Changeset - 42ef4ea26efa
[Not reviewed]
default
0 3 0
Mads Kiilerich - 6 years ago 2020-01-25 16:42:40
mads@kiilerich.com
Grafted from: 42d725799780
vcs: drop unused get_hook_location (Issue #353)

Also, the hg implementation was odd - the '.hgrc' directory name looks very
similar to 'hgrc' but could be anything, such as 'hooks'.
3 files changed with 5 insertions and 18 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/git/repository.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    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
 

	
 
import mercurial.url  # import httpbasicauthhandler, httpdigestauthhandler
 
import mercurial.util  # import url as hg_url
 
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 ascii_str, date_fromtimestamp, makedate, safe_bytes, safe_str, safe_unicode
 
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
 
    def _run_git_command(cls, cmd, cwd=None):
 
        """
 
        Runs given ``cmd`` as git command and returns output bytes in a tuple
 
        (stdout, stderr) ... or raise RepositoryError.
 

	
 
        :param cmd: git command to be executed
 
        :param cwd: passed directly to subprocess
 
        """
 
        # need to clean fix GIT_DIR !
 
        gitenv = dict(os.environ)
 
        gitenv.pop('GIT_DIR', None)
 
        gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
 

	
 
        assert isinstance(cmd, list), cmd
 
        cmd = [settings.GIT_EXECUTABLE_PATH, '-c', 'core.quotepath=false'] + cmd
 
        try:
 
            p = subprocessio.SubprocessIOChunker(cmd, cwd=cwd, env=gitenv, shell=False)
 
        except (EnvironmentError, OSError) as err:
 
            # output from the failing process is in str(EnvironmentError)
 
            msg = ("Couldn't run git command %s.\n"
 
                   "Subprocess failed with '%s': %s\n" %
 
                   (cmd, type(err).__name__, err)
 
            ).strip()
 
            log.error(msg)
 
            raise RepositoryError(msg)
 

	
 
        try:
 
            stdout = b''.join(p.output)
 
            stderr = b''.join(p.error)
 
        finally:
 
            p.close()
 
        # TODO: introduce option to make commands fail if they have any stderr output?
 
        if stderr:
 
            log.debug('stderr from %s:\n%s', cmd, stderr)
 
        else:
 
            log.debug('stderr from %s: None', cmd)
 
        return stdout, stderr
 

	
 
    def run_git_command(self, cmd):
 
        """
 
        Runs given ``cmd`` as git command with cwd set to current repo.
 
        Returns stdout as unicode str ... or raise RepositoryError.
 
        """
 
        cwd = None
 
        if os.path.isdir(self.path):
 
            cwd = self.path
 
        stdout, _stderr = self._run_git_command(cmd, cwd=cwd)
 
        return safe_unicode(stdout)
 

	
 
    @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 = mercurial.util.url(safe_bytes(url))
 
        test_uri, authinfo = url_obj.authinfo()
 
        if not test_uri.endswith('info/refs'):
 
            test_uri = test_uri.rstrip('/') + '/info/refs'
 

	
 
        url_obj.passwd = b'*****'
 
        cleaned_uri = str(url_obj)
 

	
 
        if authinfo:
 
            # create a password manager
 
            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
 
            passmgr.add_password(*authinfo)
 

	
 
            handlers.extend((mercurial.url.httpbasicauthhandler(passmgr),
 
                             mercurial.url.httpdigestauthhandler(passmgr)))
 

	
 
        o = urllib2.build_opener(*handlers)
 
        o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git
 

	
 
        req = urllib2.Request(
 
            "%s?%s" % (
 
                test_uri,
 
                urllib.urlencode({"service": 'git-upload-pack'})
 
            ))
 

	
 
        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 []
 

	
 
        rev_filter = settings.GIT_REV_FILTER
 
        cmd = ['rev-list', rev_filter, '--reverse', '--date-order']
 
        try:
 
            so = self.run_git_command(cmd)
 
        except RepositoryError:
 
            # Can be raised for empty repositories
 
            return []
 
        return so.splitlines()
 

	
 
    def _get_all_revisions2(self):
 
        # alternate implementation using dulwich
 
        includes = [ascii_str(sha) for key, (sha, type_) in self._parsed_refs.iteritems()
 
                    if type_ != b'T']
 
        return [c.commit.id for c in self._repo.get_walker(include=includes)]
 

	
 
    def _get_revision(self, revision):
 
        """
 
        Given any revision identifier, returns a 40 char string with revision hash.
 
        """
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in (None, '', 'tip', 'HEAD', 'head', -1):
 
            revision = -1
 

	
 
        if isinstance(revision, int):
 
            try:
 
                return self.revisions[revision]
 
            except IndexError:
 
                msg = "Revision %r does not exist for %s" % (revision, self.name)
 
                raise ChangesetDoesNotExistError(msg)
 

	
 
        if isinstance(revision, (str, unicode)):
 
            if revision.isdigit() and (len(revision) < 12 or len(revision) == revision.count('0')):
 
                try:
 
                    return self.revisions[int(revision)]
 
                except IndexError:
 
                    msg = "Revision %r does not exist for %s" % (revision, self)
 
                    raise ChangesetDoesNotExistError(msg)
 

	
 
            # get by branch/tag name
 
            _ref_revision = self._parsed_refs.get(safe_bytes(revision))
 
            if _ref_revision:  # and _ref_revision[1] in [b'H', b'RH', b'T']:
 
                return ascii_str(_ref_revision[0])
 

	
 
            if revision in self.revisions:
 
                return revision
 

	
 
            # maybe it's a tag ? we don't have them in self.revisions
 
            if revision in self.tags.values():
 
                return revision
 

	
 
            if SHA_PATTERN.match(revision):
 
                msg = "Revision %r does not exist for %s" % (revision, self.name)
 
                raise ChangesetDoesNotExistError(msg)
 

	
 
        raise ChangesetDoesNotExistError("Given revision %r not recognized" % revision)
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns ``GitChangeset`` object representing repository's
 
        changeset at the given ``revision``.
 
        """
 
        return self._get_revision(ref_name)
 

	
 
    def _get_archives(self, archive_name='tip'):
 

	
 
        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
 
            yield {"type": i[0], "extension": i[1], "node": archive_name}
 

	
 
    def _get_url(self, url):
 
        """
 
        Returns normalized url. If schema is not given, would fall to
 
        filesystem (``file:///``) schema.
 
