Changeset - 280ac97cb37f
[Not reviewed]
beta
0 2 0
Mads Kiilerich - 13 years ago 2012-12-12 18:11:33
madski@unity3d.com
Grafted from: 31d406b02288
don't recurse into '.hg' of removed repositories

Walking through the content of a '.hg' that has been renamed to 'rm__.hg' on
startup can take a lot of time.
2 files changed with 4 insertions and 1 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/vcs/utils/helpers.py
Show inline comments
 
"""
 
Utitlites aimed to help achieve mostly basic tasks.
 
"""
 
from __future__ import division
 

	
 
import re
 
import time
 
import datetime
 
import os.path
 
from subprocess import Popen, PIPE
 
from rhodecode.lib.vcs.exceptions import VCSError
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
from rhodecode.lib.vcs.utils.paths import abspath
 

	
 
ALIASES = ['hg', 'git']
 

	
 

	
 
def get_scm(path, search_recursively=False, explicit_alias=None):
 
    """
 
    Returns one of alias from ``ALIASES`` (in order of precedence same as
 
    shortcuts given in ``ALIASES``) and top working dir path for the given
 
    argument. If no scm-specific directory is found or more than one scm is
 
    found at that directory, ``VCSError`` is raised.
 

	
 
    :param search_recursively: if set to ``True``, this function would try to
 
      move up to parent directory every time no scm is recognized for the
 
      currently checked path. Default: ``False``.
 
    :param explicit_alias: can be one of available backend aliases, when given
 
      it will return given explicit alias in repositories under more than one
 
      version control, if explicit_alias is different than found it will raise
 
      VCSError
 
    """
 
    if not os.path.isdir(path):
 
        raise VCSError("Given path %s is not a directory" % path)
 

	
 
    def get_scms(path):
 
        return [(scm, path) for scm in get_scms_for_path(path)]
 

	
 
    found_scms = get_scms(path)
 
    while  not found_scms and search_recursively:
 
        newpath = abspath(path, '..')
 
        if newpath == path:
 
            break
 
        path = newpath
 
        found_scms = get_scms(path)
 

	
 
    if len(found_scms) > 1:
 
        for scm in found_scms:
 
            if scm[0] == explicit_alias:
 
                return scm
 
        raise VCSError('More than one [%s] scm found at given path %s'
 
                       % (','.join((x[0] for x in found_scms)), path))
 

	
 
    if len(found_scms) is 0:
 
        raise VCSError('No scm found at given path %s' % path)
 

	
 
    return found_scms[0]
 

	
 

	
 
def get_scms_for_path(path):
 
    """
 
    Returns all scm's found at the given path. If no scm is recognized
 
    - empty list is returned.
 

	
 
    :param path: path to directory which should be checked. May be callable.
 

	
 
    :raises VCSError: if given ``path`` is not a directory
 
    """
 
    from rhodecode.lib.vcs.backends import get_backend
 
    if hasattr(path, '__call__'):
 
        path = path()
 
    if not os.path.isdir(path):
 
        raise VCSError("Given path %r is not a directory" % path)
 

	
 
    result = []
 
    for key in ALIASES:
 
        dirname = os.path.join(path, '.' + key)
 
        if os.path.isdir(dirname):
 
            result.append(key)
 
            continue
 
        dirname = os.path.join(path, 'rm__.' + key)
 
        if os.path.isdir(dirname):
 
            return [None]
 
        # We still need to check if it's not bare repository as
 
        # bare repos don't have working directories
 
        try:
 
            get_backend(key)(path)
 
            result.append(key)
 
            continue
 
        except RepositoryError:
 
            # Wrong backend
 
            pass
 
        except VCSError:
 
            # No backend at all
 
            pass
 
    return result
 

	
 

	
 
def get_repo_paths(path):
 
    """
 
    Returns path's subdirectories which seems to be a repository.
 
