Exemple #1
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
    from itertools import izip, repeat
    empty = ()

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

    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:
                    bm = where(otab)
                    bitmaps[index] = (bm, len(bm))
                else:
                    # it is either the table itself, or a partition expression.
                    # Either way, returns the entire table
                    bitmaps[index] = (otab.iter_all(), otab.number_rows)

        for index, (bitmap, blen) in bitmaps.iteritems():
            prefix_gen = [repeat(index, blen)] if gen_where_index else []

            row_iter = prefix_gen + \
                [otab.mget(col, bitmap) if col is not None else repeat(None, blen)
                 for col in key_names[index]]

            for row in izip(*row_iter):
                yield row, empty
    finally:
        if otab:
            otab.close()
Exemple #2
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
    from itertools import izip, repeat
    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:
                    bm = where(otab)
                    bitmaps[index] = (bm, len(bm))
                else:
                    # it is either the table itself, or a partition expression.
                    # Either way, returns the entire table
                    bitmaps[index] = (otab.iter_all(), otab.number_rows)

        for index, (bitmap, blen) in bitmaps.iteritems():
            prefix_gen = [repeat(index, blen)] if gen_where_index else []

            row_iter = prefix_gen + \
                [otab.mget(col, bitmap) if col is not None else repeat(None, blen)
                 for col in key_names[index]]

            for row in izip(*row_iter):
                yield row, empty

    finally:
        if otab:
            otab.close()
Exemple #3
0
def hustle_input_stream(fd, size, url, params, wheres, gen_where_index,
                        key_names, limit):
    from disco import util
    from hustle.core.marble import Expr, MarbleStream
    from itertools import izip, repeat, islice, imap
    from sys import maxint
    from pyebset import BitSet

    empty = ()

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

    fle = util.localize(rest,
                        disco_data=params._task.disco_data,
                        ddfs_data=params._task.ddfs_data)

    otab = None
    try:
        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:
                    bm = where(otab)
                    if limit != maxint:
                        bs = BitSet()
                        for i in islice(bm, 0, limit):
                            bs.set(i)
                        bitmaps[index] = (bs, len(bs))
                    else:
                        bitmaps[index] = (bm, len(bm))
                else:
                    # it is either the table itself, or a partition expression.
                    # Either way, returns the entire table
                    if limit != maxint:
                        bs = BitSet()
                        for i in islice(otab.iter_all(), 0, limit):
                            bs.set(i)
                        bitmaps[index] = (bs, len(bs))
                    else:
                        bitmaps[index] = (otab.iter_all(), otab.number_rows)

        for index, (bitmap, blen) in bitmaps.iteritems():
            prefix_gen = [repeat(index, blen)] if gen_where_index else []

            # row_iter = prefix_gen + \
            # [otab.mget(col, bitmap) if col is not None else repeat(None, blen)
            # for col in key_names[index]]
            row_creators = []
            for col, column_fn in key_names[index]:
                if col is not None:
                    if column_fn is None:
                        row_creators.append(otab.mget(col, bitmap))
                    else:
                        row_creators.append(
                            imap(column_fn, otab.mget(col, bitmap)))
                else:
                    row_creators.append(repeat(None, blen))
            row_iter = prefix_gen + row_creators

            for row in izip(*row_iter):
                yield row, empty
    finally:
        if otab:
            otab.close()
Exemple #4
0
def hustle_input_stream(fd, size, url, params, wheres, gen_where_index, key_names, limit):
    from disco import util
    from hustle.core.marble import Expr, MarbleStream
    from itertools import izip, repeat, islice, imap
    from sys import maxint
    from pyebset import BitSet

    empty = ()

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

    fle = util.localize(rest, disco_data=params._task.disco_data,
                        ddfs_data=params._task.ddfs_data)

    otab = None
    try:
        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:
                    bm = where(otab)
                    if limit != maxint:
                        bs = BitSet()
                        for i in islice(bm, 0, limit):
                            bs.set(i)
                        bitmaps[index] = (bs, len(bs))
                    else:
                        bitmaps[index] = (bm, len(bm))
                else:
                    # it is either the table itself, or a partition expression.
                    # Either way, returns the entire table
                    if limit != maxint:
                        bs = BitSet()
                        for i in islice(otab.iter_all(), 0, limit):
                            bs.set(i)
                        bitmaps[index] = (bs, len(bs))
                    else:
                        bitmaps[index] = (otab.iter_all(), otab.number_rows)

        for index, (bitmap, blen) in bitmaps.iteritems():
            prefix_gen = [repeat(index, blen)] if gen_where_index else []

            # row_iter = prefix_gen + \
                # [otab.mget(col, bitmap) if col is not None else repeat(None, blen)
                 # for col in key_names[index]]
            row_creators = []
            for col, column_fn in key_names[index]:
                if col is not None:
                    if column_fn is None:
                        row_creators.append(otab.mget(col, bitmap))
                    else:
                        row_creators.append(imap(column_fn, otab.mget(col, bitmap)))
                else:
                    row_creators.append(repeat(None, blen))
            row_iter = prefix_gen + row_creators

            for row in izip(*row_iter):
                yield row, empty
    finally:
        if otab:
            otab.close()