        """
 
        url = safe_str(url)
 
        if url != 'default' and '://' not in url:
 
            url = ':///'.join(('file', url))
 
        return url
 

	
 
    def get_hook_location(self):
 
        """
 
        returns absolute path to location where hooks are stored
 
        """
 
        loc = os.path.join(self.path, 'hooks')
 
        if not self.bare:
 
            loc = os.path.join(self.path, '.git', 'hooks')
 
        return loc
 

	
 
    @LazyProperty
 
    def name(self):
 
        return os.path.basename(self.path)
 

	
 
    @LazyProperty
 
    def last_change(self):
 
        """
 
        Returns last change made on this repository as datetime object
 
        """
 
        return date_fromtimestamp(self._get_mtime(), makedate()[1])
 

	
 
    def _get_mtime(self):
 
        try:
 
            return time.mktime(self.get_changeset().date.timetuple())
 
        except RepositoryError:
 
            idx_loc = '' if self.bare else '.git'
 
            # fallback to filesystem
 
            in_path = os.path.join(self.path, idx_loc, "index")
 
            he_path = os.path.join(self.path, idx_loc, "HEAD")
 
            if os.path.exists(in_path):
 
                return os.stat(in_path).st_mtime
 
            else:
 
                return os.stat(he_path).st_mtime
 

	
 
    @LazyProperty
 
    def description(self):
 
        return safe_unicode(self._repo.get_description() or b'unknown')
 

	
 
    @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 = [(key, ascii_str(sha))
 
                     for key, (sha, type_) in self._parsed_refs.iteritems() if type_ == b'H']
 
        return OrderedDict(sorted(_branches, key=sortkey, reverse=False))
 

	
 
    @LazyProperty
 
    def closed_branches(self):
 
        return {}
 

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

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

	
 
        sortkey = lambda ctx: ctx[0]
 
        _tags = [(key, ascii_str(sha))
 
                 for key, (sha, type_) in self._parsed_refs.iteritems() if type_ == b'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[b"refs/tags/%s" % safe_bytes(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)
 
        # self._repo.refs is a DiskRefsContainer, and .path gives the full absolute path of '.git'
 
        tagpath = os.path.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 as e:
 
            raise RepositoryError(e.strerror)
 

	
 
    @LazyProperty
 
    def bookmarks(self):
 
        """
 
        Gets bookmarks for this repository
 
        """
 
        return {}
 

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

	
 
    def _get_parsed_refs(self):
 
        """Return refs as a dict, like:
 
        { b'v0.2.0': [b'599ba911aa24d2981225f3966eb659dfae9e9f30', b'T'] }
 
        """
 
        _repo = self._repo
 
        refs = _repo.get_refs()
 
        keys = [(b'refs/heads/', b'H'),
 
                (b'refs/remotes/origin/', b'RH'),
 
                (b'refs/tags/', b'T')]
 
        _refs = {}
 
        for ref, sha in refs.iteritems():
 
            for k, type_ in keys:
 
                if ref.startswith(k):
 
                    _key = ref[len(k):]
 
                    if type_ == b'T':
 
                        obj = _repo.get_object(sha)
 
                        if isinstance(obj, Tag):
 
                            sha = _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 [b'refs/heads/', b'refs/remotes/origin/']:
 
                if key.startswith(ref_key):
 
                    n = key[len(ref_key):]
 
                    if n not in [b'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
 
        return GitChangeset(repository=self, revision=self._get_revision(revision))
 

	
 
    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).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 bytes 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, cwd=self.path)
 
        # If we used 'show' command, strip first few lines (until actual diff
 
        # starts)
 
        if rev1 == self.EMPTY_CHANGESET:
 
            parts = stdout.split(b'\ndiff ', 1)
 
            if len(parts) > 1:
 
                stdout = b'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 = 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
 

	
 
        for config in gen_configs():
 
            try:
 
                return config.get(section, name)
 
            except KeyError:
 
                continue
 
        return None
 

	
 
    def get_user_name(self, config_file=None):
 
        """
 
        Returns user's name 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)
 
        """
 
        return self.get_config_value('user', 'name', config_file)
 

	
 
    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)
 
        """
 
        return self.get_config_value('user', 'email', config_file)
kallithea/lib/vcs/backends/hg/repository.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    vcs.backends.hg.repository
 
    ~~~~~~~~~~~~~~~~~~~~~~~~~~
 

	
 
    Mercurial repository implementation.
 

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

	
 
import datetime
 
import logging
 
import os
 
import time
 
import urllib
 
import urllib2
 
from collections import OrderedDict
 

	
 
import mercurial.commands
 
import mercurial.error
 
import mercurial.exchange
 
import mercurial.hg
 
import mercurial.hgweb
 
import mercurial.httppeer
 
import mercurial.match
 
import mercurial.mdiff
 
import mercurial.node
 
import mercurial.patch
 
import mercurial.scmutil
 
import mercurial.sshpeer
 
import mercurial.tags
 
import mercurial.ui
 
import mercurial.url
 
import mercurial.util
 

	
 
from kallithea.lib.vcs.backends.base import BaseRepository, CollectionGenerator
 
from kallithea.lib.vcs.exceptions import (
 
    BranchDoesNotExistError, ChangesetDoesNotExistError, EmptyRepositoryError, RepositoryError, TagAlreadyExistError, TagDoesNotExistError, VCSError)
 
from kallithea.lib.vcs.utils import ascii_str, author_email, author_name, date_fromtimestamp, makedate, safe_bytes, safe_str, safe_unicode
 
from kallithea.lib.vcs.utils.lazy import LazyProperty
 
from kallithea.lib.vcs.utils.paths import abspath
 

	
 
from .changeset import MercurialChangeset
 
from .inmemory import MercurialInMemoryChangeset
 
from .workdir import MercurialWorkdir
 

	
 

	
 
log = logging.getLogger(__name__)
 

	
 

	
 
class MercurialRepository(BaseRepository):
 
    """
 
    Mercurial repository backend
 
    """
 
    DEFAULT_BRANCH_NAME = 'default'
 
    scm = 'hg'
 

	
 
    def __init__(self, repo_path, create=False, baseui=None, src_url=None,
 
                 update_after_clone=False):
 
        """
 
        Raises RepositoryError if repository could not be find at the given
 
        ``repo_path``.
 

	
 
        :param repo_path: local path of the repository
 
        :param create=False: if set to True, would try to create repository if
 
           it does not exist rather than raising exception
 
        :param baseui=None: user data
 
        :param src_url=None: would try to clone repository from given location
 
        :param update_after_clone=False: sets update of working copy after
 
          making a clone
 
        """
 

	
 
        if not isinstance(repo_path, str):
 
            raise VCSError('Mercurial backend requires repository path to '
 
                           'be instance of <str> got %s instead' %
 
                           type(repo_path))
 
        self.path = abspath(repo_path)
 
        self.baseui = baseui or mercurial.ui.ui()
 
        # We've set path and ui, now we can set _repo itself
 
        self._repo = self._get_repo(create, src_url, update_after_clone)
 

	
 
    @property
 
    def _empty(self):
 
        """
 
        Checks if repository is empty ie. without any changesets
 
        """
 
        # TODO: Following raises errors when using InMemoryChangeset...
 
        # return len(self._repo.changelog) == 0
 
        return len(self.revisions) == 0
 

	
 
    @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()
 

	
 
    @LazyProperty
 
    def name(self):
 
        return os.path.basename(self.path)
 

	
 
    @LazyProperty
 
    def branches(self):
 
        return self._get_branches()
 

	
 
    @LazyProperty
 
    def closed_branches(self):
 
        return self._get_branches(normal=False, closed=True)
 

	
 
    @LazyProperty
 
    def allbranches(self):
 
        """
 
        List all branches, including closed branches.
 