    """
 
    repo_paths = []
 
    dirnames = (os.path.abspath(dirname) for dirname in os.listdir(path))
 
    for dirname in dirnames:
 
        try:
 
            get_scm(dirname)
 
            repo_paths.append(dirname)
 
        except VCSError:
 
            pass
 
    return repo_paths
 

	
 

	
 
def run_command(cmd, *args):
 
    """
 
    Runs command on the system with given ``args``.
 
    """
 
    command = ' '.join((cmd, args))
 
    p = Popen(command, shell=True, stdout=PIPE, stderr=PIPE)
 
    stdout, stderr = p.communicate()
 
    return p.retcode, stdout, stderr
 

	
 

	
 
def get_highlighted_code(name, code, type='terminal'):
 
    """
 
    If pygments are available on the system
 
    then returned output is colored. Otherwise
 
    unchanged content is returned.
 
    """
 
    import logging
 
    try:
 
        import pygments
 
        pygments
 
    except ImportError:
 
        return code
 
    from pygments import highlight
 
    from pygments.lexers import guess_lexer_for_filename, ClassNotFound
 
    from pygments.formatters import TerminalFormatter
 

	
 
    try:
 
        lexer = guess_lexer_for_filename(name, code)
 
        formatter = TerminalFormatter()
 
        content = highlight(code, lexer, formatter)
 
    except ClassNotFound:
 
        logging.debug("Couldn't guess Lexer, will not use pygments.")
 
        content = code
 
    return content
 

	
 
def parse_changesets(text):
 
    """
 
    Returns dictionary with *start*, *main* and *end* ids.
 

	
 
    Examples::
 

	
 
        >>> parse_changesets('aaabbb')
 
        {'start': None, 'main': 'aaabbb', 'end': None}
 
        >>> parse_changesets('aaabbb..cccddd')
 
        {'start': 'aaabbb', 'main': None, 'end': 'cccddd'}
 

	
 
    """
 
    text = text.strip()
 
    CID_RE = r'[a-zA-Z0-9]+'
 
    if not '..' in text:
 
        m = re.match(r'^(?P<cid>%s)$' % CID_RE, text)
 
        if m:
 
            return {
 
                'start': None,
 
                'main': text,
 
                'end': None,
 
            }
 
    else:
 
        RE = r'^(?P<start>%s)?\.{2,3}(?P<end>%s)?$' % (CID_RE, CID_RE)
 
        m = re.match(RE, text)
 
        if m:
 
            result = m.groupdict()
 
            result['main'] = None
 
            return result
 
    raise ValueError("IDs not recognized")
 

	
 
def parse_datetime(text):
 
    """
 
    Parses given text and returns ``datetime.datetime`` instance or raises
 
    ``ValueError``.
 

	
 
    :param text: string of desired date/datetime or something more verbose,
 
      like *yesterday*, *2weeks 3days*, etc.
 
    """
 

	
 
    text = text.strip().lower()
 

	
 
    INPUT_FORMATS = (
 
        '%Y-%m-%d %H:%M:%S',
 
        '%Y-%m-%d %H:%M',
 
        '%Y-%m-%d',
 
        '%m/%d/%Y %H:%M:%S',
 
        '%m/%d/%Y %H:%M',
 
        '%m/%d/%Y',
 
        '%m/%d/%y %H:%M:%S',
 
        '%m/%d/%y %H:%M',
 
        '%m/%d/%y',
 
    )
 
    for format in INPUT_FORMATS:
 
        try:
 
            return datetime.datetime(*time.strptime(text, format)[:6])
 
        except ValueError:
 
            pass
 

	
 
