wip: the beginnings of a storage layer

Cargo.toml

@@ -1,3 +1,3 @@
 [workspace]
 resolver = "2"
-members = ["exp1", "exp2", "offline-web-model"]
+members = ["exp1", "exp2", "offline-web-model", "offline-web-storage"]

offline-web-model/src/lib.rs

@@ -61,9 +61,9 @@ impl Reference {
         let mut cloned = self.clone();
         cloned.dependents.push(dep);
         // We ensure that our dependents are always sorted lexicographically by name.
-        cloned.dependents.sort_by(|left, right| {
-            left.name.cmp(&right.name)
-        });
+        cloned
+            .dependents
+            .sort_by(|left, right| left.name.cmp(&right.name));
         // Recalculate the ID based on dependents, content_address, and path
         let mut hasher = Self::initial_hash(&self.content_address, &self.name);
         for dependent in &cloned.dependents {

offline-web-storage/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "offline-web-storage"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+offline-web-model = { path = "../offline-web-model" }
+sqlx = { version = "0.7", features = ["runtime-tokio-rustls", "sqlite", "uuid", "chrono"] }
+tokio = { version = "1.0", features = ["full"] }
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+anyhow = "1.0"
+uuid = { version = "1.0", features = ["v4"] }
+chrono = { version = "0.4", features = ["serde"] }
+blake2 = "0.10"
+
+[dev-dependencies]

offline-web-storage/src/lib.rs

@@ -0,0 +1,473 @@
+use std::{collections::HashMap, sync::Arc};
+
+use anyhow::Result;
+use blake2::{Blake2b512, Digest};
+use offline_web_model::Reference;
+use sqlx::{Pool, Row, Sqlite, SqlitePool};
+
+pub struct ReferenceStore {
+    pool: Pool<Sqlite>,
+}
+
+impl ReferenceStore {
+    pub async fn new(database_url: &str) -> Result<Self> {
+        let pool = SqlitePool::connect(database_url).await?;
+        let store = Self { pool };
+        store.initialize_schema().await?;
+        Ok(store)
+    }
+
+    async fn initialize_schema(&self) -> Result<()> {
+        sqlx::query(
+            r#"
+            CREATE TABLE IF NOT EXISTS refs (
+                id TEXT PRIMARY KEY,
+                content_address TEXT,
+                name TEXT NOT NULL,
+                created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+                updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
+            )
+            "#,
+        )
+        .execute(&self.pool)
+        .await?;
+
+        sqlx::query(
+            r#"
+            CREATE TABLE IF NOT EXISTS reference_dependencies (
+                parent_id TEXT NOT NULL,
+                dependent_id TEXT NOT NULL,
+                PRIMARY KEY (parent_id, dependent_id)
+            )
+            "#,
+        )
+        .execute(&self.pool)
+        .await?;
+
+        sqlx::query("CREATE INDEX IF NOT EXISTS idx_refs_name ON refs(name)")
+            .execute(&self.pool)
+            .await?;
+
+        sqlx::query("CREATE INDEX IF NOT EXISTS idx_refs_content_address ON refs(content_address)")
+            .execute(&self.pool)
+            .await?;
+
+        sqlx::query("CREATE INDEX IF NOT EXISTS idx_dependencies_parent ON reference_dependencies(parent_id)")
+            .execute(&self.pool)
+            .await?;
+
+        sqlx::query("CREATE INDEX IF NOT EXISTS idx_dependencies_dependent ON reference_dependencies(dependent_id)")
+            .execute(&self.pool)
+            .await?;
+
+        sqlx::query(
+            r#"
+            CREATE TABLE IF NOT EXISTS content_objects (
+                content_address TEXT PRIMARY KEY,
+                content_data BLOB NOT NULL,
+                content_type TEXT,
+                created_at DATETIME DEFAULT CURRENT_TIMESTAMP
+            )
+            "#,
+        )
+        .execute(&self.pool)
+        .await?;
+
+        sqlx::query(
+            "CREATE INDEX IF NOT EXISTS idx_content_created ON content_objects(created_at)",
+        )
+        .execute(&self.pool)
+        .await?;
+
+        Ok(())
+    }
+
+    pub async fn store_reference(&self, reference: &Reference) -> Result<()> {
+        let mut tx = self.pool.begin().await?;
+
+        // Insert or update the reference
+        sqlx::query(
+            r#"
+            INSERT OR REPLACE INTO refs (id, content_address, name, updated_at)
+            VALUES (?, ?, ?, CURRENT_TIMESTAMP)
+            "#,
+        )
+        .bind(&reference.id)
+        .bind(&reference.content_address)
+        .bind(&reference.name)
+        .execute(&mut *tx)
+        .await?;
+
+        // Clear existing dependencies for this reference
+        sqlx::query("DELETE FROM reference_dependencies WHERE parent_id = ?")
