worker: add a job to check for potential typosquatting

This only fires when new crates are published: updates to existing
crates will not cause this job to run.

On a technical level, the major impact here is that the background
worker will keep an in memory cache of the top 3000 crates and their
owners. I don't expect the impact of this to be significant in practice.

As this is an experiment at present, configuration is hardcoded into the
new worker job module. If this becomes a longer term thing, this would
be split out into our normal configuration system for easier management.

Cargo.toml

@@ -95,6 +95,7 @@ tower = "=0.4.13"
 tower-http = { version = "=0.4.4", features = ["fs", "catch-panic"] }
 tracing = "=0.1.37"
 tracing-subscriber = { version = "=0.3.17", features = ["env-filter"] }
+typomania = { version = "=0.1.0", default-features = false }
 url = "=2.4.1"
 
 [dev-dependencies]

src/background_jobs.rs

@@ -77,6 +77,7 @@ macro_rules! job_variant_from_value {
 
 jobs! {
     pub enum Job {
+        CheckTyposquat(CheckTyposquatJob),
         DailyDbMaintenance,
         DumpDb(DumpDbJob),
         NormalizeIndex(NormalizeIndexJob),
@@ -166,6 +167,12 @@ impl Job {
         Ok(())
     }
 
+    pub fn check_typosquat<T: ToString>(krate: T) -> Self {
+        Self::CheckTyposquat(CheckTyposquatJob {
+            krate: krate.to_string(),
+        })
+    }
+
     pub fn daily_db_maintenance() -> Self {
         Self::DailyDbMaintenance
     }
@@ -250,6 +257,7 @@ impl Job {
             .as_ref()
             .expect("Application should configure a background runner environment");
         match self {
+            Job::CheckTyposquat(args) => worker::check_typosquat(env, conn, &args.krate),
             Job::DailyDbMaintenance => {
                 worker::perform_daily_db_maintenance(&mut *fresh_connection(pool)?)
             }
@@ -298,6 +306,11 @@ pub struct AddCrateJob {
     pub(super) krate: crates_io_index::Crate,
 }
 
+#[derive(Serialize, Deserialize)]
+pub struct CheckTyposquatJob {
+    pub(super) krate: String,
+}
+
 #[derive(Serialize, Deserialize)]
 pub struct UpdateCrateIndexJob {
     pub(super) crate_name: String,

src/controllers/krate/publish.rs

@@ -365,6 +365,11 @@ pub async fn publish(app: AppState, req: BytesRequest) -> AppResult<Json<GoodCra
 
             Job::enqueue_sync_to_index(&krate.name, conn)?;
 
+            // Experiment: check new crates for potential typosquatting.
+            if existing_crate.is_none() {
+                Job::check_typosquat(&krate.name).enqueue(conn)?;
+            }
+
             // The `other` field on `PublishWarnings` was introduced to handle a temporary warning
             // that is no longer needed. As such, crates.io currently does not return any `other`
             // warnings at this time, but if we need to, the field is available.

src/email.rs

@@ -91,6 +91,32 @@ or go to https://{domain}/me/pending-invites to manage all of your crate ownersh
         self.send(email, subject, &body)
     }
 
+    /// Attempts to send a notification that a new crate may be typosquatting another crate.
+    pub fn send_possible_typosquat_notification(
+        &self,
+        email: &str,
+        crate_name: &str,
+        squats: &[typomania::checks::Squat],
+    ) -> AppResult<()> {
+        let subject = "Possible typosquatting in new crate";
+        let body = format!(
+            "New crate {crate_name} may be typosquatting one or more other crates.\n
+Visit https://{domain}/crates/{crate_name} to see the offending crate.\n
+\n
+Specific squat checks that triggered:\n
+\n
+- {squats}\n",
+            domain = crate::config::domain_name(),
+            squats = squats
+                .iter()
+                .map(|squat| format!("{squat}"))
+                .collect::<Vec<_>>()
+                .join("\n- "),
+        );
+
+        self.send(email, subject, &body)
+    }
+
     /// Attempts to send an API token exposure notification email
     pub fn send_token_exposed_notification(
         &self,

src/worker/mod.rs

@@ -9,6 +9,7 @@ pub mod dump_db;
 pub mod fastly;
 mod git;
 mod readmes;
+mod typosquat;
 mod update_downloads;
 
 pub(crate) use daily_db_maintenance::perform_daily_db_maintenance;
@@ -17,4 +18,5 @@ pub(crate) use git::{
     perform_index_squash, perform_normalize_index, sync_to_git_index, sync_to_sparse_index,
 };
 pub(crate) use readmes::perform_render_and_upload_readme;
+pub(crate) use typosquat::check_typosquat;
 pub(crate) use update_downloads::perform_update_downloads;

