def __init__(self, source_col, transform, stage=None):
"""
@param source_col is the original Column or LazyColumn that this object
represents a transformed version of.
@param transform should be a Transform object.
@param stage is a string identifier for what this particular stage
of transformation represents. This is so we can later
go through a series of transforms on a Column and
search for key transformation points. Stage names should
be unique, so you cannot set a stage name that is the same
as any LazyColumn that exists in our list of ancestors.
"""
self._source_col = source_col
self._transform = transform
self._original_col = source_col.original_col
# DataColumns don't have stage names, so search all the way up until
# the end
for lazy_col in self.transformed_ancestors:
if (stage is not None) and (stage == lazy_col.stage):
raise ValueError("Cannot create LazyColumn with stage '%s' " \
"because this identifier is used in ancestor"\
" column %s " % (stage, lazy_col))
self._stage = stage
# FIXME: The hashing mechanism for transforms needs work...
self._identity_hash = hashhelper.data_hash(source_col.identity_hash,
hash(transform))
# _transformed_col says: Don't look at me from outside! I don't have a
# hash, and throwing me around outside of this warm, cozy LazyColumn
# could have bad and not immediately apparent effects on performance.
self._transformed_col = None
def __init__(self, source_col, transform, stage=None):
"""
@param source_col is the original Column or LazyColumn that this object
represents a transformed version of.
@param transform should be a Transform object.
@param stage is a string identifier for what this particular stage
of transformation represents. This is so we can later
go through a series of transforms on a Column and
search for key transformation points. Stage names should
be unique, so you cannot set a stage name that is the same
as any LazyColumn that exists in our list of ancestors.
"""
self._source_col = source_col
self._transform = transform
self._original_col = source_col.original_col
# DataColumns don't have stage names, so search all the way up until
# the end
for lazy_col in self.transformed_ancestors:
if (stage is not None) and (stage == lazy_col.stage):
raise ValueError("Cannot create LazyColumn with stage '%s' " \
"because this identifier is used in ancestor"\
" column %s " % (stage, lazy_col))
self._stage = stage
# FIXME: The hashing mechanism for transforms needs work...
self._identity_hash = hashhelper.data_hash( source_col.identity_hash,
hash(transform) )
# _transformed_col says: Don't look at me from outside! I don't have a
# hash, and throwing me around outside of this warm, cozy LazyColumn
# could have bad and not immediately apparent effects on performance.
self._transformed_col = None
def value_hash(self):
"""Return a SHA1 hash of the contents of this DataColumn."""
if self._value_hash is None:
self._value_hash = hashhelper.data_hash(self._row_values)
return self._value_hash
def value_hash(self):
"""Return a SHA1 hash of the contents of this DataColumn."""
if self._value_hash is None:
self._value_hash = hashhelper.data_hash(self._row_values)
return self._value_hash
def param_hash(self):
return hashhelper.data_hash(self._value_mapping)
def hash(self):
return hashhelper.data_hash( hashhelper.source_hash(self.__class__),
self.param_hash() )