practicing node streams – script generates a git commit-log, parses it and reformats into a day-by-day-log useful for time-tracking.

commit-timesheet.js

#!/usr/bin/env node

var fs = require('fs'),
    util = require('util'),
    spawn = require('child_process').spawn,
    stream = require('stream');


function dateString(date) { return date.toISOString().split('T')[0]; }
function timeString(date) { return date.toTimeString().slice(0,5); }

var logParser = (function() { 
    var s=new stream.Transform({objectMode:true});
    
    s._transform = function(chunk, encoding, callback) {
        var buf = chunk.toString('utf-8'),
            lines = buf.trim().split("\n");

        lines = lines.map(function(line) {
            // note: this of course correlates with git's log-format
            var parts = line.split('|');
            var ts = new Date(parts[0]),
                date = dateString(ts),
                time = timeString(ts);

            return { timestamp: ts.getTime(), date: date, time: time, author: parts[1], msg: parts[2] };
        });

        lines.forEach(this.push.bind(this));
        callback();
    };

    return s;
} ());

var dayAggregator = (function() {
    var s = new stream.Transform({objectMode:true});

    var aggregate;

    function _reset() {
        aggregate = { date: '', firstCommitAt: null, lastCommitAt: null, msgs: [] };
    }
    _reset();
    
    s._transform = function(o, encoding, callback) {
        if(o.date !== aggregate.date && aggregate.msgs.length > 0) {
            aggregate.msgs = aggregate.msgs.sort(function(a,b) { return a.timestamp - b.timestamp; });
            this.push(aggregate);

            _reset();
        }

        if(!aggregate.firstCommitAt) {
            aggregate.firstCommitAt = new Date(o.timestamp);
        }

        aggregate.date = o.date;
        aggregate.lastCommitAt = new Date(o.timestamp);
        aggregate.msgs.push(o);

        callback();
    };
    

    return s;
} ());

var dayLogWriter = (function() {
    var s = new stream.Transform({objectMode:true});
    s._write = function(o, encoding, callback) {
        this.push('\n\033[1;32m---- '+o.date+' ----\033[0m\n');

        var msgs = o.msgs.map(function(msg) {
            return '\033[36m['+msg.time+']\033[0m \033[34m' + msg.author + '\033[0m: ' + msg.msg; 
        });
        
        this.push('  '+(msgs.join('\n  '))+'\n\n');
        this.push('First commit: '+timeString(o.firstCommitAt) + '\n');
        this.push('Last commit:  '+timeString(o.lastCommitAt) + '\n');

        callback();
    };

    return s;
} ());

var dumpObjectStream = (function() { 
    var s = new stream.Writable({objectMode:true});
    s._write = function(chunk, encoding, callback) {
        console.log('dumpObjectStream:', util.inspect(chunk, {colors:true}));
        callback();
    };

    return s;
}());

// now fetch the data and format it...
var cp = spawn('git', ['log', '--reverse', '--date-order', '--format=%ai|%ae|%s'].concat(process.argv.slice(2)));
cp.stdout.pipe(logParser).pipe(dayAggregator).pipe(dayLogWriter).pipe(process.stdout);