Distinguish between html & text in user & community descriptions

9 files changed, 150 insertions, 36 deletions

diff --git a/Cargo.lock b/Cargo.lock
index 7d72fd9..3fa161f 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -220,6 +220,15 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "476e9cd489f9e121e02ffa6014a8ef220ecb15c05ed23fc34cca13925dc283fb"
 
 [[package]]
+name = "buf-min"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fa17aa1cf56bdd6bb30518767d00e58019d326f3f05d8c3e0730b549d332ea83"
+dependencies = [
+ "bytes",
+]
+
+[[package]]
 name = "bumpalo"
 version = "3.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -291,7 +300,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "19b076e143e1d9538dde65da30f8481c2a6c44040edb8e02b9bf1351edb92ce3"
 dependencies = [
  "lazy_static",
- "nom",
+ "nom 5.1.2",
  "serde",
 ]
 
@@ -469,7 +478,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2d2f06b9cac1506ece98fe3231e3cc9c4410ec3d5b1f24ae1c8946f0742cdefc"
 dependencies = [
  "backtrace",
- "version_check",
+ "version_check 0.9.2",
 ]
 
 [[package]]
@@ -713,7 +722,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ac746a5f3bbfdadd6106868134545e684693d54d9d44f6e9588a7d54af0bf980"
 dependencies = [
  "typenum",
- "version_check",
+ "version_check 0.9.2",
 ]
 
 [[package]]
@@ -1063,7 +1072,7 @@ dependencies = [
  "idna",
  "mime",
  "native-tls",
- "nom",
+ "nom 5.1.2",
  "once_cell",
  "quoted_printable",
  "r2d2",
@@ -1176,6 +1185,7 @@ dependencies = [
  "unic-langid",
  "url",
  "uuid",
+ "v_htmlescape",
 ]
 
 [[package]]
@@ -1397,13 +1407,23 @@ checksum = "e4a24736216ec316047a1fc4252e27dabb04218aa4a3f37c6e7ddbf1f9782b54"
 
 [[package]]
 name = "nom"
+version = "4.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2ad2a91a8e869eeb30b9cb3119ae87773a8f4ae617f41b1eb9c154b2905f7bd6"
+dependencies = [
+ "memchr",
+ "version_check 0.1.5",
+]
+
+[[package]]
+name = "nom"
 version = "5.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ffb4262d26ed83a1c0a33a38fe2bb15797329c85770da05e6b828ddb782627af"
 dependencies = [
  "lexical-core",
  "memchr",
- "version_check",
+ "version_check 0.9.2",
 ]
 
 [[package]]
@@ -2417,7 +2437,7 @@ version = "2.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "50f37be617794602aabbeee0be4f259dc1778fabe05e2d67ee8f79326d5cb4f6"
 dependencies = [
- "version_check",
+ "version_check 0.9.2",
 ]
 
 [[package]]
@@ -2478,6 +2498,38 @@ dependencies = [
 ]
 
 [[package]]
+name = "v_escape"
+version = "0.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f3e0ab5fab1db278a9413d2ea794cb66f471f898c5b020c3c394f6447625d9d4"
+dependencies = [
+ "buf-min",
+ "v_escape_derive",
+]
+
+[[package]]
+name = "v_escape_derive"
+version = "0.8.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c860ad1273f4eee7006cee05db20c9e60e5d24cba024a32e1094aa8e574f3668"
+dependencies = [
+ "nom 4.2.3",
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "v_htmlescape"
+version = "0.12.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1f9a8af610ad6f7fc9989c9d2590d9764bc61f294884e9ee93baa58795174572"
+dependencies = [
+ "cfg-if 1.0.0",
+ "v_escape",
+]
+
+[[package]]
 name = "vcpkg"
 version = "0.2.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -2485,6 +2537,12 @@ checksum = "6454029bf181f092ad1b853286f23e2c507d8e8194d01d92da4a55c274a5508c"
 
 [[package]]
 name = "version_check"
+version = "0.1.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "914b1a6776c4c929a602fafd8bc742e06365d4bcbe48c30f9cca5824f70dc9dd"
+
+[[package]]
+name = "version_check"
 version = "0.9.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b5a972e5669d67ba988ce3dc826706fb0a8b01471c088cb0b6110b805cc36aed"
diff --git a/Cargo.toml b/Cargo.toml
index 3d965ec..a2ac39a 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -54,6 +54,7 @@ migrant_lib = { version = "0.30.0", features = ["d-postgres"] }
 pdcm-linkify = "0.1.0"
 log = "0.4"
 env_logger = "0.8"
+v_htmlescape = "0.12.0"
 