    # Try descriptive texts
 
    if text == 'tomorrow':
 
        future = datetime.datetime.now() + datetime.timedelta(days=1)
 
        args = future.timetuple()[:3] + (23, 59, 59)
 
        return datetime.datetime(*args)
 
    elif text == 'today':
 
        return datetime.datetime(*datetime.datetime.today().timetuple()[:3])
 
    elif text == 'now':
 
        return datetime.datetime.now()
 
    elif text == 'yesterday':
 
        past = datetime.datetime.now() - datetime.timedelta(days=1)
 
        return datetime.datetime(*past.timetuple()[:3])
 
    else:
 
        days = 0
 
        matched = re.match(
 
            r'^((?P<weeks>\d+) ?w(eeks?)?)? ?((?P<days>\d+) ?d(ays?)?)?$', text)
 
        if matched:
 
            groupdict = matched.groupdict()
 
            if groupdict['days']:
 
                days += int(matched.groupdict()['days'])
 
            if groupdict['weeks']:
 
                days += int(matched.groupdict()['weeks']) * 7
 
            past = datetime.datetime.now() - datetime.timedelta(days=days)
 
            return datetime.datetime(*past.timetuple()[:3])
 

	
 
    raise ValueError('Wrong date: "%s"' % text)
 

	
 

	
 
def get_dict_for_attrs(obj, attrs):
 
    """
 
    Returns dictionary for each attribute from given ``obj``.
 
    """
 
    data = {}
 
    for attr in attrs:
 
        data[attr] = getattr(obj, attr)
 
    return data
 

	
 

	
 
def get_total_seconds(timedelta):
 
    """
 
    Backported for Python 2.5.
 

	
 
    See http://docs.python.org/library/datetime.html.
 
    """
 
    return ((timedelta.microseconds + (
 
            timedelta.seconds +
 
            timedelta.days * 24 * 60 * 60
 
        ) * 10**6) / 10**6)
rhodecode/model/scm.py
Show inline comments
 
@@ -36,385 +36,385 @@ from sqlalchemy import func
 
from pylons.i18n.translation import _
 

	
 
import rhodecode
 
from rhodecode.lib.vcs import get_backend
 
from rhodecode.lib.vcs.exceptions import RepositoryError
 
from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
from rhodecode.lib.vcs.nodes import FileNode
 
from rhodecode.lib.vcs.backends.base import EmptyChangeset
 

	
 
from rhodecode import BACKENDS
 
from rhodecode.lib import helpers as h
 
from rhodecode.lib.utils2 import safe_str, safe_unicode
 
from rhodecode.lib.auth import HasRepoPermissionAny, HasReposGroupPermissionAny
 
from rhodecode.lib.utils import get_repos as get_filesystem_repos, make_ui, \
 
    action_logger, REMOVED_REPO_PAT
 
from rhodecode.model import BaseModel
 
from rhodecode.model.db import Repository, RhodeCodeUi, CacheInvalidation, \
 
    UserFollowing, UserLog, User, RepoGroup, PullRequest
 

	
 
log = logging.getLogger(__name__)
 

	
 

	
 
class UserTemp(object):
 
    def __init__(self, user_id):
 
        self.user_id = user_id
 

	
 
    def __repr__(self):
 
        return "<%s('id:%s')>" % (self.__class__.__name__, self.user_id)
 

	
 

	
 
class RepoTemp(object):
 
    def __init__(self, repo_id):
 
        self.repo_id = repo_id
 

	
 
    def __repr__(self):
 
        return "<%s('id:%s')>" % (self.__class__.__name__, self.repo_id)
 

	
 

	
 
class CachedRepoList(object):
 
    """
 
    Cached repo list, uses in-memory cache after initialization, that is
 
    super fast
 
    """
 

	
 
    def __init__(self, db_repo_list, repos_path, order_by=None):
 
        self.db_repo_list = db_repo_list
 
        self.repos_path = repos_path
 
        self.order_by = order_by
 
        self.reversed = (order_by or '').startswith('-')
 

	
 
    def __len__(self):
 
        return len(self.db_repo_list)
 

	
 
    def __repr__(self):
 
        return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 

	
 
    def __iter__(self):
 
        # pre-propagated cache_map to save executing select statements
 
        # for each repo
 
        cache_map = CacheInvalidation.get_cache_map()
 

	
 
        for dbr in self.db_repo_list:
 
            scmr = dbr.scm_instance_cached(cache_map)
 
