Commit count by tag & author per week in CSV

amap-git-log.csv

aviz-git-lot.csv

gitLogSummary.js

/**
 * Summarize a git repos logs as a CSV file.
 * This script was made for a viz, you can find it at
 * https://observablehq.com/@austil/project-summary-git-commit-by-type-per-week
 * 
 * First : export your log in a file using `git log --pretty=format:"%h%x09%an%x09%ad%x09%s" > git-log.txt`
 * Then process it with this script `node gitLogSummary.js git-log.txt`
 * and save it in a csv : `node gitLogSummary.js git-log.txt --csvOnly > git-log.csv`
 */

// TWEAK THIS TO YOUR NEEDS
const COMMIT_TAGS = ['feat', 'fix', 'docs', 'style', 'refactor', 'test', 'chore'];
const COMMIT_TAG_REGEX = /^\w*(?=(:|\())/;
const RENAMED_AUTHORS = {
  'Dev X': 'Others',
  'Dev Y': 'Others',
  'austil': 'alacour'
};
const IGNORED_AUTHORS = ['alacour'];

// Thx to https://stackoverflow.com/a/19910622
const getWeek = (date) => {
  var onejan = new Date(date.getFullYear(),0,1);
  var millisecsInDay = 86400000;
  return Math.ceil((((date - onejan) /millisecsInDay) + onejan.getDay()+1)/7);
};
const flatten = (arr) => [].concat(...arr);

// Let's start

const fs = require('fs');
const [ , , logFile, flag] = process.argv;

const humanLog = msg => {
  if(flag !== '--csvOnly') {
    console.log(msg);
  }
};

let gitLogs = fs.readFileSync(logFile, 'utf8')
  .split('\n')
  .map(l => {
    const log = l.split('\t');
    return {
      'author': RENAMED_AUTHORS[log[1]] || log[1],
      'date': new Date(log[2]),
      'msg': log[3]
    };
  });
humanLog(`raw log : ${gitLogs.length} commits`);

gitLogs = gitLogs.filter(log => log.msg.match(/(^Merge|^Revert)/) === null);
humanLog(`after removing merge/revert : ${gitLogs.length} commits`);

gitLogs = gitLogs.filter(log => {
  const tag = log.msg.match(COMMIT_TAG_REGEX);
  return tag !== null && COMMIT_TAGS.includes(tag[0]);
});
humanLog(`after removing missing tags : ${gitLogs.length} commits`);

gitLogs = gitLogs.filter(log => !IGNORED_AUTHORS.includes(log.author));
humanLog(`after removing ignored authors : ${gitLogs.length} commits`);

const authors = Array.from(new Set(gitLogs.map(l => l.author)));
humanLog(`remaining : ${gitLogs.length} commits from ${authors.length} authors`);

const logsSummary = gitLogs
  .map(log => ({
    'week': getWeek(log.date),
    'year': log.date.getFullYear(),
    'type': log.msg.match(COMMIT_TAG_REGEX)[0],
    'author': log.author
  }))
  .reverse() // Commit are from younger to older, reverse it
  .reduce((nest, log) => {
    // Count by week and year
    if(!nest[log.year]) nest[log.year] = {};
    if(!nest[log.year][log.week]) nest[log.year][log.week] = {};
    // for each tag
    if(!nest[log.year][log.week][log.type]) {
      nest[log.year][log.week][log.type] = 1;
    } else {
      nest[log.year][log.week][log.type] += 1;
    }
    // for each author
    if(!nest[log.year][log.week][log.author]) {
      nest[log.year][log.week][log.author] = 1;
    } else {
      nest[log.year][log.week][log.author] += 1;
    }
    return nest;
  }, {});

const csvLogs = flatten(Object.entries(logsSummary)
  .map(([year, val]) => Object.entries(val)
    .map(([week, counts]) => [
      year,
      week,
      ...COMMIT_TAGS.map(t => counts[t] || 0),
      ...authors.map(a => counts[a] || 0),
    ].join(','))
  )
).join('\n');

const csvHeader = ['year', 'week', ...COMMIT_TAGS, ...authors].join(',');

humanLog('\n===== Commit count by tag & author per week =====\n');
console.log(csvHeader);
console.log(csvLogs);

ATM contributors spamming commits will appear as more active than the others.
Maybe parse the --shortstat output to compute a "change delta" to be used instead of the count of commit.