Ignore KV elements in LAYOUT_LIST

Author: Belval

textractor/entities/layout.py

@@ -139,7 +139,7 @@ def get_text_and_words(
             final_text = add_id_to_html_tag(config.list_layout_prefix, self.id, config)
             final_words = []
             for i, child in enumerate(
-                sorted(self.children, key=lambda x: x.reading_order)
+                sorted(filter(lambda c: isinstance(c, Layout), self.children), key=lambda x: x.reading_order)
             ):
                 child_text, child_words = child.get_text_and_words(config)
                 child_prefix = add_id_to_html_tag(config.list_element_prefix, child.id, config)

textractor/parsers/response_parser.py

@@ -1363,15 +1363,15 @@ def parse_document_api_response(response: dict) -> Document:
         # Using the kv_added returned by _create_table_objects, we try to match the remaining KVs
         # to existing layout elements.
         for layout in sorted(page.layouts, key=lambda x: x.bbox.y):
-            if layout.layout_type == LAYOUT_ENTITY:
+            if layout.layout_type in LAYOUT_ENTITY:
                 continue
             for kv in sorted(key_values, key=lambda x: x.bbox.y):
                 if (
                     layout.bbox.get_intersection(kv.bbox).area > THRESHOLD * kv.bbox.area
                     and kv.id not in kv_added
                 ):
                     # Ignore if the KV is already overlapping with a table
-                    if any([w.cell_id for w in kv.words]):
+                    if any([w.cell_id for w in kv.words]) or layout.layout_type == LAYOUT_LIST:
                         kv_added.add(kv.id)
                         continue
                     # Removing the duplicate words