Reland r245404 ("If the destination directory doesn't contain the desired repo, delete it")

BUG=

Review URL: https://codereview.chromium.org/131743022

git-svn-id: svn://svn.chromium.org/chrome/trunk/tools/depot_tools@249143 0039d316-1c4b-4281-b951-d872f2087c98

gclient.py

@@ -87,6 +87,7 @@ import platform
 import posixpath
 import pprint
 import re
+import socket
 import sys
 import time
 import urllib
@@ -668,6 +669,40 @@ class Dependency(gclient_utils.WorkItem, DependencySettings):
             command, options, parsed_url, self.parent.name, revision_overrides)
         self._used_scm = gclient_scm.CreateSCM(
             parsed_url, self.root.root_dir, self.name)
+
+        def enable_deletion_of_conflicting_checkouts():
+          """Determines whether to enable new checkout deletion behavior.
+
+          Initially, enables the experimental functionality on a small set of
+          bots.
+          """
+          # TODO(borenet): Remove this hack as soon as we've verified that it
+          # doesn't cause the bots to break.
+          return (os.environ.get('CHROME_HEADLESS') and
+                  socket.gethostname() in ('vm859-m1', 'build1-m1', 'vm630-m1'))
+
+        # When updating, determine whether the destination directory contains a
+        # checkout of the desired repository. If not, avoid conflicts by
+        # deleting the directory before running the update.
+        if command == 'update' and enable_deletion_of_conflicting_checkouts():
+          logging.warning('Experimental deletion of mismatching checkouts '
+                          'enabled.')
+          actual_remote_url = self._used_scm.GetRemoteURL(options)
+          url, _ = gclient_utils.SplitUrlRevision(parsed_url)
+          url = url.rstrip('/')
+          dest_dir = os.path.join(self.root.root_dir, self.name)
+          if os.path.isdir(dest_dir) and actual_remote_url != url:
+            if options.force:
+              logging.warning('%s does not contain a checkout of %s. Removing '
+                              ' %s' % (dest_dir, url, dest_dir))
+              gclient_utils.rmtree(dest_dir)
+            else:
+              raise gclient_utils.Error('%s does not contain a checkout of %s '
+                                        '(found %s instead). Please fix the '
+                                        'solution manually or run with --force '
+                                        'to delete automatically.' % (
+                                            dest_dir, url, actual_remote_url))
+
         self._got_revision = self._used_scm.RunCommand(command, options, args,
                                                        file_list)
         if file_list:

gclient_scm.py

@@ -210,6 +210,21 @@ class GitWrapper(SCMWrapper):
   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):
+      pass
+    try:
+      # This may occur if we have a git-svn checkout.
+      if scm.GIT.IsGitSvn(os.curdir):
+        return scm.GIT.Capture(['config', '--local', '--get',
+                                'svn-remote.svn.url'], os.curdir).rstrip()
+    except (OSError, subprocess2.CalledProcessError):
+      pass
+    return None
+
   def GetRevisionDate(self, _revision):
     """Returns the given revision's date in ISO-8601 format (which contains the
     time zone)."""
@@ -265,22 +280,6 @@ class GitWrapper(SCMWrapper):
 
     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.
 
@@ -367,35 +366,10 @@ class GitWrapper(SCMWrapper):
                                 '\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:
@@ -808,8 +782,7 @@ class GitWrapper(SCMWrapper):
         # 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:
@@ -1098,6 +1071,22 @@ class SVNWrapper(SCMWrapper):
   def GetCheckoutRoot(self):
     return scm.SVN.GetCheckoutRoot(self.checkout_path)
 
+  def GetRemoteURL(self, options):
+    try:
+      return scm.SVN.CaptureLocalInfo([os.curdir],
+                                      self.checkout_path).get('URL')
+    except (OSError, subprocess2.CalledProcessError):
+      pass
+    try:
+      # This may occur if we have a git-svn checkout.
+      if scm.GIT.IsGitSvn(os.curdir):
+        return scm.GIT.Capture(['config', '--local', '--get',
+                                'svn-remote.svn.url'], os.curdir).rstrip()
+    except (OSError, subprocess2.CalledProcessError):
+      pass
+    return None
+
+
   def GetRevisionDate(self, revision):
     """Returns the given revision's date in ISO-8601 format (which contains the
     time zone)."""
@@ -1137,15 +1126,10 @@ class SVNWrapper(SCMWrapper):
     Raises:
       Error: if can't get URL for relative path.
     """
-    # Only update if git or hg is not controlling the directory.
-    git_path = os.path.join(self.checkout_path, '.git')
-    if os.path.exists(git_path):
-      print('________ found .git directory; skipping %s' % self.relpath)
-      return
+    exists = os.path.exists(self.checkout_path)
 
