Find duplicate files

duplicates.js

#!/usr/bin/env node

var sys = require('sys'),
    fs = require('fs'),
    childProcess = require('child_process')

hashtable = {}
md5table = {}

fileblacklist = []
filequeue = []
md5queue = []

threadNum = 2   // this tends to be equivalent to the load avarage.
threadsWaiting = 0

stated = 0
hashed = 0
samesize = 0
duplicates = 0

debug = false

function fill(text, width) {
    text = String(text)
    while (text.length < width) text = " " + text
    return text
}

statistics = setInterval(function () {
    sys.print(
        "\rHashed: " + fill(hashed,6) + 
            "/" + fill(stated,6) + 
        " Queues: " + fill(md5queue.length,1) + 
            "/" + fill(filequeue.length,3) + 
        " Threads: " + fill(threadNum-threadsWaiting,1) + 
            "/" + fill(threadNum,1) + 
        " Dup: " + fill(duplicates,3) + 
            "/" + fill(samesize,6))
    if (!threadNum) {
        clearInterval(statistics)
        sys.puts("")
    }
}, 300)

function handleEntry () {
    // test if files should be md5-hashed
    var file = md5queue.pop()
    if (typeof file == 'string') {
        debug && dupstream.write("Hashing file " + file + "\n")
        childProcess.exec('md5sum "' + file + '"', function 
                (error, stdout, stderr) {
            hashed++
            var md5sum = /^([a-z0-9]*)/.exec(stdout)[1]
            if (md5table[md5sum]) {
                duplicates++
                md5table[md5sum].push(file)
                dupstream.write("Duplicates: " + 
                        md5table[md5sum].join(", ") + "\n")
            } else {
                md5table[md5sum] = [file]
            }
            process.nextTick(handleEntry)
        })
        return
    }
    
    // check if waiting is neccessary
    file = filequeue.pop()
    if (typeof file != 'string') {
        debug && dupstream.write("Waiting.\n")
        threadsWaiting++
        if (threadsWaiting == threadNum) {
            threadsWaiting--
            threadNum--
            return
        }
        setTimeout(function () {
            threadsWaiting--
            process.nextTick(handleEntry)
        }, 100)
        return
    }
    
    // check blacklist
    for (var i in fileblacklist) {
        if (file == fileblacklist[i]) {
            fileblacklist.splice(i, 1)
            process.nextTick(handleEntry)
            return
        }
    }
    
    // stat path
    debug && dupstream.write("Checking file " + file + "\n")
    fs.stat(file, function (err, stats) {
        if (!err) {
            stated++
            if (stats.isFile()) {
                if (hashtable[stats.size]) {
                    samesize++
                    if (hashtable[stats.size].length == 1) {
                        samesize++
                        md5queue.push(hashtable[stats.size][0])
                    }
                    md5queue.push(file)
                    hashtable[stats.size].push(file)
                } else {
                    hashtable[stats.size] = [file]
                }
            } else if (stats.isDirectory()) {
                fs.readdir(file, function (err, files) {
                    if (!err) {
                        for (var i in files)
                            filequeue.push(file + "/" + files[i])
                    }
                    process.nextTick(handleEntry)
                })
                return
            }
        }
        process.nextTick(handleEntry)
    })
}

if (process.argv.length < 4) {
    sys.puts("")
    sys.puts("Usage: node this.js" + 
        " <outfile> <path> [...] [-x <path> [...]]")
    sys.puts("")
    sys.puts("Pathes of duplicate files are written to <outfile>")
    sys.puts("All pathes are searched recursivly")
    sys.puts("-x <path>    excludes path from beeing searched")
    sys.puts("")
    process.exit(1)
}

dupstream = fs.createWriteStream(process.argv[2])

for (var i = 3 ; i < process.argv.length ; i++) {
    if (process.argv[i] == "-x") {
        i++
        fileblacklist.push(process.argv[i].replace(/\/$/, ""))
    } else
        filequeue.push(process.argv[i].replace(/\/$/, ""))
}

for (var i = 0 ; i < threadNum ; i++)
    handleEntry()

das ist aber javascript :-P
wahrscheinlich müsste das da heißen:
#!/usr/bin/env node

Der Code erinnerte mich so an Python.
Wusste gar nicht, dass JS keine Strichpunkte hat :D

Jo, war mir auch ganz neu.
Da habe ich jetzt jahre gebraucht, um mir das semikolon anzugewöhnen, und jetzt brauche ich es nicht mehr :-P