Merge pull request #2908 from BlazZupan/select-rows-annotated-data

[ENH] Select Rows: Add annotated data output

Author: lanzagar

Orange/widgets/data/owselectrows.py

@@ -1,8 +1,9 @@
 import enum
-
 from collections import OrderedDict
 from itertools import chain
 
+import numpy as np
+
 from AnyQt.QtWidgets import (
     QWidget, QTableWidget, QHeaderView, QComboBox, QLineEdit, QToolButton,
     QMessageBox, QMenu, QListView, QGridLayout, QPushButton, QSizePolicy,
@@ -26,6 +27,8 @@
 from Orange.widgets.utils import vartype
 from Orange.canvas import report
 from Orange.widgets.widget import Msg
+from Orange.widgets.utils.annotated_data import (create_annotated_table,
+                                                 ANNOTATED_DATA_SIGNAL_NAME)
 
 
 class SelectRowsContextHandler(DomainContextHandler):
@@ -79,6 +82,7 @@ class Inputs:
     class Outputs:
         matching_data = Output("Matching Data", Table, default=True)
         unmatched_data = Output("Unmatched Data", Table)
+        annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Table)
 
     want_main_area = False
 
@@ -465,6 +469,8 @@ def _values_to_floats(self, attr, values):
     def commit(self):
         matching_output = self.data
         non_matching_output = None
+        annotated_output = None
+
         self.Error.clear()
         if self.data:
             domain = self.data.domain
@@ -512,6 +518,9 @@ def commit(self):
                 self.filters.negate = True
                 non_matching_output = self.filters(self.data)
 
+                row_sel = np.in1d(self.data.ids, matching_output.ids)
+                annotated_output = create_annotated_table(self.data, row_sel)
+
             # if hasattr(self.data, "name"):
             #     matching_output.name = self.data.name
             #     non_matching_output.name = self.data.name
@@ -529,14 +538,18 @@ def commit(self):
 
                 matching_output = remover(matching_output)
                 non_matching_output = remover(non_matching_output)
+                annotated_output = remover(annotated_output)
 
         if matching_output is not None and not len(matching_output):
             matching_output = None
         if non_matching_output is not None and not len(non_matching_output):
             non_matching_output = None
+        if annotated_output is not None and not len(annotated_output):
+            annotated_output = None
 
         self.Outputs.matching_data.send(matching_output)
         self.Outputs.unmatched_data.send(non_matching_output)
+        self.Outputs.annotated_data.send(annotated_output)
 
         self.match_desc = report.describe_data_brief(matching_output)
         self.nonmatch_desc = report.describe_data_brief(non_matching_output)

Orange/widgets/data/tests/test_owselectrows.py

@@ -2,7 +2,9 @@
 # pylint: disable=missing-docstring
 from AnyQt.QtCore import QLocale, Qt
 from AnyQt.QtTest import QTest
-from AnyQt.QtWidgets import QLineEdit
+from AnyQt.QtWidgets import QLineEdit, QComboBox
+
+import numpy as np
 
 from Orange.data import (
     Table, ContinuousVariable, StringVariable, DiscreteVariable)
@@ -12,6 +14,7 @@
 
 from Orange.data.filter import FilterContinuous, FilterString
 from Orange.widgets.tests.utils import simulate, override_locale
+from Orange.widgets.utils.annotated_data import ANNOTATED_DATA_FEATURE_NAME
 
 CFValues = {
     FilterContinuous.Equal: ["5.4"],
@@ -115,32 +118,6 @@ def test_continuous_filter_with_sl_SI_locale(self):
         self.enterFilter(iris.domain[2], "is below", "5.2")
         self.assertEqual(self.widget.conditions[0][2], ("52",))
 
-    def enterFilter(self, variable, filter, value=None, value2=None):
-        row = self.widget.cond_list.model().rowCount()
-        self.widget.add_button.click()
-
-        var_combo = self.widget.cond_list.cellWidget(row, 0)
-        simulate.combobox_activate_item(var_combo, variable.name, delay=0)
-
-        oper_combo = self.widget.cond_list.cellWidget(row, 1)
-        simulate.combobox_activate_item(oper_combo, filter, delay=0)
-
-        value_inputs = self._get_value_line_edits(row)
-        for i, value in enumerate([value, value2]):
-            if value is None:
-                continue
-            QTest.mouseClick(value_inputs[i], Qt.LeftButton)
-            QTest.keyClicks(value_inputs[i], value, delay=0)
-            QTest.keyClick(value_inputs[i], Qt.Key_Enter)
-
-    def _get_value_line_edits(self, row):
-        value_inputs = self.widget.cond_list.cellWidget(row, 2)
-        if value_inputs:
-            value_inputs = [w for w in value_inputs.children()
-                            if isinstance(w, QLineEdit)]
-        return value_inputs
-
-
     @override_locale(QLocale.Slovenian)
     def test_stores_settings_in_invariant_locale(self):
         iris = Table("iris")[:5]
@@ -156,8 +133,6 @@ def test_stores_settings_in_invariant_locale(self):
         saved_condition = context.values["conditions"][0]
         self.assertEqual(saved_condition[2][0], 5.2)
 
