From c0d569968a79f234206dab9bd0d76ef825dad46b Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Thu, 15 Feb 2024 17:29:16 +0000
Subject: [PATCH] Fixing code tag removals

---
 tests.py       | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++
 utils.py       |  3 ++-
 webapp_post.py | 14 +++++++++----
 3 files changed, 65 insertions(+), 5 deletions(-)

diff --git a/tests.py b/tests.py
index 1ee2edd0d..28aa64ce5 100644
--- a/tests.py
+++ b/tests.py
@@ -187,6 +187,7 @@ from newswire import parse_feed_date
 from newswire import limit_word_lengths
 from mastoapiv1 import get_masto_api_v1id_from_nickname
 from mastoapiv1 import get_nickname_from_masto_api_v1id
+from webapp_post import remove_incomplete_code_tags
 from webapp_post import replace_link_variable
 from webapp_post import prepare_html_post_nickname
 from speaker import speaker_replace_links
@@ -5932,6 +5933,33 @@ def _test_links_within_post(base_dir: str) -> None:
     assert post_json_object['object']['content'] == content
     assert post_json_object['object']['contentMap'][system_language] == content
 
+    content = "<p>I see confusion between <code>git bulldada</code> and " + \
+        "<code>git bollocks</code>.</p><p><code>git-checkout</code> " + \
+        "changes everything or fucks up trees and rodents.</p><p>" + \
+        "<code>git vermin</code> obliterates <code>hamsters</code> and " + \
+        "<code>gerbils</code> and that is all she wrote.</p>"
+    post_json_object = \
+        create_public_post(base_dir, nickname, domain, port, http_prefix,
+                           content,
+                           False,
+                           False, True,
+                           None, None,
+                           '', '', None,
+                           test_in_reply_to, test_in_reply_to_atom_uri,
+                           test_subject, test_schedule_post,
+                           test_event_date, test_event_time,
+                           test_event_end_time, test_location,
+                           test_is_article, system_language, conversation_id,
+                           low_bandwidth, content_license_url,
+                           media_license_url, media_creator,
+                           languages_understood, translate, buy_url, chat_url,
+                           auto_cw_cache)
+    if post_json_object['object']['content'] != content:
+        print('content1: ' + post_json_object['object']['content'])
+        print('content2: ' + content)
+    assert post_json_object['object']['content'] == content
+    assert post_json_object['object']['contentMap'][system_language] == content
+
 
 def _test_mastoapi():
     print('test_masto_api')
@@ -8672,6 +8700,30 @@ def _test_check_individual_post_content():
     assert content5 == content
 
 
+def _test_remove_tags() -> None:
+    print('remove_tags')
+    content = 'This is some content'
+    result = remove_incomplete_code_tags(content)
+    assert result == content
+
+    content = '<code>This is some content'
+    result = remove_incomplete_code_tags(content)
+    assert result == 'This is some content'
+
+    content = 'This is some content</code>'
+    result = remove_incomplete_code_tags(content)
+    assert result == 'This is some content'
+
+    content = '<code>This is some content</code>. <code>Some other content'
+    result = remove_incomplete_code_tags(content)
+    assert result == 'This is some content. Some other content'
+
+    content = \
+        '<code>This is some content</code>. <code>Some other content</code>'
+    result = remove_incomplete_code_tags(content)
+    assert result == 'This is some content. Some other content'
+
+
 def run_all_tests():
     base_dir = os.getcwd()
     print('Running tests...')
@@ -8689,6 +8741,7 @@ def run_all_tests():
     _test_checkbox_names()
     _test_thread_functions()
     _test_functions()
+    _test_remove_tags()
     _test_check_individual_post_content()
     _test_uninvert2()
     _test_book_link(base_dir)
diff --git a/utils.py b/utils.py
index 084f720b8..716d88c64 100644
--- a/utils.py
+++ b/utils.py
@@ -402,7 +402,8 @@ def has_object_dict(post_json_object: {}) -> bool:
 def remove_markup_tag(html: str, tag: str) -> str:
     """Remove the given tag from the given html markup
     """
-    if '<' + tag not in html:
+    if '<' + tag not in html and \
+       '</' + tag not in html:
         return html
 
     section = html.split('<' + tag)
diff --git a/webapp_post.py b/webapp_post.py
index 63d4075ad..fcb1b38df 100644
--- a/webapp_post.py
+++ b/webapp_post.py
@@ -2041,14 +2041,20 @@ def _get_buy_footer(buy_links: {}, translate: {}) -> str:
     return buy_str
 
 
-def _remove_incomplete_code_tags(content: str) -> str:
+def remove_incomplete_code_tags(content: str) -> str:
     """Remove any uncompleted code tags
     """
     tags = ('code', 'pre')
     for tag_name in tags:
-        if '<' + tag_name not in content:
+        if '<' + tag_name not in content and \
+           '</' + tag_name not in content:
             continue
-        if '</' + tag_name not in content:
+        if '<' + tag_name not in content or \
+           '</' + tag_name not in content:
+            content = remove_markup_tag(content, tag_name)
+            continue
+        if content.count('<' + tag_name) > 1 or \
+           content.count('</' + tag_name) > 1:
             content = remove_markup_tag(content, tag_name)
     return content
 
@@ -2849,7 +2855,7 @@ def individual_post_as_html(signing_priv_key_pem: str,
             encrypted_str = translate[encrypted_str]
         object_content = '🔒 ' + encrypted_str
 
-    object_content = _remove_incomplete_code_tags(object_content)
+    object_content = remove_incomplete_code_tags(object_content)
 
     object_content = \
         '<article><span itemprop="articleBody">' + \