arozdfs/main.go

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"io"
	"io/ioutil"
	"math"
	"mime/multipart"
	"net/http"
	"os"
	"path"
	"path/filepath"
	"strconv"
	"strings"

	"github.com/FossoresLP/go-uuid-v4"
)

func check(e error) {
	if e != nil {
		panic(e)
	}
}

//ArOZ PHP-Golang Bridge
//The following sections remap the PHP functions to golang for the ease of development
func file_exists(filepath string) bool {
	if _, err := os.Stat(filepath); !os.IsNotExist(err) {
		return true
	}
	return false
}

func mkdir(filepath string) {
	os.MkdirAll(filepath, os.ModePerm)
}

func file_put_contents(file string, data string) bool {
	f, err := os.Create(file)
	check(err)
	_, err = f.WriteString(data)
	defer f.Close()
	if err != nil {
		return false
	}
	return true
}

func file_get_contents(file string) string {
	b, err := ioutil.ReadFile(file)
	check(err)
	return string(b)
}

func strtolower(word string) string {
	return strings.ToLower(word)
}

func strtoupper(word string) string {
	return strings.ToUpper(word)
}

func trim(word string) string {
	return strings.Trim(word, " ")
}

func strlen(word string) int {
	return len(word)
}

func count(array []string) int {
	return len(array)
}

func explode(key string, word string) []string {
	return strings.Split(word, key)
}

func implode(key string, array []string) string {
	return strings.Join(array[:], key)
}

func str_replace(target string, result string, word string) string {
	return strings.Replace(word, target, result, -1)
}

func in_array(a string, list []string) bool {
	for _, b := range list {
		if b == a {
			return true
		}
	}
	return false
}

func strpos(word string, target string) int {
	return strings.Index(word, target)
}

func dirname(filepath string) string {
	return path.Dir(filepath)
}

func basename(filepath string) string {
	return path.Base(filepath)
}

//End of mapping functions
//Utilities functions
func genUUIDv4() string {
	uuid, err := uuid.NewString()
	check(err)
	return uuid
}

func padZeros(thisInt string, maxval int) string {
	targetLength := len(strconv.Itoa(maxval))
	result := thisInt
	if len(thisInt) < targetLength {
		padzeros := targetLength - len(thisInt)
		for i := 0; i < padzeros; i++ {
			result = "0" + result
		}
	}
	return result

}

type clusterConfig struct {
	Prefix string `json: "prefix"`
	Port   string `json: "port`
}

//End of utilities functions

func init() {
	//Check if the required directory exists. If not, create it.
	if !file_exists("chunks/") {
		mkdir("chunks/")
	}
	if !file_exists("uploads/") {
		mkdir("uploads/")
	}
	if !file_exists("index/") {
		mkdir("index/")
	}
	if !file_exists("tmp/") {
		mkdir("tmp/")
	}
	if !file_exists("remoteDisks.config") {
		file_put_contents("remoteDisks.config", "")
	}
}

func main() {
	//arozdfs implementation in Golang
	//Refer to the help section for the usable commands and parameters

	if len(os.Args) == 1 {
		fmt.Println("ERROR. Undefined function group or operations. Type 'arozdfs help' for usage instructions. ")
		return
	}

	//For each argument, start the processing
	switch functgroup := os.Args[1]; functgroup {
	case "help":
		showHelp()
	case "slice":
		startSlicingProc()
	case "upload":
		startUploadProc()
	case "download":
		startDownloadProc()
	case "open":
		openChunkedFile()
	case "debug":
		fmt.Println(padZeros("1", 32)) //Debug function. Change this line for unit testing
	default:
		showNotFound()
	}

