Changeset - 52e8c4da4768
[Not reviewed]
default
0 1 0
Mads Kiilerich - 11 years ago 2014-08-12 13:08:23
madski@unity3d.com
hg: optimize get_ref_revision

Avoid expensive branch(x) iterating over whole repo just to take max.
1 file changed with 4 insertions and 2 deletions:
0 comments (0 inline, 0 general)
kallithea/lib/vcs/backends/hg/repository.py
Show inline comments
 
@@ -262,362 +262,364 @@ class MercurialRepository(BaseRepository
 
        :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 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 = match(self.path, '', [path])
 
        else:
 
            file_filter = None
 

	
 
        return ''.join(patch.diff(self._repo, rev1, rev2, match=file_filter,
 
                          opts=diffopts(git=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('file:'):
 
            return True
 

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

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

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

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

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

	
 
        q = {"cmd": 'between'}
 
        q.update({'pairs': "%s-%s" % ('0' * 40, '0' * 40)})
 
        qs = '?%s' % urllib.urlencode(q)
 
        cu = "%s%s" % (test_uri, qs)
 
        req = urllib2.Request(cu, None, {})
 

	
 
        try:
 
            resp = o.open(req)
 
            if resp.code != 200:
 
                raise Exception('Return Code is not 200')
 
        except Exception, e:
 
            # means it cannot be cloned
 
            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))
 

	
 
        # now check if it's a proper hg repo
 
        try:
 
            httppeer(repoui or ui.ui(), url).lookup('tip')
 
        except Exception, 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 = str(self._get_url(src_url))
 
                opts = {}
 
                if not update_after_clone:
 
                    opts.update({'noupdate': True})
 
                try:
 
                    MercurialRepository._check_url(url, self.baseui)
 
                    clone(self.baseui, url, self.path, **opts)
 
#                except urllib2.URLError:
 
#                    raise Abort("Got HTTP 404 error")
 
                except Exception:
 
                    raise
 

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

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

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

	
 
    @LazyProperty
 
    def contact(self):
 
        undefined_contact = u'Unknown'
 
        return safe_unicode(get_contact(self._repo.ui.config)
 
                            or undefined_contact)
 

	
 
    @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_hidden(self):
 
        return self._repo.ui.configbool("web", "hidden", untrusted=True)
 

	
 
    def _get_revision(self, revision):
 
        """
 
        Gets an ID revision given as str. This will always return a fill
 
        40 char revision number
 

	
 
        :param revision: str or int or None
 
        """
 
        if isinstance(revision, unicode):
 
            revision = safe_str(revision)
 

	
 
        if self._empty:
 
            raise EmptyRepositoryError("There are no changesets yet")
 

	
 
        if revision in [-1, 'tip', None]:
 
            revision = 'tip'
 

	
 
        try:
 
            revision = hex(self._repo.lookup(revision))
 
        except (LookupError, ):
 
            msg = ("Ambiguous identifier `%s` for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 
        except (IndexError, ValueError, RepoLookupError, TypeError):
 
            msg = ("Revision %s does not exist for %s" % (revision, self))
 
            raise ChangesetDoesNotExistError(msg)
 

	
 
        return revision
 

	
 
    def get_ref_revision(self, ref_type, ref_name):
 
        """
 
        Returns revision number for the given reference.
 
        """
 
        ref_name = safe_str(ref_name)
 
        # lookup up the exact node id
 
        _revset_predicates = {
 
                'branch': 'branch',
 
                'book': 'bookmark',
 
                'tag': 'tag',
 
                'rev': 'id',
 
            }
 
        rev_spec = "max(%s(%%s))" % _revset_predicates[ref_type]
 
        # avoid expensive branch(x) iteration over whole repo
 
        rev_spec = "%%s & %s(%%s)" % _revset_predicates[ref_type]
 
        try:
 
            revs = self._repo.revs(rev_spec, safe_str(ref_name))
 
            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 (IndexError, ValueError, RepoLookupError, TypeError):
 
            msg = ("Revision %s:%s does not exist for %s" % (ref_type, ref_name, self.name))
 
            raise ChangesetDoesNotExistError(msg)
 
        if revs:
 
            revision = revs[-1]
 
        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("web", "allow_archive",
 
                                         untrusted=True)
 
        for i in [('zip', '.zip'), ('gz', '.tar.gz'), ('bz2', '.tar.bz2')]:
 
            if i[0] in allowed or self._repo.ui.configbool("web",
 
                                                           "allow" + i[0],
 
                                                           untrusted=True):
 
                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 = 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``.
 
        """
 
        revision = self._get_revision(revision)
 
        changeset = MercurialChangeset(repository=self, revision=revision)
 
        return changeset
 

	
 
    def get_changesets(self, start=None, end=None, start_date=None,
 
                       end_date=None, branch_name=None, reverse=False):
 
        """
 
        Returns iterator of ``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 = self.revisions.index(start_raw_id) if start else None
 
        end_raw_id = self._get_revision(end)
 
        end_pos = self.revisions.index(end_raw_id) if end else None
 

	
 
        if None not in [start, end] 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.keys():
 
            msg = ("Branch %s not found in %s" % (branch_name, self))
 
            raise BranchDoesNotExistError(msg)
 
        if end_pos is not None:
 
            end_pos += 1
 
        #filter branches
 
        filter_ = []
 
        if branch_name:
 
            filter_.append('branch("%s")' % (branch_name))
 

	
 
        if start_date and not end_date:
 
            filter_.append('date(">%s")' % start_date)
 
        if end_date and not start_date:
 
            filter_.append('date("<%s")' % end_date)
 
        if start_date and end_date:
 
            filter_.append('date(">%s") and date("<%s")' % (start_date, end_date))
 
        if filter_:
 
            revisions = scmutil.revrange(self._repo, filter_)
 
        else:
 
            revisions = self.revisions
 

	
 
        revs = revisions[start_pos:end_pos]
 
        if reverse:
 
            revs = reversed(revs)
 

	
 
        return CollectionGenerator(self, revs)
 

	
 
    def pull(self, url):
 
        """
 
        Tries to pull changes from external location.
 
        """
 
        url = self._get_url(url)
 
        try:
 
            other = peer(self._repo, {}, url)
 
            self._repo.pull(other, heads=None, force=None)
 
        except Abort, 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
 
        for path in config_file:
 
            config.readconfig(path)
 
        return config.config(section, 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')
 
        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')
 
        if username:
 
            return author_email(username)
 
        return None
0 comments (0 inline, 0 general)