"""The `ResultSet` class which is returned as result of a rql query

:organization: Logilab

:copyright: 2001-2009 LOGILAB S.A. (Paris, FRANCE), license is LGPL v2.

:contact: http://www.logilab.fr/ -- mailto:contact@logilab.fr

:license: GNU Lesser General Public License, v2.1 - http://www.gnu.org/licenses

"""

__docformat__ = "restructuredtext en"

from logilab.common.decorators import cached, clear_cache, copy_cache

from rql import nodes, stmts

from cubicweb import NotAnEntity

class ResultSet(object):

    """a result set wrap a RQL query result. This object implements a partial

    list protocol to allow direct use as a list of result rows.

    :type rowcount: int

    :ivar rowcount: number of rows in the result

    :type rows: list

    :ivar rows: list of rows of result

    :type description: list

    :ivar description:

      result's description, using the same structure as the result itself

    :type rql: str or unicode

    :ivar rql: the original RQL query string

    """

    def __init__(self, results, rql, args=None, description=(), cachekey=None,

                 rqlst=None):

        self.rows = results

        self.rowcount = results and len(results) or 0

        # original query and arguments

        self.rql = rql

        self.args = args

        self.cachekey = cachekey

        # entity types for each cell (same shape as rows)

        # maybe discarded if specified when the query has been executed

        self.description = description

        # parsed syntax tree

        if rqlst is not None:

            rqlst.schema = None # reset schema in case of pyro transfert

        self._rqlst = rqlst

        # set to (limit, offset) when a result set is limited using the

        # .limit method

        self.limited = None

        # set by the cursor which returned this resultset

        self.vreg = None

        self.req = None

        # actions cache

        self._rsetactions = None

    def __str__(self):

        if not self.rows:

            return '<empty resultset %s>' % self.rql

        return '<resultset %s (%s rows)>' % (self.rql, len(self.rows))

    def __repr__(self):

        if not self.rows:

            return '<empty resultset for %r>' % self.rql

        rows = self.rows

        if len(rows) > 10:

            rows = rows[:10] + ['...']

        if not self.description:

            return '<resultset %r (%s rows): %s>' % (self.rql, len(self.rows),

                                                     '\n'.join(str(r) for r in rows))

        return '<resultset %r (%s rows): %s>' % (self.rql, len(self.rows),

                                                 '\n'.join('%s (%s)' % (r, d)

                                                           for r, d in zip(rows, self.description)))

    def possible_actions(self, **kwargs):

        if self._rsetactions is None:

            self._rsetactions = {}

        if kwargs:

            key = tuple(sorted(kwargs.iteritems()))

        else:

            key = None

        try:

            return self._rsetactions[key]

        except KeyError:

            actions = self.vreg['actions'].poss_visible_objects(

                self.req, rset=self, **kwargs)

            self._rsetactions[key] = actions

            return actions

    def __len__(self):

        """returns the result set's size"""

        return self.rowcount

    def __nonzero__(self):

        return self.rowcount

    def __getitem__(self, i):

        """returns the ith element of the result set"""

        return self.rows[i] #ResultSetRow(self.rows[i])

    def __getslice__(self, i, j):

        """returns slice [i:j] of the result set"""

        return self.rows[i:j]

    def __iter__(self):

        """Returns an iterator over rows"""

        return iter(self.rows)

    def __add__(self, rset):

        # XXX buggy implementation (.rql and .args attributes at least much

        # probably differ)

        # at least rql could be fixed now that we have union and sub-queries

        # but I tend to think that since we have that, we should not need this

        # method anymore (syt)

        rset = ResultSet(self.rows+rset.rows, self.rql, self.args,

                         self.description +rset.description)

        return self.req.decorate_rset(rset)

    def copy(self, rows=None, descr=None):

        if rows is None:

            rows = self.rows[:]

            descr = self.description[:]

        rset = ResultSet(rows, self.rql, self.args, descr)

        return self.req.decorate_rset(rset)

    def transformed_rset(self, transformcb):

        """ the result set according to a given column types

        :type transormcb: callable(row, desc)

        :param transformcb:

          a callable which should take a row and its type description as

          parameters, and return the transformed row and type description.

