Sanitize HTML from federated data (#ref 647)

Cargo.lock

@@ -389,6 +389,21 @@ dependencies = [
  "memchr",
 ]
 
+[[package]]
+name = "ammonia"
+version = "3.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "89eac85170f4b3fb3dc5e442c1cfb036cb8eecf9dbbd431a161ffad15d90ea3b"
+dependencies = [
+ "html5ever",
+ "lazy_static",
+ "maplit",
+ "markup5ever_rcdom",
+ "matches",
+ "tendril",
+ "url",
+]
+
 [[package]]
 name = "ansi_term"
 version = "0.11.0"
@@ -1325,6 +1340,16 @@ version = "0.3.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3dcaa9ae7725d12cdb85b3ad99a434db70b468c09ded17e012d86b5c1010f7a7"
 
+[[package]]
+name = "futf"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7c9c1ce3fa9336301af935ab852c437817d14cd33690446569392e65170aac3b"
+dependencies = [
+ "mac",
+ "new_debug_unreachable",
+]
+
 [[package]]
 name = "futures"
 version = "0.3.5"
@@ -1539,6 +1564,20 @@ dependencies = [
  "winapi 0.3.9",
 ]
 
+[[package]]
+name = "html5ever"
+version = "0.25.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "aafcf38a1a36118242d29b92e1b08ef84e67e4a5ed06e0a80be20e6a32bfed6b"
+dependencies = [
+ "log",
+ "mac",
+ "markup5ever",
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "http"
 version = "0.2.1"
@@ -1835,6 +1874,7 @@ dependencies = [
  "actix-rt",
  "actix-web",
  "actix-web-actors",
+ "ammonia",
  "anyhow",
  "async-trait",
  "awc",
@@ -2026,12 +2066,47 @@ version = "0.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7d947cbb889ed21c2a84be6ffbaebf5b4e0f4340638cba0444907e38b56be084"
 
+[[package]]
+name = "mac"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c41e0c4fef86961ac6d6f8a82609f55f31b05e4fce149ac5710e439df7619ba4"
+
 [[package]]
 name = "maplit"
 version = "1.0.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3e2e65a1a2e43cfcb47a895c4c8b10d1f4a61097f9f254f183aee60cad9c651d"
 
+[[package]]
+name = "markup5ever"
+version = "0.10.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "aae38d669396ca9b707bfc3db254bc382ddb94f57cc5c235f34623a669a01dab"
+dependencies = [
+ "log",
+ "phf",
+ "phf_codegen",
+ "serde 1.0.116",
+ "serde_derive",
+ "serde_json",
+ "string_cache",
+ "string_cache_codegen",
+ "tendril",
+]
+
+[[package]]
+name = "markup5ever_rcdom"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f015da43bcd8d4f144559a3423f4591d69b8ce0652c905374da7205df336ae2b"
+dependencies = [
+ "html5ever",
+ "markup5ever",
+ "tendril",
+ "xml5ever",
+]
+
 [[package]]
 name = "match_cfg"
 version = "0.1.0"
@@ -2183,6 +2258,12 @@ dependencies = [
  "winapi 0.3.9",
 ]
 
+[[package]]
+name = "new_debug_unreachable"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e4a24736216ec316047a1fc4252e27dabb04218aa4a3f37c6e7ddbf1f9782b54"
+
 [[package]]
 name = "nom"
 version = "4.2.3"
@@ -2417,6 +2498,44 @@ dependencies = [
  "sha-1 0.8.2",
 ]
 
+[[package]]
+name = "phf"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3dfb61232e34fcb633f43d12c58f83c1df82962dcdfa565a4e866ffc17dafe12"
+dependencies = [
+ "phf_shared",
+]
+
+[[package]]
+name = "phf_codegen"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cbffee61585b0411840d3ece935cce9cb6321f01c45477d30066498cd5e1a815"
+dependencies = [
+ "phf_generator",
+ "phf_shared",
+]
+
+[[package]]
+name = "phf_generator"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "17367f0cc86f2d25802b2c26ee58a7b23faeccf78a396094c13dced0d0182526"
+dependencies = [
+ "phf_shared",
+ "rand 0.7.3",
+]
+
+[[package]]
+name = "phf_shared"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c00cf8b9eafe68dde5e9eaa2cef8ee84a9336a47d566ec55ca16589633b65af7"
+dependencies = [
+ "siphasher",
+]
+
 [[package]]
 name = "pin-project"
 version = "0.4.23"
@@ -2482,6 +2601,12 @@ dependencies = [
  "vcpkg",
 ]
 
