|
| 1 | +import numpy as np |
| 2 | + |
| 3 | +from AnyQt.QtWidgets import QGridLayout, QLabel, QLineEdit, QSizePolicy |
| 4 | +from AnyQt.QtCore import QSize, Qt |
| 5 | + |
| 6 | +from Orange.data import StringVariable, DiscreteVariable, Domain |
| 7 | +from Orange.data.table import Table |
| 8 | +from Orange.preprocess.transformation import Transformation |
| 9 | +from Orange.widgets import gui, widget |
| 10 | +from Orange.widgets.settings import DomainContextHandler, ContextSetting |
| 11 | +from Orange.widgets.utils.itemmodels import DomainModel |
| 12 | +from Orange.widgets.widget import Msg |
| 13 | + |
| 14 | + |
| 15 | +class ValueFromSubstring(Transformation): |
| 16 | + def __init__(self, variable, values): |
| 17 | + super().__init__(variable) |
| 18 | + self.values = values |
| 19 | + |
| 20 | + def transform(self, c): |
| 21 | + c = c.flatten() |
| 22 | + res = np.empty(len(c)) |
| 23 | + for i, val in enumerate(c): |
| 24 | + for res[i], pattern in enumerate(self.values): |
| 25 | + if pattern in val: |
| 26 | + break |
| 27 | + else: |
| 28 | + res[i] = np.nan |
| 29 | + return res |
| 30 | + |
| 31 | + |
| 32 | +class OWCreateClass(widget.OWWidget): |
| 33 | + name = "Create class" |
| 34 | + description = "Create class attribute from a string attribute" |
| 35 | + icon = "icons/CreateClass.svg" |
| 36 | + category = "Data" |
| 37 | + keywords = ["data"] |
| 38 | + |
| 39 | + inputs = [("Data", Table, "set_data")] |
| 40 | + outputs = [("Data", Table)] |
| 41 | + |
| 42 | + want_main_area = False |
| 43 | + |
| 44 | + settingsHandler = DomainContextHandler() |
| 45 | + attribute = ContextSetting(None) |
| 46 | + |
| 47 | + class Warning(widget.OWWidget.Warning): |
| 48 | + no_string_variables = Msg("No string variables.") |
| 49 | + |
| 50 | + def __init__(self): |
| 51 | + super().__init__() |
| 52 | + self.data = None |
| 53 | + self.rules = [["C1", ""], ["C2", ""]] |
| 54 | + self.line_edits = [] |
| 55 | + self.remove_buttons = [] |
| 56 | + self.counts = [] |
| 57 | + self.setSizePolicy(QSizePolicy.Preferred, QSizePolicy.Maximum) |
| 58 | + |
| 59 | + box = gui.hBox(self.controlArea) |
| 60 | + gui.widgetLabel(box, "Class from column: ", addSpace=12) |
| 61 | + gui.comboBox( |
| 62 | + box, self, "attribute", callback=self.apply, |
| 63 | + model=DomainModel(valid_types=StringVariable), |
| 64 | + sizePolicy=(QSizePolicy.MinimumExpanding, QSizePolicy.Fixed)) |
| 65 | + |
| 66 | + self.rules_box = rules_box = QGridLayout() |
| 67 | + self.controlArea.layout().addLayout(self.rules_box) |
| 68 | + self.add_button = gui.button(None, self, "+", flat=True, |
| 69 | + callback=self.add_row, |
| 70 | + minimumSize=QSize(12, 20)) |
| 71 | + self.rules_box.setColumnMinimumWidth(1, 80) |
| 72 | + self.rules_box.setColumnMinimumWidth(0, 10) |
| 73 | + self.rules_box.setColumnStretch(0, 1) |
| 74 | + self.rules_box.setColumnStretch(1, 1) |
| 75 | + self.rules_box.setColumnStretch(2, 100) |
| 76 | + rules_box.addWidget(QLabel("Name"), 0, 1) |
| 77 | + rules_box.addWidget(QLabel("Pattern"), 0, 2) |
| 78 | + rules_box.addWidget(QLabel("#Instances"), 0, 3, 1, 2) |
| 79 | + self.update_rules() |
| 80 | + |
| 81 | + box = gui.hBox(self.controlArea) |
| 82 | + gui.rubber(box) |
| 83 | + gui.button(box, self, "Apply", autoDefault=False, callback=self.apply) |
| 84 | + |
| 85 | + def retrieveSpecificSettings(self): |
| 86 | + self.rules = list(self.current_context.rules) |
| 87 | + self.update_rules() |
| 88 | + |
| 89 | + def storeSpecificSettings(self): |
| 90 | + self.current_context.rules = tuple(self.rules) |
| 91 | + |
| 92 | + def rules_to_edits(self): |
| 93 | + for editr, textr in zip(self.line_edits, self.rules): |
| 94 | + for edit, text in zip(editr, textr): |
| 95 | + edit.setText(text) |
| 96 | + |
| 97 | + def edits_to_rule(self): |
| 98 | + self.rules = [[edit.text() for edit in row] for row in self.line_edits] |
| 99 | + |
| 100 | + def set_data(self, data): |
| 101 | + # We don't reset the rules before opening new context, thus keeping the |
| 102 | + # existing rules. This makes sense within a lifetime of a widget, and |
| 103 | + # it doesn't extend past it. |
| 104 | + self.closeContext() |
| 105 | + self.data = data |
| 106 | + model = self.controls.attribute.model() |
| 107 | + model.set_domain(data and data.domain) |
| 108 | + self.Warning.no_string_variables(shown=data is not None and not model) |
| 109 | + if not model: |
| 110 | + self.attribute = None |
| 111 | + self.send("Data", None) |
| 112 | + return |
