compare controller for pylons showing differences between two

repos, branches, bookmarks or tips

This file was forked by the Kallithea project in July 2014.

Original author and date, and relevant copyright and licensing information is below:

:created_on: May 6, 2012

:author: marcink

:copyright: (c) 2013 RhodeCode GmbH, and others.

:license: GPLv3, see LICENSE.md for more details.

"""

import logging

import re

from webob.exc import HTTPBadRequest

from pylons import request, tmpl_context as c, url

from pylons.controllers.util import redirect

from pylons.i18n.translation import _

from kallithea.lib.vcs.utils.hgcompat import unionrepo

from kallithea.lib import helpers as h

from kallithea.lib.base import BaseRepoController, render

from kallithea.lib.auth import LoginRequired, HasRepoPermissionAnyDecorator

from kallithea.lib import diffs

from kallithea.model.db import Repository

from kallithea.lib.diffs import LimitedDiffContainer

from kallithea.controllers.changeset import _ignorews_url,\

    _context_url, get_line_ctx, get_ignore_ws

from kallithea.lib.graphmod import graph_data

from kallithea.lib.compat import json

log = logging.getLogger(__name__)

class CompareController(BaseRepoController):

    def __before__(self):

        super(CompareController, self).__before__()

    @staticmethod

    def _get_changesets(alias, org_repo, org_rev, other_repo, other_rev):

        """

        Returns lists of changesets that can be merged from org_repo@org_rev

        to other_repo@other_rev

        ... and the other way

        ... and the ancestor that would be used for merge

        :param org_rev: the revision we want our compare to be made

            all changesets that we need to obtain

        :param other_rev: revision we want out compare to be made on other_repo

        """

        ancestor = None

        if org_rev == other_rev:

            org_changesets = []

            other_changesets = []

            ancestor = org_rev

        elif alias == 'hg':

            #case two independent repos

            if org_repo != other_repo:

                hgrepo = unionrepo.unionrepository(other_repo.baseui,

                                                   other_repo.path,

                                                   org_repo.path)

                # all ancestors of other_rev will be in other_repo and

                # rev numbers from hgrepo can be used in other_repo - org_rev ancestors cannot

            #no remote compare do it on the same repository

            else:

                hgrepo = other_repo._repo

            ancestors = hgrepo.revs("ancestor(id(%s), id(%s))", org_rev, other_rev)

            if ancestors:

                # pick arbitrary ancestor - but there is usually only one

                ancestor = hgrepo[ancestors[0]].hex()

            other_revs = hgrepo.revs("ancestors(id(%s)) and not ancestors(id(%s)) and not id(%s)",

                                     other_rev, org_rev, org_rev)

            other_changesets = [other_repo.get_changeset(rev) for rev in other_revs]

            org_revs = hgrepo.revs("ancestors(id(%s)) and not ancestors(id(%s)) and not id(%s)",

                                   org_rev, other_rev, other_rev)

            org_changesets = [org_repo.get_changeset(hgrepo[rev].hex()) for rev in org_revs]

        elif alias == 'git':

            if org_repo != other_repo:

                from dulwich.repo import Repo

                from dulwich.client import SubprocessGitClient

                gitrepo = Repo(org_repo.path)

                SubprocessGitClient(thin_packs=False).fetch(other_repo.path, gitrepo)

                gitrepo_remote = Repo(other_repo.path)

                SubprocessGitClient(thin_packs=False).fetch(org_repo.path, gitrepo_remote)

                revs = []

                for x in gitrepo_remote.get_walker(include=[other_rev],

    @hybrid_property

    def app_settings_value(self):

        v = self._app_settings_value

        _type = self.app_settings_type

        converter = self.SETTINGS_TYPES.get(_type) or self.SETTINGS_TYPES['unicode']

        return converter(v)

    @app_settings_value.setter

    def app_settings_value(self, val):

        """

        Setter that will always make sure we use unicode in app_settings_value

        :param val:

        """

        self._app_settings_value = safe_unicode(val)

    @hybrid_property

    def app_settings_type(self):

        return self._app_settings_type

    @app_settings_type.setter

    def app_settings_type(self, val):

        if val not in self.SETTINGS_TYPES:

            raise Exception('type must be one of %s got %s'

                            % (self.SETTINGS_TYPES.keys(), val))

        self._app_settings_type = val

    def __unicode__(self):

        return u"<%s('%s:%s[%s]')>" % (

            self.__class__.__name__,

            self.app_settings_name, self.app_settings_value, self.app_settings_type

        )

    @classmethod

    def get_by_name(cls, key):

        return cls.query()\

            .filter(cls.app_settings_name == key).scalar()

    @classmethod

    def get_by_name_or_create(cls, key, val='', type='unicode'):

        res = cls.get_by_name(key)

        if not res:

            res = cls(key, val, type)

        return res

    @classmethod

    def create_or_update(cls, key, val=Optional(''), type=Optional('unicode')):

