Changeset - 3725f86e8514
[Not reviewed]
default
0 2 0
domruf - 8 years ago 2017-10-17 19:27:10
dominikruf@gmail.com
api: wrap changeset file paths with safe_unicode

Fixes error in get_changesets if path contains non ascii character.
2 files changed with 13 insertions and 11 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/base.py
Show inline comments
 
@@ -300,195 +300,195 @@ class BaseRepository(object):
 

	
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        raise NotImplementedError
 

	
 
    def get_user_email(self, config_file=None):
 
        """
 
        Returns user's email from global configuration file.
 

	
 
        :param config_file: A path to file which should be used to retrieve
 
          configuration from (might also be a list of file paths)
 
        """
 
        raise NotImplementedError
 

	
 
    # =========== #
 
    # WORKDIR API #
 
    # =========== #
 

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

	
 

	
 
class BaseChangeset(object):
 
    """
 
    Each backend should implement it's changeset representation.
 

	
 
    **Attributes**
 

	
 
        ``repository``
 
            repository object within which changeset exists
 

	
 
        ``id``
 
            may be ``raw_id`` or i.e. for mercurial's tip just ``tip``
 

	
 
        ``raw_id``
 
            raw changeset representation (i.e. full 40 length sha for git
 
            backend)
 

	
 
        ``short_id``
 
            shortened (if apply) version of ``raw_id``; it would be simple
 
            shortcut for ``raw_id[:12]`` for git/mercurial backends or same
 
            as ``raw_id`` for subversion
 

	
 
        ``revision``
 
            revision number as integer
 

	
 
        ``files``
 
            list of ``FileNode`` (``Node`` with NodeKind.FILE) objects
 

	
 
        ``dirs``
 
            list of ``DirNode`` (``Node`` with NodeKind.DIR) objects
 

	
 
        ``nodes``
 
            combined list of ``Node`` objects
 

	
 
        ``author``
 
            author of the changeset, as unicode
 

	
 
        ``message``
 
            message of the changeset, as unicode
 

	
 
        ``parents``
 
            list of parent changesets
 

	
 
        ``last``
 
            ``True`` if this is last changeset in repository, ``False``
 
            otherwise; trying to access this attribute while there is no
 
            changesets would raise ``EmptyRepositoryError``
 
    """
 
    def __str__(self):
 
        return '<%s at %s:%s>' % (self.__class__.__name__, self.revision,
 
            self.short_id)
 

	
 
    def __repr__(self):
 
        return self.__str__()
 

	
 
    def __unicode__(self):
 
        return u'%s:%s' % (self.revision, self.short_id)
 

	
 
    def __eq__(self, other):
 
        return self.raw_id == other.raw_id
 

	
 
    def __json__(self, with_file_list=False):
 
        if with_file_list:
 
            return dict(
 
                short_id=self.short_id,
 
                raw_id=self.raw_id,
 
                revision=self.revision,
 
                message=self.message,
 
                date=self.date,
 
                author=self.author,
 
                added=[el.path for el in self.added],
 
                changed=[el.path for el in self.changed],
 
                removed=[el.path for el in self.removed],
 
                added=[safe_unicode(el.path) for el in self.added],
 
                changed=[safe_unicode(el.path) for el in self.changed],
 
                removed=[safe_unicode(el.path) for el in self.removed],
 
            )
 
        else:
 
            return dict(
 
                short_id=self.short_id,
 
                raw_id=self.raw_id,
 
                revision=self.revision,
 
                message=self.message,
 
                date=self.date,
 
                author=self.author,
 
            )
 

	
 
    @LazyProperty
 
    def last(self):
 
        if self.repository is None:
 
            raise ChangesetError("Cannot check if it's most recent revision")
 
        return self.raw_id == self.repository.revisions[-1]
 

	
 
    @LazyProperty
 
    def parents(self):
 
        """
 
        Returns list of parents changesets.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def children(self):
 
        """
 
        Returns list of children changesets.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def id(self):
 
        """
 
        Returns string identifying this changeset.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def raw_id(self):
 
        """
 
        Returns raw string identifying this changeset.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def short_id(self):
 
        """
 
        Returns shortened version of ``raw_id`` attribute, as string,
 
        identifying this changeset, useful for web representation.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def revision(self):
 
        """
 
        Returns integer identifying this changeset.
 

