hash.h

Go to the documentation of this file.

#ifndef CVMFS_CRYPTO_HASH_H_
#define CVMFS_CRYPTO_HASH_H_

#include <arpa/inet.h>
#include <stdint.h>

#include <cassert>
#include <cctype>
#include <cstdlib>
#include <cstring>
#include <string>

#include "util/export.h"
#include "util/logging.h"
#include "util/prng.h"
#include "util/smalloc.h"

#ifdef CVMFS_NAMESPACE_GUARD
namespace CVMFS_NAMESPACE_GUARD {
#endif

namespace shash {

enum Algorithms {
  kMd5 = 0,
  kSha1,
  kRmd160,
  kShake128,  // with 160 output bits
  kAny,
};

const char kSuffixNone         = 0;
const char kSuffixCatalog      = 'C';
const char kSuffixHistory      = 'H';
const char kSuffixMicroCatalog = 'L';  // currently unused
const char kSuffixPartial      = 'P';
const char kSuffixTemporary    = 'T';
const char kSuffixCertificate  = 'X';
const char kSuffixMetainfo     = 'M';


const unsigned kDigestSizes[] =
  {16,  20,   20,     20,       20};
// Md5  Sha1  Rmd160  Shake128  Any
const unsigned kMaxDigestSize = 20;

const unsigned kMaxContextSize = 256;

CVMFS_EXPORT extern const char *kAlgorithmIds[];
const unsigned kAlgorithmIdSizes[] =
  {0,   0,    7,       9,         0};
// Md5  Sha1  -rmd160  -shake128  Any
const unsigned kMaxAlgorithmIdentifierSize = 9;

const unsigned kBlockSizes[] =
  {64,  64,   64,     168};
// Md5  Sha1  Rmd160  Shake128

struct CVMFS_EXPORT HexPtr {
  const std::string *str;
  explicit HexPtr(const std::string &s) { str = &s; }
  bool IsValid() const;
};

struct CVMFS_EXPORT AsciiPtr {
  const std::string *str;
  explicit AsciiPtr(const std::string &s) { str = &s; }
};

typedef char Suffix;

template<unsigned digest_size_, Algorithms algorithm_>
struct CVMFS_EXPORT Digest {
  unsigned char digest[digest_size_];
  Algorithms    algorithm;
  Suffix        suffix;

  class Hex {
   public:
    explicit Hex(const Digest<digest_size_, algorithm_> *digest) :
      digest_(*digest),
      hash_length_(2 * kDigestSizes[digest_.algorithm]),
      algo_id_length_(kAlgorithmIdSizes[digest_.algorithm]) {}

    unsigned int length() const { return hash_length_ + algo_id_length_; }

    char operator[](const unsigned int position) const {
      assert(position < length());
      return (position < hash_length_)
        ? GetHashChar(position)
        : GetAlgorithmIdentifierChar(position);
    }

   protected:
    char GetHashChar(const unsigned int position) const {
      assert(position < hash_length_);
      const char digit = (position % 2 == 0)
        ? digest_.digest[position / 2] / 16
        : digest_.digest[position / 2] % 16;
      return ToHex(digit);
    }

    char GetAlgorithmIdentifierChar(const unsigned int position) const {
      assert(position >= hash_length_);
      return kAlgorithmIds[digest_.algorithm][position - hash_length_];
    }

    char ToHex(const char c) const {
      return static_cast<char>(c + ((c <= 9) ? '0' : 'a' - 10));
    }

   private:
    const Digest<digest_size_, algorithm_>  &digest_;
    const unsigned int                       hash_length_;
    const unsigned int                       algo_id_length_;
  };

  unsigned GetDigestSize() const { return kDigestSizes[algorithm]; }
  unsigned GetHexSize() const {
    return 2*kDigestSizes[algorithm] + kAlgorithmIdSizes[algorithm];
  }

  Digest() :
    algorithm(algorithm_), suffix(kSuffixNone)
  {
    SetNull();
  }

  explicit Digest(const Algorithms a, const HexPtr hex, const char s = 0) :
    algorithm(a), suffix(s)
  {
    assert((algorithm_ == kAny) || (a == algorithm_));
    const unsigned char_size = 2*kDigestSizes[a];

    const std::string *str = hex.str;
    const unsigned length = str->length();
    assert(length >= char_size);  // A suffix won't hurt

    for (unsigned i = 0; i < char_size; i += 2) {
      this->digest[i/2] =
        ((*str)[i] <= '9' ? (*str)[i] -'0' : (*str)[i] - 'a' + 10)*16 +
        ((*str)[i+1] <= '9' ? (*str)[i+1] - '0' : (*str)[i+1] - 'a' + 10);
    }
  }