            # check permission at this level
 
            if not HasRepoPermissionAny(
 
                'repository.read', 'repository.write', 'repository.admin'
 
            )(dbr.repo_name, 'get repo check'):
 
                continue
 

	
 
            if scmr is None:
 
                log.error(
 
                    '%s this repository is present in database but it '
 
                    'cannot be created as an scm instance' % dbr.repo_name
 
                )
 
                continue
 

	
 
            last_change = scmr.last_change
 
            tip = h.get_changeset_safe(scmr, 'tip')
 

	
 
            tmp_d = {}
 
            tmp_d['name'] = dbr.repo_name
 
            tmp_d['name_sort'] = tmp_d['name'].lower()
 
            tmp_d['raw_name'] = tmp_d['name'].lower()
 
            tmp_d['description'] = dbr.description
 
            tmp_d['description_sort'] = tmp_d['description'].lower()
 
            tmp_d['last_change'] = last_change
 
            tmp_d['last_change_sort'] = time.mktime(last_change.timetuple())
 
            tmp_d['tip'] = tip.raw_id
 
            tmp_d['tip_sort'] = tip.revision
 
            tmp_d['rev'] = tip.revision
 
            tmp_d['contact'] = dbr.user.full_contact
 
            tmp_d['contact_sort'] = tmp_d['contact']
 
            tmp_d['owner_sort'] = tmp_d['contact']
 
            tmp_d['repo_archives'] = list(scmr._get_archives())
 
            tmp_d['last_msg'] = tip.message
 
            tmp_d['author'] = tip.author
 
            tmp_d['dbrepo'] = dbr.get_dict()
 
            tmp_d['dbrepo_fork'] = dbr.fork.get_dict() if dbr.fork else {}
 
            yield tmp_d
 

	
 

	
 
class SimpleCachedRepoList(CachedRepoList):
 
    """
 
    Lighter version of CachedRepoList without the scm initialisation
 
    """
 

	
 
    def __iter__(self):
 
        for dbr in self.db_repo_list:
 
            # check permission at this level
 
            if not HasRepoPermissionAny(
 
                'repository.read', 'repository.write', 'repository.admin'
 
            )(dbr.repo_name, 'get repo check'):
 
                continue
 

	
 
            tmp_d = {}
 
            tmp_d['name'] = dbr.repo_name
 
            tmp_d['name_sort'] = tmp_d['name'].lower()
 
            tmp_d['raw_name'] = tmp_d['name'].lower()
 
            tmp_d['description'] = dbr.description
 
            tmp_d['description_sort'] = tmp_d['description'].lower()
 
            tmp_d['dbrepo'] = dbr.get_dict()
 
            tmp_d['dbrepo_fork'] = dbr.fork.get_dict() if dbr.fork else {}
 
            yield tmp_d
 

	
 

	
 
class GroupList(object):
 

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

	
 
    def __len__(self):
 
        return len(self.db_repo_group_list)
 

	
 
    def __repr__(self):
 
        return '<%s (%s)>' % (self.__class__.__name__, self.__len__())
 

	
 
    def __iter__(self):
 
        for dbgr in self.db_repo_group_list:
 
            # check permission at this level
 
            if not HasReposGroupPermissionAny(
 
                'group.read', 'group.write', 'group.admin'
 
            )(dbgr.group_name, 'get group repo check'):
 
                continue
 

	
 
            yield dbgr
 

	
 

	
 
class ScmModel(BaseModel):
 
    """
 
    Generic Scm Model
 
    """
 

	
 
    def __get_repo(self, instance):
 
        cls = Repository
 
        if isinstance(instance, cls):
 
            return instance
 
        elif isinstance(instance, int) or safe_str(instance).isdigit():
 
            return cls.get(instance)
 
        elif isinstance(instance, basestring):
 
            return cls.get_by_repo_name(instance)
 
        elif instance:
 
            raise Exception('given object must be int, basestr or Instance'
 
