def test_notebook_dependency(fxsoldir, fxnewfile):
# run notebook first
nbdir = get_notebook_dir()
path = os.path.join(nbdir, 'test-notebook3.ipynb')
assert os.path.isfile(path)
with HDF('haje01') as hdf:
if 'test' in hdf.store:
del hdf.store['test']
update_notebook_by_run(path)
manifest = Manifest(True, path)
assert manifest._prev_files_chksum == manifest._dep_files_chksum
with HDF('haje01') as hdf:
prev_hdf_chksum = dataframe_checksum(hdf.store['test'])
print "prev_hdf_chksum {}".format(prev_hdf_chksum)
print len(hdf.store['test'])
# add new file
with open(fxnewfile, 'w') as f:
f.write('2014-03-05 23:30 [ERROR] - Async\n')
manifest = Manifest(False, path)
assert manifest._depend_files_changed
assert manifest._prev_files_chksum != manifest._dep_files_chksum
# run notebok again
update_notebook_by_run(path)
with HDF('haje01') as hdf:
new_hdf_chksum = dataframe_checksum(hdf.store['test'])
print "new_hdf_chksum {}".format(new_hdf_chksum)
print len(hdf.store['test'])
# check check
assert prev_hdf_chksum != new_hdf_chksum
def _append(self, value, aformat, append, columns, dropna, **kwargs):
import pandas as pd
ktuple = tuple(((k, v if isinstance(v, tuple) else tuple(v)) for k,
v in kwargs.items()))
_args = (dataframe_checksum(value), self.uname, self.sname, aformat,
append, columns, dropna, ktuple)
# escape None for hashing
args = map(lambda x: 0 if x is None else x, _args)
chksum = hash(tuple(args))
# logging.debug(' chksum: {}'.format(chksum))
self.manifest._out_hdf_chksum = chksum
logging.debug("HDFStore append - _out_hdf_chksum {}".format(chksum))
with HDF(self.uname) as hdf:
hdf.store.append(self.sname, value, aformat, append, columns,
dropna, **kwargs)
# write checksum
hsname = HDF_CHKSUM_FMT.format(self.sname)
if append:
pchksum = int(hdf.store[hsname])
chksum = pchksum + chksum
hdf.store.append(hsname, pd.Series([str(chksum)]), append=False)