Example #1
0
def stat_input_stream(fd, size, url, params):
    from disco import util
    from hustle.core.marble import MarbleStream

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

    otab = None
    try:
        # print "FLurlG: %s" % url
        fle = util.localize(rest, disco_data=params._task.disco_data,
                            ddfs_data=params._task.ddfs_data)
        # print "FLOGLE: %s" % fle
        otab = MarbleStream(fle)
        rows = otab.number_rows
        frows = float(rows)
        rval = {'_': rows, }
        for field, (subdb, subindexdb, _, column, _) in otab.dbs.iteritems():
            if subindexdb:
                rval[field] = subindexdb.stat(otab.txn)['ms_entries'] / frows
        yield '', rval
    except Exception as e:
        print "Gibbers: %s" % e
        raise e
    finally:
        if otab:
            otab.close()
Example #2
0
def stat_input_stream(fd, size, url, params):
    from disco import util
    from hustle.core.marble import MarbleStream

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

    otab = None
    try:
        # print "FLurlG: %s" % url
        fle = util.localize(rest,
                            disco_data=params._task.disco_data,
                            ddfs_data=params._task.ddfs_data)
        # print "FLOGLE: %s" % fle
        otab = MarbleStream(fle)
        rows = otab.number_rows
        frows = float(rows)
        rval = {
            '_': rows,
        }
        for field, (subdb, subindexdb, _, column, _) in otab.dbs.iteritems():
            if subindexdb:
                rval[field] = subindexdb.stat(otab.txn)['ms_entries'] / frows
        yield '', rval
    except Exception as e:
        print "Gibbers: %s" % e
        raise e
    finally:
        if otab:
            otab.close()
Example #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
    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()
Example #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()
Example #5
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()
Example #6
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()
Example #7
0
def Open(url, task=None):
    if task:
        disco_data = task.disco_data
        ddfs_data = task.ddfs_data
    else:
        from disco.settings import DiscoSettings
        settings = DiscoSettings()
        disco_data = settings['DISCO_DATA']
        ddfs_data = settings['DDFS_DATA']
    scheme, netloc, rest = util.urlsplit(url)
    path, rest = rest.split('!', 1) if '!' in rest else (rest, '')
    discodb = DiscoDB.load(open(util.localize(path, disco_data=disco_data,
                                ddfs_data=ddfs_data)))

    if rest:
        method_name, arg = rest.split('/', 1) if '/' in rest else (rest, None)
        method = getattr(discodb, method_name)
        if method_name in ('metaquery', 'query'):
            return method(Q.urlscan(arg))
        return method(*filter(None, arg))
    return discodb
Example #8
0
def filename_input_stream(fd, size, url, params):
    """This input_stream simply returns the path of the local disk file for this map job"""
    from disco import util
    from disco.worker.classic import worker

    try:
        scheme, netloc, rest = util.urlsplit(url)
        netloc = "%s:%s" % netloc if netloc[1] else netloc[0]
    except Exception as e:
        msg = "Error handling hustle_input_stream for %s. %s" % (url, e)
        raise util.DataError(msg, url)

    if scheme == 'file':
        yield url, "/%s" % rest
    else:
        # print url, rest
        fle = util.localize(rest,
                            disco_data=worker.Task.disco_data,
                            ddfs_data=worker.Task.ddfs_data)

        yield url, fle
Example #9
0
def filename_input_stream(fd, size, url, params):
    """This input_stream simply returns the path of the local disk file for this map job"""
    from disco import util
    from disco.worker.classic import worker

    try:
        scheme, netloc, rest = util.urlsplit(url)
        netloc = "%s:%s" % netloc if netloc[1] else netloc[0]
    except Exception as e:
        msg = "Error handling hustle_input_stream for %s. %s" % (url, e)
        raise util.DataError(msg, url)

    if scheme == 'file':
        yield url, "/%s" % rest
    else:
        # print url, rest
        fle = util.localize(rest,
                            disco_data=worker.Task.disco_data,
                            ddfs_data=worker.Task.ddfs_data)

        yield url, fle
Example #10
0
def Open(url, task=None):
    if task:
        disco_data = task.disco_data
        ddfs_data = task.ddfs_data
    else:
        from disco.settings import DiscoSettings
        settings = DiscoSettings()
        disco_data = settings['DISCO_DATA']
        ddfs_data = settings['DDFS_DATA']
    scheme, netloc, rest = util.urlsplit(url)
    path, rest = rest.split('!', 1) if '!' in rest else (rest, '')
    discodb = DiscoDB.load(
        open(util.localize(path, disco_data=disco_data, ddfs_data=ddfs_data)))

    if rest:
        method_name, arg = rest.split('/', 1) if '/' in rest else (rest, None)
        method = getattr(discodb, method_name)
        if method_name in ('metaquery', 'query'):
            return method(Q.urlscan(arg))
        return method(*filter(None, arg))
    return discodb
Example #11
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()
Example #12
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()