From fea10f6d408ae8f438357cf7fc545140e092c29a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sebasti=C3=A1n=20Galkin?= <code@amisdelabc.com>
Date: Wed, 10 Dec 2025 18:35:42 -0300
Subject: [PATCH] Port: More fixes for virtual chunk url escapes (#1474)

---
 icechunk/src/virtual_chunks.rs      | 16 ++++++++--------
 icechunk/tests/test_virtual_refs.rs | 22 ++++++++++++++++++++++
 2 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/icechunk/src/virtual_chunks.rs b/icechunk/src/virtual_chunks.rs
index 27e2f412f..6a591cac9 100644
--- a/icechunk/src/virtual_chunks.rs
+++ b/icechunk/src/virtual_chunks.rs
@@ -295,11 +295,11 @@ impl VirtualChunkResolver {
         &self,
         chunk_location: &Url,
     ) -> Result<Arc<dyn ChunkFetcher>, VirtualReferenceError> {
-        let cont = self
-            .matching_container(chunk_location.to_string().as_str())
-            .ok_or_else(|| {
-                VirtualReferenceErrorKind::NoContainerForUrl(chunk_location.to_string())
-            })?;
+        let location = chunk_location.to_string();
+        let location = urlencoding::decode(location.as_str())?;
+        let cont = self.matching_container(location.as_ref()).ok_or_else(|| {
+            VirtualReferenceErrorKind::NoContainerForUrl(chunk_location.to_string())
+        })?;
 
         let cache_key = fetcher_cache_key(cont, chunk_location)?;
         // TODO: we shouldn't need to clone the container name
@@ -416,7 +416,7 @@ impl VirtualChunkResolver {
                 };
 
                 let bucket_name = if let Some(host) = chunk_location.host_str() {
-                    host.to_string()
+                    urlencoding::decode(host)?.into_owned()
                 } else {
                     Err(VirtualReferenceErrorKind::CannotParseBucketName(
                         "No bucket name found".to_string(),
@@ -475,7 +475,7 @@ impl VirtualChunkResolver {
                 };
 
                 let container = if let Some(host) = chunk_location.host_str() {
-                    host.to_string()
+                    urlencoding::decode(host)?.into_owned()
                 } else {
                     Err(VirtualReferenceErrorKind::CannotParseBucketName(
                         "No bucket name found".to_string(),
@@ -565,7 +565,7 @@ impl ChunkFetcher for S3Fetcher {
         checksum: Option<&Checksum>,
     ) -> Result<Box<dyn Buf + Unpin + Send>, VirtualReferenceError> {
         let bucket_name = if let Some(host) = chunk_location.host_str() {
-            host.to_string()
+            urlencoding::decode(host)?.into_owned()
         } else {
             Err(VirtualReferenceErrorKind::CannotParseBucketName(
                 "No bucket name found".to_string(),
diff --git a/icechunk/tests/test_virtual_refs.rs b/icechunk/tests/test_virtual_refs.rs
index 6db877cd0..e0c6c3d9e 100644
--- a/icechunk/tests/test_virtual_refs.rs
+++ b/icechunk/tests/test_virtual_refs.rs
@@ -210,6 +210,19 @@ async fn create_minio_repository() -> Repository {
             }),
         )
         .unwrap(),
+        VirtualChunkContainer::new(
+            "s3://testbucket/path with spaces/".to_string(),
+            ObjectStoreConfig::S3Compatible(S3Options {
+                region: Some(String::from("us-east-1")),
+                endpoint_url: Some("http://localhost:9000".to_string()),
+                anonymous: false,
+                allow_http: true,
+                force_path_style: true,
+                network_stream_timeout_seconds: None,
+                requester_pays: false,
+            }),
+        )
+        .unwrap(),
         VirtualChunkContainer::new(
             "az://testcontainer/".to_string(),
             ObjectStoreConfig::Azure(HashMap::from([
@@ -230,6 +243,15 @@ async fn create_minio_repository() -> Repository {
                 expires_after: None,
             }))),
         ),
+        (
+            "s3://testbucket/path with spaces/".to_string(),
+            Some(Credentials::S3(S3Credentials::Static(S3StaticCredentials {
+                access_key_id: "minio123".to_string(),
+                secret_access_key: "minio123".to_string(),
+                session_token: None,
+                expires_after: None,
+            }))),
+        ),
         (
             "az://testcontainer/".to_string(),
             Some(Credentials::Azure(AzureCredentials::FromEnv)),