Basic Emoji suppport (#157)

4 files changed, 301 insertions, 19 deletions

diff --git a/activities/models/__init__.py b/activities/models/__init__.py
index aa34c0f..15a08a2 100644
--- a/activities/models/__init__.py
+++ b/activities/models/__init__.py
@@ -1,3 +1,4 @@
+from .emoji import Emoji, EmojiStates  # noqa
 from .fan_out import FanOut, FanOutStates  # noqa
 from .hashtag import Hashtag, HashtagStates  # noqa
 from .post import Post, PostStates  # noqa
diff --git a/activities/models/emoji.py b/activities/models/emoji.py
new file mode 100644
index 0000000..00f6e67
--- /dev/null
+++ b/activities/models/emoji.py
@@ -0,0 +1,261 @@
+import re
+from functools import partial
+from typing import ClassVar, cast
+
+import urlman
+from asgiref.sync import sync_to_async
+from django.conf import settings
+from django.core.exceptions import ValidationError
+from django.db import models
+from django.utils.safestring import mark_safe
+
+from core.files import get_remote_file
+from core.html import strip_html
+from core.models import Config
+from core.uploads import upload_emoji_namer
+from core.uris import AutoAbsoluteUrl, RelativeAbsoluteUrl, StaticAbsoluteUrl
+from stator.models import State, StateField, StateGraph, StatorModel
+from users.models import Domain
+
+
+class EmojiStates(StateGraph):
+    outdated = State(try_interval=300, force_initial=True)
+    updated = State()
+
+    outdated.transitions_to(updated)
+
+    @classmethod
+    async def handle_outdated(cls, instance: "Emoji"):
+        """
+        Fetches remote emoji and uploads to file for local caching
+        """
+        if instance.remote_url and not instance.file:
+            file, mimetype = await get_remote_file(
+                instance.remote_url,
+                timeout=settings.SETUP.REMOTE_TIMEOUT,
+                max_size=settings.SETUP.EMOJI_MAX_IMAGE_FILESIZE_KB * 1024,
+            )
+            if file:
+                instance.file = file
+                instance.mimetype = mimetype
+                await sync_to_async(instance.save)()
+
+        return cls.updated
+
+
+class EmojiQuerySet(models.QuerySet):
+    def usable(self, domain: Domain | None = None):
+        public_q = models.Q(public=True)
+        if Config.system.emoji_unreviewed_are_public:
+            public_q |= models.Q(public__isnull=True)
+
+        qs = self.filter(public_q)
+        if domain:
+            if domain.local:
+                qs = qs.filter(local=True)
+            else:
+                qs = qs.filter(domain=domain)
+        return qs
+
+
+class EmojiManager(models.Manager):
+    def get_queryset(self):
+        return EmojiQuerySet(self.model, using=self._db)
+
+    def usable(self, domain: Domain | None = None):
+        return self.get_queryset().usable(domain)
+
+
+class Emoji(StatorModel):
+
+    # Normalized Emoji without the ':'
+    shortcode = models.SlugField(max_length=100, db_index=True)
+
+    domain = models.ForeignKey(
+        "users.Domain", null=True, blank=True, on_delete=models.CASCADE
+    )
+    local = models.BooleanField(default=True)
+
+    # Should this be shown in the public UI?
+    public = models.BooleanField(null=True)
+
+    object_uri = models.CharField(max_length=500, blank=True, null=True, unique=True)
+
+    mimetype = models.CharField(max_length=200)
+
+    # Files may not be populated if it's remote and not cached on our side yet
+    file = models.ImageField(
+        upload_to=partial(upload_emoji_namer, "emoji"),
+        null=True,
+        blank=True,
+    )
+
+    # A link to the custom emoji
+    remote_url = models.CharField(max_length=500, blank=True, null=True)
+
+    # Used for sorting custom emoji in the picker
+    category = models.CharField(max_length=100, blank=True, null=True)
+
+    # State of this Emoji
+    state = StateField(EmojiStates)
+
+    created = models.DateTimeField(auto_now_add=True)
+    updated = models.DateTimeField(auto_now=True)
+
+    objects = EmojiManager()
+
+    # Cache of the local emojis {shortcode: Emoji}
+    locals: ClassVar["dict[str, Emoji]"]
+
+    class Meta:
+        unique_together = ("domain", "shortcode")
+
+    class urls(urlman.Urls):
+        root = "/admin/emoji/"
+        create = "{root}/create/"
+        edit = "{root}{self.Emoji}/"
+        delete = "{edit}delete/"
+
+    emoji_regex = re.compile(r"\B:([a-zA-Z0-9(_)-]+):\B")
+
+    def clean(self):
+        super().clean()
+        if self.local ^ (self.domain is None):
+            raise ValidationError("Must be local or have a domain")
+
+    def __str__(self):
+        return f"{self.id}-{self.shortcode}"
+
+    @classmethod
+    def load_locals(cls) -> dict[str, "Emoji"]:
+        return {x.shortcode: x for x in Emoji.objects.usable().filter(local=True)}
+
+    @property
+    def fullcode(self):
+        return f":{self.shortcode}:"
+
+    @property
+    def is_usable(self) -> bool:
+        """
+        Return True if this Emoji is usable.
+        """
+        return self.public or (