-
-
     @override_locale(QLocale.C)
     def test_restores_continuous_filter_in_c_locale(self):
         iris = Table("iris")[:5]
@@ -226,23 +201,15 @@ def test_is_defined_on_continuous_variable(self):
         data = Table(datasets.path("testing_dataset_cls"))
         self.send_signal(self.widget.Inputs.data, data)
 
-
         self.enterFilter(data.domain["c2"], "is defined")
         self.assertFalse(self.widget.Error.parsing_error.is_shown())
         self.assertEqual(len(self.get_output("Matching Data")), 3)
         self.assertEqual(len(self.get_output("Unmatched Data")), 1)
+        self.assertEqual(len(self.get_output("Data")), len(data))
 
         # Test saving of settings
         self.widget.settingsHandler.pack_data(self.widget)
 
-    def widget_with_context(self, domain, conditions):
-        ch = SelectRowsContextHandler()
-        context = ch.new_context(domain, *ch.encode_domain(domain))
-        context.values = dict(conditions=conditions)
-        settings = dict(context_settings=[context])
-
-        return self.create_widget(OWSelectRows, settings)
-
     def test_output_filter(self):
         """
         None on output when there is no data.
@@ -255,7 +222,66 @@ def test_output_filter(self):
         self.enterFilter(data.domain[0], "is below", "-1")
         self.assertIsNone(self.get_output("Matching Data"))
         self.assertEqual(len(self.get_output("Unmatched Data")), len_data)
+        self.assertEqual(len(self.get_output("Data")), len_data)
         self.widget.remove_all_button.click()
         self.enterFilter(data.domain[0], "is below", "10")
         self.assertIsNone(self.get_output("Unmatched Data"))
         self.assertEqual(len(self.get_output("Matching Data")), len_data)
+        self.assertEqual(len(self.get_output("Data")), len_data)
+
+    def test_annotated_data(self):
+        iris = Table("iris")
+        self.send_signal(self.widget.Inputs.data, iris)
+
+        self.enterFilter(iris.domain["iris"], "is", "Iris-setosa")
+
+        annotated = self.get_output(self.widget.Outputs.annotated_data)
+        self.assertEqual(len(annotated), 150)
+        annotations = annotated.get_column_view(ANNOTATED_DATA_FEATURE_NAME)[0]
+        np.testing.assert_equal(annotations[:50], True)
+        np.testing.assert_equal(annotations[50:], False)
+
+    def widget_with_context(self, domain, conditions):
+        ch = SelectRowsContextHandler()
+        context = ch.new_context(domain, *ch.encode_domain(domain))
+        context.values = dict(conditions=conditions)
+        settings = dict(context_settings=[context])
+
+        return self.create_widget(OWSelectRows, settings)
+
+    def enterFilter(self, variable, filter, value1=None, value2=None):
+        row = self.widget.cond_list.model().rowCount()
+        self.widget.add_button.click()
+
+        var_combo = self.widget.cond_list.cellWidget(row, 0)
+        simulate.combobox_activate_item(var_combo, variable.name, delay=0)
+
+        oper_combo = self.widget.cond_list.cellWidget(row, 1)
+        simulate.combobox_activate_item(oper_combo, filter, delay=0)
+
+        value_inputs = self.__get_value_widgets(row)
+        for i, value in enumerate([value1, value2]):
+            if value is None:
+                continue
+            self.__set_value(value_inputs[i], value)
+
+    def __get_value_widgets(self, row):
+        value_inputs = self.widget.cond_list.cellWidget(row, 2)
+        if value_inputs:
+            if isinstance(value_inputs, QComboBox):
+                value_inputs = [value_inputs]
+            else:
+                value_inputs = [
+                    w for w in value_inputs.children()
+                    if isinstance(w, QLineEdit)]
+        return value_inputs
+
+    def __set_value(self, widget, value):
+        if isinstance(widget, QLineEdit):
+            QTest.mouseClick(widget, Qt.LeftButton)
+            QTest.keyClicks(widget, value, delay=0)
+            QTest.keyClick(widget, Qt.Key_Enter)
+        elif isinstance(widget, QComboBox):
+            simulate.combobox_activate_item(widget, value)
+        else:
+            raise ValueError("Unsupported widget {}".format(widget))

doc/visual-programming/source/widgets/data/selectrows.rst

@@ -12,6 +12,8 @@ Outputs
         instances that match the conditions
     Non-Matching Data
         instances that do not match the conditions
+    Data
+        data with an additional column showing whether a instance is selected
 
 
 This widget selects a subset from an input dataset, based on user-defined