 [dev-dependencies]
 rand = "0.7.3"
diff --git a/migrations/20201231175007_description-html/down.sql b/migrations/20201231175007_description-html/down.sql
new file mode 100644
index 0000000..d24bd35
--- /dev/null
+++ b/migrations/20201231175007_description-html/down.sql
@@ -0,0 +1,4 @@
+BEGIN;
+	ALTER TABLE person DROP COLUMN description_html;
+	ALTER TABLE community DROP COLUMN description_html;
+COMMIT;
diff --git a/migrations/20201231175007_description-html/up.sql b/migrations/20201231175007_description-html/up.sql
new file mode 100644
index 0000000..44e9d07
--- /dev/null
+++ b/migrations/20201231175007_description-html/up.sql
@@ -0,0 +1,4 @@
+BEGIN;
+	ALTER TABLE person ADD COLUMN description_html TEXT;
+	ALTER TABLE community ADD COLUMN description_html TEXT;
+COMMIT;
diff --git a/src/apub_util/ingest.rs b/src/apub_util/ingest.rs
index e85cdfd..95cb9c6 100644
--- a/src/apub_util/ingest.rs
+++ b/src/apub_util/ingest.rs
@@ -215,10 +215,9 @@ pub async fn ingest_object(
                         .and_then(|maybe| maybe.iter().filter_map(|x| x.as_xsd_string()).next())
                 })
                 .unwrap_or("");
-            let description = group
+            let description_html = group
                 .summary()
-                .and_then(|maybe| maybe.iter().filter_map(|x| x.as_xsd_string()).next())
-                .unwrap_or("");
+                .and_then(|maybe| maybe.iter().filter_map(|x| x.as_xsd_string()).next());
             let inbox = group.inbox_unchecked().as_str();
             let shared_inbox = group
                 .endpoints_unchecked()
@@ -236,8 +235,8 @@ pub async fn ingest_object(
                 .and_then(|key| key.signature_algorithm.as_deref());
 
             let id = CommunityLocalID(db.query_one(
-                "INSERT INTO community (name, local, ap_id, ap_inbox, ap_shared_inbox, public_key, public_key_sigalg, description) VALUES ($1, FALSE, $2, $3, $4, $5, $6, $7) ON CONFLICT (ap_id) DO UPDATE SET ap_inbox=$3, ap_shared_inbox=$4, public_key=$5, public_key_sigalg=$6, description=$7 RETURNING id",
-                &[&name, &ap_id.as_str(), &inbox, &shared_inbox, &public_key, &public_key_sigalg, &description],
+                "INSERT INTO community (name, local, ap_id, ap_inbox, ap_shared_inbox, public_key, public_key_sigalg, description_html) VALUES ($1, FALSE, $2, $3, $4, $5, $6, $7) ON CONFLICT (ap_id) DO UPDATE SET ap_inbox=$3, ap_shared_inbox=$4, public_key=$5, public_key_sigalg=$6, description_html=$7 RETURNING id",
+                &[&name, &ap_id.as_str(), &inbox, &shared_inbox, &public_key, &public_key_sigalg, &description_html],
             ).await?.get(0));
 
             Ok(Some(IngestResult::Actor(
@@ -291,10 +290,9 @@ pub async fn ingest_object(
                 .public_key
                 .as_ref()
                 .and_then(|key| key.signature_algorithm.as_deref());
-            let description = person
+            let description_html = person
                 .summary()
-                .and_then(|maybe| maybe.iter().filter_map(|x| x.as_xsd_string()).next())
-                .unwrap_or("");
+                .and_then(|maybe| maybe.iter().filter_map(|x| x.as_xsd_string()).next());
 
             let avatar = person.icon().and_then(|icon| {
                 icon.iter()
@@ -316,8 +314,8 @@ pub async fn ingest_object(
                 .map(|x| x.as_str());
 
             let id = UserLocalID(db.query_one(
-                "INSERT INTO person (username, local, created_local, ap_id, ap_inbox, ap_shared_inbox, public_key, public_key_sigalg, description, avatar) VALUES ($1, FALSE, localtimestamp, $2, $3, $4, $5, $6, $7, $8) ON CONFLICT (ap_id) DO UPDATE SET ap_inbox=$3, ap_shared_inbox=$4, public_key=$5, public_key_sigalg=$6, description=$7, avatar=$8 RETURNING id",
-                &[&username, &ap_id.as_str(), &inbox, &shared_inbox, &public_key, &public_key_sigalg, &description, &avatar],
+                "INSERT INTO person (username, local, created_local, ap_id, ap_inbox, ap_shared_inbox, public_key, public_key_sigalg, description_html, avatar) VALUES ($1, FALSE, localtimestamp, $2, $3, $4, $5, $6, $7, $8) ON CONFLICT (ap_id) DO UPDATE SET ap_inbox=$3, ap_shared_inbox=$4, public_key=$5, public_key_sigalg=$6, description_html=$7, avatar=$8 RETURNING id",
+                &[&username, &ap_id.as_str(), &inbox, &shared_inbox, &public_key, &public_key_sigalg, &description_html, &avatar],
             ).await?.get(0));
 
