feat: allow passing azure client/tenant ID through remote SDK (#3102)

Prior to this commit we supported passing the azure storage account name
to the lancedb remote SDK through headers. This adds support for client
ID and tenant ID as well.

rust/lancedb/src/connection.rs

@@ -566,8 +566,11 @@ pub struct ConnectBuilder {
 }
 
 #[cfg(feature = "remote")]
-const ENV_VARS_TO_STORAGE_OPTS: [(&str, &str); 1] =
-    [("AZURE_STORAGE_ACCOUNT_NAME", "azure_storage_account_name")];
+const ENV_VARS_TO_STORAGE_OPTS: [(&str, &str); 3] = [
+    ("AZURE_STORAGE_ACCOUNT_NAME", "azure_storage_account_name"),
+    ("AZURE_CLIENT_ID", "azure_client_id"),
+    ("AZURE_TENANT_ID", "azure_tenant_id"),
+];
 
 impl ConnectBuilder {
     /// Create a new [`ConnectOptions`] with the given database URI.

rust/lancedb/src/remote/client.rs

@@ -446,13 +446,23 @@ impl<S: HttpSend> RestfulLanceDbClient<S> {
                 })?,
             );
         }
-        if let Some(v) = options.0.get("azure_storage_account_name") {
-            headers.insert(
-                HeaderName::from_static("x-azure-storage-account-name"),
-                HeaderValue::from_str(v).map_err(|_| Error::InvalidInput {
-                    message: format!("non-ascii storage account name '{}' provided", db_name),
-                })?,
-            );
+        // Map azure storage options to x-azure-* headers.
+        // The option key uses underscores (e.g. "azure_client_id") while the
+        // header uses hyphens (e.g. "x-azure-client-id").
+        let azure_opts: [(&str, &str); 3] = [
+            ("azure_storage_account_name", "x-azure-storage-account-name"),
+            ("azure_client_id", "x-azure-client-id"),
+            ("azure_tenant_id", "x-azure-tenant-id"),
+        ];
+        for (opt_key, header_name) in azure_opts {
+            if let Some(v) = options.0.get(opt_key) {
+                headers.insert(
+                    HeaderName::from_static(header_name),
+                    HeaderValue::from_str(v).map_err(|_| Error::InvalidInput {
+                        message: format!("non-ascii value '{}' for option '{}'", v, opt_key),
+                    })?,
+                );
+            }
         }
 
         for (key, value) in &config.extra_headers {
@@ -1075,4 +1085,34 @@ mod tests {
             _ => panic!("Expected Runtime error"),
         }
     }
+
+    #[test]
+    fn test_default_headers_azure_opts() {
+        let mut opts = HashMap::new();
+        opts.insert(
+            "azure_storage_account_name".to_string(),
+            "myaccount".to_string(),
+        );
+        opts.insert("azure_client_id".to_string(), "my-client-id".to_string());
+        opts.insert("azure_tenant_id".to_string(), "my-tenant-id".to_string());
+        let remote_opts = RemoteOptions::new(opts);
+
+        let headers = RestfulLanceDbClient::<Sender>::default_headers(
+            "test-key",
+            "us-east-1",
+            "testdb",
+            false,
+            &remote_opts,
+            None,
+            &ClientConfig::default(),
+        )
+        .unwrap();
+
+        assert_eq!(
+            headers.get("x-azure-storage-account-name").unwrap(),
+            "myaccount"
+        );
+        assert_eq!(headers.get("x-azure-client-id").unwrap(), "my-client-id");
+        assert_eq!(headers.get("x-azure-tenant-id").unwrap(), "my-tenant-id");
+    }
 }

rust/lancedb/src/remote/db.rs

@@ -777,7 +777,12 @@ impl RemoteOptions {
 
 impl From<StorageOptions> for RemoteOptions {
     fn from(options: StorageOptions) -> Self {
-        let supported_opts = vec!["account_name", "azure_storage_account_name"];
+        let supported_opts = vec![
+            "account_name",
+            "azure_storage_account_name",
+            "azure_client_id",
+            "azure_tenant_id",
+        ];
         let mut filtered = HashMap::new();
         for opt in supported_opts {
             if let Some(v) = options.0.get(opt) {