From b136d913cc51f5bd0d899954c78170abbe48a70f Mon Sep 17 00:00:00 2001
From: rob-maron <132852777+rob-maron@users.noreply.github.com>
Date: Thu, 19 Mar 2026 14:13:15 -0400
Subject: [PATCH 1/2] trim whitespace + control chars

---
 shared/data-provider/src/hub.rs | 31 +++++++++++++++++++++----------
 1 file changed, 21 insertions(+), 10 deletions(-)

diff --git a/shared/data-provider/src/hub.rs b/shared/data-provider/src/hub.rs
index 13a575b84..7206fb58e 100644
--- a/shared/data-provider/src/hub.rs
+++ b/shared/data-provider/src/hub.rs
@@ -1,11 +1,11 @@
 use crate::errors::UploadError;
 use crate::hub::model::HubRepo;
 use hf_hub::{
-    Cache, Repo, RepoType,
     api::{
-        Siblings,
         tokio::{ApiError, UploadSource},
+        Siblings,
     },
+    Cache, Repo, RepoType,
 };
 use psyche_coordinator::model;
 use psyche_core::FixedString;
@@ -16,6 +16,13 @@ use tracing::{error, info};
 const MODEL_EXTENSIONS: [&str; 3] = [".safetensors", ".json", ".py"];
 const DATASET_EXTENSIONS: [&str; 1] = [".parquet"];
 
+/// Strip leading/trailing whitespace and control characters from a repo identifier.
+/// TODO: Remove once https://github.com/PsycheFoundation/nousnet/pull/636 is merged
+fn sanitize_repo_id(raw: &str) -> String {
+    raw.trim_matches(|c: char| c.is_whitespace() || c.is_control())
+        .to_string()
+}
+
 fn check_extensions(sibling: &Siblings, extensions: &[&'static str]) -> bool {
     match extensions.is_empty() {
         true => true,
@@ -90,10 +97,11 @@ pub async fn download_model_repo_async(
     max_concurrent_downloads: Option<usize>,
     progress_bar: bool,
 ) -> Result<Vec<PathBuf>, ApiError> {
+    let repo_id = sanitize_repo_id(repo_id);
     download_repo_async(
         match revision {
-            Some(revision) => Repo::with_revision(repo_id.to_string(), RepoType::Model, revision),
-            None => Repo::model(repo_id.to_string()),
+            Some(revision) => Repo::with_revision(repo_id.clone(), RepoType::Model, revision),
+            None => Repo::model(repo_id),
         },
         cache,
         token,
@@ -112,10 +120,11 @@ pub async fn download_dataset_repo_async(
     max_concurrent_downloads: Option<usize>,
     progress_bar: bool,
 ) -> Result<Vec<PathBuf>, ApiError> {
+    let repo_id = sanitize_repo_id(&repo_id);
     download_repo_async(
         match revision {
-            Some(revision) => Repo::with_revision(repo_id.to_owned(), RepoType::Dataset, revision),
-            None => Repo::new(repo_id.to_owned(), RepoType::Dataset),
+            Some(revision) => Repo::with_revision(repo_id.clone(), RepoType::Dataset, revision),
+            None => Repo::new(repo_id, RepoType::Dataset),
         },
         cache,
         token,
@@ -162,10 +171,11 @@ pub fn download_model_repo_sync(
     token: Option<String>,
     progress_bar: bool,
 ) -> Result<Vec<PathBuf>, hf_hub::api::sync::ApiError> {
+    let repo_id = sanitize_repo_id(repo_id);
     download_repo_sync(
         match revision {
-            Some(revision) => Repo::with_revision(repo_id.to_owned(), RepoType::Model, revision),
-            None => Repo::model(repo_id.to_owned()),
+            Some(revision) => Repo::with_revision(repo_id.clone(), RepoType::Model, revision),
+            None => Repo::model(repo_id),
         },
         cache,
         token,
@@ -181,10 +191,11 @@ pub fn download_dataset_repo_sync(
     token: Option<String>,
     progress_bar: bool,
 ) -> Result<Vec<PathBuf>, hf_hub::api::sync::ApiError> {
+    let repo_id = sanitize_repo_id(repo_id);
     download_repo_sync(
         match revision {
-            Some(revision) => Repo::with_revision(repo_id.to_owned(), RepoType::Dataset, revision),
-            None => Repo::new(repo_id.to_owned(), RepoType::Dataset),
+            Some(revision) => Repo::with_revision(repo_id.clone(), RepoType::Dataset, revision),
+            None => Repo::new(repo_id, RepoType::Dataset),
         },
         cache,
         token,

From 7fc91ce2a6c0d27214a18abda2b020d9c79fceaf Mon Sep 17 00:00:00 2001
From: rob-maron <132852777+rob-maron@users.noreply.github.com>
Date: Thu, 19 Mar 2026 14:15:56 -0400
Subject: [PATCH 2/2] fmt

---
 shared/data-provider/src/hub.rs | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/shared/data-provider/src/hub.rs b/shared/data-provider/src/hub.rs
index 7206fb58e..5b44a01d3 100644
--- a/shared/data-provider/src/hub.rs
+++ b/shared/data-provider/src/hub.rs
@@ -1,11 +1,11 @@
 use crate::errors::UploadError;
 use crate::hub::model::HubRepo;
 use hf_hub::{
+    Cache, Repo, RepoType,
     api::{
-        tokio::{ApiError, UploadSource},
         Siblings,
+        tokio::{ApiError, UploadSource},
     },
-    Cache, Repo, RepoType,
 };
 use psyche_coordinator::model;
 use psyche_core::FixedString;