             Ok(Some(IngestResult::Actor(super::ActorLocalInfo::User {
diff --git a/src/routes/api/communities.rs b/src/routes/api/communities.rs
index 4392983..d3f79c8 100644
--- a/src/routes/api/communities.rs
+++ b/src/routes/api/communities.rs
@@ -13,6 +13,8 @@ struct RespCommunityInfo<'a> {
     base: RespMinimalCommunityInfo<'a>,
 
     description: &'a str,
+    description_html: Option<String>,
+    description_text: Option<&'a str>,
 
     #[serde(skip_serializing_if = "Option::is_none")]
     you_are_moderator: Option<bool>,
@@ -25,6 +27,20 @@ struct RespYourFollowInfo {
     accepted: bool,
 }
 
+fn get_community_description_fields<'a>(
+    description_text: &'a str,
+    description_html: Option<&'a str>,
+) -> (&'a str, Option<&'a str>, Option<String>) {
+    match description_html {
+        Some(description_html) => (
+            description_html,
+            None,
+            Some(ammonia::clean(description_html)),
+        ),
+        None => (description_text, Some(description_text), None),
+    }
+}
+
 async fn route_unstable_communities_list(
     _: (),
     ctx: Arc<crate::RouteContext>,
@@ -43,7 +59,7 @@ async fn route_unstable_communities_list(
 
     let query: CommunitiesListQuery = serde_urlencoded::from_str(req.uri().query().unwrap_or(""))?;
 
-    let mut sql = String::from("SELECT id, name, local, ap_id, description");
+    let mut sql = String::from("SELECT id, name, local, ap_id, description, description_html");
     let mut values: Vec<&(dyn tokio_postgres::types::ToSql + Sync)> = Vec::new();
 
     let db = ctx.db_pool.get().await?;
@@ -77,7 +93,9 @@ async fn route_unstable_communities_list(
             let name = row.get(1);
             let local = row.get(2);
             let ap_id = row.get(3);
-            let description = row.get(4);
+
+            let (description, description_text, description_html) =
+                get_community_description_fields(row.get(4), row.get(5));
 
             let host = crate::get_actor_host_or_unknown(local, ap_id, &ctx.local_hostname);
 
@@ -91,13 +109,16 @@ async fn route_unstable_communities_list(
                 },
 
                 description,
+                description_html,
+                description_text,
+
                 you_are_moderator: if query.include_your {
-                    Some(row.get(6))
+                    Some(row.get(7))
                 } else {
                     None
                 },
                 your_follow: if query.include_your {
-                    Some(match row.get(5) {
+                    Some(match row.get(6) {
                         Some(accepted) => Some(RespYourFollowInfo { accepted }),
                         None => None,
                     })
@@ -204,12 +225,12 @@ async fn route_unstable_communities_get(
         (if query.include_your {
             let user = crate::require_login(&req, &db).await?;
             db.query_opt(
-                "SELECT name, local, ap_id, description, (SELECT accepted FROM community_follow WHERE community=community.id AND follower=$2), EXISTS(SELECT 1 FROM community_moderator WHERE community=community.id AND person=$2) FROM community WHERE id=$1",
+                "SELECT name, local, ap_id, description, description_html, (SELECT accepted FROM community_follow WHERE community=community.id AND follower=$2), EXISTS(SELECT 1 FROM community_moderator WHERE community=community.id AND person=$2) FROM community WHERE id=$1",
                 &[&community_id.raw(), &user.raw()],
             ).await?
         } else {
             db.query_opt(
-                "SELECT name, local, ap_id, description FROM community WHERE id=$1",
+                "SELECT name, local, ap_id, description, description_html FROM community WHERE id=$1",
                 &[&community_id.raw()],
             ).await?
         })
@@ -224,6 +245,9 @@ async fn route_unstable_communities_get(
     let community_local = row.get(1);
     let community_ap_id: Option<&str> = row.get(2);
 
+    let (description, description_text, description_html) =
+        get_community_description_fields(row.get(3), row.get(4));
+
     let info = RespCommunityInfo {
         base: RespMinimalCommunityInfo {
             id: community_id,
@@ -239,14 +263,16 @@ async fn route_unstable_communities_get(
             },
             remote_url: community_ap_id,
         },
-        description: row.get(3),
+        description,
+        description_html,
+        description_text,
         you_are_moderator: if query.include_your {
-            Some(row.get(5))
+            Some(row.get(6))
         } else {
             None
         },
         your_follow: if query.include_your {
-            Some(match row.get(4) {
+            Some(match row.get(5) {
                 Some(accepted) => Some(RespYourFollowInfo { accepted }),
                 None => None,
             })
diff --git a/src/routes/api/users.rs b/src/routes/api/users.rs
index ad1d77d..efe25da 100644
--- a/src/routes/api/users.rs
+++ b/src/routes/api/users.rs
@@ -143,6 +143,8 @@ struct RespUserInfo<'a> {
     base: RespMinimalAuthorInfo<'a>,
 
     description: &'a str,
+    description_html: Option<&'a str>,
+    description_text: Option<&'a str>,
     #[serde(skip_serializing_if = "Option::is_none")]
     suspended: Option<bool>,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -252,7 +254,8 @@ async fn route_unstable_users_patch(
 
