You need to have docker and docker compose installed.
Run docker compose-up -d to start the services.
Run (or simply visit the address): curl http://localhost:8080/get-address-to-fund and fund the address with the corresponding funds, this is an address in your control, see To make backups.
This should make the Bee service work properly. To check this, run: docker-compose logs -f bee-1.
If you want to check the logs of the API container: docker-compose logs -f uploader
Create a request to get-batch-id and save its return value.
When that all is set, you can go ahead and call /upload with your zim file and the batch id.
Docker Compose will start 3 services: Bee and bee-clef instances that will interact with Swarm. And a Node API running on port 8080, that has the following:
/upload: recieves an url to a zim file as a parameter, and an id parameter (this is a stamp batch id) that you can get from/get-batch-id, it then processes it and uploades it to Swarm. Example:http://localhost:8080/upload?url=https://download.kiwix.org/zim/wikipedia/wikipedia_bm_all_maxi_2022-02.zim&bid=01077a80d594c2d694efae01f93f3294aeabaf7d48bfcfd7f5f319b46726223e. This will return an object with the hash and a tag to check for the upload./get-batch-id: Generates a batch of stamps and returns the id./check-upload: Will receive a tag id and return the status of the corresponding upload. Example:http://localhost:8080/check-upload?tag=:1043824072/get-address-to-fund: This is the address you need to fund with xDai and BZZ in the Gnosis Chain in order to get the bee node to interact with the network.
The main files to take note of are:
docker-compose.yml: Handles creating containers of swarm and clef (which might not bee needed if we move to use a file as a key) and a node container that handles running:src/app.ts: express server with endpoints for interacting with the Bee container.scripts/prepareFiles.sh: handles decompressing files with zimtools, and calling.src/utils.ts: handles upload of files with mantaray-js. For more information on this see the credits.hashes/the hashes folder will store the hashes created as a backup.
There are some ideas that were implemented and that I still have on this repo like:
- A Next.js application that solves the redirection issues (
/bzz/HASH) and acts as a single page application with the content of the zim file. This could be expanded upon if there are specific needs in the future, for example generating extra pages on top of the wiki files as links to other versions of the pages in Swarm.
docker cp swarm-zim-uploader_clef-1_1:/app clef
docker cp swarm-zim-uploader_bee-1_1:/home/bee/.bee/ beeBEE_SWAP_ENDPOINT=https://gno.getblock.io/mainnet/?api_key=YOUR-API-KEY
BEE_PASSWORD=YOUR-PASSWORD
- Reading a funded wallet from a file instead of generating one and storing it in a volume.
- Code improvements: modularity.
- Endpoint to check and renew stamps.
- Creating a frontend for the application.
- Improve the index html file generated.
- Create a frontend app for interacting with this, it would basically serve as an intermediary between a person that wants to provide technical resources and a person who'd like to see a particular set of websites backed up in Swarm.
- Create a browser extension that allows for saving a particular wiki article.
When developing app-wiki an error comes out (similar to this). What's happening docker-compose automatically installs dependencies, which makes the docker user the owner. This can be fixed by running npm install on the local machine.
- Waiting for the stamp to propagate. Solutions:
- For this I decided to go with an extra endpoint to create the stamp, but I see the potential for something akin to a frontend for this API that could handle notifying the user when the stamps are ready to be used.
This would not have worked out well without the help of ldeffenb. The upload functionality in chunks was grabbed from his repository. We talked on discord regarding mixing this solution with his and he was kind enough to agree in mixing them to get this work out there. See his special considerations for more.
If you don't want to clone and run the uploader, you can see the results on the mainnet using either the following reference in your own mainnet swarm node, or via the bzz.link gateway. Note that the gateway's access throttle can make images fail to appear until several refreshes are done.
bzz/2fb90d693e7202791e8148ca5d5cbc46aaae2f9c6f606eb7211ba2a08dedf31a/
https://bah5acgzaf64q22j6oibhshubjdff2xf4i2vk4l44n5qg5nzbdorkbdpn6mna.bzz.link/