packer-cn/common/download.go

304 lines
7.2 KiB
Go
Raw Normal View History

2013-06-12 20:41:44 -04:00
package common
import (
"bytes"
"crypto/md5"
"crypto/sha1"
"crypto/sha256"
"crypto/sha512"
2013-06-12 20:41:44 -04:00
"encoding/hex"
"errors"
"fmt"
"hash"
"io"
"log"
2013-06-12 20:41:44 -04:00
"net/http"
"net/url"
"os"
"runtime"
2013-06-12 20:41:44 -04:00
)
// DownloadConfig is the configuration given to instantiate a new
// download instance. Once a configuration is used to instantiate
// a download client, it must not be modified.
type DownloadConfig struct {
// The source URL in the form of a string.
Url string
// This is the path to download the file to.
TargetPath string
// DownloaderMap maps a schema to a Download.
DownloaderMap map[string]Downloader
// If true, this will copy even a local file to the target
// location. If false, then it will "download" the file by just
// returning the local path to the file.
CopyFile bool
// The hashing implementation to use to checksum the downloaded file.
Hash hash.Hash
// The checksum for the downloaded file. The hash implementation configuration
// for the downloader will be used to verify with this checksum after
// it is downloaded.
Checksum []byte
// What to use for the user agent for HTTP requests. If set to "", use the
// default user agent provided by Go.
UserAgent string
2013-06-12 20:41:44 -04:00
}
// A DownloadClient helps download, verify checksums, etc.
type DownloadClient struct {
config *DownloadConfig
downloader Downloader
}
// HashForType returns the Hash implementation for the given string
// type, or nil if the type is not supported.
func HashForType(t string) hash.Hash {
switch t {
case "md5":
return md5.New()
case "sha1":
return sha1.New()
case "sha256":
return sha256.New()
case "sha512":
return sha512.New()
default:
return nil
}
}
2013-06-12 20:41:44 -04:00
// NewDownloadClient returns a new DownloadClient for the given
// configuration.
func NewDownloadClient(c *DownloadConfig) *DownloadClient {
if c.DownloaderMap == nil {
c.DownloaderMap = map[string]Downloader{
"http": &HTTPDownloader{userAgent: c.UserAgent},
"https": &HTTPDownloader{userAgent: c.UserAgent},
2013-06-12 20:41:44 -04:00
}
}
return &DownloadClient{config: c}
}
// A downloader is responsible for actually taking a remote URL and
// downloading it.
type Downloader interface {
Cancel()
Download(*os.File, *url.URL) error
2013-06-12 20:41:44 -04:00
Progress() uint
Total() uint
}
func (d *DownloadClient) Cancel() {
// TODO(mitchellh): Implement
}
func (d *DownloadClient) Get() (string, error) {
// If we already have the file and it matches, then just return the target path.
if verify, _ := d.VerifyChecksum(d.config.TargetPath); verify {
2015-08-19 16:15:23 -04:00
log.Println("[DEBUG] Initial checksum matched, no download needed.")
2013-06-12 20:41:44 -04:00
return d.config.TargetPath, nil
}
u, err := url.Parse(d.config.Url)
2013-06-12 20:41:44 -04:00
if err != nil {
return "", err
}
log.Printf("Parsed URL: %#v", u)
2013-06-12 20:41:44 -04:00
// Files when we don't copy the file are special cased.
2015-06-22 15:14:35 -04:00
var f *os.File
2013-06-12 20:41:44 -04:00
var finalPath string
sourcePath := ""
if u.Scheme == "file" && !d.config.CopyFile {
// This is special case for relative path in this case user specify
// file:../ and after parse destination goes to Opaque
if u.Path != "" {
// If url.Path is set just use this
finalPath = u.Path
} else if u.Opaque != "" {
// otherwise try url.Opaque
finalPath = u.Opaque
}
// This is a special case where we use a source file that already exists
// locally and we don't make a copy. Normally we would copy or download.
2015-08-19 16:15:23 -04:00
log.Printf("[DEBUG] Using local file: %s", finalPath)
2013-08-03 16:34:48 -04:00
// Remove forward slash on absolute Windows file URLs before processing
if runtime.GOOS == "windows" && len(finalPath) > 0 && finalPath[0] == '/' {
finalPath = finalPath[1:]
}
// Keep track of the source so we can make sure not to delete this later
sourcePath = finalPath
if _, err = os.Stat(finalPath); err != nil {
return "", err
}
2013-06-12 20:41:44 -04:00
} else {
finalPath = d.config.TargetPath
2013-06-12 20:41:44 -04:00
var ok bool
d.downloader, ok = d.config.DownloaderMap[u.Scheme]
2013-06-12 20:41:44 -04:00
if !ok {
return "", fmt.Errorf("No downloader for scheme: %s", u.Scheme)
2013-06-12 20:41:44 -04:00
}
// Otherwise, download using the downloader.
