From d9973a13dd2386083be86bcc55b146753720b00d Mon Sep 17 00:00:00 2001
From: dal <dallin@buster.so>
Date: Wed, 5 Feb 2025 11:58:43 -0700
Subject: [PATCH] bugfix(datasets): add delete dataset route

Implement a new DELETE route for removing datasets by their ID
---
 .../rest/routes/datasets/delete_dataset.rs    | 81 +++++++++++++++++++
 api/src/routes/rest/routes/datasets/mod.rs    |  4 +-
 2 files changed, 84 insertions(+), 1 deletion(-)
 create mode 100644 api/src/routes/rest/routes/datasets/delete_dataset.rs

diff --git a/api/src/routes/rest/routes/datasets/delete_dataset.rs b/api/src/routes/rest/routes/datasets/delete_dataset.rs
new file mode 100644
index 000000000..2d9bb1a56
--- /dev/null
+++ b/api/src/routes/rest/routes/datasets/delete_dataset.rs
@@ -0,0 +1,81 @@
+use anyhow::{anyhow, Result};
+use axum::{extract::Path, http::StatusCode, Extension};
+use chrono::Utc;
+use diesel::{update, ExpressionMethods, QueryDsl};
+use diesel_async::RunQueryDsl;
+use serde_json::Value;
+use uuid::Uuid;
+
+use crate::database::{lib::get_pg_pool, models::User, schema::datasets};
+
+pub async fn delete_dataset(
+    Extension(user): Extension<User>,
+    Path(dataset_id): Path<Uuid>,
+) -> Result<StatusCode, (StatusCode, String)> {
+    match delete_dataset_handler(&user, dataset_id).await {
+        Ok(_) => Ok(StatusCode::NO_CONTENT),
+        Err(e) => {
+            tracing::error!("Error deleting dataset: {}", e);
+            Err((StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))
+        }
+    }
+}
+
+async fn delete_dataset_handler(user: &User, dataset_id: Uuid) -> Result<()> {
+    let mut conn = get_pg_pool()
+        .get()
+        .await
+        .map_err(|e| anyhow!("Unable to get connection from pool: {}", e))?;
+
+    // Get dataset's organization_id
+    let dataset = datasets::table
+        .select(datasets::organization_id)
+        .filter(datasets::id.eq(dataset_id))
+        .filter(datasets::deleted_at.is_null())
+        .first::<Uuid>(&mut conn)
+        .await
+        .map_err(|e| match e {
+            diesel::result::Error::NotFound => anyhow!("Dataset not found"),
+            _ => anyhow!("Error getting dataset: {}", e),
+        })?;
+
+    // Check user's organization and role
+    let user_org_id = user
+        .attributes
+        .get("organization_id")
+        .and_then(|v| match v {
+            Value::String(s) => Some(s.as_str()),
+            _ => None,
+        })
+        .ok_or_else(|| anyhow!("User organization id not found"))?;
+
+    let user_org_id =
+        Uuid::parse_str(user_org_id).map_err(|_| anyhow!("Invalid organization id format"))?;
+
+    if user_org_id != dataset {
+        return Err(anyhow!("User does not belong to dataset's organization"));
+    }
+
+    let user_role = user
+        .attributes
+        .get("organization_role")
+        .and_then(|v| match v {
+            Value::String(s) => Some(s.as_str()),
+            _ => None,
+        })
+        .ok_or_else(|| anyhow!("User role not found"))?;
+
+    if !["workspace_admin", "data_admin"].contains(&user_role) {
+        return Err(anyhow!("User does not have required permissions"));
+    }
+
+    // Soft delete the dataset
+    update(datasets::table)
+        .filter(datasets::id.eq(dataset_id))
+        .set(datasets::deleted_at.eq(Some(Utc::now())))
+        .execute(&mut conn)
+        .await
+        .map_err(|e| anyhow!("Error updating dataset: {}", e))?;
+
+    Ok(())
+}
diff --git a/api/src/routes/rest/routes/datasets/mod.rs b/api/src/routes/rest/routes/datasets/mod.rs
index 6bc84f0e9..7bc000077 100644
--- a/api/src/routes/rest/routes/datasets/mod.rs
+++ b/api/src/routes/rest/routes/datasets/mod.rs
@@ -1,4 +1,5 @@
 mod assets;
+mod delete_dataset;
 mod deploy_datasets;
 mod get_dataset;
 mod get_dataset_data_sample;
@@ -6,7 +7,7 @@ mod list_datasets;
 mod post_dataset;
 
 use axum::{
-    routing::{get, post},
+    routing::{get, post, delete},
     Router,
 };
 
@@ -16,6 +17,7 @@ pub fn router() -> Router {
         .route("/", post(post_dataset::post_dataset))
         .route("/deploy", post(deploy_datasets::deploy_datasets))
         .route("/:dataset_id", get(get_dataset::get_dataset))
+        .route("/:dataset_id", delete(delete_dataset::delete_dataset))
         .route(
             "/:dataset_id/data/sample",
             get(get_dataset_data_sample::get_dataset_data_sample),