fix: parse unexpected elements in lists as text

enex2notion/note_parser_elements.py

@@ -1,7 +1,8 @@
 import logging
 
-from bs4 import Tag
+from bs4 import NavigableString, Tag
 
+from enex2notion.notion_blocks import NotionTextBlock, TextProp
 from enex2notion.notion_blocks_header import (
     NotionHeaderBlock,
     NotionSubheaderBlock,
@@ -36,11 +37,27 @@ def parse_list(element: Tag):
             nodes[-1].children.extend(parse_list(subelement))
 
         else:
-            logger.warning(f"Unexpected tag inside list: {subelement.name}")
+            li_odd_item = _parse_odd_item(subelement)
+            if li_odd_item is None:
+                continue
+
+            nodes.append(li_odd_item)
 
     return nodes
 
 
+def _parse_odd_item(element: Tag):
+    if isinstance(element, NavigableString):
+        if not element.text.strip():
+            return None
+
+        logger.warning("Non-empty string element inside list")
+        return NotionTextBlock(text_prop=TextProp(text=element.text.strip()))
+
+    logger.warning(f"Unexpected tag inside list: {element.name}, parsing as text")
+    return NotionTextBlock(text_prop=extract_string(element))
+
+
 def _parse_list_item(list_item, is_ul):
     li_text = extract_string(list_item)
 

tests/test_note_parser.py

@@ -119,6 +119,32 @@ def test_list_ul_nested():
     assert parse_note_dom(test_note) == expected
 
 
+def test_list_ul_strings_inside(caplog):
+    test_note = parse_html("<ul><li><div>test1</div></li>test2</ul>")
+
+    with caplog.at_level(logging.WARNING):
+        result_blocks = parse_note_dom(test_note)
+
+    assert "Non-empty string element inside list" in caplog.records[0].message
+    assert result_blocks == [
+        NotionBulletedListBlock(text_prop=TextProp("test1")),
+        NotionTextBlock(text_prop=TextProp("test2")),
+    ]
+
+
+def test_list_ul_unexpected_inside(caplog):
+    test_note = parse_html("<ul><li><div>test1</div></li><span>test2</span></ul>")
+
+    with caplog.at_level(logging.WARNING):
+        result_blocks = parse_note_dom(test_note)
+
+    assert "Unexpected tag inside list" in caplog.records[0].message
+    assert result_blocks == [
+        NotionBulletedListBlock(text_prop=TextProp("test1")),
+        NotionTextBlock(text_prop=TextProp("test2")),
+    ]
+
+
 def test_table():
     test_note = parse_html(
         "<table>"