	
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def committer(self):
 
        """
 
        Returns Committer for given commit
 
        """
 

	
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def committer_name(self):
 
        """
 
        Returns Author name for given commit
 
        """
 

	
 
        return author_name(self.committer)
 

	
 
    @LazyProperty
 
    def committer_email(self):
 
        """
 
        Returns Author email address for given commit
 
        """
 

	
 
        return author_email(self.committer)
 

	
 
    @LazyProperty
 
    def author(self):
 
        """
 
        Returns Author for given commit
 
        """
 

	
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def author_name(self):
 
        """
 
@@ -574,195 +574,195 @@ class BaseChangeset(object):
 
        """
 
        Returns iterable archive. Tiny wrapper around ``fill_archive`` method.
 

	
 
        :param chunk_size: extra parameter which controls size of returned
 
            chunks. Default:8k.
 
        """
 

	
 
        chunk_size = kwargs.pop('chunk_size', 8192)
 
        stream = kwargs.get('stream')
 
        self.fill_archive(**kwargs)
 
        while True:
 
            data = stream.read(chunk_size)
 
            if not data:
 
                break
 
            yield data
 

	
 
    @LazyProperty
 
    def root(self):
 
        """
 
        Returns ``RootNode`` object for this changeset.
 
        """
 
        return self.get_node('')
 

	
 
    def next(self, branch=None):
 
        """
 
        Returns next changeset from current, if branch is gives it will return
 
        next changeset belonging to this branch
 

	
 
        :param branch: show changesets within the given named branch
 
        """
 
        raise NotImplementedError
 

	
 
    def prev(self, branch=None):
 
        """
 
        Returns previous changeset from current, if branch is gives it will
 
        return previous changeset belonging to this branch
 

	
 
        :param branch: show changesets within the given named branch
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def added(self):
 
        """
 
        Returns list of added ``FileNode`` objects.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def changed(self):
 
        """
 
        Returns list of modified ``FileNode`` objects.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def removed(self):
 
        """
 
        Returns list of removed ``FileNode`` objects.
 
        """
 
        raise NotImplementedError
 

	
 
    @LazyProperty
 
    def size(self):
 
        """
 
        Returns total number of bytes from contents of all filenodes.
 
        """
 
        return sum((node.size for node in self.get_filenodes_generator()))
 

	
 
    def walk(self, topurl=''):
 
        """
 
        Similar to os.walk method. Instead of filesystem it walks through
 
        changeset starting at given ``topurl``.  Returns generator of tuples
 
        (topnode, dirnodes, filenodes).
 
        """
 
        topnode = self.get_node(topurl)
 
        yield (topnode, topnode.dirs, topnode.files)
 
        for dirnode in topnode.dirs:
 
            for tup in self.walk(dirnode.path):
 
                yield tup
 

	
 
    def get_filenodes_generator(self):
 
        """
 
        Returns generator that yields *all* file nodes.
 
        """
 
        for topnode, dirs, files in self.walk():
 
            for node in files:
 
                yield node
 

	
 
    def as_dict(self):
 
        """
 
        Returns dictionary with changeset's attributes and their values.
 