        """

        :param key:

        :param val:

        :param type:

        :return:

        """

        res = cls.get_by_name(key)

        if not res:

            val = Optional.extract(val)

            type = Optional.extract(type)

            res = cls(key, val, type)

        else:

            res.app_settings_name = key

            if not isinstance(val, Optional):

                # update if set

                res.app_settings_value = val

            if not isinstance(type, Optional):

                # update if set

                res.app_settings_type = type

        return res

    @classmethod

    def get_app_settings(cls, cache=False):

        ret = cls.query()

        if cache:

            ret = ret.options(FromCache("sql_cache_short", "get_hg_settings"))

        if not ret:

            raise Exception('Could not get application settings !')

        settings = {}

        for each in ret:

            settings[each.app_settings_name] = \

                each.app_settings_value

        return settings

    @classmethod

    def get_auth_plugins(cls, cache=False):

        auth_plugins = cls.get_by_name("auth_plugins").app_settings_value

        return auth_plugins

    @classmethod

    def get_auth_settings(cls, cache=False):

        ret = cls.query()\

                .filter(cls.app_settings_name.startswith('auth_')).all()

        fd = {}

        active = v.StringBoolean(if_missing=False)

        firstname = v.UnicodeString(strip=True, min=1, not_empty=False)

        lastname = v.UnicodeString(strip=True, min=1, not_empty=False)

        email = All(v.Email(not_empty=True), v.UniqSystemEmail(old_data))

        extern_name = v.UnicodeString(strip=True)

        extern_type = v.UnicodeString(strip=True)

        chained_validators = [v.ValidPasswordsMatch()]

    return _UserForm

def UserGroupForm(edit=False, old_data={}, available_members=[]):

    class _UserGroupForm(formencode.Schema):

        allow_extra_fields = True

        filter_extra_fields = True

        users_group_name = All(

            v.UnicodeString(strip=True, min=1, not_empty=True),

            v.ValidUserGroup(edit, old_data)

        )

        user_group_description = v.UnicodeString(strip=True, min=1,

                                                 not_empty=False)

        users_group_active = v.StringBoolean(if_missing=False)

        if edit:

            users_group_members = v.OneOf(

                available_members, hideList=False, testValueList=True,

                if_missing=None, not_empty=False

            )

    return _UserGroupForm

def RepoGroupForm(edit=False, old_data={}, available_groups=[],

                   can_create_in_root=False):

    class _RepoGroupForm(formencode.Schema):

        allow_extra_fields = True

        filter_extra_fields = False

        group_name = All(v.UnicodeString(strip=True, min=1, not_empty=True),

                         v.SlugifyName(),

                         v.ValidRegex(msg=_('Name must not contain only digits'))(r'(?!^\d+$)^.+$'))

        group_description = v.UnicodeString(strip=True, min=1,

                                            not_empty=False)

        group_copy_permissions = v.StringBoolean(if_missing=False)

        if edit:

            #FIXME: do a special check that we cannot move a group to one of

            pass

        group_parent_id = All(v.CanCreateGroup(can_create_in_root),

                              v.OneOf(available_groups, hideList=False,

                                      testValueList=True,

                                      if_missing=None, not_empty=True))

        enable_locking = v.StringBoolean(if_missing=False)

        chained_validators = [v.ValidRepoGroup(edit, old_data)]

    return _RepoGroupForm

def RegisterForm(edit=False, old_data={}):

    class _RegisterForm(formencode.Schema):

        allow_extra_fields = True

        filter_extra_fields = True

        username = All(

            v.ValidUsername(edit, old_data),

            v.UnicodeString(strip=True, min=1, not_empty=True)

        )

        password = All(

            v.ValidPassword(),

            v.UnicodeString(strip=False, min=6, not_empty=True)

        )

        password_confirmation = All(

            v.ValidPassword(),

            v.UnicodeString(strip=False, min=6, not_empty=True)

        )

        active = v.StringBoolean(if_missing=False)

        firstname = v.UnicodeString(strip=True, min=1, not_empty=False)

        lastname = v.UnicodeString(strip=True, min=1, not_empty=False)

        email = All(v.Email(not_empty=True), v.UniqSystemEmail(old_data))

        chained_validators = [v.ValidPasswordsMatch()]

    return _RegisterForm

def PasswordResetForm():

    class _PasswordResetForm(formencode.Schema):

        allow_extra_fields = True

        filter_extra_fields = True

        email = All(v.ValidSystemEmail(), v.Email(not_empty=True))

    return _PasswordResetForm

def RepoForm(edit=False, old_data={}, supported_backends=BACKENDS.keys(),

             repo_groups=[], landing_revs=[]):

    class _RepoForm(formencode.Schema):