  Digest(const Algorithms a,
         const unsigned char *digest_buffer,
         const Suffix s = kSuffixNone) :
    algorithm(a), suffix(s)
  {
    memcpy(digest, digest_buffer, kDigestSizes[a]);
  }

  void Randomize() {
    Prng prng;
    prng.InitLocaltime();
    Randomize(&prng);
  }

  void Randomize(const uint64_t seed) {
    Prng prng;
    prng.InitSeed(seed);
    Randomize(&prng);
  }

  void Randomize(Prng *prng) {
    const unsigned bytes = GetDigestSize();
    for (unsigned i = 0; i < bytes; ++i) {
      digest[i] = prng->Next(256);
    }
  }

  bool HasSuffix() const { return suffix != kSuffixNone; }
  void set_suffix(const Suffix s) { suffix = s; }

  std::string ToString(const bool with_suffix = false) const {
    Hex hex(this);
    const bool     use_suffix  = with_suffix && HasSuffix();
    const unsigned string_length = hex.length() + use_suffix;
    std::string result(string_length, 0);

    for (unsigned int i = 0; i < hex.length(); ++i) {
      result[i] = hex[i];
    }

    if (use_suffix) {
      result[string_length - 1] = suffix;
    }

    assert(result.length() == string_length);
    return result;
  }

  std::string ToFingerprint(const bool with_suffix = false) const {
    Hex hex(this);
    const bool     use_suffix  = with_suffix && HasSuffix();
    const unsigned string_length =
      hex.length() + kDigestSizes[algorithm] - 1 + use_suffix;
    std::string result(string_length, 0);

    unsigned l = hex.length();
    for (unsigned int hex_i = 0, result_i = 0; hex_i < l; ++hex_i, ++result_i) {
      result[result_i] = toupper(hex[hex_i]);
      if ((hex_i < 2 * kDigestSizes[algorithm] - 1) && (hex_i % 2 == 1)) {
        result[++result_i] = ':';
      }
    }

    if (use_suffix) {
      result[string_length - 1] = suffix;
    }

    assert(result.length() == string_length);
    return result;
  }

  std::string ToStringWithSuffix() const {
    return ToString(true);
  }

  std::string MakePath() const {
    return MakePathExplicit(1, 2, suffix);
  }

  std::string MakeAlternativePath() const {
    return ".cvmfsalt-" + ToStringWithSuffix();
  }

  std::string MakePathWithoutSuffix() const {
    return MakePathExplicit(1, 2, kSuffixNone);
  }

  std::string MakePathExplicit(const unsigned dir_levels,
                               const unsigned digits_per_level,
                               const Suffix   hash_suffix = kSuffixNone) const {
    Hex hex(this);

    // figure out how big the output string needs to be
    const bool use_suffix = (hash_suffix != kSuffixNone);
    const unsigned string_length = hex.length() + dir_levels + use_suffix;
    std::string result;
    result.resize(string_length);

    // build hexified hash and path delimiters
    unsigned i   = 0;
    unsigned pos = 0;
    for (; i < hex.length(); ++i) {
      if (i > 0 && (i % digits_per_level == 0)
                && (i / digits_per_level <= dir_levels)) {
        result[pos++] = '/';
      }
      result[pos++] = hex[i];
    }

    // (optionally) add hash hint suffix
    if (use_suffix) {
      result[pos++] = hash_suffix;
    }

    assert(i   == hex.length());
    assert(pos == string_length);
    return result;
  }

  bool IsNull() const {
    for (unsigned i = 0; i < kDigestSizes[algorithm]; ++i) {
      if (digest[i] != 0)
        return false;
    }
    return true;
  }

  uint32_t Partial32() const {
    const uint32_t *partial = (const uint32_t *)digest;
    return ntohl(*partial);
  }


  void SetNull() {
    memset(digest, 0, digest_size_);
  }


  bool operator ==(const Digest<digest_size_, algorithm_> &other) const {
    if (this->algorithm != other.algorithm)
      return false;
    for (unsigned i = 0; i < kDigestSizes[algorithm]; ++i) {
      if (this->digest[i] != other.digest[i])
        return false;
    }
    return true;
  }

  bool operator !=(const Digest<digest_size_, algorithm_> &other) const {
    return !(*this == other);
  }

  bool operator <(const Digest<digest_size_, algorithm_> &other) const {
    if (this->algorithm != other.algorithm)
      return (this->algorithm < other.algorithm);
    for (unsigned i = 0; i < kDigestSizes[algorithm]; ++i) {
      if (this->digest[i] > other.digest[i])
        return false;
      if (this->digest[i] < other.digest[i])
        return true;
    }
    return false;
  }