src/worker/typosquat.rs

@@ -0,0 +1,99 @@
+use diesel::PgConnection;
+use typomania::Package;
+
+use crate::{background_jobs::Environment, swirl::PerformError, Emails};
+
+use self::types::OwnedCrate;
+
+mod cache;
+mod config;
+mod types;
+
+#[cfg(test)]
+mod test_util;
+
+#[instrument(skip_all, fields(krate.name = ?name))]
+pub fn check_typosquat(
+    env: &Environment,
+    conn: &mut PgConnection,
+    name: &str,
+) -> Result<(), PerformError> {
+    check_typosquat_inner(env.emails(), conn, name)
+}
+
+fn check_typosquat_inner(
+    emails: &Emails,
+    conn: &mut PgConnection,
+    name: &str,
+) -> Result<(), PerformError> {
+    info!("Checking new crate for potential typosquatting");
+
+    let krate: Box<dyn Package> = Box::new(OwnedCrate::from_name(conn, name)?);
+    let squats = cache::get_harness(conn)?.check_package(name, krate)?;
+    if !squats.is_empty() {
+        // Well, well, well. For now, the only action we'll take is to e-mail people who hopefully
+        // care to check into things more closely.
+        info!(?squats, "Found potential typosquatting");
+
+        for email in config::NOTIFY_EMAILS.iter() {
+            if let Err(e) = emails.send_possible_typosquat_notification(email, name, &squats) {
+                error!(?e, ?email, "sending possible typosquat notification");
+            }
+        }
+    }
+
+    Ok(())
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::test_util::pg_connection;
+
+    use super::test_util::Faker;
+    use super::*;
+
+    #[test]
+    fn integration() -> Result<(), PerformError> {
+        let emails = Emails::new_in_memory();
+        let mut faker = Faker::new(pg_connection());
+
+        // Set up a user and a crate to match against.
+        let user = faker.user("a")?;
+        faker.crate_and_version("my-crate", "It's awesome", &user, 100)?;
+
+        // Prime the cache so it only includes the crate we just created.
+        //
+        // Note that there's theoretical flakiness here if the test takes longer to run than the
+        // cache TTL. Of course, since the cache TTL is currently set to 12 hours, that would
+        // probably indicate bigger problems.
+        let _harness = super::cache::get_harness(faker.borrow_conn())?;
+
+        // Now we'll create new crates: one problematic, one not so.
+        let other_user = faker.user("b")?;
+        let (angel, _version) = faker.crate_and_version(
+            "innocent-crate",
+            "I'm just a simple, innocent crate",
+            &other_user,
+            0,
+        )?;
+        let (demon, _version) = faker.crate_and_version(
+            "mycrate",
+            "I'm even more innocent, obviously",
+            &other_user,
+            0,
+        )?;
+
+        // OK, we're done faking stuff.
+        let mut conn = faker.into_conn();
+
+        // Run the check with a crate that shouldn't cause problems.
+        check_typosquat_inner(&emails, &mut conn, &angel.name)?;
+        assert!(emails.mails_in_memory().unwrap().is_empty());
+
+        // Now run the check with a less innocent crate.
+        check_typosquat_inner(&emails, &mut conn, &demon.name)?;
+        assert!(!emails.mails_in_memory().unwrap().is_empty());
+
+        Ok(())
+    }
+}

src/worker/typosquat/cache.rs

@@ -0,0 +1,100 @@
+use std::{
+    sync::{Arc, Mutex},
+    time::Instant,
+};
+
+use diesel::PgConnection;
+use typomania::{
+    checks::{Bitflips, Omitted, SwappedWords, Typos},
+    Harness,
+};
+
+use crate::swirl::PerformError;
+
+use super::{config, types::TopCrates};
+
+/// Gets the typomania harness for the cached top crates, regenerating it if it is out of date.
+pub(super) fn get_harness(
+    conn: &mut PgConnection,
+) -> Result<Arc<Harness<TopCrates>>, PerformError> {
+    HARNESS_CACHE.get(conn)
+}
+
+static HARNESS_CACHE: Cache = Cache::new();
+
+struct Cache(Mutex<Inner>);
+
+impl Cache {
+    const fn new() -> Self {
+        Self(Mutex::new(Inner::new()))
+    }
+
+    fn get(&self, conn: &mut PgConnection) -> Result<Arc<Harness<TopCrates>>, PerformError> {
+        let mut inner = self.0.lock().unwrap();
+
+        Ok(if let Some(harness) = inner.get() {
+            harness
+        } else {
+            let harness = Arc::new(new_harness(conn)?);
+
+            inner.update(harness.clone());
+            harness
+        })
+    }
+}
+
+#[instrument(skip_all)]
+fn new_harness(conn: &mut PgConnection) -> Result<Harness<TopCrates>, PerformError> {
+    debug!("Rebuilding top crate cache");
+    let start = Instant::now();
+
+    let top_crates = TopCrates::new(conn, config::TOP_CRATES)?;
+
+    // This is essentially the standard set of checks that was implemented by typogard-crates.
+    let harness = Harness::builder()
+        .with_check(Bitflips::new(
+            config::CRATE_NAME_ALPHABET,
+            top_crates.iter_names(),
+        ))
+        .with_check(Omitted::new(config::CRATE_NAME_ALPHABET))
+        .with_check(SwappedWords::new("-_"))
+        .with_check(Typos::new(config::TYPOS.iter().map(|(c, typos)| {
+            (*c, typos.iter().map(|ss| ss.to_string()).collect())
+        })))
+        .build(top_crates);
+
+    let elapsed = Instant::now() - start;
+    debug!(?elapsed, "Top crate cache rebuilt");
+
+    Ok(harness)
+}
+
+struct Inner {
+    harness: Option<Arc<Harness<TopCrates>>>,
+    last_update: Option<Instant>,
+}
+
+impl Inner {
+    const fn new() -> Self {
+        Self {
+            harness: None,
+            last_update: None,
+        }
+    }
+
+    fn get(&self) -> Option<Arc<Harness<TopCrates>>> {
+        if let Some(harness) = &self.harness {
+            if let Some(when) = self.last_update {
+                if when >= Instant::now() - config::CACHE_TTL {
+                    return Some(harness.clone());
+                }
+            }
+        }
+        None
+    }
+
+    fn update(&mut self, harness: Arc<Harness<TopCrates>>) {
+        self.harness = Some(harness);
+        self.last_update = Some(Instant::now());
+    }
+}