From ffa23deaab855b4b3ffbf1d5cf4c5d5460968b28 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Fri, 8 Apr 2022 17:16:16 +0100
Subject: [PATCH] Fix for empty paragraphs during long string check

---
 content.py |  2 ++
 tests.py   | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 39 insertions(+)

diff --git a/content.py b/content.py
index 34dd1bd48..5bd414974 100644
--- a/content.py
+++ b/content.py
@@ -745,6 +745,7 @@ def remove_long_words(content: str, max_word_length: int,
                 else:
                     content = content[:max_word_length]
                 return content
+    content = content.replace('<p></p>', '<p> </p>')
     words = content.split(' ')
     if not long_words_list:
         long_words_list = []
@@ -806,6 +807,7 @@ def remove_long_words(content: str, max_word_length: int,
     if content.startswith('<p>'):
         if not content.endswith('</p>'):
             content = content.strip() + '</p>'
+    content = content.replace('<p> </p>', '<p></p>')
     return content
 
 
diff --git a/tests.py b/tests.py
index fd0fed64c..ec64f752e 100644
--- a/tests.py
+++ b/tests.py
@@ -2845,6 +2845,43 @@ def _test_create_person_account(base_dir: str):
     assert test_post_json['object']['content']
     assert '(yawn)' in test_post_json['object']['content']
 
+    content = \
+        'I would regard fediverse as being things based on ActivityPub ' + \
+        'or OStatus. i.e. things whose protocol lineage can be traced ' + \
+        'back to identica/statusnet/pumpio.\n' + \
+        '\nFediverse is a vague term though ' + \
+        'and I know some people regard Matrix and Diaspora as being ' + \
+        'fediverse. If fediverse just means any federated system ' + \
+        'then email would be somequitelongword.\nAnotherlongwordhere sentence.'
+    test_post_json = \
+        create_public_post(base_dir, nickname, domain, port, http_prefix,
+                           content, followers_only, save_to_file,
+                           client_to_server,
+                           comments_enabled, attach_image_filename, media_type,
+                           'Not suitable for Vogons', 'London, England',
+                           test_in_reply_to, test_in_reply_to_atom_uri,
+                           test_subject, test_schedule_post,
+                           test_event_date, test_event_time, test_location,
+                           test_is_article, system_language, conversation_id,
+                           low_bandwidth, content_license_url,
+                           languages_understood)
+    assert test_post_json
+    assert test_post_json.get('object')
+    assert test_post_json['object']['content']
+    assert 'Fediverse' in test_post_json['object']['content']
+    content_str = test_post_json['object']['content']
+    object_content = remove_long_words(content_str, 40, [])
+    assert 'Fediverse' in object_content
+    bold_reading = False
+    object_content = remove_text_formatting(object_content, bold_reading)
+    assert 'Fediverse' in object_content
+    object_content = limit_repeated_words(object_content, 6)
+    assert 'Fediverse' in object_content
+    object_content = html_replace_email_quote(object_content)
+    assert 'Fediverse' in object_content
+    object_content = html_replace_quote_marks(object_content)
+    assert 'Fediverse' in object_content
+
     os.chdir(curr_dir)
     shutil.rmtree(base_dir, ignore_errors=False, onerror=None)