lapsync.go

package main

import _ "net/http/pprof"
import (
    "bufio"
    "crypto/md5"
    "flag"
    "fmt"
    "io"
    "os"
    "path/filepath"
    "runtime"
    "strings"
    "sync"
    "net/http"
    "log"
)

type entry struct {
    path string
    info os.FileInfo
    err  error
}

// md5sum does an md5sum on the contents of the file provided
func md5sum(file_path string, info os.FileInfo) string {
    var result string
    file, err := os.Open(file_path)
    if err != nil {
        return result
    }
    defer file.Close()

    reader := bufio.NewReader(file)

    hash := md5.New()
    if _, err := io.Copy(hash, reader); err != nil {
        return result
    }

    result = fmt.Sprintf("%x", hash.Sum(nil))
//    fmt.Println("MD5SUM: ", file_path, " : ", result)
    return result
}

// copyFile copies a file from a source to a destination
// worth noting that src_path and dst_path are the full path to the source and destination paths
func copyFile(src_path string, dst_path string) error {
    df, err := os.Create(dst_path)
    if err != nil {
        return err
    }
    defer df.Close()

    sf, err := os.Open(src_path)
    if err != nil {
        return err
    }
    defer sf.Close()

    _, err = io.Copy(df, sf)
    if err != nil {
        return err
    }

    return nil
}

// syncEntry does the work of checking src and dst looking for differences and then syncing to destination.
// It's worth noting that the src_path argument is the full path of the source file and dst_path is the
// full path of the root of the destination - we derive the full path of the destination file by joining
// dst_path and the unique portion of the src_path (to maintain src directory structure).
func (e *entry) syncEntry(src_path *string, dst_path *string) {
    src_spec := strings.TrimPrefix(e.path, *src_path)
    dst_full_path := filepath.Join(*dst_path, src_spec)

    // just letting us know what we're doing - i.e. debug info
//    fmt.Println("src:", e.path, "---> dst:", dst_full_path)

    // create destination directory structure if it doesn't exist
    base_dir_path := dst_full_path
    if !e.info.IsDir() {
        base_dir_path = filepath.Dir(dst_full_path)
    }
    err := os.MkdirAll(base_dir_path, 0777)
    if err != nil {
        e.err = err
        return
    }

    // if the destination doesn't exist, go ahead and copy (denoted with "+++" prepended output)
    dst_file, err := os.Open(dst_full_path)
    if err != nil {
        if os.IsNotExist(err) {
//            fmt.Println("+++", e.path)
            err := copyFile(e.path, dst_full_path)
            if err != nil {
                e.err = err
            }
            return
        } else {
            e.err = err
            return
        }
    }
    defer dst_file.Close()

    // if the destination does exist but the file sizes are different, go ahead and copy ("!!!" prepended output)
    dst_file_info, err := dst_file.Stat()
    if err != nil {
        e.err = err
        return
    }
    if e.info.Size() != dst_file_info.Size() {
//        fmt.Println("!!!", e.path)
        dst_file.Close()
        copyFile(e.path, dst_full_path)
        if err != nil {
            e.err = err
        }
        return
    }

    // if the md5 checksum doesn't match, go ahead and copy ("~~~" prepended output)
    if md5sum(e.path, e.info) != md5sum(dst_full_path, dst_file_info) {
//        fmt.Println("~~~", e.path)
        dst_file.Close()
        copyFile(e.path, dst_full_path)
        if err != nil {
            e.err = err
        }
        return
    }

    // should be binary identical ("===", e.path)
//    fmt.Println("===", e.path)

}

// syncPath is the worker goroutine
func syncPath(src_path *string, dst_path *string, queue chan *entry, wg *sync.WaitGroup) {
    for e := range queue {
        e.syncEntry(src_path, dst_path)
        if e.err != nil {
            fmt.Println("**ERROR:", e.err)
        }
        wg.Done()
    }
}

func parseArgs() {
    flag.Parse()

    if ( len(*src_path) < 1 ) {
        fmt.Println("Please provide a source directory to sync")
        os.Exit(1)
    }

    if ( len(*dst_path) < 1 ) {
        fmt.Println("Please provide a dest directory to sync")
        os.Exit(1)
    }

    fmt.Println("source directory is: ", *src_path)
    fmt.Println("dest directory is: ", *dst_path)
}

var src_path = flag.String("source", "", "source directory")
var dst_path = flag.String("dest", "", "destination directory")

func main() {

    parseArgs()

    go func() {
        log.Println(http.ListenAndServe("localhost:6060", nil))

    }()

    // want twice as many goroutines as we have cores... lots of io wait going so it shouldn't destroy the system
    goRoutines := runtime.NumCPU() * 2

    // work queue
    queue := make(chan *entry)

    // the "thread" barrier
    var wg sync.WaitGroup

    // start up the goroutines
    for i := 0; i < goRoutines; i++ {
        go syncPath(src_path, dst_path, queue, &wg)
    }

    // fill up the work queue with work
    filepath.Walk(*src_path, func(path string, info os.FileInfo, err error) error {
        wg.Add(1)
        queue <- &entry{path, info, err}
        return err
    })

    // wait for it to finish
    wg.Wait()

}