Changeset - 59456e7a12b6
[Not reviewed]
default
0 2 0
Mads Kiilerich - 6 years ago 2019-12-31 13:23:19
mads@kiilerich.com
Grafted from: 3b391e655755
vcs: change get_diff to use _run_git_command directly

This will allow run_git_command to return strings while get_diff still return
raw bytes.
2 files changed with 22 insertions and 23 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/git/repository.py
Show inline comments
 
@@ -541,194 +541,193 @@ class GitRepository(BaseRepository):
 
        if start_date:
 
            cmd += ['--since', start_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if end_date:
 
            cmd += ['--until', end_date.strftime('%m/%d/%y %H:%M:%S')]
 
        if branch_name:
 
            cmd.append(branch_name)
 
        else:
 
            cmd.append(settings.GIT_REV_FILTER)
 

	
 
        revs = self.run_git_command(cmd)[0].splitlines()
 
        start_pos = 0
 
        end_pos = len(revs)
 
        if start:
 
            _start = self._get_revision(start)
 
            try:
 
                start_pos = revs.index(_start)
 
            except ValueError:
 
                pass
 

	
 
        if end is not None:
 
            _end = self._get_revision(end)
 
            try:
 
                end_pos = revs.index(_end)
 
            except ValueError:
 
                pass
 

	
 
        if None not in [start, end] and start_pos > end_pos:
 
            raise RepositoryError('start cannot be after end')
 

	
 
        if end_pos is not None:
 
            end_pos += 1
 

	
 
        revs = revs[start_pos:end_pos]
 
        if reverse:
 
            revs.reverse()
 

	
 
        return CollectionGenerator(self, revs)
 

	
 
    def get_diff(self, rev1, rev2, path=None, ignore_whitespace=False,
 
                 context=3):
 
        """
 
        Returns (git like) *diff*, as plain text. Shows changes introduced by
 
        ``rev2`` since ``rev1``.
 

	
 
        :param rev1: Entry point from which diff is shown. Can be
 
          ``self.EMPTY_CHANGESET`` - in this case, patch showing all
 
          the changes since empty state of the repository until ``rev2``
 
        :param rev2: Until which revision changes should be shown.
 
        :param ignore_whitespace: If set to ``True``, would not show whitespace
 
          changes. Defaults to ``False``.
 
        :param context: How many lines before/after changed lines should be
 
          shown. Defaults to ``3``. Due to limitations in Git, if
 
          value passed-in is greater than ``2**31-1``
 
          (``2147483647``), it will be set to ``2147483647``
 
          instead. If negative value is passed-in, it will be set to
 
          ``0`` instead.
 
        """
 

	
 
        # Git internally uses a signed long int for storing context
 
        # size (number of lines to show before and after the
 
        # differences). This can result in integer overflow, so we
 
        # ensure the requested context is smaller by one than the
 
        # number that would cause the overflow. It is highly unlikely
 
        # that a single file will contain that many lines, so this
 
        # kind of change should not cause any realistic consequences.
 
        overflowed_long_int = 2**31
 

	
 
        if context >= overflowed_long_int:
 
            context = overflowed_long_int - 1
 

	
 
        # Negative context values make no sense, and will result in
 
        # errors. Ensure this does not happen.
 
        if context < 0:
 
            context = 0
 

	
 
        flags = ['-U%s' % context, '--full-index', '--binary', '-p', '-M', '--abbrev=40']
 
        if ignore_whitespace:
 
            flags.append('-w')
 

	
 
        if hasattr(rev1, 'raw_id'):
 
            rev1 = getattr(rev1, 'raw_id')
 

	
 
        if hasattr(rev2, 'raw_id'):
 
            rev2 = getattr(rev2, 'raw_id')
 

	
 
        if rev1 == self.EMPTY_CHANGESET:
 
            rev2 = self.get_changeset(rev2).raw_id
 
            cmd = ['show'] + flags + [rev2]
 
        else:
 
            rev1 = self.get_changeset(rev1).raw_id
 
            rev2 = self.get_changeset(rev2).raw_id
 
            cmd = ['diff'] + flags + [rev1, rev2]
 

	
 
        if path:
 
            cmd += ['--', path]
 

	
 
        stdout, stderr = self.run_git_command(cmd)
 
        # TODO: don't ignore stderr
 
        stdout, stderr = self._run_git_command(cmd, cwd=self.path)
 
        # If we used 'show' command, strip first few lines (until actual diff
 
        # starts)
 
        if rev1 == self.EMPTY_CHANGESET:
 
            parts = stdout.split('\ndiff ', 1)
 
            if len(parts) > 1:
 
                stdout = 'diff ' + parts[1]
 
        return stdout
 

	
 
    @LazyProperty
 
    def in_memory_changeset(self):
 
        """
 
        Returns ``GitInMemoryChangeset`` object for this repository.
 