	/*
		//Examples for using the Go-PHP bridge functions
		file_put_contents("Hello World.txt", "This is the content of the file.")
		fmt.Println(file_get_contents("Hello World.txt"))

		array := explode(",", "Apple,Orange,Pizza")
		fmt.Println(array)
		newstring := implode(",", array)
		fmt.Println(newstring)
		fmt.Println(in_array("Pizza", array))
		fmt.Println(strpos(newstring, "Pizza"))
		fmt.Println(strtoupper("hello world"))
		fmt.Println(str_replace("Pizza", "Ramen", newstring))
	*/
}

func startDownloadProc() {

}

func startSlicingProc() {
	infile := ""
	slice := 64 //Default 64MB per file chunk
	fileUUID := genUUIDv4()
	storepath := fileUUID + "/"
	for i, arg := range os.Args {
		if strpos(arg, "-") == 0 {
			//This is a parameter defining keyword
			if arg == "-infile" {
				infile = os.Args[i+1]
			} else if arg == "-storepath" {
				storepath = os.Args[i+1]
				//Check if the storepath is end with /. if not, append it into the pathname
				if storepath[len(storepath)-1:] != "/" {
					storepath = storepath + "/"
				}
			} else if arg == "-slice" {
				sliceSize, err := strconv.Atoi(os.Args[i+1])
				check(err)
				slice = sliceSize
			}
		}
	}
	if slice <= 0 {
		fmt.Println("ERROR. slice size cannot be smaller or equal to 0")
		os.Exit(0)
	}
	if storepath != "" && infile != "" {
		fmt.Println(storepath + " " + infile + " " + strconv.Itoa(slice) + " " + fileUUID)
		splitFileChunks(infile, "chunks/"+storepath, fileUUID, slice)
		fmt.Println(fileUUID)
	} else {
		fmt.Println("ERROR. Undefined storepath or infile.")
	}

}

func splitFileChunks(rawfile string, outputdir string, outfilename string, chunksize int) bool {
	if !file_exists(outputdir) {
		mkdir(outputdir)
	}
	fileToBeChunked := rawfile
	file, err := os.Open(fileToBeChunked)

	if err != nil {
		return false
	}

	defer file.Close()

	fileInfo, _ := file.Stat()

	var fileSize int64 = fileInfo.Size()
	var fileChunk = float64(chunksize * 1024 * 1024) // chunksize in MB

	// calculate total number of parts the file will be chunked into

	totalPartsNum := uint64(math.Ceil(float64(fileSize) / float64(fileChunk)))
	fmt.Printf("Splitting to %d pieces.\n", totalPartsNum)
	for i := uint64(0); i < totalPartsNum; i++ {
		partSize := int(math.Min(fileChunk, float64(fileSize-int64(i*uint64(fileChunk)))))
		partBuffer := make([]byte, partSize)
		file.Read(partBuffer)
		// write to disk
		fileName := outputdir + outfilename + "_" + padZeros(strconv.FormatUint(i, 10), int(totalPartsNum))
		_, err := os.Create(fileName)
		if err != nil {
			return false
		}
		// write/save buffer to disk
		ioutil.WriteFile(fileName, partBuffer, os.ModeAppend)
		fmt.Println("Split to : ", fileName)
	}
	return true
}

func openChunkedFile() {
	storepath := "tmp/"
	uuid := ""
	outfile := ""
	removeAfterMerge := 0
	for i, arg := range os.Args {
		if strpos(arg, "-") == 0 {
			//This is a parameter defining keyword
			if arg == "-uuid" {
				uuid = os.Args[i+1]
			} else if arg == "-storepath" {
				storepath = os.Args[i+1]
				//Check if the storepath is end with /. if not, append it into the pathname
				if storepath[len(storepath)-1:] != "/" {
					storepath = storepath + "/"
				}
			} else if arg == "-outfile" {
				outfile = os.Args[i+1]
			} else if arg == "-c" {
				//Remove the file chunks after the merging process
				removeAfterMerge = 1
			}
		}
	}
	if storepath != "" && uuid != "" && outfile != "" {
		fmt.Println(storepath + " " + uuid + " " + outfile)
		if joinFileChunks(storepath+uuid, outfile) {
			//Do checksum here

