If the destination directory doesn't contain the desired repo, delete it

gclient_scm.py

 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """Gclient-specific SCM-specific operations."""
 
 import collections
 import logging
 import os
 import posixpath
@@ skipping 204 matching lines @@
       result, version = scm.GIT.AssertVersion('1.7')
       if not result:
         raise gclient_utils.Error('Git version is older than 1.7: %s' % version)
       return result
     except OSError:
       return False
 
   def GetCheckoutRoot(self):
     return scm.GIT.GetCheckoutRoot(self.checkout_path)
 
+  def GetRemoteURL(self, options, cwd=None):
+    try:
+      return self._Capture(['config', 'remote.%s.url' % self.remote],
+                           cwd=cwd or self.checkout_path).rstrip()
+    except (OSError, subprocess2.CalledProcessError):
+      return None
+
   def GetRevisionDate(self, _revision):
     """Returns the given revision's date in ISO-8601 format (which contains the
     time zone)."""
     # TODO(floitsch): get the time-stamp of the given revision and not just the
     # time-stamp of the currently checked out revision.
     return self._Capture(['log', '-n', '1', '--format=%ai'])
 
   @staticmethod
   def cleanup(options, args, file_list):
     """'Cleanup' the repo.
@@ skipping 37 matching lines @@
       # Not a fatal error, or even very interesting in a non-git-submodule
       # world.  So just keep it quiet.
       pass
     try:
       gclient_utils.CheckCallAndFilter(cmd3, **kwargs)
     except subprocess2.CalledProcessError:
       gclient_utils.CheckCallAndFilter(cmd3 + ['always'], **kwargs)
 
     gclient_utils.CheckCallAndFilter(cmd4, **kwargs)
 
-  def _FetchAndReset(self, revision, file_list, options):
-    """Equivalent to git fetch; git reset."""
-    quiet = []
-    if not options.verbose:
-      quiet = ['--quiet']
-    self._UpdateBranchHeads(options, fetch=False)
-
-    fetch_cmd = [
-      '-c', 'core.deltaBaseCacheLimit=2g', 'fetch', self.remote, '--prune']
-    self._Run(fetch_cmd + quiet, options, retry=True)
-    self._Run(['reset', '--hard', revision] + quiet, options)
-    self.UpdateSubmoduleConfig()
-    if file_list is not None:
-      files = self._Capture(['ls-files']).splitlines()
-      file_list.extend([os.path.join(self.checkout_path, f) for f in files])
-
   def update(self, options, args, file_list):
     """Runs git to update or transparently checkout the working copy.
 
     All updated files will be appended to file_list.
 
     Raises:
       Error: if can't get URL for relative path.
     """
     if args:
       raise gclient_utils.Error("Unsupported argument(s): %s" % ",".join(args))
@@ skipping 66 matching lines @@
       return self._Capture(['rev-parse', '--verify', 'HEAD'])
 
     if not os.path.exists(os.path.join(self.checkout_path, '.git')):
       raise gclient_utils.Error('\n____ %s%s\n'
                                 '\tPath is not a git repo. No .git dir.\n'
                                 '\tTo resolve:\n'
                                 '\t\trm -rf %s\n'
                                 '\tAnd run gclient sync again\n'
                                 % (self.relpath, rev_str, self.relpath))
 
-    # See if the url has changed (the unittests use git://foo for the url, let
-    # that through).
-    current_url = self._Capture(['config', 'remote.%s.url' % self.remote])
-    return_early = False
-    # TODO(maruel): Delete url != 'git://foo' since it's just to make the
-    # unit test pass. (and update the comment above)
-    # Skip url auto-correction if remote.origin.gclient-auto-fix-url is set.
-    # This allows devs to use experimental repos which have a different url
-    # but whose branch(s) are the same as official repos.
-    if (current_url != url and
-        url != 'git://foo' and
-        subprocess2.capture(
-            ['git', 'config', 'remote.%s.gclient-auto-fix-url' % self.remote],
-            cwd=self.checkout_path).strip() != 'False'):
-      print('_____ switching %s to a new upstream' % self.relpath)
-      # Make sure it's clean
-      self._CheckClean(rev_str)
-      # Switch over to the new upstream
-      self._Run(['remote', 'set-url', self.remote, url], options)
-      self._FetchAndReset(revision, file_list, options)
-      return_early = True
-
     # Need to do this in the normal path as well as in the post-remote-switch
     # path.
     self._PossiblySwitchCache(url, options)
 
