# SD Download Zipped Artifacts

#### Context

The current [API endpoint](https://github.com/screwdriver-cd/screwdriver/blob/master/plugins/builds/artifacts/get.js) allows downloading artifacts for a single file. This is quite limited, as users often need to download multiple files simultaneously.

#### Proposal

###### Asynchronous Implementation (preferred for a foolproof solution)

```mermaid
sequenceDiagram
  Client->>API: Download dir
  API->>DB: Save request
  API->>Client: Return 201 with a link to get download status
  Zipper->>DB: Consume request
  create participant Store
  Zipper->>Store: Stream files
  destroy Store
  Zipper->>Store: Close connection
  Zipper->>Zipper: zipping
  create participant S3
  Zipper->>S3: upload zipped file
  destroy S3
  S3->>Zipper: send the URL info
  destroy Zipper
  Zipper->>DB: update request with S3 URL
  Client->>API: Download dir status
  destroy DB
  API->>DB: Get download URL based on request ID
  API->>Client: Return 200 with S3 URL
```

###### Async Implementation Flows

1. User requests to download a directory.
2. The client sends a download directory request to the API.
3. The API saves the request in the database.
4. The API returns a 201 response to the client with a link to check the download status.
5. The Zipper service consumes the request from the database.
6. The Zipper service calls Store and streams the files from the store.
7. The Zipper service starts zipping the streamed files.
8. The Zipper service uploads the zipped file to S3.
9. The S3 sends the URL information back to the Zipper service.
10. The Zipper service updates the request in the database with the S3 URL.
11. The client sends a request to check the download directory status to the API.
12. The API retrieves the download URL based on the request ID from the database.
13. The API returns a 200 response to the client with the S3 URL.

###### Synchronous Implementation (fast development time)

```mermaid
sequenceDiagram
  Client->>API: Download dir
  loop Directory tree
    API->>Store: Stream files
  end
  API->>API: zipping
  API->>Client: return the zipped file
```

###### Synchronous Implementation Flows

1. User requests to download a directory.
2. The client sends a download directory request to the API.
3. The API begins streaming the files from the store in a loop, recursively traversing the entire directory tree.
4. The API zips the streamed files.
5. The API returns the zipped file to the client.
6. The client receives the zipped file.

#### Design Caveat

Depending on the number of files being streamed, there’s a potential risk of running into memory issues. To prevent this, we should implement a check to estimate the total download size, such as 1GB.