        :type col: int

        :param col: the column index

        :rtype: `ResultSet`

        """

        rows, descr = [], []

        rset = self.copy(rows, descr)

        for row, desc in zip(self.rows, self.description):

            nrow, ndesc = transformcb(row, desc)

            if ndesc: # transformcb returns None for ndesc to skip that row

                rows.append(nrow)

                descr.append(ndesc)

        rset.rowcount = len(rows)

        return rset

    def filtered_rset(self, filtercb, col=0):

        """filter the result set according to a given filtercb

        :type filtercb: callable(entity)

        :param filtercb:

          a callable which should take an entity as argument and return

          False if it should be skipped, else True

        :type col: int

        :param col: the column index

        :rtype: `ResultSet`

        """

        rows, descr = [], []

        rset = self.copy(rows, descr)

        for i in xrange(len(self)):

            if not filtercb(self.get_entity(i, col)):

                continue

            rows.append(self.rows[i])

            descr.append(self.description[i])

        rset.rowcount = len(rows)

        return rset

    def sorted_rset(self, keyfunc, reverse=False, col=0):

        """sorts the result set according to a given keyfunc

        :type keyfunc: callable(entity)

        :param keyfunc:

          a callable which should take an entity as argument and return

          the value used to compare and sort

        :type reverse: bool

        :param reverse: if the result should be reversed

        :type col: int

        :param col: the column index. if col = -1, the whole row are used

        :rtype: `ResultSet`

        """

        rows, descr = [], []

        rset = self.copy(rows, descr)

        if col >= 0:

            entities = sorted(enumerate(self.entities(col)),

                              key=lambda (i, e): keyfunc(e), reverse=reverse)

        else:

            entities = sorted(enumerate(self),

                              key=lambda (i, e): keyfunc(e), reverse=reverse)

        for index, _ in entities:

            rows.append(self.rows[index])

            descr.append(self.description[index])

        rset.rowcount = len(rows)

        return rset

    def split_rset(self, keyfunc=None, col=0, return_dict=False):

        """Splits the result set in multiple result set according to a given key

        :type keyfunc: callable(entity or FinalType)

        :param keyfunc:

          a callable which should take a value of the rset in argument and

          return the value used to group the value. If not define, raw value

          of the specified columns is used.

        :type col: int

        :param col: the column index. if col = -1, the whole row are used

        :type return_dict: Boolean

        :param return_dict: If true, the function return a mapping

            (key -> rset) instead of a list of rset

        :rtype: List of `ResultSet` or mapping of  `ResultSet`

        """

        result = []

        mapping = {}

        for idx, line in enumerate(self):

            if col >= 0:

                try:

                    key = self.get_entity(idx, col)

                except NotAnEntity:

                    key = line[col]

            else:

                key = line

            if keyfunc is not None:

                key = keyfunc(key)

            if key not in mapping:

                rows, descr = [], []

                rset = self.copy(rows, descr)

                mapping[key] = rset

                result.append(rset)

            else:

                rset = mapping[key]

            rset.rows.append(self.rows[idx])

            rset.description.append(self.description[idx])

        for rset in result:

            rset.rowcount = len(rset.rows)

        if return_dict:

            return mapping

        else:

            return result

    def limited_rql(self):

        """return a printable rql for the result set associated to the object,

        with limit/offset correctly set according to maximum page size and

        currently displayed page when necessary

        """

        # try to get page boundaries from the navigation component

        # XXX we should probably not have a ref to this component here (eg in

        #     cubicweb)

        nav = self.vreg['components'].select_or_none('navigation', self.req,

                                                     rset=self)

        if nav:

            start, stop = nav.page_boundaries()

            rql = self._limit_offset_rql(stop - start, start)

        # result set may have be limited manually in which case navigation won't

        # apply

        elif self.limited:

            rql = self._limit_offset_rql(*self.limited)