-    if return_early:
-      return self._Capture(['rev-parse', '--verify', 'HEAD'])
-
     cur_branch = self._GetCurrentBranch()
 
     # Cases:
     # 0) HEAD is detached. Probably from our initial clone.
     #   - make sure HEAD is contained by a named ref, then update.
     # Cases 1-4. HEAD is a branch.
     # 1) current branch is not tracking a remote branch (could be git-svn)
     #   - try to rebase onto the new hash or branch
     # 2) current branch is tracking a remote branch with local committed
     #    changes, but the DEPS file switched to point to a hash
@@ skipping 385 matching lines @@
         if use_reference:
           cmd += ['--reference', os.path.abspath(self.checkout_path)]
 
         self._Run(cmd + [url, folder],
                   options, filter_fn=filter_fn, cwd=self.cache_dir, retry=True)
       else:
         # For now, assert that host/path/to/repo.git is identical. We may want
         # to relax this restriction in the future to allow for smarter cache
         # repo update schemes (such as pulling the same repo, but from a
         # different host).
-        existing_url = self._Capture(['config', 'remote.%s.url' % self.remote],
-                                     cwd=folder)
+        existing_url = self.GetRemoteURL(options, cwd=folder)
         assert self._NormalizeGitURL(existing_url) == self._NormalizeGitURL(url)
 
         if use_reference:
           with open(altfile, 'w') as f:
             f.write(os.path.abspath(checkout_objects))
 
         # Would normally use `git remote update`, but it doesn't support
         # --progress, so use fetch instead.
         self._Run(['fetch'] + v + ['--multiple', '--progress', '--all'],
                   options, filter_fn=filter_fn, cwd=folder, retry=True)
@@ skipping 248 matching lines @@
       result, version = scm.SVN.AssertVersion('1.4')
       if not result:
         raise gclient_utils.Error('SVN version is older than 1.4: %s' % version)
       return result
     except OSError:
       return False
 
   def GetCheckoutRoot(self):
     return scm.SVN.GetCheckoutRoot(self.checkout_path)
 
+  def GetRemoteURL(self, options):
+    try:
+      local_info = scm.SVN.CaptureLocalInfo([os.curdir], self.checkout_path)
+    except (OSError, subprocess2.CalledProcessError):
+      return None
+    return local_info.get('URL')
+
   def GetRevisionDate(self, revision):
     """Returns the given revision's date in ISO-8601 format (which contains the
     time zone)."""
     date = scm.SVN.Capture(
         ['propget', '--revprop', 'svn:date', '-r', revision],
         os.path.join(self.checkout_path, '.'))
     return date.strip()
 
   def cleanup(self, options, args, _file_list):
     """Cleanup working copy."""
@@ skipping 19 matching lines @@
         filter_fn=SvnDiffFilterer(self.relpath).Filter)
 
   def update(self, options, args, file_list):
     """Runs svn to update or transparently checkout the working copy.
 
     All updated files will be appended to file_list.
 
     Raises:
       Error: if can't get URL for relative path.
     """
     # Only update if git or hg is not controlling the directory.

[M-A Ruel, 2014/01/10 16:41:47]

Do you want to keep these? From what I understand, it is becoming a no-op since
the check is done *before* this function is being called.

[borenet, 2014/01/10 19:27:23]

It seems that these checks shouldn't be needed, but I'm a little scared to touch
much more of this.