        """
 
        return self._get_branches(closed=True)
 

	
 
    def _get_branches(self, normal=True, closed=False):
 
        """
 
        Gets branches for this repository
 
        Returns only not closed branches by default
 

	
 
        :param closed: return also closed branches for mercurial
 
        :param normal: return also normal branches
 
        """
 

	
 
        if self._empty:
 
            return {}
 

	
 
        bt = OrderedDict()
 
        for bn, _heads, node, isclosed in sorted(self._repo.branchmap().iterbranches()):
 
            if isclosed:
 
                if closed:
 
                    bt[safe_unicode(bn)] = ascii_str(mercurial.node.hex(node))
 
            else:
 
                if normal:
 
                    bt[safe_unicode(bn)] = ascii_str(mercurial.node.hex(node))
 
        return bt
 

	
 
    @LazyProperty
 
    def tags(self):
 
        """
 
        Gets tags for this repository
 
        """
 
        return self._get_tags()
 

	
 
    def _get_tags(self):
 
        if self._empty:
 
            return {}
 

	
 
        return OrderedDict(sorted(
 
            ((safe_unicode(n), ascii_str(mercurial.node.hex(h))) for n, h in self._repo.tags().items()),
 
            reverse=True,
 
            key=lambda x: x[0],  # sort by name
 
        ))
 

	
 
    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)
 
        local = kwargs.setdefault('local', False)
 

	
 
        if message is None:
 
            message = "Added tag %s for changeset %s" % (name,
 
                changeset.short_id)
 

	
 
        if date is None:
 
            date = safe_bytes(datetime.datetime.now().strftime('%a, %d %b %Y %H:%M:%S'))
 

	
 
        try:
 
            mercurial.tags.tag(self._repo, safe_bytes(name), changeset._ctx.node(), safe_bytes(message), local, safe_bytes(user), date)
 
        except mercurial.error.Abort as e:
 
            raise RepositoryError(e.message)
 

	
 
        # Reinitialize tags
 
        self.tags = self._get_tags()
 
        tag_id = self.tags[name]
 

	
 
        return self.get_changeset(revision=tag_id)
 

	
 
    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)
 
        if message is None:
 
            message = "Removed tag %s" % name
 
        if date is None:
 
            date = safe_bytes(datetime.datetime.now().strftime('%a, %d %b %Y %H:%M:%S'))
 
        local = False
 

	
 
        try:
 
            mercurial.tags.tag(self._repo, safe_bytes(name), mercurial.commands.nullid, safe_bytes(message), local, safe_bytes(user), date)
 
            self.tags = self._get_tags()
 
        except mercurial.error.Abort as e:
 
            raise RepositoryError(e.message)
 

	
 
    @LazyProperty
 
    def bookmarks(self):
 
        """
 
        Gets bookmarks for this repository
 
        """
 
        return self._get_bookmarks()
 

	
 
    def _get_bookmarks(self):
 
        if self._empty:
 
            return {}
 

	
 
        return OrderedDict(sorted(
 
            ((safe_unicode(n), ascii_str(h)) for n, h in self._repo._bookmarks.items()),
 
            reverse=True,
 
            key=lambda x: x[0],  # sort by name
 
        ))
 

	
 
    def _get_all_revisions(self):
 
        return [ascii_str(self._repo[x].hex()) for x in self._repo.filtered(b'visible').changelog.revs()]
 

	
 
    def get_diff(self, rev1, rev2, path='', 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``. If negative value is passed-in, it will be
 
          set to ``0`` instead.
 
        """
 

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

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

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

	
 
        # Check if given revisions are present at repository (may raise
 
        # ChangesetDoesNotExistError)
 
        if rev1 != self.EMPTY_CHANGESET:
 
            self.get_changeset(rev1)
 
        self.get_changeset(rev2)
 
        if path:
 
            file_filter = mercurial.match.exact(path)
 
        else:
 
            file_filter = None
 

	
 
        return b''.join(mercurial.patch.diff(self._repo, rev1, rev2, match=file_filter,
 
                          opts=mercurial.mdiff.diffopts(git=True,
 
                                        showfunc=True,
 
                                        ignorews=ignore_whitespace,
 
                                        context=context)))
 

	
 
    @classmethod
 
    def _check_url(cls, url, repoui=None):
 
        """
 
        Function will check given url and try to verify if it's a valid
 
        link. Sometimes it may happened that mercurial 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(b'file:'):
 
            return True
 

	
 
        if url.startswith(b'ssh:'):
 
            # in case of invalid uri or authentication issues, sshpeer will
 
            # throw an exception.
 
            mercurial.sshpeer.instance(repoui or mercurial.ui.ui(), url, False).lookup(b'tip')
 
            return True
 

	
 
        url_prefix = None
 
        if b'+' in url[:url.find(b'://')]:
 
            url_prefix, url = url.split(b'+', 1)
 

	
 
        handlers = []
 
        url_obj = mercurial.util.url(url)
 
        test_uri, authinfo = url_obj.authinfo()
 
        url_obj.passwd = b'*****'
 
        cleaned_uri = str(url_obj)
 

	
 
        if authinfo:
 
            # create a password manager
 
            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
 
            passmgr.add_password(*authinfo)
 

	
 
            handlers.extend((mercurial.url.httpbasicauthhandler(passmgr),
 
                             mercurial.url.httpdigestauthhandler(passmgr)))
 

	
 
        o = urllib2.build_opener(*handlers)
 
        o.addheaders = [('Content-Type', 'application/mercurial-0.1'),
 
                        ('Accept', 'application/mercurial-0.1')]
 

	
 
        req = urllib2.Request(
 
            "%s?%s" % (
 
                test_uri,
 
                urllib.urlencode({
 
                    'cmd': 'between',
 
                    'pairs': "%s-%s" % ('0' * 40, '0' * 40),
 
                })
 
            ))
 

	
 
        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))
 

	
 
        if not url_prefix: # skip svn+http://... (and git+... too)
 
            # now check if it's a proper hg repo
 
            try:
 
                mercurial.httppeer.instance(repoui or mercurial.ui.ui(), url, False).lookup(b'tip')
 
            except Exception as e:
 
                raise urllib2.URLError(
 
                    "url [%s] does not look like an hg repo org_exc: %s"
 
                    % (cleaned_uri, e))
 

	
 
        return True
 

	
 
    def _get_repo(self, create, src_url=None, update_after_clone=False):
 
        """
 
        Function will check for mercurial repository in given path and return
 
        a localrepo object. If there is no repository in that path it will
 
        raise an exception unless ``create`` parameter is set to True - in
 
        that case repository would be created and returned.
 
