SHA-256 Merkle tree

SHA-256 Merkle tree
You are encouraged to solve this task according to the task description, using any language you may know.

As described in its documentation, Amazon S3 Glacier requires that all uploaded files come with a checksum computed as a Merkle Tree using SHA-256.

Specifically, the SHA-256 hash is computed for each 1MiB block of the file. And then, starting from the beginning of the file, the raw hashes of consecutive blocks are paired up and concatenated together, and a new hash is computed from each concatenation. Then these are paired up and concatenated and hashed, and the process continues until there is only one hash left, which is the final checksum. The hexadecimal representation of this checksum is the value that must be included with the AWS API call to upload the object (or complete a multipart upload).

Implement this algorithm in your language; you can use the code from the SHA-256 task for the actual hash computations. For better manageability and portability, build the tree using a smaller block size of only 1024 bytes, and demonstrate it on the RosettaCode title image with that block size. The final result should be the hexadecimal digest value a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c.

C

Library: GLib

<lang c>#include <glib.h>

1. include <stdlib.h>
2. include <stdio.h>
3. include <string.h>

guchar* sha256_merkle_tree(FILE* in, size_t block_size) {

   gchar* buffer = g_malloc(block_size);
   GPtrArray* hashes = g_ptr_array_new_with_free_func(g_free);
   gssize digest_length = g_checksum_type_get_length(G_CHECKSUM_SHA256);
   GChecksum* checksum = g_checksum_new(G_CHECKSUM_SHA256);
   size_t bytes;
   while ((bytes = fread(buffer, 1, block_size, in)) > 0) {
       g_checksum_reset(checksum);
       g_checksum_update(checksum, (guchar*)buffer, bytes);
       gsize len = digest_length;
       guchar* digest = g_malloc(len);
       g_checksum_get_digest(checksum, digest, &len);
       g_ptr_array_add(hashes, digest);
   }
   g_free(buffer);
   guint hashes_length = hashes->len;
   if (hashes_length == 0) {
       g_ptr_array_free(hashes, TRUE);
       g_checksum_free(checksum);
       return NULL;
   }
   while (hashes_length > 1) {
       guint j = 0;
       for (guint i = 0; i < hashes_length; i += 2, ++j) {
           guchar* digest1 = g_ptr_array_index(hashes, i);
           guchar* digest_out = g_ptr_array_index(hashes, j);
           if (i + 1 < hashes_length) {
               guchar* digest2 = g_ptr_array_index(hashes, i + 1);
               g_checksum_reset(checksum);
               g_checksum_update(checksum, digest1, digest_length);
               g_checksum_update(checksum, digest2, digest_length);
               gsize len = digest_length;
               g_checksum_get_digest(checksum, digest_out, &len);
           } else {
               memcpy(digest_out, digest1, digest_length);
           }
       }
       hashes_length = j;
   }
   guchar* result = g_ptr_array_steal_index(hashes, 0);
   g_ptr_array_free(hashes, TRUE);
   g_checksum_free(checksum);
   return result;

}

int main(int argc, char** argv) {

   if (argc != 2) {
       fprintf(stderr, "usage: %s filename\n", argv[0]);
       return EXIT_FAILURE;
   }
   FILE* in = fopen(argv[1], "rb");
   if (in) {
       guchar* digest = sha256_merkle_tree(in, 1024);
       fclose(in);
       if (digest) {
           gssize length = g_checksum_type_get_length(G_CHECKSUM_SHA256);
           for (gssize i = 0; i < length; ++i)
               printf("%02x", digest[i]);
           printf("\n");
           g_free(digest);
       }
   } else {
       perror(argv[1]);
       return EXIT_FAILURE;
   }
   return EXIT_SUCCESS;

}</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

C++

Library: OpenSSL

<lang cpp>#include <cstdlib>

1. include <fstream>
2. include <iomanip>
3. include <iostream>
4. include <vector>
5. include <openssl/sha.h>

class sha256_exception : public std::exception { public:

   const char* what() const noexcept override {
       return "SHA-256 error";
   }

};

class sha256 { public:

   sha256() { reset(); }
   sha256(const sha256&) = delete;
   sha256& operator=(const sha256&) = delete;
   void reset() {
       if (SHA256_Init(&context_) == 0)
           throw sha256_exception();
   }
   void update(const void* data, size_t length) {
       if (SHA256_Update(&context_, data, length) == 0)
           throw sha256_exception();
   }
   std::vector<unsigned char> digest() {
       std::vector<unsigned char> digest(SHA256_DIGEST_LENGTH);
       if (SHA256_Final(digest.data(), &context_) == 0)
           throw sha256_exception();
       return digest;
   }

private:

   SHA256_CTX context_;

};

void print_digest(std::ostream& out, const std::vector<unsigned char>& digest) {

   out << std::hex << std::setfill('0');
   for (size_t i = 0; i < digest.size(); ++i)
       out << std::setw(2) << static_cast<int>(digest[i]);
   out << '\n';

