def aggregate(self, cell, measures=None, drill_down=None):
"""See :meth:`cubes.browsers.cell.aggregate`."""
###################################################
# 1. Prepare cell selector
if drill_down:
drill_dimension = self.cube.dimension(drill_down)
else:
drill_dimension = None
selector = self.selector_object(cell, drill_dimension)
condition = {self.cell_selector_name: selector}
condition[self.aggregate_flag_field] = True
if drill_down:
drill_dimension = self.cube.dimension(drill_down)
else:
drill_dimension = None
###################################################
# 2. Prepare dimension filter conditions
dim_conditions = {}
for cut in cell.cuts:
if type(cut) != PointCut:
raise AttributeError(
"only point cuts are currently supported for mongo aggregation browsing"
)
dimension = self.cube.dimension(cut.dimension)
path = cut.path
dim_levels = dimension.default_hierarchy.levels
# Get physical field names from field mappings specified in cube and use them
# in selection condition
for i, value in enumerate(path):
level = dim_levels[i]
mapped = base.dimension_field_mapping(self.cube, dimension,
level.key)
dim_conditions[mapped[0]] = value
# Expand dictionary: convert key1.key2 = value into 'key1 : { key2 : value}'
dim_conditions = cubes.common.expand_dictionary(dim_conditions)
condition.update(dim_conditions)
###################################################
# 3. Perform selection - find records in collection
cursor = self.collection.find(spec=condition)
return cursor
def aggregate(self, cell, measures = None, drill_down = None):
"""See :meth:`cubes.browsers.cell.aggregate`."""
###################################################
# 1. Prepare cell selector
if drill_down:
drill_dimension = self.cube.dimension(drill_down)
else:
drill_dimension = None
selector = self.selector_object(cell, drill_dimension)
condition = { self.cell_selector_name: selector }
condition[self.aggregate_flag_field] = True
if drill_down:
drill_dimension = self.cube.dimension(drill_down)
else:
drill_dimension = None
###################################################
# 2. Prepare dimension filter conditions
dim_conditions = {}
for cut in cell.cuts:
if type(cut) != PointCut:
raise AttributeError("only point cuts are currently supported for mongo aggregation browsing")
dimension = self.cube.dimension(cut.dimension)
path = cut.path
dim_levels = dimension.default_hierarchy.levels
# Get physical field names from field mappings specified in cube and use them
# in selection condition
for i, value in enumerate(path):
level = dim_levels[i]
mapped = base.dimension_field_mapping(self.cube, dimension, level.key)
dim_conditions[mapped[0]] = value
# Expand dictionary: convert key1.key2 = value into 'key1 : { key2 : value}'
dim_conditions = cubes.util.expand_dictionary(dim_conditions)
condition.update(dim_conditions)
###################################################
# 3. Perform selection - find records in collection
cursor = self.collection.find(spec = condition)
return cursor
def compute_cell(self, selector):
"""
Compute aggregation for cell specified by selector. cell is computed using MongoDB
aggregate_ function. Computed records are inserted into `cube_collection` and they contain:
* key fields used for grouping
* aggregated measures suffixed with `_sum`, for example: `amount_sum`
* record count in `record_count`
* cell selector as `_selector` (configurable) with dimension names as keys and current
dimension levels as values, for example: {"date": ["year", "month"] }
* cell reference as `_cell` (configurable) with dimension names as keys and level
keys forming dimension paths as values, for example: {"date": [2010, 10] }
.. _aggregate: http://www.mongodb.org/display/DOCS/Aggregation#Aggregation-Group
:Arguments:
* `selector` is a list of tuples: (dimension, level_names)
.. note::
Only 'sum' aggregation is being computed. Other aggregations might be implemented in the
future, such as average, min, max, rank, ...