        If ``src_url`` is given, would try to clone repository from the
 
        location at given clone_point. Additionally it'll make update to
 
        working copy accordingly to ``update_after_clone`` flag
 
        """
 
        try:
 
            if src_url:
 
                url = safe_bytes(self._get_url(src_url))
 
                opts = {}
 
                if not update_after_clone:
 
                    opts.update({'noupdate': True})
 
                MercurialRepository._check_url(url, self.baseui)
 
                mercurial.commands.clone(self.baseui, url, safe_bytes(self.path), **opts)
 

	
 
                # Don't try to create if we've already cloned repo
 
                create = False
 
            return mercurial.localrepo.instance(self.baseui, safe_bytes(self.path), create=create)
 
        except (mercurial.error.Abort, mercurial.error.RepoError) as err:
 
            if create:
 
                msg = "Cannot create repository at %s. Original error was %s" \
 
                    % (self.name, err)
 
            else:
 
                msg = "Not valid repository at %s. Original error was %s" \
 
                    % (self.name, err)
 
            raise RepositoryError(msg)
 

	
 
    @LazyProperty
 
    def in_memory_changeset(self):
 
        return MercurialInMemoryChangeset(self)
 

	
 
    @LazyProperty
 
    def description(self):
 
        _desc = self._repo.ui.config(b'web', b'description', None, untrusted=True)
 
        return safe_unicode(_desc or b'unknown')
 

	
 
    @LazyProperty
 
    def contact(self):
 
        return safe_unicode(mercurial.hgweb.common.get_contact(self._repo.ui.config)
 
                            or b'Unknown')
 

	
 
    @LazyProperty
 
    def last_change(self):
 
        """
 
        Returns last change made on this repository as datetime object
 
        """
 
        return date_fromtimestamp(self._get_mtime(), makedate()[1])
 

	
 
    def _get_mtime(self):
 
        try:
 
            return time.mktime(self.get_changeset().date.timetuple())
 
        except RepositoryError:
 
            # fallback to filesystem
 
            cl_path = os.path.join(self.path, '.hg', "00changelog.i")
 
            st_path = os.path.join(self.path, '.hg', "store")
 
            if os.path.exists(cl_path):
 
                return os.stat(cl_path).st_mtime
 
            else:
 
                return os.stat(st_path).st_mtime
 

	
 
    def _get_revision(self, revision):
 
        """
 
        Given any revision identifier, returns a 40 char string with revision hash.
 

	
 
        :param revision: str or int or None
 
        """
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in [-1, None]:
 
            revision = b'tip'
 
        elif isinstance(revision, unicode):
 
            revision = safe_bytes(revision)
 

	
 
        try:
 
            if isinstance(revision, int):
 
                return ascii_str(self._repo[revision].hex())
 
            return ascii_str(mercurial.scmutil.revsymbol(self._repo, revision).hex())
 
        except (IndexError, ValueError, mercurial.error.RepoLookupError, TypeError):
 
            msg = "Revision %r does not exist for %s" % (safe_unicode(revision), self.name)
 
            raise ChangesetDoesNotExistError(msg)
 
        except (LookupError, ):
 
            msg = "Ambiguous identifier `%s` for %s" % (safe_unicode(revision), self.name)
 
            raise ChangesetDoesNotExistError(msg)
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns revision number for the given reference.
 
        """
 
        ref_name = safe_str(ref_name)
 
        if ref_type == 'rev' and not ref_name.strip('0'):
 
            return self.EMPTY_CHANGESET
 
        # lookup up the exact node id
 
        _revset_predicates = {
 
                'branch': 'branch',
 
                'book': 'bookmark',
 
                'tag': 'tag',
 
                'rev': 'id',
 
            }
 
        # avoid expensive branch(x) iteration over whole repo
 
        rev_spec = "%%s & %s(%%s)" % _revset_predicates[ref_type]
 
        try:
 
            revs = self._repo.revs(rev_spec, ref_name, ref_name)
 
        except LookupError:
 
            msg = "Ambiguous identifier %s:%s for %s" % (ref_type, ref_name, self.name)
 
            raise ChangesetDoesNotExistError(msg)
 
        except mercurial.error.RepoLookupError:
 
            msg = "Revision %s:%s does not exist for %s" % (ref_type, ref_name, self.name)
 
            raise ChangesetDoesNotExistError(msg)
 
        if revs:
 
            revision = revs.last()
 
        else:
 
            # TODO: just report 'not found'?
 
            revision = ref_name
 

	
 
        return self._get_revision(revision)
 

	
 
    def _get_archives(self, archive_name='tip'):
 
        allowed = self.baseui.configlist(b"web", b"allow_archive",
 
                                         untrusted=True)
 
        for name, ext in [(b'zip', '.zip'), (b'gz', '.tar.gz'), (b'bz2', '.tar.bz2')]:
 
            if name in allowed or self._repo.ui.configbool(b"web",
 
                                                           b"allow" + name,
 
                                                           untrusted=True):
 
                yield {"type": name, "extension": ext, "node": archive_name}
 

	
 
    def _get_url(self, url):
 
        """
 
        Returns normalized url. If schema is not given, would fall
 
        to filesystem
 
        (``file:///``) schema.
 
        """
 
        url = safe_str(url)
 
        if url != 'default' and '://' not in url:
 
            url = "file:" + urllib.pathname2url(url)
 
        return url
 

	
 
    def get_hook_location(self):
 
        """
 
        returns absolute path to location where hooks are stored
 
        """
 
        return os.path.join(self.path, '.hg', '.hgrc')
 

	
 
    def get_changeset(self, revision=None):
 
        """
 
        Returns ``MercurialChangeset`` object representing repository's
 
        changeset at the given ``revision``.
 
        """
 
        return MercurialChangeset(repository=self, revision=self._get_revision(revision))
 

	
 
    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 ``MercurialChangeset`` objects from start to end
 
        (both are inclusive)
 

	
 
        :param start: None, str, int or mercurial lookup format
 
        :param end:  None, str, int or mercurial lookup format
 
        :param start_date:
 
        :param end_date:
 
        :param branch_name:
 
        :param reversed: return changesets in reversed order
 
        """
 
        start_raw_id = self._get_revision(start)
 
        start_pos = None if start is None else self.revisions.index(start_raw_id)
 
        end_raw_id = self._get_revision(end)
 
        end_pos = None if end is None else self.revisions.index(end_raw_id)
 

	
 
        if start_pos is not None and end_pos is not None and start_pos > end_pos:
 
            raise RepositoryError("Start revision '%s' cannot be "
 
                                  "after end revision '%s'" % (start, end))
 

	
 
        if branch_name and branch_name not in self.allbranches:
 
            msg = "Branch %r not found in %s" % (branch_name, self.name)
 
            raise BranchDoesNotExistError(msg)
 
        if end_pos is not None:
 
            end_pos += 1
 
        # filter branches
 
        filter_ = []
 
        if branch_name:
 
            filter_.append(b'branch("%s")' % safe_bytes(branch_name))
 
        if start_date:
 
            filter_.append(b'date(">%s")' % safe_bytes(str(start_date)))
 
        if end_date:
 
            filter_.append(b'date("<%s")' % safe_bytes(str(end_date)))
 
        if filter_ or max_revisions:
 
            if filter_:
 
                revspec = b' and '.join(filter_)
 
            else:
 
                revspec = b'all()'
 
            if max_revisions:
 
                revspec = b'limit(%s, %d)' % (revspec, max_revisions)
 
            revisions = mercurial.scmutil.revrange(self._repo, [revspec])
 
        else:
 
            revisions = self.revisions
 

	
 
        # this is very much a hack to turn this into a list; a better solution
 
        # would be to get rid of this function entirely and use revsets
 
        revs = list(revisions)[start_pos:end_pos]
 
        if reverse:
 
            revs.reverse()
 

	
 
        return CollectionGenerator(self, revs)
 

	
 
    def pull(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        other = mercurial.hg.peer(self._repo, {}, safe_bytes(self._get_url(url)))
 
        try:
 
            mercurial.exchange.pull(self._repo, other, heads=None, force=None)
 
        except mercurial.error.Abort as err:
 