                            ' of %s got %s' % (type(cls), type(instance)))
 

	
 
    @LazyProperty
 
    def repos_path(self):
 
        """
 
        Get's the repositories root path from database
 
        """
 

	
 
        q = self.sa.query(RhodeCodeUi).filter(RhodeCodeUi.ui_key == '/').one()
 

	
 
        return q.ui_value
 

	
 
    def repo_scan(self, repos_path=None):
 
        """
 
        Listing of repositories in given path. This path should not be a
 
        repository itself. Return a dictionary of repository objects
 

	
 
        :param repos_path: path to directory containing repositories
 
        """
 

	
 
        if repos_path is None:
 
            repos_path = self.repos_path
 

	
 
        log.info('scanning for repositories in %s' % repos_path)
 

	
 
        baseui = make_ui('db')
 
        repos = {}
 

	
 
        for name, path in get_filesystem_repos(repos_path, recursive=True):
 
            # skip removed repos
 
            if REMOVED_REPO_PAT.match(name):
 
            if REMOVED_REPO_PAT.match(name) or path[0] is None:
 
                continue
 

	
 
            # name need to be decomposed and put back together using the /
 
            # since this is internal storage separator for rhodecode
 
            name = Repository.url_sep().join(name.split(os.sep))
 

	
 
            try:
 
                if name in repos:
 
                    raise RepositoryError('Duplicate repository name %s '
 
                                          'found in %s' % (name, path))
 
                else:
 

	
 
                    klass = get_backend(path[0])
 

	
 
                    if path[0] == 'hg' and path[0] in BACKENDS.keys():
 
                        repos[name] = klass(safe_str(path[1]), baseui=baseui)
 

	
 
                    if path[0] == 'git' and path[0] in BACKENDS.keys():
 
                        repos[name] = klass(path[1])
 
            except OSError:
 
                continue
 

	
 
        return repos
 

	
 
    def get_repos(self, all_repos=None, sort_key=None, simple=False):
 
        """
 
        Get all repos from db and for each repo create it's
 
        backend instance and fill that backed with information from database
 

	
 
        :param all_repos: list of repository names as strings
 
            give specific repositories list, good for filtering
 

	
 
        :param sort_key: initial sorting of repos
 
        :param simple: use SimpleCachedList - one without the SCM info
 
        """
 
        if all_repos is None:
 
            all_repos = self.sa.query(Repository)\
 
                        .filter(Repository.group_id == None)\
 
                        .order_by(func.lower(Repository.repo_name)).all()
 
        if simple:
 
            repo_iter = SimpleCachedRepoList(all_repos,
 
                                             repos_path=self.repos_path,
 
                                             order_by=sort_key)
 
        else:
 
            repo_iter = CachedRepoList(all_repos,
 
                                       repos_path=self.repos_path,
 
                                       order_by=sort_key)
 

	
 
        return repo_iter
 

	
 
    def get_repos_groups(self, all_groups=None):
 
        if all_groups is None:
 
            all_groups = RepoGroup.query()\
 
                .filter(RepoGroup.group_parent_id == None).all()
 
        group_iter = GroupList(all_groups)
 

	
 
        return group_iter
 

	
 
    def mark_for_invalidation(self, repo_name):
 
        """
 
        Puts cache invalidation task into db for
 
        further global cache invalidation
 

	
 
        :param repo_name: this repo that should invalidation take place
 
        """
 
        CacheInvalidation.set_invalidate(repo_name=repo_name)
 

	
 
    def toggle_following_repo(self, follow_repo_id, user_id):
 

	
 
        f = self.sa.query(UserFollowing)\
 
            .filter(UserFollowing.follows_repo_id == follow_repo_id)\
 
            .filter(UserFollowing.user_id == user_id).scalar()
 

