Merge 33e040d into 9c85afa

src/cr/cube/__init__.py

@@ -0,0 +1,5 @@
+# encoding: utf-8
+
+'''Shared constants'''
+
+ITEM_DIMENSION_TYPES = ('categorical_array', 'multiple_response')

src/cr/cube/crunch_cube.py

@@ -1,9 +1,12 @@
+# encoding: utf-8
+
 '''Home of the CrunchCube class.
 
 This module contains the definition of the CrunchCube class. It represents
 the open-source library used for manipulating the crunch cubes (JSON responses
 from the Crunch.io platform).
 '''
+
 from __future__ import division
 
 import json
@@ -17,6 +20,7 @@
 from .measures.index import Index
 from .measures.scale_means import ScaleMeans
 from .utils import lazyproperty
+from . import ITEM_DIMENSION_TYPES
 
 np.seterr(divide='ignore', invalid='ignore')
 
@@ -147,7 +151,7 @@ def _transform(self, res, include_transforms_for_dims,
                          i - dim_offset in include_transforms_for_dims)
             if dim.type == 'multiple_response':
                 dim_offset += 1
-            if not transform or dim.type == 'categorical_array':
+            if not transform or dim.type in ITEM_DIMENSION_TYPES:
                 continue
             # Perform transformations
             insertions = self._insertions(res, dim, i)
@@ -947,6 +951,7 @@ def hs_dims_for_den(hs_dims, axis):
 
         # Calculate nominator from table (include all H&S dimensions).
         num = self._transform(table, include_transforms_for_dims, inflate=True)
+        # num = self._transform(table, hs_dims, inflate=True)
 
         res = self._fix_shape(num / den)
 
@@ -1041,7 +1046,7 @@ def index(self, weighted=True, prune=False):
 
     def _calculate_std_res(self, counts, total, colsum, rowsum, slice_):
         dim_types = slice_.dim_types
-        has_mr_or_ca = 'categorical_array' in dim_types or 'multiple_response' in dim_types
+        has_mr_or_ca = set(dim_types) & set(ITEM_DIMENSION_TYPES)
         # if self.has_mr or self.ca_dim_ind is not None:
         if has_mr_or_ca:
             if not self.is_double_mr and (self.mr_dim_ind == 0 or self.mr_dim_ind == 1 and self.ndim == 3):

src/cr/cube/dimension.py

@@ -1,9 +1,12 @@
+# encoding: utf-8
+
 '''Contains implementation of the Dimension class, for Crunch Cubes.'''
 
 import numpy as np
 
 from .subtotal import Subtotal
 from .utils import lazyproperty, memoize
+from . import ITEM_DIMENSION_TYPES
 
 
 class Dimension(object):
@@ -115,8 +118,8 @@ def _elements(self):
     @property
     def inserted_hs_indices(self):
         '''Returns inserted H&S indices for the dimension.'''
-        if (self.type == 'categorical_array' or not self.subtotals):
-            return []  # For CA subvariables, we don't do H&S insertions
+        if (self.type in ITEM_DIMENSION_TYPES or not self.subtotals):
+            return []  # For CA and MR items, we don't do H&S insertions
 
         elements = self.elements()
         element_ids = [element['id'] for element in elements]

src/cr/cube/utils/__init__.py

@@ -1,6 +1,7 @@
 '''Utility functions for crunch cube, as well as other modules.'''
 import collections
 import functools
+import numpy as np
 
 try:
     from itertools import ifilterfalse
@@ -122,3 +123,26 @@ def clear():
 
 
 memoize = lru_cache(100)
+
+
+def compress_pruned(table):
+    """Compress table based on pruning mask.
+
+    Only the rows/cols in which all of the elements are masked need to be
+    pruned.
+    """
+    if not isinstance(table, np.ma.core.MaskedArray):
+        return table
+
+    if table.ndim == 0:
+        return table.data
+
+    if table.ndim == 1:
+        return np.ma.compressed(table)
+
+    row_inds = ~table.mask.all(axis=1)
+    col_inds = ~table.mask.all(axis=0)
+    table = table[row_inds, :][:, col_inds]
+    if table.dtype == float and table.mask.any():
+        table[table.mask] = np.nan
+    return table

tests/integration/fixtures/__init__.py

@@ -141,3 +141,4 @@ def _load(cube_file):
 GENDER_X_WEIGHT = _load('gender-x-weight.json')
 CAT_X_CAT_PRUNING_HS = _load('cat-x-cat-pruning-hs.json')
 CA_ITEMS_X_CA_CAT_X_CAT = _load('ca-items-x-ca-cat-x-cat.json')
+CAT_X_MR_X_CAT = _load('cat-x-mr-x-cat.json')