Disallow <code> tags in bio

src/models/posts/mod.rs

@@ -3,3 +3,4 @@ pub mod helpers;
 pub mod mentions;
 pub mod queries;
 pub mod types;
+mod validators;

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_strict;
+use super::validators::clean_content;
 
 #[derive(Clone, Debug, PartialEq)]
 pub enum Visibility {
@@ -263,15 +263,7 @@ impl PostCreateData {
     /// Validate and clean post data (only for local posts).
     pub fn clean(&mut self, character_limit: usize) -> Result<(), ValidationError> {
         assert!(self.object_id.is_none());
-        if self.content.chars().count() > character_limit {
-            return Err(ValidationError("post is too long"));
-        };
-        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"));
-        };
-        self.content = content_trimmed.to_string();
+        self.content = clean_content(&self.content, character_limit)?;
         Ok(())
     }
 }
@@ -288,16 +280,7 @@ mod tests {
     const POST_CHARACTER_LIMIT: usize = 1000;
 
     #[test]
-    fn test_post_data_empty() {
-        let mut post_data_1 = PostCreateData {
-            content: "  ".to_string(),
-            ..Default::default()
-        };
-        assert_eq!(post_data_1.clean(POST_CHARACTER_LIMIT).is_ok(), false);
-    }
-
-    #[test]
-    fn test_post_data_trimming() {
+    fn test_post_data_clean() {
         let mut post_data_2 = PostCreateData {
             content: "test ".to_string(),
             ..Default::default()

src/models/posts/validators.rs

@@ -0,0 +1,40 @@
+use crate::errors::ValidationError;
+use crate::utils::html::clean_html_strict;
+
+const CONTENT_ALLOWED_TAGS: [&str; 4] = ["a", "br", "pre", "code"];
+
+pub fn clean_content(
+    content: &str,
+    character_limit: usize,
+) -> Result<String, ValidationError> {
+    if content.chars().count() > character_limit {
+        return Err(ValidationError("post is too long"));
+    };
+    let content_safe = clean_html_strict(content, &CONTENT_ALLOWED_TAGS);
+    let content_trimmed = content_safe.trim();
+    if content_trimmed.is_empty() {
+        return Err(ValidationError("post can not be empty"));
+    };
+    Ok(content_trimmed.to_string())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    const POST_CHARACTER_LIMIT: usize = 1000;
+
+    #[test]
+    fn test_clean_content_empty() {
+        let content = "  ";
+        let result = clean_content(content, POST_CHARACTER_LIMIT);
+        assert_eq!(result.is_ok(), false);
+    }
+
+    #[test]
+    fn test_clean_content_trimming() {
+        let content = "test ";
+        let cleaned = clean_content(content, POST_CHARACTER_LIMIT).unwrap();
+        assert_eq!(cleaned, "test");
+    }
+}

src/models/profiles/validators.rs

@@ -4,6 +4,11 @@ use crate::utils::html::{clean_html, clean_html_strict};
 use super::types::ExtraField;
 
 const USERNAME_RE: &str = r"^[a-zA-Z0-9_\.-]+$";
+const DISPLAY_NAME_MAX_LENGTH: usize = 200;
+const BIO_MAX_LENGTH: usize = 10000;
+const BIO_ALLOWED_TAGS: [&str; 2] = ["a", "br"];
+const FIELD_NAME_MAX_SIZE: usize = 500;
+const FIELD_VALUE_MAX_SIZE: usize = 5000;
 
 pub fn validate_username(username: &str) -> Result<(), ValidationError> {
     if username.is_empty() {
@@ -19,8 +24,6 @@ pub fn validate_username(username: &str) -> Result<(), ValidationError> {
     Ok(())
 }
 
-const DISPLAY_NAME_MAX_LENGTH: usize = 200;
-
 pub fn validate_display_name(display_name: &str)
     -> Result<(), ValidationError>
 {
@@ -30,8 +33,6 @@ pub fn validate_display_name(display_name: &str)
     Ok(())
 }
 
-const BIO_MAX_LENGTH: usize = 10000;
-
 pub fn clean_bio(bio: &str, is_remote: bool) -> Result<String, ValidationError> {
     let cleaned_bio = if is_remote {
         // Remote profile
@@ -42,14 +43,11 @@ pub fn clean_bio(bio: &str, is_remote: bool) -> Result<String, ValidationError>
         if bio.chars().count() > BIO_MAX_LENGTH {
             return Err(ValidationError("bio is too long"));
         };
-        clean_html_strict(bio)
+        clean_html_strict(bio, &BIO_ALLOWED_TAGS)
     };
     Ok(cleaned_bio)
 }
 
-const FIELD_NAME_MAX_SIZE: usize = 500;
-const FIELD_VALUE_MAX_SIZE: usize = 5000;
-
 /// Validates extra fields and removes fields with empty labels
 pub fn clean_extra_fields(extra_fields: &[ExtraField])
     -> Result<Vec<ExtraField>, ValidationError>
@@ -57,7 +55,7 @@ pub fn clean_extra_fields(extra_fields: &[ExtraField])
     let mut cleaned_extra_fields = vec![];
     for mut field in extra_fields.iter().cloned() {
         field.name = field.name.trim().to_string();
-        field.value = clean_html_strict(&field.value);
+        field.value = clean_html_strict(&field.value, &BIO_ALLOWED_TAGS);
         if field.name.is_empty() {
             continue;
         };

src/utils/html.rs

@@ -1,4 +1,5 @@
 use std::collections::HashSet;
+use std::iter::FromIterator;
 
 use ammonia::Builder;
 
@@ -12,13 +13,12 @@ pub fn clean_html(unsafe_html: &str) -> 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");
-    allowed_tags.insert("pre");
-    allowed_tags.insert("code");
-
+pub fn clean_html_strict(
+    unsafe_html: &str,
+    allowed_tags: &[&str],
+) -> String {
+    let allowed_tags =
+        HashSet::from_iter(allowed_tags.iter().copied());
     let safe_html = Builder::default()
         .tags(allowed_tags)
         .clean(unsafe_html)
@@ -47,7 +47,7 @@ mod tests {
     #[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);
+        let safe_html = clean_html_strict(unsafe_html, &["a", "br", "code"]);
         assert_eq!(safe_html, r#"test bold with <a href="https://example.com" rel="noopener noreferrer">link</a> and <code>code</code>"#);
     }