class NormalizationModule(SingleMethodModule):
attribute = 'normalizer'
title = 'Normalization'
toggle_enabled = True
enabled = settings.Setting(False)
methods = [
preprocess.PorterStemmer,
preprocess.SnowballStemmer,
preprocess.WordNetLemmatizer,
preprocess.UDPipeLemmatizer,
]
SNOWBALL = 1
UDPIPE = 3
snowball_language = settings.Setting('English')
udpipe_language = settings.Setting('English')
udpipe_tokenizer = settings.Setting(False)
def __init__(self, master):
super().__init__(master)
label = gui.label(self, self, 'Language:')
label.setAlignment(Qt.AlignRight)
self.method_layout.addWidget(label, self.SNOWBALL, 1)
snowball_box = widgets.ComboBox(self, 'snowball_language',
items=preprocess.SnowballStemmer.supported_languages)
snowball_box.currentIndexChanged.connect(self.change_language)
self.method_layout.addWidget(snowball_box, self.SNOWBALL, 2)
self.methods[self.SNOWBALL].language = self.snowball_language
self.udpipe_tokenizer_box = QCheckBox("UDPipe tokenizer", self,
checked=self.udpipe_tokenizer)
self.udpipe_tokenizer_box.stateChanged.connect(self.change_tokenizer)
self.method_layout.addWidget(self.udpipe_tokenizer_box, self.UDPIPE, 1)
self.udpipe_label = gui.label(self, self, 'Language:')
self.udpipe_label.setAlignment(Qt.AlignRight)
self.method_layout.addWidget(self.udpipe_label, self.UDPIPE, 2)
self.udpipe_models = UDPipeModels()
self.create_udpipe_box()
self.udpipe_online = self.udpipe_models.online
self.on_off_button.stateChanged.connect(self.check_udpipe_online)
self.check_udpipe_online()
self.methods[self.UDPIPE].language = self.udpipe_language
self.methods[self.UDPIPE].use_tokenizer = self.udpipe_tokenizer
def create_udpipe_box(self):
if not self.udpipe_models.supported_languages:
self.group.button(self.UDPIPE).setEnabled(False)
self.udpipe_tokenizer_box.setEnabled(False)
self.udpipe_label.setEnabled(False)
self.udpipe_box = widgets.ComboBox(self, 'udpipe_language', items=[''])
self.udpipe_box.setEnabled(False)
else:
self.group.button(self.UDPIPE).setEnabled(True)
self.udpipe_tokenizer_box.setEnabled(True)
self.udpipe_label.setEnabled(True)
self.udpipe_box = widgets.ComboBox(self, 'udpipe_language',
items=self.udpipe_models.supported_languages)
self.udpipe_box.currentIndexChanged.connect(self.change_language)
self.method_layout.addWidget(self.udpipe_box, self.UDPIPE, 3)
def check_udpipe_online(self):
current_state = self.udpipe_models.online
if self.udpipe_online != current_state:
self.create_udpipe_box()
self.udpipe_online = current_state
self.master.Warning.udpipe_offline.clear()
self.master.Warning.udpipe_offline_no_models.clear()
if not current_state and self.enabled:
if self.udpipe_models.supported_languages:
self.master.Warning.udpipe_offline()
else:
self.master.Warning.udpipe_offline_no_models()
def change_language(self):
if self.methods[self.SNOWBALL].language != self.snowball_language:
self.methods[self.SNOWBALL].language = self.snowball_language
if self.method_index == self.SNOWBALL:
self.change_signal.emit()
if self.methods[self.UDPIPE].language != self.udpipe_language:
self.methods[self.UDPIPE].language = self.udpipe_language
if self.method_index == self.UDPIPE:
self.change_signal.emit()
def change_tokenizer(self):
self.udpipe_tokenizer = self.udpipe_tokenizer_box.isChecked()
if self.methods[self.UDPIPE].use_tokenizer != self.udpipe_tokenizer:
self.methods[self.UDPIPE].use_tokenizer = self.udpipe_tokenizer
