Dropbox Content Hash

dropbox_content_hash.go

package dropbox

import (
	"crypto/sha256"
	"fmt"
	"os"

	"github.com/pkg/errors"
)

// ContentHash performs the dropbox content hash sum for a file on disk
func ContentHash(path string) (string, error) {
	file, err := os.Open(path)
	if err != nil {
		return "", errors.Wrap(err, "failed to open the file for reading")
	}
	defer file.Close()

	fi, err := file.Stat()
	if err != nil {
		return "", errors.Wrap(err, "failed to check the size of the file")
	}

	remainingBytes := fi.Size()
	blockSize := int64(4 * 1024 * 1024)

	// bytes from the shasums of each block
	blockSHABytes := []byte{}
	for {
		// bytes to read this time round
		bytesToRead := remainingBytes
		// don't read more than a block
		if bytesToRead > blockSize {
			bytesToRead = blockSize
		}

		// read that number of bytes
		byteSlice := make([]byte, bytesToRead)
		bytesRead, err := file.Read(byteSlice)
		if err != nil {
			return "", errors.Wrap(err, "failed to read block of bytes from file")
		}

		// update the remainingBytes, less the bytes we just read
		remainingBytes -= int64(bytesRead)

		// compute the shasum bytes for that block
		shaData := sha256.Sum256(byteSlice)
		// append the shasum bytes to the shasum bytes for the other blocks
		blockSHABytes = append(blockSHABytes, shaData[:]...)

		// exit if we don't need to read another block
		if int64(bytesRead) < blockSize {
			break
		}
	}

	// return a hex formatted string of the concatenated shasums, shasummed again
	return fmt.Sprintf("%x", sha256.Sum256(blockSHABytes)), nil
}