        """
 
        data = get_dict_for_attrs(self, ['id', 'raw_id', 'short_id',
 
            'revision', 'date', 'message'])
 
        data['author'] = {'name': self.author_name, 'email': self.author_email}
 
        data['added'] = [node.path for node in self.added]
 
        data['changed'] = [node.path for node in self.changed]
 
        data['removed'] = [node.path for node in self.removed]
 
        data['added'] = [safe_unicode(node.path) for node in self.added]
 
        data['changed'] = [safe_unicode(node.path) for node in self.changed]
 
        data['removed'] = [safe_unicode(node.path) for node in self.removed]
 
        return data
 

	
 
    @LazyProperty
 
    def closesbranch(self):
 
        return False
 

	
 
    @LazyProperty
 
    def obsolete(self):
 
        return False
 

	
 
    @LazyProperty
 
    def bumped(self):
 
        return False
 

	
 
    @LazyProperty
 
    def divergent(self):
 
        return False
 

	
 
    @LazyProperty
 
    def extinct(self):
 
        return False
 

	
 
    @LazyProperty
 
    def unstable(self):
 
        return False
 

	
 
    @LazyProperty
 
    def phase(self):
 
        return ''
 

	
 

	
 
class BaseWorkdir(object):
 
    """
 
    Working directory representation of single repository.
 

	
 
    :attribute: repository: repository object of working directory
 
    """
 

	
 
    def __init__(self, repository):
 
        self.repository = repository
 

	
 
    def get_branch(self):
 
        """
 
        Returns name of current branch.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_changeset(self):
 
        """
 
        Returns current changeset.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_added(self):
 
        """
 
        Returns list of ``FileNode`` objects marked as *new* in working
 
        directory.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_changed(self):
 
        """
 
        Returns list of ``FileNode`` objects *changed* in working directory.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_removed(self):
 
        """
 
        Returns list of ``RemovedFileNode`` objects marked as *removed* in
 
        working directory.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_untracked(self):
 
        """
 
        Returns list of ``FileNode`` objects which are present within working
 
        directory however are not tracked by repository.
 
        """
 
        raise NotImplementedError
 

	
 
    def get_status(self):
 
        """
 
        Returns dict with ``added``, ``changed``, ``removed`` and ``untracked``
 
        lists.
 
        """
 
        raise NotImplementedError
 

	
 
    def commit(self, message, **kwargs):
 
        """
 
        Commits local (from working directory) changes and returns newly
 
        created
 
        ``Changeset``. Updates repository's ``revisions`` list.
 

	
 
        :raises ``CommitError``: if any error occurs while committing
 
        """
 
        raise NotImplementedError
kallithea/tests/vcs/test_changesets.py
Show inline comments
 
# encoding: utf8
 

	
 
import time
 
import datetime
 
from kallithea.lib import vcs
 
from kallithea.tests.vcs.base import _BackendTestMixin
 
from kallithea.tests.vcs.conf import SCM_TESTS
 

	
 
from kallithea.lib.vcs.backends.base import BaseChangeset
 
from kallithea.lib.vcs.nodes import (
 
    FileNode, AddedFileNodesGenerator,
 
    ChangedFileNodesGenerator, RemovedFileNodesGenerator
 
)
 
from kallithea.lib.vcs.exceptions import (
 
    BranchDoesNotExistError, ChangesetDoesNotExistError,
 
    RepositoryError, EmptyRepositoryError
 
)
 
from kallithea.lib.vcs.utils.compat import unittest
 
from kallithea.tests.vcs.conf import get_new_dir
 

	
 

	
 
class TestBaseChangeset(unittest.TestCase):
 

	
 
    def test_as_dict(self):
 
        changeset = BaseChangeset()
 
        changeset.id = 'ID'
 
        changeset.raw_id = 'RAW_ID'
 
        changeset.short_id = 'SHORT_ID'
 
        changeset.revision = 1009
 
        changeset.date = datetime.datetime(2011, 1, 30, 1, 45)
 
        changeset.message = 'Message of a commit'
 
        changeset.author = 'Joe Doe <joe.doe@example.com>'
 
        changeset.added = [FileNode('foo/bar/baz'), FileNode('foobar')]
 
