From eaf0fbf531b529b8146112223544d8ae7dc06942 Mon Sep 17 00:00:00 2001
From: Felix Ableitner <me@nutomic.com>
Date: Thu, 6 Jul 2023 12:08:14 +0200
Subject: [PATCH] Only allow http(s) scheme for urls (ref #3505)

With this change only http(s) schemes are allowed for post.url
field. This is checked for incoming api and federation requests.
Existing posts in database which are sent to clients are not
checked. Neither does it check urls in markdown.
---
 crates/api_crud/src/post/create.rs   |  3 ++-
 crates/api_crud/src/post/update.rs   |  3 ++-
 crates/apub/src/objects/post.rs      |  2 ++
 crates/utils/src/utils/validation.rs | 19 +++++++++++++++++++
 4 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/crates/api_crud/src/post/create.rs b/crates/api_crud/src/post/create.rs
index 1dcc90241..13d896b41 100644
--- a/crates/api_crud/src/post/create.rs
+++ b/crates/api_crud/src/post/create.rs
@@ -31,7 +31,7 @@ use lemmy_utils::{
   error::LemmyError,
   utils::{
     slurs::{check_slurs, check_slurs_opt},
-    validation::{clean_url_params, is_valid_body_field, is_valid_post_title},
+    validation::{check_url_scheme, clean_url_params, is_valid_body_field, is_valid_post_title},
   },
 };
 use tracing::{warn, Instrument};
@@ -58,6 +58,7 @@ impl PerformCrud for CreatePost {
 
     is_valid_post_title(&data.name)?;
     is_valid_body_field(&data.body, true)?;
+    check_url_scheme(&data.url)?;
 
     check_community_ban(local_user_view.person.id, data.community_id, context.pool()).await?;
     check_community_deleted_or_removed(data.community_id, context.pool()).await?;
diff --git a/crates/api_crud/src/post/update.rs b/crates/api_crud/src/post/update.rs
index 632bc5ae7..a8c5715bb 100644
--- a/crates/api_crud/src/post/update.rs
+++ b/crates/api_crud/src/post/update.rs
@@ -20,7 +20,7 @@ use lemmy_utils::{
   error::LemmyError,
   utils::{
     slurs::check_slurs_opt,
-    validation::{clean_url_params, is_valid_body_field, is_valid_post_title},
+    validation::{check_url_scheme, clean_url_params, is_valid_body_field, is_valid_post_title},
   },
 };
 
@@ -50,6 +50,7 @@ impl PerformCrud for EditPost {
     }
 
     is_valid_body_field(&data.body, true)?;
+    check_url_scheme(&data.url)?;
 
     let post_id = data.post_id;
     let orig_post = Post::read(context.pool(), post_id).await?;
diff --git a/crates/apub/src/objects/post.rs b/crates/apub/src/objects/post.rs
index 8064c771b..d73642be6 100644
--- a/crates/apub/src/objects/post.rs
+++ b/crates/apub/src/objects/post.rs
@@ -44,6 +44,7 @@ use lemmy_utils::{
     markdown::markdown_to_html,
     slurs::{check_slurs_opt, remove_slurs},
     time::convert_datetime,
+    validation::check_url_scheme,
   },
 };
 use std::ops::Deref;
@@ -191,6 +192,7 @@ impl Object for ApubPost {
       } else {
         None
       };
+      check_url_scheme(&url)?;
 
       let local_site = LocalSite::read(context.pool()).await.ok();
       let allow_sensitive = local_site_opt_to_sensitive(&local_site);
diff --git a/crates/utils/src/utils/validation.rs b/crates/utils/src/utils/validation.rs
index e43ddbbb5..b5ae5b74f 100644
--- a/crates/utils/src/utils/validation.rs
+++ b/crates/utils/src/utils/validation.rs
@@ -302,12 +302,22 @@ pub fn check_site_visibility_valid(
   Ok(())
 }
 
+pub fn check_url_scheme(url: &Option<Url>) -> LemmyResult<()> {
+  if let Some(url) = url {
+    if url.scheme() != "http" && url.scheme() != "https" {
+      return Err(LemmyError::from_message("invalid_url_scheme"));
+    }
+  }
+  Ok(())
+}
+
 #[cfg(test)]
 mod tests {
   use super::build_totp_2fa;
   use crate::utils::validation::{
     build_and_check_regex,
     check_site_visibility_valid,
+    check_url_scheme,
     clean_url_params,
     generate_totp_2fa_secret,
     is_valid_actor_name,
@@ -519,4 +529,13 @@ mod tests {
     assert!(check_site_visibility_valid(false, false, &Some(true), &None).is_ok());
     assert!(check_site_visibility_valid(false, false, &None, &Some(true)).is_ok());
   }
+
+  #[test]
+  fn test_check_url_scheme() {
+    assert!(check_url_scheme(&None).is_ok());
+    assert!(check_url_scheme(&Some(Url::parse("http://example.com").unwrap())).is_ok());
+    assert!(check_url_scheme(&Some(Url::parse("https://example.com").unwrap())).is_ok());
+    assert!(check_url_scheme(&Some(Url::parse("ftp://example.com").unwrap())).is_err());
+    assert!(check_url_scheme(&Some(Url::parse("javascript:void").unwrap())).is_err());
+  }
 }