+            .bind(&reference.id)
+            .execute(&mut *tx)
+            .await?;
+
+        // Insert new dependencies
+        for dependent in &reference.dependents {
+            sqlx::query(
+                "INSERT INTO reference_dependencies (parent_id, dependent_id) VALUES (?, ?)",
+            )
+            .bind(&reference.id)
+            .bind(&dependent.id)
+            .execute(&mut *tx)
+            .await?;
+        }
+
+        tx.commit().await?;
+        Ok(())
+    }
+
+    pub async fn get_reference(&self, id: &str) -> Result<Option<Reference>> {
+        // Get the reference record
+        let row = sqlx::query("SELECT id, content_address, name FROM refs WHERE id = ?")
+            .bind(id)
+            .fetch_optional(&self.pool)
+            .await?;
+
+        let Some(row) = row else {
+            return Ok(None);
+        };
+
+        let reference_id: String = row.get("id");
+        let content_address: Option<String> = row.get("content_address");
+        let name: String = row.get("name");
+
+        // Get dependencies
+        let dependents = self.get_dependents(&reference_id).await?;
+
+        Ok(Some(Reference {
+            id: reference_id,
+            content_address,
+            name,
+            dependents,
+        }))
+    }
+
+    fn get_dependents<'a>(
+        &'a self,
+        parent_id: &'a str,
+    ) -> std::pin::Pin<Box<dyn std::future::Future<Output = Result<Vec<Arc<Reference>>>> + Send + 'a>>
+    {
+        Box::pin(async move {
+            let rows = sqlx::query(
+                r#"
+                SELECT r.id, r.content_address, r.name
+                FROM refs r
+                JOIN reference_dependencies rd ON r.id = rd.dependent_id
+                WHERE rd.parent_id = ?
+                ORDER BY r.name
+                "#,
+            )
+            .bind(parent_id)
+            .fetch_all(&self.pool)
+            .await?;
+
+            let mut dependents = Vec::new();
+            for row in rows {
+                let id: String = row.get("id");
+                let content_address: Option<String> = row.get("content_address");
+                let name: String = row.get("name");
+
+                // Recursively get dependents for each dependent
+                let nested_dependents = self.get_dependents(&id).await?;
+
+                dependents.push(Arc::new(Reference {
+                    id,
+                    content_address,
+                    name,
+                    dependents: nested_dependents,
+                }));
+            }
+
+            Ok(dependents)
+        })
+    }
+
+    pub async fn get_references_by_name(&self, name: &str) -> Result<Vec<Reference>> {
+        let rows = sqlx::query("SELECT id FROM refs WHERE name = ?")
+            .bind(name)
+            .fetch_all(&self.pool)
+            .await?;
+
+        let mut references = Vec::new();
+        for row in rows {
+            let id: String = row.get("id");
+            if let Some(reference) = self.get_reference(&id).await? {
+                references.push(reference);
+            }
+        }
+
+        Ok(references)
+    }
+
+    pub async fn get_references_by_content_address(
+        &self,
+        content_address: &str,
+    ) -> Result<Vec<Reference>> {
+        let rows = sqlx::query("SELECT id FROM refs WHERE content_address = ?")