}

std::vector<unsigned char> sha256_merkle_tree(std::istream& in, size_t block_size) {

   std::vector<std::vector<unsigned char>> hashes;
   std::vector<char> buffer(block_size);
   sha256 md;
   while (in) {
       in.read(buffer.data(), block_size);
       size_t bytes = in.gcount();
       if (bytes == 0)
           break;
       md.reset();
       md.update(buffer.data(), bytes);
       hashes.push_back(md.digest());
   }
   if (hashes.empty())
       return {};
   size_t length = hashes.size();
   while (length > 1) {
       size_t j = 0;
       for (size_t i = 0; i < length; i += 2, ++j) {
           auto& digest1 = hashes[i];
           auto& digest_out = hashes[j];
           if (i + 1 < length) {
               auto& digest2 = hashes[i + 1];
               md.reset();
               md.update(digest1.data(), digest1.size());
               md.update(digest2.data(), digest2.size());
               digest_out = md.digest();
           } else {
               digest_out = digest1;
           }
       }
       length = j;
   }
   return hashes[0];

}

int main(int argc, char** argv) {

   if (argc != 2) {
       std::cerr << "usage: " << argv[0] << " filename\n";
       return EXIT_FAILURE;
   }
   std::ifstream in(argv[1], std::ios::binary);
   if (!in) {
       std::cerr << "Cannot open file " << argv[1] << ".\n";
       return EXIT_FAILURE;
   }
   try {
       print_digest(std::cout, sha256_merkle_tree(in, 1024));
   } catch (const std::exception& ex) {
       std::cerr << ex.what() << "\n";
       return EXIT_FAILURE;
   }
   return EXIT_SUCCESS;

}</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Factor

Works with: Factor version 0.99 2020-08-14

<lang factor>USING: checksums checksums.sha fry grouping io io.encodings.binary io.files kernel make math math.parser namespaces sequences ;

each-block ( ... size quot: ( ... block -- ... ) -- ... )

   input-stream get spin (each-stream-block) ; inline

>sha-256 ( seq -- newseq ) sha-256 checksum-bytes ;

(hash-read) ( path encoding chunk-size -- )

   '[ _ [ >sha-256 , ] each-block ] with-file-reader ;

! Read a file in chunks as a sequence of sha-256 hashes, so as ! not to store a potentially large file in memory all at once.

hash-read ( path chunk-size -- seq )

   binary swap [ (hash-read) ] { } make ;

hash-combine ( seq -- newseq )

   2 <groups>
   [ dup length 1 > [ concat >sha-256 ] [ first ] if ] map ;

merkle-hash ( path chunk-size -- str )

   hash-read [ dup length 1 = ] [ hash-combine ] until first
   bytes>hex-string ;

"title.png" 1024 merkle-hash print</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Go

<lang go>package main

import (

   "crypto/sha256"
   "fmt"
   "io"
   "log"
   "os"

)

func main() {

   const blockSize = 1024
   f, err := os.Open("title.png")
   if err != nil {
       log.Fatal(err)
   }
   defer f.Close()

   var hashes [][]byte
   buffer := make([]byte, blockSize)
   h := sha256.New()
   for {
       bytesRead, err := f.Read(buffer)
       if err != nil {
           if err != io.EOF {
               log.Fatal(err)
           }
           break
       }
       h.Reset()
       h.Write(buffer[:bytesRead])
       hashes = append(hashes, h.Sum(nil))
   }
   buffer = make([]byte, 64)
   for len(hashes) > 1 {
       var hashes2 [][]byte
       for i := 0; i < len(hashes); i += 2 {
           if i < len(hashes)-1 {                
               copy(buffer, hashes[i])
               copy(buffer[32:], hashes[i+1])
               h.Reset()
               h.Write(buffer)
               hashes2 = append(hashes2, h.Sum(nil))
           } else {
               hashes2 = append(hashes2, hashes[i])
           }
       }
       hashes = hashes2
   }
   fmt.Printf("%x", hashes[0])
   fmt.Println()

}</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Java

<lang java>import java.io.*; import java.security.*; import java.util.*;

public class SHA256MerkleTree {

   public static void main(String[] args) {
       try {
           if (args.length != 1) {
               System.err.println("missing file argument");
               System.exit(1);
           }
           try (InputStream in = new BufferedInputStream(new FileInputStream(args[0]))) {
               byte[] digest = sha256MerkleTree(in, 1024);
               if (digest != null)
                   System.out.println(digestToString(digest));
           }
       } catch (Exception e) {
           e.printStackTrace();
       }
   }

   private static String digestToString(byte[] digest) {
       StringBuilder result = new StringBuilder();
       for (int i = 0; i < digest.length; ++i)
           result.append(String.format("%02x", digest[i]));
       return result.toString();
   }

