Skip to content

Latest commit

 

History

History
204 lines (140 loc) · 4.95 KB

README-en.md

File metadata and controls

204 lines (140 loc) · 4.95 KB

Github Growth Hack

中文文档

A framework to help you do growth hacking for your open source projects.

Motivation

This project wants to solve these problems:

  • Push

How to make the project owners invest more man-days in their open source projects?

  • Efficiency improvement

How to automate some chores and save some technical man-days?

  • Planning

Given the limited resource of technical man-days,what should we do to (try to) maximize the benefits?

Features

It can:

  1. Generate a weekly or monthly report for the github projects you care about
  • calculate an "activity score" for every project
  • use the scores to rank these projects
  • send the report to IM group
  • have a dashboard to show them

Screenshots of the dashboard and reports:

excel

  1. Find those issues without any comments and warn you

Screenshots of the warnings:

Usage

Generate Weekly Report as Excel

  1. clone the repo
git clone https://github.com/seeflood/github-growth-hack.git
  1. Modify the configuration options in src/index.js
// Modify these configuration items as you like
// 1. Your github API token
// We use your token to invoke github graphql api to query some repo data and won't do any modification.
// see https://docs.github.com/en/graphql/guides/forming-calls-with-graphql#authenticating-with-graphql
const token = ``;

// 2. Time range.
// modify it as you like
// e.g.
// let since = `2021-06-12T00:32:13Z`;
// let since = utils.lastSunday();
// let since = `2021-10-09T00:32:13Z`;
let since = utils.sevenDaysBefore();
// let to = utils.thisFriday23_59_59();
let to = utils.today();

// 3. Which repositries do you care about?
// Parameters in the repos array are:
// [owner, repo]
let repos = [
  ["mosn", "mosn"],
  ["mosn", "layotto"],
];

// 4. (Optional) Send messages to dingtalk group
const dingTalkGroupConfig = {
  url: "",
  keyword: "",
  owners: new Map(),
};

dingTalkGroupConfig.owners.set("project name", "dingtalk uid");
dingTalkGroupConfig.owners.set("layotto", "193555");
  1. Run it
# install packages
npm install
npm run week

  1. Now you have the weekly report for your repositories!

    result

    Check the generated .csv files:

    csv

    You can open them with Excel:

    excel

Send issue warning to dingtalk group

  1. Modify the configuration options in src/index.js

  2. Run npm run scan

Generate Weekly Report as Grafana Dashboard

To generate a Grafana Dashboard,you have to load the report.csv data into a MYSQL table and start a Grafana server integrated with the MYSQL table.

  1. Create a MYSQL Table
CREATE TABLE `github_repo_weekly` (
  `id` int(11) unsigned NOT NULL AUTO_INCREMENT,
  `date_from` datetime DEFAULT NULL,
  `date_to` datetime DEFAULT NULL,
  `record_date` datetime DEFAULT NULL,
  `rank` int(11) NOT NULL,
  `score` int(11) DEFAULT NULL,
  `owner` varchar(40) DEFAULT NULL,
  `project` varchar(100) DEFAULT NULL,
  `new_stars` int(11) DEFAULT NULL,
  `new_contributors` int(11) DEFAULT NULL,
  `new_forks` int(11) DEFAULT NULL,
  `new_pr` int(11) DEFAULT NULL,
  `closed_pr` int(11) DEFAULT NULL,
  `new_issues` int(11) DEFAULT NULL,
  `closed_issues` int(11) DEFAULT NULL,
  `pr_comment` int(11) DEFAULT NULL,
  `issue_comment` int(11) DEFAULT NULL,
  PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 AVG_ROW_LENGTH=150 ROW_FORMAT=DYNAMIC COMMENT='github_repo_weekly'
;

  1. Run a Grafana server and add the MYSQL Table as a new datasource.

  2. Set up your Grafana dashboard using the config file grafana/dashboard.json

  3. Set up the configuration in src/index.js(see above).

Make sure you have set the mysql config:

// (Optional) Write the reports into Mysql
let mysqlConfig = {
  host: "",
  user: "",
  password: "",
  database: "",
};
  1. Run it
# install packages
npm install
npm run week
  1. Open your Grafana Dashboard:

Generate "Monthly Active Contributors(MAC)" report as Excel

npm run month

Scan issues without any comments for at least 5 days

npm run scan

Rationale

How is the 'score' calculated?

Based on the formula proposed by open-digger,but I add some new factors

score