| 113 | + self.attribute = model[0] |
| 114 | + self.openContext(data) |
| 115 | + self.update_rules() |
| 116 | + |
| 117 | + def update_rules(self): |
| 118 | + self.adjust_n_rule_rows() |
| 119 | + self.rules_to_edits() |
| 120 | + self.update_counts() |
| 121 | + |
| 122 | + def adjust_n_rule_rows(self): |
| 123 | + def _add_line(): |
| 124 | + self.line_edits.append([]) |
| 125 | + n_lines = len(self.line_edits) |
| 126 | + for coli in range(1, 3): |
| 127 | + edit = QLineEdit() |
| 128 | + self.line_edits[-1].append(edit) |
| 129 | + self.rules_box.addWidget(edit, n_lines, coli) |
| 130 | + edit.textChanged.connect(self.sync_edit) |
| 131 | + button = gui.button( |
| 132 | + None, self, label='×', flat=True, height=20, |
| 133 | + styleSheet='* {font-size: 16pt; color: silver}' |
| 134 | + '*:hover {color: black}', |
| 135 | + callback=self.remove_row) |
| 136 | + button.setMinimumSize(QSize(12, 20)) |
| 137 | + self.remove_buttons.append(button) |
| 138 | + self.rules_box.addWidget(button, n_lines, 0) |
| 139 | + self.counts.append([]) |
| 140 | + for coli in range(3, 5): |
| 141 | + label = QLabel(alignment=Qt.AlignRight if coli == 3 |
| 142 | + else Qt.AlignLeft) |
| 143 | + self.counts[-1].append(label) |
| 144 | + self.rules_box.addWidget(label, n_lines, coli) |
| 145 | + |
| 146 | + def _remove_line(): |
| 147 | + for edit in self.line_edits.pop(): |
| 148 | + edit.deleteLater() |
| 149 | + self.remove_buttons.pop().deleteLater() |
| 150 | + for label in self.counts.pop(): |
| 151 | + label.deleteLater() |
| 152 | + |
| 153 | + def _fix_tab_order(): |
| 154 | + prev = None |
| 155 | + for row, rule in zip(self.line_edits, self.rules): |
| 156 | + for col_idx, edit in enumerate(row): |
| 157 | + edit.row, edit.col_idx = rule, col_idx |
| 158 | + if prev is not None: |
| 159 | + self.setTabOrder(prev, edit) |
| 160 | + prev = edit |
| 161 | + |
| 162 | + n = len(self.rules) |
| 163 | + while n > len(self.line_edits): |
| 164 | + _add_line() |
| 165 | + while len(self.line_edits) > n: |
| 166 | + _remove_line() |
| 167 | + self.rules_box.addWidget(self.add_button, len(self.rules) + 1, 0) |
| 168 | + _fix_tab_order() |
| 169 | + |
| 170 | + def add_row(self): |
| 171 | + self.rules.append(["", ""]) |
| 172 | + self.adjust_n_rule_rows() |
| 173 | + |
| 174 | + def remove_row(self): |
| 175 | + remove_idx = self.remove_buttons.index(self.sender()) |
| 176 | + del self.rules[remove_idx] |
| 177 | + self.update_rules() |
| 178 | + |
| 179 | + def sync_edit(self, text): |
| 180 | + edit = self.sender() |
| 181 | + edit.row[edit.col_idx] = text |
| 182 | + self.update_counts() |
| 183 | + |
| 184 | + def update_counts(self): |
| 185 | + if not self.attribute: |
| 186 | + for label_row in self.counts: |
| 187 | + for label in label_row: |
| 188 | + label.setText("") |
| 189 | + return |
| 190 | + data = [str(x) for x in self.data[:, self.attribute]] |
| 191 | + remaining = data.copy() |
| 192 | + for (lab_match, lab_all), (_, pattern) in zip(self.counts, self.rules): |
| 193 | + new_remaining = [x for x in remaining if pattern not in x] |
| 194 | + n_matched = len(remaining) - len(new_remaining) |
| 195 | + n_match_in_all = sum(pattern in x for x in data) |
| 196 | + lab_match.setText("{}".format(n_matched) if remaining else "") |
| 197 | + lab_all.setText("({})".format(n_match_in_all) |
| 198 | + if remaining and n_match_in_all != n_matched |
| 199 | + else "") |
| 200 | + remaining = new_remaining |
| 201 | + |
| 202 | + def apply(self): |
| 203 | + if not self.data: |
| 204 | + return |
| 205 | + domain = self.data.domain |
| 206 | + names, patterns = \ |
| 207 | + zip(*((name.strip(), pattern) |
| 208 | + for name, pattern in self.rules if name.strip())) |
| 209 | + new_class = DiscreteVariable( |
| 210 | + "class", names, |
| 211 | + compute_value=ValueFromSubstring(self.attribute, patterns)) |
| 212 | + new_domain = Domain(domain.attributes, new_class, |
| 213 | + domain.metas + domain.class_vars) |
| 214 | + new_data = Table(new_domain, self.data) |
| 215 | + self.send("Data", new_data) |
| 216 | + |
| 217 | + |
| 218 | +def main(): |
| 219 | + import sys |
| 220 | + from AnyQt.QtWidgets import QApplication |
| 221 | + |
| 222 | + a = QApplication(sys.argv) |
| 223 | + table = Table("zoo") |
| 224 | + ow = OWCreateClass() |
| 225 | + ow.show() |
| 226 | + ow.set_data(table) |
| 227 | + a.exec() |
| 228 | + ow.saveSettings() |
| 229 | + |
| 230 | +if __name__ == "__main__": |
| 231 | + main() |
0 commit comments