Accept portuguese for gliner

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "trainable-entity-extractor"
-version = "2024.12.19.4"
+version = "2024.12.19.5"
 description = "This tool is a trainable text/PDF to entity extractor"
 license = { file = "LICENSE" }
 authors = [{ name = "HURIDOCS" }]

src/trainable_entity_extractor/data/TrainingSample.py

@@ -41,7 +41,7 @@ def from_pdf(pdf_path: str | Path, label_text: str, language_iso: str = "en"):
         segmentation_data = SegmentationData(
             page_width=pdf_features.pages[0].page_width if pdf_features.pages else 0,
             page_height=pdf_features.pages[0].page_height if pdf_features.pages else 0,
-            xml_segments_boxes=[],
+            xml_segments_boxes=[SegmentBox(left=0, top=0, width=343434342, height=300, page_number=1)],
             label_segments_boxes=[],
         )
         pdf_data.set_segments_from_segmentation_data(segmentation_data=segmentation_data)

src/trainable_entity_extractor/extractors/text_to_text_extractor/methods/GlinerDateParserMethod.py

@@ -10,8 +10,7 @@ class GlinerDateParserMethod(ToTextExtractorMethod):
 
     @staticmethod
     def get_alphanumeric_text_with_spaces(text):
-        alphanumeric_pattern = re.compile(r"[A-Za-z0-9 ]+")
-        return "".join(alphanumeric_pattern.findall(text))
+        return "".join([letter for letter in text if letter.isalnum() or letter.isspace()])
 
     @staticmethod
     def get_date(tags_texts: list[str]):
@@ -36,3 +35,7 @@ def predict(self, predictions_samples: list[PredictionSample]) -> list[str]:
         ]
         predictions = [date.strftime("%Y-%m-%d") if date else "" for date in predictions_dates]
         return predictions
+
+
+if __name__ == "__main__":
+    print(GlinerDateParserMethod.get_alphanumeric_text_with_spaces("21 DE MARÇO DE 2023"))

src/trainable_entity_extractor/extractors/text_to_text_extractor/methods/test/test_gliner_date_method.py

@@ -25,6 +25,12 @@ def test_predict_special_character(self):
         predictions = gliner_method.predict([PredictionSample.from_text("SENTENÇA DE 1° DE JULHO DE 2009")])
         self.assertEqual(["2009-07-01"], predictions)
 
+    def test_predict_portuguese(self):
+        gliner_method = GlinerDateParserMethod(extraction_identifier)
+
+        predictions = gliner_method.predict([PredictionSample.from_text("SENTENÇA DE 1° DE MARÇO DE 2010")])
+        self.assertEqual(["2010-03-01"], predictions)
+
     def test_predict_multiple_dates_spanish(self):
         text = "Informe no. 52/16 caso 12.521 fondo Maria Laura órdenes guerra y otros Chile 30 de noviembre de 2016"
         sample = TrainingSample(