Exemple #1
0
 def test_marble_stream_get(self):
     for date, file in self.files.iteritems():
         stream = MarbleStream(file)
         rowid = 1
         for album in self.albums:
             if album[_PARTITIONS] != date:
                 continue
             # test 'get' first
             for k, v in album.iteritems():
                 self.assertEqual(v, stream.get(k, rowid))
             rowid += 1
         stream.close()
Exemple #2
0
 def test_marble_stream_get(self):
     for date, file in self.files.iteritems():
         stream = MarbleStream(file)
         rowid = 1
         for album in self.albums:
             if album[_PARTITIONS] != date:
                 continue
             # test 'get' first
             for k, v in album.iteritems():
                 self.assertEqual(v, stream.get(k, rowid))
             rowid += 1
         stream.close()
Exemple #3
0
def hustle_input_stream(fd, size, url, params, wheres, gen_where_index,
                        key_names):
    from disco import util
    from hustle.core.marble import Expr, MarbleStream
    empty = ()

    try:
        scheme, netloc, rest = util.urlsplit(url)
    except Exception as e:
        print "Error handling hustle_input_stream for %s. %s" % (url, e)
        raise e

    fle = util.localize(rest,
                        disco_data=params._task.disco_data,
                        ddfs_data=params._task.ddfs_data)
    # print "FLOGLE: %s %s" % (url, fle)

    otab = None
    try:
        # import sys
        # sys.path.append('/Library/Python/2.7/site-packages/pycharm-debug.egg')
        # import pydevd
        # pydevd.settrace('localhost', port=12999, stdoutToServer=True, stderrToServer=True)
        otab = MarbleStream(fle)
        bitmaps = {}
        for index, where in enumerate(wheres):
            # do not process where clauses that have nothing to do with this marble
            if where._name == otab.marble._name:
                if type(where) is Expr and not where.is_partition:
                    bitmaps[index] = where(otab)
                else:
                    # it is either the table itself, or a partition expression.  either way,
                    # return the entire table
                    bitmaps[index] = otab.iter_all()

        for index, bitmap in bitmaps.iteritems():
            prefix = [index] if gen_where_index else []
            for row_id in bitmap:
                record = [
                    otab.get(col, row_id) if col else None
                    for col in key_names[index]
                ]
                # print "Gibbled: %s" % repr(record)
                record[
                    0:
                    0] = prefix  # this looks odd, but is faster than 'prefix + record'
                yield tuple(record), empty
    finally:
        if otab:
            otab.close()
Exemple #4
0
def hustle_input_stream(fd, size, url, params, wheres, gen_where_index, key_names):
    from disco import util
    from hustle.core.marble import Expr, MarbleStream
    empty = ()

    try:
        scheme, netloc, rest = util.urlsplit(url)
    except Exception as e:
        print "Error handling hustle_input_stream for %s. %s" % (url, e)
        raise e

    fle = util.localize(rest, disco_data=params._task.disco_data, ddfs_data=params._task.ddfs_data)
    # print "FLOGLE: %s %s" % (url, fle)

    otab = None
    try:
        # import sys
        # sys.path.append('/Library/Python/2.7/site-packages/pycharm-debug.egg')
        # import pydevd
        # pydevd.settrace('localhost', port=12999, stdoutToServer=True, stderrToServer=True)
        otab = MarbleStream(fle)
        bitmaps = {}
        for index, where in enumerate(wheres):
            # do not process where clauses that have nothing to do with this marble
            if where._name == otab.marble._name:
                if type(where) is Expr and not where.is_partition:
                    bitmaps[index] = where(otab)
                else:
                    # it is either the table itself, or a partition expression.  either way,
                    # return the entire table
                    bitmaps[index] = otab.iter_all()

        for index, bitmap in bitmaps.iteritems():
            prefix = [index] if gen_where_index else []
            for row_id in bitmap:
                record = [otab.get(col, row_id) if col else None for col in key_names[index]]
                # print "Gibbled: %s" % repr(record)
                record[0:0] = prefix  # this looks odd, but is faster than 'prefix + record'
                yield tuple(record), empty
    finally:
        if otab:
            otab.close()