        Gets the repositories root path from database

        """

        q = self.sa.query(Ui).filter(Ui.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):

            # name need to be decomposed and put back together using the /

            # since this is internal storage separator for kallithea

            name = Repository.normalize_repo_name(name)

            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

        log.debug('found %s paths with repositories' % (len(repos)))

        return repos

    def get_repos(self, all_repos=None, sort_key=None, simple=False):

        """

        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_repo_groups(self, all_groups=None):

        if all_groups is None:

            all_groups = RepoGroup.query()\

                .filter(RepoGroup.group_parent_id == None).all()

        return [x for x in RepoGroupList(all_groups)]

    def mark_for_invalidation(self, repo_name, delete=False):

        """

        Mark caches of this repo invalid in the database.

        :param repo_name: the repo for which caches should be marked invalid

        """

        CacheInvalidation.set_invalidate(repo_name, delete=delete)

        repo = Repository.get_by_repo_name(repo_name)

        if repo:

            repo.update_changeset_cache()

    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:

        :param repo: SCM instance

        """

        user = self._get_user(user)

        IMC = self._get_IMC_module(repo.alias)

        # decoding here will force that we have proper encoded values

        # in any other case this will throw exceptions and deny commit

        content = safe_str(content)

        path = safe_str(f_path)

        # message and author needs to be unicode

        # proper backend should then translate that into required type

        message = safe_unicode(message)

        author = safe_unicode(author)

        imc = IMC(repo)

        imc.change(FileNode(path, content, mode=cs.get_file_mode(f_path)))

        try:

            tip = imc.commit(message=message, author=author,

                             parents=[cs], branch=cs.branch)

        except Exception, e:

            log.error(traceback.format_exc())

            raise IMCCommitError(str(e))

        finally:

            # always clear caches, if commit fails we want fresh object also

            self.mark_for_invalidation(repo_name)

        self._handle_push(repo,

                          username=user.username,

                          action='push_local',

                          repo_name=repo_name,

                          revisions=[tip.raw_id])

        return tip

    def _sanitize_path(self, f_path):

        if f_path.startswith('/') or f_path.startswith('.') or '../' in f_path:

            raise NonRelativePathError('%s is not an relative path' % f_path)

        if f_path:

            f_path = os.path.normpath(f_path)

        return f_path

    def get_nodes(self, repo_name, revision, root_path='/', flat=True):

        """

        recursive walk in root dir and return a set of all path in that dir

        based on repository walk function

        :param repo_name: name of repository

        :param revision: revision for which to list nodes

        :param root_path: root path to list

        """

        _files = list()

        _dirs = list()

        try:

            _repo = self.__get_repo(repo_name)

            changeset = _repo.scm_instance.get_changeset(revision)

            root_path = root_path.lstrip('/')

            for topnode, dirs, files in changeset.walk(root_path):

                for f in files:

                    _files.append(f.path if flat else {"name": f.path,

                                                       "type": "file"})

                for d in dirs:

                    _dirs.append(d.path if flat else {"name": d.path,

                                                      "type": "dir"})

        except RepositoryError:

            log.debug(traceback.format_exc())

            raise

        return _dirs, _files

    def create_nodes(self, user, repo, message, nodes, parent_cs=None,

                     author=None, trigger_push_hook=True):

        """

        Commits given multiple nodes into repo

        :param user: Kallithea User object or user_id, the commiter

        :param repo: Kallithea Repository object

        :param message: commit message

        :param nodes: mapping {filename:{'content':content},...}

        :param parent_cs: parent changeset, can be empty than it's initial commit

        :param author: author of commit, cna be different that commiter only for git

        :param trigger_push_hook: trigger push hooks

        :returns: new commited changeset

        """

        user = self._get_user(user)

        scm_instance = repo.scm_instance_no_cache()

        processed_nodes = []

        for f_path in nodes:

            f_path = self._sanitize_path(f_path)

            content = nodes[f_path]['content']

            f_path = safe_str(f_path)

            # decoding here will force that we have proper encoded values

            # in any other case this will throw exceptions and deny commit

            if isinstance(content, (basestring,)):

            if op == 'add':

                imc.add(filenode)

            elif op == 'del':

                imc.remove(filenode)

            elif op == 'mod':

                if filename != old_filename:

                    #TODO: handle renames, needs vcs lib changes

                    imc.remove(filenode)

                    imc.add(FileNode(filename, content=content))

                else:

                    imc.change(filenode)

        # commit changes

        tip = imc.commit(message=message,

                         author=author,

                         parents=parents,

                         branch=parent_cs.branch)

        self.mark_for_invalidation(repo.repo_name)

        if trigger_push_hook:

            self._handle_push(scm_instance,

                              username=user.username,

                              action='push_local',

                              repo_name=repo.repo_name,

                              revisions=[tip.raw_id])

    def delete_nodes(self, user, repo, message, nodes, parent_cs=None,

                     author=None, trigger_push_hook=True):

