Fix hashtagging of HTML entities

2 files changed, 10 insertions, 4 deletions

diff --git a/activities/models/hashtag.py b/activities/models/hashtag.py
index 7085b15..9cfb21d 100644
--- a/activities/models/hashtag.py
+++ b/activities/models/hashtag.py
@@ -123,7 +123,7 @@ class Hashtag(StatorModel):
         delete = "{edit}delete/"
         timeline = "/tags/{self.hashtag}/"
 
-    hashtag_regex = re.compile(r"((?:\B#)([a-zA-Z0-9(_)]{1,}\b))")
+    hashtag_regex = re.compile(r"\B#([a-zA-Z0-9(_)]+\b)(?!;)")
 
     def save(self, *args, **kwargs):
         self.hashtag = self.hashtag.lstrip("#")
@@ -182,7 +182,7 @@ class Hashtag(StatorModel):
     @classmethod
     def linkify_hashtags(cls, content) -> str:
         def replacer(match):
-            hashtag = match.group()
-            return f'<a class="hashtag" href="/tags/{hashtag.lstrip("#").lower()}/">{hashtag}</a>'
+            hashtag = match.group(1)
+            return f'<a class="hashtag" href="/tags/{hashtag.lower()}/">#{hashtag}</a>'
 
         return mark_safe(Hashtag.hashtag_regex.sub(replacer, content))
diff --git a/tests/core/test_html.py b/tests/core/test_html.py
index 012a0ce..d4c74dc 100644
--- a/tests/core/test_html.py
+++ b/tests/core/test_html.py
@@ -1,4 +1,4 @@
-from core.html import html_to_plaintext
+from core.html import html_to_plaintext, sanitize_post
 
 
 def test_html_to_plaintext():
@@ -13,3 +13,9 @@ def test_html_to_plaintext():
         html_to_plaintext("<p>Hi!</p>\n\n<p>How are<br> you?</p><p>today</p>")
         == "Hi!\n\nHow are\n you?\n\ntoday"
     )
+
+
+def test_sanitize_post():
+
+    assert sanitize_post("<p>Hello!</p>") == "<p>Hello!</p>"
+    assert sanitize_post("<p>It&#39;s great</p>") == "<p>It&#39;s great</p>"