Merge bf75278 into 736470d

readabilipy/simplifiers/html.py

@@ -196,7 +196,11 @@ def insert_paragraph_breaks(soup):
                     new_p_element = soup.new_tag("p")
                     new_p_element.string = text_fragment
                     parent_element.insert_after(new_p_element)
-                parent_element.string.replace_with(text_fragments[0])
+                # Replace the parent string if it exists or add one if not
+                if parent_element.string:
+                    parent_element.string.replace_with(text_fragments[0])
+                else:
+                    parent_element.string = text_fragments[0]
             # Otherwise we want to simply include all the text fragments as independent NavigableStrings (that will be wrapped later)
             else:
                 # Iterate in reverse order as we are repeatedly adding new elements directly after the original one

tests/test_weird_html.py

@@ -51,3 +51,17 @@ def test_tags_inside_words():
         """a<a href="http://example.com">i</a>sle""",
         "<div><p>aisle</p></div>"
     )
+
+
+# Test splitting for unclosed tags inside paragraphs
+def test_paragraph_splitting_with_unclosed_tags():
+    """Ensure that paragraphs with unclosed tags inside them split correctly."""
+    check_exact_html_output(
+        """
+        <p>
+            <meta charset="utf-8">First paragraph.
+            <br><br>
+            Second paragraph.
+        </p>""",
+        "<div><p>First paragraph.</p><p>Second paragraph.</p></div>"
+    )