From 2d9f4a3681af33e942b7e6c5aa0c920bf4e7e7e1 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Sun, 19 Feb 2023 11:36:35 +0000
Subject: [PATCH] Remove css style from post content

This could otherwise mess up the UI
---
 tests.py       | 18 ++++++++++++++++++
 utils.py       | 20 ++++++++++++++++++++
 webapp_post.py |  5 +++++
 3 files changed, 43 insertions(+)

diff --git a/tests.py b/tests.py
index 026f1a11a..ce6281887 100644
--- a/tests.py
+++ b/tests.py
@@ -54,6 +54,7 @@ from follow import clear_followers
 from follow import send_follow_request_via_server
 from follow import send_unfollow_request_via_server
 from siteactive import site_is_active
+from utils import remove_style_within_html
 from utils import html_tag_has_closing
 from utils import remove_inverted_text
 from utils import remove_square_capitals
@@ -7865,6 +7866,22 @@ def _test_html_closing_tag() -> None:
     assert not html_tag_has_closing('code', content)
 
 
+def _test_remove_style() -> None:
+    print('remove_style')
+    html_str = '<p>this is a test</p>'
+    result = remove_style_within_html(html_str)
+    assert result == html_str
+
+    html_str = \
+        '<span style="font-size: 200%" class="mfm _mfm_x2_">something</span>'
+    result = remove_style_within_html(html_str)
+    expected = \
+        '<span class="mfm _mfm_x2_">something</span>'
+    if result != expected:
+        print(expected + '\n\n' + result)
+    assert result == expected
+
+
 def run_all_tests():
     base_dir = os.getcwd()
     print('Running tests...')
@@ -7882,6 +7899,7 @@ def run_all_tests():
     _test_checkbox_names()
     _test_thread_functions()
     _test_functions()
+    _test_remove_style()
     _test_html_closing_tag()
     _test_replace_remote_tags()
     _test_replace_variable()
diff --git a/utils.py b/utils.py
index e8324fda2..d5fa9c724 100644
--- a/utils.py
+++ b/utils.py
@@ -655,6 +655,26 @@ def remove_html(content: str) -> str:
     return result
 
 
+def remove_style_within_html(content: str) -> str:
+    """Removes style="something" within html post content.
+    Used to ensure that styles
+    """
+    if '<' not in content:
+        return content
+    if ' style="' not in content:
+        return content
+    sections = content.split(' style="')
+    result = ''
+    ctr = 0
+    for section_text in sections:
+        if ctr > 0:
+            result += section_text.split('"', 1)[1]
+        else:
+            result = section_text
+        ctr = 1
+    return result
+
+
 def first_paragraph_from_string(content: str) -> str:
     """Get the first paragraph from a blog post
     to be used as a summary in the newswire feed
diff --git a/webapp_post.py b/webapp_post.py
index 0bb9592bf..54e15755d 100644
--- a/webapp_post.py
+++ b/webapp_post.py
@@ -25,6 +25,7 @@ from posts import post_is_muted
 from posts import get_person_box
 from posts import download_announce
 from posts import populate_replies_json
+from utils import remove_style_within_html
 from utils import license_link_from_name
 from utils import dont_speak_hashtags
 from utils import remove_eol
@@ -2423,6 +2424,10 @@ def individual_post_as_html(signing_priv_key_pem: str,
         languages_understood = get_actor_languages_list(actor_json)
     content_str = get_content_from_post(post_json_object, system_language,
                                         languages_understood)
+
+    # remove any css styling within the post itself
+    content_str = remove_style_within_html(content_str)
+
     content_language = \
         get_language_from_post(post_json_object, system_language,
                                languages_understood)