-    hg_path = os.path.join(self.checkout_path, '.hg')
-    if os.path.exists(hg_path):
-      print('________ found .hg directory; skipping %s' % self.relpath)
+    if exists and scm.GIT.IsGitSvn(self.checkout_path):
+      print '________ %s looks like git-svn; skipping.' % self.relpath
       return
 
     if args:
@@ -1154,8 +1138,6 @@ class SVNWrapper(SCMWrapper):
     # 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.
@@ -1174,7 +1156,6 @@ class SVNWrapper(SCMWrapper):
       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(
@@ -1310,53 +1291,6 @@ class SVNWrapper(SCMWrapper):
       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:
@@ -1423,11 +1357,8 @@ class SVNWrapper(SCMWrapper):
       return self.update(options, [], file_list)
 
     if not os.path.isdir(os.path.join(self.checkout_path, '.svn')):
-      if os.path.isdir(os.path.join(self.checkout_path, '.git')):
-        print('________ found .git directory; skipping %s' % self.relpath)
-        return
-      if os.path.isdir(os.path.join(self.checkout_path, '.hg')):
-        print('________ found .hg directory; skipping %s' % self.relpath)
+      if scm.GIT.IsGitSvn(self.checkout_path):
+        print '________ %s looks like git-svn; skipping.' % self.relpath
         return
       if not options.force:
         raise gclient_utils.Error('Invalid checkout path, aborting')

tests/gclient_smoketest.py

@@ -13,6 +13,7 @@ This test assumes GClientSmokeBase.URL_BASE is valid.
 import logging
 import os
 import re
+import socket
 import subprocess
 import sys
 import unittest
@@ -109,9 +110,9 @@ class GClientSmokeBase(FakeReposTestBase):
               not re.match(
                   r'_____ [^ ]+ : Attempting rebase onto [0-9a-f]+...',
                   line) and
-              not re.match(r'_____ [^ ]+ at [^ ]+', line)):
-            # The two regexp above are a bit too broad, they are necessary only
-            # for git checkouts.
+              not re.match(r'_____ [^ ]+ at [^ ]+', line) and not
+              re.match(r'________ (.*) looks like git-svn; skipping.', line)):
+            # The regexp above are a bit too broad.
             self.fail(line)
         else:
           results.append([[match.group(1), match.group(2), match.group(3)]])
@@ -776,7 +777,42 @@ class GClientSmokeSVN(GClientSmokeBase):
 
     # Cripple src/third_party/foo and make sure gclient still succeeds.
     gclient_utils.rmtree(join(third_party, 'foo', '.svn'))
-    self.assertEquals(0, self.gclient(cmd)[-1])
+    self.assertEquals(0, self.gclient(cmd + ['--force'])[-1])
+
+  def testSkipGitSvn(self):
+    # Check that gclient skips git-svn checkouts.
+    if not self.enabled:
+      return
+
+    # Create the .gclient file.
+    svn_url = self.svn_base + 'trunk/src'
+    self.gclient(['config', svn_url], cwd=self.root_dir)
+
+    # Create a git-svn checkout.
+    subprocess2.check_call(['git', 'svn', 'clone', svn_url], cwd=self.root_dir)
+
+    # Ensure that gclient skips the git-svn checkout.
+    stdout, stderr, rc = self.gclient(['sync', '--jobs', '1'])
+    self.assertEquals(rc, 0)
+    self.assertFalse(stderr)
+    self.assertTrue('________ src looks like git-svn; skipping.' in stdout)
+    self.checkBlock(stdout, [
+        ['running', self.root_dir],
+        ['running', os.path.join(self.root_dir, 'src', 'file', 'other')],
+        ['running', self.root_dir],
+        ['running', self.root_dir],
+        ['running', self.root_dir],
+        ['running', self.root_dir],
+    ])
+
+    # But, we still need the DEPS to be checked out...
+    foo_dir = os.path.join(self.root_dir, 'src', 'third_party', 'foo')
+    foo_rev = subprocess2.check_output(['svnversion', foo_dir]).strip()
+    self.assertEquals(foo_rev, '1')
+
+    other_dir = os.path.join(self.root_dir, 'src', 'other')
+    other_rev = subprocess2.check_output(['svnversion', other_dir]).strip()
+    self.assertEquals(other_rev, '2')
 
 
 class GClientSmokeSVNTransitive(GClientSmokeBase):
@@ -1103,7 +1139,7 @@ class GClientSmokeGIT(GClientSmokeBase):
     self.assertTree(tree)
 