[M-A Ruel, 2014/01/13 17:31:38]

I'd recommend to explicitly test it.

Create a dummy svn .gclient, checkout a git repository instead and gclient sync
to see what happens. If this code is not triggered, remove it.

Technically, this should be a proper test but I'll defer if this is a
requirement to Robbie.

[borenet, 2014/01/13 18:17:57]

I tested this out locally, and as expected the checkout was either removed
(using --force) or an error was thrown before we get to this point.  Removed
these sections in the latest patch set.

     git_path = os.path.join(self.checkout_path, '.git')
     if os.path.exists(git_path):
       print('________ found .git directory; skipping %s' % self.relpath)
       return
 
     hg_path = os.path.join(self.checkout_path, '.hg')

[M-A Ruel, 2014/01/10 16:41:47]

https://codereview.chromium.org/3601007 is opaque to me. I don't know if is it
still desired.

[borenet, 2014/01/10 19:27:23]

I defer to someone else's judgment.

     if os.path.exists(hg_path):
       print('________ found .hg directory; skipping %s' % self.relpath)
       return
 
     if args:
       raise gclient_utils.Error("Unsupported argument(s): %s" % ",".join(args))
 
     # revision is the revision to match. It is None if no revision is specified,
     # i.e. the 'deps ain't pinned'.
     url, revision = gclient_utils.SplitUrlRevision(self.url)
-    # Keep the original unpinned url for reference in case the repo is switched.
-    base_url = url
     managed = True
     if options.revision:
       # Override the revision number.
       revision = str(options.revision)
     if revision:
       if revision != 'unmanaged':
         forced_revision = True
         # Reconstruct the url.
         url = '%s@%s' % (url, revision)
         rev_str = ' at %s' % revision
       else:
         managed = False
         revision = None
     else:
       forced_revision = False
       rev_str = ''
 
     # Get the existing scm url and the revision number of the current checkout.
     exists = os.path.exists(self.checkout_path)
     if exists and managed:
       try:
         from_info = scm.SVN.CaptureLocalInfo(

[M-A Ruel, 2014/01/10 16:41:47]

From what I understand, this code is not necessary anymore?

[borenet, 2014/01/10 19:27:23]

I guess I'm not familiar enough with this code to know, but it looks like this
is used quite a bit below...

[M-A Ruel, 2014/01/13 17:31:38]

Why not make an assert at the top to ensure the checkout is in a predetermined
state before even calling this function?

             [], os.path.join(self.checkout_path, '.'))
       except (gclient_utils.Error, subprocess2.CalledProcessError):
         if options.reset and options.delete_unversioned_trees:
           print 'Removing troublesome path %s' % self.checkout_path
           gclient_utils.rmtree(self.checkout_path)
           exists = False
         else:
           msg = ('Can\'t update/checkout %s if an unversioned directory is '
                  'present. Delete the directory and try again.')
           raise gclient_utils.Error(msg % self.checkout_path)
@@ skipping 64 matching lines @@
       command = ['checkout', url, self.checkout_path]
       command = self._AddAdditionalUpdateFlags(command, options, revision)
       self._RunAndGetFileList(command, options, file_list, self._root_dir)
       return self.Svnversion()
 
     if not managed:
       print ('________ unmanaged solution; skipping %s' % self.relpath)
       return self.Svnversion()
 
     if 'URL' not in from_info:
       raise gclient_utils.Error(

[M-A Ruel, 2014/01/13 17:31:38]

I also assume this cannot happen anymore.