+            .bind(content_address)
+            .fetch_all(&self.pool)
+            .await?;
+
+        let mut references = Vec::new();
+        for row in rows {
+            let id: String = row.get("id");
+            if let Some(reference) = self.get_reference(&id).await? {
+                references.push(reference);
+            }
+        }
+
+        Ok(references)
+    }
+
+    pub async fn delete_reference(&self, id: &str) -> Result<bool> {
+        let result = sqlx::query("DELETE FROM refs WHERE id = ?")
+            .bind(id)
+            .execute(&self.pool)
+            .await?;
+
+        Ok(result.rows_affected() > 0)
+    }
+
+    pub async fn list_all_references(&self) -> Result<Vec<Reference>> {
+        let rows = sqlx::query("SELECT id FROM refs ORDER BY name")
+            .fetch_all(&self.pool)
+            .await?;
+
+        let mut references = Vec::new();
+        for row in rows {
+            let id: String = row.get("id");
+            if let Some(reference) = self.get_reference(&id).await? {
+                references.push(reference);
+            }
+        }
+
+        Ok(references)
+    }
+
+    pub async fn update_reference_graph(
+        &self,
+        updated_references: &HashMap<String, Arc<Reference>>,
+    ) -> Result<()> {
+        let mut tx = self.pool.begin().await?;
+
+        for (_, reference) in updated_references {
+            // Update the reference
+            sqlx::query(
+                r#"
+                INSERT OR REPLACE INTO refs (id, content_address, name, updated_at)
+                VALUES (?, ?, ?, CURRENT_TIMESTAMP)
+                "#,
+            )
+            .bind(&reference.id)
+            .bind(&reference.content_address)
+            .bind(&reference.name)
+            .execute(&mut *tx)
+            .await?;
+
+            // Clear existing dependencies
+            sqlx::query("DELETE FROM reference_dependencies WHERE parent_id = ?")
+                .bind(&reference.id)
+                .execute(&mut *tx)
+                .await?;
+
+            // Insert new dependencies
+            for dependent in &reference.dependents {
+                sqlx::query(
+                    "INSERT INTO reference_dependencies (parent_id, dependent_id) VALUES (?, ?)",
+                )
+                .bind(&reference.id)
+                .bind(&dependent.id)
+                .execute(&mut *tx)
+                .await?;
+            }
+        }
+
+        tx.commit().await?;
+        Ok(())
+    }
+
+    pub fn calculate_content_address(content: &[u8]) -> String {
+        let mut hasher = Blake2b512::new();
+        hasher.update(content);
+        format!("{:x}", hasher.finalize())
+    }
+
+    pub async fn store_content(
+        &self,
+        content: &[u8],
+        content_type: Option<String>,
+    ) -> Result<String> {
+        let content_address = Self::calculate_content_address(content);
+
+        // Check if content already exists (deduplication)
+        let exists = sqlx::query("SELECT 1 FROM content_objects WHERE content_address = ?")
+            .bind(&content_address)
+            .fetch_optional(&self.pool)
+            .await?
+            .is_some();
+
+        if !exists {
+            sqlx::query(
+                r#"
+                INSERT INTO content_objects (content_address, content_data, content_type)
+                VALUES (?, ?, ?)
+                "#,
+            )
+            .bind(&content_address)
+            .bind(content)
+            .bind(&content_type)
+            .execute(&self.pool)
+            .await?;
+        }
+
+        Ok(content_address)
+    }
+
+    pub async fn get_content(&self, content_address: &str) -> Result<Option<Vec<u8>>> {
+        let row = sqlx::query("SELECT content_data FROM content_objects WHERE content_address = ?")
