         self.copies = []
         self.wc = None
         self.cwd = os.getcwd()
 
         path = os.path.realpath(path)
 
         created = False
         if os.path.isfile(os.path.join(path, '.svn', 'entries')):
             self.wc = path
             self.run0('update')
         else:
             wcpath = os.path.join(os.getcwd(), os.path.basename(path) + '-wc')
 
             if os.path.isdir(os.path.dirname(path)):
                 if not os.path.exists(os.path.join(path, 'db', 'fs-type')):
                     ui.status(_('initializing svn repo %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
 
             ui.status(_('initializing svn wc %r\n') % os.path.basename(wcpath))
             self.run0('checkout', path, wcpath)
 
             self.wc = wcpath
         self.opener = util.opener(self.wc)
         self.wopener = util.opener(self.wc)
         self.childmap = mapfile(ui, self.join('hg-childmap'))
         self.is_exec = util.checkexec(self.wc) and util.is_exec or None
 
         if created:
             hook = os.path.join(created, 'hooks', 'pre-revprop-change')
             fp = open(hook, 'w')
             fp.write(pre_revprop_change)
             fp.close()
             util.set_flags(hook, False, True)
 
         xport = transport.SvnRaTransport(url=geturl(path))
         self.uuid = svn.ra.get_uuid(xport.ra)
 
     def wjoin(self, *names):
         return os.path.join(self.wc, *names)
 
     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(filename, 'w').write(data)
 
             if self.is_exec:
                 was_exec = self.is_exec(self.wjoin(filename))
             else:
                 # On filesystems not supporting execute-bit, there is no way
                 # to know if it is set but asking subversion. Setting it
                 # systematically is just as expensive and much simpler.
                 was_exec = 'x' not in flags
 
             util.set_flags(self.wjoin(filename), False, 'x' in flags)
             if was_exec:
                 if 'x' not in flags:
                     self.delexec.append(filename)
             else:
                 if 'x' in flags:
                     self.setexec.append(filename)
 
     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.exists(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:
             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)
             for i in strutil.rfindall(f, '/'):
                 dirs.add(f[:i])
         return dirs
 
     def add_dirs(self, files):
         add_dirs = [d for d in sorted(self.dirs_of(files))
                     if not os.path.exists(self.wjoin(d, '.svn', 'entries'))]
         if add_dirs:
             self.xargs(add_dirs, 'add', non_recursive=True, quiet=True)
         return add_dirs
 
     def add_files(self, files):
         if files:
             self.xargs(files, 'add', quiet=True)
         return files
 
     def tidy_dirs(self, names):
         deleted = []
         for d in sorted(self.dirs_of(names), reverse=True):
             wd = self.wjoin(d)
             if os.listdir(wd) == '.svn':
                 self.run0('delete', d)
                 deleted.append(d)
         return deleted
 
     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):
         # Apply changes to working copy
         for f, v in files:
             try:
                 data = source.getfile(f, v)
             except IOError:
                 self.delete.append(f)
             else:
                 e = source.getmode(f, v)
                 self.putfile(f, e, data)
                 if f in copies:
                     self.copies.append([copies[f], f])
         files = [f[0] for f in files]
 
         for parent in parents:
             try:
                 return self.revid(self.childmap[parent])
             except KeyError:
                 pass
         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')
             self.delete = []
         entries.update(self.add_files(files.difference(entries)))
         entries.update(self.tidy_dirs(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, 'w')
         fp.write(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:
                 self.ui.warn(_('unexpected svn output:\n'))
                 self.ui.warn(output)
                 raise util.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(_('XXX TAGS NOT IMPLEMENTED YET\n'))

mercurial/tags.py

@@ -1,339 +1,338
 # tags.py - read tag info from local repository
 #
 # Copyright 2009 Matt Mackall <[e-mail]>
 # Copyright 2009 Greg Ward <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2, incorporated herein by reference.
 
