Significantly better hashtag link parsing

Fixes #203

activities/models/post.py

@@ -731,7 +731,7 @@ class Post(StatorModel):
                 if tag["type"].lower() == "mention":
                     mention_identity = Identity.by_actor_uri(tag["href"], create=True)
                     post.mentions.add(mention_identity)
-                elif tag["type"].lower() == "hashtag":
+                elif tag["type"].lower() in ["_:hashtag", "hashtag"]:
                     post.hashtags.append(tag["name"].lower().lstrip("#"))
                 elif tag["type"].lower() in ["toot:emoji", "emoji"]:
                     emoji = Emoji.by_ap_tag(post.author.domain, tag, create=True)

core/html.py

@@ -34,11 +34,10 @@ def sanitize_html(post_html: str) -> str:
     Only allows a, br, p and span tags, and class attributes.
     """
     cleaner = bleach.Cleaner(
-        tags=["br", "p"],
+        tags=["br", "p", "a"],
         attributes={  # type:ignore
             "a": allow_a,
             "p": ["class"],
-            "span": ["class"],
         },
         filters=[partial(LinkifyFilter, url_re=url_regex)],
         strip=True,
@@ -148,16 +147,26 @@ class ContentRenderer:
     def linkify_hashtags(self, html, identity) -> str:
         from activities.models import Hashtag
 
-        def replacer(match):
+        def replacer(attrs, new=False):
-            hashtag = match.group(1)
+            # See if the text in this link looks like a hashtag
+            if not Hashtag.hashtag_regex.match(attrs.get("_text", "")):
+                return attrs
+            hashtag = attrs["_text"].strip().lstrip("#")
+            attrs[None, "class"] = "hashtag"
+            if (None, "rel") in attrs:
+                del attrs[None, "rel"]
             if self.local:
-                return (
+                attrs[None, "href"] = f"/tags/{hashtag.lower()}/"
-                    f'<a class="hashtag" href="/tags/{hashtag.lower()}/">#{hashtag}</a>'
-                )
             else:
-                return f'<a class="hashtag" href="https://{identity.domain.uri_domain}/tags/{hashtag.lower()}/">#{hashtag}</a>'
+                attrs[
+                    None, "href"
+                ] = f"https://{identity.domain.uri_domain}/tags/{hashtag.lower()}/"
+            return attrs
 
-        return Hashtag.hashtag_regex.sub(replacer, html)
+        linker = bleach.linkifier.Linker(
+            url_re=Hashtag.hashtag_regex, callbacks=[replacer]
+        )
+        return linker.linkify(html)
 
     def imageify_emojis(self, html: str, identity, include_local: bool = True):
         """

tests/core/test_html.py

@@ -1,4 +1,8 @@
-from core.html import html_to_plaintext, sanitize_html
+from unittest.mock import Mock
+
+import pytest
+
+from core.html import ContentRenderer, html_to_plaintext, sanitize_html
 
 
 def test_html_to_plaintext():
@@ -31,3 +35,29 @@ def test_sanitize_post():
         sanitize_html("<p>@someone@subdomain.some-domain.com</p>")
         == "<p>@someone@subdomain.some-domain.com</p>"
     )
+
+
+@pytest.mark.django_db
+def test_link_preservation(emoji_locals):
+    """
+    We want to:
+     - Preserve incoming links from other servers
+     - Linkify mentions and hashtags
+     - Not have these all step on each other!
+    """
+    renderer = ContentRenderer(local=True)
+    fake_mention = Mock()
+    fake_mention.username = "andrew"
+    fake_mention.domain_id = "aeracode.org"
+    fake_mention.urls.view = "/@andrew@aeracode.org/"
+    fake_post = Mock()
+    fake_post.mentions.all.return_value = [fake_mention]
+    fake_post.author.domain.uri_domain = "example.com"
+
+    assert (
+        renderer.render_post(
+            'Hello @andrew, I want to link to this <span>#</span>hashtag: <a href="http://example.com/@andrew/#notahashtag">here</a> and rewrite <a href="https://example.com/tags/thishashtag/">#thishashtag</a>',
+            fake_post,
+        )
+        == 'Hello <a href="/@andrew@aeracode.org/">@andrew</a>, I want to link to this <a href="/tags/hashtag/" class="hashtag">#hashtag</a>: <a href="http://example.com/@andrew/#notahashtag" rel="nofollow">here</a> and rewrite <a href="/tags/thishashtag/" class="hashtag">#thishashtag</a>'
+    )