Changeset - e0da2c0ecee1
[Not reviewed]
beta
0 2 0
Marcin Kuzminski - 13 years ago 2012-10-29 22:26:44
marcin@python-works.com
fixes #625 Git-Tags are not displayed in Shortlog
- improved git extraction of git tags if they are associated to a commit
- added not implemented tags() function for git changesets
2 files changed with 14 insertions and 3 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/vcs/backends/git/changeset.py
Show inline comments
 
import re
 
from itertools import chain
 
from dulwich import objects
 
from subprocess import Popen, PIPE
 
from rhodecode.lib.vcs.conf import settings
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
from rhodecode.lib.vcs.exceptions import ChangesetError
 
from rhodecode.lib.vcs.exceptions import NodeDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import VCSError
 
from rhodecode.lib.vcs.exceptions import ChangesetDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import ImproperArchiveTypeError
 
from rhodecode.lib.vcs.backends.base import BaseChangeset, EmptyChangeset
 
from rhodecode.lib.vcs.nodes import FileNode, DirNode, NodeKind, RootNode, \
 
    RemovedFileNode, SubModuleNode, ChangedFileNodesGenerator,\
 
    AddedFileNodesGenerator, RemovedFileNodesGenerator
 
from rhodecode.lib.vcs.utils import safe_unicode
 
from rhodecode.lib.vcs.utils import date_fromtimestamp
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 

	
 

	
 
class GitChangeset(BaseChangeset):
 
    """
 
    Represents state of the repository at single revision.
 
    """
 

	
 
    def __init__(self, repository, revision):
 
        self._stat_modes = {}
 
        self.repository = repository
 

	
 
        try:
 
            commit = self.repository._repo.get_object(revision)
 
            if isinstance(commit, objects.Tag):
 
                revision = commit.object[1]
 
                commit = self.repository._repo.get_object(commit.object[1])
 
        except KeyError:
 
            raise RepositoryError("Cannot get object with id %s" % revision)
 
        self.raw_id = revision
 
        self.id = self.raw_id
 
        self.short_id = self.raw_id[:12]
 
        self._commit = commit
 

	
 
        self._tree_id = commit.tree
 
        self._commiter_property = 'committer'
 
        self._date_property = 'commit_time'
 
        self._date_tz_property = 'commit_timezone'
 
        self.revision = repository.revisions.index(revision)
 

	
 
        self.message = safe_unicode(commit.message)
 
        #self.branch = None
 
        self.tags = []
 

	
 
        self.nodes = {}
 
        self._paths = {}
 

	
 
    @LazyProperty
 
    def author(self):
 
        return safe_unicode(getattr(self._commit, self._commiter_property))
 

	
 
    @LazyProperty
 
    def date(self):
 
        return date_fromtimestamp(getattr(self._commit, self._date_property),
 
                                  getattr(self._commit, self._date_tz_property))
 

	
 
    @LazyProperty
 
    def _timestamp(self):
 
        return getattr(self._commit, self._date_property)
 

	
 
    @LazyProperty
 
    def status(self):
 
        """
 
        Returns modified, added, removed, deleted files for current changeset
 
        """
 
        return self.changed, self.added, self.removed
 

	
 
    @LazyProperty
 
    def tags(self):
 
        _tags = []
 
        for tname, tsha in self.repository.tags.iteritems():
 
            if tsha == self.raw_id:
 
                _tags.append(tname)
 
        return _tags
 

	
 
    @LazyProperty
 
    def branch(self):
 

	
 
        heads = self.repository._heads(reverse=False)
 

	
 
        ref = heads.get(self.raw_id)
 
        if ref:
 
            return safe_unicode(ref)
 

	
 
    def _fix_path(self, path):
 
        """
 
        Paths are stored without trailing slash so we need to get rid off it if
 
        needed.
 