        changeset.added = [FileNode('foo/bar/baz'), FileNode(u'foobar'), FileNode(u'blåbærgrød')]
 
        changeset.changed = []
 
        changeset.removed = []
 
        self.assertEqual(changeset.as_dict(), {
 
            'id': 'ID',
 
            'raw_id': 'RAW_ID',
 
            'short_id': 'SHORT_ID',
 
            'revision': 1009,
 
            'date': datetime.datetime(2011, 1, 30, 1, 45),
 
            'message': 'Message of a commit',
 
            'author': {
 
                'name': 'Joe Doe',
 
                'email': 'joe.doe@example.com',
 
            },
 
            'added': ['foo/bar/baz', 'foobar'],
 
            'added': ['foo/bar/baz', 'foobar', u'bl\xe5b\xe6rgr\xf8d'],
 
            'changed': [],
 
            'removed': [],
 
        })
 

	
 

	
 
class _ChangesetsWithCommitsTestCaseixin(_BackendTestMixin):
 
    recreate_repo_per_test = True
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        start_date = datetime.datetime(2010, 1, 1, 20)
 
        for x in xrange(5):
 
            yield {
 
                'message': 'Commit %d' % x,
 
                'author': 'Joe Doe <joe.doe@example.com>',
 
                'date': start_date + datetime.timedelta(hours=12 * x),
 
                'added': [
 
                    FileNode('file_%d.txt' % x, content='Foobar %d' % x),
 
                ],
 
            }
 

	
 
    def test_new_branch(self):
 
        self.imc.add(vcs.nodes.FileNode('docs/index.txt',
 
            content='Documentation\n'))
 
        foobar_tip = self.imc.commit(
 
            message=u'New branch: foobar',
 
            author=u'joe',
 
            branch='foobar',
 
        )
 
        self.assertTrue('foobar' in self.repo.branches)
 
        self.assertEqual(foobar_tip.branch, 'foobar')
 
        # 'foobar' should be the only branch that contains the new commit
 
        self.assertNotEqual(*self.repo.branches.values())
 

	
 
    def test_new_head_in_default_branch(self):
 
        tip = self.repo.get_changeset()
 
        self.imc.add(vcs.nodes.FileNode('docs/index.txt',
 
            content='Documentation\n'))
 
        foobar_tip = self.imc.commit(
 
            message=u'New branch: foobar',
 
            author=u'joe',
 
            branch='foobar',
 
            parents=[tip],
 
        )
 
        self.imc.change(vcs.nodes.FileNode('docs/index.txt',
 
            content='Documentation\nand more...\n'))
 
        newtip = self.imc.commit(
 
            message=u'At default branch',
 
            author=u'joe',
 
            branch=foobar_tip.branch,
 
            parents=[foobar_tip],
 
        )
 

	
 
        newest_tip = self.imc.commit(
 
            message=u'Merged with %s' % foobar_tip.raw_id,
 
            author=u'joe',
 
            branch=self.backend_class.DEFAULT_BRANCH_NAME,
 
            parents=[newtip, foobar_tip],
 
        )
 

	
 
        self.assertEqual(newest_tip.branch,
 
            self.backend_class.DEFAULT_BRANCH_NAME)
 

	
 
    def test_get_changesets_respects_branch_name(self):
 
        tip = self.repo.get_changeset()
 
        self.imc.add(vcs.nodes.FileNode('docs/index.txt',
 
            content='Documentation\n'))
 
        doc_changeset = self.imc.commit(
 
            message=u'New branch: docs',
 
            author=u'joe',
 
            branch='docs',
 
        )
 
        self.imc.add(vcs.nodes.FileNode('newfile', content=''))
 
        self.imc.commit(
 
            message=u'Back in default branch',
 
            author=u'joe',
 
            parents=[tip],
 
        )
 
        default_branch_changesets = self.repo.get_changesets(
 
            branch_name=self.repo.DEFAULT_BRANCH_NAME)
 
        self.assertNotIn(doc_changeset, default_branch_changesets)
 