"""
self.log.info("computing selector")
key_maps = []
attrib_maps = []
selector_record = {}
for dimsel in selector:
dim = dimsel[0]
levels = dimsel[1]
self.log.info("-- dimension: %s levels: %s", dim.name, levels)
level_names = []
for level in levels:
level_names.append(level.name)
mapped = base.dimension_field_mapping(self.cube, dim,
level.key)
key_maps.append(mapped)
for field in level.attributes:
mapped = base.dimension_field_mapping(
self.cube, dim, field)
attrib_maps.append((mapped[0], field))
selector_record[dim.name] = level_names
###########################################
# Prepare group command parameters
#
# condition - filter condition for find() (check for existence of keys)
# keys - list of keys to be used for grouping
# measures - list of measures
condition = {}
keys = []
for mapping in key_maps:
mapped_key = mapping[0]
condition[mapped_key] = {"$exists": True}
keys.append(mapped_key)
fields = []
for mapping in attrib_maps:
mapped = mapping[0]
fields.append(mapped)
for measure in self.measures:
mapping = base.fact_field_mapping(self.cube, measure)
fields.append(mapping[0])
self.log.info("condition: %s", condition)
self.log.info("keys: %s", keys)
self.log.info("fields: %s", fields)
# Exclude aggregates:
condition[self.aggregate_flag_field] = {'$ne': True}
####################################################
# Prepare group functions + reduce + finalize
#
initial = {"record_count": 0}
aggregate_lines = []
for measure in self.measures:
measure_agg_name = measure + "_sum"
line = "out.%s += doc.%s;" % (measure_agg_name, measure)
aggregate_lines.append(line)
initial[measure_agg_name] = 0
reduce_function = '''
function(doc, out) {
out.record_count ++;
%(aggregate_lines)s
}\n''' % {
"aggregate_lines": "\n".join(aggregate_lines)
}
finalize_function = None
cursor = self.fact_collection.group(key=keys,
condition=condition,
initial=initial,
reduce=reduce_function,
finalize=finalize_function)
for record in cursor:
# use: cubes.commons.expand_dictionary(record)
cell = {}
for dimsel in selector:
dimension, levels = dimsel
path = []
for level in levels:
mapped = base.dimension_field_mapping(
self.cube, dimension, level.key)
path.append(record[mapped[0]])
cell[dimension.name] = path
record = self.construct_record(record)
record[self.aggregate_flag_field] = True
record[self.cell_record_name] = selector_record
record[self.cell_reference_record_name] = cell
self.cube_collection.insert(record)
def compute_cell(self, selector):
"""
Compute aggregation for cell specified by selector. cell is computed using MongoDB
aggregate_ function. Computed records are inserted into `cube_collection` and they contain:
* key fields used for grouping
* aggregated measures suffixed with `_sum`, for example: `amount_sum`
* record count in `record_count`
* cell selector as `_selector` (configurable) with dimension names as keys and current
dimension levels as values, for example: {"date": ["year", "month"] }
* cell reference as `_cell` (configurable) with dimension names as keys and level
keys forming dimension paths as values, for example: {"date": [2010, 10] }
.. _aggregate: http://www.mongodb.org/display/DOCS/Aggregation#Aggregation-Group
:Arguments:
* `selector` is a list of tuples: (dimension, level_names)
.. note::
Only 'sum' aggregation is being computed. Other aggregations might be implemented in the
future, such as average, min, max, rank, ...
"""
self.log.info("computing selector")
key_maps = []
attrib_maps = []
selector_record = {}
for dimsel in selector:
dim = dimsel[0]
levels = dimsel[1]
self.log.info("-- dimension: %s levels: %s", dim.name, levels)
level_names = []
for level in levels:
level_names.append(level.name)
mapped = base.dimension_field_mapping(self.cube, dim, level.key)
key_maps.append(mapped)
for field in level.attributes:
mapped = base.dimension_field_mapping(self.cube, dim, field)
attrib_maps.append((mapped[0], field))
selector_record[dim.name] = level_names
###########################################
# Prepare group command parameters
#
# condition - filter condition for find() (check for existence of keys)
# keys - list of keys to be used for grouping
# measures - list of measures
condition = {}
keys = []
for mapping in key_maps:
mapped_key = mapping[0]
condition[mapped_key] = { "$exists" : True}
keys.append(mapped_key)
fields = []
for mapping in attrib_maps:
mapped = mapping[0]
fields.append(mapped)
for measure in self.measures:
mapping = base.fact_field_mapping(self.cube, measure)
fields.append(mapping[0])
self.log.info("condition: %s", condition)
self.log.info("keys: %s", keys)
self.log.info("fields: %s", fields)
# Exclude aggregates:
condition[self.aggregate_flag_field] = {'$ne': True}
####################################################
# Prepare group functions + reduce + finalize
#
initial = { "record_count": 0 }
aggregate_lines = []
for measure in self.measures:
measure_agg_name = measure + "_sum"
line = "out.%s += doc.%s;" % (measure_agg_name, measure)
aggregate_lines.append(line)
initial[measure_agg_name] = 0
reduce_function = '''
function(doc, out) {
out.record_count ++;
%(aggregate_lines)s
}\n''' % {"aggregate_lines": "\n".join(aggregate_lines)}
finalize_function = None
cursor = self.fact_collection.group(key = keys, condition = condition,
initial = initial, reduce = reduce_function,
finalize = finalize_function)
for record in cursor:
# use: cubes.commons.expand_dictionary(record)
cell = {}
for dimsel in selector:
dimension, levels = dimsel
path = []
for level in levels:
mapped = base.dimension_field_mapping(self.cube, dimension, level.key)
path.append(record[mapped[0]])
cell[dimension.name] = path
record = self.construct_record(record)
record[self.aggregate_flag_field] = True
record[self.cell_record_name] = selector_record
record[self.cell_reference_record_name] = cell
self.cube_collection.insert(record)