        """
 
        if path.endswith('/'):
 
            path = path.rstrip('/')
 
        return path
 

	
 
    def _get_id_for_path(self, path):
 

	
 
        # FIXME: Please, spare a couple of minutes and make those codes cleaner;
 
        if not path in self._paths:
 
            path = path.strip('/')
 
            # set root tree
 
            tree = self.repository._repo[self._tree_id]
 
            if path == '':
 
                self._paths[''] = tree.id
 
                return tree.id
 
            splitted = path.split('/')
 
            dirs, name = splitted[:-1], splitted[-1]
 
            curdir = ''
 

	
 
            # initially extract things from root dir
 
            for item, stat, id in tree.iteritems():
 
                if curdir:
 
                    name = '/'.join((curdir, item))
 
                else:
 
                    name = item
 
                self._paths[name] = id
 
                self._stat_modes[name] = stat
 

	
 
            for dir in dirs:
 
                if curdir:
 
                    curdir = '/'.join((curdir, dir))
 
                else:
 
                    curdir = dir
 
                dir_id = None
 
                for item, stat, id in tree.iteritems():
 
                    if dir == item:
 
                        dir_id = id
 
                if dir_id:
 
                    # Update tree
 
                    tree = self.repository._repo[dir_id]
 
                    if not isinstance(tree, objects.Tree):
 
                        raise ChangesetError('%s is not a directory' % curdir)
 
                else:
 
                    raise ChangesetError('%s have not been found' % curdir)
 

	
 
                # cache all items from the given traversed tree
 
                for item, stat, id in tree.iteritems():
 
                    if curdir:
 
                        name = '/'.join((curdir, item))
 
                    else:
 
                        name = item
 
                    self._paths[name] = id
 
                    self._stat_modes[name] = stat
 
            if not path in self._paths:
 
                raise NodeDoesNotExistError("There is no file nor directory "
 
                    "at the given path %r at revision %r"
 
                    % (path, self.short_id))
 
        return self._paths[path]
 

	
 
    def _get_kind(self, path):
 
        obj = self.repository._repo[self._get_id_for_path(path)]
 
        if isinstance(obj, objects.Blob):
 
            return NodeKind.FILE
 
        elif isinstance(obj, objects.Tree):
 
            return NodeKind.DIR
 

	
 
    def _get_file_nodes(self):
 
        return chain(*(t[2] for t in self.walk()))
 

	
 
    @LazyProperty
 
    def parents(self):
 
        """
 
        Returns list of parents changesets.
 
        """
 
        return [self.repository.get_changeset(parent)
 
                for parent in self._commit.parents]
 

	
 
    def next(self, branch=None):
 

	
 
        if branch and self.branch != branch:
 
            raise VCSError('Branch option used on changeset not belonging '
 
                           'to that branch')
 

	
 
        def _next(changeset, branch):
rhodecode/lib/vcs/backends/git/repository.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    vcs.backends.git
 
    ~~~~~~~~~~~~~~~~
 

	
 
    Git backend implementation.
 

	
 
    :created_on: Apr 8, 2010
 
    :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.
 
"""
 

	
 
import os
 
import re
 
import time
 
import posixpath
 
import logging
 
import traceback
 
import urllib
 
import urllib2
 
from dulwich.repo import Repo, NotGitRepository
 
#from dulwich.config import ConfigFile
 
from dulwich.objects import Tag
 
from string import Template
 
from subprocess import Popen, PIPE
 
from rhodecode.lib.vcs.backends.base import BaseRepository
 
from rhodecode.lib.vcs.exceptions import BranchDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import ChangesetDoesNotExistError
 
from rhodecode.lib.vcs.exceptions import EmptyRepositoryError
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
from rhodecode.lib.vcs.exceptions import TagAlreadyExistError
 
from rhodecode.lib.vcs.exceptions import TagDoesNotExistError
 
from rhodecode.lib.vcs.utils import safe_unicode, makedate, date_fromtimestamp
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
from rhodecode.lib.vcs.utils.ordered_dict import OrderedDict
 
from rhodecode.lib.vcs.utils.paths import abspath
 
from rhodecode.lib.vcs.utils.paths import get_user_home
 
from .workdir import GitWorkdir
 
from .changeset import GitChangeset
 
from .inmemory import GitInMemoryChangeset
 
from .config import ConfigFile
 
from rhodecode.lib import subprocessio
 

	
 

	
 
log = logging.getLogger(__name__)
 

	
 

	
 
class GitRepository(BaseRepository):
 
    """
 
    Git repository backend.
 