			//Remove all files if -c is used
			if removeAfterMerge == 1 {
				matches, _ := filepath.Glob(storepath + uuid + "_*")
				for j := 0; j < len(matches); j++ {
					os.Remove(matches[j])
				}
			}
		} else {
			fmt.Println("ERROR. Unable to merge file chunks.")
		}
	} else {
		fmt.Println("ERROR. Undefined storepath, outfile or uuid.")
	}
}

func joinFileChunks(fileuuid string, outfilename string) bool {
	matches, _ := filepath.Glob(fileuuid + "_*")
	if len(matches) == 0 {
		fmt.Println("ERROR. No filechunk file for this uuid.")
		return false
	}
	outfile, err := os.Create(outfilename)
	if err != nil {
		return false
	}
	//For each file chunk, merge them into the output file
	for j := 0; j < len(matches); j++ {
		b, _ := ioutil.ReadFile(matches[j])
		outfile.Write(b)
	}
	return true
}

func startUploadProc() {
	push := "remoteDisks.config"
	storepath := "tmp/"
	uuid := ""
	vdir := ""
	for i, arg := range os.Args {
		if strpos(arg, "-") == 0 {
			//This is a parameter defining keyword
			if arg == "-uuid" {
				uuid = os.Args[i+1]
			} else if arg == "-storepath" {
				storepath = os.Args[i+1]
				//Check if the storepath is end with /. if not, append it into the pathname
				if storepath[len(storepath)-1:] != "/" {
					storepath = storepath + "/"
				}
			} else if arg == "-vdir" {
				vdir = os.Args[i+1]
			} else if arg == "-push" {
				//Remove the file chunks after the merging process
				push = os.Args[i+1]
			}
		}
	}
	//Check if the input data are valid
	if uuid == "" || vdir == "" {
		fmt.Println("ERROR. Undefined uuid or vdir.")
		os.Exit(0)
	}
	if !file_exists("clusterSetting.config") {
		fmt.Println("ERROR. clusterSetting configuration not found")
		os.Exit(0)
	}
	if file_exists("index/" + vdir + string(".index")) {
		fmt.Println("ERROR. Given file already exists in vdir. Please use remove before uploading a new file on the same vdir location.")
		os.Exit(0)
	}

	//Starting the uuid to ip conversion process

	var ipList []string
	//Read cluster setting from clusterSetting.config
	jsonFile, _ := os.Open("clusterSetting.config")
	byteValue, _ := ioutil.ReadAll(jsonFile)
	var config clusterConfig
	var uuiddata []string
	json.Unmarshal(byteValue, &config)
	//Read cluster uuid list from remoteDisks.config
	if file_exists(push) {
		clusteruuids := file_get_contents(push)
		if trim(clusteruuids) == "" {
			fmt.Println("ERROR. remoteDisks not found or it is empty! ")
			os.Exit(0)
		}
		clusteruuids = trim(strings.Trim(clusteruuids, "\n"))
		uuiddata = explode("\n", clusteruuids)
		//Generate iplist and ready for posting file chunks
		for i := 0; i < len(uuiddata); i++ {
			thisip := resolveUUID(uuiddata[i])
			clusterConfig := ":" + string(config.Port) + "/" + string(config.Prefix) + "/"
			fullip := "http://" + thisip + clusterConfig
			ipList = append(ipList, fullip)
		}
	} else {
		fmt.Println("ERROR. remoteDisks not found or it is empty! ")
		os.Exit(0)
	}
	fmt.Println(ipList)
	//Ready to push. Create index file.
	file_put_contents("index/"+vdir+string(".index"), "")
	fileList, _ := filepath.Glob(storepath + uuid + "_*")
	var pushResultTarget []string
	var pushResultFilename []string
	var failed []string
	var failedTarget []string
	for i := 0; i < len(fileList); i++ {
		uploadIP := (ipList[i%len(ipList)])
		r := pushFileChunk(uuid, uploadIP, fileList[i])
		if trim(r) == "DONE" {
			//This upload process is doing fine. Append to the result list
			pushResultTarget = append(pushResultTarget, uuiddata[i%len(ipList)])
			pushResultFilename = append(pushResultFilename, filepath.Base(fileList[i]))
			fmt.Println("[OK] " + fileList[i] + " uploaded.")
		} else {
			failed = append(failed, fileList[i])
			failedTarget = append(failedTarget, uuid)
		}
	}

	for j := 0; j < len(pushResultTarget); j++ {
		f, _ := os.OpenFile("index/"+vdir+string(".index"), os.O_APPEND|os.O_WRONLY, 0600)
		defer f.Close()
		f.WriteString(pushResultFilename[j])
		f.WriteString(",")
		f.WriteString(pushResultTarget[j])
		f.WriteString("\n")
	}
	fmt.Println("[OK] All chunks uploaded.")
}

func pushFileChunk(uuid string, targetEndpoint string, filename string) string {
	response := string(SendPostRequest(targetEndpoint+"SystemAOB/functions/arozdfs/upload.php", filename, "file"))
	return response
}

func resolveUUID(uuid string) string {
	tmp := []byte(uuid)
	uuid = string(bytes.Trim(tmp, "\xef\xbb\xbf"))
	uuid = strings.Trim(strings.Trim(uuid, "\n"), "\r")
	if file_exists("../cluster/mappers/") {
		if file_exists("../cluster/mappers/" + uuid + ".inf") {
			return file_get_contents("../cluster/mappers/" + uuid + ".inf")
		} else {
			fmt.Println("ERROR. UUID not found. Please perform a scan first before using arozdfs functions")
		}
	} else {
		fmt.Println("ERROR. Unable to resolve UUID to IP: cluster services not found. Continuing with UUID as IP address.")

	}
	return uuid
}

func SendPostRequest(url string, filename string, fieldname string) []byte {
	file, err := os.Open(filename)

	if err != nil {
		panic(err)
	}
	defer file.Close()

	body := &bytes.Buffer{}
	writer := multipart.NewWriter(body)
	part, err := writer.CreateFormFile(fieldname, filepath.Base(file.Name()))

	if err != nil {
		panic(err)
	}

	io.Copy(part, file)
	writer.Close()
	request, err := http.NewRequest("POST", url, body)

	if err != nil {
		panic(err)
	}

	request.Header.Add("Content-Type", writer.FormDataContentType())
	client := &http.Client{}

	response, err := client.Do(request)

	if err != nil {
		panic(err)
	}
	defer response.Body.Close()

	content, err := ioutil.ReadAll(response.Body)

	if err != nil {
		panic(err)
	}

	return content
}

func showHelp() {
	fmt.Println(`[arozdfs - Distributed File Storage Management Tool for ArOZ Online Cloud System]
	This is a command line tool build for the ArOZ Online distributed cloud platform file chunking and redundant data storage. 
	Please refer to the ArOZ Online Documentaion for more information.
	`)
	fmt.Println(`Supported commands:
	help --> show all the help information

	[Uploading to arozdfs commands]
	slice
	-infile <filename> --> declare the input file
	-slice <filesize> --> declare the slicing filesize
	-storepath <pathname> (Optional) --> Relative path for storing the sliced chunk files, default ./{file-uuid}

	upload
	-push <remoteDisks.config> --> push to a list of clusters and sync file index to other clusters
	-storepath <pathname> --> The location where the file chunks are stored
	-uuid <file uuid> --> uuid of the file to be uploaded
	-vdir <file.index> --> where the file.index should be stored. (Use for file / folder navigation)

	[Download from arozdfs commands]
	download
	-vdir <file.index> --> file.index location
	-storepath <tmp directory> (Optional) --> define a special directory for caching the downloaded data chunks, default ./tmp

	open
	-storepath <tmp directory> --> the file chunks tmp folder, default ./tmp
	-uuid <file uuid> --> the uuid which the file is stored
	-outfile <filename> --> filepath for the exported and merged file
	-c --> remove all stored file chunks after merging the file chunks.

	[File Operations]
	remove <file.index> --> remove all chunks related to thie file index
	rename <file.index> <newfile.index> --> rename all records related to this file
	move <filepath/file.index> <newpath/file.index> --> move the file to a new path in index directory


	[System checking commands]
	checkfile <file.index> --> check if a file contains all chunks which has at least two copies of each chunks
	rebuild --> Check all files on the system and fix all chunks which has corrupted
	migrate <remoteDisks.config> --> Move all chunks from this host to other servers in the list.`)
}

func showNotFound() {
	fmt.Println("ERROR. Command not found for function group: " + os.Args[1])
}