2024-05-20 05:18:09 +02:00
|
|
|
use apb::{Node, Base, Object, Document};
|
2024-05-25 05:31:10 +02:00
|
|
|
use sea_orm::{sea_query::Expr, ActiveValue::Set, ColumnTrait, EntityTrait, IntoActiveModel, QueryFilter};
|
2024-05-20 02:49:43 +02:00
|
|
|
use crate::{errors::UpubError, model, server::Context};
|
|
|
|
|
|
|
|
use super::fetcher::Fetcher;
|
|
|
|
|
|
|
|
#[axum::async_trait]
|
|
|
|
pub trait Normalizer {
|
|
|
|
async fn insert_object(&self, obj: impl apb::Object, server: Option<String>) -> crate::Result<model::object::Model>;
|
|
|
|
}
|
|
|
|
|
|
|
|
#[axum::async_trait]
|
|
|
|
impl Normalizer for super::Context {
|
|
|
|
async fn insert_object(&self, object_node: impl apb::Object, server: Option<String>) -> crate::Result<model::object::Model> {
|
2024-05-25 05:31:10 +02:00
|
|
|
let oid = object_node.id().ok_or_else(UpubError::bad_request)?.to_string();
|
|
|
|
let uid = object_node.attributed_to().id();
|
|
|
|
let mut object_model = model::object::ActiveModel::new(&object_node)?;
|
2024-05-20 02:49:43 +02:00
|
|
|
if let Some(server) = server {
|
|
|
|
// make sure we're allowed to create this object
|
2024-05-25 05:31:10 +02:00
|
|
|
if let Set(Some(object_author)) = &object_model.attributed_to {
|
2024-05-20 02:49:43 +02:00
|
|
|
if server != Context::server(object_author) {
|
|
|
|
return Err(UpubError::forbidden());
|
|
|
|
}
|
2024-05-25 05:31:10 +02:00
|
|
|
} else if server != Context::server(&oid) {
|
2024-05-20 02:49:43 +02:00
|
|
|
return Err(UpubError::forbidden());
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
|
|
|
// make sure content only contains a safe subset of html
|
2024-05-25 05:31:10 +02:00
|
|
|
if let Set(Some(content)) = object_model.content {
|
|
|
|
object_model.content = Set(Some(mdhtml::safe_html(&content)));
|
2024-05-20 02:49:43 +02:00
|
|
|
}
|
|
|
|
|
2024-05-24 00:21:02 +02:00
|
|
|
// fix context for remote posts
|
|
|
|
// > note that this will effectively recursively try to fetch the parent object, in order to find
|
|
|
|
// > the context (which is id of topmost object). there's a recursion limit of 16 hidden inside
|
|
|
|
// > btw! also if any link is broken or we get rate limited, the whole insertion fails which is
|
|
|
|
// > kind of dumb. there should be a job system so this can be done in waves. or maybe there's
|
|
|
|
// > some whole other way to do this?? im thinking but misskey aaaa!! TODO
|
2024-05-25 05:31:10 +02:00
|
|
|
if let Set(Some(ref reply)) = object_model.in_reply_to {
|
|
|
|
if let Some(o) = model::object::Entity::find_by_ap_id(reply).one(self.db()).await? {
|
|
|
|
object_model.context = Set(o.context);
|
2024-05-24 00:21:02 +02:00
|
|
|
} else {
|
2024-05-25 05:31:10 +02:00
|
|
|
object_model.context = Set(None); // TODO to be filled by some other task
|
2024-05-24 00:21:02 +02:00
|
|
|
}
|
|
|
|
} else {
|
2024-05-25 05:31:10 +02:00
|
|
|
object_model.context = Set(Some(oid.clone()));
|
2024-05-20 02:49:43 +02:00
|
|
|
}
|
|
|
|
|
2024-05-23 23:27:45 +02:00
|
|
|
model::object::Entity::insert(object_model.clone().into_active_model()).exec(self.db()).await?;
|
2024-05-25 05:31:10 +02:00
|
|
|
let object = model::object::Entity::find_by_ap_id(&oid).one(self.db()).await?.ok_or_else(UpubError::internal_server_error)?;
|
2024-05-23 23:27:45 +02:00
|
|
|
|
2024-05-20 02:49:43 +02:00
|
|
|
// update replies counter
|
2024-05-25 05:31:10 +02:00
|
|
|
if let Set(Some(ref in_reply_to)) = object_model.in_reply_to {
|
2024-05-20 02:49:43 +02:00
|
|
|
if self.fetch_object(in_reply_to).await.is_ok() {
|
|
|
|
model::object::Entity::update_many()
|
|
|
|
.filter(model::object::Column::Id.eq(in_reply_to))
|
2024-05-25 05:31:10 +02:00
|
|
|
.col_expr(model::object::Column::Replies, Expr::col(model::object::Column::Replies).add(1))
|
2024-05-20 02:49:43 +02:00
|
|
|
