Improve internal ordering of KeyValue set elements

Author: Belval

textractor/data/text_linearization_config.py

@@ -100,9 +100,7 @@ class TextLinearizationConfig:
 
     text_suffix: str = ""  #: Suffix for text layout elements
 
-    key_value_layout_prefix: str = (
-        os.linesep * 2
-    )  #: Prefix for key_value layout elements (not for individual key-value elements)
+    key_value_layout_prefix: str = ""  #: Prefix for key_value layout elements (not for individual key-value elements)
 
     key_value_layout_suffix: str = ""  #: Suffix for key_value layout elements (not for individual key-value elements)
 

textractor/parsers/response_parser.py

@@ -768,8 +768,6 @@ def _create_layout_objects(
                             [line_by_id[line_id] for line_id in relationship["Ids"] if line_id in line_by_id]
                         )
         else:
-            if block["BlockType"] in (LAYOUT_KEY_VALUE):
-                continue
             leaf_layouts.append(
                 Layout(
                     entity_id=block["Id"],

textractor/utils/text_utils.py

@@ -62,10 +62,7 @@ def should_group(line, group):
 
     current_group = [sorted_elements[0]]
     for element in sorted_elements[1:]:
-        if (
-            "Table" in element.__class__.__name__
-            or "KeyValue" in element.__class__.__name__
-        ):
+        if "Table" in element.__class__.__name__:
             if current_group:
                 grouped_elements.append(current_group)
             grouped_elements.append([element])
@@ -184,7 +181,7 @@ def part_of_same_paragraph(element1, element2, config=config):
                 result += config.same_paragraph_separator + text_element
                 words_output += words_element
             else:
-                result += config.layout_element_separator + text_element
+                result += config.same_layout_element_separator + text_element
                 words_output += words_element
 
             # FIXME: Seems like this would be mostly needed
@@ -193,6 +190,8 @@ def part_of_same_paragraph(element1, element2, config=config):
 
         if is_layout_table:
             result += config.table_row_separator
+        else:
+            result += config.same_layout_element_separator
 
         # We make a dummy line element with the bbox from the previous group
         prev_element = Line(