+            self.public is None and Config.system.emoji_unreviewed_are_public
+        )
+
+    def full_url(self) -> RelativeAbsoluteUrl:
+        if self.is_usable:
+            if self.file:
+                return AutoAbsoluteUrl(self.file.url)
+            elif self.remote_url:
+                return AutoAbsoluteUrl(f"/proxy/emoji/{self.pk}/")
+        return StaticAbsoluteUrl("img/blank-emoji-128.png")
+
+    def as_html(self):
+        if self.is_usable:
+            return mark_safe(
+                f'<img src="{self.full_url().relative}" class="emoji" alt="Emoji {self.shortcode}">'
+            )
+        return self.fullcode
+
+    @classmethod
+    def imageify_emojis(
+        cls,
+        content: str,
+        *,
+        emojis: list["Emoji"] | EmojiQuerySet | None = None,
+        include_local: bool = True,
+    ):
+        """
+        Find :emoji: in content and convert to <img>. If include_local is True,
+        the local emoji will be used as a fallback for any shortcodes not defined
+        by emojis.
+        """
+        emoji_set = (
+            cast(list[Emoji], list(cls.locals.values())) if include_local else []
+        )
+
+        if emojis:
+            if isinstance(emojis, (EmojiQuerySet, list)):
+                emoji_set.extend(list(emojis))
+            else:
+                raise TypeError("Unsupported type for emojis")
+
+        possible_matches = {
+            emoji.shortcode: emoji.as_html() for emoji in emoji_set if emoji.is_usable
+        }
+
+        def replacer(match):
+            fullcode = match.group(1).lower()
+            if fullcode in possible_matches:
+                return possible_matches[fullcode]
+            return match.group()
+
+        return mark_safe(Emoji.emoji_regex.sub(replacer, content))
+
+    @classmethod
+    def emojis_from_content(cls, content: str, domain: Domain) -> list[str]:
+        """
+        Return a parsed and sanitized of emoji found in content without
+        the surrounding ':'.
+        """
+        emoji_hits = cls.emoji_regex.findall(strip_html(content))
+        emojis = sorted({emoji.lower() for emoji in emoji_hits})
+        return list(
+            cls.objects.filter(local=domain is None)
+            .usable(domain)
+            .filter(shortcode__in=emojis)
+        )
+
+    def to_ap_tag(self):
+        """
+        Return this Emoji as an ActivityPub Tag
+        http://joinmastodon.org/ns#Emoji
+        """
+        return {
+            "id": self.object_uri,
+            "type": "Emoji",
+            "name": self.shortcode,
+            "icon": {
+                "type": "Image",
+                "mediaType": self.mimetype,
+                "url": self.full_url().absolute,
+            },
+        }
+
+    @classmethod
+    def by_ap_tag(cls, domain: Domain, data: dict, create: bool = False):
+        """ """
+        try:
+            return cls.objects.get(object_uri=data["id"])
+        except cls.DoesNotExist:
+            if not create:
+                raise KeyError(f"No emoji with ID {data['id']}", data)
+
+        # create
+        shortcode = data["name"].lower().strip(":")
+        icon = data["icon"]
+        category = (icon.get("category") or "")[:100]
+        emoji = cls.objects.create(
+            shortcode=shortcode,
+            domain=None if domain.local else domain,
+            local=domain.local,
+            object_uri=data["id"],
+            mimetype=icon["mediaType"],
+            category=category,
+            remote_url=icon["url"],
+        )
+        return emoji
+
+    ### Mastodon API ###
+
+    def to_mastodon_json(self):
+        url = self.full_url().absolute
+        data = {
+            "shortcode": self.shortcode,
+            "url": url,
+            "static_url": self.remote_url or url,
+            "visible_in_picker": self.public,
+            "category": self.category or "",
+        }
+        return data
diff --git a/activities/models/fan_out.py b/activities/models/fan_out.py
index 8f4f342..fd52acd 100644
--- a/activities/models/fan_out.py
+++ b/activities/models/fan_out.py
@@ -184,8 +184,14 @@ class FanOut(StatorModel):
         """
         Returns a version of the object with all relations pre-loaded
         """
-        return await FanOut.objects.select_related(
-            "identity",
-            "subject_post",
-            "subject_post_interaction",
-        ).aget(pk=self.pk)
+        return (
+            await FanOut.objects.select_related(
+                "identity",
+                "subject_post",
+                "subject_post_interaction",
+            )
+            .prefetch_related(
+                "subject_post__emojis",
+            )
+            .aget(pk=self.pk)
+        )
diff --git a/activities/models/post.py b/activities/models/post.py
index 8e355bf..2b0a7c2 100644
--- a/activities/models/post.py
+++ b/activities/models/post.py
@@ -12,8 +12,10 @@ from django.template.defaultfilters import linebreaks_filter
 from django.utils import timezone
 from django.utils.safestring import mark_safe
 
