Deep understanding of Base64 underlying principles

Base64 Is a common data encoding method , Commonly used for data transmission . For mobile developers , Network requests often use . Yes JSON Familiar students all know ,JSON The serialization tool of does not support byte Array directly into JSON In the data , For this binary data , In dealing with , All need to be done Base64 code , Then put the encoded string into JSON in . From the beginning to now ,Base64 It's a very common function , From the Internet “ Carry ” Base64 Codec tools to use Android SDK The built-in Base64. Although skilled in use , But I didn't understand it deeply Base64 The underlying mechanism of .

Coincidentally , In the near future , I met again Base64 Related matters of , So I went to have a relatively in-depth understanding Base64 Related content of . The following is only for your own learning records .

1. Base64 Definition

Base64 It's based on 64 A representation of binary data that represents three printable characters , In human words, you can call Binary system Data is printed on the console , It can copy and transmit well .

Used in coding 64 Printable characters only need 6 individual Bit You can say ： $2^6=64$

In the computer world ,3 individual Byte Binary data of is equivalent to 24 individual Bit , Therefore, when encoding, you need to correspond 4 individual Base64 To represent .

Base64 The printable character set of contains letters A-Z、a-Z、 Numbers 0-9, That's it 62 Characters , The other two may be different in different systems , We usually use + 、 /, stay URL Safe In the mode of -、_

Base64 clock ：

2. Base64 transformation

We will byte Every one of the arrays byte All converted into 8 position bit, Form a string of binary data , Then press 6 Bit by bit , Then every 6 Bits are converted to a bit , Will be converted bit As index , Go to the code table to find the corresponding displayed characters .

The coding example is as follows ：

As shown in the figure above , take LZC convert to Base64 The encoded printable string is TFpD. From this example, we can see , Just enough to 3 individual byte Turn into 4 individual byte.

Smart you may have found the existing problems , If you want to encode the original byte[] What should I do if the length of cannot be divided by three ？

In the code table , Except for those used in coding 64 Characters , There is also a padding character at the end ： = , Used to fill in insufficient digits .

Illustrate with examples :

3. Base64 encode Code implementation （Java edition ）

1. Put the existing 3 individual byte a, b, c Code as Base64 String of

final char[] base64 = {
    
    'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
    'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
    'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
    'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
    '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/'
};
byte a = 'L';
byte b = 'Z';
byte c = 'C';

int index1 = (a >> 2) & 0b00111111;
int index2 = ((a << 4) & 0b00110000) | ((b >> 4) & 0b00001111);
int index3 = ((b << 2) & 0b00111100) | ((c >> 6) & 0b00000011);
int index4 = c & 0b00111111;

byte[] result = new byte[]{
    
    (byte) base64[index1], (byte) base64[index2], (byte) base64[index3], (byte) base64[index4]
};
System.out.println(new String(result));

// TFpD

2. Put the existing 2 individual byte a, b Code as Base64 String of

byte a = 'Z';
byte b = 'C';

int index1 = (a >> 2) & 0b00111111;
int index2 = ((a << 4) & 0b00110000) | ((b >> 4) & 0b00001111);
int index3 = ((b << 2) & 0b00111111);
byte[] result = new byte[]{
    
    (byte) base64[index1], (byte) base64[index2], (byte) base64[index3], '='
};
System.out.println(new String(result));

// WkM=

3. Put the existing 1 individual byte a Code as Base64 String of

byte a = 'L';
int index1 = (a >> 2) & 0b00111111;
int index2 = ((a << 4) & 0b00111111);
byte[] result = new byte[]{
    
    (byte) base64[index1], (byte) base64[index2], '=', '='
};
System.out.println(new String(result));

// TA==

​

From code , You can see the conversion steps described above . But in calculating index2 as well as index3 When , It's still a little winding , Is there a better way to calculate ？

stay JDK Of Base64 In the implementation of , See a new implementation . stay Java in , One int Number of type 4 individual byte , So you can convert 3 position byte In a int In the data .

As shown in the figure below , take LZC Three byte Put one in int Type variable ：

Code implementation ：

byte a = 'L';
byte b = 'Z';
byte c = 'C';
int bits = (a & 0xFF) << 16 | (b & 0xFF) << 8 | (c & 0xFF);

// 10011000101101001000011

Get bits after , You can directly perform bit operations , Take out the corresponding 4 individual 6 Bit binary number . The code implementation is as follows ：

// 0x3f == 0b00111111
int index1 = (bits >> 18) & 0x3F;
int index2 = (bits >> 12) & 0x3F;
int index3 = (bits >> 6) & 0x3F;
int index4 = bits  & 0x3F;


byte[] result = new byte[]{
    
  (byte) base64[index1], (byte) base64[index2], (byte) base64[index3], (byte) base64[index4]
};
System.out.println(new String(result));

4. Think about a ：Base64 Coding must add padding （Padding） Do you ？

In the first article , For lack of 3 individual byte In the process of data conversion , You need to add padding characters to the encoded string = , Let the last output Base64 The string of is exactly 4 Integer multiple . So here comes the question , Why use = Number to fill ？ Is it OK not to fill ？

stay android.utils.Base64 In the implementation of , One called NO_PADDING Of flag. If in use , Passed in this flag, The final output result is no embedding = Result . therefore , When coding , For lack of 3 individual byte Data processing of , You can also not fill .

Of course , If filling , In decoding , It can be handled more simply , Directly encode the string ,4 A group of characters is processed , Without judging the number of characters , Take different decoding logic . Of course ,Padding It can also be used to identify the end of coding , To prevent multiple Base64 After splicing together , Decoding failed .

5. Think about the second ：Base64 Can it be considered as an encryption method ？

Base64 It's a way of encoding , Convert the original data into a printable string . At work , It is often called Base64 encryption , according to encryption The definition of ：

Change the original information data with a special algorithm , Enables unauthorized users to obtain encrypted information , But I don't know how to decrypt it , Still unable to understand the content of the information .

about Base64 Speaking of , In use , It has already complied with common coding standards ： RFC 4648 as well as RFC 2045, Use a standard code table for processing , It can't be regarded as “ encryption ”, But if , In the process of encoding and decoding , Use a special code table , It can make the encoded content not easy to be decoded , It can also realize the original content “ encryption ” function .

6. Last

In the text , of Base64 The principle of the coding process and the code implementation have already existed , If you are interested , You can try to realize the coding of decoding , We can better understand the principle . Know what it is , Know why , Writing code is like God's help .