        """
 
        return GitInMemoryChangeset(self)
 

	
 
    def clone(self, url, update_after_clone=True, bare=False):
 
        """
 
        Tries to clone changes from external location.
 

	
 
        :param update_after_clone: If set to ``False``, git won't checkout
 
          working directory
 
        :param bare: If set to ``True``, repository would be cloned into
 
          *bare* git repository (no working directory at all).
 
        """
 
        url = self._get_url(url)
 
        cmd = ['clone', '-q']
 
        if bare:
 
            cmd.append('--bare')
 
        elif not update_after_clone:
 
            cmd.append('--no-checkout')
 
        cmd += ['--', url, self.path]
 
        # If error occurs run_git_command raises RepositoryError already
 
        self.run_git_command(cmd)
 

	
 
    def pull(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        url = self._get_url(url)
 
        cmd = ['pull', '--ff-only', url]
 
        # If error occurs run_git_command raises RepositoryError already
 
        self.run_git_command(cmd)
 

	
 
    def fetch(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        url = self._get_url(url)
 
        so, se = self.run_git_command(['ls-remote', '-h', url])
 
        cmd = ['fetch', url, '--']
 
        for line in (x for x in so.splitlines()):
 
            sha, ref = line.split('\t')
 
            cmd.append('+%s:%s' % (ref, ref))
 
        self.run_git_command(cmd)
 

	
 
    def _update_server_info(self):
 
        """
 
        runs gits update-server-info command in this repo instance
 
        """
 
        from dulwich.server import update_server_info
 
        try:
 
            update_server_info(self._repo)
 
        except OSError as e:
 
            if e.errno not in [errno.ENOENT, errno.EROFS]:
 
                raise
 
            # Workaround for dulwich crashing on for example its own dulwich/tests/data/repos/simple_merge.git/info/refs.lock
 
            log.error('Ignoring %s running update-server-info: %s', type(e).__name__, e)
 

	
 
    @LazyProperty
 
    def workdir(self):
 
        """
 
        Returns ``Workdir`` instance for this repository.
 
        """
 
        return GitWorkdir(self)
 

	
 
    def get_config_value(self, section, name, config_file=None):
 
        """
 
        Returns configuration value for a given [``section``] and ``name``.
 

	
 
        :param section: Section we want to retrieve value from
 
        :param name: Name of configuration we want to retrieve
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        if config_file is None:
 
            config_file = []
 
        elif isinstance(config_file, basestring):
 
            config_file = [config_file]
 

	
 
        def gen_configs():
 
            for path in config_file + self._config_files:
 
                try:
 
                    yield ConfigFile.from_path(path)
 
                except (IOError, OSError, ValueError):
 
                    continue
 

	
kallithea/tests/vcs/test_git.py
Show inline comments
 
@@ -572,251 +572,251 @@ class TestGitChangeset(object):
 
        assert node.not_changed
 
        assert not node.removed
 

	
 
        # If node has REMOVED state then trying to fetch it would raise
 
        # ChangesetError exception
 
        chset = self.repo.get_changeset(
 
            'fa6600f6848800641328adbf7811fd2372c02ab2')
 
        path = 'vcs/backends/BaseRepository.py'
 
        with pytest.raises(NodeDoesNotExistError):
 
            chset.get_node(path)
 
        # but it would be one of ``removed`` (changeset's attribute)
 
        assert path in [rf.path for rf in chset.removed]
 

	
 
        chset = self.repo.get_changeset(
 
            '54386793436c938cff89326944d4c2702340037d')
 
        changed = ['setup.py', 'tests/test_nodes.py', 'vcs/backends/hg.py',
 
            'vcs/nodes.py']
 
        assert set(changed) == set([f.path for f in chset.changed])
 

