Fix inbox processing errors from pinned posts and non-Mastodon servers (#596)

If a post (interaction) comes in from AP inbox but no local author profile exists,
fetch_actor will pull in both the identity AND its pinned posts, which the incoming
post might have been. This would case a database integrity violation. We check
for post existing again after syncing the actor.

Post processing also barfed on posts where content didn't follow Mastodon specs.
For example, Kbin sets tag names in 'tag' attribute, instead of 'name' attribute.

activities/models/post.py

@@ -12,9 +12,11 @@ from asgiref.sync import async_to_sync
 from django.contrib.postgres.indexes import GinIndex
 from django.contrib.postgres.search import SearchVector
 from django.db import models, transaction
+from django.db.utils import IntegrityError
 from django.template import loader
 from django.template.defaultfilters import linebreaks_filter
 from django.utils import timezone
+from pyld.jsonld import JsonLdError
 
 from activities.models.emoji import Emoji
 from activities.models.fan_out import FanOut
@@ -842,6 +844,9 @@ class Post(StatorModel):
                 if author.domain is None:
                     if fetch_author:
                         async_to_sync(author.fetch_actor)()
+                        # perhaps the entire "try again" logic below
+                        # could be replaced with TryAgainLater for
+                        # _all_ fetches, to let it handle pinned posts?
                         if author.domain is None:
                             raise TryAgainLater()
                     else:
@@ -849,22 +854,46 @@ class Post(StatorModel):
                 # If the post is from a blocked domain, stop and drop
                 if author.domain.blocked:
                     raise cls.DoesNotExist("Post is from a blocked domain")
-                post = cls.objects.create(
-                    object_uri=data["id"],
-                    author=author,
-                    content="",
-                    local=False,
-                    type=data["type"],
-                )
-                created = True
+                try:
+                    # try again, because fetch_actor() also fetches pinned posts
+                    post = cls.objects.select_related("author__domain").get(
+                        object_uri=data["id"]
+                    )
+                except cls.DoesNotExist:
+                    # finally, create a stub
+                    try:
+                        post = cls.objects.create(
+                            object_uri=data["id"],
+                            author=author,
+                            content="",
+                            local=False,
+                            type=data["type"],
+                        )
+                        created = True
+                    except IntegrityError as dupe:
+                        # there's still some kind of race condition here
+                        # it's far more rare, but sometimes we fire an
+                        # IntegrityError on activities_post_object_uri_key
+                        # this transaction is now aborted and anything following
+                        # in the caller function will fail in the database.
+                        raise TryAgainLater() from dupe
             else:
                 raise cls.DoesNotExist(f"No post with ID {data['id']}", data)
         if update or created:
             post.type = data["type"]
             if post.type in (cls.Types.article, cls.Types.question):
                 post.type_data = PostTypeData(__root__=data).__root__
-            post.content = get_value_or_map(data, "content", "contentMap")
-            post.summary = data.get("summary")
+            try:
+                # apparently sometimes posts (Pages?) in the fediverse
+                # don't have content?!
+                post.content = get_value_or_map(data, "content", "contentMap")
+            except KeyError:
+                post.content = None
+            # Document types have names, not summaries
+            post.summary = data.get("summary") or data.get("name")
+            if not post.content and post.summary:
+                post.content = post.summary
+                post.summary = None
             post.sensitive = data.get("sensitive", False)
             post.url = data.get("url", data["id"])
             post.published = parse_ld_date(data.get("published"))
@@ -878,10 +907,13 @@ class Post(StatorModel):
                     mention_identity = Identity.by_actor_uri(tag["href"], create=True)
                     post.mentions.add(mention_identity)
                 elif tag_type in ["_:hashtag", "hashtag"]:
+                    # kbin produces tags with 'tag' instead of 'name'
+                    if "tag" in tag and "name" not in tag:
+                        name = get_value_or_map(tag, "tag", "tagMap")
+                    else:
+                        name = get_value_or_map(tag, "name", "nameMap")
                     post.hashtags.append(
-                        get_value_or_map(tag, "name", "nameMap")
-                        .lower()
-                        .lstrip("#")[: Hashtag.MAXIMUM_LENGTH]
+                        name.lower().lstrip("#")[: Hashtag.MAXIMUM_LENGTH]
                     )
                 elif tag_type in ["toot:emoji", "emoji"]:
                     emoji = Emoji.by_ap_tag(post.author.domain, tag, create=True)
@@ -907,6 +939,10 @@ class Post(StatorModel):
             # These have no IDs, so we have to wipe them each time
             post.attachments.all().delete()
             for attachment in get_list(data, "attachment"):
+                if "url" not in attachment.keys():
+                    # sometimes attachments don't have URLs. Skip them.
+                    print(f"no URL for {attachment} in {post}")
+                    continue
                 if "focalPoint" in attachment:
                     try:
                         focal_x, focal_y = attachment["focalPoint"]
@@ -982,8 +1018,10 @@ class Post(StatorModel):
                         update=True,
                         fetch_author=True,
                     )
-                except (json.JSONDecodeError, ValueError):
-                    raise cls.DoesNotExist(f"Invalid ld+json response for {object_uri}")
+                except (json.JSONDecodeError, ValueError, JsonLdError) as err:
+                    raise cls.DoesNotExist(
+                        f"Invalid ld+json response for {object_uri}"
+                    ) from err
                 # We may need to fetch the author too
                 if post.author.state == IdentityStates.outdated:
                     async_to_sync(post.author.fetch_actor)()

activities/models/post_interaction.py

@@ -471,8 +471,9 @@ class PostInteraction(StatorModel):
                 # TODO: Limited retry state?
                 return
 
-            interaction.post.calculate_stats()
-            interaction.post.calculate_type_data()
+            if interaction and interaction.post:
+                interaction.post.calculate_stats()
+                interaction.post.calculate_type_data()
 
     @classmethod
     def handle_undo_ap(cls, data):

users/services/identity.py

@@ -4,6 +4,7 @@ from django.template.defaultfilters import linebreaks_filter
 from activities.models import FanOut, Post, PostInteraction, PostInteractionStates
 from core.files import resize_image
 from core.html import FediverseHtmlParser
+from stator.exceptions import TryAgainLater
 from users.models import (
     Block,
     BlockStates,
@@ -201,6 +202,10 @@ class IdentityService:
                 except Post.DoesNotExist:
                     # ignore 404s...
                     pass
+                except TryAgainLater:
+                    # when fetching a post -> author -> post we can
+                    # get into a state. Ignore this round.
+                    pass
             for removed in PostInteraction.objects.filter(
                 type=PostInteraction.Types.pin,
                 identity=self.identity,