        # navigation component doesn't apply and rset has not been limited, no

        # need to limit query

        else:

            rql = self.printable_rql()

        return rql

    def _limit_offset_rql(self, limit, offset):

        rqlst = self.syntax_tree()

        if len(rqlst.children) == 1:

            select = rqlst.children[0]

            olimit, ooffset = select.limit, select.offset

            select.limit, select.offset = limit, offset

            rql = rqlst.as_string(kwargs=self.args)

            # restore original limit/offset

            select.limit, select.offset = olimit, ooffset

        else:

            newselect = stmts.Select()

            newselect.limit = limit

            newselect.offset = offset

            aliases = [nodes.VariableRef(newselect.get_variable(vref.name, i))

                       for i, vref in enumerate(rqlst.selection)]

            newselect.set_with([nodes.SubQuery(aliases, rqlst)], check=False)

            newunion = stmts.Union()

            newunion.append(newselect)

            rql = rqlst.as_string(kwargs=self.args)

            rqlst.parent = None

        return rql

    def limit(self, limit, offset=0, inplace=False):

        """limit the result set to the given number of rows optionaly starting

        from an index different than 0

        :type limit: int

        :param limit: the maximum number of results

        :type offset: int

        :param offset: the offset index

        :type inplace: bool

        :param inplace:

          if true, the result set is modified in place, else a new result set

          is returned and the original is left unmodified

        :rtype: `ResultSet`

        """

        stop = limit+offset

        rows = self.rows[offset:stop]

        descr = self.description[offset:stop]

        if inplace:

            rset = self

            rset.rows, rset.description = rows, descr

            rset.rowcount = len(rows)

            clear_cache(rset, 'description_struct')

            if offset:

                clear_cache(rset, 'get_entity')

            # we also have to fix/remove from the request entity cache entities

            # which get a wrong rset reference by this limit call

            for entity in self.req.cached_entities():

                if entity.cw_rset is self:

                    if offset <= entity.cw_row < stop:

                        entity.cw_row = entity.cw_row - offset

                    else:

                        self.req.drop_entity_cache(entity.eid)

        else:

            rset = self.copy(rows, descr)

            if not offset:

                # can copy built entity caches

                copy_cache(rset, 'get_entity', self)

        rset.limited = (limit, offset)

        return rset

    def printable_rql(self, encoded=False):

        """return the result set's origin rql as a string, with arguments

        substitued

        """

        encoding = self.req.encoding

        rqlstr = self.syntax_tree().as_string(encoding, self.args)

        # sounds like we get encoded or unicode string due to a bug in as_string

        if not encoded:

            if isinstance(rqlstr, unicode):

                return rqlstr

            return unicode(rqlstr, encoding)

        else:

            if isinstance(rqlstr, unicode):

                return rqlstr.encode(encoding)

            return rqlstr

    # client helper methods ###################################################

    def entities(self, col=0):

        """iter on entities with eid in the `col` column of the result set"""

        for i in xrange(len(self)):

            # may have None values in case of outer join (or aggregat on eid

            # hacks)

            if self.rows[i][col] is not None:

                yield self.get_entity(i, col)

    def complete_entity(self, row, col=0, skip_bytes=True):

        """short cut to get an completed entity instance for a particular

        row (all instance's attributes have been fetched)

        """

        entity = self.get_entity(row, col)

        entity.complete(skip_bytes=skip_bytes)

        return entity

    @cached

    def get_entity(self, row, col=None):

        """special method for query retreiving a single entity, returns a

        partially initialized Entity instance.

        WARNING: due to the cache wrapping this function, you should NEVER

                 give row as a named parameter (i.e. rset.get_entity(req, 0)

                 is OK but rset.get_entity(row=0, req=req) isn't

        :type row,col: int, int

        :param row,col:

          row and col numbers localizing the entity among the result's table

        :return: the partially initialized `Entity` instance

        """

        if col is None:

            from warnings import warn

            msg = '[3.2] col parameter will become mandatory in future version'

            warn(msg, DeprecationWarning, stacklevel=3)

            col = 0

        etype = self.description[row][col]

        try:

            eschema = self.vreg.schema.eschema(etype)

            if eschema.final:

                raise NotAnEntity(etype)

        except KeyError: