Phase 1: storage + catalog layer

- storaged: object_store backend (LocalFileSystem), PUT/GET/DELETE/LIST endpoints
- shared: arrow_helpers with Parquet roundtrip + schema fingerprinting (2 tests)
- catalogd: in-memory registry with write-ahead manifest persistence to object storage
- catalogd: POST/GET /datasets, GET /datasets/by-name/{name}
- gateway: wires storaged + catalogd with shared object_store state
- Phase tracker updated: Phase 0 + Phase 1 gates passed

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/storaged/src/backend.rs

@@ -0,0 +1,10 @@
+use object_store::ObjectStore;
+use object_store::local::LocalFileSystem;
+use std::sync::Arc;
+
+/// Initialize the object store backend.
+/// Starts with LocalFileSystem. Swap to S3/RustFS by changing this function.
+pub fn init_local(root: &str) -> Arc<dyn ObjectStore> {
+    std::fs::create_dir_all(root).expect("failed to create storage root");
+    Arc::new(LocalFileSystem::new_with_prefix(root).expect("failed to init local object store"))
+}

crates/storaged/src/lib.rs

@@ -1 +1,3 @@
+pub mod backend;
+pub mod ops;
 pub mod service;

crates/storaged/src/ops.rs

@@ -0,0 +1,30 @@
+use bytes::Bytes;
+use futures::TryStreamExt;
+use object_store::{ObjectStore, path::Path};
+use std::sync::Arc;
+
+pub async fn put(store: &Arc<dyn ObjectStore>, key: &str, data: Bytes) -> Result<(), String> {
+    let path = Path::from(key);
+    store.put(&path, data.into()).await.map_err(|e| e.to_string())?;
+    Ok(())
+}
+
+pub async fn get(store: &Arc<dyn ObjectStore>, key: &str) -> Result<Bytes, String> {
+    let path = Path::from(key);
+    let result = store.get(&path).await.map_err(|e| e.to_string())?;
+    let bytes = result.bytes().await.map_err(|e| e.to_string())?;
+    Ok(bytes)
+}
+
+pub async fn delete(store: &Arc<dyn ObjectStore>, key: &str) -> Result<(), String> {
+    let path = Path::from(key);
+    store.delete(&path).await.map_err(|e| e.to_string())?;
+    Ok(())
+}
+
+pub async fn list(store: &Arc<dyn ObjectStore>, prefix: Option<&str>) -> Result<Vec<String>, String> {
+    let prefix_path = prefix.map(Path::from);
+    let stream = store.list(prefix_path.as_ref());
+    let items: Vec<_> = stream.try_collect().await.map_err(|e| e.to_string())?;
+    Ok(items.into_iter().map(|m| m.location.to_string()).collect())
+}

crates/storaged/src/service.rs

@@ -1,9 +1,75 @@
-use axum::{Router, routing::get};
+use axum::{
+    Router,
+    extract::{Path, Query, State},
+    http::StatusCode,
+    response::IntoResponse,
+    routing::{delete, get, put},
+};
+use bytes::Bytes;
+use object_store::ObjectStore;
+use serde::Deserialize;
+use std::sync::Arc;
 
-pub fn router() -> Router {
-    Router::new().route("/health", get(health))
+use crate::ops;
+
+pub type StorageState = Arc<dyn ObjectStore>;
+
+pub fn router(store: Arc<dyn ObjectStore>) -> Router {
+    Router::new()
+        .route("/health", get(health))
+        .route("/objects/{*key}", put(put_object))
+        .route("/objects/{*key}", get(get_object))
+        .route("/objects/{*key}", delete(delete_object))
+        .route("/objects", get(list_objects))
+        .with_state(store)
 }
 
 async fn health() -> &'static str {
     "storaged ok"
 }
+
+async fn put_object(
+    State(store): State<StorageState>,
+    Path(key): Path<String>,
+    body: Bytes,
+) -> impl IntoResponse {
+    match ops::put(&store, &key, body).await {
+        Ok(()) => (StatusCode::CREATED, format!("stored: {key}")),
+        Err(e) => (StatusCode::INTERNAL_SERVER_ERROR, e),
+    }
+}
+
+async fn get_object(
+    State(store): State<StorageState>,
+    Path(key): Path<String>,
+) -> impl IntoResponse {
+    match ops::get(&store, &key).await {
+        Ok(data) => Ok(data),
+        Err(e) => Err((StatusCode::NOT_FOUND, e)),
+    }
+}
+
+async fn delete_object(
+    State(store): State<StorageState>,
+    Path(key): Path<String>,
+) -> impl IntoResponse {
+    match ops::delete(&store, &key).await {
+        Ok(()) => (StatusCode::OK, format!("deleted: {key}")),
+        Err(e) => (StatusCode::INTERNAL_SERVER_ERROR, e),
+    }
+}
+
+#[derive(Deserialize)]
+struct ListQuery {
+    prefix: Option<String>,
+}
+
+async fn list_objects(
+    State(store): State<StorageState>,
+    Query(q): Query<ListQuery>,
+) -> impl IntoResponse {
+    match ops::list(&store, q.prefix.as_deref()).await {
+        Ok(keys) => Ok(axum::Json(keys)),
+        Err(e) => Err((StatusCode::INTERNAL_SERVER_ERROR, e)),
+    }
+}