 # Currently this module only deals with reading and caching tags.
 # Eventually, it could take care of updating (adding/removing/moving)
 # tags too.
 
 import os
 from node import nullid, bin, hex, short
 from i18n import _
 import encoding
 import error
 
 def _debugalways(ui, *msg):
     ui.write(*msg)
 
 def _debugconditional(ui, *msg):
     ui.debug(*msg)
 
 def _debugnever(ui, *msg):
     pass
 
 _debug = _debugalways
 _debug = _debugnever
 
 def findglobaltags1(ui, repo, alltags, tagtypes):
     '''Find global tags in repo by reading .hgtags from every head that
     has a distinct version of it.  Updates the dicts alltags, tagtypes
     in place: alltags maps tag name to (node, hist) pair (see _readtags()
     below), and tagtypes maps tag name to tag type ('global' in this
     case).'''
 
     seen = set()
     fctx = None
     ctxs = []                       # list of filectx
     for node in repo.heads():
         try:
             fnode = repo[node].filenode('.hgtags')
         except error.LookupError:
             continue
         if fnode not in seen:
             seen.add(fnode)
             if not fctx:
                 fctx = repo.filectx('.hgtags', fileid=fnode)
             else:
                 fctx = fctx.filectx(fnode)
             ctxs.append(fctx)
 
     # read the tags file from each head, ending with the tip
     for fctx in reversed(ctxs):
         filetags = _readtags(
             ui, repo, fctx.data().splitlines(), fctx)
         _updatetags(filetags, "global", alltags, tagtypes)
 
 def findglobaltags2(ui, repo, alltags, tagtypes):
     '''Same as findglobaltags1(), but with caching.'''
     # This is so we can be lazy and assume alltags contains only global
     # tags when we pass it to _writetagcache().
     assert len(alltags) == len(tagtypes) == 0, \
            "findglobaltags() should be called first"
 
     (heads, tagfnode, cachetags, shouldwrite) = _readtagcache(ui, repo)
     if cachetags is not None:
         assert not shouldwrite
         # XXX is this really 100% correct?  are there oddball special
         # cases where a global tag should outrank a local tag but won't,
         # because cachetags does not contain rank info?
         _updatetags(cachetags, 'global', alltags, tagtypes)
         return
 
     _debug(ui, "reading tags from %d head(s): %s\n"
            % (len(heads), map(short, reversed(heads))))
     seen = set()                    # set of fnode
     fctx = None
     for head in reversed(heads):        # oldest to newest
         assert head in repo.changelog.nodemap, \
                "tag cache returned bogus head %s" % short(head)
 
         fnode = tagfnode.get(head)
         if fnode and fnode not in seen:
             seen.add(fnode)
             if not fctx:
                 fctx = repo.filectx('.hgtags', fileid=fnode)
             else:
                 fctx = fctx.filectx(fnode)
 
             filetags = _readtags(ui, repo, fctx.data().splitlines(), fctx)
             _updatetags(filetags, 'global', alltags, tagtypes)
 
     # and update the cache (if necessary)
     if shouldwrite:
         _writetagcache(ui, repo, heads, tagfnode, alltags)
 
 # Set this to findglobaltags1 to disable tag caching.
 findglobaltags = findglobaltags2
 
 def readlocaltags(ui, repo, alltags, tagtypes):
     '''Read local tags in repo.  Update alltags and tagtypes.'''
     try:
         # localtags is in the local encoding; re-encode to UTF-8 on
         # input for consistency with the rest of this module.
         data = repo.opener("localtags").read()
         filetags = _readtags(
             ui, repo, data.splitlines(), "localtags",
             recode=encoding.fromlocal)
         _updatetags(filetags, "local", alltags, tagtypes)
     except IOError:
         pass
 
 def _readtags(ui, repo, lines, fn, recode=None):
     '''Read tag definitions from a file (or any source of lines).
     Return a mapping from tag name to (node, hist): node is the node id
     from the last line read for that name, and hist is the list of node
     ids previously associated with it (in file order).  All node ids are
     binary, not hex.'''
 