     # Pre-DEPS hooks run when syncing with --nohooks.
-    self.gclient(['sync', '--deps', 'mac', '--nohooks',
+    self.gclient(['sync', '--deps', 'mac', '--nohooks', '--force',
                   '--revision', 'src@' + self.githash('repo_5', 2)])
     tree = self.mangle_git_tree(('repo_5@2', 'src'),
                                 ('repo_1@2', 'src/repo1'),
@@ -1115,7 +1151,7 @@ class GClientSmokeGIT(GClientSmokeBase):
     os.remove(join(self.root_dir, 'src', 'git_pre_deps_hooked'))
 
     # Pre-DEPS hooks don't run with --noprehooks
-    self.gclient(['sync', '--deps', 'mac', '--noprehooks',
+    self.gclient(['sync', '--deps', 'mac', '--noprehooks', '--force',
                   '--revision', 'src@' + self.githash('repo_5', 2)])
     tree = self.mangle_git_tree(('repo_5@2', 'src'),
                                 ('repo_1@2', 'src/repo1'),
@@ -1343,6 +1379,81 @@ class GClientSmokeBoth(GClientSmokeBase):
 
     self.assertEquals(sorted(entries), sorted(expected))
 
+  # TODO(borenet): Enable this at the same time that the guard is removed in
+  # gclient.
+  if (os.environ.get('CHROME_HEADLESS') and
+      socket.gethostname() in ('vm859-m1', 'build1-m1', 'vm630-m1')):
+    def testDeleteConflictingCheckout(self):
+      if not self.enabled:
+        return
+  
+      # Create an initial svn checkout.
+      self.gclient(['config', '--spec',
+          'solutions=['
+          '{"name": "src",'
+          ' "url": "' + self.svn_base + 'trunk/src"},'
+          ']'
+      ])
+      results = self.gclient(['sync', '--deps', 'mac'])
+      self.assertEqual(results[2], 0, 'Sync failed!')
+  
+      # Verify that we have the expected svn checkout.
+      results = self.gclient(['revinfo', '--deps', 'mac'])
+      actual = results[0].splitlines()
+      expected = [
+        'src: %strunk/src' % self.svn_base,
+        'src/file/other: File("%strunk/other/DEPS")' % self.svn_base,
+        'src/other: %strunk/other' % self.svn_base,
+        'src/third_party/foo: %strunk/third_party/foo@1' % self.svn_base,
+      ]
+      self.assertEquals(actual, expected)
+  
+      # Change the desired checkout to git.
+      self.gclient(['config', '--spec',
+          'solutions=['
+          '{"name": "src",'
+          ' "url": "' + self.git_base + 'repo_1"},'
+          ']'
+      ])
+  
+      # Verify that the sync succeeds with --force.
+      results = self.gclient(['sync', '--deps', 'mac', '--force'])
+      self.assertEqual(results[2], 0, 'Sync failed!')
+  
+      # Verify that we got the desired git checkout.
+      results = self.gclient(['revinfo', '--deps', 'mac'])
+      actual = results[0].splitlines()
+      expected = [
+        'src: %srepo_1' % self.git_base,
+        'src/repo2: %srepo_2@%s' % (self.git_base,
+                                    self.githash('repo_2', 1)[:7]),
+        'src/repo2/repo_renamed: %srepo_3' % self.git_base,
+      ]
+      self.assertEquals(actual, expected)
+  
+      # Change the desired checkout back to svn.
+      self.gclient(['config', '--spec',
+          'solutions=['
+          '{"name": "src",'
+          ' "url": "' + self.svn_base + 'trunk/src"},'
+          ']'
+      ])
+  
+      # Verify that the sync succeeds.
+      results = self.gclient(['sync', '--deps', 'mac', '--force'])
+      self.assertEqual(results[2], 0, 'Sync failed!')
+  
+      # Verify that we have the expected svn checkout.
+      results = self.gclient(['revinfo', '--deps', 'mac'])
+      actual = results[0].splitlines()
+      expected = [
+        'src: %strunk/src' % self.svn_base,
+        'src/file/other: File("%strunk/other/DEPS")' % self.svn_base,
+        'src/other: %strunk/other' % self.svn_base,
+        'src/third_party/foo: %strunk/third_party/foo@1' % self.svn_base,
+      ]
+      self.assertEquals(actual, expected)
+
 
 class GClientSmokeFromCheckout(GClientSmokeBase):
   # WebKit abuses this. It has a .gclient and a DEPS from a checkout.