Hi Jeff,
Il 19/01/2012 14:08, Jeff Reback ha scritto:
> Hi,
>
> using the configuration:
>
> pytables 2.3.1
> numexpr 1.4.1
> python 2.7.1 (on adm64 debian and win64)
>
> using a readWhere to select rows from a table, if I give a selector with
> multiple operands on the index column (e.g. (column > value) & ( column <
> value2))
> doesn't seem to work (though works fine with a single operand and on a
> non-indexed table)
>
> in the test output the 4th case (index with start and stop operands), I don't
> receive any selection (contrast with the 2nd case which shows non-indexed
> behavior)
>
> is this behavior expected?
>
> thanks,
>
> Jeff
>
> ----------- test script -----------
>
> #!/usr/local/bin/python
>
> import tables
> import numpy as np
> import datetime, time
>
> # create table
> def create(name, add_index = False):
> test_file = "%s.hdf" % name
> handle = tables.openFile(test_file, "w")
>
> table = handle.createTable(handle.root, 'table', dict(
> index = tables.Time64Col(),
> column = tables.StringCol(25),
> values = tables.FloatCol(shape=(3)),
> ))
>
> # add data
> date = datetime.datetime(2011,1,1,8,0,0)
>
> r = table.row
> for i in xrange(100):
> r['index'] = time.mktime((date +
> datetime.timedelta(days=i)).timetuple())
> r['column'] = ("str-%d" % (i % 5))
> r['values'] = np.arange(3)
> r.append()
> table.flush()
>
> if add_index:
> col = table.cols._f_col('index')
> col.createIndex(filters = None)
>
> handle.close()
> return test_file
>
> def select(name, start, stop = None):
> """ start and stop are dates """
>
> test_file = "%s.hdf" % name
> handle = tables.openFile(test_file,"r")
> selectors = []
>
> # index selector
> selectors.append("(index >= %s)" % time.mktime(start.timetuple()))
> if stop is not None:
> selectors.append("(index <= %s)" % time.mktime(stop.timetuple()))
>
> # column selector
> selectors.append("((column == 'str-0') | (column == 'str-1'))")
>
> selector = ' & '.join(selectors)
>
> print "selector -> [f->%s,start->%s,stop->%s] --> %s" %
> (test_file,start,stop,selector)
> ans = getattr(handle.root,'table').readWhere(selector)
> print "ans -> %s" % ans
> handle.close()
>
>
> # no indexing
> create('no_index',add_index = False)
> select('no_index', start = datetime.datetime(2011,2,1,0,0,0))
> select('no_index', start = datetime.datetime(2011,2,1,0,0,0), stop =
> datetime.datetime(2011,3,1,0,0,0))
>
> # with indexing
> create('with_index',add_index = True)
> select('with_index', start = datetime.datetime(2011,2,1,0,0,0))
> select('with_index', start = datetime.datetime(2011,2,1,0,0,0), stop =
> datetime.datetime(2011,3,1,0,0,0))
>
> ------ptdump -v on the no_index.hdf--------
> [cow-jreback-/tmp] ptdump -v no_index.hdf
> / (RootGroup) ''
> /table (Table(100,)) ''
> description := {
> "column": StringCol(itemsize=25, shape=(), dflt='', pos=0),
> "index": Time64Col(shape=(), dflt=0.0, pos=1),
> "values": Float64Col(shape=(3,), dflt=0.0, pos=2)}
> byteorder := 'little'
> chunkshape := (1149,)
>
> ------ ptdump -v on the with_index.hdf --------
>
> [cow-jreback-/tmp] ptdump -v with_index.hdf
> / (RootGroup) ''
> /table (Table(100,)) ''
> description := {
> "column": StringCol(itemsize=25, shape=(), dflt='', pos=0),
> "index": Time64Col(shape=(), dflt=0.0, pos=1),
> "values": Float64Col(shape=(3,), dflt=0.0, pos=2)}
> byteorder := 'little'
> chunkshape := (1149,)
> autoIndex := True
> colindexes := {
> "index": Index(6, medium, shuffle, zlib(1)).is_CSI=False}
>
> --------test output -------
> selector -> [f->no_index.hdf,start->2011-02-01 00:00:00,stop->None] -->
> (index >= 1296536400.0) & ((column == 'str-0') | (column == 'str-1'))
> ans -> [('str-1', 1296565200.0, [0.0, 1.0, 2.0])
> ('str-0', 1296910800.0, [0.0, 1.0, 2.0])
> ('str-1', 1296997200.0, [0.0, 1.0, 2.0])
> ('str-0', 1297342800.0, [0.0, 1.0, 2.0])
> ('str-1', 1297429200.0, [0.0, 1.0, 2.0])
> ('str-0', 1297774800.0, [0.0, 1.0, 2.0])
> ('str-1', 1297861200.0, [0.0, 1.0, 2.0])
[CUT]
IMHO your problem is related to a known issue [1].
Can you please confirm?
I linked your message to the issue but please feel free to add
additional info.
thanks for reporting.
We will try to fix the issue ASAP.
best regards
[1] https://github.com/PyTables/PyTables/issues/119
--
Antonio Valentino
------------------------------------------------------------------------------
Keep Your Developer Skills Current with LearnDevNow!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-d2d
_______________________________________________
Pytables-users mailing list
Pytables-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/pytables-users