     filetags = {}               # map tag name to (node, hist)
     count = 0
 
     def warn(msg):
         ui.warn(_("%s, line %s: %s\n") % (fn, count, msg))
 
     for line in lines:
         count += 1
         if not line:
             continue
         try:
             (nodehex, name) = line.split(" ", 1)
         except ValueError:
             warn(_("cannot parse entry"))
             continue
         name = name.strip()
         if recode:
             name = recode(name)
         try:
             nodebin = bin(nodehex)
         except TypeError:
             warn(_("node '%s' is not well formed") % nodehex)
             continue
         if nodebin not in repo.changelog.nodemap:
             # silently ignore as pull -r might cause this
             continue
 
         # update filetags
         hist = []
         if name in filetags:
             n, hist = filetags[name]
             hist.append(n)
         filetags[name] = (nodebin, hist)
     return filetags
 
 def _updatetags(filetags, tagtype, alltags, tagtypes):
     '''Incorporate the tag info read from one file into the two
     dictionaries, alltags and tagtypes, that contain all tag
     info (global across all heads plus local).'''
 
     for name, nodehist in filetags.iteritems():
         if name not in alltags:
             alltags[name] = nodehist
             tagtypes[name] = tagtype
             continue
 
         # we prefer alltags[name] if:
         #  it supercedes us OR
         #  mutual supercedes and it has a higher rank
         # otherwise we win because we're tip-most
         anode, ahist = nodehist
         bnode, bhist = alltags[name]
         if (bnode != anode and anode in bhist and
             (bnode not in ahist or len(bhist) > len(ahist))):
             anode = bnode
         ahist.extend([n for n in bhist if n not in ahist])
         alltags[name] = anode, ahist
         tagtypes[name] = tagtype
 
 
 # The tag cache only stores info about heads, not the tag contents
 # from each head.  I.e. it doesn't try to squeeze out the maximum
 # performance, but is simpler has a better chance of actually
 # working correctly.  And this gives the biggest performance win: it
 # avoids looking up .hgtags in the manifest for every head, and it
 # can avoid calling heads() at all if there have been no changes to
 # the repo.
 
 def _readtagcache(ui, repo):
     '''Read the tag cache and return a tuple (heads, fnodes, cachetags,
     shouldwrite).  If the cache is completely up-to-date, cachetags is a
     dict of the form returned by _readtags(); otherwise, it is None and
     heads and fnodes are set.  In that case, heads is the list of all
     heads currently in the repository (ordered from tip to oldest) and
     fnodes is a mapping from head to .hgtags filenode.  If those two are
     set, caller is responsible for reading tag info from each head.'''
 
     try:
         cachefile = repo.opener('tags.cache', 'r')
         _debug(ui, 'reading tag cache from %s\n' % cachefile.name)
     except IOError:
         cachefile = None
 
     # The cache file consists of lines like
     #   <headrev> <headnode> [<tagnode>]
     # where <headrev> and <headnode> redundantly identify a repository
     # head from the time the cache was written, and <tagnode> is the
     # filenode of .hgtags on that head.  Heads with no .hgtags file will
     # have no <tagnode>.  The cache is ordered from tip to oldest (which
     # is part of why <headrev> is there: a quick visual check is all
     # that's required to ensure correct order).
-    # 
+    #
     # This information is enough to let us avoid the most expensive part
     # of finding global tags, which is looking up <tagnode> in the
     # manifest for each head.
     cacherevs = []                      # list of headrev
     cacheheads = []                     # list of headnode
     cachefnode = {}                     # map headnode to filenode
     if cachefile:
         for line in cachefile:
             if line == "\n":
                 break
             line = line.rstrip().split()
             cacherevs.append(int(line[0]))
             headnode = bin(line[1])
             cacheheads.append(headnode)
             if len(line) == 3:
                 fnode = bin(line[2])
                 cachefnode[headnode] = fnode
 
     tipnode = repo.changelog.tip()
     tiprev = len(repo.changelog) - 1
 
     # Case 1 (common): tip is the same, so nothing has changed.
     # (Unchanged tip trivially means no changesets have been added.
     # But, thanks to localrepository.destroyed(), it also means none
     # have been destroyed by strip or rollback.)
     if cacheheads and cacheheads[0] == tipnode and cacherevs[0] == tiprev:
         _debug(ui, "tag cache: tip unchanged\n")
         tags = _readtags(ui, repo, cachefile, cachefile.name)
         cachefile.close()
         return (None, None, tags, False)
     if cachefile:
         cachefile.close()               # ignore rest of file
-        
+
     repoheads = repo.heads()
-
     # Case 2 (uncommon): empty repo; get out quickly and don't bother
     # writing an empty cache.
     if repoheads == [nullid]:
         return ([], {}, {}, False)
 