+            .bind(content_address)
+            .fetch_optional(&self.pool)
+            .await?;
+
+        Ok(row.map(|row| row.get::<Vec<u8>, _>("content_data")))
+    }
+
+    pub async fn get_content_info(&self, content_address: &str) -> Result<Option<ContentInfo>> {
+        let row = sqlx::query(
+            "SELECT content_type, created_at FROM content_objects WHERE content_address = ?",
+        )
+        .bind(content_address)
+        .fetch_optional(&self.pool)
+        .await?;
+
+        Ok(row.map(|row| ContentInfo {
+            content_address: content_address.to_string(),
+            content_type: row.get("content_type"),
+            size: 0, // Size not stored in database
+            created_at: row.get("created_at"),
+        }))
+    }
+
+    pub async fn content_exists(&self, content_address: &str) -> Result<bool> {
+        let exists = sqlx::query("SELECT 1 FROM content_objects WHERE content_address = ?")
+            .bind(content_address)
+            .fetch_optional(&self.pool)
+            .await?
+            .is_some();
+
+        Ok(exists)
+    }
+
+    pub async fn store_reference_with_content(
+        &self,
+        name: String,
+        content: &[u8],
+        content_type: Option<String>,
+    ) -> Result<Reference> {
+        // Store the content and get its address
+        let content_address = self.store_content(content, content_type).await?;
+
+        // Create the reference
+        let reference = Reference::new(Some(content_address), name);
+
+        // Store the reference
+        self.store_reference(&reference).await?;
+
+        Ok(reference)
+    }
+
+    pub async fn get_reference_with_content(
+        &self,
+        id: &str,
+    ) -> Result<Option<(Reference, Option<Vec<u8>>)>> {
+        let reference = self.get_reference(id).await?;
+
+        if let Some(ref reference) = reference {
+            if let Some(ref content_address) = reference.content_address {
+                let content = self.get_content(content_address).await?;
+                return Ok(Some((reference.clone(), content)));
+            }
+        }
+
+        Ok(reference.map(|r| (r, None)))
+    }
+
+    pub async fn delete_content(&self, content_address: &str) -> Result<bool> {
+        let result = sqlx::query("DELETE FROM content_objects WHERE content_address = ?")
+            .bind(content_address)
+            .execute(&self.pool)
+            .await?;
+
+        Ok(result.rows_affected() > 0)
+    }
+
+    pub async fn list_unreferenced_content(&self) -> Result<Vec<String>> {
+        let rows = sqlx::query(
+            r#"
+            SELECT co.content_address
+            FROM content_objects co
+            LEFT JOIN refs r ON co.content_address = r.content_address
+            WHERE r.content_address IS NULL
+            "#,
+        )
+        .fetch_all(&self.pool)
+        .await?;
+
+        Ok(rows
+            .into_iter()
+            .map(|row| row.get("content_address"))
+            .collect())
+    }
+
+    pub async fn cleanup_unreferenced_content(&self) -> Result<usize> {
+        let result = sqlx::query(
+            r#"
+            DELETE FROM content_objects
+            WHERE content_address IN (
+                SELECT co.content_address
+                FROM content_objects co
+                LEFT JOIN refs r ON co.content_address = r.content_address
+                WHERE r.content_address IS NULL
+            )
+            "#,
+        )
+        .execute(&self.pool)
+        .await?;
+
+        Ok(result.rows_affected() as usize)
+    }
+
+    pub async fn get_storage_stats(&self) -> Result<StorageStats> {
+        let content_count = sqlx::query("SELECT COUNT(*) as count FROM content_objects")
+            .fetch_one(&self.pool)
+            .await?;
+
+        let reference_count = sqlx::query("SELECT COUNT(*) as count FROM refs")
+            .fetch_one(&self.pool)
+            .await?;
+
+        Ok(StorageStats {
+            content_object_count: content_count.get::<i64, _>("count") as usize,
+            total_content_size: 0, // Size not tracked
+            reference_count: reference_count.get::<i64, _>("count") as usize,
+        })
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct ContentInfo {
+    pub content_address: String,
+    pub content_type: Option<String>,
+    pub size: usize,
+    pub created_at: String,
+}
+
+#[derive(Debug, Clone)]
+pub struct StorageStats {
+    pub content_object_count: usize,
+    pub total_content_size: usize,
+    pub reference_count: usize,
+}

offline-web-storage/tests/integration_tests.rs

@@ -0,0 +1,431 @@
+use std::{collections::HashMap, sync::Arc};
+
+use offline_web_model::Reference;
+use offline_web_storage::ReferenceStore;
+
+async fn create_test_store() -> ReferenceStore {
+    let store = ReferenceStore::new("sqlite::memory:").await.unwrap();
+    store
+}
+
+#[tokio::test]
+async fn test_store_and_retrieve_reference() {
+    let store = create_test_store().await;
+
+    let reference = Reference::new(Some("abc123".to_string()), "test.txt".to_string());
+
+    // Store the reference
+    store.store_reference(&reference).await.unwrap();
+
+    // Retrieve it
+    let retrieved = store.get_reference(&reference.id).await.unwrap();
+    assert!(retrieved.is_some());
+
+    let retrieved = retrieved.unwrap();
+    assert_eq!(retrieved.id, reference.id);
+    assert_eq!(retrieved.content_address, reference.content_address);
+    assert_eq!(retrieved.name, reference.name);
+    assert_eq!(retrieved.dependents.len(), 0);
+}
+
+#[tokio::test]
+async fn test_store_reference_with_dependents() {
+    let store = create_test_store().await;
+
+    let dep1 = Arc::new(Reference::new(
+        Some("dep1".to_string()),
+        "dep1.txt".to_string(),
+    ));
+    let dep2 = Arc::new(Reference::new(
+        Some("dep2".to_string()),
+        "dep2.txt".to_string(),
+    ));
+
+    // Store dependencies first
+    store.store_reference(&dep1).await.unwrap();
+    store.store_reference(&dep2).await.unwrap();
+
+    let mut parent = Reference::new(Some("parent".to_string()), "parent.txt".to_string());
+    parent = parent.add_dep(dep1.clone());
+    parent = parent.add_dep(dep2.clone());
+
+    // Store parent with dependencies
+    store.store_reference(&parent).await.unwrap();
+
+    // Retrieve and verify
+    let retrieved = store.get_reference(&parent.id).await.unwrap().unwrap();
+    assert_eq!(retrieved.dependents.len(), 2);
+
+    let dep_names: Vec<_> = retrieved.dependents.iter().map(|d| &d.name).collect();
+    assert!(dep_names.contains(&&"dep1.txt".to_string()));
+    assert!(dep_names.contains(&&"dep2.txt".to_string()));
+}
+
+#[tokio::test]
+async fn test_get_references_by_name() {
+    let store = create_test_store().await;
+
+    let ref1 = Reference::new(Some("abc1".to_string()), "test.txt".to_string());
+    let ref2 = Reference::new(Some("abc2".to_string()), "test.txt".to_string());
+    let ref3 = Reference::new(Some("abc3".to_string()), "other.txt".to_string());
+
+    store.store_reference(&ref1).await.unwrap();
+    store.store_reference(&ref2).await.unwrap();
+    store.store_reference(&ref3).await.unwrap();
+
+    let results = store.get_references_by_name("test.txt").await.unwrap();
+    assert_eq!(results.len(), 2);
+
+    let results = store.get_references_by_name("other.txt").await.unwrap();
+    assert_eq!(results.len(), 1);
+
+    let results = store
+        .get_references_by_name("nonexistent.txt")
+        .await
+        .unwrap();
+    assert_eq!(results.len(), 0);
+}
+
+#[tokio::test]
+async fn test_get_references_by_content_address() {
+    let store = create_test_store().await;
+
+    let ref1 = Reference::new(Some("same_content".to_string()), "file1.txt".to_string());
+    let ref2 = Reference::new(Some("same_content".to_string()), "file2.txt".to_string());
+    let ref3 = Reference::new(
+        Some("different_content".to_string()),
+        "file3.txt".to_string(),
+    );
+
+    store.store_reference(&ref1).await.unwrap();
+    store.store_reference(&ref2).await.unwrap();
+    store.store_reference(&ref3).await.unwrap();
+
+    let results = store
+        .get_references_by_content_address("same_content")
+        .await
+        .unwrap();
+    assert_eq!(results.len(), 2);
+
+    let results = store
+        .get_references_by_content_address("different_content")
+        .await
+        .unwrap();
+    assert_eq!(results.len(), 1);
+}
+
+#[tokio::test]
+async fn test_delete_reference() {
+    let store = create_test_store().await;
+
+    let reference = Reference::new(Some("abc123".to_string()), "test.txt".to_string());
+    store.store_reference(&reference).await.unwrap();
+
+    // Verify it exists
+    let retrieved = store.get_reference(&reference.id).await.unwrap();
+    assert!(retrieved.is_some());
+
+    // Delete it
+    let deleted = store.delete_reference(&reference.id).await.unwrap();
+    assert!(deleted);
+
+    // Verify it's gone
+    let retrieved = store.get_reference(&reference.id).await.unwrap();
+    assert!(retrieved.is_none());
+
+    // Try to delete again
+    let deleted = store.delete_reference(&reference.id).await.unwrap();
+    assert!(!deleted);
+}
+
+#[tokio::test]
+async fn test_list_all_references() {
+    let store = create_test_store().await;
+
+    let ref1 = Reference::new(Some("abc1".to_string()), "a.txt".to_string());
+    let ref2 = Reference::new(Some("abc2".to_string()), "b.txt".to_string());
+    let ref3 = Reference::new(Some("abc3".to_string()), "c.txt".to_string());
+
+    store.store_reference(&ref1).await.unwrap();
+    store.store_reference(&ref2).await.unwrap();
+    store.store_reference(&ref3).await.unwrap();
+
+    let all_refs = store.list_all_references().await.unwrap();
+    assert_eq!(all_refs.len(), 3);
+
+    // Should be sorted by name
+    assert_eq!(all_refs[0].name, "a.txt");
+    assert_eq!(all_refs[1].name, "b.txt");
+    assert_eq!(all_refs[2].name, "c.txt");
+}
+
+#[tokio::test]
+async fn test_update_reference_graph() {
+    let store = create_test_store().await;
+
+    let ref1 = Arc::new(Reference::new(
+        Some("abc1".to_string()),
+        "file1.txt".to_string(),
+    ));
+    let ref2 = Arc::new(Reference::new(
+        Some("abc2".to_string()),
+        "file2.txt".to_string(),
+    ));
+
+    let mut updated_refs = HashMap::new();
+    updated_refs.insert(ref1.id.clone(), ref1.clone());
+    updated_refs.insert(ref2.id.clone(), ref2.clone());
+
+    store.update_reference_graph(&updated_refs).await.unwrap();
+
+    // Verify both references were stored
+    let retrieved1 = store.get_reference(&ref1.id).await.unwrap();
+    let retrieved2 = store.get_reference(&ref2.id).await.unwrap();
+
+    assert!(retrieved1.is_some());
+    assert!(retrieved2.is_some());
+    assert_eq!(retrieved1.unwrap().name, "file1.txt");
+    assert_eq!(retrieved2.unwrap().name, "file2.txt");
+}
+
+#[tokio::test]
+async fn test_store_and_retrieve_content() {
+    let store = create_test_store().await;
+
+    let content = b"Hello, world!";
+    let content_type = Some("text/plain".to_string());
+
+    // Store content
+    let content_address = store
+        .store_content(content, content_type.clone())
+        .await
+        .unwrap();
+
+    // Retrieve content
+    let retrieved_content = store.get_content(&content_address).await.unwrap();
+    assert!(retrieved_content.is_some());
+    assert_eq!(retrieved_content.unwrap(), content);
+
+    // Check content info
+    let content_info = store.get_content_info(&content_address).await.unwrap();
+    assert!(content_info.is_some());
+    let info = content_info.unwrap();
+    assert_eq!(info.content_address, content_address);
+    assert_eq!(info.content_type, content_type);
+
+    // Check content exists
+    assert!(store.content_exists(&content_address).await.unwrap());
+}
+
+#[tokio::test]
+async fn test_content_deduplication() {
+    let store = create_test_store().await;
+
+    let content = b"Duplicate content";
+
+    // Store same content twice
+    let addr1 = store.store_content(content, None).await.unwrap();
+    let addr2 = store.store_content(content, None).await.unwrap();
+
+    // Should get same address
+    assert_eq!(addr1, addr2);
+
+    // Should only have one copy in storage
+    let stats = store.get_storage_stats().await.unwrap();
+    assert_eq!(stats.content_object_count, 1);
+}
+
+#[tokio::test]
+async fn test_store_reference_with_content() {
+    let store = create_test_store().await;
+
+    let content = b"File content here";
+    let name = "test_file.txt".to_string();
+    let content_type = Some("text/plain".to_string());
+
+    // Store reference with content
+    let reference = store
+        .store_reference_with_content(name.clone(), content, content_type)
+        .await
+        .unwrap();
+
+    assert_eq!(reference.name, name);
+    assert!(reference.content_address.is_some());
+
+    // Retrieve reference with content
+    let (retrieved_ref, retrieved_content) = store
+        .get_reference_with_content(&reference.id)
+        .await
+        .unwrap()
+        .unwrap();
+
+    assert_eq!(retrieved_ref.id, reference.id);
+    assert_eq!(retrieved_ref.name, name);
+    assert!(retrieved_content.is_some());
+    assert_eq!(retrieved_content.unwrap(), content);
+}
+
+#[tokio::test]
+async fn test_calculate_content_address() {
+    let content1 = b"Same content";
+    let content2 = b"Same content";
+    let content3 = b"Different content";
+
+    let addr1 = ReferenceStore::calculate_content_address(content1);
+    let addr2 = ReferenceStore::calculate_content_address(content2);
+    let addr3 = ReferenceStore::calculate_content_address(content3);
+
+    assert_eq!(addr1, addr2);
+    assert_ne!(addr1, addr3);
+
+    // Should be a valid hex string
+    assert!(addr1.chars().all(|c| c.is_ascii_hexdigit()));
+    assert_eq!(addr1.len(), 128); // Blake2b512 produces 64 bytes = 128 hex chars
+}
+
+#[tokio::test]
+async fn test_delete_content() {
+    let store = create_test_store().await;
+
+    let content = b"Content to delete";
+    let content_address = store.store_content(content, None).await.unwrap();
+
+    // Verify content exists
+    assert!(store.content_exists(&content_address).await.unwrap());
+
+    // Delete content
+    let deleted = store.delete_content(&content_address).await.unwrap();
+    assert!(deleted);
+
+    // Verify content is gone
+    assert!(!store.content_exists(&content_address).await.unwrap());
+    let retrieved = store.get_content(&content_address).await.unwrap();
+    assert!(retrieved.is_none());
+
+    // Try to delete again
+    let deleted_again = store.delete_content(&content_address).await.unwrap();
+    assert!(!deleted_again);
+}
+
+#[tokio::test]
+async fn test_cleanup_unreferenced_content() {
+    let store = create_test_store().await;
+
+    // Store some content with references
+    let content1 = b"Referenced content";
+    let ref1 = store
+        .store_reference_with_content("file1.txt".to_string(), content1, None)
+        .await
+        .unwrap();
+
+    // Store content without references
+    let content2 = b"Unreferenced content 1";
+    let content3 = b"Unreferenced content 2";
+    let _addr2 = store.store_content(content2, None).await.unwrap();
+    let _addr3 = store.store_content(content3, None).await.unwrap();
+
+    // Initial stats
+    let stats = store.get_storage_stats().await.unwrap();
+    assert_eq!(stats.content_object_count, 3);
+    assert_eq!(stats.reference_count, 1);
+
+    // List unreferenced content
+    let unreferenced = store.list_unreferenced_content().await.unwrap();
+    assert_eq!(unreferenced.len(), 2);
+
+    // Cleanup unreferenced content
+    let cleaned_up = store.cleanup_unreferenced_content().await.unwrap();
+    assert_eq!(cleaned_up, 2);
+
+    // Check final stats
+    let final_stats = store.get_storage_stats().await.unwrap();
+    assert_eq!(final_stats.content_object_count, 1);
+    assert_eq!(final_stats.reference_count, 1);
+
+    // Referenced content should still exist
+    let (retrieved_ref, retrieved_content) = store
+        .get_reference_with_content(&ref1.id)
+        .await
+        .unwrap()
+        .unwrap();
+    assert_eq!(retrieved_ref.id, ref1.id);
+    assert_eq!(retrieved_content.unwrap(), content1);
+}
+
+#[tokio::test]
+async fn test_storage_stats() {
+    let store = create_test_store().await;
+
+    // Initial stats should be empty
+    let stats = store.get_storage_stats().await.unwrap();
+    assert_eq!(stats.content_object_count, 0);
+    assert_eq!(stats.total_content_size, 0);
+    assert_eq!(stats.reference_count, 0);
+
+    // Add some content and references
+    let content1 = b"First file";
+    let content2 = b"Second file content";
+
+    let _ref1 = store
+        .store_reference_with_content("file1.txt".to_string(), content1, None)
+        .await
+        .unwrap();
+    let _ref2 = store
+        .store_reference_with_content("file2.txt".to_string(), content2, None)
+        .await
+        .unwrap();
+
+    // Check updated stats
+    let final_stats = store.get_storage_stats().await.unwrap();
+    assert_eq!(final_stats.content_object_count, 2);
+    assert_eq!(final_stats.reference_count, 2);
+}
+
+#[tokio::test]
+async fn test_reference_with_content_and_dependencies() {
+    let store = create_test_store().await;
+
+    // Create dependencies with content
+    let dep1_content = b"Dependency 1 content";
+    let dep2_content = b"Dependency 2 content";
+
+    let dep1 = store
+        .store_reference_with_content("dep1.txt".to_string(), dep1_content, None)
+        .await
+        .unwrap();
+    let dep2 = store
+        .store_reference_with_content("dep2.txt".to_string(), dep2_content, None)
+        .await
+        .unwrap();
+
+    // Create parent with content and dependencies
+    let parent_content = b"Parent content";
+    let parent = store
+        .store_reference_with_content("parent.txt".to_string(), parent_content, None)
+        .await
+        .unwrap();
+
+    // Add dependencies to parent
+    let parent_with_deps = parent.add_dep(Arc::new(dep1)).add_dep(Arc::new(dep2));
+    store.store_reference(&parent_with_deps).await.unwrap();
+
+    // Retrieve parent with content
+    let (retrieved_parent, retrieved_content) = store
+        .get_reference_with_content(&parent_with_deps.id)
+        .await
+        .unwrap()
+        .unwrap();
+
+    assert_eq!(retrieved_parent.dependents.len(), 2);
+    assert_eq!(retrieved_content.unwrap(), parent_content);
+
+    // Check that dependencies also have their content
+    for dep in &retrieved_parent.dependents {
+        let (_, dep_content) = store
+            .get_reference_with_content(&dep.id)
+            .await
+            .unwrap()
+            .unwrap();
+        assert!(dep_content.is_some());
+    }
+}