vcs.backends.git.repository

    ~~~~~~~~~~~~~~~~~~~~~~~~~~~

    Git repository implementation.

    :created_on: Apr 8, 2010

    :copyright: (c) 2010-2011 by Marcin Kuzminski, Lukasz Balcerzak.

"""

import errno

import logging

import os

import re

import time

import urllib

import urllib2

from collections import OrderedDict

from dulwich.config import ConfigFile

from dulwich.objects import Tag

from dulwich.repo import NotGitRepository, Repo

from kallithea.lib.vcs import subprocessio

from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator

from kallithea.lib.vcs.conf import settings

from kallithea.lib.vcs.exceptions import (

    BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError, RepositoryError, TagAlreadyExistError, TagDoesNotExistError)

from kallithea.lib.vcs.utils import date_fromtimestamp, makedate, safe_str, safe_unicode

from kallithea.lib.vcs.utils.hgcompat import hg_url, httpbasicauthhandler, httpdigestauthhandler

from kallithea.lib.vcs.utils.lazy import LazyProperty

from kallithea.lib.vcs.utils.paths import abspath, get_user_home

from .changeset import GitChangeset

from .inmemory import GitInMemoryChangeset

from .workdir import GitWorkdir

SHA_PATTERN = re.compile(r'^([0-9a-fA-F]{12}|[0-9a-fA-F]{40})$')

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 = safe_unicode(abspath(repo_path))

        self.repo = self._get_repo(create, src_url, update_after_clone, bare)

        self.bare = self.repo.bare

    @property

    def _config_files(self):

        return [

            self.bare and abspath(self.path, 'config')

                      or abspath(self.path, '.git', 'config'),

             abspath(get_user_home(), '.gitconfig'),

         ]

    @property

    def _repo(self):

        return self.repo

    @property

    def head(self):

        try:

            return self._repo.head()

        except KeyError:

            return None

    @property

    def _empty(self):

        """

        Checks if repository is empty ie. without any changesets

        """

        try:

            self.revisions[0]

        except (KeyError, IndexError):

            return True

        return False

    @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()

    @classmethod

        """

        :param cmd: git command to be executed

        """

        # need to clean fix GIT_DIR !

        gitenv['GIT_CONFIG_NOGLOBAL'] = '1'

        try:

        finally:

            p.close()

    def run_git_command(self, cmd):

        if os.path.isdir(self.path):

    @classmethod

    def _check_url(cls, url):

        """

        Function will check given url and try to verify if it's a valid

        link. Sometimes it may happened that git will issue basic

        auth request that can cause whole API to hang when used from python

        or other external calls.

        On failures it'll raise urllib2.HTTPError, exception is also thrown

        when the return code is non 200

        """

        # check first if it's not an local url

        if os.path.isdir(url) or url.startswith('file:'):

            return True

        if url.startswith('git://'):

            return True

        if '+' in url[:url.find('://')]:

            url = url[url.find('+') + 1:]

        handlers = []

        url_obj = hg_url(url)

        test_uri, authinfo = url_obj.authinfo()

        url_obj.passwd = '*****'

        cleaned_uri = str(url_obj)

        if not test_uri.endswith('info/refs'):

            test_uri = test_uri.rstrip('/') + '/info/refs'

        if authinfo:

            # create a password manager

            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()

            passmgr.add_password(*authinfo)

            handlers.extend((httpbasicauthhandler(passmgr),

                             httpdigestauthhandler(passmgr)))

        o = urllib2.build_opener(*handlers)

        o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git

        q = {"service": 'git-upload-pack'}

        qs = '?%s' % urllib.urlencode(q)

        cu = "%s%s" % (test_uri, qs)

        req = urllib2.Request(cu, None, {})

        try:

            resp = o.open(req)

            if resp.code != 200:

                raise Exception('Return Code is not 200')

        except Exception as e:

            # means it cannot be cloned

            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))

        # now detect if it's proper git repo

        gitdata = resp.read()

        if 'service=git-upload-pack' not in gitdata:

            raise urllib2.URLError(

                "url [%s] does not look like an git" % cleaned_uri)

        return True

    def _get_repo(self, create, src_url=None, update_after_clone=False,

                  bare=False):

        if create and os.path.exists(self.path):

            raise RepositoryError("Location already exist")

        if src_url and not create:

            raise RepositoryError("Create should be set to True if src_url is "

                                  "given (clone operation creates repository)")

        try:

            if create and src_url:

                GitRepository._check_url(src_url)

                self.clone(src_url, update_after_clone, bare)

                return Repo(self.path)

            elif create:

                os.makedirs(self.path)

                if bare:

                    return Repo.init_bare(self.path)

                else:

                    return Repo.init(self.path)

            else:

                return Repo(self.path)

        except (NotGitRepository, OSError) as err:

            raise RepositoryError(err)

    def _get_all_revisions(self):

        # we must check if this repo is not empty, since later command

        # fails if it is. And it's cheaper to ask than throw the subprocess

        # errors

        try:

            self._repo.head()

        except KeyError:

            return []

            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, max_revisions=None):

        """

        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)

        # actually we should check now if it's not an empty repo to not spaw

        # subprocess commands

        if self._empty:

            raise EmptyRepositoryError("There are no changesets yet")

        # %H at format means (full) commit hash, initial hashes are retrieved

        # in ascending date order

        cmd = ['log', '--date-order', '--reverse', '--pretty=format:%H']

        if max_revisions:

            cmd += ['--max-count=%s' % max_revisions]

        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):

        """

        :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, 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)