     # Case 3 (uncommon): cache file missing or empty.
     if not cacheheads:
         _debug(ui, 'tag cache: cache file missing or empty\n')
 
     # Case 4 (uncommon): tip rev decreased.  This should only happen
     # when we're called from localrepository.destroyed().  Refresh the
     # cache so future invocations will not see disappeared heads in the
     # cache.
     elif cacheheads and tiprev < cacherevs[0]:
         _debug(ui,
                'tag cache: tip rev decremented (from %d to %d), '
                'so we must be destroying nodes\n'
                % (cacherevs[0], tiprev))
 
     # Case 5 (common): tip has changed, so we've added/replaced heads.
     else:
         _debug(ui,
                'tag cache: tip has changed (%d:%s); must find new heads\n'
                % (tiprev, short(tipnode)))
 
     # Luckily, the code to handle cases 3, 4, 5 is the same.  So the
     # above if/elif/else can disappear once we're confident this thing
     # actually works and we don't need the debug output.
 
     # N.B. in case 4 (nodes destroyed), "new head" really means "newly
     # exposed".
     newheads = [head
                 for head in repoheads
                 if head not in set(cacheheads)]
     _debug(ui, 'tag cache: found %d head(s) not in cache: %s\n'
            % (len(newheads), map(short, newheads)))
 
     # Now we have to lookup the .hgtags filenode for every new head.
     # This is the most expensive part of finding tags, so performance
     # depends primarily on the size of newheads.  Worst case: no cache
     # file, so newheads == repoheads.
     for head in newheads:
         cctx = repo[head]
         try:
             fnode = cctx.filenode('.hgtags')
             cachefnode[head] = fnode
         except error.LookupError:
             # no .hgtags file on this head
             pass
 
     # Caller has to iterate over all heads, but can use the filenodes in
     # cachefnode to get to each .hgtags revision quickly.
     return (repoheads, cachefnode, None, True)
 
 def _writetagcache(ui, repo, heads, tagfnode, cachetags):
 
     cachefile = repo.opener('tags.cache', 'w', atomictemp=True)
     _debug(ui, 'writing cache file %s\n' % cachefile.name)
 
     realheads = repo.heads()            # for sanity checks below
     for head in heads:
         # temporary sanity checks; these can probably be removed
         # once this code has been in crew for a few weeks
         assert head in repo.changelog.nodemap, \
                'trying to write non-existent node %s to tag cache' % short(head)
         assert head in realheads, \
                'trying to write non-head %s to tag cache' % short(head)
         assert head != nullid, \
                'trying to write nullid to tag cache'
 
         # This can't fail because of the first assert above.  When/if we
         # remove that assert, we might want to catch LookupError here
         # and downgrade it to a warning.
         rev = repo.changelog.rev(head)
 
         fnode = tagfnode.get(head)
         if fnode:
             cachefile.write('%d %s %s\n' % (rev, hex(head), hex(fnode)))
         else:
             cachefile.write('%d %s\n' % (rev, hex(head)))
 
     # Tag names in the cache are in UTF-8 -- which is the whole reason
     # we keep them in UTF-8 throughout this module.  If we converted
     # them local encoding on input, we would lose info writing them to
     # the cache.
     cachefile.write('\n')
     for (name, (node, hist)) in cachetags.iteritems():
         cachefile.write("%s %s\n" % (hex(node), name))
 
     cachefile.rename()
     cachefile.close()

mercurial/ui.py

@@ -1,381 +1,381
 # ui.py - user interface bits for mercurial
 #
 # Copyright 2005-2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2, incorporated herein by reference.
 