	
 
    def test_get_changeset_by_branch(self):
 
        for branch, sha in self.repo.branches.iteritems():
 
            self.assertEqual(sha, self.repo.get_changeset(branch).raw_id)
 

	
 
    def test_get_changeset_by_tag(self):
 
        for tag, sha in self.repo.tags.iteritems():
 
            self.assertEqual(sha, self.repo.get_changeset(tag).raw_id)
 

	
 
    def test_get_changeset_parents(self):
 
        for test_rev in [1, 2, 3]:
 
            sha = self.repo.get_changeset(test_rev-1)
 
            self.assertEqual([sha], self.repo.get_changeset(test_rev).parents)
 

	
 
    def test_get_changeset_children(self):
 
@@ -248,145 +248,147 @@ class _ChangesetsTestCaseMixin(_BackendT
 

	
 
    def test_get_filenodes_generator(self):
 
        tip = self.repo.get_changeset()
 
        filepaths = [node.path for node in tip.get_filenodes_generator()]
 
        self.assertItemsEqual(filepaths, ['file_%d.txt' % x for x in xrange(5)])
 

	
 
    def test_size(self):
 
        tip = self.repo.get_changeset()
 
        size = 5 * len('Foobar N') # Size of 5 files
 
        self.assertEqual(tip.size, size)
 

	
 
    def test_author(self):
 
        tip = self.repo.get_changeset()
 
        self.assertEqual(tip.author, u'Joe Doe <joe.doe@example.com>')
 

	
 
    def test_author_name(self):
 
        tip = self.repo.get_changeset()
 
        self.assertEqual(tip.author_name, u'Joe Doe')
 

	
 
    def test_author_email(self):
 
        tip = self.repo.get_changeset()
 
        self.assertEqual(tip.author_email, u'joe.doe@example.com')
 

	
 
    def test_get_changesets_raise_changesetdoesnotexist_for_wrong_start(self):
 
        with self.assertRaises(ChangesetDoesNotExistError):
 
            list(self.repo.get_changesets(start='foobar'))
 

	
 
    def test_get_changesets_raise_changesetdoesnotexist_for_wrong_end(self):
 
        with self.assertRaises(ChangesetDoesNotExistError):
 
            list(self.repo.get_changesets(end='foobar'))
 

	
 
    def test_get_changesets_raise_branchdoesnotexist_for_wrong_branch_name(self):
 
        with self.assertRaises(BranchDoesNotExistError):
 
            list(self.repo.get_changesets(branch_name='foobar'))
 

	
 
    def test_get_changesets_raise_repositoryerror_for_wrong_start_end(self):
 
        start = self.repo.revisions[-1]
 
        end = self.repo.revisions[0]
 
        with self.assertRaises(RepositoryError):
 
            list(self.repo.get_changesets(start=start, end=end))
 

	
 
    def test_get_changesets_numerical_id_reversed(self):
 
        with self.assertRaises(RepositoryError):
 
            [x for x in self.repo.get_changesets(start=3, end=2)]
 

	
 
    def test_get_changesets_numerical_id_respects_both_start_and_end_last(self):
 
        with self.assertRaises(RepositoryError):
 
            last = len(self.repo.revisions)
 
            list(self.repo.get_changesets(start=last-1, end=last-2))
 

	
 
    def test_get_changesets_numerical_id_last_zero_error(self):
 
        with self.assertRaises(RepositoryError):
 
            last = len(self.repo.revisions)
 
            list(self.repo.get_changesets(start=last-1, end=0))
 

	
 

	
 
class _ChangesetsChangesTestCaseMixin(_BackendTestMixin):
 
    recreate_repo_per_test = False
 

	
 
