--- layout: default title: 'Supported Locations' nav_order: 2 description: 'Supported Locations' parent: Reference permalink: /docs/locations --- ## Supported Location URLs ReplayWeb.Page uses URLs to uniquely identify web archives. In addition HTTP/S, the system supports loading from additional cloud providers and via peer-to-peer decentralized services. The goal is to support loading web archives from any source a web browser can connect to. The following is a list of currently supported URL schemes and details below: | Scheme | Status | |:-----------------|:------------| | `https://` | Supported | | `http://` | Supported | | `file://` | Supported | | `s3://` | Supported | | `googledrive://` | Supported | | `ipfs://` | Supported | | `hyper://` | Planned | ### HTTP and HTTPS While `https://replayweb.page` can load from any URL, a site must allow access by enabling [CORS](https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS) for the web archive. HTTP URLs may not be accessible from `https://replayweb.page` since it requires HTTPS. If you need to provide access to web archives stored on your own domain or over HTTP, consider deploying a copy on your server. See: Running your Own. ### Local Files Local files selected via the UI file chooser, and will be given a `file://` URL. Unlike other web archive URLs, file URLs can not be shared with others. ### Amazon S3 The `s3://` URL scheme is provided as a convenience to reference S3 URLs. The URL is converted to an HTTPS URL for access. The S3 Bucket must be public and also provide access to `https://replayweb.page` via CORS. Downloading via S3 credentials is not currently supported. ### Google Drive The `googledrive://` URL scheme is used when accessing a web archive shared over Google Drive. It provides a unique identifier provided via Google Drive. When accessing a Google Drive URL, the user will be asked to authenticate to be able to access the file. ### IPFS {: .d-inline-block } Help Wanted {: .label .label-yellow } The `ipfs://` URL scheme is provide as a convenience and resolves to using the Cloudflare IPFS Gateway to download archives from IPFS. A better implementation with IPFS directly is also possible. ### DAT/Hyper {: .d-inline-block } Help Wanted {: .label .label-yellow } Support for `hyper://` is planned to allow true peer-to-peer transfer of web archives.