 from i18n import _
 import errno, getpass, os, socket, sys, tempfile, traceback
 import config, util, error
 
 _booleans = {'1': True, 'yes': True, 'true': True, 'on': True,
              '0': False, 'no': False, 'false': False, 'off': False}
 
 class ui(object):
     def __init__(self, src=None):
         self._buffers = []
         self.quiet = self.verbose = self.debugflag = self._traceback = False
         self._reportuntrusted = True
         self._ocfg = config.config() # overlay
         self._tcfg = config.config() # trusted
         self._ucfg = config.config() # untrusted
         self._trustusers = set()
         self._trustgroups = set()
 
         if src:
             self._tcfg = src._tcfg.copy()
             self._ucfg = src._ucfg.copy()
             self._ocfg = src._ocfg.copy()
             self._trustusers = src._trustusers.copy()
             self._trustgroups = src._trustgroups.copy()
             self.fixconfig()
         else:
             # we always trust global config files
             for f in util.rcpath():
                 self.readconfig(f, trust=True)
 
     def copy(self):
         return self.__class__(self)
 
     def _is_trusted(self, fp, f):
         st = util.fstat(fp)
         if util.isowner(st):
             return True
 
         tusers, tgroups = self._trustusers, self._trustgroups
         if '*' in tusers or '*' in tgroups:
             return True
 
         user = util.username(st.st_uid)
         group = util.groupname(st.st_gid)
         if user in tusers or group in tgroups or user == util.username():
             return True
 
         if self._reportuntrusted:
             self.warn(_('Not trusting file %s from untrusted '
                         'user %s, group %s\n') % (f, user, group))
         return False
 
     def readconfig(self, filename, root=None, trust=False,
                    sections=None, remap=None):
         try:
             fp = open(filename)
         except IOError:
             if not sections: # ignore unless we were looking for something
                 return
             raise
 
         cfg = config.config()
         trusted = sections or trust or self._is_trusted(fp, filename)
 
         try:
             cfg.read(filename, fp, sections=sections, remap=remap)
         except error.ConfigError, inst:
             if trusted:
                 raise
             self.warn(_("Ignored: %s\n") % str(inst))
 
         if trusted:
             self._tcfg.update(cfg)
             self._tcfg.update(self._ocfg)
         self._ucfg.update(cfg)
         self._ucfg.update(self._ocfg)
 
         if root is None:
             root = os.path.expanduser('~')
         self.fixconfig(root=root)
 
     def fixconfig(self, root=None):
         # translate paths relative to root (or home) into absolute paths
         root = root or os.getcwd()
         for c in self._tcfg, self._ucfg, self._ocfg:
             for n, p in c.items('paths'):
                 if p and "://" not in p and not os.path.isabs(p):
                     c.set("paths", n, os.path.normpath(os.path.join(root, p)))
 
         # update ui options
         self.debugflag = self.configbool('ui', 'debug')
         self.verbose = self.debugflag or self.configbool('ui', 'verbose')
         self.quiet = not self.debugflag and self.configbool('ui', 'quiet')
         if self.verbose and self.quiet:
             self.quiet = self.verbose = False
         self._reportuntrusted = self.configbool("ui", "report_untrusted", True)
         self._traceback = self.configbool('ui', 'traceback', False)
 
         # update trust information
         self._trustusers.update(self.configlist('trusted', 'users'))
         self._trustgroups.update(self.configlist('trusted', 'groups'))
 
     def setconfig(self, section, name, value):
         for cfg in (self._ocfg, self._tcfg, self._ucfg):
             cfg.set(section, name, value)
         self.fixconfig()
 
     def _data(self, untrusted):
         return untrusted and self._ucfg or self._tcfg
 
     def configsource(self, section, name, untrusted=False):
         return self._data(untrusted).source(section, name) or 'none'
 
     def config(self, section, name, default=None, untrusted=False):
         value = self._data(untrusted).get(section, name, default)
         if self.debugflag and not untrusted and self._reportuntrusted:
             uvalue = self._ucfg.get(section, name)
             if uvalue is not None and uvalue != value:
                 self.debug(_("ignoring untrusted configuration option "
                              "%s.%s = %s\n") % (section, name, uvalue))
         return value
 