	
 
        if f is not None:
 
            try:
 
                self.sa.delete(f)
 
                action_logger(UserTemp(user_id),
 
                              'stopped_following_repo',
 
                              RepoTemp(follow_repo_id))
 
                return
 
            except:
 
                log.error(traceback.format_exc())
 
                raise
 

	
 
        try:
 
            f = UserFollowing()
 
            f.user_id = user_id
 
            f.follows_repo_id = follow_repo_id
 
            self.sa.add(f)
 

	
 
            action_logger(UserTemp(user_id),
 
                          'started_following_repo',
 
                          RepoTemp(follow_repo_id))
 
        except:
 
            log.error(traceback.format_exc())
 
            raise
 

	
 
    def toggle_following_user(self, follow_user_id, user_id):
 
        f = self.sa.query(UserFollowing)\
 
            .filter(UserFollowing.follows_user_id == follow_user_id)\
 
            .filter(UserFollowing.user_id == user_id).scalar()
 

	
 
        if f is not None:
 
            try:
 
                self.sa.delete(f)
 
                return
 
            except:
 
                log.error(traceback.format_exc())
 
                raise
 

	
 
        try:
 
            f = UserFollowing()
 
            f.user_id = user_id
 
            f.follows_user_id = follow_user_id
 
            self.sa.add(f)
 
        except:
 
            log.error(traceback.format_exc())
 
            raise
 

	
 
    def is_following_repo(self, repo_name, user_id, cache=False):
 
        r = self.sa.query(Repository)\
 
            .filter(Repository.repo_name == repo_name).scalar()
 

	
 
        f = self.sa.query(UserFollowing)\
 
            .filter(UserFollowing.follows_repository == r)\
 
            .filter(UserFollowing.user_id == user_id).scalar()
 

	
 
        return f is not None
 

	
 
    def is_following_user(self, username, user_id, cache=False):
 
        u = User.get_by_username(username)
 

	
 
        f = self.sa.query(UserFollowing)\
 
            .filter(UserFollowing.follows_user == u)\
 
            .filter(UserFollowing.user_id == user_id).scalar()
 

	
 
        return f is not None
 

	
 
    def get_followers(self, repo):
 
        repo = self._get_repo(repo)
 

	
 
        return self.sa.query(UserFollowing)\
 
                .filter(UserFollowing.follows_repository == repo).count()
 

	
 
    def get_forks(self, repo):
 
        repo = self._get_repo(repo)
 
        return self.sa.query(Repository)\
 
                .filter(Repository.fork == repo).count()
 

	
 
    def get_pull_requests(self, repo):
 
        repo = self._get_repo(repo)
 
        return self.sa.query(PullRequest)\
 
                .filter(PullRequest.other_repo == repo).count()
 

	
 
    def mark_as_fork(self, repo, fork, user):
 
        repo = self.__get_repo(repo)
 
        fork = self.__get_repo(fork)
 
        if fork and repo.repo_id == fork.repo_id:
 
            raise Exception("Cannot set repository as fork of itself")
 
        repo.fork = fork
 
        self.sa.add(repo)
 
        return repo
 

	
 
    def pull_changes(self, repo, username):
 
        dbrepo = self.__get_repo(repo)
 
        clone_uri = dbrepo.clone_uri
 
        if not clone_uri:
 
            raise Exception("This repository doesn't have a clone uri")
 

	
 
        repo = dbrepo.scm_instance
 
        from rhodecode import CONFIG
 
        try:
 
            extras = {
 
                'ip': '',
 
                'username': username,
 
                'action': 'push_remote',
 
                'repository': dbrepo.repo_name,
 
                'scm': repo.alias,
 
                'config': CONFIG['__file__'],
 
                'make_lock': None,
 
                'locked_by': [None, None]
 
            }
 

	
 
            Repository.inject_ui(repo, extras=extras)
 

	
 
            if repo.alias == 'git':
 
                repo.fetch(clone_uri)
 
            else:
 
                repo.pull(clone_uri)
 
            self.mark_for_invalidation(dbrepo.repo_name)
 
        except:
 
            log.error(traceback.format_exc())
0 comments (0 inline, 0 general)