if self.method_index == self.UDPIPE:
self.change_signal.emit()
class NormalizeEditor(ScBaseEditor):
DEFAULT_GROUP_BY = False
DEFAULT_GROUP_VAR = None
DEFAULT_METHOD = Normalize.CPM
def __init__(self, parent=None, master=None, **kwargs):
super().__init__(parent, **kwargs)
self._group_var = self.DEFAULT_GROUP_VAR
self._master = master
self._master.input_data_changed.connect(self._set_model)
self.setLayout(QVBoxLayout())
form = QFormLayout()
cpm_b = QRadioButton("Counts per million", checked=True)
med_b = QRadioButton("Median")
self.group = QButtonGroup()
self.group.buttonClicked.connect(self._on_button_clicked)
for i, button in enumerate([cpm_b, med_b]):
index = index_to_enum(Normalize.Method, i).value
self.group.addButton(button, index - 1)
form.addRow(button)
self.group_by_check = QCheckBox("Cell Groups: ",
enabled=self.DEFAULT_GROUP_BY)
self.group_by_check.clicked.connect(self.edited)
self.group_by_combo = QComboBox(enabled=self.DEFAULT_GROUP_BY)
self.group_by_model = DomainModel(order=(DomainModel.METAS,
DomainModel.CLASSES),
valid_types=DiscreteVariable,
alphabetical=True)
self.group_by_combo.setModel(self.group_by_model)
self.group_by_combo.currentIndexChanged.connect(self.changed)
self.group_by_combo.activated.connect(self.edited)
form.addRow(self.group_by_check, self.group_by_combo)
self.layout().addLayout(form)
self._set_model()
def _set_model(self):
data = self._master.data
self.group_by_model.set_domain(data and data.domain)
enable = bool(self.group_by_model)
self.group_by_check.setChecked(False)
self.group_by_check.setEnabled(enable)
self.group_by_combo.setEnabled(enable)
if self.group_by_model:
self.group_by_combo.setCurrentIndex(0)
if self._group_var and self._group_var in data.domain:
index = self.group_by_model.indexOf(self._group_var)
self.group_by_combo.setCurrentIndex(index)
else:
self.group_by_combo.setCurrentText(None)
def _on_button_clicked(self):
self.changed.emit()
self.edited.emit()
def setParameters(self, params):
method = params.get("method", self.DEFAULT_METHOD)
index = enum_to_index(Normalize.Method, method)
self.group.buttons()[index].setChecked(True)
self._group_var = params.get("group_var", self.DEFAULT_GROUP_VAR)
group = bool(self._group_var and self.group_by_model)
if group:
index = self.group_by_model.indexOf(self._group_var)
self.group_by_combo.setCurrentIndex(index)
group_by = params.get("group_by", self.DEFAULT_GROUP_BY)
self.group_by_check.setChecked(group_by and group)
def parameters(self):
index = self.group_by_combo.currentIndex()
group_var = self.group_by_model[index] if index > -1 else None
group_by = self.group_by_check.isChecked()
method = index_to_enum(Normalize.Method, self.group.checkedId())
return {"group_var": group_var, "group_by": group_by, "method": method}
@staticmethod
def createinstance(params):
group_var = params.get("group_var")
group_by = params.get("group_by", NormalizeEditor.DEFAULT_GROUP_BY)
method = params.get("method", NormalizeEditor.DEFAULT_METHOD)
return NormalizeGroups(group_var, method) \
if group_by and group_var else NormalizeSamples(method)
def __repr__(self):
method = self.group.button(self.group.checkedId()).text()
index = self.group_by_combo.currentIndex()
group_var = self.group_by_model[index] if index > -1 else None
group_by = self.group_by_check.isChecked()
group_text = ", Grouped by: {}".format(group_var) if group_by else ""
return "Method: {}".format(method) + group_text