            # Propagate error but with vcs's type
 
            raise RepositoryError(str(err))
 

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

	
 
    def get_config_value(self, section, name=None, 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]
 

	
 
        config = self._repo.ui
 
        if config_file:
 
            config = mercurial.ui.ui()
 
            for path in config_file:
 
                config.readconfig(safe_bytes(path))
 
        return config.config(safe_bytes(section), safe_bytes(name))
 

	
 
    def get_user_name(self, config_file=None):
 
        """
 
        Returns user's name 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)
 
        """
 
        username = self.get_config_value('ui', 'username', config_file=config_file)
 
        if username:
 
            return author_name(username)
 
        return None
 

	
 
    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)
 
        """
 
        username = self.get_config_value('ui', 'username', config_file=config_file)
 
        if username:
 
            return author_email(username)
 
        return None
kallithea/tests/vcs/test_git.py
Show inline comments
 
import datetime
 
import os
 
import sys
 
import urllib2
 

	
 
import mock
 
import pytest
 

	
 
from kallithea.lib.vcs.backends.git import GitChangeset, GitRepository
 
from kallithea.lib.vcs.exceptions import NodeDoesNotExistError, RepositoryError, VCSError
 
from kallithea.lib.vcs.nodes import DirNode, FileNode, NodeKind, NodeState
 
from kallithea.model.scm import ScmModel
 
from kallithea.tests.vcs.base import _BackendTestMixin
 
from kallithea.tests.vcs.conf import TEST_GIT_REPO, TEST_GIT_REPO_CLONE, TESTS_TMP_PATH, get_new_dir
 

	
 

	
 
class TestGitRepository(object):
 

	
 
    def __check_for_existing_repo(self):
 
        if os.path.exists(TEST_GIT_REPO_CLONE):
 
            pytest.fail('Cannot test git clone repo as location %s already '
 
                      'exists. You should manually remove it first.'
 
                      % TEST_GIT_REPO_CLONE)
 

	
 
    def setup_method(self):
 
        self.repo = GitRepository(TEST_GIT_REPO)
 

	
 
    def test_wrong_repo_path(self):
 
        wrong_repo_path = os.path.join(TESTS_TMP_PATH, 'errorrepo')
 
        with pytest.raises(RepositoryError):
 
            GitRepository(wrong_repo_path)
 

	
 
    def test_git_cmd_injection(self):
 
        repo_inject_path = TEST_GIT_REPO + '; echo "Cake";'
 
        with pytest.raises(urllib2.URLError):
 
            # Should fail because URL will contain the parts after ; too
 
            GitRepository(get_new_dir('injection-repo'), src_url=repo_inject_path, update_after_clone=True, create=True)
 

	
 
        with pytest.raises(RepositoryError):
 
            # Should fail on direct clone call, which as of this writing does not happen outside of class
 
            clone_fail_repo = GitRepository(get_new_dir('injection-repo'), create=True)
 
            clone_fail_repo.clone(repo_inject_path, update_after_clone=True,)
 

	
 
        # Verify correct quoting of evil characters that should work on posix file systems
 
        if sys.platform == 'win32':
 
            # windows does not allow '"' in dir names
 
            # and some versions of the git client don't like ` and '
 
            tricky_path = get_new_dir("tricky-path-repo-$")
 
        else:
 
            tricky_path = get_new_dir("tricky-path-repo-$'\"`")
 
        successfully_cloned = GitRepository(tricky_path, src_url=TEST_GIT_REPO, update_after_clone=True, create=True)
 
        # Repo should have been created
 
        assert not successfully_cloned._repo.bare
 

	
 
        if sys.platform == 'win32':
 
            # windows does not allow '"' in dir names
 
            # and some versions of the git client don't like ` and '
 
            tricky_path_2 = get_new_dir("tricky-path-2-repo-$")
 
        else:
 
            tricky_path_2 = get_new_dir("tricky-path-2-repo-$'\"`")
 
        successfully_cloned2 = GitRepository(tricky_path_2, src_url=tricky_path, bare=True, create=True)
 
        # Repo should have been created and thus used correct quoting for clone
 
        assert successfully_cloned2._repo.bare
 

	
 
        # Should pass because URL has been properly quoted
 
        successfully_cloned.pull(tricky_path_2)
 
        successfully_cloned2.fetch(tricky_path)
 

	
 
    def test_repo_create_with_spaces_in_path(self):
 
        repo_path = get_new_dir("path with spaces")
 
        repo = GitRepository(repo_path, src_url=None, bare=True, create=True)
 
        # Repo should have been created
 
        assert repo._repo.bare
 

	
 
    def test_repo_clone(self):
 
        self.__check_for_existing_repo()
 
        repo = GitRepository(TEST_GIT_REPO)
 
        repo_clone = GitRepository(TEST_GIT_REPO_CLONE,
 
            src_url=TEST_GIT_REPO, create=True, update_after_clone=True)
 
        assert len(repo.revisions) == len(repo_clone.revisions)
 
        # Checking hashes of changesets should be enough
 
        for changeset in repo.get_changesets():
 
            raw_id = changeset.raw_id
 
            assert raw_id == repo_clone.get_changeset(raw_id).raw_id
 

	
 
    def test_repo_clone_with_spaces_in_path(self):
 
        repo_path = get_new_dir("path with spaces")
 
        successfully_cloned = GitRepository(repo_path, src_url=TEST_GIT_REPO, update_after_clone=True, create=True)
 
        # Repo should have been created
 
        assert not successfully_cloned._repo.bare
 

	
 
        successfully_cloned.pull(TEST_GIT_REPO)
 
        self.repo.fetch(repo_path)
 

	
 
    def test_repo_clone_without_create(self):
 
        with pytest.raises(RepositoryError):
 
            GitRepository(TEST_GIT_REPO_CLONE + '_wo_create', src_url=TEST_GIT_REPO)
 

	
 
    def test_repo_clone_with_update(self):
 
        repo = GitRepository(TEST_GIT_REPO)
 
        clone_path = TEST_GIT_REPO_CLONE + '_with_update'
 
        repo_clone = GitRepository(clone_path,
 
            create=True, src_url=TEST_GIT_REPO, update_after_clone=True)
 
        assert len(repo.revisions) == len(repo_clone.revisions)
 

	
 
        # check if current workdir was updated
 
        fpath = os.path.join(clone_path, 'MANIFEST.in')
 
        assert os.path.isfile(fpath) == True, 'Repo was cloned and updated but file %s could not be found' % fpath
 

	
 
    def test_repo_clone_without_update(self):
 
        repo = GitRepository(TEST_GIT_REPO)
 
        clone_path = TEST_GIT_REPO_CLONE + '_without_update'
 
        repo_clone = GitRepository(clone_path,
 
            create=True, src_url=TEST_GIT_REPO, update_after_clone=False)
 
        assert len(repo.revisions) == len(repo_clone.revisions)
 
        # check if current workdir was *NOT* updated
 
        fpath = os.path.join(clone_path, 'MANIFEST.in')
 
        # Make sure it's not bare repo
 
        assert not repo_clone._repo.bare
 
        assert os.path.isfile(fpath) == False, 'Repo was cloned and updated but file %s was found' % fpath
 