+from activities.models.emoji import Emoji
 from activities.models.fan_out import FanOut
 from activities.models.hashtag import Hashtag
+from activities.templatetags.emoji_tags import imageify_emojis
 from core.html import sanitize_post, strip_html
 from core.ld import canonicalise, format_ld_date, get_list, parse_ld_date
 from stator.models import State, StateField, StateGraph, StatorModel
@@ -218,6 +220,12 @@ class Post(StatorModel):
     # Hashtags in the post
     hashtags = models.JSONField(blank=True, null=True)
 
+    emojis = models.ManyToManyField(
+        "activities.Emoji",
+        related_name="posts_using_emoji",
+        blank=True,
+    )
+
     # When the post was originally created (as opposed to when we received it)
     published = models.DateTimeField(default=timezone.now)
 
@@ -328,8 +336,11 @@ class Post(StatorModel):
         """
         Returns the content formatted for local display
         """
-        return Hashtag.linkify_hashtags(
-            self.linkify_mentions(sanitize_post(self.content), local=True)
+        return imageify_emojis(
+            Hashtag.linkify_hashtags(
+                self.linkify_mentions(sanitize_post(self.content), local=True)
+            ),
+            self.author.domain,
         )
 
     def safe_content_remote(self):
@@ -379,6 +390,8 @@ class Post(StatorModel):
                     visibility = reply_to.Visibilities.local_only
             # Find hashtags in this post
             hashtags = Hashtag.hashtags_from_content(content) or None
+            # Find emoji in this post
+            emojis = Emoji.emojis_from_content(content, author.domain)
             # Strip all HTML and apply linebreaks filter
             content = linebreaks_filter(strip_html(content))
             # Make the Post object
@@ -395,6 +408,7 @@ class Post(StatorModel):
             post.object_uri = post.urls.object_uri
             post.url = post.absolute_object_uri()
             post.mentions.set(mentions)
+            post.emojis.set(emojis)
             if attachments:
                 post.attachments.set(attachments)
             post.save()
@@ -416,6 +430,7 @@ class Post(StatorModel):
             self.edited = timezone.now()
             self.hashtags = Hashtag.hashtags_from_content(content) or None
             self.mentions.set(self.mentions_from_content(content, self.author))
+            self.emojis.set(Emoji.emojis_from_content(content, self.author.domain))
             self.attachments.set(attachments or [])
             self.save()
 
@@ -520,14 +535,11 @@ class Post(StatorModel):
             value["updated"] = format_ld_date(self.edited)
         # Mentions
         for mention in self.mentions.all():
-            value["tag"].append(
-                {
-                    "href": mention.actor_uri,
-                    "name": "@" + mention.handle,
-                    "type": "Mention",
-                }
-            )
+            value["tag"].append(mention.to_ap_tag())
             value["cc"].append(mention.actor_uri)
+        # Emoji
+        for emoji in self.emojis.all():
+            value["tag"].append(emoji.to_ap_tag())
         # Attachments
         for attachment in self.attachments.all():
             value["attachment"].append(attachment.to_ap())
@@ -616,7 +628,9 @@ class Post(StatorModel):
         # Do we have one with the right ID?
         created = False
         try:
-            post = cls.objects.get(object_uri=data["id"])
+            post = cls.objects.select_related("author__domain").get(
+                object_uri=data["id"]
+            )
         except cls.DoesNotExist:
             if create:
                 # Resolve the author
@@ -645,10 +659,10 @@ class Post(StatorModel):
                     mention_identity = Identity.by_actor_uri(tag["href"], create=True)
                     post.mentions.add(mention_identity)
                 elif tag["type"].lower() == "as:hashtag":
-                    post.hashtags.append(tag["name"].lstrip("#"))
+                    post.hashtags.append(tag["name"].lower().lstrip("#"))
                 elif tag["type"].lower() == "http://joinmastodon.org/ns#emoji":
-                    # TODO: Handle incoming emoji
-                    pass
+                    emoji = Emoji.by_ap_tag(post.author.domain, tag, create=True)
+                    post.emojis.add(emoji)
                 else:
                     raise ValueError(f"Unknown tag type {tag['type']}")
             # Visibility and to
@@ -818,7 +832,7 @@ class Post(StatorModel):
                 if self.hashtags
                 else []
             ),
-            "emojis": [],
+            "emojis": [emoji.to_mastodon_json() for emoji in self.emojis.usable()],
             "reblogs_count": self.interactions.filter(type="boost").count(),
             "favourites_count": self.interactions.filter(type="like").count(),
             "replies_count": 0,