     #[derive(Deserialize)]
     struct UsersEditBody<'a> {
-        description: Option<Cow<'a, str>>,
+        #[serde(alias = "description")]
+        description_text: Option<Cow<'a, str>>,
         email_address: Option<Cow<'a, str>>,
         password: Option<String>,
         avatar: Option<Cow<'a, str>>,
@@ -266,7 +269,7 @@ async fn route_unstable_users_patch(
 
     let mut changes = Vec::<(&str, &(dyn tokio_postgres::types::ToSql + Sync))>::new();
 
-    if let Some(description) = body.description.as_ref() {
+    if let Some(description) = body.description_text.as_ref() {
         changes.push(("description", description));
     }
     if let Some(email_address) = body.email_address.as_ref() {
@@ -534,7 +537,7 @@ async fn route_unstable_users_get(
 
     let row = db
         .query_opt(
-            "SELECT username, local, ap_id, description, avatar, suspended FROM person WHERE id=$1",
+            "SELECT username, local, ap_id, description, description_html, avatar, suspended FROM person WHERE id=$1",
             &[&user_id],
         )
         .await?;
@@ -548,7 +551,7 @@ async fn route_unstable_users_get(
 
     let local = row.get(1);
     let ap_id = row.get(2);
-    let avatar: Option<&str> = row.get(4);
+    let avatar: Option<&str> = row.get(5);
 
     let info = RespMinimalAuthorInfo {
         id: user_id,
@@ -561,10 +564,24 @@ async fn route_unstable_users_get(
         }),
     };
 
+    let (description, description_text, description_html) = match row.get(4) {
+        Some(description_html) => (
+            description_html,
+            None,
+            Some(ammonia::clean(description_html)),
+        ),
+        None => {
+            let description_text: &str = row.get(3);
+            (description_text, Some(description_text), None)
+        }
+    };
+
     let info = RespUserInfo {
         base: info,
-        description: row.get(3),
-        suspended: if local { Some(row.get(5)) } else { None },
+        description,
+        description_html: description_html.as_deref(),
+        description_text,
+        suspended: if local { Some(row.get(6)) } else { None },
         your_note,
     };
 
diff --git a/src/routes/apub/communities.rs b/src/routes/apub/communities.rs
index dd249f7..f9fe71b 100644
--- a/src/routes/apub/communities.rs
+++ b/src/routes/apub/communities.rs
@@ -88,7 +88,7 @@ async fn handler_communities_get(
 
     match db
         .query_opt(
-            "SELECT name, local, public_key, description FROM community WHERE id=$1",
+            "SELECT name, local, public_key, description, description_html FROM community WHERE id=$1",
             &[&community_id],
         )
         .await?
@@ -116,7 +116,10 @@ async fn handler_communities_get(
                             None
                         }
                     });
-            let description: &str = row.get(3);
+            let description = match row.get(4) {
+                Some(description_html) => description_html,
+                None => v_htmlescape::escape(row.get(3)).to_string(),
+            };
 
             let community_ap_id =
                 crate::apub_util::get_local_community_apub_id(community_id, &ctx.host_url_apub);
diff --git a/src/routes/apub/mod.rs b/src/routes/apub/mod.rs
index d6a9583..b352cd4 100644
--- a/src/routes/apub/mod.rs
+++ b/src/routes/apub/mod.rs
@@ -118,7 +118,7 @@ async fn handler_users_get(
 
     match db
         .query_opt(
-            "SELECT username, local, public_key, description, avatar FROM person WHERE id=$1",
+            "SELECT username, local, public_key, description, description_html, avatar FROM person WHERE id=$1",
             &[&user_id.raw()],
         )
         .await?
@@ -148,9 +148,12 @@ async fn handler_users_get(
                         }
                     });
 
-            let description: &str = row.get(3);
+            let description = match row.get(4) {
+                Some(description_html) => description_html,
+                None => v_htmlescape::escape(row.get(3)).to_string(),
+            };
 
-            let avatar: Option<&str> = row.get(4);
+            let avatar: Option<&str> = row.get(5);
 
             let user_ap_id =
                 crate::apub_util::get_local_person_apub_id(user_id, &ctx.host_url_apub);