Validate content of incoming Note objects

docs/openapi.yaml

@@ -337,7 +337,9 @@ paths:
               type: object
               properties:
                 status:
-                  description: Text content of the post.
+                  description: |
+                    Text content of the post.
+                    Allowed HTML tags: a, br, pre, code.
                   type: string
                 'media_ids[]':
                   description: Array of Attachment ids to be attached as media.

src/activitypub/receiver.rs

@@ -37,6 +37,7 @@ use crate::models::relationships::queries::{
     unfollow,
 };
 use crate::models::users::queries::get_user_by_name;
+use crate::utils::html::clean_html;
 use super::activity::{Object, Activity, create_activity_accept_follow};
 use super::actor::Actor;
 use super::constants::AP_PUBLIC;
@@ -132,6 +133,16 @@ fn get_object_id(object: Value) -> Result<String, ValidationError> {
     Ok(object_id)
 }
 
+const CONTENT_MAX_SIZE: usize = 100000;
+
+fn clean_note_content(content: &str) -> Result<String, ValidationError> {
+    if content.len() > CONTENT_MAX_SIZE {
+        return Err(ValidationError("content is too long"));
+    };
+    let content_safe = clean_html(content);
+    Ok(content_safe)
+}
+
 pub async fn process_note(
     config: &Config,
     db_client: &mut impl GenericClient,
@@ -221,6 +232,7 @@ pub async fn process_note(
         ).await?;
         let content = object.content
             .ok_or(ValidationError("no content"))?;
+        let content_cleaned = clean_note_content(&content)?;
         let mut attachments: Vec<Uuid> = Vec::new();
         if let Some(list) = object.attachment {
             let mut downloaded = vec![];
@@ -375,7 +387,7 @@ pub async fn process_note(
             Visibility::Direct
         };
         let post_data = PostCreateData {
-            content,
+            content: content_cleaned,
             in_reply_to_id,
             repost_of_id: None,
             visibility,

src/models/posts/types.rs

@@ -10,7 +10,7 @@ use crate::database::int_enum::{int_enum_from_sql, int_enum_to_sql};
 use crate::errors::{ConversionError, DatabaseError, ValidationError};
 use crate::models::attachments::types::DbMediaAttachment;
 use crate::models::profiles::types::DbActorProfile;
-use crate::utils::html::clean_html;
+use crate::utils::html::clean_html_strict;
 
 #[derive(Clone, Debug, PartialEq)]
 pub enum Visibility {
@@ -220,7 +220,7 @@ impl PostCreateData {
         if self.content.chars().count() > character_limit {
             return Err(ValidationError("post is too long"));
         };
-        let content_safe = clean_html(&self.content);
+        let content_safe = clean_html_strict(&self.content);
         let content_trimmed = content_safe.trim();
         if content_trimmed.is_empty() {
             return Err(ValidationError("post can not be empty"));

src/models/profiles/types.rs

@@ -11,7 +11,7 @@ use uuid::Uuid;
 use crate::activitypub::actor::Actor;
 use crate::activitypub::views::get_actor_url;
 use crate::errors::ValidationError;
-use crate::utils::html::clean_html;
+use crate::utils::html::clean_html_strict;
 use super::validators::{
     validate_username,
     validate_display_name,
@@ -173,12 +173,12 @@ pub struct ProfileUpdateData {
 impl ProfileUpdateData {
     pub fn clean(&mut self) -> Result<(), ValidationError> {
         // Validate and clean bio
-        self.bio = self.bio.as_ref().map(|val| clean_html(val));
+        self.bio = self.bio.as_ref().map(|val| clean_html_strict(val));
         // Clean extra fields and remove fields with empty labels
         self.extra_fields = self.extra_fields.iter().cloned()
             .map(|mut field| {
                 field.name = field.name.trim().to_string();
-                field.value = clean_html(&field.value);
+                field.value = clean_html_strict(&field.value);
                 field
             })
             .filter(|field| !field.name.is_empty())

src/utils/html.rs

@@ -3,6 +3,16 @@ use std::collections::HashSet;
 use ammonia::Builder;
 
 pub fn clean_html(unsafe_html: &str) -> String {
+    let safe_html = Builder::default()
+        .add_generic_attributes(&["class"])
+        .add_tag_attributes("a", &["rel", "target"])
+        .link_rel(None)
+        .clean(unsafe_html)
+        .to_string();
+    safe_html
+}
+
+pub fn clean_html_strict(unsafe_html: &str) -> String {
     let mut allowed_tags = HashSet::new();
     allowed_tags.insert("a");
     allowed_tags.insert("br");
@@ -22,8 +32,15 @@ mod tests {
 
     #[test]
     fn test_clean_html() {
-        let unsafe_html = r#"<p>test <b>bold</b><script>dangerous</script> with <a href="https://example.com">link</a> and <code>code</code></p>"#;
+        let unsafe_html = r#"<p><span class="h-card"><a href="https://example.com/user" class="u-url mention" rel="ugc">@<span>user</span></a></span> test</p>"#;
         let safe_html = clean_html(unsafe_html);
+        assert_eq!(safe_html, r#"<p><span class="h-card"><a href="https://example.com/user" class="u-url mention" rel="ugc">@<span>user</span></a></span> test</p>"#);
+    }
+
+    #[test]
+    fn test_clean_html_strict() {
+        let unsafe_html = r#"<p>test <b>bold</b><script>dangerous</script> with <a href="https://example.com">link</a> and <code>code</code></p>"#;
+        let safe_html = clean_html_strict(unsafe_html);
         assert_eq!(safe_html, r#"test bold with <a href="https://example.com" rel="noopener noreferrer">link</a> and <code>code</code>"#);
     }
 }