f, err = os.OpenFile(finalPath, os.O_RDWR|os.O_CREATE, os.FileMode(0666))
2013-06-12 20:41:44 -04:00
if err != nil {
return "", err
}
log.Printf("[DEBUG] Downloading: %s", u.String())
err = d.downloader.Download(f, u)
2015-06-22 15:17:29 -04:00
f.Close()
if err != nil {
return "", err
}
2013-06-12 20:41:44 -04:00
}
if d.config.Hash != nil {
var verify bool
verify, err = d.VerifyChecksum(finalPath)
if err == nil && !verify {
// Only delete the file if we made a copy or downloaded it
if sourcePath != finalPath {
os.Remove(finalPath)
}
2015-06-22 15:17:29 -04:00
err = fmt.Errorf(
"checksums didn't match expected: %s",
hex.EncodeToString(d.config.Checksum))
2013-06-12 20:41:44 -04:00
}
}
return finalPath, err
}
// PercentProgress returns the download progress as a percentage.
func (d *DownloadClient) PercentProgress() int {
if d.downloader == nil {
return -1
}
return int((float64(d.downloader.Progress()) / float64(d.downloader.Total())) * 100)
2013-06-12 20:41:44 -04:00
}
// VerifyChecksum tests that the path matches the checksum for the
// download.
func (d *DownloadClient) VerifyChecksum(path string) (bool, error) {
if d.config.Checksum == nil || d.config.Hash == nil {
return false, errors.New("Checksum or Hash isn't set on download.")
}
f, err := os.Open(path)
if err != nil {
return false, err
}
defer f.Close()
log.Printf("Verifying checksum of %s", path)
2013-06-12 20:41:44 -04:00
d.config.Hash.Reset()
io.Copy(d.config.Hash, f)
2017-03-28 21:29:55 -04:00
return bytes.Equal(d.config.Hash.Sum(nil), d.config.Checksum), nil
2013-06-12 20:41:44 -04:00
}
// HTTPDownloader is an implementation of Downloader that downloads
// files over HTTP.
type HTTPDownloader struct {
progress uint
total uint
userAgent string
2013-06-12 20:41:44 -04:00
}
func (*HTTPDownloader) Cancel() {
// TODO(mitchellh): Implement
}
func (d *HTTPDownloader) Download(dst *os.File, src *url.URL) error {
log.Printf("Starting download: %s", src.String())
2015-06-22 15:14:35 -04:00
// Seek to the beginning by default
if _, err := dst.Seek(0, 0); err != nil {
return err
}
// Reset our progress
d.progress = 0
2015-06-22 15:14:35 -04:00
// Make the request. We first make a HEAD request so we can check
// if the server supports range queries. If the server/URL doesn't
// support HEAD requests, we just fall back to GET.
req, err := http.NewRequest("HEAD", src.String(), nil)
if err != nil {
return err
}
if d.userAgent != "" {
req.Header.Set("User-Agent", d.userAgent)
}
httpClient := &http.Client{
Transport: &http.Transport{
Proxy: http.ProxyFromEnvironment,
},
}
resp, err := httpClient.Do(req)
2015-06-22 15:14:35 -04:00
if err == nil && (resp.StatusCode >= 200 && resp.StatusCode < 300) {
// If the HEAD request succeeded, then attempt to set the range
// query if we can.
if resp.Header.Get("Accept-Ranges") == "bytes" {
if fi, err := dst.Stat(); err == nil {
if _, err = dst.Seek(0, os.SEEK_END); err == nil {
req.Header.Set("Range", fmt.Sprintf("bytes=%d-", fi.Size()))
d.progress = uint(fi.Size())
}
}
}
}
2015-06-22 15:14:35 -04:00
// Set the request to GET now, and redo the query to download
req.Method = "GET"
resp, err = httpClient.Do(req)
if err != nil {
return err
}
d.total = d.progress + uint(resp.ContentLength)
2013-06-12 20:41:44 -04:00
var buffer [4096]byte
for {
n, err := resp.Body.Read(buffer[:])
if err != nil && err != io.EOF {
return err
}
d.progress += uint(n)
if _, werr := dst.Write(buffer[:n]); werr != nil {
return werr
}
if err == io.EOF {
break
}
}
return nil
}
func (d *HTTPDownloader) Progress() uint {
return d.progress
}
func (d *HTTPDownloader) Total() uint {
return d.total
}