	
 
    def test_commit_message_is_unicode(self):
 
        for cs in self.repo:
 
            assert isinstance(cs.message, unicode)
 

	
 
    def test_changeset_author_is_unicode(self):
 
        for cs in self.repo:
 
            assert isinstance(cs.author, unicode)
 

	
 
    def test_repo_files_content_is_unicode(self):
 
        changeset = self.repo.get_changeset()
 
        for node in changeset.get_node('/'):
 
            if node.is_file():
 
                assert isinstance(node.content, unicode)
 

	
 
    def test_wrong_path(self):
 
        # There is 'setup.py' in the root dir but not there:
 
        path = 'foo/bar/setup.py'
 
        tip = self.repo.get_changeset()
 
        with pytest.raises(VCSError):
 
            tip.get_node(path)
 

	
 
    def test_author_email(self):
 
        assert 'marcin@python-blog.com' == self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3').author_email
 
        assert 'lukasz.balcerzak@python-center.pl' == self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b').author_email
 
        assert '' == self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992').author_email
 

	
 
    def test_author_username(self):
 
        assert 'Marcin Kuzminski' == self.repo.get_changeset('c1214f7e79e02fc37156ff215cd71275450cffc3').author_name
 
        assert 'Lukasz Balcerzak' == self.repo.get_changeset('ff7ca51e58c505fec0dd2491de52c622bb7a806b').author_name
 
        assert 'marcink none@none' == self.repo.get_changeset('8430a588b43b5d6da365400117c89400326e7992').author_name
 

	
 

	
 
class TestGitSpecificWithRepo(_BackendTestMixin):
 
    backend_alias = 'git'
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        return [
 
            {
 
                'message': 'Initial',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 20),
 
                'added': [
 
                    FileNode('foobar/static/js/admin/base.js', content='base'),
 
                    FileNode('foobar/static/admin', content='admin',
 
                        mode=0o120000), # this is a link
 
                    FileNode('foo', content='foo'),
 
                ],
 
            },
 
            {
 
                'message': 'Second',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 22),
 
                'added': [
 
                    FileNode('foo2', content='foo2'),
 
                ],
 
            },
 
        ]
 

	
 
    def test_paths_slow_traversing(self):
 
        cs = self.repo.get_changeset()
 
        assert cs.get_node('foobar').get_node('static').get_node('js').get_node('admin').get_node('base.js').content == 'base'
 

	
 
    def test_paths_fast_traversing(self):
 
        cs = self.repo.get_changeset()
 
        assert cs.get_node('foobar/static/js/admin/base.js').content == 'base'
 

	
 
    def test_workdir_get_branch(self):
 
        self.repo.run_git_command(['checkout', '-b', 'production'])
 
        # Regression test: one of following would fail if we don't check
 
        # .git/HEAD file
 
        self.repo.run_git_command(['checkout', 'production'])
 
        assert self.repo.workdir.get_branch() == 'production'
 
        self.repo.run_git_command(['checkout', 'master'])
 
        assert self.repo.workdir.get_branch() == 'master'
 

	
 
    def test_get_diff_runs_git_command_with_hashes(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1)
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1)])
 
             self.repo._get_revision(0), self.repo._get_revision(1)], cwd=self.repo.path)
 

	
 
    def test_get_diff_runs_git_command_with_str_hashes(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(self.repo.EMPTY_CHANGESET, 1)
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['show', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(1)])
 
             self.repo._get_revision(1)], cwd=self.repo.path)
 

	
 
    def test_get_diff_runs_git_command_with_path_if_its_given(self):
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo')
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'], cwd=self.repo.path)
 

	
 
    def test_get_diff_does_not_sanitize_valid_context(self):
 
        almost_overflowed_long_int = 2**31-1
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=almost_overflowed_long_int)
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(almost_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'], cwd=self.repo.path)
 

	
 
    def test_get_diff_sanitizes_overflowing_context(self):
 
        overflowed_long_int = 2**31
 
        sanitized_overflowed_long_int = overflowed_long_int-1
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=overflowed_long_int)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(sanitized_overflowed_long_int), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'], cwd=self.repo.path)
 

