Add to db methods

Author: gabriel-piles

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "trainable-entity-extractor"
-version = "2025.02.14.07"
+version = "2025.03.10"
 description = "This tool is a trainable text/PDF to entity extractor"
 license = { file = "LICENSE" }
 authors = [{ name = "HURIDOCS" }]

src/multilingual_paragraph_extractor/domain/ParagraphFeatures.py

@@ -256,3 +256,15 @@ def get_distance(self, next_paragraph: "ParagraphFeatures") -> float:
             return 0
 
         return (next_paragraph.first_token_bounding_box.top - self.last_token_bounding_box.bottom) / self.page_height
+
+    def to_db(self):
+        return ParagraphFeatures(
+            index=self.index,
+            page_height=self.page_height,
+            page_width=self.page_width,
+            paragraph_type=self.paragraph_type,
+            page_number=self.page_number,
+            bounding_box=self.bounding_box,
+            text_cleaned=self.text_cleaned,
+            original_text=self.original_text,
+        )

src/multilingual_paragraph_extractor/domain/ParagraphsFromLanguage.py

@@ -432,3 +432,10 @@ def remove_big_no_text_paragraphs(self):
             fixed_paragraphs.append(paragraph)
 
         self.paragraphs = fixed_paragraphs
+
+    def to_db(self):
+        return ParagraphsFromLanguage(
+            language=self.language,
+            paragraphs=[x.to_db() for x in self.paragraphs],
+            is_main_language=self.is_main_language,
+        )