Merge pull request #924 from PrimozGodec/langdedect-create-corpus

[ENH] Create Corpus - add language to corpus
biolab · Jan 11, 2023 · 971969e · 971969e
2 parents 25952ea + 83e0446
commit 971969e
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 2 deletions.
diff --git a/orangecontrib/text/widgets/owcreatecorpus.py b/orangecontrib/text/widgets/owcreatecorpus.py
@@ -16,6 +16,7 @@
 from orangewidget.settings import Setting
 
 from orangecontrib.text import Corpus
+from orangecontrib.text.language import LANG2ISO, DEFAULT_LANGUAGE, LanguageModel
 
 
 class EditorsVerticalScrollArea(gui.VerticalScrollArea):
@@ -76,13 +77,26 @@ class Outputs:
 
     want_main_area = False
 
+    language: str = Setting(DEFAULT_LANGUAGE)
     texts: List[Tuple[str, str]] = Setting([("", "")] * 3)
     auto_commit: bool = Setting(True)
 
     def __init__(self):
         super().__init__()
         self.editors = []
 
+        gui.comboBox(
+            self.controlArea,
+            self,
+            "language",
+            model=LanguageModel(),
+            box="Language",
+            orientation=Qt.Horizontal,
+            callback=self.commit.deferred,
+            sendSelectedValue=True,
+            searchable=True,
+        )
+
         scroll_area = EditorsVerticalScrollArea()
         self.editor_vbox = gui.vBox(self.controlArea, spacing=0)
         self.editor_vbox.layout().setSpacing(10)
@@ -142,12 +156,13 @@ def commit(self):
             np.empty((len(self.texts), 0)),
             metas=np.array(self.texts),
             text_features=[doc_var],
+            language=LANG2ISO[self.language],
         )
         corpus.set_title_variable(title_var)
         self.Outputs.corpus.send(corpus)
 
     def sizeHint(self) -> QSize:
-        return QSize(600, 600)
+        return QSize(600, 650)
 
 
 if __name__ == "__main__":

diff --git a/orangecontrib/text/widgets/tests/test_owcreatecorpus.py b/orangecontrib/text/widgets/tests/test_owcreatecorpus.py
@@ -3,7 +3,8 @@
 import numpy as np
 from Orange.data import StringVariable
 from Orange.widgets.tests.base import WidgetTest
-from AnyQt.QtWidgets import QPushButton
+from AnyQt.QtWidgets import QPushButton, QComboBox
+from orangewidget.tests.utils import simulate
 
 from orangecontrib.text.widgets.owcreatecorpus import OWCreateCorpus
 
@@ -197,6 +198,15 @@ def test_output(self):
         self.assertListEqual(["Test 3"], corpus.documents)
         np.testing.assert_array_equal([["Document 3", "Test 3"]], corpus.metas)
 
+    def test_language(self):
+        corpus = self.get_output(self.widget.Outputs.corpus)
+        self.assertEqual("en", corpus.language)
+
+        combo = self.widget.controlArea.findChild(QComboBox)
+        simulate.combobox_activate_index(combo, 2)
+        corpus = self.get_output(self.widget.Outputs.corpus)
+        self.assertEqual("am", corpus.language)
+
 
 if __name__ == "__main__":
     unittest.main()