	
 
    def test_get_diff_does_not_sanitize_zero_context(self):
 
        zero_context = 0
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=zero_context)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U' + str(zero_context), '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'], cwd=self.repo.path)
 

	
 
    def test_get_diff_sanitizes_negative_context(self):
 
        negative_context = -10
 

	
 
        self.repo.run_git_command = mock.Mock(return_value=['', ''])
 
        self.repo._run_git_command = mock.Mock(return_value=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=negative_context)
 

	
 
        self.repo.run_git_command.assert_called_once_with(
 
        self.repo._run_git_command.assert_called_once_with(
 
            ['diff', '-U0', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'])
 
             self.repo._get_revision(0), self.repo._get_revision(1), '--', 'foo'], cwd=self.repo.path)
 

	
 

	
 
class TestGitRegression(_BackendTestMixin):
 
    backend_alias = 'git'
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        return [
 
            {
 
                'message': 'Initial',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 20),
 
                'added': [
 
                    FileNode('bot/__init__.py', content='base'),
 
                    FileNode('bot/templates/404.html', content='base'),
 
                    FileNode('bot/templates/500.html', content='base'),
 
                ],
 
            },
 
            {
 
                'message': 'Second',
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 22),
 
                'added': [
 
                    FileNode('bot/build/migrations/1.py', content='foo2'),
 
                    FileNode('bot/build/migrations/2.py', content='foo2'),
 
                    FileNode('bot/build/static/templates/f.html', content='foo2'),
 
                    FileNode('bot/build/static/templates/f1.html', content='foo2'),
 
                    FileNode('bot/build/templates/err.html', content='foo2'),
 
                    FileNode('bot/build/templates/err2.html', content='foo2'),
 
                ],
 
            },
 
        ]
 

	
 
    def test_similar_paths(self):
 
        cs = self.repo.get_changeset()
 
        paths = lambda *n: [x.path for x in n]
 
        assert paths(*cs.get_nodes('bot')) == ['bot/build', 'bot/templates', 'bot/__init__.py']
 
        assert paths(*cs.get_nodes('bot/build')) == ['bot/build/migrations', 'bot/build/static', 'bot/build/templates']
 
        assert paths(*cs.get_nodes('bot/build/static')) == ['bot/build/static/templates']
 
        # this get_nodes below causes troubles !
 
        assert paths(*cs.get_nodes('bot/build/static/templates')) == ['bot/build/static/templates/f.html', 'bot/build/static/templates/f1.html']
 
        assert paths(*cs.get_nodes('bot/build/templates')) == ['bot/build/templates/err.html', 'bot/build/templates/err2.html']
 
        assert paths(*cs.get_nodes('bot/templates/')) == ['bot/templates/404.html', 'bot/templates/500.html']
 

	
 

	
 
class TestGitHooks(object):
 
    """
 
    Tests related to hook functionality of Git repositories.
 
    """
 

	
 
    def setup_method(self):
 
        # For each run we want a fresh repo.
 
        self.repo_directory = get_new_dir("githookrepo")
 
        self.repo = GitRepository(self.repo_directory, create=True)
 

	
 
        # Create a dictionary where keys are hook names, and values are paths to
 
        # them. Deduplicates code in tests a bit.
 
        self.hook_directory = self.repo.get_hook_location()
 
        self.kallithea_hooks = dict((h, os.path.join(self.hook_directory, h)) for h in ("pre-receive", "post-receive"))
 

	
 
    def test_hooks_created_if_missing(self):
 
        """
 
        Tests if hooks are installed in repository if they are missing.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            if os.path.exists(hook_path):
 
                os.remove(hook_path)
 

	
 
        ScmModel().install_git_hooks(repo=self.repo)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            assert os.path.exists(hook_path)
 

	
 
    def test_kallithea_hooks_updated(self):
 
        """
 
        Tests if hooks are updated if they are Kallithea hooks already.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
 
                f.write("KALLITHEA_HOOK_VER=0.0.0\nJUST_BOGUS")
 

	
 
        ScmModel().install_git_hooks(repo=self.repo)
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path) as f:
 
                assert "JUST_BOGUS" not in f.read()
 

	
 
    def test_custom_hooks_untouched(self):
 
        """
 
        Tests if hooks are left untouched if they are not Kallithea hooks.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
0 comments (0 inline, 0 general)