    """
 
    DEFAULT_BRANCH_NAME = 'master'
 
    scm = 'git'
 

	
 
    def __init__(self, repo_path, create=False, src_url=None,
 
                 update_after_clone=False, bare=False):
 

	
 
        self.path = abspath(repo_path)
 
        self._repo = self._get_repo(create, src_url, update_after_clone, bare)
 
        #temporary set that to now at later we will move it to constructor
 
        baseui = None
 
        if baseui is None:
 
            from mercurial.ui import ui
 
            baseui = ui()
 
        # patch the instance of GitRepo with an "FAKE" ui object to add
 
        # compatibility layer with Mercurial
 
        setattr(self._repo, 'ui', baseui)
 

	
 
        try:
 
            self.head = self._repo.head()
 
        except KeyError:
 
            self.head = None
 

	
 
        self._config_files = [
 
            bare and abspath(self.path, 'config') or abspath(self.path, '.git',
 
                'config'),
 
            abspath(get_user_home(), '.gitconfig'),
 
        ]
 
        self.bare = self._repo.bare
 

	
 
    @LazyProperty
 
    def revisions(self):
 
        """
 
        Returns list of revisions' ids, in ascending order.  Being lazy
 
        attribute allows external tools to inject shas from cache.
 
        """
 
        return self._get_all_revisions()
 

	
 
    def run_git_command(self, cmd):
 
        """
 
        Runs given ``cmd`` as git command and returns tuple
 
        (returncode, stdout, stderr).
 

	
 
        .. note::
 
           This method exists only until log/blame functionality is implemented
 
           at Dulwich (see https://bugs.launchpad.net/bugs/645142). Parsing
 
           os command's output is road to hell...
 

	
 
        :param cmd: git command to be executed
 
        """
 

	
 
        _copts = ['-c', 'core.quotepath=false', ]
 
        _str_cmd = False
 
        if isinstance(cmd, basestring):
 
            cmd = [cmd]
 
            _str_cmd = True
 

	
 
        gitenv = os.environ
 
        # need to clean fix GIT_DIR !
 
        if 'GIT_DIR' in gitenv:
 
            del gitenv['GIT_DIR']
 
        gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
 

	
 
        cmd = ['git'] + _copts + cmd
 
        if _str_cmd:
 
            cmd = ' '.join(cmd)
 
        try:
 
            opts = dict(
 
                env=gitenv,
 
@@ -308,192 +308,196 @@ class GitRepository(BaseRepository):
 
    @LazyProperty
 
    def description(self):
 
        idx_loc = '' if self.bare else '.git'
 
        undefined_description = u'unknown'
 
        description_path = os.path.join(self.path, idx_loc, 'description')
 
        if os.path.isfile(description_path):
 
            return safe_unicode(open(description_path).read())
 
        else:
 
            return undefined_description
 

	
 
    @LazyProperty
 
    def contact(self):
 
        undefined_contact = u'Unknown'
 
        return undefined_contact
 

	
 
    @property
 
    def branches(self):
 
        if not self.revisions:
 
            return {}
 
        sortkey = lambda ctx: ctx[0]
 
        _branches = [(x[0], x[1][0])
 
                     for x in self._parsed_refs.iteritems() if x[1][1] == 'H']
 
        return OrderedDict(sorted(_branches, key=sortkey, reverse=False))
 

	
 
    @LazyProperty
 
    def tags(self):
 
        return self._get_tags()
 

	
 
    def _get_tags(self):
 
        if not self.revisions:
 
            return {}
 

	
 
        sortkey = lambda ctx: ctx[0]
 
        _tags = [(x[0], x[1][0])
 
                 for x in self._parsed_refs.iteritems() if x[1][1] == 'T']
 
        return OrderedDict(sorted(_tags, key=sortkey, reverse=True))
 

	
 
    def tag(self, name, user, revision=None, message=None, date=None,
 
            **kwargs):
 
        """
 
        Creates and returns a tag for the given ``revision``.
 

	
 
        :param name: name for new tag
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param revision: changeset id for which new tag would be created
 
        :param message: message of the tag's commit
 
        :param date: date of tag's commit
 

	
 
        :raises TagAlreadyExistError: if tag with same name already exists
 
        """
 
        if name in self.tags:
 
            raise TagAlreadyExistError("Tag %s already exists" % name)
 
        changeset = self.get_changeset(revision)
 
        message = message or "Added tag %s for commit %s" % (name,
 
            changeset.raw_id)
 
        self._repo.refs["refs/tags/%s" % name] = changeset._commit.id
 

	
 
        self._parsed_refs = self._get_parsed_refs()
 
        self.tags = self._get_tags()
 
        return changeset
 

	
 
    def remove_tag(self, name, user, message=None, date=None):
 
        """
 
        Removes tag with the given ``name``.
 

