DF-PY

CLI-based Data Farming (DF) & veOCEAN (VE) backend. It's used for weekly "dispense" ops and to create data for VE/DF frontend.

Usage: in console, type dftool to see further commads.

Data flow and csvs: See "Data Flow in dftool" GSlides 2&3

Installation

Prerequisites

Ensure prerequisites:

• Linux/MacOS
• Python 3.8.5+
• mypy, pylint, black. sudo apt install mypy pylint black
• solc 0.8.0+ [Instructions]
• Any Ocean Barge pre-requisites. See here
• nvm 16.13.2, not nvm 17. To install: nvm install 16.13.2; nvm use 16.13.2. [Details]

Setup for Local: Install & Run Barge

We use Ocean Barge to run ganache, deploy contracts to Ganache, and run TheGraph with Ocean subgraphs. The deployed contracts come from github.com/oceanprotocol/contracts. df-py has a local redundant copy in its directory so that the system easily knows what objects look like.

Let's get Barge going. Open a new terminal and:

#get repo
git clone [email protected]:oceanprotocol/barge.git
cd barge

#clean up old containers (to be sure)
docker system prune -a --volumes

#run barge
#-deploys ocean contracts with addresses at ~/.ocean/ocean-contracts/artifacts/address.json
#-only runs the components it needs
export GANACHE_FORK="london"
./start_ocean.sh --no-aquarius --no-elasticsearch --no-provider --no-dashboard --with-thegraph

Install df-py

Then, open a new terminal and:

#clone repo
git clone https://github.com/oceanprotocol/df-py.git
cd df-py

#create & activate virtual env't
python -m venv venv
source venv/bin/activate

#install dependencies
pip install wheel
pip install -r requirements.txt

#install openzeppelin library, to import from .sol
npm install @openzeppelin/contracts

#add pwd to bash path
export PATH=$PATH:.

#set judge private key
#  - get at private repo https://github.com/oceanprotocol/private-keys
#  - CI automatically sees it via Github Actions Secrets
export JUDGE_PRIVATE_KEY=<judge key>

CLI

dftool is the main tool. In main terminal:

#top-level help, lists all tools
dftool

#see help for key functions
dftool calc
dftool dispense
...

Then, simply follow the usage directions:)

Running Tests

In terminal:

#run tests for one method, with print statements to console. "-s" is to show output
pytest df_py/volume/test/test_calcrewards.py::test_simple -s

#run tests for one module
pytest df_py/volume/test/test_calcrewards.py

#run all tests. Note: util is the only directory _with_ tests
pytest test df_py/util

#run static type-checking. By default, uses config mypy.ini. Note: pytest does dynamic type-checking.
mypy ./

#run linting on code style. Uses .pylintrc
pylint *

#auto-fix some pylint complaints
black ./

Setup for Remote Networks

Examples so far were on a local chain. Let's do a one-time setup for remote networks. In console:

export POLYGON_RPC_URL=https://polygon-rpc.com/
export MUMBAI_RPC_URL=https://polygon-mumbai.blockpi.network/v1/rpc/public

Or using an Infura ID, here is an example for polygon. Please note that you do not need to include the Infura Id in the URL, they will be glued together at runtime. However, you need to specify which networks use the infura project id, by a comma separated value list in the INFURA_NETWORKS env var.

export WEB3_INFURA_PROJECT_ID=***
export POLYGON_RPC_URL=https://polygon-mainnet.infura.io/v3/
export INFURA_NETWORKS=polygon,mumbai

Now, you can use those networks simply by specifying a different chainid in dftool calls.

Rewards Distribution Ops

Happens via regularly-scheduled Github Actions:

• Active: flow.yml

More info: README-crons-ops.md

DFRewards Owner Control Ops

See README-control-ops.md

Docker

Here's how to use df-py running inside a Docker container.

Install & Use Docker Locally

Build the docker image.

docker build . -t dfpy

Usage: ./dfpy_docker args

Docker will mount /tmp/dfpy:/app/data Which will be located inside of /root/.dfcsv

Example usage with docker:

./dfpy_docker help  # prints help

$ ./dfpy_docker get_rate OCEAN 2022-01-01 2022-01-02 /app/data

Arguments: ST=2022-01-01, FIN=2022-01-02, CSV_DIR=/app/data
rate = $0.8774 / OCEAN
Created /app/data/rate-OCEAN.csv

Docker Using Remote Networks

Expand configured networks in the Docker container by editing the Dockerfile as follows:

...
RUN export POLYGON_URL=https://polygon-rpc.com/
...
COPY . .
RUN rm -rf build

Docker Using Contract Addresses

Here's how to configure docker to access contracts.

First, address.json on contracts github to local.

Then, configure dfpy_docker like so:

docker run --env-file ./.env -v /tmp/dfpy:/app/data -v /app/df-py/address.json:/address.json --rm dfpy $@

End-to-end With SQL and Webapp

Let's have an end-to-end flow of df-py, df-sql, and df-web.

Overall flow:

• df-sql runs df-py/dftool to generate csvs, and then serve them up in a SQL API.
• Then, df-web consumes this API

More info:

• High-level diagram (from df-sql repo)
• df-sql reads csvs in ~/.dfcsv/

Here's how to set up this flow.

First, create folders, if they don't yet exist.

mkdir /tmp/dfpy/
mkdir ~/.dfcsv/ # df-sql reads csvs from here

Then, create a new script named getAllRecords-df-sql.sh, and fill it with the following content.

cd /app/df-py/
date=`date -dlast-wednesday '+%Y-%m-%d'`
now=`date '+%Y-%m-%d'`

/app/df-py/dfpy_docker get_rate OCEAN $date $now /app/data &&
/app/df-py/dfpy_docker query $date latest 1 /app/data 1 &&
/app/df-py/dfpy_docker query $date latest 1 /app/data 137 &&
/app/df-py/dfpy_docker calc /app/data 10000 OCEAN &&
mv /tmp/dfpy/* ~/.dfcsv/

Now, add the script to our crontab:

*/10 * * * * /app/df-py/getAllRecords-df-sql.sh

You can adjust this by changing this path in both repositories and redeploying.

Gotchas and workarounds

This section provides tactics if you encounter issues like KeyError "development".

Gotcha: KeyError "development"

If you run a test and get an error like KeyError: "development"

Then your problem is: barge wasn't able to deploy contracts to ganache. Then it couldn't update ~/.ocean/ocean-contracts/artifacts/address.json with a new sub-dict holding all the addresses at "development" chain (ganache). See #562.

How to fix:

• Tactic: in barge console: ./cleanup.sh
• Tactic: in any console: rm -rf ~/.ocean
• More tactics at barge README

Release process (developers)

In order to release a new version of df-py, simply create a new GitHub release. As opposed to other Ocean Protocol python repos, there is no need to use bumpversion.