        """

        Deletes given multiple nodes into repo

        :param user: Kallithea User object or user_id, the commiter

        :param repo: Kallithea Repository object

        :param message: commit message

        :param nodes: mapping {filename:{'content':content},...}

        :param parent_cs: parent changeset, can be empty than it's initial commit

        :param author: author of commit, cna be different that commiter only for git

        :param trigger_push_hook: trigger push hooks

        :returns: new commited changeset after deletion

        """

        user = self._get_user(user)

        scm_instance = repo.scm_instance_no_cache()

        processed_nodes = []

        for f_path in nodes:

            f_path = self._sanitize_path(f_path)

            # structure as add_nodes

            content = nodes[f_path].get('content')

            processed_nodes.append((f_path, content))

        message = safe_unicode(message)

        commiter = user.full_contact

        author = safe_unicode(author) if author else commiter

        IMC = self._get_IMC_module(scm_instance.alias)

        imc = IMC(scm_instance)

        if not parent_cs:

            parent_cs = EmptyChangeset(alias=scm_instance.alias)

        if isinstance(parent_cs, EmptyChangeset):

            # EmptyChangeset means we we're editing empty repository

            parents = None

        else:

            parents = [parent_cs]

        # add multiple nodes

        for path, content in processed_nodes:

            imc.remove(FileNode(path, content=content))

        tip = imc.commit(message=message,

                         author=author,

                         parents=parents,

                         branch=parent_cs.branch)

        self.mark_for_invalidation(repo.repo_name)

        if trigger_push_hook:

            self._handle_push(scm_instance,

                              username=user.username,

                              action='push_local',

                              repo_name=repo.repo_name,

                              revisions=[tip.raw_id])

        return tip

    def get_unread_journal(self):

        return self.sa.query(UserLog).count()

    def get_repo_landing_revs(self, repo=None):

        """

        Generates select option with tags branches and bookmarks (for hg only)

        grouped by type

        :param repo:

        """

                run_task(tasks.send_email, [user_email],

                         _("Password reset link"), body, body)

                log.info('send new password mail to %s' % user_email)

            else:

                log.debug("password reset email %s not found" % user_email)

        except Exception:

            log.error(traceback.format_exc())

            return False

        return True

    def reset_password(self, data):

        from kallithea.lib.celerylib import tasks, run_task

        from kallithea.lib import auth

        user_email = data['email']

        pre_db = True

        try:

            user = User.get_by_email(user_email)

            new_passwd = auth.PasswordGenerator().gen_password(8,

                            auth.PasswordGenerator.ALPHABETS_BIG_SMALL)

            if user:

                user.password = auth.get_crypt_password(new_passwd)

                Session().add(user)

                Session().commit()

                log.info('change password for %s' % user_email)

            if new_passwd is None:

                raise Exception('unable to generate new password')

            pre_db = False

            run_task(tasks.send_email, [user_email],

                     _('Your new password'),

                     _('Your new Kallithea password:%s') % (new_passwd,))

            log.info('send new password mail to %s' % user_email)

        except Exception:

            log.error('Failed to update user password')

            log.error(traceback.format_exc())

            if pre_db:

                # we rollback only if local db stuff fails. If it goes into

                # run_task, we're pass rollback state this wouldn't work then

                Session().rollback()

        return True

    def fill_data(self, auth_user, user_id=None, api_key=None, username=None):

        """

        Fetches auth_user by user_id,or api_key if present.

        Fills auth_user attributes with those taken from database.

        present in database

        :param auth_user: instance of user to set attributes

        :param user_id: user id to fetch by

        :param api_key: api key to fetch by

        :param username: username to fetch by

        """

        if user_id is None and api_key is None and username is None:

            raise Exception('You need to pass user_id, api_key or username')

        try:

            dbuser = None

            if user_id:

                dbuser = self.get(user_id)

            elif api_key:

                dbuser = self.get_by_api_key(api_key)

            elif username:

                dbuser = self.get_by_username(username)

            if dbuser is not None and dbuser.active:

                log.debug('filling %s data' % dbuser)

                for k, v in dbuser.get_dict().iteritems():

                    if k not in ['api_keys', 'permissions']:

                        setattr(auth_user, k, v)

            else:

                return False

        except Exception:

            log.error(traceback.format_exc())

            auth_user.is_authenticated = False

            return False

        return True

    def has_perm(self, user, perm):

        perm = self._get_perm(perm)

        user = self._get_user(user)

        return UserToPerm.query().filter(UserToPerm.user == user)\

            .filter(UserToPerm.permission == perm).scalar() is not None

    def grant_perm(self, user, perm):

        """

        Grant user global permissions

        :param user:

        :param perm:

        """