	
 
        :param name: name of the tag to be removed
 
        :param user: full username, i.e.: "Joe Doe <joe.doe@example.com>"
 
        :param message: message of the tag's removal commit
 
        :param date: date of tag's removal commit
 

	
 
        :raises TagDoesNotExistError: if tag with given name does not exists
 
        """
 
        if name not in self.tags:
 
            raise TagDoesNotExistError("Tag %s does not exist" % name)
 
        tagpath = posixpath.join(self._repo.refs.path, 'refs', 'tags', name)
 
        try:
 
            os.remove(tagpath)
 
            self._parsed_refs = self._get_parsed_refs()
 
            self.tags = self._get_tags()
 
        except OSError, e:
 
            raise RepositoryError(e.strerror)
 

	
 
    @LazyProperty
 
    def _parsed_refs(self):
 
        return self._get_parsed_refs()
 

	
 
    def _get_parsed_refs(self):
 
        refs = self._repo.get_refs()
 
        keys = [('refs/heads/', 'H'),
 
                ('refs/remotes/origin/', 'RH'),
 
                ('refs/tags/', 'T')]
 
        _refs = {}
 
        for ref, sha in refs.iteritems():
 
            for k, type_ in keys:
 
                if ref.startswith(k):
 
                    _key = ref[len(k):]
 
                    if type_ == 'T':
 
                        obj = self._repo.get_object(sha)
 
                        if isinstance(obj, Tag):
 
                            sha = self._repo.get_object(sha).object[1]
 
                    _refs[_key] = [sha, type_]
 
                    break
 
        return _refs
 

	
 
    def _heads(self, reverse=False):
 
        refs = self._repo.get_refs()
 
        heads = {}
 

	
 
        for key, val in refs.items():
 
            for ref_key in ['refs/heads/', 'refs/remotes/origin/']:
 
                if key.startswith(ref_key):
 
                    n = key[len(ref_key):]
 
                    if n not in ['HEAD']:
 
                        heads[n] = val
 

	
 
        return heads if reverse else dict((y, x) for x, y in heads.iteritems())
 

	
 
    def get_changeset(self, revision=None):
 
        """
 
        Returns ``GitChangeset`` object representing commit from git repository
 
        at the given revision or head (most recent commit) if None given.
 
        """
 
        if isinstance(revision, GitChangeset):
 
            return revision
 
        revision = self._get_revision(revision)
 
        changeset = GitChangeset(repository=self, revision=revision)
 
        return changeset
 

	
 
    def get_changesets(self, start=None, end=None, start_date=None,
 
           end_date=None, branch_name=None, reverse=False):
 
        """
 
        Returns iterator of ``GitChangeset`` objects from start to end (both
 
        are inclusive), in ascending date order (unless ``reverse`` is set).
 

	
 
        :param start: changeset ID, as str; first returned changeset
 
        :param end: changeset ID, as str; last returned changeset
 
        :param start_date: if specified, changesets with commit date less than
 
          ``start_date`` would be filtered out from returned set
 
        :param end_date: if specified, changesets with commit date greater than
 
          ``end_date`` would be filtered out from returned set
 
        :param branch_name: if specified, changesets not reachable from given
 
          branch would be filtered out from returned set
 
        :param reverse: if ``True``, returned generator would be reversed
 
          (meaning that returned changesets would have descending date order)
 

	
 
        :raise BranchDoesNotExistError: If given ``branch_name`` does not
 
            exist.
 
        :raise ChangesetDoesNotExistError: If changeset for given ``start`` or
 
          ``end`` could not be found.
 

	
 
        """
 
        if branch_name and branch_name not in self.branches:
 
            raise BranchDoesNotExistError("Branch '%s' not found" \
 
                                          % branch_name)
 
        # %H at format means (full) commit hash, initial hashes are retrieved
 
        # in ascending date order
 
        cmd_template = 'log --date-order --reverse --pretty=format:"%H"'
 
        cmd_params = {}
 
        if start_date:
 
            cmd_template += ' --since "$since"'
 
            cmd_params['since'] = start_date.strftime('%m/%d/%y %H:%M:%S')
 
        if end_date:
 
            cmd_template += ' --until "$until"'
 
            cmd_params['until'] = end_date.strftime('%m/%d/%y %H:%M:%S')
 
        if branch_name:
 
            cmd_template += ' $branch_name'
 
            cmd_params['branch_name'] = branch_name
 
        else:
 
            cmd_template += ' --all'
 

	
 
        cmd = Template(cmd_template).safe_substitute(**cmd_params)
 
        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:
0 comments (0 inline, 0 general)