Remove link tracking gloop from post urls

content.py

@@ -663,6 +663,27 @@ def _contains_academic_references(content: str) -> bool:
     return False
 
 
+def remove_link_trackers_from_content(content: str) -> str:
+    """ Removes any link trackers from urls within the content
+    """
+    if '?utm_' not in content:
+        return content
+    sections = content.split('?utm_')
+    ctr = 0
+    new_content = ''
+    for section_str in sections:
+        if ctr == 0:
+            new_content = section_str
+            ctr = 1
+            continue
+        if '"' in section_str:
+            new_content += '"' + section_str.split('"', 1)[1]
+        else:
+            new_content += section_str
+        ctr += 1
+    return new_content
+
+
 def remove_link_tracking(url: str) -> str:
     """ Removes any web link tracking, such as utm_medium, utm_campaign
     or utm_source

tests.py

@@ -147,6 +147,7 @@ from inbox import valid_inbox
 from inbox import valid_inbox_filenames
 from inbox import cache_svg_images
 from categories import guess_hashtag_category
+from content import remove_link_trackers_from_content
 from content import remove_link_tracking
 from content import format_mixed_right_to_left
 from content import replace_remote_hashtags
@@ -8778,6 +8779,24 @@ def _test_link_tracking() -> None:
         'habits-and-then-sell-that-to-letter-agencies'
     assert remove_link_tracking(url) == expected
 
+    content = 'Some content'
+    expected = content
+    assert remove_link_trackers_from_content(content) == expected
+
+    content = \
+        'Some <a href="dreadfulsite.com/abc?utm_medium=gloop">content</a>'
+    expected = \
+        'Some <a href="dreadfulsite.com/abc">content</a>'
+    assert remove_link_trackers_from_content(content) == expected
+
+    content = \
+        'Some <a href="dreadfulsite.com/abc?utm_medium=gloop">content</a> ' + \
+        '<a href="surveillancecrap.com/def?utm_campaign=ohno">scurrilous</a>'
+    expected = \
+        'Some <a href="dreadfulsite.com/abc">content</a> ' + \
+        '<a href="surveillancecrap.com/def">scurrilous</a>'
+    assert remove_link_trackers_from_content(content) == expected
+
 
 def run_all_tests():
     base_dir = os.getcwd()

webapp_post.py

@@ -78,6 +78,7 @@ from utils import get_attributed_to
 from utils import get_reply_to
 from utils import get_actor_from_post
 from utils import resembles_url
+from content import remove_link_trackers_from_content
 from content import format_mixed_right_to_left
 from content import replace_remote_hashtags
 from content import detect_dogwhistles
@@ -2845,6 +2846,7 @@ def individual_post_as_html(signing_priv_key_pem: str,
                not post_is_blog:
                 content_str = bold_reading_string(content_str)
 
+            object_content = remove_link_trackers_from_content(content_str)
             object_content = \
                 remove_long_words(content_str, 40, [])
             object_content = \