From 709dc8616232076e295b85d570119765415e833f Mon Sep 17 00:00:00 2001
From: Andrew Godwin <andrew@aeracode.org>
Date: Wed, 3 May 2023 23:11:34 -0600
Subject: [PATCH] Fixed #559: Trim hashtags to 100 chars or less

---
 activities/models/hashtag.py         |  2 ++
 activities/models/post.py            | 16 +++++++++++----
 tests/activities/models/test_post.py | 30 +++++++++++++++++++++++++++-
 3 files changed, 43 insertions(+), 5 deletions(-)

diff --git a/activities/models/hashtag.py b/activities/models/hashtag.py
index 3a212e4..894973c 100644
--- a/activities/models/hashtag.py
+++ b/activities/models/hashtag.py
@@ -87,6 +87,8 @@ class HashtagManager(models.Manager):
 
 class Hashtag(StatorModel):
 
+    MAXIMUM_LENGTH = 100
+
     # Normalized hashtag without the '#'
     hashtag = models.SlugField(primary_key=True, max_length=100)
 
diff --git a/activities/models/post.py b/activities/models/post.py
index aaf2a83..6f2b184 100644
--- a/activities/models/post.py
+++ b/activities/models/post.py
@@ -489,7 +489,10 @@ class Post(StatorModel):
             # Strip all unwanted HTML and apply linebreaks filter, grabbing hashtags on the way
             parser = FediverseHtmlParser(linebreaks_filter(content), find_hashtags=True)
             content = parser.html
-            hashtags = sorted(parser.hashtags) or None
+            hashtags = (
+                sorted([tag[: Hashtag.MAXIMUM_LENGTH] for tag in parser.hashtags])
+                or None
+            )
             # Make the Post object
             post = cls.objects.create(
                 author=author,
@@ -529,7 +532,10 @@ class Post(StatorModel):
             # Strip all HTML and apply linebreaks filter
             parser = FediverseHtmlParser(linebreaks_filter(content), find_hashtags=True)
             self.content = parser.html
-            self.hashtags = sorted(parser.hashtags) or None
+            self.hashtags = (
+                sorted([tag[: Hashtag.MAXIMUM_LENGTH] for tag in parser.hashtags])
+                or None
+            )
             self.summary = summary or None
             self.sensitive = bool(summary) if sensitive is None else sensitive
             self.visibility = visibility
@@ -577,7 +583,7 @@ class Post(StatorModel):
         if self.hashtags:
             for hashtag in self.hashtags:
                 tag, _ = await Hashtag.objects.aget_or_create(
-                    hashtag=hashtag,
+                    hashtag=hashtag[: Hashtag.MAXIMUM_LENGTH],
                 )
                 await tag.atransition_perform(HashtagStates.outdated)
 
@@ -876,7 +882,9 @@ class Post(StatorModel):
                     post.mentions.add(mention_identity)
                 elif tag_type in ["_:hashtag", "hashtag"]:
                     post.hashtags.append(
-                        get_value_or_map(tag, "name", "nameMap").lower().lstrip("#")
+                        get_value_or_map(tag, "name", "nameMap")
+                        .lower()
+                        .lstrip("#")[: Hashtag.MAXIMUM_LENGTH]
                     )
                 elif tag_type in ["toot:emoji", "emoji"]:
                     emoji = Emoji.by_ap_tag(post.author.domain, tag, create=True)
diff --git a/tests/activities/models/test_post.py b/tests/activities/models/test_post.py
index 6688ef6..58d5e05 100644
--- a/tests/activities/models/test_post.py
+++ b/tests/activities/models/test_post.py
@@ -1,7 +1,7 @@
 import pytest
 from pytest_httpx import HTTPXMock
 
-from activities.models import Post, PostStates
+from activities.models import Hashtag, Post, PostStates
 from activities.models.post_types import QuestionData
 from users.models import Identity, InboxMessage
 
@@ -57,6 +57,34 @@ def test_post_create_edit(identity: Identity, config_system):
     assert list(post.mentions.all()) == []
 
 
+@pytest.mark.django_db
+def test_ensure_hashtag(identity: Identity, config_system, stator):
+    """
+    Tests that normal hashtags get a Hashtag object created, and a hashtag
+    over our limit of 100 characters is truncated.
+    """
+    # Normal length hashtag
+    post = Post.create_local(
+        author=identity,
+        content="Hello, #testtag",
+    )
+    stator.run_single_cycle_sync()
+    assert post.hashtags == ["testtag"]
+    assert Hashtag.objects.filter(hashtag="testtag").exists()
+    # Excessively long hashtag
+    post = Post.create_local(
+        author=identity,
+        content="Hello, #thisisahashtagthatiswaytoolongandissignificantlyaboveourmaximumlimitofonehundredcharacterswhytheywouldbethislongidontknow",
+    )
+    stator.run_single_cycle_sync()
+    assert post.hashtags == [
+        "thisisahashtagthatiswaytoolongandissignificantlyaboveourmaximumlimitofonehundredcharacterswhytheywou"
+    ]
+    assert Hashtag.objects.filter(
+        hashtag="thisisahashtagthatiswaytoolongandissignificantlyaboveourmaximumlimitofonehundredcharacterswhytheywou"
+    ).exists()
+
+
 @pytest.mark.django_db
 def test_linkify_mentions_remote(
     identity, identity2, remote_identity, remote_identity2