           ('gclient is confused. Couldn\'t get the url for %s.\n'
            'Try using @unmanaged.\n%s') % (
             self.checkout_path, from_info))
 
     # Look for locked directories.
     dir_info = scm.SVN.CaptureStatus(
         None, os.path.join(self.checkout_path, '.'))
     if any(d[0][2] == 'L' for d in dir_info):
       try:
         self._Run(['cleanup', self.checkout_path], options)
@@ skipping 19 matching lines @@
               if d[0][0] == '!':
                 print 'You can pass --force to enable automatic removal.'
               raise e
 
     # Retrieve the current HEAD version because svn is slow at null updates.
     if options.manually_grab_svn_rev and not revision:
       from_info_live = scm.SVN.CaptureRemoteInfo(from_info['URL'])
       revision = str(from_info_live['Revision'])
       rev_str = ' at %s' % revision
 
-    if from_info['URL'] != base_url:
-      # The repository url changed, need to switch.
-      try:
-        to_info = scm.SVN.CaptureRemoteInfo(url)
-      except (gclient_utils.Error, subprocess2.CalledProcessError):
-        # The url is invalid or the server is not accessible, it's safer to bail
-        # out right now.
-        raise gclient_utils.Error('This url is unreachable: %s' % url)
-      can_switch = ((from_info['Repository Root'] != to_info['Repository Root'])
-                    and (from_info['UUID'] == to_info['UUID']))
-      if can_switch:
-        print('\n_____ relocating %s to a new checkout' % self.relpath)
-        # We have different roots, so check if we can switch --relocate.
-        # Subversion only permits this if the repository UUIDs match.
-        # Perform the switch --relocate, then rewrite the from_url
-        # to reflect where we "are now."  (This is the same way that
-        # Subversion itself handles the metadata when switch --relocate
-        # is used.)  This makes the checks below for whether we
-        # can update to a revision or have to switch to a different
-        # branch work as expected.
-        # TODO(maruel):  TEST ME !
-        command = ['switch', '--relocate',
-                   from_info['Repository Root'],
-                   to_info['Repository Root'],
-                   self.relpath]
-        self._Run(command, options, cwd=self._root_dir)
-        from_info['URL'] = from_info['URL'].replace(
-            from_info['Repository Root'],
-            to_info['Repository Root'])
-      else:
-        if not options.force and not options.reset:
-          # Look for local modifications but ignore unversioned files.
-          for status in scm.SVN.CaptureStatus(None, self.checkout_path):
-            if status[0][0] != '?':
-              raise gclient_utils.Error(
-                  ('Can\'t switch the checkout to %s; UUID don\'t match and '
-                   'there is local changes in %s. Delete the directory and '
-                   'try again.') % (url, self.checkout_path))
-        # Ok delete it.
-        print('\n_____ switching %s to a new checkout' % self.relpath)
-        gclient_utils.rmtree(self.checkout_path)
-        # We need to checkout.
-        command = ['checkout', url, self.checkout_path]
-        command = self._AddAdditionalUpdateFlags(command, options, revision)
-        self._RunAndGetFileList(command, options, file_list, self._root_dir)
-        return self.Svnversion()
-
     # If the provided url has a revision number that matches the revision
     # number of the existing directory, then we don't need to bother updating.
     if not options.force and str(from_info['Revision']) == revision:
       if options.verbose or not forced_revision:
         print('\n_____ %s%s' % (self.relpath, rev_str))
     else:
       command = ['update', self.checkout_path]
       command = self._AddAdditionalUpdateFlags(command, options, revision)
       self._RunAndGetFileList(command, options, file_list, self._root_dir)
 
@@ skipping 169 matching lines @@
       new_command.append('--force')
       if command[0] != 'checkout' and scm.SVN.AssertVersion('1.6')[0]:
         new_command.extend(('--accept', 'theirs-conflict'))
     elif options.manually_grab_svn_rev:
       new_command.append('--force')
       if command[0] != 'checkout' and scm.SVN.AssertVersion('1.6')[0]:
         new_command.extend(('--accept', 'postpone'))
     elif command[0] != 'checkout' and scm.SVN.AssertVersion('1.6')[0]:
       new_command.extend(('--accept', 'postpone'))
     return new_command