	
 
    def test_repo_clone_into_bare_repo(self):
 
        repo = GitRepository(TEST_GIT_REPO)
 
        clone_path = TEST_GIT_REPO_CLONE + '_bare.git'
 
        repo_clone = GitRepository(clone_path, create=True,
 
            src_url=repo.path, bare=True)
 
        assert repo_clone._repo.bare
 

	
 
    def test_create_repo_is_not_bare_by_default(self):
 
        repo = GitRepository(get_new_dir('not-bare-by-default'), create=True)
 
        assert not repo._repo.bare
 

	
 
    def test_create_bare_repo(self):
 
        repo = GitRepository(get_new_dir('bare-repo'), create=True, bare=True)
 
        assert repo._repo.bare
 

	
 
    def test_revisions(self):
 
        # there are 112 revisions (by now)
 
        # so we can assume they would be available from now on
 
        subset = set([
 
            'c1214f7e79e02fc37156ff215cd71275450cffc3',
 
            '38b5fe81f109cb111f549bfe9bb6b267e10bc557',
 
            'fa6600f6848800641328adbf7811fd2372c02ab2',
 
            '102607b09cdd60e2793929c4f90478be29f85a17',
 
            '49d3fd156b6f7db46313fac355dca1a0b94a0017',
 
            '2d1028c054665b962fa3d307adfc923ddd528038',
 
            'd7e0d30fbcae12c90680eb095a4f5f02505ce501',
 
            'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
 
            'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
 
            '8430a588b43b5d6da365400117c89400326e7992',
 
            'd955cd312c17b02143c04fa1099a352b04368118',
 
            'f67b87e5c629c2ee0ba58f85197e423ff28d735b',
 
            'add63e382e4aabc9e1afdc4bdc24506c269b7618',
 
            'f298fe1189f1b69779a4423f40b48edf92a703fc',
 
            'bd9b619eb41994cac43d67cf4ccc8399c1125808',
 
            '6e125e7c890379446e98980d8ed60fba87d0f6d1',
 
            'd4a54db9f745dfeba6933bf5b1e79e15d0af20bd',
 
            '0b05e4ed56c802098dfc813cbe779b2f49e92500',
 
            '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
 
            '45223f8f114c64bf4d6f853e3c35a369a6305520',
 
            'ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
 
            'f5ea29fc42ef67a2a5a7aecff10e1566699acd68',
 
            '27d48942240f5b91dfda77accd2caac94708cc7d',
 
            '622f0eb0bafd619d2560c26f80f09e3b0b0d78af',
 
            'e686b958768ee96af8029fe19c6050b1a8dd3b2b'])
 
        assert subset.issubset(set(self.repo.revisions))
 

	
 
    def test_slicing(self):
 
        # 4 1 5 10 95
 
        for sfrom, sto, size in [(0, 4, 4), (1, 2, 1), (10, 15, 5),
 
                                 (10, 20, 10), (5, 100, 95)]:
 
            revs = list(self.repo[sfrom:sto])
 
            assert len(revs) == size
 
            assert revs[0] == self.repo.get_changeset(sfrom)
 
            assert revs[-1] == self.repo.get_changeset(sto - 1)
 

	
 
    def test_branches(self):
 
        # TODO: Need more tests here
 
        # Removed (those are 'remotes' branches for cloned repo)
 
        #assert 'master' in self.repo.branches
 
        #assert 'gittree' in self.repo.branches
 
        #assert 'web-branch' in self.repo.branches
 
        for name, id in self.repo.branches.items():
 
            assert isinstance(self.repo.get_changeset(id), GitChangeset)
 

	
 
    def test_tags(self):
 
        # TODO: Need more tests here
 
        assert 'v0.1.1' in self.repo.tags
 
        assert 'v0.1.2' in self.repo.tags
 
        for name, id in self.repo.tags.items():
 
            assert isinstance(self.repo.get_changeset(id), GitChangeset)
 

	
 
    def _test_single_changeset_cache(self, revision):
 
        chset = self.repo.get_changeset(revision)
 
        assert revision in self.repo.changesets
 
        assert chset is self.repo.changesets[revision]
 

	
 
    def test_initial_changeset(self):
 
        id = self.repo.revisions[0]
 
        init_chset = self.repo.get_changeset(id)
 
        assert init_chset.message == 'initial import\n'
 
        assert init_chset.author == 'Marcin Kuzminski <marcin@python-blog.com>'
 
        for path in ('vcs/__init__.py',
 
                     'vcs/backends/BaseRepository.py',
 
                     'vcs/backends/__init__.py'):
 
            assert isinstance(init_chset.get_node(path), FileNode)
 
        for path in ('', 'vcs', 'vcs/backends'):
 
            assert isinstance(init_chset.get_node(path), DirNode)
 

	
 
        with pytest.raises(NodeDoesNotExistError):
 
            init_chset.get_node(path='foobar')
 

	
 
        node = init_chset.get_node('vcs/')
 
        assert hasattr(node, 'kind')
 
        assert node.kind == NodeKind.DIR
 

	
 
        node = init_chset.get_node('vcs')
 
        assert hasattr(node, 'kind')
 
        assert node.kind == NodeKind.DIR
 

	
 
        node = init_chset.get_node('vcs/__init__.py')
 
        assert hasattr(node, 'kind')
 
        assert node.kind == NodeKind.FILE
 

	
 
    def test_not_existing_changeset(self):
 
        with pytest.raises(RepositoryError):
 
            self.repo.get_changeset('f' * 40)
 

	
 
    def test_changeset10(self):
 

	
 
        chset10 = self.repo.get_changeset(self.repo.revisions[9])
 
        readme = b"""===
 
VCS
 
===
 

	
 
Various Version Control System management abstraction layer for Python.
 

	
 
Introduction
 
------------
 

	
 
TODO: To be written...
 

	
 
"""
 
        node = chset10.get_node('README.rst')
 
        assert node.kind == NodeKind.FILE
 
        assert node.content == readme
 

	
 

	
 
class TestGitChangeset(object):
 

	
 
    def setup_method(self):
 
        self.repo = GitRepository(TEST_GIT_REPO)
 

	
 
    def test_default_changeset(self):
 
        tip = self.repo.get_changeset()
 
        assert tip == self.repo.get_changeset(None)
 
        assert tip == self.repo.get_changeset('tip')
 

	
 
    def test_root_node(self):
 
        tip = self.repo.get_changeset()
 
        assert tip.root is tip.get_node('')
 

	
 
    def test_lazy_fetch(self):
 
        """
 
        Test if changeset's nodes expands and are cached as we walk through
 
        the revision. This test is somewhat hard to write as order of tests
 
        is a key here. Written by running command after command in a shell.
 