.exec(self.db())
|
|
|
|
.await?;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// update statuses counter
|
|
|
|
if let Some(object_author) = uid {
|
2024-05-25 05:31:10 +02:00
|
|
|
model::actor::Entity::update_many()
|
|
|
|
.col_expr(model::actor::Column::StatusesCount, Expr::col(model::actor::Column::StatusesCount).add(1))
|
|
|
|
.filter(model::actor::Column::Id.eq(&object_author))
|
2024-05-20 02:49:43 +02:00
|
|
|
.exec(self.db())
|
|
|
|
.await?;
|
|
|
|
}
|
|
|
|
|
|
|
|
for attachment in object_node.attachment().flat() {
|
|
|
|
let attachment_model = match attachment {
|
|
|
|
Node::Empty => continue,
|
|
|
|
Node::Array(_) => {
|
|
|
|
tracing::warn!("ignoring array-in-array while processing attachments");
|
|
|
|
continue
|
|
|
|
},
|
|
|
|
Node::Link(l) => model::attachment::ActiveModel {
|
2024-05-25 05:31:10 +02:00
|
|
|
internal: sea_orm::ActiveValue::NotSet,
|
2024-05-20 02:49:43 +02:00
|
|
|
url: Set(l.href().to_string()),
|
2024-05-25 05:31:10 +02:00
|
|
|
object: Set(object.internal),
|
2024-05-20 02:49:43 +02:00
|
|
|
document_type: Set(apb::DocumentType::Page),
|
|
|
|
name: Set(l.link_name().map(|x| x.to_string())),
|
|
|
|
media_type: Set(l.link_media_type().unwrap_or("link").to_string()),
|
|
|
|
created: Set(chrono::Utc::now()),
|
|
|
|
},
|
|
|
|
Node::Object(o) => model::attachment::ActiveModel {
|
2024-05-25 05:31:10 +02:00
|
|
|
internal: sea_orm::ActiveValue::NotSet,
|
2024-05-20 02:49:43 +02:00
|
|
|
url: Set(o.url().id().unwrap_or_else(|| o.id().map(|x| x.to_string()).unwrap_or_default())),
|
2024-05-25 05:31:10 +02:00
|
|
|
object: Set(object.internal),
|
2024-05-20 02:49:43 +02:00
|
|
|
document_type: Set(o.as_document().map_or(apb::DocumentType::Document, |x| x.document_type().unwrap_or(apb::DocumentType::Page))),
|
|
|
|
name: Set(o.name().map(|x| x.to_string())),
|
|
|
|
media_type: Set(o.media_type().unwrap_or("link").to_string()),
|
2024-05-20 05:18:09 +02:00
|
|
|
created: Set(o.published().unwrap_or_else(chrono::Utc::now)),
|
2024-05-20 02:49:43 +02:00
|
|
|
},
|
|
|
|
};
|
|
|
|
model::attachment::Entity::insert(attachment_model)
|
|
|
|
.exec(self.db())
|
|
|
|
.await?;
|
|
|
|
}
|
|
|
|
// lemmy sends us an image field in posts, treat it like an attachment i'd say
|
|
|
|
if let Some(img) = object_node.image().get() {
|
|
|
|
// TODO lemmy doesnt tell us the media type but we use it to display the thing...
|
|
|
|
let img_url = img.url().id().unwrap_or_default();
|
|
|
|
let media_type = if img_url.ends_with("png") {
|
|
|
|
Some("image/png".to_string())
|
|
|
|
} else if img_url.ends_with("webp") {
|
|
|
|
Some("image/webp".to_string())
|
|
|
|
} else if img_url.ends_with("jpeg") || img_url.ends_with("jpg") {
|
|
|
|
Some("image/jpeg".to_string())
|
|
|
|
} else {
|
|
|
|
None
|
|
|
|
};
|
|
|
|
|
|
|
|
let attachment_model = model::attachment::ActiveModel {
|
2024-05-25 05:31:10 +02:00
|
|
|
internal: sea_orm::ActiveValue::NotSet,
|
2024-05-20 02:49:43 +02:00
|
|
|
url: Set(img.url().id().unwrap_or_else(|| img.id().map(|x| x.to_string()).unwrap_or_default())),
|
2024-05-25 05:31:10 +02:00
|
|
|
object: Set(object.internal),
|
2024-05-20 02:49:43 +02:00
|
|
|
document_type: Set(img.as_document().map_or(apb::DocumentType::Document, |x| x.document_type().unwrap_or(apb::DocumentType::Page))),
|
|
|
|
name: Set(img.name().map(|x| x.to_string())),
|
|
|
|
media_type: Set(img.media_type().unwrap_or(media_type.as_deref().unwrap_or("link")).to_string()),
|
2024-05-20 05:18:09 +02:00
|
|
|
created: Set(img.published().unwrap_or_else(chrono::Utc::now)),
|
2024-05-20 02:49:43 +02:00
|
|
|
};
|
|
|
|
model::attachment::Entity::insert(attachment_model)
|
|
|
|
.exec(self.db())
|
|
|
|
.await?;
|
|
|
|
}
|
|
|
|
|
2024-05-25 05:31:10 +02:00
|
|
|
Ok(object)
|
2024-05-20 02:49:43 +02:00
|
|
|
}
|
|
|
|
}
|