   private static byte[] sha256MerkleTree(InputStream in, int blockSize) throws Exception {
       byte[] buffer = new byte[blockSize];
       int bytes;
       MessageDigest md = MessageDigest.getInstance("SHA-256");
       List<byte[]> digests = new ArrayList<>();
       while ((bytes = in.read(buffer)) > 0) {
           md.reset();
           md.update(buffer, 0, bytes);
           digests.add(md.digest());
       }
       int length = digests.size();
       if (length == 0)
           return null;
       while (length > 1) {
           int j = 0;
           for (int i = 0; i < length; i += 2, ++j) {
               byte[] digest1 = digests.get(i);
               if (i + 1 < length) {
                   byte[] digest2 = digests.get(i + 1);
                   md.reset();
                   md.update(digest1);
                   md.update(digest2);
                   digests.set(j, md.digest());
               } else {
                   digests.set(j, digest1);
               }
           }
           length = j;
       }
       return digests.get(0);
   }

}</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Julia

<lang julia>using SHA

function merkletree(filename="title.png", blocksize=1024)

   bytes = codeunits(read(filename, String))
   len = length(bytes)
   hsh = [sha256(view(bytes. i:min(i+blocksize-1, len)])) for i in 1:1024:len]
   len = length(hsh)
   while len > 1
       hsh = [i == len ? hsh[i] : sha256(vcat(hsh[i], hsh[i + 1])) for i in 1:2:len]
       len = length(hsh)
   end
   return bytes2hex(hsh[1])

end

println(merkletree())

</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Phix

<lang Phix>include builtins\libcurl.e include builtins\sha256.e

constant ONE_MB = 1024 * 1024

function merkle(string filename, url, integer block_size=ONE_MB)

   if not file_exists(filename) then
       printf(1,"Downloading %s...\n",{filename})
       CURLcode res = curl_easy_get_file(url,"",filename) -- (no proxy)
       if res!=CURLE_OK then
           string error = sprintf("%d",res)
           if res=CURLE_COULDNT_RESOLVE_HOST then
               error &= " [CURLE_COULDNT_RESOLVE_HOST]"
           end if
           crash("Error %s downloading file\n", {error})
       end if  
   end if  
   string data = get_text(filename)
   sequence blocks = {}
   for i=1 to length(data) by block_size do
       blocks = append(blocks,sha256(data[i..min(i+block_size-1,length(data))]))
   end for
   while length(blocks)>1 do
       integer l = 0
       for i=1 to length(blocks) by 2 do
           l += 1
           blocks[l] = iff(i<length(blocks)?sha256(blocks[i]&blocks[i+1])
                                           :blocks[i])
       end for
       blocks = blocks[1..l]
   end while
   return blocks[1]            

end function

function asHex(string s) string res = ""

   for i=1 to length(s) do
       res &= sprintf("%02X",s[i])
   end for
   return res

end function

printf(1,"%s\n",asHex(merkle("title.png", "https://rosettacode.org/mw/title.png", 1024)))</lang>

Output:

a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Python

This version attempts to combine blocks as soon as possible to minimize the memory footprint.

<lang Python>#!/usr/bin/env python

1. compute the root label for a SHA256 Merkle tree built on blocks of a given
2. size (default 1MB) taken from the given file(s)

import argh import hashlib import sys

@argh.arg('filename', nargs='?', default=None) def main(filename, block_size=1024*1024):

   if filename:
       fin = open(filename, 'rb')
   else: 
       fin = sys.stdin
   
   stack = []
   block = fin.read(block_size)
   while block:
       # a node is a pair: ( tree-level, hash )
       node = (0, hashlib.sha256(block).digest())
       stack.append(node)

       # concatenate adjacent pairs at the same level
       while len(stack) >= 2 and stack[-2][0] == stack[-1][0]:
           a = stack[-2]
           b = stack[-1]
           l = a[0]
           stack[-2:] = [(l+1, hashlib.sha256(a[1] + b[1]).digest())]

       block = fin.read(block_size)
   
   while len(stack) > 1:
       # at the end we have to concatenate even across levels
       a = stack[-2]
       b = stack[-1]
       al = a[0]
       bl = b[0]
       stack[-2:] = [(max(al, bl)+1, hashlib.sha256(a[1] + b[1]).digest())]

   print(stack[0][1].hex())

argh.dispatch_command(main) </lang>

Output:

$ sha256tree.py --block-size=1024 title.png
a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c

Raku

<lang perl6>use Digest::SHA256::Native;

unit sub MAIN(Int :b(:$block-size) = 1024 × 1024, *@args);

my $in = @args ?? IO::CatHandle.new(@args) !! $*IN;

my @blocks = do while my $block = $in.read: $block-size { sha256 $block };

while @blocks > 1 {

 @blocks = @blocks.batch(2).map: { $_ > 1 ?? sha256([~] $_) !! .[0] }

}

say @blocks[0]».fmt('%02x').join;</lang>

Output:

$ sha256tree --block-size=1024 title.png
a4f902cf9d51fe51eda156a6792e1445dff65edf3a217a1f3334cc9cf1495c2c