        """
 
        commit_id = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
 
        assert commit_id in self.repo.revisions
 
        chset = self.repo.get_changeset(commit_id)
 
        assert len(chset.nodes) == 0
 
        root = chset.root
 
        assert len(chset.nodes) == 1
 
        assert len(root.nodes) == 8
 
        # accessing root.nodes updates chset.nodes
 
        assert len(chset.nodes) == 9
 

	
 
        docs = root.get_node('docs')
 
        # we haven't yet accessed anything new as docs dir was already cached
 
        assert len(chset.nodes) == 9
 
        assert len(docs.nodes) == 8
 
        # accessing docs.nodes updates chset.nodes
 
        assert len(chset.nodes) == 17
 

	
 
        assert docs is chset.get_node('docs')
 
        assert docs is root.nodes[0]
 
        assert docs is root.dirs[0]
 
        assert docs is chset.get_node('docs')
 

	
 
    def test_nodes_with_changeset(self):
 
        commit_id = '2a13f185e4525f9d4b59882791a2d397b90d5ddc'
 
        chset = self.repo.get_changeset(commit_id)
 
        root = chset.root
 
        docs = root.get_node('docs')
 
        assert docs is chset.get_node('docs')
 
        api = docs.get_node('api')
 
        assert api is chset.get_node('docs/api')
 
        index = api.get_node('index.rst')
 
        assert index is chset.get_node('docs/api/index.rst')
 
        assert index is chset.get_node('docs') \
 
                             .get_node('api') \
 
                             .get_node('index.rst')
 

	
 
    def test_branch_and_tags(self):
 
        # Those tests seem to show wrong results:
 
        # in Git, only heads have a branch - most changesets don't
 
        rev0 = self.repo.revisions[0]
 
        chset0 = self.repo.get_changeset(rev0)
 
        assert chset0.branch is None # should be 'master'?
 
        assert chset0.branches == [] # should be 'master'?
 
        assert chset0.tags == []
 

	
 
        rev10 = self.repo.revisions[10]
 
        chset10 = self.repo.get_changeset(rev10)
 
        assert chset10.branch is None # should be 'master'?
 
        assert chset10.branches == [] # should be 'master'?
 
        assert chset10.tags == []
 

	
 
        rev44 = self.repo.revisions[44]
 
        chset44 = self.repo.get_changeset(rev44)
 
        assert chset44.branch is None # should be 'web-branch'?
 
        assert chset44.branches == [] # should be 'web-branch'?
 

	
 
        tip = self.repo.get_changeset('tip')
 
        assert 'tip' not in tip.tags # it should be?
 
        assert not tip.tags # how it is!
 

	
 
    def _test_slices(self, limit, offset):
 
        count = self.repo.count()
 
        changesets = self.repo.get_changesets(limit=limit, offset=offset)
 
        idx = 0
 
        for changeset in changesets:
 
            rev = offset + idx
 
            idx += 1
 
            rev_id = self.repo.revisions[rev]
 
            if idx > limit:
 
                pytest.fail("Exceeded limit already (getting revision %s, "
 
                    "there are %s total revisions, offset=%s, limit=%s)"
 
                    % (rev_id, count, offset, limit))
 
            assert changeset == self.repo.get_changeset(rev_id)
 
        result = list(self.repo.get_changesets(limit=limit, offset=offset))
 
        start = offset
 
        end = limit and offset + limit or None
 
        sliced = list(self.repo[start:end])
 
        pytest.failUnlessEqual(result, sliced,
 
            msg="Comparison failed for limit=%s, offset=%s"
 
            "(get_changeset returned: %s and sliced: %s"
 
            % (limit, offset, result, sliced))
 

	
 
    def _test_file_size(self, revision, path, size):
 
        node = self.repo.get_changeset(revision).get_node(path)
 
        assert node.is_file()
 
        assert node.size == size
 

	
 
    def test_file_size(self):
 
        to_check = (
 
            ('c1214f7e79e02fc37156ff215cd71275450cffc3',
 
                'vcs/backends/BaseRepository.py', 502),
 
            ('d7e0d30fbcae12c90680eb095a4f5f02505ce501',
 
                'vcs/backends/hg.py', 854),
 
            ('6e125e7c890379446e98980d8ed60fba87d0f6d1',
 
                'setup.py', 1068),
 
            ('d955cd312c17b02143c04fa1099a352b04368118',
 
                'vcs/backends/base.py', 2921),
 
            ('ca1eb7957a54bce53b12d1a51b13452f95bc7c7e',
 
                'vcs/backends/base.py', 3936),
 
            ('f50f42baeed5af6518ef4b0cb2f1423f3851a941',
 
                'vcs/backends/base.py', 6189),
 
        )
 
        for revision, path, size in to_check:
 
            self._test_file_size(revision, path, size)
 

	
 
    def _test_dir_size(self, revision, path, size):
 
        node = self.repo.get_changeset(revision).get_node(path)
 
        assert node.size == size
 

	
 
    def test_dir_size(self):
 
        to_check = (
 
            ('5f2c6ee195929b0be80749243c18121c9864a3b3', '/', 674076),
 
            ('7ab37bc680b4aa72c34d07b230c866c28e9fc204', '/', 674049),
 
            ('6892503fb8f2a552cef5f4d4cc2cdbd13ae1cd2f', '/', 671830),
 
        )
 
        for revision, path, size in to_check:
 
            self._test_dir_size(revision, path, size)
 

	
 
    def test_repo_size(self):
 
        assert self.repo.size == 674076
 

	
 
    def test_file_history(self):
 
