Added to_text to ontology element

Author: MaksOpp

unstructured/documents/ontology.py

@@ -89,6 +89,9 @@ def to_html(self, add_children=True) -> str:
 
         return result_html
 
+    def to_text(self, add_children=True) -> str:
+        return BeautifulSoup(self.to_html(), "html.parser").get_text().strip()
+
     def _construct_attribute_string(self, attributes: dict) -> str:
         return " ".join(
             f'{key}="{value}"' if value else f"{key}" for key, value in attributes.items()
@@ -481,6 +484,9 @@ def to_html(self, add_children=True) -> str:
 
         return f"<{self.html_tag_name} {combined_attr_str} />"
 
+    def to_text(self, add_children=True) -> str:
+        return super().to_text() + self.text
+
 
 class Checkbox(OntologyElement):
     description: str = Field("A small box that can be checked or unchecked", frozen=True)

unstructured/partition/html/transformations.py

@@ -96,10 +96,8 @@ def ontology_to_unstructured_elements(
         ]
         element_class = TYPE_TO_TEXT_ELEMENT_MAP[unstructured_element_class_name]
         html_code_of_ontology_element = ontology_element.to_html()
-        element_text = (
-            BeautifulSoup(html_code_of_ontology_element, "html.parser").get_text().strip()
-        )
-        # TODO value attribute from form input should be added to the text
+        element_text = ontology_element.to_text()
+
         unstructured_element = element_class(
             text=element_text,
             element_id=ontology_element.id,