How do you convert a byte array to a hexadecimal s

2楼-- · 2018-12-30 23:33

There's a class called SoapHexBinary that does exactly what you want.

using System.Runtime.Remoting.Metadata.W3cXsd2001;

public static byte[] GetStringToBytes(string value)
{
    SoapHexBinary shb = SoapHexBinary.Parse(value);
    return shb.Value;
}

public static string GetBytesToString(byte[] value)
{
    SoapHexBinary shb = new SoapHexBinary(value);
    return shb.ToString();
}

0人赞添加讨论(0) 举报

谁念西风独自凉

3楼-- · 2018-12-30 23:34

This problem could also be solved using a look-up table. This would require a small amount of static memory for both the encoder and decoder. This method will however be fast:

Encoder table 512 bytes or 1024 bytes (twice the size if both upper and lower case is needed)
Decoder table 256 bytes or 64 KiB (either a single char look-up or dual char look-up)

My solution uses 1024 bytes for the encoding table, and 256 bytes for decoding.

Decoding

private static readonly byte[] LookupTable = new byte[] {
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF
};

private static byte Lookup(char c)
{
  var b = LookupTable[c];
  if (b == 255)
    throw new IOException("Expected a hex character, got " + c);
  return b;
}

public static byte ToByte(char[] chars, int offset)
{
  return (byte)(Lookup(chars[offset]) << 4 | Lookup(chars[offset + 1]));
}

Encoding

private static readonly char[][] LookupTableUpper;
private static readonly char[][] LookupTableLower;

static Hex()
{
  LookupTableLower = new char[256][];
  LookupTableUpper = new char[256][];
  for (var i = 0; i < 256; i++)
  {
    LookupTableLower[i] = i.ToString("x2").ToCharArray();
    LookupTableUpper[i] = i.ToString("X2").ToCharArray();
  }
}

public static char[] ToCharLower(byte[] b, int bOffset)
{
  return LookupTableLower[b[bOffset]];
}

public static char[] ToCharUpper(byte[] b, int bOffset)
{
  return LookupTableUpper[b[bOffset]];
}

Comparison

StringBuilderToStringFromBytes:   106148
BitConverterToStringFromBytes:     15783
ArrayConvertAllToStringFromBytes:  54290
ByteManipulationToCharArray:        8444
TableBasedToCharArray:              5651 *

* this solution

Note

During decoding IOException and IndexOutOfRangeException could occur (if a character has a too high value > 256). Methods for de/encoding streams or arrays should be implemented, this is just a proof of concept.

0人赞添加讨论(0) 举报

临风纵饮

4楼-- · 2018-12-30 23:34

From Microsoft's developers, a nice, simple conversion:

public static string ByteArrayToString(byte[] ba) 
{
    // Concatenate the bytes into one long string
    return ba.Aggregate(new StringBuilder(32),
                            (sb, b) => sb.Append(b.ToString("X2"))
                            ).ToString();
}

While the above is clean an compact, performance junkies will scream about it using enumerators. You can get peak performance with an improved version of Tomolak's original answer:

public static string ByteArrayToString(byte[] ba)   
{   
   StringBuilder hex = new StringBuilder(ba.Length * 2);   

   for(int i=0; i < ga.Length; i++)       // <-- Use for loop is faster than foreach   
       hex.Append(ba[i].ToString("X2"));   // <-- ToString is faster than AppendFormat   

   return hex.ToString();   
}

This is the fastest of all the routines I've seen posted here so far. Don't just take my word for it... performance test each routine and inspect its CIL code for yourself.

0人赞添加讨论(0) 举报

与风俱净

5楼-- · 2018-12-30 23:35

Safe versions:

public static class HexHelper
{
    [System.Diagnostics.Contracts.Pure]
    public static string ToHex(this byte[] value)
    {
        if (value == null)
            throw new ArgumentNullException("value");

        const string hexAlphabet = @"0123456789ABCDEF";

        var chars = new char[checked(value.Length * 2)];
        unchecked
        {
            for (int i = 0; i < value.Length; i++)
            {
                chars[i * 2] = hexAlphabet[value[i] >> 4];
                chars[i * 2 + 1] = hexAlphabet[value[i] & 0xF];
            }
        }
        return new string(chars);
    }

    [System.Diagnostics.Contracts.Pure]
    public static byte[] FromHex(this string value)
    {
        if (value == null)
            throw new ArgumentNullException("value");
        if (value.Length % 2 != 0)
            throw new ArgumentException("Hexadecimal value length must be even.", "value");

        unchecked
        {
            byte[] result = new byte[value.Length / 2];
            for (int i = 0; i < result.Length; i++)
            {
                // 0(48) - 9(57) -> 0 - 9
                // A(65) - F(70) -> 10 - 15
                int b = value[i * 2]; // High 4 bits.
                int val = ((b - '0') + ((('9' - b) >> 31) & -7)) << 4;
                b = value[i * 2 + 1]; // Low 4 bits.
                val += (b - '0') + ((('9' - b) >> 31) & -7);
                result[i] = checked((byte)val);
            }
            return result;
        }
    }
}

Unsafe versions For those who prefer performance and do not afraid of unsafeness. About 35% faster ToHex and 10% faster FromHex.

