2023-08-30 12:39:38 +01:00
|
|
|
package storage
|
|
|
|
|
|
|
|
|
|
import (
|
|
|
|
|
"bytes"
|
2023-09-02 12:07:30 +01:00
|
|
|
"context"
|
2023-08-30 12:39:38 +01:00
|
|
|
"crypto/sha256"
|
|
|
|
|
"encoding/hex"
|
|
|
|
|
"fmt"
|
|
|
|
|
"hash"
|
|
|
|
|
"io"
|
|
|
|
|
"math"
|
|
|
|
|
"net/http"
|
|
|
|
|
"os"
|
2023-09-02 12:14:02 +01:00
|
|
|
"path/filepath"
|
2023-08-30 12:39:38 +01:00
|
|
|
|
|
|
|
|
"github.com/dustin/go-humanize"
|
|
|
|
|
"github.com/pkg/errors"
|
|
|
|
|
)
|
|
|
|
|
|
2023-09-02 12:07:30 +01:00
|
|
|
func (s *Storage) download(ctx context.Context, url string, hash []byte, out string, applyReaderMiddleware func(io.Reader) io.Reader) error {
|
2023-09-02 12:14:02 +01:00
|
|
|
outClean := filepath.Clean(out)
|
|
|
|
|
|
2023-08-30 12:39:38 +01:00
|
|
|
var created, success bool
|
|
|
|
|
|
|
|
|
|
defer func() {
|
|
|
|
|
if created && !success {
|
2023-09-02 12:14:02 +01:00
|
|
|
_ = os.Remove(outClean)
|
2023-08-30 12:39:38 +01:00
|
|
|
}
|
|
|
|
|
}()
|
|
|
|
|
|
2023-09-02 12:14:02 +01:00
|
|
|
_, err := os.Stat(outClean)
|
2023-08-30 12:39:38 +01:00
|
|
|
if err == nil {
|
|
|
|
|
return errors.Wrap(err, "file already exists")
|
2023-09-02 11:47:58 +01:00
|
|
|
} else if !errors.Is(err, os.ErrNotExist) {
|
|
|
|
|
return errors.Wrap(err, "stat out path")
|
2023-08-30 12:39:38 +01:00
|
|
|
}
|
|
|
|
|
|
2023-09-02 12:14:02 +01:00
|
|
|
f, err := os.OpenFile(outClean, os.O_CREATE|os.O_WRONLY, 0400)
|
2023-08-30 12:39:38 +01:00
|
|
|
if err != nil {
|
|
|
|
|
return errors.Wrap(err, "open file")
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
created = true
|
|
|
|
|
|
|
|
|
|
defer func() { _ = f.Close() }()
|
|
|
|
|
|
2023-09-02 12:07:30 +01:00
|
|
|
req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
|
|
|
|
|
if err != nil {
|
|
|
|
|
return errors.Wrap(err, "create new http get request")
|
|
|
|
|
}
|
|
|
|
|
|
2023-09-02 12:14:02 +01:00
|
|
|
s.logger.Info("Starting to download file", "from", url, "to", outClean)
|
2023-08-30 12:39:38 +01:00
|
|
|
|
2023-09-02 12:07:30 +01:00
|
|
|
resp, err := http.DefaultClient.Do(req)
|
2023-08-30 12:39:38 +01:00
|
|
|
if err != nil {
|
|
|
|
|
return errors.Wrap(err, "http get")
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
defer func() { _ = resp.Body.Close() }()
|
|
|
|
|
|
2023-08-30 13:13:08 +01:00
|
|
|
knownSize := resp.ContentLength
|
|
|
|
|
|
|
|
|
|
var readFrom io.Reader
|
|
|
|
|
if applyReaderMiddleware != nil {
|
|
|
|
|
readFrom = applyReaderMiddleware(resp.Body)
|
|
|
|
|
knownSize = 0
|
|
|
|
|
} else {
|
|
|
|
|
readFrom = resp.Body
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
n, err := copyWithProgressAndHash(f, readFrom, 1024, hash, func(downloaded int) {
|
|
|
|
|
var percent float64
|
|
|
|
|
if knownSize != 0 {
|
|
|
|
|
percent = float64(downloaded) / float64(knownSize)
|
|
|
|
|
}
|
|
|
|
|
|
2023-09-02 12:14:02 +01:00
|
|
|
lg := s.logger.With("out", outClean, "done", humanize.Bytes(uint64(downloaded)))
|
2023-08-30 13:13:08 +01:00
|
|
|
if percent != 0 {
|
|
|
|
|
lg.Info("Downloading file", "percent", math.Round(percent*100*100)/100)
|
|
|
|
|
} else {
|
|
|
|
|
lg.Info("Downloading compressed file", "percent", "N/A")
|
|
|
|
|
}
|
2023-08-30 12:39:38 +01:00
|
|
|
})
|
|
|
|
|
if err != nil {
|
|
|
|
|
return errors.Wrap(err, "copy resp to file")
|
|
|
|
|
}
|
|
|
|
|
|
2023-09-02 12:14:02 +01:00
|
|
|
s.logger.Info("Successfully downloaded file", "from", url, "to", outClean, "out-size", humanize.Bytes(uint64(n)))
|
2023-08-30 12:39:38 +01:00
|
|
|
|
|
|
|
|
success = true
|
|
|
|
|
|
|
|
|
|
return nil
|
|
|
|
|
}
|
|
|
|
|
|
2023-08-30 13:13:08 +01:00
|
|
|
func copyWithProgressAndHash(dst io.Writer, src io.Reader, blockSize int, wantHash []byte, report func(int)) (int, error) {
|
2023-08-30 12:39:38 +01:00
|
|
|
block := make([]byte, blockSize)
|
|
|
|
|
|
|
|
|
|
var h hash.Hash
|
|
|
|
|
if wantHash != nil {
|
|
|
|
|
h = sha256.New()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
var progress int
|
|
|
|
|
|
|
|
|
|
for {
|
|
|
|
|
read, err := src.Read(block)
|
|
|
|
|
if read > 0 {
|
|
|
|
|
written, err := dst.Write(block[:read])
|
|
|
|
|
if err != nil {
|
|
|
|
|
return progress, errors.Wrap(err, "write")
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if h != nil {
|
|
|
|
|
h.Write(block[:read])
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
progress += written
|
|
|
|
|
}
|
|
|
|
|
if err != nil {
|
|
|
|
|
if errors.Is(err, io.EOF) {
|
|
|
|
|
break
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return progress, errors.Wrap(err, "read")
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if progress%1000000 == 0 {
|
2023-08-30 13:13:08 +01:00
|
|
|
report(progress)
|
2023-08-30 12:39:38 +01:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if h != nil {
|
|
|
|
|
sum := h.Sum(nil)
|
|
|
|
|
if !bytes.Equal(sum, wantHash) {
|
|
|
|
|
return progress, fmt.Errorf("hash mismach: want '%v', have '%v'", hex.EncodeToString(wantHash), hex.EncodeToString(sum))
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return progress, nil
|
|
|
|
|
}
|