  bool operator >(const Digest<digest_size_, algorithm_> &other) const {
    if (this->algorithm != other.algorithm)
      return (this->algorithm > other.algorithm);
    for (unsigned i = 0; i < kDigestSizes[algorithm]; ++i) {
      if (this->digest[i] < other.digest[i])
        return false;
      if (this->digest[i] > other.digest[i])
        return true;
    }
    return false;
  }
};


struct CVMFS_EXPORT Md5 : public Digest<16, kMd5> {
  Md5() : Digest<16, kMd5>() { }
  explicit Md5(const AsciiPtr ascii);
  explicit Md5(const HexPtr hex) : Digest<16, kMd5>(kMd5, hex) { }
  Md5(const char *chars, const unsigned length);

  Md5(const uint64_t lo, const uint64_t hi);
  void ToIntPair(uint64_t *lo, uint64_t *hi) const;
};

struct CVMFS_EXPORT Sha1 : public Digest<20, kSha1> { };
struct CVMFS_EXPORT Rmd160 : public Digest<20, kRmd160> { };
struct CVMFS_EXPORT Shake128 : public Digest<20, kShake128> { };

struct CVMFS_EXPORT Any : public Digest<kMaxDigestSize, kAny> {
  Any() : Digest<kMaxDigestSize, kAny>() { }

  explicit Any(const Algorithms a,
               const char       s = kSuffixNone) :
    Digest<kMaxDigestSize, kAny>() { algorithm = a; suffix = s; }

  Any(const Algorithms     a,
      const unsigned char *digest_buffer,
      const Suffix         suffix = kSuffixNone) :
    Digest<kMaxDigestSize, kAny>(a, digest_buffer, suffix) { }

  explicit Any(const Algorithms  a,
               const HexPtr      hex,
               const char        suffix = kSuffixNone) :
    Digest<kMaxDigestSize, kAny>(a, hex, suffix) { }

  Md5 CastToMd5();
};


CVMFS_EXPORT unsigned GetContextSize(const Algorithms algorithm);

class CVMFS_EXPORT ContextPtr {
 public:
  Algorithms  algorithm;
  void       *buffer;
  unsigned    size;

  ContextPtr() : algorithm(kAny), buffer(NULL), size(0) {}

  explicit ContextPtr(const Algorithms a) :
    algorithm(a), buffer(NULL), size(GetContextSize(a)) {}
  ContextPtr(const Algorithms a, void *b) :
    algorithm(a), buffer(b), size(GetContextSize(a)) {}
};

CVMFS_EXPORT void Init(ContextPtr context);
CVMFS_EXPORT void Update(const unsigned char *buffer,
                         const unsigned buffer_size,
                         ContextPtr context);
CVMFS_EXPORT void Final(ContextPtr context, Any *any_digest);
CVMFS_EXPORT bool HashFile(const std::string &filename, Any *any_digest);
CVMFS_EXPORT bool HashFd(int fd, Any *any_digest);
CVMFS_EXPORT void HashMem(const unsigned char *buffer,
                          const unsigned buffer_size,
                          Any *any_digest);
CVMFS_EXPORT void HashString(const std::string &content, Any *any_digest);
CVMFS_EXPORT void Hmac(const std::string &key,
                       const unsigned char *buffer,
                       const unsigned buffer_size,
                       Any *any_digest);
inline void HmacString(const std::string &key, const std::string &content,
                       Any *any_digest)
{
  Hmac(key,
       reinterpret_cast<const unsigned char *>(content.data()),
       content.size(),
       any_digest);
}

CVMFS_EXPORT std::string Hmac256(const std::string &key,
                                 const std::string &content,
                                 bool raw_output = false);
CVMFS_EXPORT std::string Sha256File(const std::string &filename);
CVMFS_EXPORT std::string Sha256Mem(const unsigned char *buffer,
                                   const unsigned buffer_size);
CVMFS_EXPORT std::string Sha256String(const std::string &content);

CVMFS_EXPORT
Algorithms ParseHashAlgorithm(const std::string &algorithm_option);
CVMFS_EXPORT
Any MkFromHexPtr(const HexPtr hex, const Suffix suffix = kSuffixNone);
CVMFS_EXPORT Any MkFromSuffixedHexPtr(const HexPtr hex);

}  // namespace shash

#ifdef CVMFS_NAMESPACE_GUARD
}  // namespace CVMFS_NAMESPACE_GUARD
#endif

#endif  // CVMFS_CRYPTO_HASH_H_