+[[package]]
+name = "precomputed-hash"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "925383efa346730478fb4838dbe9137d2a47675ad789c546d150a6e1dd4ab31c"
+
 [[package]]
 name = "proc-macro-hack"
 version = "0.5.18"
@@ -2576,7 +2701,7 @@ dependencies = [
  "rand_isaac",
  "rand_jitter",
  "rand_os",
- "rand_pcg",
+ "rand_pcg 0.1.2",
  "rand_xorshift",
  "winapi 0.3.9",
 ]
@@ -2592,6 +2717,7 @@ dependencies = [
  "rand_chacha 0.2.2",
  "rand_core 0.5.1",
  "rand_hc 0.2.0",
+ "rand_pcg 0.2.1",
 ]
 
 [[package]]
@@ -2700,6 +2826,15 @@ dependencies = [
  "rand_core 0.4.2",
 ]
 
+[[package]]
+name = "rand_pcg"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "16abd0c1b639e9eb4d7c50c0b8100b0d0f849be2349829c740fe8e6eb4816429"
+dependencies = [
+ "rand_core 0.5.1",
+]
+
 [[package]]
 name = "rand_xorshift"
 version = "0.1.1"
@@ -3095,6 +3230,12 @@ dependencies = [
  "num-traits 0.2.12",
 ]
 
+[[package]]
+name = "siphasher"
+version = "0.3.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fa8f3741c7372e75519bd9346068370c9cdaabcc1f9599cbcf2a2719352286b7"
+
 [[package]]
 name = "slab"
 version = "0.4.2"
@@ -3189,6 +3330,31 @@ version = "0.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "213701ba3370744dcd1a12960caa4843b3d68b4d1c0a5d575e0d65b2ee9d16c0"
 
+[[package]]
+name = "string_cache"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2940c75beb4e3bf3a494cef919a747a2cb81e52571e212bfbd185074add7208a"
+dependencies = [
+ "lazy_static",
+ "new_debug_unreachable",
+ "phf_shared",
+ "precomputed-hash",
+ "serde 1.0.116",
+]
+
+[[package]]
+name = "string_cache_codegen"
+version = "0.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f24c8e5e19d22a726626f1a5e16fe15b132dcf21d10177fa5a45ce7962996b97"
+dependencies = [
+ "phf_generator",
+ "phf_shared",
+ "proc-macro2",
+ "quote",
+]
+
 [[package]]
 name = "strsim"
 version = "0.8.0"
@@ -3244,6 +3410,17 @@ dependencies = [
  "winapi 0.3.9",
 ]
 
+[[package]]
+name = "tendril"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "707feda9f2582d5d680d733e38755547a3e8fb471e7ba11452ecfd9ce93a5d3b"
+dependencies = [
+ "futf",
+ "mac",
+ "utf-8",
+]
+
 [[package]]
 name = "termcolor"
 version = "1.1.0"
@@ -3576,6 +3753,12 @@ dependencies = [
  "serde 1.0.116",
 ]
 
+[[package]]
+name = "utf-8"
+version = "0.7.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "05e42f7c18b8f902290b009cde6d651262f956c98bc51bca4cd1d511c9cd85c7"
+
 [[package]]
 name = "uuid"
 version = "0.7.4"
@@ -3842,3 +4025,15 @@ name = "xdg"
 version = "2.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d089681aa106a86fade1b0128fb5daf07d5867a509ab036d99988dec80429a57"
+
+[[package]]
+name = "xml5ever"
+version = "0.16.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b1b52e6e8614d4a58b8e70cf51ec0cc21b256ad8206708bcff8139b5bbd6a59"
+dependencies = [
+ "log",
+ "mac",
+ "markup5ever",
+ "time 0.1.44",
+]

Cargo.toml

@@ -59,6 +59,7 @@ anyhow = "1.0"
 thiserror = "1.0"
 background-jobs = " 0.8"
 reqwest = { version = "0.10", features = ["json"] }
+ammonia = "3.1.0"
 
 [dev-dependencies.cargo-husky]
 version = "1"

src/apub/comment.rs

@@ -38,6 +38,7 @@ use activitystreams::{
   public,
 };
 use actix_web::{body::Body, web, web::Path, HttpResponse};
+use ammonia::clean_text;
 use anyhow::Context;
 use itertools::Itertools;
 use lemmy_db::{
@@ -183,7 +184,7 @@ impl FromApub for CommentForm {
       creator_id: creator.id,
       post_id: post.id,
       parent_id,
-      content: content_slurs_removed,
+      content: clean_text(&content_slurs_removed),
       removed: None,
       read: None,
       published: note.published().map(|u| u.to_owned().naive_local()),

src/apub/community.rs

@@ -36,6 +36,7 @@ use activitystreams::{
 };
 use activitystreams_ext::Ext2;
 use actix_web::{body::Body, web, HttpResponse};
+use ammonia::clean_text;
 use anyhow::Context;
 use itertools::Itertools;
 use lemmy_db::{
@@ -376,7 +377,7 @@ impl FromApub for CommunityForm {
       .content()
       .map(|s| s.as_single_xsd_string())
       .flatten()
-      .map(|s| s.to_string());
+      .map(|s| clean_text(s));
     check_slurs(&name)?;
     check_slurs(&title)?;
     check_slurs_opt(&description)?;
@@ -408,8 +409,8 @@ impl FromApub for CommunityForm {
     };
 