     def configbool(self, section, name, default=False, untrusted=False):
         v = self.config(section, name, None, untrusted)
         if v is None:
             return default
         if v.lower() not in _booleans:
             raise error.ConfigError(_("%s.%s not a boolean ('%s')")
                                     % (section, name, v))
         return _booleans[v.lower()]
 
     def configlist(self, section, name, default=None, untrusted=False):
         """Return a list of comma/space separated strings"""
         result = self.config(section, name, untrusted=untrusted)
         if result is None:
             result = default or []
         if isinstance(result, basestring):
             result = result.replace(",", " ").split()
         return result
 
     def has_section(self, section, untrusted=False):
         '''tell whether section exists in config.'''
         return section in self._data(untrusted)
 
     def configitems(self, section, untrusted=False):
         items = self._data(untrusted).items(section)
         if self.debugflag and not untrusted and self._reportuntrusted:
             for k, v in self._ucfg.items(section):
                 if self._tcfg.get(section, k) != v:
                     self.debug(_("ignoring untrusted configuration option "
                                 "%s.%s = %s\n") % (section, k, v))
         return items
 
     def walkconfig(self, untrusted=False):
         cfg = self._data(untrusted)
         for section in cfg.sections():
             for name, value in self.configitems(section, untrusted):
                 yield section, name, str(value).replace('\n', '\\n')
 
     def username(self):
         """Return default username to be used in commits.
 
         Searched in this order: $HGUSER, [ui] section of hgrcs, $EMAIL
         and stop searching if one of these is set.
         If not found and ui.askusername is True, ask the user, else use
         ($LOGNAME or $USER or $LNAME or $USERNAME) + "@full.hostname".
         """
         user = os.environ.get("HGUSER")
         if user is None:
             user = self.config("ui", "username")
         if user is None:
             user = os.environ.get("EMAIL")
         if user is None and self.configbool("ui", "askusername"):
             user = self.prompt(_("enter a commit username:"), default=None)
         if user is None:
             try:
                 user = '%s@%s' % (util.getuser(), socket.getfqdn())
                 self.warn(_("No username found, using '%s' instead\n") % user)
             except KeyError:
                 pass
         if not user:
             raise util.Abort(_("Please specify a username."))
         if "\n" in user:
             raise util.Abort(_("username %s contains a newline\n") % repr(user))
         return user
 
     def shortuser(self, user):
         """Return a short representation of a user name or email address."""
         if not self.verbose: user = util.shortuser(user)
         return user
 
     def _path(self, loc):
         p = self.config('paths', loc)
         if p and '%%' in p:
             self.warn('(deprecated \'%%\' in path %s=%s from %s)\n' %
                     (loc, p, self.configsource('paths', loc)))
             p = p.replace('%%', '%')
         return p
 
     def expandpath(self, loc, default=None):
         """Return repository location relative to cwd or from [paths]"""
         if "://" in loc or os.path.isdir(os.path.join(loc, '.hg')):
             return loc
 
         path = self._path(loc)
         if not path and default is not None:
             path = self._path(default)
         return path or loc
 
     def pushbuffer(self):
         self._buffers.append([])
 
     def popbuffer(self):
         return "".join(self._buffers.pop())
 
     def write(self, *args):
         if self._buffers:
             self._buffers[-1].extend([str(a) for a in args])
         else:
             for a in args:
                 sys.stdout.write(str(a))
 
     def write_err(self, *args):
         try:
             if not sys.stdout.closed: sys.stdout.flush()
             for a in args:
                 sys.stderr.write(str(a))
             # stderr may be buffered under win32 when redirected to files,
             # including stdout.
             if not sys.stderr.closed: sys.stderr.flush()
         except IOError, inst:
             if inst.errno != errno.EPIPE:
                 raise
 
     def flush(self):
         try: sys.stdout.flush()
         except: pass
         try: sys.stderr.flush()
         except: pass
 
     def interactive(self):
         i = self.configbool("ui", "interactive", None)
         if i is None:
             return sys.stdin.isatty()
         return i
 
