Merge pull request #2084 from VesnaT/janez-merge-data

OWMergeData: New GUI, heuristic to suggest matching attributes

Author: janezd

Orange/widgets/data/owmergedata.py

@@ -0,0 +1,7 @@
+gender	region	name
+d	d	s
+
+f	1	Ann
+m	2	Peter
+m	1	John
+m	3	Jose

Orange/widgets/data/tests/data-gender-region.tab

@@ -0,0 +1,5 @@
+id	state	continent
+d	s	d
+1	UK	Europe
+2	Russia	Europe
+3	Mexico	America

Orange/widgets/data/tests/data-regions.tab

@@ -867,8 +867,6 @@ def unique(seq):
             yield item
             seen.add(item)
 
-from Orange.widgets.data.owmergedata import group_table_indices
-
 
 def unique_non_nan(ar):
     # metas have sometimes object dtype, but values are numpy floats
@@ -1651,6 +1649,19 @@ def filter_vars(vars):
     return Orange.data.Table.from_table(domain, data)
 
 
+def group_table_indices(table, key_var):
+    """
+    Group table indices based on values of selected columns (`key_vars`).
+
+    Return a dictionary mapping all unique value combinations (keys)
+    into a list of indices in the table where they are present.
+    """
+    groups = defaultdict(list)
+    for i, inst in enumerate(table):
+        groups[str(inst[key_var])].append(i)
+    return groups
+
+
 def test():
     from Orange.evaluation import ShuffleSplit
 

Orange/widgets/data/tests/test_owmergedata.py

@@ -3,6 +3,7 @@
 
 import unittest
 import numpy as np
+from collections import defaultdict
 
 from Orange.data import (Table, Domain, StringVariable,
                          DiscreteVariable, ContinuousVariable)
@@ -13,7 +14,8 @@
                                                     table_concat,
                                                     varying_between,
                                                     drop_columns,
-                                                    OWVennDiagram)
+                                                    OWVennDiagram,
+                                                    group_table_indices)
 from Orange.tests import test_filename
 
 
@@ -174,3 +176,17 @@ def test_no_data(self):
         self.send_signal(self.signal_name, self.data[:100], 1)
         self.send_signal(self.signal_name, self.data[50:], 2)
         self.send_signal(self.signal_name, self.data[:0], 3)
+
+
+class GroupTableIndicesTest(unittest.TestCase):
+    def test_group_table_indices(self):
+        table = Table(test_filename("test9.tab"))
+        dd = defaultdict(list)
+        dd["1"] = [0, 1]
+        dd["huh"] = [2]
+        dd["hoy"] = [3]
+        dd["?"] = [4]
+        dd["2"] = [5]
+        dd["oh yeah"] = [6]
+        dd["3"] = [7]
+        self.assertEqual(dd, group_table_indices(table, "g"))