     Ok(CommunityForm {
-      name,
+      name: clean_text(&name),
-      title,
+      title: clean_text(&title),
       description,
       category_id: group.ext_one.category.identifier.parse::<i32>()?,
       creator_id: creator.id,

src/apub/post.rs

@@ -34,6 +34,7 @@ use activitystreams::{
 };
 use activitystreams_ext::Ext1;
 use actix_web::{body::Body, web, HttpResponse};
+use ammonia::{clean, clean_text};
 use anyhow::Context;
 use lemmy_db::{
   community::Community,
@@ -168,17 +169,17 @@ fn extract_embed_from_apub(
         .flatten()
         .map(|s| s.as_xsd_string())
         .flatten()
-        .map(|s| s.to_string());
+        .map(|s| clean_text(s));
       let description = preview_page
         .summary()
         .map(|s| s.as_single_xsd_string())
         .flatten()
-        .map(|s| s.to_string());
+        .map(|s| clean_text(s));
       let html = preview_page
         .content()
         .map(|c| c.as_single_xsd_string())
         .flatten()
-        .map(|s| s.to_string());
+        .map(|s| clean(s));
       Ok(EmbedType {
         title,
         description,
@@ -262,11 +263,11 @@ impl FromApub for PostForm {
       .as_ref()
       .map(|c| c.as_single_xsd_string())
       .flatten()
-      .map(|s| s.to_string());
+      .map(|s| clean_text(s));
     check_slurs(&name)?;
     let body_slurs_removed = body.map(|b| remove_slurs(&b));
     Ok(PostForm {
-      name,
+      name: clean_text(&name),
       url,
       body: body_slurs_removed,
       creator_id: creator.id,

src/apub/private_message.rs

@@ -26,6 +26,7 @@ use activitystreams::{
   object::{kind::NoteType, Note, Tombstone},
   prelude::*,
 };
+use ammonia::clean_text;
 use anyhow::Context;
 use lemmy_db::{
   private_message::{PrivateMessage, PrivateMessageForm},
@@ -96,16 +97,17 @@ impl FromApub for PrivateMessageForm {
     let recipient = get_or_fetch_and_upsert_user(&recipient_actor_id, context).await?;
     let ap_id = note.id_unchecked().context(location_info!())?.to_string();
     check_is_apub_id_valid(&Url::parse(&ap_id)?)?;
+    let content = note
+      .content()
+      .context(location_info!())?
+      .as_single_xsd_string()
+      .context(location_info!())?
+      .to_string();
 
     Ok(PrivateMessageForm {
       creator_id: creator.id,
       recipient_id: recipient.id,
-      content: note
+      content: clean_text(&content),
-        .content()
-        .context(location_info!())?
-        .as_single_xsd_string()
-        .context(location_info!())?
-        .to_string(),
       published: note.published().map(|u| u.to_owned().naive_local()),
       updated: note.updated().map(|u| u.to_owned().naive_local()),
       deleted: None,

src/apub/user.rs

@@ -26,6 +26,7 @@ use activitystreams::{
 };
 use activitystreams_ext::Ext1;
 use actix_web::{body::Body, web, HttpResponse};
+use ammonia::clean_text;
 use anyhow::Context;
 use lemmy_db::{
   naive_now,
@@ -242,19 +243,19 @@ impl FromApub for UserForm {
       .as_xsd_string()
       .context(location_info!())?
       .to_string();
-    let preferred_username = person.inner.preferred_username().map(|u| u.to_string());
+    let preferred_username = person.inner.preferred_username().map(|u| clean_text(u));
     let bio = person
       .inner
       .summary()
       .map(|s| s.as_single_xsd_string())
       .flatten()
-      .map(|s| s.to_string());
+      .map(|s| clean_text(s));
     check_slurs(&name)?;
     check_slurs_opt(&preferred_username)?;
     check_slurs_opt(&bio)?;
 
     Ok(UserForm {
-      name,
+      name: clean_text(&name),
       preferred_username,
       password_encrypted: "".to_string(),
       admin: false,