public static class HexUnsafeHelper
{
    [System.Diagnostics.Contracts.Pure]
    public static unsafe string ToHex(this byte[] value)
    {
        if (value == null)
            throw new ArgumentNullException("value");

        const string alphabet = @"0123456789ABCDEF";

        string result = new string(' ', checked(value.Length * 2));
        fixed (char* alphabetPtr = alphabet)
        fixed (char* resultPtr = result)
        {
            char* ptr = resultPtr;
            unchecked
            {
                for (int i = 0; i < value.Length; i++)
                {
                    *ptr++ = *(alphabetPtr + (value[i] >> 4));
                    *ptr++ = *(alphabetPtr + (value[i] & 0xF));
                }
            }
        }
        return result;
    }

    [System.Diagnostics.Contracts.Pure]
    public static unsafe byte[] FromHex(this string value)
    {
        if (value == null)
            throw new ArgumentNullException("value");
        if (value.Length % 2 != 0)
            throw new ArgumentException("Hexadecimal value length must be even.", "value");

        unchecked
        {
            byte[] result = new byte[value.Length / 2];
            fixed (char* valuePtr = value)
            {
                char* valPtr = valuePtr;
                for (int i = 0; i < result.Length; i++)
                {
                    // 0(48) - 9(57) -> 0 - 9
                    // A(65) - F(70) -> 10 - 15
                    int b = *valPtr++; // High 4 bits.
                    int val = ((b - '0') + ((('9' - b) >> 31) & -7)) << 4;
                    b = *valPtr++; // Low 4 bits.
                    val += (b - '0') + ((('9' - b) >> 31) & -7);
                    result[i] = checked((byte)val);
                }
            }
            return result;
        }
    }
}

BTW For benchmark testing initializing alphabet every time convert function called is wrong, alphabet must be const (for string) or static readonly (for char[]). Then alphabet-based conversion of byte[] to string becomes as fast as byte manipulation versions.

And of course test must be compiled in Release (with optimization) and with debug option "Suppress JIT optimization" turned off (same for "Enable Just My Code" if code must be debuggable).

0人赞添加讨论(0) 举报

春风洒进眼中

6楼-- · 2018-12-30 23:36

For performance I would go with drphrozens solution. A tiny optimization for the decoder could be to use a table for either char to get rid of the "<< 4".

Clearly the two method calls are costly. If some kind of check is made either on input or output data (could be CRC, checksum or whatever) the if (b == 255)... could be skipped and thereby also the method calls altogether.

Using offset++ and offset instead of offset and offset + 1 might give some theoretical benefit but I suspect the compiler handles this better than me.

private static readonly byte[] LookupTableLow = new byte[] {
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF
};

private static readonly byte[] LookupTableHigh = new byte[] {
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0x00, 0x10, 0x20, 0x30, 0x40, 0x50, 0x60, 0x70, 0x80, 0x90, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xA0, 0xB0, 0xC0, 0xD0, 0xE0, 0xF0, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xA0, 0xB0, 0xC0, 0xD0, 0xE0, 0xF0, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
  0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF
};

private static byte LookupLow(char c)
{
  var b = LookupTableLow[c];
  if (b == 255)
    throw new IOException("Expected a hex character, got " + c);
  return b;
}

private static byte LookupHigh(char c)
{
  var b = LookupTableHigh[c];
  if (b == 255)
    throw new IOException("Expected a hex character, got " + c);
  return b;
}

public static byte ToByte(char[] chars, int offset)
{
  return (byte)(LookupHigh(chars[offset++]) | LookupLow(chars[offset]));
}

This is just off the top of my head and has not been tested or benchmarked.

0人赞添加讨论(0) 举报

琉璃瓶的回忆

7楼-- · 2018-12-30 23:37

When writing crypto code it's common to avoid data dependent branches and table lookups to ensure the runtime doesn't depend on the data, since data dependent timing can lead to side-channel attacks.

It's also pretty fast.

static string ByteToHexBitFiddle(byte[] bytes)
{
    char[] c = new char[bytes.Length * 2];
    int b;
    for (int i = 0; i < bytes.Length; i++) {
        b = bytes[i] >> 4;
        c[i * 2] = (char)(55 + b + (((b-10)>>31)&-7));
        b = bytes[i] & 0xF;
        c[i * 2 + 1] = (char)(55 + b + (((b-10)>>31)&-7));
    }
    return new string(c);
}

_{Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn}

Abandon all hope, ye who enter here

An explanation of the weird bit fiddling:

bytes[i] >> 4 extracts the high nibble of a byte
bytes[i] & 0xF extracts the low nibble of a byte
b - 10
is < 0 for values b < 10, which will become a decimal digit
is >= 0 for values b > 10, which will become a letter from A to F.
Using i >> 31 on a signed 32 bit integer extracts the sign, thanks to sign extension. It will be -1 for i < 0 and 0 for i >= 0.
Combining 2) and 3), shows that (b-10)>>31 will be 0 for letters and -1 for digits.
Looking at the case for letters, the last summand becomes 0, and b is in the range 10 to 15. We want to map it to A(65) to F(70), which implies adding 55 ('A'-10).
Looking at the case for digits, we want to adapt the last summand so it maps b from the range 0 to 9 to the range 0(48) to 9(57). This means it needs to become -7 ('0' - 55).
Now we could just multiply with 7. But since -1 is represented by all bits being 1, we can instead use & -7 since (0 & -7) == 0 and (-1 & -7) == -7.

Some further considerations:

I didn't use a second loop variable to index into c, since measurement shows that calculating it from i is cheaper.
Using exactly i < bytes.Length as upper bound of the loop allows the JITter to eliminate bounds checks on bytes[i], so I chose that variant.
Making b an int allows unnecessary conversions from and to byte.

0人赞添加讨论(0) 举报