        # we can only check if those revisions are present in the history
 
        # as we cannot update this test every time file is changed
 
        files = {
 
            'setup.py': [
 
                '54386793436c938cff89326944d4c2702340037d',
 
                '51d254f0ecf5df2ce50c0b115741f4cf13985dab',
 
                '998ed409c795fec2012b1c0ca054d99888b22090',
 
                '5e0eb4c47f56564395f76333f319d26c79e2fb09',
 
                '0115510b70c7229dbc5dc49036b32e7d91d23acd',
 
                '7cb3fd1b6d8c20ba89e2264f1c8baebc8a52d36e',
 
                '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
 
                '191caa5b2c81ed17c0794bf7bb9958f4dcb0b87e',
 
                'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
 
            ],
 
            'vcs/nodes.py': [
 
                '33fa3223355104431402a888fa77a4e9956feb3e',
 
                'fa014c12c26d10ba682fadb78f2a11c24c8118e1',
 
                'e686b958768ee96af8029fe19c6050b1a8dd3b2b',
 
                'ab5721ca0a081f26bf43d9051e615af2cc99952f',
 
                'c877b68d18e792a66b7f4c529ea02c8f80801542',
 
                '4313566d2e417cb382948f8d9d7c765330356054',
 
                '6c2303a793671e807d1cfc70134c9ca0767d98c2',
 
                '54386793436c938cff89326944d4c2702340037d',
 
                '54000345d2e78b03a99d561399e8e548de3f3203',
 
                '1c6b3677b37ea064cb4b51714d8f7498f93f4b2b',
 
                '2d03ca750a44440fb5ea8b751176d1f36f8e8f46',
 
                '2a08b128c206db48c2f0b8f70df060e6db0ae4f8',
 
                '30c26513ff1eb8e5ce0e1c6b477ee5dc50e2f34b',
 
                'ac71e9503c2ca95542839af0ce7b64011b72ea7c',
 
                '12669288fd13adba2a9b7dd5b870cc23ffab92d2',
 
                '5a0c84f3e6fe3473e4c8427199d5a6fc71a9b382',
 
                '12f2f5e2b38e6ff3fbdb5d722efed9aa72ecb0d5',
 
                '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
 
                'f50f42baeed5af6518ef4b0cb2f1423f3851a941',
 
                'd7e390a45f6aa96f04f5e7f583ad4f867431aa25',
 
                'f15c21f97864b4f071cddfbf2750ec2e23859414',
 
                'e906ef056cf539a4e4e5fc8003eaf7cf14dd8ade',
 
                'ea2b108b48aa8f8c9c4a941f66c1a03315ca1c3b',
 
                '84dec09632a4458f79f50ddbbd155506c460b4f9',
 
                '0115510b70c7229dbc5dc49036b32e7d91d23acd',
 
                '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
 
                '3bf1c5868e570e39569d094f922d33ced2fa3b2b',
 
                'b8d04012574729d2c29886e53b1a43ef16dd00a1',
 
                '6970b057cffe4aab0a792aa634c89f4bebf01441',
 
                'dd80b0f6cf5052f17cc738c2951c4f2070200d7f',
 
                'ff7ca51e58c505fec0dd2491de52c622bb7a806b',
 
            ],
 
            'vcs/backends/git.py': [
 
                '4cf116ad5a457530381135e2f4c453e68a1b0105',
 
                '9a751d84d8e9408e736329767387f41b36935153',
 
                'cb681fb539c3faaedbcdf5ca71ca413425c18f01',
 
                '428f81bb652bcba8d631bce926e8834ff49bdcc6',
 
                '180ab15aebf26f98f714d8c68715e0f05fa6e1c7',
 
                '2b8e07312a2e89e92b90426ab97f349f4bce2a3a',
 
                '50e08c506174d8645a4bb517dd122ac946a0f3bf',
 
                '54000345d2e78b03a99d561399e8e548de3f3203',
 
            ],
 
        }
 
        for path, revs in files.items():
 
            node = self.repo.get_changeset(revs[0]).get_node(path)
 
            node_revs = [chset.raw_id for chset in node.history]
 
            assert set(revs).issubset(set(node_revs)), "We assumed that %s is subset of revisions for which file %s " \
 
                "has been changed, and history of that node returned: %s" \
 
                % (revs, path, node_revs)
 

	
 
    def test_file_annotate(self):
 
        files = {
 
            'vcs/backends/__init__.py': {
 
                'c1214f7e79e02fc37156ff215cd71275450cffc3': {
 
                    'lines_no': 1,
 
                    'changesets': [
 
                        'c1214f7e79e02fc37156ff215cd71275450cffc3',
 
                    ],
 
                },
 
                '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647': {
 
                    'lines_no': 21,
 
                    'changesets': [
 
                        '49d3fd156b6f7db46313fac355dca1a0b94a0017',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                    ],
 
                },
 
                'e29b67bd158580fc90fc5e9111240b90e6e86064': {
 
                    'lines_no': 32,
 
                    'changesets': [
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '5eab1222a7cd4bfcbabc218ca6d04276d4e27378',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '54000345d2e78b03a99d561399e8e548de3f3203',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '78c3f0c23b7ee935ec276acb8b8212444c33c396',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '2a13f185e4525f9d4b59882791a2d397b90d5ddc',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '78c3f0c23b7ee935ec276acb8b8212444c33c396',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '992f38217b979d0b0987d0bae3cc26dac85d9b19',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                        '16fba1ae9334d79b66d7afed2c2dfbfa2ae53647',
 
                    ],
 
                },
 
            },
 
        }
 

	
 
        for fname, revision_dict in files.items():
 
            for rev, data in revision_dict.items():
 
                cs = self.repo.get_changeset(rev)
 

	
 
                l1_1 = [x[1] for x in cs.get_file_annotate(fname)]
 
                l1_2 = [x[2]().raw_id for x in cs.get_file_annotate(fname)]
 
                assert l1_1 == l1_2
 
                l1 = l1_1
 
                l2 = files[fname][rev]['changesets']
 
                assert l1 == l2, "The lists of revision for %s@rev %s" \
 
                    "from annotation list should match each other, " \
 
                    "got \n%s \nvs \n%s " % (fname, rev, l1, l2)
 

	
 
    def test_files_state(self):
 
        """
 
        Tests state of FileNodes.
 
        """
 
        node = self.repo \
 
            .get_changeset('e6ea6d16e2f26250124a1f4b4fe37a912f9d86a0') \
 
            .get_node('vcs/utils/diffs.py')
 
        assert node.state, NodeState.ADDED
 
        assert node.added
 
        assert not node.changed
 
        assert not node.not_changed
 
        assert not node.removed
 

	
 
        node = self.repo \
 
            .get_changeset('33fa3223355104431402a888fa77a4e9956feb3e') \
 
            .get_node('.hgignore')
 
        assert node.state, NodeState.CHANGED
 
        assert not node.added
 
        assert node.changed
 
        assert not node.not_changed
 
        assert not node.removed
 

	
 
        node = self.repo \
 
            .get_changeset('e29b67bd158580fc90fc5e9111240b90e6e86064') \
 
            .get_node('setup.py')
 
        assert node.state, NodeState.NOT_CHANGED
 
        assert not node.added
 
        assert not node.changed
 
        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_bytes(self):
 
        changeset = self.repo.get_changeset()
 
        for node in changeset.get_node('/'):
 
            if node.is_file():
 
                assert isinstance(node.content, bytes)
 

	
 
    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 == b'base'
 

	
 
    def test_paths_fast_traversing(self):
 
        cs = self.repo.get_changeset()
 
        assert cs.get_node('foobar/static/js/admin/base.js').content == b'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=(b'', b''))
 
        self.repo.get_diff(0, 1)
 
        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)], cwd=self.repo.path)
 

	
 
    def test_get_diff_runs_git_command_with_str_hashes(self):
 
        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(
 
            ['show', '-U3', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             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=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo')
 
        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'], 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=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=almost_overflowed_long_int)
 
        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'], 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=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=overflowed_long_int)
 

	
 
        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'], 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=(b'', b''))
 
        self.repo.get_diff(0, 1, 'foo', context=zero_context)
 

	
 
        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'], cwd=self.repo.path)
 

	
 
    def test_get_diff_sanitizes_negative_context(self):
 
        negative_context = -10
 

	
 
        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(
 
            ['diff', '-U0', '--full-index', '--binary', '-p', '-M', '--abbrev=40',
 
             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"))
 
        # them in the non-bare repo. Deduplicates code in tests a bit.
 
        self.kallithea_hooks = {
 
            "pre-receive": os.path.join(self.repo.path, '.git', 'hooks', "pre-receive"),
 
            "post-receive": os.path.join(self.repo.path, '.git', 'hooks', "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:
 
                f.write("#!/bin/bash\n#CUSTOM_HOOK")
 

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

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path) as f:
 
                assert "CUSTOM_HOOK" in f.read()
 

	
 
    def test_custom_hooks_forced_update(self):
 
        """
 
        Tests if hooks are forcefully updated even though they are custom hooks.
 
        """
 

	
 
        for hook, hook_path in self.kallithea_hooks.iteritems():
 
            with open(hook_path, "w") as f:
 
                f.write("#!/bin/bash\n#CUSTOM_HOOK")
 

	
 
        ScmModel().install_git_hooks(repo=self.repo, force_create=True)
 

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