     def _readline(self, prompt=''):
         if sys.stdin.isatty():
             try:
                 # magically add command line editing support, where
                 # available
                 import readline
                 # force demandimport to really load the module
                 readline.read_history_file
                 # windows sometimes raises something other than ImportError
             except Exception:
                 pass
         line = raw_input(prompt)
         # When stdin is in binary mode on Windows, it can cause
         # raw_input() to emit an extra trailing carriage return
         if os.linesep == '\r\n' and line and line[-1] == '\r':
             line = line[:-1]
         return line
 
     def prompt(self, msg, default="y"):
         """Prompt user with msg, read response.
         If ui is not interactive, the default is returned.
         """
         if not self.interactive():
             self.write(msg, ' ', default, "\n")
             return default
         try:
             r = self._readline(msg + ' ')
             if not r:
                 return default
             return r
         except EOFError:
             raise util.Abort(_('response expected'))
 
     def promptchoice(self, msg, choices, default=0):
         """Prompt user with msg, read response, and ensure it matches
         one of the provided choices. The index of the choice is returned.
         choices is a sequence of acceptable responses with the format:
-        ('&None', 'E&xec', 'Sym&link') Responses are case insensitive. 
+        ('&None', 'E&xec', 'Sym&link') Responses are case insensitive.
         If ui is not interactive, the default is returned.
         """
         resps = [s[s.index('&')+1].lower() for s in choices]
         while True:
             r = self.prompt(msg, resps[default])
             if r.lower() in resps:
                 return resps.index(r.lower())
             self.write(_("unrecognized response\n"))
 
 
     def getpass(self, prompt=None, default=None):
         if not self.interactive(): return default
         try:
             return getpass.getpass(prompt or _('password: '))
         except EOFError:
             raise util.Abort(_('response expected'))
     def status(self, *msg):
         if not self.quiet: self.write(*msg)
     def warn(self, *msg):
         self.write_err(*msg)
     def note(self, *msg):
         if self.verbose: self.write(*msg)
     def debug(self, *msg):
         if self.debugflag: self.write(*msg)
     def edit(self, text, user):
         (fd, name) = tempfile.mkstemp(prefix="hg-editor-", suffix=".txt",
                                       text=True)
         try:
             f = os.fdopen(fd, "w")
             f.write(text)
             f.close()
 
             editor = self.geteditor()
 
             util.system("%s \"%s\"" % (editor, name),
                         environ={'HGUSER': user},
                         onerr=util.Abort, errprefix=_("edit failed"))
 
             f = open(name)
             t = f.read()
             f.close()
         finally:
             os.unlink(name)
 
         return t
 
     def traceback(self):
         '''print exception traceback if traceback printing enabled.
         only to call in exception handler. returns true if traceback
         printed.'''
         if self._traceback:
             traceback.print_exc()
         return self._traceback
 
     def geteditor(self):
         '''return editor to use'''
         return (os.environ.get("HGEDITOR") or
                 self.config("ui", "editor") or
                 os.environ.get("VISUAL") or
                 os.environ.get("EDITOR", "vi"))
 
     def progress(self, topic, pos, item="", unit="", total=None):
         '''show a progress message
 
         With stock hg, this is simply a debug message that is hidden
         by default, but with extensions or GUI tools it may be
         visible. 'topic' is the current operation, 'item' is a
         non-numeric marker of the current position (ie the currently
         in-process file), 'pos' is the current numeric position (ie
         revision, bytes, etc.), units is a corresponding unit label,
         and total is the highest expected pos.
 
         Multiple nested topics may be active at a time. All topics
         should be marked closed by setting pos to None at termination.
         '''
 
         if pos == None or not self.debugflag:
             return
 
         if units:
             units = ' ' + units
         if item:
             item = ' ' + item
 
         if total:
             pct = 100.0 * pos / total
             ui.debug('%s:%s %s/%s%s (%4.2g%%)\n'
                      % (topic, item, pos, total, units, pct))
         else:
             ui.debug('%s:%s %s%s\n' % (topic, item, pos, units))