    @classmethod
 
    def _get_commits(cls):
 
        return [
 
            {
 
                'message': u'Initial',
 
                'author': u'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 20),
 
                'added': [
 
                    FileNode('foo/bar', content='foo'),
 
                    FileNode('foo/bał', content='foo'),
 
                    FileNode('foobar', content='foo'),
 
                    FileNode('qwe', content='foo'),
 
                ],
 
            },
 
            {
 
                'message': u'Massive changes',
 
                'author': u'Joe Doe <joe.doe@example.com>',
 
                'date': datetime.datetime(2010, 1, 1, 22),
 
                'added': [FileNode('fallout', content='War never changes')],
 
                'changed': [
 
                    FileNode('foo/bar', content='baz'),
 
                    FileNode('foobar', content='baz'),
 
                ],
 
                'removed': [FileNode('qwe')],
 
            },
 
        ]
 

	
 
    def test_initial_commit(self):
 
        changeset = self.repo.get_changeset(0)
 
        self.assertItemsEqual(changeset.added, [
 
            changeset.get_node('foo/bar'),
 
            changeset.get_node('foo/bał'),
 
            changeset.get_node('foobar'),
 
            changeset.get_node('qwe'),
 
        ])
 
        self.assertItemsEqual(changeset.changed, [])
 
        self.assertItemsEqual(changeset.removed, [])
 
        assert u'foo/ba\u0142' in changeset.as_dict()['added']
 
        assert u'foo/ba\u0142' in changeset.__json__(with_file_list=True)['added']
 

	
 
    def test_head_added(self):
 
        changeset = self.repo.get_changeset()
 
        self.assertTrue(isinstance(changeset.added, AddedFileNodesGenerator))
 
        self.assertItemsEqual(changeset.added, [
 
            changeset.get_node('fallout'),
 
        ])
 
        self.assertTrue(isinstance(changeset.changed, ChangedFileNodesGenerator))
 
        self.assertItemsEqual(changeset.changed, [
 
            changeset.get_node('foo/bar'),
 
            changeset.get_node('foobar'),
 
        ])
 
        self.assertTrue(isinstance(changeset.removed, RemovedFileNodesGenerator))
 
        self.assertEqual(len(changeset.removed), 1)
 
        self.assertEqual(list(changeset.removed)[0].path, 'qwe')
 

	
 
    def test_get_filemode(self):
 
        changeset = self.repo.get_changeset()
 
        self.assertEqual(33188, changeset.get_file_mode('foo/bar'))
 

	
 
    def test_get_filemode_non_ascii(self):
 
        changeset = self.repo.get_changeset()
 
        self.assertEqual(33188, changeset.get_file_mode('foo/bał'))
 
        self.assertEqual(33188, changeset.get_file_mode(u'foo/bał'))
 

	
 

	
 
# For each backend create test case class
 
for alias in SCM_TESTS:
 
    attrs = {
 
        'backend_alias': alias,
 
    }
 
    # tests with additional commits
 
    cls_name = ''.join(('%s changesets with commits test' % alias).title().split())
 
    cls_name = alias.title() + 'ChangesetsWithCommitsTest'
 
    bases = (_ChangesetsWithCommitsTestCaseixin, unittest.TestCase)
 
    globals()[cls_name] = type(cls_name, bases, attrs)
 

	
 
    # tests without additional commits
 
    cls_name = ''.join(('%s changesets test' % alias).title().split())
 
    cls_name = alias.title() + 'ChangesetsTest'
 
    bases = (_ChangesetsTestCaseMixin, unittest.TestCase)
 
    globals()[cls_name] = type(cls_name, bases, attrs)
 

	
 
    # tests changes
 
    cls_name = ''.join(('%s changesets changes test' % alias).title().split())
 
    cls_name = alias.title() + 'ChangesetsChangesTest'
 
    bases = (_ChangesetsChangesTestCaseMixin, unittest.TestCase)
 
    globals()[cls_name] = type(cls_name, bases, attrs)
 

	
 

	
 
if __name__ == '__main__':
 
    unittest.main()
0 comments (0 inline, 0 general)